Denzin v Nutrasweet
[1999] NSWSC 106
•22 February 1999
CITATION: Denzin & ors v Nutrasweet & ors [1999] NSWSC 106 CURRENT JURISDICTION: Common Law Division FILE NUMBER(S): 10927/90; 11890/98; 12141/89; 12142/89; 13864/89; 13983/89; 11354/90; 123255/90; 13528/93 HEARING DATE(S): JUDGMENT DATE:
22 February 1999PARTIES :
Fiona Mary Denzin P1
Margaret Ann Lane P2
Margaret Orders P3
Brenda Southern P4
Sonya Lee Moylan P5
Sharon Gai Robertson P6
Karen Cynthia Tabbah P7
Deborah Lea Gentle P8
Wendy Janene Lee P9
The Nutrasweet Company D1JUDGMENT OF: Bruce J
COUNSEL : SOLICITORS: CATCHWORDS: DECISION: for the defendants
THE SUPREME COURT
OF NEW SOUTH WALES
COMMON LAW DIVISIONCORAM: BRUCE J
MONDAY 22 FEBRUARY 1999
010927/90 FIONA MARY DENZIN v THE NUTRASWEET COMPANY & 5 ORS.
011890/89 MARGARET ANN LANE v THE NUTRASWEET COMPANY 6 ORS.
012141/89 MARGARET ORDERS v THE NUTRASWEET COMPANY 5 ORS.
012142/89 BRENDA SOUTHERN v THE NUTRASWEET COMPANY & 5 ORS.
013864/89 SONYA LEE MOYLAN v THE NUTRASWEET COMPANY & 5 ORS.
013983/89 SHARON GAI ROBERTSON v THE NUTRASWEET COMPANY & 5 ORS.
011354/90 KAREN CYNTHIA TABBAH v THE NUTRASWEET COMPANY & 5 ORS.
0123255/90 DEBORAH LEA GENTLE v THE NUTRASWEET COMPANY & 5 ORS.
013528/93 WENDY JANENE LEE v THE NUTRASWEET COMPANY & 3 ORS.
JUDGMENT
TABLE OF CONTENTSSummary (i)-(iii)
Page
1. Introduction 1-43
2. Pelvic Inflammatory Disease 44-65
3. Causation 66-98
4. Probability of Causation 99
4.1 Epidemiology 100-207
4.2 Scientific Analysis 208-243
4.3 Clinical Observations 244
5. Warnings 245
6. Conclusion 246-247
Schedule I 248-408
Schedule II 409-410Schedule III 411-412
SUMMARY
The plaintiffs in these proceedings have each suffered considerable physical and emotional injury as a result of contracting pelvic inflammatory disease which has affected their fertility. There is no doubt they have each suffered considerably.They have sued the manufacturer and suppliers of the IUD device they each used essentially on the basis that the manufacturer was negligent in the design of the device and failed to warn them of risks to their health of which they were aware.
The litigation was the longest and most complex product liability litigation in Australian legal history. The hearing of evidence commenced on 24 January 1996 and concluded on 13 September 1996. The oral evidence comprised substantially approximately 15 lineal metres of folders of exhibits. The written submissions of the parties comprised approximately 3 lineal metres of folders of submissions and supporting documents. Oral submissions were delivered between 27 October, 1997 and 23 December, 1997.
It was common ground that injury of the type sustained by the plaintiffs could be sustained without the use of the defendants’ device and that it could also be sustained with the use of the defendants’ device. The plaintiffs submitted that if they showed any negligence on the part of the defendants which materially increased the risk to the users of the devices then that was sufficient to establish causation of their injuries so as to entitle them to succeed against the defendants unless the defendants established that the injury was due to some other cause.
In my view this is not the law and until it is changed by either a decision of the High Court of Australia or by the Parliament of New South Wales it is necessary for the plaintiffs to establish that their injuries were, on the balance of probability, caused by or materially contributed to by the negligent acts of the defendants.
In my view the plaintiffs have not established that the defendants were negligent in any material way nor that the actions of the defendants caused or materially contributed to the injuries to the plaintiffs.
Further, in my view, the plaintiffs have not established any breach of any duty to warn the plaintiffs of any matter.
There will be judgment for the defendants in each of the cases.
1. INTRODUCTION
This litigation involves nine lead cases selected from some two hundred commenced in this Court claiming damages for injury said to arise from the use of an interuterine device known as Copper 7 (Cu-7) or Gravigard or from the use of another interuterine device known as the Mini-Gravigard which was a smaller version of the first device. These devices will be referred to respectively as “ Cu-7 ” and “ Mini ”.The devices used fine copper wire carried on a plastic vehicle which is somewhat in the shape of the number 7 for contraceptive purposes. The devices were inserted into the uterus of the user and there was a tail of polypropylene attached to the device which after insertion passed from the device through the cervix of the user into her vagina.
The devices were apparently extremely successful in preventing pregnancies occurring while they were in place. The mechanism by which the devices achieved their objective is complex and not known with certainty even by the manufacturer but an understanding of the mechanism is not necessary for the determination of the issues which arise in these cases.
The Cu-7 was initially developed by a Chilean doctor, Hime Zipper. A licence agreement was entered into on 13 April 1970 between a corporation with which Dr Zipper was associated and one of the defendants which is a United States corporation. That licence agreement ran until 31 December 1985 when there was an option to purchase the intellectual property associated with the device. By about that time the Cu-7 was no longer marketed in the United States although it continued to be marketed in other countries including Australia until 1990.
The Mini was marketed in Australia but not in the United States. This device was apparently regarded as being more appropriate than the Cu-7 for use by women who had never conceived.
Each of the nine plaintiffs had one or more Cu-7s inserted in the period between 1974 and 1981.
In the selection of the nine lead cases before the Court four plaintiffs were identified as parous, that is they had already borne children at the time they used the Cu-7 devices, namely Moylan, Ottaway, Robertson, Southren and four as nulliparous, that is they had never borne a child, namely Gentle, Lane, Lee and Orders.
One of the plaintiffs, Denzin, was nulliparous when she began using the Cu-7 and when she subsequently used the Mini, the first in 1983 and the second in 1985.
Four of the plaintiffs, Denzin, Gentle, Lee and Ottaway had documented string retractions, that is the polypropolene string attached to the plastic device retracted from the vagina into the uterus after the device had been inserted.
Four of the plaintiffs, Denzin, Orders, Ottaway and Southren had an ectopic pregnancy or ectopic pregnancies.
One of the plaintiffs, Gentle, had a septic abortion with her Cu-7 in place.
All of the plaintiffs except Gentle suffered either complete infertility or impaired fertility.
The defendants are all members of what is now the Monsanto Group of companies. The first defendant carried on all aspects of the business associated with the Cu-7 at all times at least up to 24 May 1978. At that time it was known as G D Searle and Co., but subsequently changed its name to The NutraSweet Company.
The second defendant was called Searle Australia Pty Limited at all relevant times and is now known as Monsanto Australia Pty Limited. It carried on the business of distribution, sales, promotion, marketing and provision of information and warnings for health care providers and users within Australia at all relevant times.
The third defendant carried on the business associated with the Cu-7 in the period from 24 May 1978 to 1 January 1986 and the fourth defendant carried on that business from 1 January 1986. The fourth defendant changed its name to G D Searle & Co. on 2 January 1986. However, almost all references to “G D Searle & Co.” are references to the first defendant.
The plaintiffs’ case was in essence that use of the devices caused the plaintiffs to contract Pelvic Inflammatory Disease (PID) causing damage to their reproductive organs and thus either permanent or partial infertility and, in some instances, damage to the fallopian tubes resulting in ectopic pregnancies.
They submitted that the use of the devices put the users at greater risk of the introduction of infection into the uterine cavity or the fallopian tubes or to the ovaries than was the risk to non-users of the devices.
The plaintiffs claimed exemplary damages on the basis that the defendants acted in knowing and contumelious disregard of their rights, health and welfare.
The questions to be determined are:
1. Did the plaintiffs contract PID?
2. If so, was use of the defendants' products causally
connected with the contraction of PID by the plaintiffs?
3. If so, how were the defendants' products causally connected?
4. Did that connection or connections or any conduct of the
defendants or any of them constitute a breach or
breaches of the duty of care owed by the defendants to the
plaintiffs?
5. If there was a breach, did the defendants or any of them act in
knowing and contumelious disregard of the plaintiffs' rights,
health and welfare?
The hearing of these cases commenced on 29 January, 1996 and the evidence concluded on 13 September, 1996. The addresses concluded on 23 December, 1997.The oral evidence comprised substantially cross-examination of expert witnesses on written statements of evidence tendered by them. The documentary evidence comprised approximately 15 lineal metres of folders of exhibits and included numerous photocopies of medical and scientific papers and publications. A great deal of the exhibits are difficult to read.
The quality and complexity of the evidence was such that I clearly informed the parties a number of times during the hearing that the issues which would be determined were:-
1. Those properly raised by the pleadings and in accordance
with the particulars provided, and
2. Those which were the explicit subject of submissions by the
parties.I also informed the parties that the only evidence which would be considered would be that evidence which was explicitly referred to in the submissions.
The submissions of the parties were partly written and partly oral. The written submissions are contained in approximately 3 lineal metres of folders of submissions and supporting documents. The oral submissions were delivered between 27 October, 1997 and 23 December, 1997.
In order to bring some order to the vast quantity of material before the Court I have included the following schedules in this judgment:
Schedule I A chronological list of relevant facts, documents
and publications.
The chronology does not include the facts relevant to each individual plaintiff except for the dates of insertion of the defendants’ devices.
Schedule II Overview of Plaintiffs’ Pleadings - Common
Law Negligence - Gravigard (Cu-7)
prepared by the defendants.
Schedule III Overview of Common Law Pleadings:
- Design/Manufacture
- Clinical Trials
- Marketing
prepared by the defendants.SUMMARY OF PLAINTIFFS' CASE
The two pillars of the plaintiffs’ cases (T.7412, l.46) are based on negligence. They are:
1. Failure to warn
2. Defective design
1. Failure to warn
The plaintiffs’ case on this basis is that
1.1 A woman was placed at greater risk of contracting PID if
she used an IUD including a Cu-7.
1.2 That greater risk was known to the defendants.
1.3 The defendants failed to warn:
1.3.1 the plaintiffs; or
1.3.2 the plaintiffs’ doctors.
The threshold question in determining liability on this count is whether a woman was placed at greater risk of contracting PID if she used a Cu-7. The plaintiffs submitted that the affirmative of this question was established by:
1. Epidemiological evidence.
2. The scientific case as to the role of the IUD and its design
defects in increasing the risk of PID.
3. The causes of increased risk of PID as explained from
clinical expertise.
2. Defective Design
The plaintiffs allege that the Cu-7 had the following defects of which the manufacturer was aware but which it failed to rectify:
2.1 Fraying of the tailstring
2.2 “Memory” of the tailstring which caused it to retract into the
uterus
2.3 The inserter tube had the string on the outside of the tube.2.1 Fraying of the tailstring.
This was said to have materially increased the risk of injury
from ascending infection by:-
(i) fraying creating rough surfaces thereby increasing the area to which bacteria could adhere;
(ii) bacteria prefer rough surfaces;
(iii) frayed surfaces allowed more bacteria to adhere;
(iv) more bacteria could colonise the frayed tailstring;
(v) the frayed surface provided a locus for infectious agents
which can affect the uterus.The plaintiffs’ evidence called to support the case pleaded that the
tailstring had a design defect by virtue of being made of polypropylene and
subjected to tension in manufacture making it prone to fray.It was not disputed that Cu-7 tailstrings frayed and suffered other
morphological (although not chemical) breakdown. This extended to all
Cu-7 tailstrings over time. The disputes as to fraying related to how much
and where.The plaintiffs tendered a great many documents dating from 1974
(the “ dogs breakfast ” memorandum - see document 1 dated 17
September 1974 in the fraying bundle attached to Part 5, ‘Documentary
Case’), including documents from Searle Australia and from the New
South Wales Family Planning Association which show that the fraying was
readily apparent to the naked eye and a source of real concern. Searle’s
documents state that “ Fraying of some degree exists throughout our
production .” (Exhibit P8, Tab 2, page 30).The plaintiffs led evidence to show that Cu-7 tailstrings frayed
throughout their length. Professor Fives-Taylor gave evidence as to
research performed by her group at the University of Vermont which
undertook quantitative scoring of fraying along the length of the tailstrings.
She said they found frays along the length of the tailstrings.Dr Piziali gave evidence for the defendants, that a number of new
and used tailstrings were examined optically or by scanning electron
microscope. These observations were that no or few fibrils appeared on
polyethylene or nylon tailstrings whereas polypropylene tailstrings did
exhibit fibrillation. However the observations reported by Dr Piziali’s team
were of fraying on the vaginal portion only of such strings. In this respect
Dr Piziali’s observations differ from those of Professor Fives-Taylor. They
also differ from the observations made by the Australian National
Biological Standards laboratory which in February 1980 examined a
tailstring and reported to the Commonwealth Department of Health that
“ numerous fraying parts were evident along the entire length of the
filament together with some lengths of split ends protruding from the
filament .” (Document No. 134 in Part 5, ‘Documentary Case’).There is no dispute between the parties as to the fraying being
caused by the way in which the tailstrings were manufactured. Professor
Robertson gave evidence that application of tension (or drawing) to the
Cu-7 tailstring aligned polymer chains along the length of the tailstring.
This had the effect of strengthening the tailstring along its length at the
cost of weakening lateral bonds within the tailstring. The result was that
the tailstring would withstand a greater pull before breaking but became
prone to fray. His evidence applies to the entirety of the length of the
tailstring.It was submitted that the evidence of Professor Fives-Taylor and
Professor Robertson and the letter from the National Biological Standards
Laboratory support the finding that Cu-7 tailstrings, being made of highly
drawn polypropylene, frayed throughout their length.I accept that this was so.
The defendants argued that fraying by itself was solely an aesthetic
problem without medical consequences. The plaintiffs submitted that a
frayed tailstring will have a larger and rougher surface area than a
non-fraying tailstring and will thereby attract bacterial adhesion and
colonisation, increasing the risk of pelvic inflammatory disease.It appears that there is no dispute between the parties as to
Professor Fives-Taylor’s evidence that the body’s primary defence against
infection by micro-organisms is desquamation, that is, the defence of
flushing out micro-organisms by shedding cells on the outer layer of
organs. If a micro-organism has adhered to a surface which does not
desquamate, then it is able to stay in the body. Professor Fives-Taylor
described adhesion as “ the first and most primary virulence factor ” for a
micro-organism in the body.A Cu-7 or Mini, including its tailstring, cannot desquamate. Any
biomedical implant becomes coated with a thin layer of host proteins
immediately upon being placed into the body. An IUD is no exception.
Micro-organisms can attach to this layer of proteins through specific
adhesion. Specific adhesion involves the formation of bonds between the
micro-organism and the surface layer on the tailstring.Professor Ellner gave evidence that the initial attachment is due to
physical and chemical forces but that within a few hours after attachment
micro-organisms which have adhered to an IUD can exude a substance
which forms a “ slime ” layer over the IUD (para 16 of Professor Ellner’s
first statement, Tab 1, Exhibit P282). This, together with material from the
host, forms a biofilm which rapidly covers the whole of the surface of the
device. Professor Ellner said micro-organisms can and do colonise the
surface and interior of this biofilm.The plaintiffs submitted that because of the fibrillation which causes
frayed and rough surfaces, bacteria are better able to adhere to the Cu-7
tailstring than to other tailstrings. This was said to be supported by the evidence of Professor Katz.Professor Katz gave evidence that bacteria are better able to adhere
to a rough surface than a smooth surface and that bacteria are more likely
to colonise along the frayed areas. (Paragraph 17, Statement of Professor
Katz dated 25 August 1995, Tab 1, Exhibit P192). Professor Kaplan
agreed that bacteria might be able to adhere better to rough surfaces than
smooth. (T5370/L57-5371-L3). Professor Robertson and Professor
Fives-Taylor observed that the biofilm preferentially concentrated in the
grooves of the tailstring created by fibrillation, frayed areas and areas of
pits, crevices and roughness. Professor Fives-Taylor undertook
experiments which indicated that up to 10 times as many bacteria adhered
to frayed parts of the Cu-7 tailstring as to smooth parts. The defendants
accept that bacteria can adhere to IUD tailstrings.Professor Fives-Taylor conducted experiments to measure bacterial
adhesion to various tailstrings and polymers. She measured specific
adherence of radioactively “ tagged ” bacteria to unprocessed beads of
polyethylene, polypropylene and co-polymer. While some bacterial species
adhered better to one material and some to another, there was little or no
overall preference shown by various bacterial species for one over another.
Six of ten bacterial species tested adhered better to polyethylene and four
adhered better to polypropylene.She then measured specific adherence of bacteria to tailstrings of
IUDs. The results demonstrated much greater adhesion to the
polypropylene tailstring of the Cu-7 over the tailstrings of other devices.
This held true across bacterial species. Twelve of the thirteen bacterial
species tested adhered better to polypropylene in tailstring form. There
was no result for the thirteenth.Professor Fives-Taylor concluded that it was the processing of the
material, rather than the choice of polypropylene itself, which accounted
for the greater adherence of bacteria to Cu-7 tailstrings over the tailstrings
of other materials.The defendants’ witness, Professor Kaplan, an expert in
microbiology from the University of Texas Medical School, criticised a
number of techniques used in Professor Fives-Taylor’s bacterial adherence
experiments. His evidence was that the use of human saliva rather than
oestrus fluid as a medium for coating the tailstrings deprived the results of
significance. He criticised the application of the method of counting
bacteria and suggested that the experimental technique was sloppy,
pointing to the recording in the laboratory notebooks of a number of
experiments being spoiled by accidents in the laboratory. Professor
Fives-Taylor’s group conducted more than a thousand experiments,
including replicates and gave evidence that some accidents were to be
expected. Professor Kaplan did not point to any scientific
literature which suggested that Professor Fives-Taylor’s experiments
should not be given weight.The plaintiffs submitted that the Cu-7 tailstring has been shown to
have certain characteristics resulting from its manufacturing process. The
fibrillation and morphological breakdown of the tailstring led to it having a
greater surface area and greater surface roughness than tailstrings of the
same diameter made of other materials.The plaintiffs submitted that Professor Fives-Taylor’s conclusion,
that the fraying and degradation of the Cu-7 tailstring promote the
adhesion of bacteria should be accepted because:
· Professor Fives-Taylor and her group undertook many experiments and replicates so even though some are criticised, the overall weight is convincing.
· there are no contrary experiments.
· the only experimental work of the defendants about which there is evidence is supportive in that it indicates that E. coli, a recognised PID pathogen, can adhere to the Cu-7 tailstring.
· Professor Ellner gave evidence to similar effect including citations of the published science which is itself supportive (Statement 31
July 1995, Exhibit P 282, Tab 1) and Exhibits P 205 (Wilkins et al
1989 IJP 57 1-7) and D347 (Wilkins et al Contraception Vol
39, No 2, p. 205), being two of the footnotes to paragraph 16 of
Professor Ellner’s statement.
· the defendants concede that bacteria adhere to tailstrings so there is nothing unlikely in Professor Fives-Taylor’s results.Medical Risks Due To Defects
The defendants argued that, even if there are medical risks due to defects, these nevertheless have no clinical implications.Professor Kaplan agreed that microbial adherence based on biofilm
is a fundamental and pivotal factor in sepsis involving biomaterials.
(T5218/L54-T5219/L5) but was not prepared to extrapolate from this
position, which applied to a wide range of other implanted devices, to the
case of IUDs. He expressed the view that because the epidemiological
data cleared IUDs from involvement in infections, IUDs were necessarily
different from other biomedical devices.Professor Kaplan accepted that cells might detach from the biofilm
of an IUD in situ and, in a host whose defences are lowered sufficiently,
might cause infection. However he suggested that the probability would be
that the large numbers of pathogens constantly accessing the upper genital
tract would cause PID rather than pathogens detaching from the biofilm-
covered IUD. (T5266/L9-41).It was submitted that Professor Kaplan’s opinion that pathogens
detaching from the biofilm were not likely to cause PID was not based on
his expertise in microbiology but rested on his understanding of the
epidemiology as exonerating the Cu-7. He agreed that his opinions were
not informed by any expertise as an epidemiologist (T5192/L25-332)
although he has read the literature and formed opinions. If the
epidemiological prop to his argument is removed, it was submitted that his
concession as to the microbiological point goes a long way to support the
plaintiffs’ case.Professor Ellner expressed a quite different view as to the
importance of pathogens leaving the biofilm. He expressed the view (as
did Professor Katz) that the biofilm protects bacteria from the host immune
system and antibiotics, allowing pathogens to survive in the upper genital
tract, an area of the body where they would ordinarily be eliminated by the
immune system. His evidence was that the biofilm on an IUD can
concentrate nutrients, enabling the bacteria to multiply within the biofilm.
He stated that the biofilm covered Cu-7 forms a “ nidus ” or infective
reservoir from which micro-organisms can emerge to infect the host. His
evidence was that whether the bacterial colonisation of the Cu-7 results in
infection or not depends upon many factors involving the state of host
defences and the numbers and virulence of the micro-organisms. However
the role of the fraying as a design defect was to provide a reservoir of
pathogens able to infect the uterus.The danger of frayed strings was said to be given support by Dr
Stewart who declared that frayed strings were “ medically unacceptable ”
(Exhibit P 486). A number of Searle documents refer to the dangers of
infection from frayed tailstrings. The plaintiffs submitted that the evidence
of Professor Ellner, supported by the plaintiffs’ and defendants’ clinicians,
(who agreed that they would remove the Cu-7 in the event of infection),
should be preferred to that of Professor Kaplan.Colonisation
The plaintiffs alleged in their pleadings that the Cu-7 and Mini
tailstring provided a suitable environment for colonisation by bacteria
and/or other infectious organisms.There is no issue that the normal uterus is subject to periodic
incursions of bacteria which are almost always effectively eliminated by
the immune system. There is also no issue that a uterus which has an IUD
in it will always have permanent colonies of micro-organisms associated
with the biofilm on that IUD. These micro-organisms are protected by the
biofilm from the action of the immune system.It is common ground that it is only upon ascension through the
cervix into the upper genital tract that micro-organisms, including virulent
pathogens such as gonococcus and chlamydia trachomatis, and anaerobes,
can cause PID.An issue between the parties was the extent, if any, to which the
tailstring of the Cu-7 could assist a potential pathogen to “ ascend ”. One
mechanism pleaded in relation to this is the contamination of the device
with bacteria from the vagina or lower cervix upon insertion. There is no
dispute that this occurs. Another mechanism pleaded is the retraction of
the tailstring into the uterus.Leaving these mechanisms aside, it was submitted that there does
not seem to be any dispute that, upon insertion into the body, the Cu-7
rapidly becomes coated by a biofilm which is colonised by bacteria and
that some of these bacteria can be pathogenic in the upper genital tract.
There was no dispute as to the evidence of Professor Ellner and Professor
Fives-Taylor that there are colonies of bacteria in the device biofilm which
have bypassed the cervical defences and which enjoy a measure of
protection from the immune system whenever an IUD is in situ. These
colonies of bacteria are said to be able to exist in the upper genital tract
solely by reason of the presence of the device. In this sense it was
submitted there is no real argument that the presence of the device has
assisted in the establishment of micro-organisms in the upper genital tract.
The question is whether they can do any harm.There is consensus that all IUDs (and other biomedical implants)
rapidly become covered in biofilm upon insertion into the body. There is
debate as to how much of the biofilm is produced by adherent bacteria and
how much is host material, an issue which does not need to be decided and which on the evidence cannot be decided.Professor Ellner gave evidence that biofilm confers a substantial
number of advantages upon bacteria which permit them to survive
permanently in the upper genital tract. Professor Katz supported this
evidence as did Professor Fives-Taylor.Professor Kaplan suggested that biofilm might “ entomb ” bacteria
based upon a reference made by Professor Costerton.It was submitted by the plaintiffs that the views of Professors Ellner,
Katz, Fives-Taylor, Robertson should be preferred to those of Professor
Kaplan. Their views were said to be supported by the published science in
the area, including recent work of Professor Costerton. Professor Kaplan
put forward no published work other than the older work of Professor
Costerton (covered by his disavowal of pre-scanning confocal microscope
work), to support it and Professor Kaplan admitted that his use of terms
such as “ cement ” to describe biofilm and “ entombed ” to describe living
bacteria was “ theatrical ”. (T5394/L1-15).There was disagreement between the parties as to the mechanisms
which may allow bacteria to move into the upper genital tract. Professor
Ellner identified mechanisms by which he said the Cu-7 and Mini assisted
bacterial ascension including:
(a) movement of the perimeter of individual colonies,
(b) “ pollination ” from existing colonies on the string, and
(c) the establishment of new colonies by planktonic bacteria attaching to the tailstring, whether or not they had emerged from other colonies on the tailstring.Professor Robertson identified a number of mechanisms including
the extension of boundaries of a colony by lateral extension driven by
multiplication of organisms causing growth of the colony. He also
identified “ pollination ” in which micro-organisms or entire colonies detach
from the string and reattach at different places.Professor Katz also identified a tendency of motile organisms to
move along the tailstring surface towards a source of nourishment, a
process known as chemotaxis.Professor Coughlin demonstrated that micro-organisms can and do
grow along Cu-7 tailstrings in the laboratory in preference to random
growth. The defendants criticised the laboratory techniques used in the
experiments and indeed the applicability of in vitro experiments at all but
did not call evidence to the effect that the mechanisms identified by the
plaintiffs’ scientific experts did not operate.The only evidence as to experiments carried out by the defendants
was tendered by the plaintiffs. This related to experiments carried out in
1986/7 which suggested that at least one micro-organism associated with
PID, E. coli, was able to attach and to move through mucus in the presence
of the tailstring.It was submitted that the evidence of Professors Ellner, Katz,
Robertson and Coughlin that the Cu-7 and Mini tailstring can facilitate the
transport of micro-organisms into the upper genital tract should be
accepted.Design Criteria
The plaintiffs pleaded that a number of negligent actions were
involved in the design of the tailstring of the Cu-7 and the Mini. An aspect
of this is the case the plaintiffs sought to make that the defendants failed to
apply applicable design criteria in the decision to use highly drawn
polypropylene as the tailstring material.Professor Robertson gave evidence in his reply statement to the
effect that criteria are and were applicable, and were routinely applied, to
medical devices such as IUDs at the relevant times. He referred to criteria
for safety of biomedical implants which were formulated by Hegelyi in an
article published in 1971. (Annexure to reply statement of Professor
Robertson, Exhibit P.136, Tab 2).These criteria included consideration whether properties of the
materials used might exert an effect on the health of the user and whether
the biological environment might exert an effect on the material. Dr
Piziali’s claim said that such criteria only applied to prostheses or
materials used to repair or replace organs. (Paragraph 2 of Professor
Robertson’s reply statement, Exhibit P.136, responding to paragraph 44 of
Dr Piziali’s statement).Professor Robertson’s evidence was that the (primary) principle
applicable to the design of biomedical devices was to eliminate risks or, if
that could not be achieved, then to reduce them as far as possible and warn
of remaining risks. His evidence was that Searle’s polypropylene tailstring
failed to meet these criteria. He said that at the time the Cu-7 was being
designed there was available literature as to the nature of polypropylene
and the migration of bacteria and gave a number of references. He was not
challenged in cross-examination on this material.It was submitted that Professor Robertson’s opinions should be
given greater weight because he had direct experience with the design of
another IUD in 1970/1, the relevant period. He gave evidence as to the
design process for this device. This involved the pharmaceutical
manufacturer, Alza, assembling a team of scientific experts to undertake
the design. This was said to contrast with the action of Searle in handing
device design and development to its in-house Packaging Development
group, which according to an internal Searle document, did not have the
resources in number or expertise to deal with the issues but accepted the
challenge anyway.Dr Piziali gave evidence that an IUD tailstring should be made of a
material which resists biodegradation and is biocompatible with the
environment to reduce interaction with host tissues. (Paragraph 42 of Dr
Piziali’s first statement of 19.10.95, Exhibit D.306). This was not
challenged by the plaintiffs’ experts in biomedical devices, Professor
Robertson and Professor Katz.Dr Piziali suggested that Searle had complied with these
requirements because polypropylene was more resistant to degradation
than other choices, although he appeared to be referring to chemical rather
than morphological degradation. Professor Robertson however said that it
was all kinds of degradation which could pose a risk to the health of the
user which ought be considered. In this regard he was critical of the choice
of drawn polypropylene as the polymer used for the Cu-7 tailstring.
Professor Katz was also critical of the use of a material which degraded
and was unpredictable in its action in the Cu-7/Mini tailstring. It was
submitted that the views of Professor Robertson and Professor Katz should
be preferred to those of Dr Piziali and it should be found that Searle failed
to comply with design criteria applicable and accepted in the early 1970s.2.2 “Memory” of the tailstring which caused it to retract into
the user’s uterus.
This was said to materially increase the risk of injury from
ascending infection by:
(i) string retraction into the uterus introducing more potential
pathogens than otherwise would be introduced;
(ii) string retraction required additional instrumentation of the uterus to remove device the device which instrumentation increased risk of introducing infection;
(iii) the defendants admitted that uterine instrumentation increases the risk of introducing infection.Following the successful insertion of a Cu-7 in the uterus, the
tailstring ran from the bottom of the Cu-7 in the uterus through the
cervix and into the vagina to provide the user or her doctor with an easy
method of checking that the device was still in place and to facilitate the
removal of the device.The Cu-7 is shaped like a figure 7 with a small rounded cap which
fuses the horizontal arm and the curved vertical stem together. Copper
wire is wound around the vertical stem. The polypropylene tailstring is
attached to the lower end of the vertical stem. When the device is
packaged, the Copper 7 stem sits in the inserter tube. The tailstring joins at
the base of the stem (inside the inserter tube), and turns to pass up the tube
beside the device before emerging from the top of the inserter. It then
loops through 180 degrees to run down the outside of the inserter tube.
This can be seen clearly in Exhibit P3, a box of packaged Copper 7s.The plaintiffs alleged in each case that in its packaged configuration
the tailstring was looped out of the top of the inserter tube rather than run
down the inside of the inserter tube, such loop resulting in a string
“ memory ” which caused the tailstring to have a propensity to retract into
the uterus resulting in:
(A) the tailstring transferring bacteria and/or other infectious organisms from the vagina into the uterus;
(B) the need for instrumentation of the uterus; and an increased risk of Pelvic Inflammatory Disease (“ PID ”).The plaintiffs also pleaded, as particulars of the exemplary damages
sought, that Searle was aware that the Copper 7 tailstring had a tendency
to retract into the uterus increasing the risk but chose not to remedy the
defect.The defendants do not dispute that retraction of the Cu-7 tailstring
into the uterus occurs. They say that this retraction phenomenon is
unrelated to the design of the Cu-7, and, in any event, does not
increase the risk of upper genital tract infection.Dr O’Brien, the Associate Director of Clinical Research of Searle
between July 1974 and September 1985, gave evidence for the defendants.
His evidence was that Searle received reports of retractions during the
clinical trial and following marketing of the Cu-7 and that he became
aware of them shortly after they were received. Reports were received
from August 1974 through to 5 December 1983. Dr O’Brien explained
that the efforts Searle made to find a way to eliminate the string retraction
effect were not motivated by any concerns that the retraction of the string
might promote infection by the drawing up of micro-organisms into the
uterus but rather that these efforts were made for marketing purposes. He
went on to say that all IUD tailstrings retract and there was no evidence
that the Cu-7 retracted at a rate greater than did the tailstrings of the
Lippes Loop. Finally he stated that there was no evidence that retraction
promoted infection.The defendants also called Dr Piziali. He is a consultant with a
background in mechanical engineering who has experience in the fields ofbiomechanics and bioengineering. He disclaimed particular expertise in
polymer chemistry. His evidence was that while there was a memory effect
with the Cu-7 tailstring, it was not of sufficient strength or persistence to
account for retraction into the uterus. His evidence included a video of a
Cu-7 inside a plastic uterus. While this video showed movement in the
tailstring, he was of the opinion that the forces involved would not be
sufficient to account for the retractions. He said that although he had given
the matter considerable thought, he did not have a theory as to why
tailstrings retracted but was confident that it could not be the memory.The defendants relied upon his evidence to counterbalance the
exhibits tendered by the plaintiffs many of which are Searle documents
referring to the effect that the special characteristic of the polypropylene
(the “ memory ”) used in the tailstring accounted for the retraction
phenomenon. These exhibits document reports commencing prior to
the period in which all insertions in the lead cases occurred.The defendants’ case was that Searle employees believed at the time
that there was a memory effect inherent in the polypropylene which
accounted for the retractions, but that they were wrong. Searle said that its
employees did not have sufficient expertise to know why the tailstrings
retracted. The plaintiffs refer to statements to a similar effect from
Battelle, expert design engineers employed by Searle to assist in solving
the retraction problem. The documents include statements that the
alternative polyethylene string Searle proposed for use on the Copper
(used on Searle’s Tatum T) was a definite improvement in that it had
almost no memory. It was submitted that this suggested that Searle
employees were right to ascribe the memory problem to the particular
characteristics of the highly drawn polypropylene tailstring.Doctors using the Cu-7 reported similar experiences to Searle. For
example, Dr Michaelow, described in an internal Searle memo as a
“ Gravigard devotee ” reported that up to 80% of Cu-7 tailstrings retracted
in his experience of 2,000 insertions and that the phenomenon was
definitely not occurring with other devices. Mr Edmunds of Searle South
Africa asked “ Why does this not happen with our Tatum-T thread or
threads of other devices ?”The plaintiffs have relied on the Searle documents and the evidence
of Professor Channing Robertson and Professor David Katz to establish
that the polypropylene used by Searle in the tailstring had a propensity to
retract. They submitted that unless these documents are ignored or proved
to be all wrong on the question, the plaintiffs are entitled to a finding in
their favour.Professor Robertson also gave evidence for the plaintiffs. He is a
Professor of Chemical Engineering at Stanford University. He is the
witness in the case with the greatest expertise in polymer chemistry.
Polypropylene is a polymer. It was common ground that the memory effect
was a result of the stresses within the polypropylene tailstring deriving
from the chemical conformation of polypropylene processed by drawing.
Searle designed the packaging so that the tailstring looped above the 7 and
then ran down along the outside of the inserter tube. Professor Robertson
gave evidence as to the effect of the packaging of the tailstring. His
evidence was that by using polypropylene and packaging the Cu-7
with a loop in the tailstring, the result was that the tailstring was prone to
resume the shape it had within the packaging. The effect of this was that
the tailstring after insertion of the device would have a tendency to
withdraw into the uterus, playing a role in the introduction of infection
(para 40, Statement dated 28 July 1995, Tab A, Exhibit P 136).Professor David Katz is a Professor in both the Department of
Biomedical Engineering and in the Department of Obstetrics and
Gynaecology at Duke University, North Carolina. He had previously held
a position as a Professor in the Department of Chemical Engineering at the
University of California, Davis. He has expertise in polymer chemistry.
His evidence was that one of polypropylene’s characteristics was to have a
memory, in the sense of a tendency to resume a position in which the
material had previously been placed, and that this characteristic cannot be
removed and cannot be predicted with certainty.It was submitted that the evidence of Professors Robertson and Katz
should be accepted and a finding made that the memory characteristic of
the polypropylene used in the Cu-7 tailstring caused the tailstring to have a
propensity to retract and that this characteristic existed to a markedly
greater degree in Cu-7 tailstrings than in the polyethylene alternative which
Searle used on its Tatum-T and proposed to use on the Cu-7 and Mini.The second dispute about retraction is whether it increased the risk
of PID.Dr O’Brien gave evidence that there was no study which concluded
that string retraction played any role in the development of PID. His
evidence was that the efforts made to resolve the tailstring retraction
problem were made for marketing reasons (i.e. presumably that doctors
were reluctant to insert a device that might cause problems) rather than out
of concerns about safety of users.Denzin, Gentle, Lee and Ottaway suffered tailstring retractions.
It is common ground between the parties that any instrumentation of the
uterus increases the risk of contracting, or exacerbating, PID. Each of
Denzin, Gentle, Lee and Ottaway required instrumentation of the
uterus to remove her Cu-7 because the tailstring had retracted, although in
Gentle’s case this was associated with treatment of septic abortion. It was
submitted that there is no real dispute that the tailstring retractions
increased the risk of injury to the plaintiffs by requiring instrumentation.Professor Ellner gave evidence that it was obvious that retraction of
the tailstring was a mechanism for contaminating the uterus with bacteria,
causing infection. Professors Eschenbach, Robertson and Katz gave
evidence to the same effect. In addition, the defendants’ expert medical
witness, Professor Fraser, confirmed that the retraction of the tailstring
might introduce bacteria into the uterus and it would depend upon the type
of bacteria and the health of the woman as to whether this would create an
infectious process.The plaintiffs submitted that:
(a) the evidence of Professors Ellner, Robertson, Katz and Fraser as to the direct contamination of the uterus by string retraction combined with
(b) the Searle documents,
(c) the consensus as to the risk of instrumentation (necessitated by the retraction in at least three of the lead cases) and
(d) the concerns as to the potential of the frayed tailstring to pose a risk of infection independently of fraying to support a finding that the retraction of the tailstring provided a means whereby bacteria could be carried from the vagina into the uterus, thereby increasing the risk of PID for users of the Cu-7.2.3 The design of the inserter of the Cu-7 had the string on the
outside of the inserter tube
As a result of this, the string trailed through the bacteria laden
parts of the lower genital tract and trans-cervically introduced
potential pathogens into the uterus.In the case of each of the nine plaintiffs it was submitted that
two or more of the alleged defects increased the risk of injury:
1. Denzin: memory plus fraying plus inserter tube
2. Gentle: memory plus fraying (septic abortion
case)
3. Lane: fraying plus inserter tube
4. Lee: memory plus fraying plus inserter tube
5. Moylan: fraying plus inserter tube
6. Orders: fraying plus inserter tube
7. Ottaway: memory plus fraying plus inserter
8. Robertson: fraying plus inserter tube
9. Southren: fraying plus inserter tubeThe plaintiffs submitted that the general matters which prove liability on the part of the defendants are:
(a) the use of the Cu-7 increased the risk of PID and its sequelae;
(b) the defendants were aware of the epidemiological data which
proved or provided compelling evidence that use of IUDs in general
and of the Cu-7 in particular caused an increased risk;
(c) the defendants knew or ought to have known that the design defects
caused or were likely to cause an increased risk;
(d) the defendants, notwithstanding their knowledge as set out in (b)
and (c) above, negligently failed (i) to warn of the increased risk;
and (ii) to remedy the defects;
(e) the plaintiffs suffered injury, namely PID and its sequelae, and for
Gentle a septic abortion, as a result of the defects and/or the
increased risks;
(f) the plaintiffs would not have suffered the injury had each been
warned of the increased risks;
(g) the plaintiffs were not warned of the increased risks.It was submitted that each of the three alleged design defects dealt with above was a material contribution to the risk of injury and not 'de minimis' because:
1. (i) the defendants' own behaviour revealed its recognition that the problem (ie the added risk) was not de minimis.
(ii) contemporary medical opinion: eg Dr Hatcher, Dr Tyrer of the Planned Parenthood Federation of America, the Food and Drug Administration, the Australian National Biological Standards Laboratory was of that view.
(iii) epidemiological data showed that users of Cu-7 were at an increased risk of PID and its sequelae.
(iv) the foreign body effect potentiates and prolongs the risk of infection from additional bacteria introduced because of design defects.2. No alternative causes were proved by defendants.
The plaintiffs submitted that each of the three defects were negligently incorporated in the design of the Cu7 by the defendants and that independently and jointly they made a material contribution to the risk of injury to each plaintiff.
The plaintiffs submitted that each of the plaintiffs suffered tubal
damage which is said to be explained by a prior episode of PID.
Therefore it was said the task for the medical experts was to identify
when the episode was most likely to have occurred, rather than whether it
occurred.In the diagnosis of PID, the parties did not disagree as to the fact that
the disease could present with severe symptoms ie classic PID, or with
less obvious and fewer symptoms ie atypical or unrecognised PID.
However on behalf of the plaintiffs Dr Sweet gave evidence that a doctor
could identify the signs and symptoms of PID in a woman who had
suffered an unrecognised attack if a careful history was subsequently
obtained from her.In the case of each plaintiff, it is impossible to identify the pathogen
which caused the PID suffered by the plaintiffs.The defendants attempted in all cases except Mrs Denzin to prove
that it must have been a sexually transmitted disease (STD), not an IUD,
which caused any episode of PID suffered by the plaintiffs.However the plaintiffs submitted that because the IUD plays an
independent and additive role in the presence of infection it is irrelevant
to enquire whether the pathogen causing the PID was an STD or non-STD
micro organism.They also submitted that the attempts by the defendants to establish
the occurrence of STD failed because:
(i) no identification of an STD pathogen causative of PID was ever made;
(ii) the only STD micro-organisms ever identified were "innocent" in the sense that they were not causative of PID eg genital warts or trichomonads;
(iii) it was never argued that if a woman had evidence of an "innocent" STD micro-organism she must have also had a "pathogenic" STD capable of causing PID viz. Neisseria gonorrhoea or Chlamydia trachomatis;
(iv) the defendants' experts never committed themselves to an opinion that a plaintiff's history and records prior to using a Cu-7 proved an attack of PID;
(v) any signs or symptoms of PID identified in association with the hypothetical STD were much less persuasive of PID than the signs and symptoms in association with the use of the Cu-7.
The defendants also sought to implicate other occasions in the
histories of the individual plaintiffs as establishing an alternative to the
IUD as the cause of PID. These included appendectomies, termination of
pregnancy, hysterosalpingogram and post-partum infection.The plaintiffs' experts were of the view that in each case the PID
occurred during IUD use and not at any other time alleged by the
defendants’ experts.The plaintiffs' case also included the proposition that the IUDs
The plaintiffs submitted that the opinions of the plaintiffs' medical experts as to the occurrence of PID during Cu-7 use (Gravigard and Mini-Gravigard) and the role of the IUD in causing the PID should be preferred to those of the defendants’ medical experts who were, it was submitted, not able to offer any convincing alternative explanation for the PID suffered by the plaintiffs. It was also submitted that the expert opinions of the plaintiffs' medical witnesses as to the causal role of the Cu-7 in PID were consistent with and supported by the epidemiological data, the medical principles relating to infections and the foreign body effect, and their own clinical experience.
materially contributed to the PID by reason of their effect on the body's
immune system.
2…PELVIC INFLAMMATORY DISEASE (PID)
1. The Female Reproductive SystemThe female reproductive system above the vagina consists of:
1.1 The Uterus
The uterus is a hollow muscular organ in which the fertilised ovum normally becomes embedded and in which the developing embryo and foetus is nourished.It has a mucous membrane lining known as the endometrium.
Its cavity opens into the vagina below and above into two Fallopian Tubes, one on either side of the uterus.
At the lower portion of the uterus there is a constricted portion known as the isthmus leading to the cervical canal which is the narrow lower end of the uterus between the isthmus and the opening of the cervix into the vagina. The opening of the cervix into the vagina is known as the cervical os.
1.2 Fallopian Tubes
There are two fallopian tubes, one on each side of the uterus. They extend laterally from the uterus and terminate in the peritoneal cavity near each ovary. They serve to convey the ovum from the ovary to the uterus and spermatozoa from the uterus toward the ovary. In the event that an ovum is fertilised by spermatozoa the fertilized ovum is conveyed, in normal cases, to the uterus where it is embedded in the endometrium.1.3 Ovaries
There are two Ovaries, one on each side of the body near the ends of the fallopian tubes.The ovaries are glands which produce the reproductive cells, the ovum, which pass into the fallopian tubes and then to the uterus.
2. PID - Definition
The term “ pelvic inflammatory disease ” is used to describe infections of various parts of the upper reproductive organs covering:
1. endometritis - an infection to the endometrium (the lining of the
uterus). In the course of his evidence Eschenbach said that most patients with pelvic inflammatory disease have endometritis, because the infection " starts in the lower genital tract and the lower parts of the uterus and moves up the uterus and into the fallopian tubes and out. So most individuals who have pelvic inflammatory disease would also have endometritis. " (T1664/L25-L30).2. salpingitis - an infection of the fallopian tubes;
3. oophoritis - an infection of the surface of the ovaries;
4. pelvic peritonitis - in diffuse cases all of the above may be infected
together with the pelvic peritoneum.PID may be:
1. Acute
2. Chronic
3. SilentOn gynaecological examination tubal scarring or blockage is found and this leads to a retrospective diagnosis of silent PID.
3. The cause of PID
The cause of the infections known as PID is usually pathogens which ascend from the vagina to the cervix and then through the cervix into the endometrial cavity where they break the lining of the uterus and group in sufficient numbers to produce an infection. That infection can then move into the fallopiana tubes and to the ovaries, and possibly the abdominal cavity (Eschenbach, Ex. 108 Tab 1, para 12).Eschenbach and Newton’s view was that there are two major groups of organisms which, when found in the uterine cavity have been found to lead to PID:
(a) exogenous organisms including STD organisms, in particular
gonorrhoea and Chlamydia trachomatis; and
(b) endogenous bacteria, both aerobic and anaerobic bacteria which
are commonly isolated from the vagina and /or vulva and
perineum of healthy women.Gonorrhoea and Chlamydia are most commonly, although not exclusively, transmitted to the women's body through sexual contact
(Ex. 108 Tab 1, para 14).The presence of gonorrhoea and chlamydial infections in the vagina or cervix may cause localised infections but they will only have serious complications related to a woman's fertility if they are introduced into the upper reproductive tract (Ex.108 Tab 1, para 17). They do not automatically ascend to the uterus and are often prevented by the barrier of the cervix and the body's defence mechanisms.
Sweet said that in the absence of a factor such as an IUD, cervical gonorrhoea or chlamydia will cause PID in approximately 10% to 20% of cases (T3070/L38 para 24 as corrected in memorandum of 28 April 1996. Tab 1). Similarly, the vaginal bacteria do not usually ascend into the uterus unless aided in some way (para 25).
Sweet said that approximately 65% of women with PID have chlamydia and/or gonorrhoea, and of these about half also displayed evidence of infection with aerobic and/or anaerobic bacteria (Par.26). Approximately 30% of women with PID have only aerobic and/or anaerobic bacteria present. (para 26).
Polymicrobial bacteria exist naturally within the vagina, most commonly aerobic and anaerobic bacteria including streptococci, lactobacillus, e. coli, gardnerella, vaginalis, peptostrepococci and acteroides (Ex. 108 Tab 1, para 15). These bacteria are harmless when they form part of the vaginal flora, but if they are introduced into the upper genital tract they can be pathogenic (i.e. they can cause disease)
(Ex. 108 Tab 1, para 16).By the 1970s it had become evident, following the widespread development of anaerobic bacteriology, that multiple organisms were often isolated from infected sites, particularly from intra-abdominal abscesses. Polymicrobial infections, consisting of both aerobic and anaerobic bacteria, were commonly identified among women with pelvic inflammatory disease and vaginitis (Ex. P108 Tab 1, para 142).
In 1984 Dr Eschenbach presented a paper “Acute pelvic inflammatory disease” at a symposium on sexually transmitted diseases in which he said: (Ex. P108. Tab.12).
“Primary pelvic inflammatory disease is usually an acute infection
in which organisms ascend into the uterus and fallopian tubes from the cervix. Chronic active infections are unusual except in neglected cases and in actonomyces infection. However, sterile chronic inflammatory adhesions are common residuals of an acute infection.” (Tab 12, pp 65-66).As well as sexually transmitted organisms, Dr Eschenbach said that non-sexually transmitted bacteria can also cause pelvic inflammatory disease:
“Sexually transmitted organisms, most noticeably Neisseria
gonorrhea, Chlamydia trachomatis, and genital mycoplasmas, are recovered from the majority of women who develop pelvic inflammatory disease. However, non-sexually transmitted aerobic and anaerobic endogenous cervicovaginal bacteria also cause pelvic inflammatory disease in a significant number of incidences. The rate with which these micro-organisms have been isolated with pelvic inflammatory disease differs widely in various studies, probably because of the differences between the various populations studies and differences in infection.” (Tab 12 p 67 right hand column, paragraph headed ‘Microbial Etiology’).The complexity of the infectious position is such that it is possible
to:
1. have a sexually transmitted disease in the vagina
e.g. gonorrhoea, and not have PID;
2. have a sexually transmitted disease eg gonorrhea, and have
PID;
3. not have a sexually transmitted disease and not have PID;
4. not have a sexually transmitted disease and have PID.By 1982, when Eschenbach published " New concepts of obstetric and gynaecological infection " (Ex.P108, Tab 11) research had established that polymicrobial infections of aerobic and anaerobic bacteria had been detected in 20% to 60% of women with PID in the United States (he cited Eschenbach 1975; Sweet et al. 1979 and Monif et al. 1976). Eschenbach concluded:
" Although it had become clear that PID is sexually transmitted including by N. gonorrhoea and by C. trachomatis, it is also non-sexually transmitted by polymicrobial infection involving aerobic and anaerobic bacteria " (Tab 1, Par.145).Eschenbach said that N. gonorrhoea and C. trachomatis could be found in anywhere between 50% and 60% of PID cases (T166/L7). However, the actual proportion is very "population dependant" and may be as low as 30% (T1668/L29). In Eschenbach's opinion, he would include in the top category of pathogens causing PID, gonorrhoea, Chlamydia, and commensals such as E. coli and anaerobic bacteria (T1683/L9).
Fraser, in para 79 of his first witness statement said that:
“Using careful investigations, specific sexually transmissible
infection is identifiable in 60% to 80% of cases.”The body’s defence mechanism
Dr Eschenbach said in his first witness statement that the inoculum (i.e. the number of bacteria required to cause infection to a particular organ) varies and depends upon the woman’s immunity and other factors (Tab.1 Par. 18;), (T1690/L10-L35).Among the other factors to which he referred is the effect of a foreign body (in this case an intrauterine device) inserted in the uterus. Dr Eschenbach said (second witness statement, Tab 17. Par. 2) that the immune system within the uterus and within other organs is similar for non-pregnant women. The effect of inserting a foreign body in the uterus will be similar to inserting a foreign body elsewhere. “The hypothesis of the uterus as an immunologically-privileged organ only exists during pregnancy, and then, only for a small portion of the intra-uterine contents, the foetal - decidual interface” (Tab.17. Par. 2). It was said that the presence of the IUD modifies the body’s normal defence mechanisms, leading in some instances to pelvic infection. Eschenbach reported in March 1976 (Ex. P108. Tab. 5) “Acute pelvic inflammatory disease: aetiology, risk factors and pathogenesis” (NEJM pp.147-169) that the intrauterine device was a risk factor for pelvic inflammatory disease, a major public health problem and that prospective studies of the magnitude of the problem were long overdue (Tab. 5. p.166 at 2). Dr Eschenbach said in that article:
“The role of the IUD in the pathogenesis of PID may be similar
to that of a foreign body in the presence of bacteria elsewhere in human tissue: it allows colonisation of bacteria on its surface, reducing the effectiveness of normal defence mechanisms. The concentration of anti-body, phagocytes, and other components of the defence mechanism surrounding an IUD is unknown but possibly is inadequate to prevent bacterial colonisation from resulting in infection. Perhaps well-vascularised tissues, without the presence of a foreign body, are more likely to possess defences capable of preventing infection.” (Tab. 1 Par.66).It was said that the presence of the intrauterine device, being a foreign body also lowers the inoculum or the amount of bacteria that may be needed to cause an infection. (Tab. 1 Par.21).
Dr Eschenbach described a number of ways in which the presence of a Copper 7 adversely affects a woman’s defence mechanisms:
1. the string of the Copper 7 acts as a bridge for bacteria which
attach to the string and produce a biofilm which protects them from the body’s defence mechanism; the biofilm prevents the white cells from destroying the bacteria; and the bacteria can also transfer along the string from the vagina into the uterus. (Tab 1. Par. 22(i)). Elstein had shown in 1967 greater evidence of signs and symptoms suggestive of PID in women using an IUD with a tail. (Ex. D113, first Par.). Dr Eschenbach said with regard to bacterial motility, that bacteria had a capability of using the contiguous spread mechanism as well as the walls of the uterus to ascend to the tubes. (T1997/L5.).2. the arm of the Copper 7 or its projection can embed in the
lining of the uterus giving bacteria access to the basal layers of tissue, creating a more resistant and difficult infection to treat. (Tab 1. Par. 22(ii)).3. the Copper 7 by increasing the amount of bleeding
experienced by the women during and or between menstruation provides nutrients for bacteria to grow within the uterus (Tab. 1. Par. 22(iii)).4. the Copper 7 affects the uterine environment by lowering the
oxygen and permitting preferential growth of anaerobic bacteria, thus increasing the likelihood of uterine infection (Tab. 1. Par. 22(iv); T1907/L45-L55; see also T1984/L5-L20).5. the device affects the natural barriers in the woman’s
reproductive system designed to protect the uterus from infection namely the cervix, cervical mucous and epithelial lining; because of the string, the body has no barrier to infection being transmitted between the uterus and the fallopian tubes (Tab 1, Par.23).6. the retraction of the string associated with the Copper 7
provides another mechanism by which pathogens within the vagina can be transferred to the uterus and cause infection (Tab. 1. Par.24).It was submitted by the plaintiffs that HSG studies reported by Elstein in 1969 (Ex.D113) established an increased incidence of tubal damage and occlusion in women with signs and symptoms of PID which occurred in association with the use of an IUD with a tail (p.279, second last Par.).
Effects of PID
PID can cause tubal damage leading to a greater risk of impaired fertility, infertility and ectopic pregnancy.PID often takes the form of salpingitis which causes damage to the fallopian tubes in a number of ways:
1. Inflammation or obstruction of the fallopian tube may prevent the egg and the sperm from meeting.
2. The cilia or finger-like projections within the fallopian tube and the secretory cells within the tubes may be damaged, such that the ability of the tubes to keep the egg alive and move the egg and sperm together is destroyed.
3. Adhesions between the tubes and ovaries may be created such that the fallopian tubes cannot move over the ovary during the menstrual cycle to pick up the egg and move it along.
Eight per cent of patients who experience infections to the fallopian tube are rendered infertile. 85%-90% of women who develop tubo-ovarian abscesses are rendered infertile (Dr Eschenbach, Ex P108, Tab 1, Paras 27 & 28).
There is also the possibility of suffering tubo-ovarian abscesses.
Dr Eschenbach said that the most serious form of pelvic inflammatory disease is tubo-ovarian abscesses which are very difficult to treat because the body “walls off the abscess through production of collagen walls, particularly where anaerobic bacteria have been the cause of the abscesses. Given that the collagen walls have no blood vessels within, it is difficult to get the antibiotics to the abscess.” (Tab.1 Par.33). Tubo-ovarian abscesses are typically polymicrobial: that is “they are normally produced by infections relating to the bacteria found within the vagina, most normally bacteroides and peptostreptococci. As such, they are not normally associated with STD PID alone, but instead are more commonly associated with the ascension of vaginal bacteria through external means such as through the introduction of an IUD .” (Tab.1 Par.33, 34).
Ectopic pregnancies are a further risk. Infections in the fallopian tubes create holes or craters in the tubal area which provide traps for the fertilised egg, or the egg alone, so that the risk of ectopic pregnancies is increased. The seriousness of the health risk from ectopic pregnancy is not in dispute. As Dr Eschenbach said: “The real danger is the rupturing of the ectopic pregnancy which causes great haemorrhaging and can lead to a woman’s death.”
(Ex. P108. Tab 1, paras 29 & 30).Dr Keeping also gave evidence to the effect that PID created an increased rate of ectopic pregnancy (T 6387/L51 - T6388/L12) as did Professor Newton (Ex. D299, para 87).
Chronic pain is experienced by approximately 20% of women after contracting pelvic inflammatory disease.
Dr Eschenbach summarised the consequences of PID (chronic pain, subsequent episodes of PID, infertility, and ectopic pregnancy) (Ex. P108 Tab 11, p2041) as follows:
Frequency %
Consequences
15
Chronic pain
20-25
Subsequent PID
Infertility*
1 Infection
2 Infections
3 Infections
10-15
25-35
50-757
Ectopic pregnancy
(with first subsequent pregnancy)
* 8% of patients who experience infections to the fallopian tubes are rendered infertile
Infertile
Ectopic Pregnancy
Intrauterine Pregnancy
Number
Category
Total Women
Number
Rate
Number
Rate
Rate433
Control
448
8
2%
6
1%
97%
PID Patients
852
1st episode
1,025
112
11%
61
6%
83%124
2nd episode
198
50
25%
24
12%
63%
> 3rd episode
69
30
43%
15
22%
24
35%Diagnosis of PID
PID may be categorised as:
1. Acute PID
This presents with classic symptoms which are usually quite severe.
Pelvic inflammatory disease may manifest itself by "classic symptoms" evidenced by inflammation or infection which is accompanied by abdominal tenderness, cervical motion tenderness and adnexal tenderness, but it may also occur without these symptoms. The classic full-blown syndrome of PID includes fever, elevated white count, cervical motion tenderness, adnexal tenderness.2. Atypical PID
This may be:
2.1 subacute; or
2.2 chronic
Each of these subcategories presents with lesser signs and symptoms.
They are usually characterised by inter-menstrual bleeding and pain. It is often termed ' asymptomatic ' though it is probably more accurately a case of ' unrecognised ' pelvic inflammatory disease.Atypical PID has also been called “silent” PID. This suggests that
there are no recognisable symptoms, but Dr Sweet gave evidence to the effect that a survey of histories of patients with “silent” PID would reveal episodes from which a retrospective identification of the presence of PID can be made (T 3055/L33-35, Ex. P247, p3163).The damage which might be suffered by a woman who suffers from
atypical PID may not be any less than that suffered by a woman with acute PID, including severe tubal damage (Dr Keeping, T6153/L10; Dr Sweet,
T 3054-7).Dr Sweet said that a person can have acute pelvic inflammatory disease that is unrecognised and "we are beginning to understand that probably the majority of pelvic inflammatory disease is actually of the unrecognised or atypical which physicians have not been attuned to addressing. It is very similar to the point I tried to make to the FDA back in 1977, that we needed to be aware that these patients with abnormal bleeding, with pain who did not have fever did indeed have pelvic inflammatory disease."
(T3054/L49-T3055/L2).PID may be acute, involving infection of the fallopian tubes, but not manifested by significant severe pain or fever, presenting as Dr Sweet said a more subtle picture (T3055/L15-L20).
In the case of atypical PID, a woman may not even elicit tenderness on bimanual examination:
"That can happen, we now recognise that women may have
endometritis and have low grade infection, may have some
bleeding, spotting, pain, but may not actually have the
tenderness." (T3066/L10).In the case of unrecognised PID, Dr Sweet said that the doctor typically sees things such as "abnormal bleeding, complaints of abdominal or pelvic discomfort, you may get discharge, maybe discomfort with sexual intercourse, those types of things" (T3066/L35).
The degree of bleeding may be heavy which the patient would notice, or simply something that the doctor might notice at the time of a speculum examination (T3067/L24) or it may simply be spotting which may not occur on the days the patient sees the doctor (T3067/L29). The pelvic discomfort may be cramps or aches, commonly symptoms reported by menstruating women (T3068/L39), but the patients usually know the pattern of their own menstruation established over time and are aware of differences (T3069/L5).
In the case of vaginal discharge, particularly with bacterial vaginosis, 40% of those women "have histologic and microbiologic evidence of endometritis" with bacterial vaginosis being present (T3069/L30). The presence of dyspareunia is something which the patient may describe. (T3070/L5).
Dr Sweet said that the studies by Wolner-Hanssen showed that "....If you went back and took a careful history from the women, asked questions about pain, discomfort, that the overwhelming majority of women had had those types of symptoms that had not been diagnosed as PID because physicians were relying on the classis criterion. When you questioned the patients closely they could have symptoms, that is why its has moved to be unrecognised or atypical" (T3055/L40-L45)
Dr Sweet said that the mild clinical PID can have the same outcome for the patient as full-blown PID (T3057/L55).
A question to be determined in each of the nine cases is whether the plaintiffs suffered PID in association with Cu-7 use. The plaintiffs submitted that in eight of the cases (Denzin, Lane, Lee, Moylan, Orders, Ottaway, Robertson and Southren) the existence of PID has been confirmed by objective tests so that the task of the expert witnesses was to identify whether or not there were signs or symptoms of PID in association with Cu-7 use.
They submit that the presence of tubal damage in the absence of any alternative cause supports the finding that it was caused by PID.
3. CAUSATIONThe evidence established that women who have not used the
Cu-7 device both do and do not contract PID and that women who do use the Cu-7 device both do and do not contract PID.Thus if a woman uses a Cu-7 and contracts PID the question to be determined is whether the Cu-7 was causative of the PID in the particular case.
The plaintiffs stated the legal principle upon which their case is based as follows (in the Summary of Plaintiffs’ Case B.1.):
“The defendant is liable for negligent defects which result in injury where those defects materially increase the risk of injury, i.e. other than de minimis, unless the defendant proves that the injury was in fact caused by some other cause unrelated to its negligent act.”This statement appears to me to beg the question and should be stated properly to reflect their position as follows:
“The defendant is liable for injury to the plaintiffs where there are negligent defects which materially increase the risk of injury, i.e. other than de minimus, unless the defendant proves that the injury was in fact caused by some other cause unrelated to its negligent act.”Their proposition was stated orally by their counsel at
T.7435-6:
“If you use a Copper 7, you are more likely to get PID than if you don’t, and that is our client’s case. That is to say, applying the reasoning, for example, in Bonnington Castings and in McLean if there is negligence on the part of the defendant that places the woman at an increased risk of contracting PID, then if the injury is suffered, that will be sufficient to establish legal causation.……it is always open to the defendant to attempt to prove that irrespective of the increased risk of using the IUD, the injury was suffered because of some other entirely different event. For instance, if there was a case of a ruptured appendix as a young woman and if it was shown of course that damage was suffered and caused by a ruptured appendix….the plaintiff cannot succeed because the injury would have been suffered irrespective of the IUD…
If the defendant wants to prove an alternative cause, then it must do so according to the laws of proof. It must satisfy the Court that there is another cause….”
The first proposition that if you use a Copper 7 “ you are more likely to get PID than if you don’t ” is a factual question to which I shall come later.
The next proposition is that if the first proposition is established and the woman has suffered PID, that in itself (subject to the third proposition) is enough to establish causation in the legal sense.
The third proposition was that it is open to the defendant to attempt to prove that irrespective of the increased risk of using the IUD, the injury was suffered because of some other entirely different event.
In Bonnington Castings Ltd v Wardlaw (1956) AC 613 an employee was exposed to silica dust by his employer from two sources. One source of exposure did not render the employer legally liable but the other source did.
Lord Reid said (p.622):
“I think that the position can be shortly stated in this way. It may be that, of the noxious dust in the general atmosphere of the shop, more came from the pneumatic hammers than from the swing grinders, but I think it is sufficiently proved that the dust from the grinders made a substantial contribution. The respondent, however, did not only inhale the general atmosphere of the shop; when he was working his hammer his face was directly over it and it must often have happened that dust from his hammer substantially increased the concentration of noxious dust in the air which he inhaled. It is therefore probable that much the greater proportion of the noxious dust which he inhaled over the whole period came from the hammers. But, on the other hand, some certainly came from the swing grinders, and I cannot avoid the conclusion that the proportion which came from the swing grinders was not negligible. He was inhaling the general atmosphere all the time, and there is no evidence to show that his hammer gave off noxious dust so frequently or that the concentration of noxious dust above it when it was producing dust was so much greater than the concentration in the general atmosphere, that that special concentration of dust could be said to be substantially the sole cause of his disease.”At p.623 his Lordship said:
“No doubt the total amount from both sources in the atmosphere was small at any one time, but the combined effect over a period of eight years was to cause the respondent’s disease. The importance of evidence of the fellow-workmen is that it shows that the visible dust and therefore also the invisible dust from the swing grinders was not immediately dispersed, and therefore that the respondent was bound to inhale some of the invisible noxious dust from the swing grinders. On this matter Lord Carmont said: “Even if the majority of the pursuer’s inhalations took place near the source where the silica dust was produced, i.e., at his hammer, a minority of inhalations from the general atmosphere of the shop needlessly contaminated owing to the breakdown of the extracting hood, duct and fan at the swing grinders may well have contributed a quota of silica dust to the pursuer’s lungs and so helped to produce the disease.” On his view of the onus of proof Lord Carmont did not require to go farther than that. In my opinion, it is proved not only that the swing grinders may well have contributed but that they did in fact contribute a quota of silica dust which was not negligible to the pursuer’s lungs and therefore did help to produce the disease. That is sufficient to establish liability against the appellants, and I am therefore of opinion that this appeal should be dismissed.”This decision turned on a factual finding that the exposure of the employee to silica dust negligently by his employer did in fact materially contribute to the onset of the disease from which he suffered.
The plaintiffs accepted that the onus of proving that the defendant’s negligent conduct caused the plaintiffs’ injury remains on them but submitted that the onus is discharged by proving exposure to increased risk that could materially contribute to the injury.
They submitted that establishing increased exposure to risk caused by the negligence of the defendant prima facie establishes causation (T.7438).
The determination of causation is governed by March v E & MH Stramare PtyLtd (1991) 171 CLR 506. It is essentially a question of fact to be answered by reference to commonsense and experience and one into which considerations of policy and value judgments necessarily enter.
The plaintiffs referred to authorities cited by Mason CJ at page 514 of Stramare in support of the statement:
“Nonetheless, the law’s recognition that concurrent or successive tortious acts may each amount to a cause of the injuries sustained by a plaintiff is reflected in the proposition that it is for the plaintiff to establish his or her injuries are ‘ caused or materially contributed to ’ by the defendant’s wrongful conduct.”Those authorities were:
1. Bonnington Castings Ltd v Wardlaw (1956) AC 613.
2. McGhee v National Coal Board (1973) 1 WLR1.
In McGhee , Lord Reid referred to Bonnington Castings . He said at p.4D:
“It has always been the law that a pursuer succeeds if he can show that fault of the defendant caused or materially contributed to his injury. There may have been two separate causes but it is enough if one of the causes arose from fault of the defendant. The pursuer does not have to prove that this cause would of itself have been enough to cause his injury. That is well illustrated by decision of this House in Bonnington Castings ….”The medical evidence in that case could not explain why it was that the employee’s cycling home caked in grime and sweat had added to the risk that the disease might develop. Lord Reid said:
“It does not and could not explain why this is so. But experience shows that it is so. Plainly that must be because what happens while the man remains unwashed can have a causative effect, though just how the cause operates is uncertain…. From a broad and practical viewpoint I can see no substantial difference between saying what the defendant did materially increased the risk of injury to the pursuer and saying that what the defendant did made a material contribution to his injury.”3. Duyvelsheff v Cathcart & Ritchie Ltd (1973) 47 ALJR 410. At 417 Gibbs J said:
“The plaintiff had to prove that the defendant’s breach of duty caused or materially contributed to the injury. It was sufficient where there was a breach and an injury of the kind that might thereby be caused to justify the drawing of an inference, in the absence of any sufficient reason to the contrary, that in fact the injury did occur because of the breach.”4. Tubemakers of Australia Ltd v Fernandez (1976) 50 ALJR 720.
Mason J said (p.724E) that where the issue of causation is not within the realm of common knowledge and experience and “ fails to be determined by reference to expert medical evidence ” , if the expert evidence provided a sufficient basis for the inference that on the probabilities a particular act or event caused or materially contributed to the occurrence of the medical condition, then liability is established.It was submitted that the following conclusions arise from these authorities:
1. Where medical evidence does not explain a causative effect,
nevertheless such an effect will be found where clinical experience or observation confirms the causal link.
2. Where the injury may be caused by more than one factor, if the defendant’s conduct materially contributed to the injury in the sense that it is not de minimis, then the defendant will be liable.
3. Conduct of the defendant which materially increased the risk of injury will be found to have materially contributed to the injury.The plaintiffs also relied on Commonwealth of Australia v McLean (1997) 41 NSWLR 389.
The plaintiff in that case suffered post-traumatic stress disorder and, as a result, over the next thirty years and more he drank alcohol and smoked tobacco, both to excess, which caused him to develop throat cancer.
Handley JA and Beazley JA referred inter alia to Bonnington Castings Limited and McGhee and said (p.409A):
“These cases decide that where medical science does not permit a clear answer, the court can find that a disease was caused by a breach of duty which materially increased the risk of contracting it: see also Duyvelshaff v Cathcart & Ritchie Ltd (1973) 47 ALJR 410 at 417, per Gibbs J.”Their Honours then referred to Wilsher v Essex Area Health Authority (1980) AC 1074. This was a medical negligence case where there were five possible causes of the blindness suffered by the plaintiff only one of which arose from the negligence of the defendant. The English Court of Appeal followed McGhee and the plaintiff was successful.
In the House of Lords, Lord Bridge said that in McGhee the House had done no more than adopt a robust and pragmatic approach to the drawing of an inference of causation, but the decision did not assist a plaintiff where there were four other possible causes.
Causation is a question of fact to be resolved as a matter of commonsense. Their Honours said (p.410B):
“Some of the plaintiff’s consumption was not caused by the tort because the ‘ but for’ test was not satisfied. However on the jury’s findings a significant part of his total consumption would not have occurred ‘but for’ the tort seems to us that the rule in Bonnington Castings v Wardlaw and the related cases applies and that the nature of the duty and the chain of causation under consideration were not fundamental to those decisions. They supply a principle for deciding responsibility where injury results from a noxious substance situation where only some of the exposure was caused by the tort.Those cases do not stand alone. A tribunal of fact is entitled to find causation as a matter of commonsense from the sequence of events, although medical science does not support an affirmative answer, provided it does not exclude such a finding: see Adelaide Stevedoring Co Ltd v Forst (1940) 64 CLR 535 at 563-564, 569 and Tubemakers of Australia Ltd v Fernandez (1976) 50 ALR 720.
The principle established in Purkess v Crittenden (1965) 114 CLR 164 is also relevant. The majority there said (at 168):
“….where a plaintiff has….made out a prima facie case that incapacity has resulted from the defendant’s negligence, the onus of adducing evidence that his incapacity is wholly or partly the result of some pre-existing condition or that incapacity, either total or partial, would, in any event, have resulted from a pre-existing condition, rests upon the defendant.”
The plaintiff’s pre-collision smoking and drinking habits were not literally a ‘ pre-existing condition’. However the defendant’s case that the throat cancer would ‘ in any event’ have resulted from those habits is equivalent to the situations referred to in this passage. The later reasoning of the majority (at 168) therefore applies:
“…it is not enough for the defendant merely to suggest the existence of a progressive pre-existing condition in the plaintiff or a relationship between any such condition and the plaintiff’s present incapacity. On the contrary…both the pre-existing condition and its future probable effects or its actual relationship to that incapacity must be the subject of evidence which, if accepted, would establish with some reasonable measure of precision what the pre-existing condition was and what its future effects both as to their nature and their future development and progress, were likely to be. That being done, it is for the plaintiff upon the whole of the evidence to satisfy the tribunal of fact of the extent of the injury caused by the defendant’s negligence.”
The defendant’s evidence did not satisfy these tests. The defendant did no more than ‘ suggest’ that the plaintiff’s consumption of alcohol and tobacco, without any increase due to the tort, could have caused the cancer in any event. In our judgment the direction to the jury that they were entitled to award damages for the cancer if they found that the increase in consumption caused by the tort was significant was correct.
The issue was causation in fact of an actual event, and not causation of an hypothetical event where the court allows for the probability that the event would or would not have occurred: see Malec v J C Hutton Pty Ltd (1990) 169 CLR 638. Causation in fact of an actual event is an all or nothing issue. The plaintiff was not entitled to recover anything for his throat cancer unless he established, on the balance of probabilities, that it had been caused by the tort. Once the plaintiff succeeded on that issue the defendant was not entitled to any deduction for the chance that the plaintiff might have contracted throat cancer in any event: see Hotson v East Berkshire Health Authority (1987) AC 750 at 783.”
The question was again dealt with by the Court of Appeal (Mason P, Beazley JA and Stein JA) in Bendix-Mintex Pty Ltd & Ors v Barnes (1997) 42 NSWLR 307.
The learned President said (p.311D):
“It is sufficient for a plaintiff to establish that his or her injuries were “ caused or materially contributed to” by the defendant’s wrongful conduct: March v E & MH Stramare Pty Ltd (1991) 171 CLR 506 at 514. The basal principle does not change when the plaintiff sues more than one defendant. The test must be satisfied as against each defendant from whom the plaintiff seeks a verdict before such a verdict will be entered. Hope JA summed-up the principle in Kilgannon v Sharpe Bros Pty Ltd (1986) 4 NSWLR 600 at 628 when he said that cases such as Nesterczuk v Mortimore (1965) 115 CLR 140 and Host v Bassett (1983) 57 ALJR 681; 48 ALR 404:“…require a plaintiff to be able to point, on a balance of probabilities, to the particular defendant or defendants who was or were negligent; it is not enough that he can say that one of them must have been and all of them may have been negligent.”
The law might, and in some jurisdictions does, distinguish between proof of unreasonable want of care and causation of damage. But it does not do so in Australia, on my understanding. While the law is prepared to adopt “ a robust and pragmatic approach” (Wilsher v Essex Area Health Authority) (1988) AC 1074 at 1090, per Lord (Bridge) to proof of causation in fact, the principles I have referred to extend to proof of damage as I shall endeavour to demonstrate.
In this sense, where only one of two events could have caused or materially contributed to an injury, there is no difference in principle between a plaintiff who is in doubt as to which of several defendants is liable and a plaintiff who is in doubt as to which event (one innocent, the other negligent) touching a single defendant was the cause of compensable injury.”
At page 312A he said:
“The sufficiency of material contribution is well attested: March v E & MH Stramare (at 514); Bennett v Minister of Community Welfare (1992) 176 CLR 408 at 420. However, introduction of the notion of increased risk involves a different universe of discourse, albeit one which is not unhelpful, provided that the plaintiff’s ultimate onus of proof is not displaced.”At page 312B he said:
“Before the House of Lords decided Wilsher v Essex Area Health Authority, English courts had moved to the idea that a plaintiff who was unable to weigh the scales down in favour of one of several competing but independent causative events could still succeed if the defendant(s) owed a duty of care and was/were engaged in conduct that created the risk of the injury which occurred, even though the plaintiff was unable to ascertain the existence or extent of the contribution made by the breach(es) to the injury suffered. The trend reached its final and highest exposition in the judgment of Mustill LJ in the Court of Appeal in Wilsher v Essex Area Health Authority (1987) QB 730 at 771-772, where he said:
“There is, however, one problem which must be tackled. I had at one time believed that the present case is on all fours with the McGhee case (1973) 1 WLR 1 and that any apparent difference between the two simply stemmed from the way in which the problem happened to be expressed. I am now persuaded that this is not so, and that the two situations really are different. In the McGhee case there was only one risk operating, namely that contact of a sweaty skin with brick dust would lead to dermatitis. The fact that such contact did cause the injury was not in dispute. Just as in Bonnington Castings Ltd v Wardlaw (1956) AC 613 the defenders’ fault lay in not taking proper steps to reduce that single risk. The uncertainty was whether the fault had tipped the scale. In the present case there is a greater uncertainty. Instead of a single risk factor known to have caused the injury there is a list of factors, which cannot be fully enumerated in the current state of medical science, any one of which might have caused the injury. What the defendants did was not to enhance the risk that the known factor would lead to injury, but to add to the list of factors which might do so. I acknowledge that this is much further from the fact of Bonnington Castings Ltd v Wardlaw, which was the springboard for the McGhee case than were the facts of the McGhee case itself.
The question is whether this makes a crucial difference. The root of the problem lies in the fact that, for reasons of policy their Lordships’ House mitigated the rigour of the rule that the plaintiff must prove that the breach caused the loss, in the interests of achieving a result which was considered to be just. Given that this was a decision based on policy, rather than a chain of direct reasoning, the difficulty is to know whether a similar approach can properly be adopted in the different circumstances of the present case. After much hesitation I have come to the conclusion that it can. Reading all the speeches together, the principle applied by the House seems to me to amount to this. If it is an established fact that conduct of a particular kind creates a risk that injury will be caused to another or increases an existing risk that injury will ensue; and if the two parties stand in such a relationship that the one party owes a duty not to conduct himself in that way; and if the first party does conduct himself in that way; and if the other party does suffer injury of the kind to which the risk related; then the first party is taken to have caused the injury by his breach of duty, even though the existence and extent of the contribution made by the breach cannot be ascertained.”
Expressed in such terms, an onus was placed on the defendant to whom the principle applied to prove that he or she was not in fact responsible for the injury suffered: see also McGhee v National Coal Board (1973) 3 All ER 1008 at 1012-1013, per Lord Wilberforce.
This beneficial approach to proof of causation by plaintiffs was applied to a single defendant where it was uncertain whether a negligent or innocent act was the cause of injury ( McGhee v National Coal Board) as understood before Wilsher v Essex Area Health Authority ; and Wilsher v Essex Area Health Authority in the Court of Appeal) and to multiple defendants ( Fitzgerald v Lane (1987) QB 781; reversed on other grounds (1989) AC 328). And it was seen to offer a just solution to plaintiffs facing difficulties of proving causation due to the absence of factual evidence of which lay persons might attest ( Fitzgerald v Lane ) or due to the imperfect state of medical knowledge of which expert testimony was required ( Wilsher v Essex Area Health Authority (at 766E-F, 771B-C, 776E); Fitzgerald v Lane (at 799F-G). Fitzgerald v Lane was an example of the former and Wilsher v Essex Area Health Authority (in the English Court of Appeal) an example of the latter.
In Fitzgerald v Lane , there were three possibilities touching a complicated motor accident (two affecting one defendant and the third affecting the other defendant) that could account for the plaintiff’s tetraplegia. The trial judge held that each of the three possibilities was equally probable and that he was unable to distinguish between them. The Court of Appeal recognised that this did not mean that all three combined to produce the injury by cumulative effect: see at 795D. Nevertheless each defendant was held liable. In Wilsher v Essex Area Health Authority , the plaintiff’s blindness could have been caused either by one of four medical conditions associated with the plaintiff’s premature birth, or by the negligent saturation of oxygen by medical staff. Medical science was unable to identify the causal culprit. Since, however, the administration of excess oxygen increased the risk statistically in the sense of doubling the “ contingency” (19887) 1 QB at 771) a causal connection was “ presumed” (at 771) by the majority of the English Court of Appeal. Sir Nicolas Browne-Wilkinson V-C dissented.
This approach put English law in line with that then prevailing in North America with respect to what Prosser and Keeton describe as “ clearly established double fault and alternative liability ”: Prosser and Keeton on Torts, 5th ed (1984) at 271. The classic example is where two defendants negligently shoot but the plaintiff is struck by one bullet which might have been fired from either gun. As the matter stood in 1984, the North American case law took the position that, where negligence on the part of both defendants was clear, and it was only the issue of causation which was in doubt, the onus shifted to each defendant to displace that burden. If it could not be displaced, then a verdict should be entered against each defendant with equal contribution towards satisfying the verdict: Summers v Tice 199 P 2d 1 (1948); Cook v Lewis (1952) ` DLR 1. (The two cases are criticised by T B Hogan in “ Cook v Lewis Re-examined ” (1961) 24 Mod LR 331). A similar principle has been applied to products liability cases where a number of manufacturers were negligent in the preparation of a common drug but where the plaintiff was unable to identify the manufacturer of the precise product which led to his or her injuries. However, in that case each manufacturer’s liability was limited to its market share: Sindell v Abbott Laboratories 607 P 2d 924 (1980) (cert denied 449 US 912).
However, the Court of Appeal (majority) decision in Wilsher v Essex Area Health Authority was overturned in the House of Lords. The proceedings were remitted for a re-trial on the issue of causation. Significantly, the House of Lords adopted the dissenting judgment in the Court of Appeal (that of the Vice-Chancellor) who had said (1987) 1 QB at 779) that:
“….A failure to take preventative measures against one out of five possible causes is no evidence as to which of those five caused the injury.” (see (1988) AC at 1091).
In Wilsher v Essex Area Health Authority , Lord Wilberforce’s speech in McGhee v National Coal Board was identified as adopting a minority position which should be rejected: see Wilsher v Essex Area Health Authority (at 1087). Nevertheless the conclusion was affirmed, McGhee v National Coal Board being explained as promoting a robust and pragmatic approach to the particular facts which allowed an inference of negligence to be drawn even though medical or scientific expertise could not arrive at a definite or more probable conclusion. This was because the risk of contracting the particular disease in McGhee v National Coal Board (dermatitis) was increased the longer the brick dust remained on the body. In those circumstances, the trier of fact was able to infer that a failure to provide washing facilities materially contributed to the disease by materially increasing the risk.
The Supreme Court of Canada followed Wilsher v Essex Area Health Authority in Snell v Farrell (1990) 72 DLR (4th) 289, holding that the onus of proof does not shift in medical malpractice cases. In doing so, the Court explained that causation need not be determined by scientific precision; and that the common situation where the facts lie particularly within the knowledge of the defendant in malpractice cases means that very little affirmative evidence on the part of the plaintiff would justify the drawing of an inference of causation in the absence of evidence to the contrary: see at 299-301. Nevertheless, the reasoning was confined to medical malpractice cases, and Cook v Lewis was inferentially re-affirmed (at 299) on the basis that:
“…Reversing the burden of proof may be justified where
two defendants negligently fire in the direction of the plaintiff and then by their tortious conduct destroy the means of proof at his disposal. In such a case it is clear that the injury was not caused by neutral conduct. It is quite a different matter to compensate a plaintiff by reversing the burden of proof for an injury that may very well be due to factors unconnected to the defendant and not the fault of anyone.”See also Hollis v Dow Corning Corporation (1996) 129 CLR (4th) 609 at 639.
In Snell v Farrell , a robust factual approach to causation issues was compared to the theoretical reversal of onus of proof suggested by Lord Wilberforce in McGhee v National Coal Board . The Supreme Court endorsed Lord Bridge’s speech in Wilsher v Essex Area Health Authority (1988) AC at 1088 where he explained McGhee v National Coal Board in the following terms:
“…..where the layman is told by the doctors that the longer the brick dust remains on the body, the greater the risk of dermatitis, although the doctors cannot identify the process of causation scientifically, there seems to be nothing irrational in drawing the inference, as a matter of common sense, that the consecutive periods when brick dust remained on the body contributed cumulatively to the causation of the dermatitis. I believe that a process of inferential reasoning on these general lines underlies the decision of the majority in McGhee’s case.”
In Bennett v Minister for Community Welfare (at 421), Gaudron J notes that the debate in Canada on a possible shift in the onus of proof is seemingly resolved in the manner indicated by Lord Bridge in Wilsher v Essex Area Health Authority.
Where stands the current law in Australia? I am not aware of rejection of Wilsher’s reassessment of McGhee v National Coal Board . In March v E & MH Stramare (at 524), Mason J cites various passages in McGhee v National Coal Board , including a passage from the speech of Lord Wilberforce. It is significant, however, that he does so in support of the uncontroversial proposition that the plaintiff must prove that his or her injuries are “ caused or materially contributed to” by the defendant’s wrongful conduct. In St George Club Ltd v Hines (1961) 35 ALJR 106 at 107; (1962) ALR 30 at 41, the High Court, in a joint judgment of five justices, said:
“In an action of law a plaintiff does not prove his case merely
by showing that it was possible that his injury was caused by the defendant’s default, Bonnington Castings Ltd v Wardlaw (1956) AC 613; nor does proof of default followed by injury show that the default caused the injury, for as Viscount Simonds said in Quinn v Cameron & Robertson Ltd (1958) AC 9 at 23, “‘ Post hoc, ergo propter hoc ’” is a fallacy in respect of a breach of a statutory regulation as it is in respect of any other event in life’. Similar principles apply to a claim for compensation where it is necessary to show that the employment contributed to the injury.”See also Australian Iron & Steel Ltd v Connell (1959) 102 CLR 522 at 531-532; EMI (Australia) Ltd v Bes (1970) 2 NSWR 238; Duyvelshaff v Cathcart & Ritchie Ltd (1973) 47 ALJR 410 at 417; 1 ALR 125 at 138; Tubemakers of Australia Ltd v Fernandez (1976) 50 ALJR 720 at 724; 10 ALR 303 at 310; Wintle v Conaust (Vic) Pty Ltd (1989) VR 951 (discussed below); Western Australia v Watson (1990) WAR 248 at 286; Chance v Alcoa of Australia Ltd (190) Aust Torts Rep 67, 719.”
At page 316E he stated:
“In the end I do not think it open to an immediate court of appeal to take this path, in the light of the many Australian authorities starting with St. George Club Ltd v Hines which I have cited above; and the more recent developments in England and Canada.
In Pibworth v Bevan M Roberts Pty Ltd (1994) 176 LSJS 39 (otherwise unreported, Supreme Court of South Australia, Full Court, 15 April 1994) the Full Court of the Supreme Court of South Australia followed Wilsher v Essex Area Health Authority and adopted the summary of what the case stands for offered by Andrew Grubb in the case note (“ Causation and Medical Negligence”) in (1988) Cam LJ 350. That summary states, as its third proposition (at 350-351): “The law does not equate the situation where the defendant had materially increased the risk of injury with one where he had materially contributed to the injury.”
I respectfully agree. Only parliament or the High Court may change this.The difficulties which these principles place in the way of a plaintiff can be very considerable. To say that they are the logical outcome of a legal system that requires proof of fault before damages flow against a particular defendant does not remove a lingering sense of injustice, particularly in the case of multiple defendants where each has been shown to have negligently exposed a plaintiff to a risk of injury. Nevertheless our law remains wedded to the principle that if the risk does not come home in a way that is causally linked to the particular defendant’s negligence, then it is the plaintiff who must bear the loss: see Rhesa Shipping Co SA v Edmunds, The “Popi M” (1985) 2 Lloyd’s Rep 1 at 5. These principles have been reversed by legislation in several jurisdictions, including Ireland, Germany and the Netherlands: Civil Liabilities Act 1961 (Eire), s 11(3); J G Fleming, “ Probabilistic Causation in Tort Law” (1989) 68 Can Bar Rev 661 at 671; M Vranken, “ The first decennium of the European Product Liability Directive: a cause for celebration?” (1996) 4 Torts LJ 225 at 236-238.
In Kilgannon v Sharpe Bros Pty Ltd, Kirby P suggests that the principles of res ipsa loquitur may bring about a similar result in some circumstances involving multiple defendants all having “ control” of the relevant situation: see at 6140619. However, this view was rejected by the majority of the Court (Hope JA and Priestley JA). More importantly for present purposes, Kirby P recognised that it would be unreasonable and possibly unjust to apply the principle he espoused where “ there is a possible explanation, in an extraneous cause, originating in a party who is not before the Court” (at 617D; see also at 618B). The absence of the Royal Navy as a defendant in the present case is therefore critical, even on Kirby P’s approach.
It can be demonstrated that the common law is not unsympathetic to the plight of plaintiffs who are faced with multiple defendants yet uncertain as to which of them was legally responsible, where it appears that not all of them were……..”
At page 318A he said:
“But none of these procedural or adjectival concessions in favour of an uncertain plaintiff support the principle that a party who negligently exposes a plaintiff to a risk of injury will be liable unless the plaintiff can persuade the trier of fact that it was probable that the risk came home. The basal principle remains that: “ The law never gives judgment in favour of a plaintiff when the only finding is equally consistent with liability and non-liability.” (Moriarty v Evans Medical Supplies (1958) 1 WLR 66 at 91, per Lord Denning.)At page 319C he said:
“I agree with Beazley JA on the issue of causation. My view may be summarized as follows:
1. ………
2. The ultimate issue, in the light of the principles which I have endeavoured to summarise, is which (if any) of the persons sued by the respondent were shown, on the balance of probabilities, to have caused or materially contributed to Mr Barnes’ mesothelioma.3. Expert evidence, which the trier of fact is entitled to accept despite the weight of evidence to the contrary, established that the appellants might have caused or materially contributed to Mr Barnes’ fatal illness.. (I refer to the evidence of Dr Burns and Dr Joseph.)
4. Nevertheless, the evidence as to the type of fibre to which the respondent was exposed in the Royal Navy (crocidolite), the dose to which he was then exposed and the lag time involved all pointed strongly in the direction of the Royal Navy. Although that evidence, in the final analysis, merely supported the greater risk of the fatal disease having been contracted by the respondent during this period of his working life, that evidence (taken cumulatively) can, and in my mind does, support the finding that this was the employer more probably than not whose act or omission caused the fatal disease. As against the Ministry of Defence (UK) it is a situation where an inference of causation is warranted in that there is no practical difference between materially contributing to the risk of harm and materially contributing to the harm itself: cf Snell v Farrell (at 296G). Another way of addressing the issue is to hold that, were the Ministry of Defence (UK) the sole defendant, then a verdict against that defendant would be clearly open on the evidence. Judge Maguire reached a similar conclusion in the passage quoted early in this judgment.
5. None of this precludes a verdict against one or more of the appellants each of whom have been shown to have negligently exposed Mr Barnes to a risk of contracting the fatal disease. But that is not enough. Has it been shown to be probable that the risk they or any of them created actually caused injury? It will be sufficient if their several acts caused or contributed materially to the onset of mesothelioma.
6. As Beazley JA demonstrates, the evidence of the respondent’s principal expert witnesses (Dr Burn and Dr Joseph) does not rise higher than demonstrating that the chrysotile in the brake linings was a potential contributor to Mr Barnes’ mesothelioma. Unlike the appellants’ expert witnesses, Dr Burns considered that chrysotile was “just as likely to be harmful as any other asbestos.” This, if accepted, provided the evidentiary basis upon which the trier of fact might conclude that the appellant were responsible in fact and law for the fatal disease. But it does not compel that conclusion. I read the evidence about repeated stimuli increasing the chances of getting mesothelioma in the same sense.
The effect of the evidence of Mr Barnes’ exposure to greater quantities of the more risky crocidolite at a more likely time cannot be ignored. Nor can one ignore the evidence of lighter exposure to the less lethal cyrysotile in the more recent periods involving the appellants. The evidence of Dr Burns and Dr Joseph simply fails to offer a scientific or indeed a rational basis for preferring the chrysotile of the appellants over the crocidolite of the Royal Navy; or for concluding that the later exposure had a harmful cumulative effect, like the brick dust in McGhee v National Coal Board on prolonged exposure to noise. The essence of this view is that the scales remain equipoised because medical science cannot say which exposure was significant. If the Royal Navy exposure had not occurred, the trier of fact (and, one infers, even the appellants’ experts) would have found it probable that the appellants were liable. But the presence of the Royal Navy cannot be ignored, especially since the evidence points strongly to it in the sense that it, more than any other person, materially increased the risk of contracting mesothelioma.
7. It would be speculation, not inference, to conclude that it was more probable than not that the appellants or any of them materially contributed to the respondent’s disease. The highest that the respondent’s case can be put is that the appellants’ involvement increased the statistical contingency that their acts or omissions may have caused or materially contributed to the mesothelioma. The inability of medical science to identify the aetiology of the disease or to establish a likelihood let alone probability that the appellants’ asbestos contributed to the disease, means that the respondent has not established against any party from whom he seeks a verdict that that party is liable in the tort of negligence.
Beazley JA also dealt with causation, reaching the same conclusion as the learned President.
At page 335 her Honour said:
“Causation is a question of fact to be determined by the application of commonsense to the facts of each case: March v E & MH Stramare Pty Ltd (1991) 171 CLR 506 at 515, per Mason CJ; Bennett v Minister of Community Welfare (1992) 176 CLR 408. Notwithstanding this exceedingly simple formulation of principle, its application in a case of this sort is not easy of resolution. Mesothelioma is a condition of uncertain aetiology and Mr Barnes was serially exposed to asbestos which the evidence indicated possibly had different degrees of toxicity depending upon the particular fibre, the first exposure being to crocidolite, which is considered to be a more virulent causative agent of mesothelioma than chrysotile, to which Mr Barnes was later exposed.The factual question which calls for resolution in the first instance is whether, on the balance of probabilities, the exposure to chrysotile during the period 1962 to 1985 or any relevant part of this period, caused or materially contributed to Mr Barnes’ contraction of mesothelioma: Bonnington Castings Ltd v Wardlaw (1956) AC 613.
In Bonnington Castings Ltd v Wardlaw , a steel dresser was exposed to silica dust from two sources in the course of his employment: a pneumatic hammer and swing grinders. There was no dust extraction plant known or practicable for use with the hammer. Although the swing grinders were fitted with such equipment, they were not kept free from obstruction. The employer conceded it was thereby in breach of relevant regulations which governed its operations. The steel dresser contracted pneumoconiosis. The question which arose was whether the worker had established that the cause of his pneumoconiosis was due to the breach of statutory duty. Lord Reid stated (at 620-621):
“…In my judgment, the employee must in all cases prove his case by the ordinary standard of proof in civil actions: he must make it appear at least that on balance of probabilities the breach of duty caused or materially contributed to his injury….
….I am in agreement with much of the Lord President’s opinion in this case, but I cannot agree that the question is: which was the most probable source of the respondent’s disease, the dust from the pneumatic hammers or the dust from the swing grinders? It appears to me that the source of his disease was the dust from both sources, and the real question is whether the dust from the swing grinders materially contributed to the disease. What is a material contribution must be a question of degree. A contribution which comes within the exception de minimis non curat lex is not material, but I think that any contribution which does not fall within that exception must be material. I do not see how there can be something too large to come within the de minimis principle but yet too small to be material….”
Dealing with the particular facts in the case, Lord Reid stated (at 622):
“I think that the position can be shortly stated in this way. It may be that, of the noxious dust in the general atmosphere of the shop, more came from the pneumatic hammers than from the swing grinders, but I think it is sufficiently proved that the dust from the grinders made a substantial contribution. The respondent, however, did not only inhale the general atmosphere of the shop: when he was working his hammer his face was directly over it and it must often have happened that dust from his hammer substantially increased the concentration of noxious dust in the air which he inhaled. It is therefore probable that much the greater proportion of the noxious dust which he inhaled over the whole period came from the hammers. But, on the other hand, some certainly came from the swing grinders, and I cannot avoid the conclusion that the proportion which came from the swing grinders was not negligible.”
Lord Reid reiterated the principle he had formulated in
Bonnington Castings Ltd v Wardlaw in McGhee v National Coal Board (1973) 1 WLR 1 at 4; (1972) 3 All ER 1008 at 1010, stating:“It has always been the law that a pursuer succeeds if he can show that fault of the defender caused or materially contributed to his injury. There may have been two separate causes but it is enough if one of the causes arose from fault of the defnder. The pursuer does not have to prove that this cause would of itself have been enough to cause him injury. That is well illustrated by the decision of this House in Bonnington Castings Ltd v Wardlaw (1956) AC 613.”
The causation issue in McGhee v National Coal Board was slightly different to the problem encountered in Bonnington Castings Ltd v Wardlaw . McGhee had been a long time employee in brickworks and during the course of his daily employment would become covered in brick dust. The respondent failed to provide him with showering facilities so that at the end of the work day, he would cycle home still covered in the brick dust. He contracted dermatitis. Lord Reid noted that the evidence did not establish how dermatitis of the type suffered by the worker was contracted. Two possible theories were advanced, one to the effect that there had t be multiple tiny abrasions to the horny layer of the skin before the dermatitis developed, the other to the effect that a single abrasion was sufficient, the effect of multiple abrasions being to increase the possible sites where the disease could start. Lord Reid considered that the evidence tendered to the former view. However, the lack of certainty as to the precise cause of the disease did not mean that a plaintiff could not establish causation. As Lord Reid further stated (at 4; 1011):
“…in a field where so little appears to be known with certainty I could not say that that is proved. If it were, then this case would be indistinguishable from Wardlaw’s case. But I think that in cases like this we must take a broader view of causation. The medical evidence is to the effect that the fact that the man had to cycle home caked with grime and sweat added materially to the risk that this disease might develop. It does not and could not explain just why that is so. But experience shows that it is so.”
See also Lord Simon (at 8), Lord Kibrandon (at 10) and Lord Salmon (at 12).
Lord Reid added that he saw no distinction between “ materially increasing the risk that the disease will occur and making a material contribution to its occurrence.”
Bonnington Castings Ltd v Wardlaw and McGhee v National Coal Board were explained in Wilsher v Essex Area Health Authority (1988) AC 1074 at 1087, by Lord Bridge of Harwich in there terms:
“A distinction is, of course, apparent between the fact of Bonnington Castings Ltd v Wardlaw , where the ‘ innocent’ and guilty’ silica dust particles which together caused the pursuer’s lung disease were inhaled concurrently and the fact of McGhee v National Coal Board (1973) 1 WLR 1 where the ‘innocent’ and ‘ guilty’ brick dust was present on the pursuer’s body for consecutive periods. In the one case the concurrent inhalation of ‘ innocent’ and ‘ guilty’ dust must both have contributed to the cause of the disease. In the other case the consecutive periods when ‘ innocent’ and ‘ guilty’ brick dust was present on the pursuer’s body may both have contributed to the cause of the disease or, theoretically at least, one or the other may have been the sole cause. But where the layman is told by the doctors that the longer the brick dust remains on the body, the greater the risk of dermatitis, although the doctors cannot identify the process of causation scientifically, there seems to be nothing irrational in drawing the inference, as a matter of common sense, that the consecutive periods when brick dust remained on the body probably contributed cumulatively to the causation of the dermatitis. I believe that a process of inferential reasoning on these general lines underlies the decision of the majority in McGhee’s case.”
Lord Bridge further stated (at 1090):
“The conclusion I draw from these passages is that McGhee v National Coal Board (1973) 1 WLR 1 laid down no new principle of law whatever. On the contrary, it affirmed the principle that the onus of proving causation lies on the pursuer or plaintiff. Adopting a robust and pragmatic approach to the undisputed primary facts of the case, the majority concluded that it was a legitimate inference of fact that the defenders’ negligence had materially contributed to the pursuer’s injury. The decision, in my opinion, is of no greater significance than that and to attempt to extract from it some esoteric principle which in some way modifies, as a matter of law, the nature of the burden of proof of causation which the plaintiff or pursuer must discharge once he has established a relevant breach of duty is a fruitless one.”
In Wilsher v Essex Area Health Authority, the defendant’s staff negligently allowed the plaintiff to be exposed to excess oxygen which could have caused his blindness. During the same period he suffered from four other conditions which could also have caused his blindness. The Court of Appeal, purporting to apply McGhee v National Coal Board held that the breach of duty established a presumptive inference that the exposure to the oxygen was the cause of the blindness. However, the House of Lords rejected the proposition that there was any reversal of onus in such a case such that the defendant had to prove that its breach had not materially contributed to the plaintiff’s injury. They endorsed the dissenting view of Browne-Wilkinson V-C (at 1091) that:
“A failure to take preventative measures against one out of
five possible causes is no evidence as to which of those five caused the injury.”The principle established by Bonnington Castings Ltd v Wardlaw and McGhee v National Coal Board has been consistently applied in England and Australia. In March v E & MH Stramare , Mason CJ stated (at 514):
“….the law’s recognition that concurrent or successive tortious acts may each amount to a cause of the injuries sustained by a plaintiff is reflected in the proposition that it is for the plaintiff to establish that his or her injuries are ‘ caused or materially contributed to’ by the defendant’s wrongful conduct: Duyvelshaff v Cathcart & Ritchie Ltd (1973) 47 ALJR 410 at 417; 1 ALR 125 at 138, per Gibbs J; Tubemakers of Australia Ltd v Fernandez (1976) 50 ALJR 720 at 724; 10 ALR 303 at 310, per Mason J:
Bonnington Castings Ltd v Wardlaw (1956) AC 613 at 620; per Lord Reid; McGhee v National Coal Board (1973) 1 WLR 1 at 4,6,8 and 12. Generally speaking, that causal connexion is established if it appears that the plaintiff would not have sustained his or her injuries had the defendant not been negligent: see ICIANZ Ltd v Murphy (1973) 47 ALJR 122 at 127-128. But, as the decision in that case illustrates, it is often extremely difficult to demonstrate what would have happened in the absence of the defendant’s negligent conduct.”See also Bennett v Minister of Community Welfare; Commonwealth v McLean; Birkholtz v R J Gilbertson Pty Ltd (1985) 38 SASR 121; Seltsam Ltd v Minahan.
Wintle v Conaust (Vic) Pty Ltd (1989) VR 951 was a mesothelioma case in which the applicant sought an extension of time within which to commence proceedings. Crockett J and Gray J dismissed an appeal from the trial judge’s refusal of the application, stating (at 953):
“The difficulty simply is - as the judge pointed out - that the evidence does not disclose for how long (constantly or intermittently) a person must be exposed to asbestos dust before it can be said that a subsequently contracted mesothelioma was as a matter of probability caused by that exposure.”
Their Honours added:
“In cases such as that with which the court is now concerned the burden of proof of actionable negligence against any defendant remains firmly with the plaintiff. Despite this court’s rejection of the hearsay in Pastras v The Commonwealth (1967) VR 161, there have remained suggestions that a defendant carried the burden of disproof if it were shown that he was one of a number any one or more of whom must have been negligent. These have now been firmly rejected by the House of Lords: Wilsher v Essex Area Health Authority (1988) 2 WLR 557.”
The clear effect of these cases is that the onus remains on the plaintiff to prove causation on the balance of probabilities. The onus is not discharged by establishing that a particular matter cannot be excluded as a cause of the injury: see Sydney County Council v Furner (1991) 7 NSWCCR 210, where Hope A-JA said (at 214):
“…To find that a particular cause cannot be excluded as the cause of the relevant injury does not establish that on the probabilities it was the cause of that injury; it may have been ‘ the’ or ‘ a’ cause. In other words, to establish that medical evidence supports a conclusion that the cause could not be excluded as a cause of an injury does not establish, without more, that particular cause in fact resulted in or caused the injury.”
Subsequently the Court of Appeal decided ICI Operations Pty Limited v Watel & Ors . (Unreported 3 October, 1997). The bench consisted of the President and Justices Handley and Meagher. The learned President shortly restated his views as expressed in Bendix. Handley JA (one of the members of the Court in McLean ) agreed with the views of the President.
In my view the law of New South Wales is as was stated by the learned President in Bendix. It is not, as he says, for an intermediate Court of Appeal to change it let alone for a trial judge such as myself.
Whatever may be one’s views of the social desirability of the High Court and Parliament of changing the law to in effect shift the burden of proof to a defendant of showing that his negligent actions did not cause injury to the plaintiff where the plaintiff has shown that the actions of the defendant increased the risk of the plaintiff sustaining the injury which was sustained, it is not open to me to decide this case on that basis.
Simply establishing that a causal connection between the actions of the defendant and the injury to the plaintiff is possible does not justify a finding of causation unless the evidence justifies an inference of probable connection.
Accordingly I do not accept the submission on behalf of the plaintiffs in respect to the proof of causation. In order to succeed the plaintiffs must, according to the law of New South Wales, establish that the or a probable cause of their injury was the negligence of the defendants.
4. PROBABILITY OF CAUSATION
The evidence of causation relied on by the plaintiffs as pointing to the conclusion that Cu-7 use caused or materially contributed to the acquisition of PID and its sequelae and to septic abortion was as follows:
4.1 Epidemiology
4.2 Scientific analysis
4.3 Clinical observation
The plaintiffs’ case is that use of IUDs “enhanced” or “materially contributed ” to PID.
4.1 EPIDEMIOLOGYThe threshold question is whether a user of a Cu-7 was placed at a greater risk of PID than if she had not used a device. The parties were diametrically opposed on the answer to this question.
The plaintiffs assert that there was an increased risk. The defendants assert that the only increased risk involved was a risk associated with the process of insertion of the device into the uterus (an iatrogenic risk).
The plaintiffs seek to establish the existence of an increased risk by the use of epidemiological evidence. They submitted that throughout the decade from 1975 when the first insertion of an IUD relevant to these proceedings occurred “ there was a drum beat of concern in the medical literature about the increased risk from IUD use, PID and its sequelae.”
The defendants conceded that virtually all of the numerous case studies published reported a statistical association between PID and the use of an IUD but submitted that there are “ cogent reasons why these studies, when properly interpreted, cannot establish a causal link between the use of the Gravigard/Mini and PID with its sequelae. ”
Epidemiology is the statistical study of disease in relation to such things as prevalence or factors associated with suffers of the disease which may be causative of it.
One method of statistically determining cause is by conducting what is known as an observational cohort study. An example of this is given by the defendants in Chapter 7 Volume 1 Page 6 of their submissions. That example is that if an investigator suspects that tea drinking causes renal stones, a group of four thousand tea drinkers and a group of four thousand non-tea drinkers are assembled. They are then followed for ten years and results of their renal stone incidents are tabulated. If there are fewer renal stone incidents among non-tea drinkers than there are among tea-drinkers then that may suggest that tea drinking may cause renal stones. The difference in the number of incidents will indicate the seriousness of the hazard.
An alternative method of determining cause is a case control study which follows a reverse direction.
The investigator studies people who already have the disease and a control group is assembled comprising people in whom the disease has not been noted. The two groups are then followed in a backward direction with enquiries aimed at determining each member’s antecedent exposure to the alleged aetiological agent or cause of the disease.
The defendants submitted that one can conclude that any difference which appears in the results of a case control study only reflects some causal link if it is possible to exclude all other plausible explanations for the difference. They refer to “ non-causal” explanations which they submit must be excluded as:
(a) Selection bias:
This is said to mean that there is some fundamental difference
between the members of the two groups “ rather than the tea
drinking having any effect. ”In support of this proposition the defendants refer to an article
by Grimes “ Biases: the Saboteurs of Silence “.
(Exhibit D384 Tab 10). In that article Grimes refers to an
example of methological weaknesses in cohort studies as
examples, which methological weaknesses may have
damaged a study’s validity. The first he refers to is selection
bias where he opines that “ selection bias implies that
subjects in one study group are different from those in
another in some feature other than the exposure of interest.
If this feature is not measured it cannot be controlled for
in the analysis and thus may bias the results. ”
He concluded: “ In a cohort study, both groups should be
comparable in all respects, except for the exposure of
interest (here hormone therapy) ”.
Grimes is referring to cohort studies in the article referred to
but it is difficult to see why there should be any difference in
principle in this aspect between cohort studies and case
control studies.
(b) Assessment or data bias:
The diagnosis is not assessed in identical ways in the two
groups that are being studied.
(c) Chance:
Despite there being no “ real” difference sampling error or
random variation has resulted in difference in the study data.The studies referred to in these proceedings are essentially case control studies. They are “ retrospective studies. ” They compare women who had the problem under investigation and “ controls ”, women who did not have the same problem.
Professor Newton said that the disadvantage of this type of investigation is that “ bias is inherent and difficult to control.”
(Exhibit D299 para 105(b)).Another difficulty with case control studies are what are known as confounders. The most common confounder in medical studies is the age of the persons studied. In a comparison there should be an adjustment for age so that the comparison is between people of the same age.
Professor Daling who gave evidence on behalf of the plaintiffs said (T2770) that it is not possible to entirely remove bias but “ I mean we do the best we can to remove all possible bias. ”
At T2773 Professor Daling said, when referring to the elimination of the effect of other factors relating to a disease, “ you have taken all those steps made possible by your data collection and your statistical tools to reduce as much as possible. ”
Studies may suffer from bias for various reasons and the effect of that is that you may get an estimate of the risk which is not the true estimate of the risk. (Daling, T2768).
The defendants in their submissions dealt simply with the definition of a number of relevant epidemiological terms and it is convenient to deal with those definitions at this stage. They were illustrated by the following example.
5,000 men and 5,000 women crossed Elizabeth Street and the number of each group who stumbled were counted.
Of the 5,000 men 100 or 2% stumbled. Of the 5,000 women 150 or 3% stumbled.
Those percentages are both incidence rates or simply the likelihood of stumbling. From the figures appearing in this example it would be correct to say that women have an increased incidence rate of stumbling compared to men.
The relative risk is 3%/2% or 1.5 which means that women are 1.5 times more likely to stumble while crossing Elizabeth Street as men. One can then say that women have an increased risk of stumbling compared to men.
A case control study does not obtain data which enables incidence rates to be calculated so what is known as an “ odds ratio ” is used to estimate the relative risk and adding a confidence interval.
If it is sought to use data from a case study to estimate relative risk one assumes that the larger study is the more accurate will be the estimate.
A confidence interval is calculated. This is an interval within which the true relative risk probably falls. In simple terms, it is likely that the larger the data set the greater is the confidence interval. A 95% confidence interval is used to indicate that the true odds ratio or relative risk is 95% likely to be at the figure or band determined.
If the relative risk was 1.5 a 95% confidence interval means that if repeated samples of the same size were taken from the population study and studied, ninety-five times out of one hundred the relative risk calculated would fall within the confidence limit.
The defendants submitted that it is not possible to come to a conclusion regarding causation from statistical results obtained in observational epidemiology or at least coming to a conclusion is fraught with difficulty. The defendants refer to Feinstien and Horwitz (Exhibit D291) where the author attacked the scientific methods used by epidemiologists.
Keeping gave evidence in relation to the use of epidemiology at some length and concluded that the procedures and processes “ still falls short of the only true method, which would be a control trial.”
There have been two significant reviews of the literature relevant to the issue of causal connection between IUD use and PID and its sequelae including ectopic pregnancy.. First was Senanayke and Kramer 1980 and second was the WHO Technical Group Report 1987 .
The conclusion reached by Senanayke and Kramer was:
“Taken together, the reported studies on the IUD and PID
deal with most of biases that could explain the apparent
increases in risk with the IUD. Results of the studies are
consistent in direction despite varying study designs,
researchers, and locations, and they generally point to an
increased risk of 150% to 400% for IUD users when
compared with all non-users of contraceptives.”
(p.857, LHC, last para).Review Articles are articles which overview the relevant medical literature and attempt to synthesise the state of knowledge researched at the time of writing. Professor Daling gave evidence that one criterion to be looked at is what is known as “ consistencies across the studies ”, namely a pattern mostly in one direction or the other across different study populations observed by different researchers.
Senanayke and Kramer 1980 referred to above, canvassed sixteen studies. They are set out in Attachment 2 which forms part of the review. The WHO Technical Group Report 1987 canvassed those sixteen studies together with nine others.
In all sixteen studies reported on by Senanayke and Kramer that there was an increased risk of PID from IUD use whether the comparison was made with women using other forms of contraception (the results show an increased risk varying from 1.9 to 9.3) or using no contraception at all (the results show an increased risk varying from 1.5 to 7.3).
The nine further studies dealt with by the WHO Report show on their face an increased risk for IUD users varying between 2.1 and 10.5 compared to women using other forms of contraception and between 1.5 and 11.5 compared to women using no contraception.
The conclusion reached by the WHO Technical Group was:
“Since 1980, when 16 reports were reviewed by Senanayke
and Kramer, 9 additional reports of significantly increased risk of pelvic inflammatory disease or tubal infertility among IUD users have been published. These 25 studies, conducted in many different countries by different investigators, albeit with different diagnostic criteria, have all found an increased risk of the disease or its sequelae among IUD users, which argue strongly for a causal relationship” (WHO Report p.38, LHC, para 7.4).Eschenbach gave evidence (T1822, L46-57) that the three studies of Wright and Laemmle 1968, Noonan and Adams 1974 and Targum and Wright 1974 were enough to lead to the conclusion that that there was an increased risk of PID from IUD use.
Ory said of the six epidemiological studies he reviewed:
“In summary, then, I think that these six studies point in the
same direction, that IUD use is associated with the increased risk of acute PID. The best estimate suggests that IUD users have from three to five times the risk of developing PID compared to non-users, that teenagers appear as likely to be affected as older women, that copper IUD’s do not offer any protection from gonococcal PID, and finally, there is the intriguing but as yet unconfirmed suggestion that nulliparous IUD users are more likely to develop acute PID than are multi-parous users.
As an epidemiologist, I kind of look at these six studies, you know, done by five different investigators at three different countries, all showing the same results, and that, to me, provides some evidence that the association is one of cause and effect. Each study had its strengths and its weaknesses. However, all together, they rule out virtually every major source of bias as a possible reason for observing this association.
In addition, when you couple it with Dick Soderstrom’s recent report of the histologic changes and the possibility that this is altering tubal defense mechanisms; also if you couple it with the fact that previously rare syndromes, or so they are said to be in the literature, such as unilateral tubo-ovarian abscess and infective agents like Actinomycoses are now being reported in association with IUD use, I think biologic plausibility is added to a causal interpretation of these six studies.” (Ex. P228, Tab 13, pp.72-73).The plaintiffs sought a finding that at the time of each IUD insertion relied upon by the plaintiffs, Searle was in possession of strong evidence from epidemiological studies which pointed to a causal association between IUD use and PID. The association, as revealed by the studies referred to above, was so strong and consistent that Searle was under a legal obligation to act on the basis of that causal association.
The plaintiffs submitted that if this finding is made and if subsequent scientific research has not reviewed the conclusion in so far as it applies to Cu-7 then their case relating to a failure to warn is made out.
The defendants submitted that later epidemiology was able to isolate the risk associated with Cu-7 itself and established that there was no increased rate of PID associated with Cu-7 use. The defendants further submitted that there was no increased risk among women in stable monogamous relationships.
The defendants dealt with the papers referred to in the two overview papers to attempt to demonstrate that they could not be relied upon to support the conclusion contended for by the plaintiffs.
I shall deal with each of the papers in the order in which they appear in the reports.
Wright & Laemmle 1968
This is one of the first studies which attempted to look at the aetiology and epidemiology of PID within a large family planning program:
The study was carried out in 1965 at the Grady Memorial Hospital, Atlanta, Georgia amongst a population of black indigent women.
The methodology was that 902 first episode cases of acute PID were obtained from the Hospital records. The denominator used was the estimated total population of indigent women aged between 15 to 44 years in the 2 counties surrounding the hospital. A rounded-up figure of 50,000 was used to calculate the rates.
A definite diagnosis of acute PID was made if a patient presented with a fever plus 2 or more of the following:
· lower abdominal tenderness with or without guarding;
· cervical motion tenderness;
· uni- or bi-lateral adnexal tenderness with or without masses.An exception was made if there was actual visualization of PID at laparotomy.
The authors looked at different factors which might have been relevant to the incidence of PID including age and marital status. They also looked at attack rates of acute PID among women according to the type of contraceptive method that was currently being used.
The results of an analysis of 108 subjects with PID appear in Table VIII on page 985 of the paper. A summary of the table prepared by the defendants appears below:
IUD
% Selecting Method
Woman Years
PID
PID Per 1000 Woman Years43
1238
82
66.2
OC
44
1267
17
13.4
Vaginal Foam
11
317
8
25.2
Other
<2
58
1
Total
100
2880
108
37.5 (Average)
The summary table shows a significantly higher incidence of PIDamong IUD users than among the other types of contraceptors reported
on.
The defendants point to the following alleged shortcomings:
1. No information is given in the paper as to the type of IUD used
by the subjects in this study but they would not have included
Cu-7s. (It was not on the market then).
2. There is no information in the paper as to the duration of use and
the incidence of PID among cases.
3. No relative risk estimates were reported by the authors.
4. The authors state that the study findings should be considered
as being tentative.
5. It appears that most of the IUD-associated attacks of PID
referred to in the paper were mild and were treated with the
device in place.Given the above matters the defendants submitted that extrapolation from the results of the study is “ fraught with difficulty ”.
Noonan & Adams 1974
The aim of this study was to determine the relationship between the prevalence of gonorrhoea and the presence of symptoms and the use of various contraceptive methods.
This study was also carried out in the Family Planning Clinic of Grady Memorial Hospital in Atlanta, Georgia amongst a population of black indigent women.
1,706 women were selected from the clinic being every third non-post partum patient presenting at the clinic. They were screened for cervical gonorrhoea and questioned for a history of PID symptoms.
A non-specific definition of PID was used. Subjects were classified as being symptomatic if:
(a) they complained of pelvic pain or cramping abdominal pain at
times other than during menstruation; or
(b) if the examining physician elicited pelvic tenderness that the reviewer felt was pathological. Dyspareunia alone was not considered a symptom.Among women with symptoms of PID, 20.6% used oral contraceptives, 33.3% used IUDs and 16.2% used other methods of contraception.
Among women diagnosed with gonococcal PID, 0.7% used oral contraceptives, 3.1% used IUDs and 0.2% used other methods of contraception.
There were no relative risk estimates reported. However it was subsequently reported in Senanayake & Kramer 1980 that women with IUDs had symptoms of PID 2.5 times more frequently than women not using any contraception and 1.8 times more frequently than women using all other methods of contraception combined (Ex. P124 page 854).
No information was given as to the type of IUDs being used by subjects in this study but Cu-7s were not used.
The authors stated that the prevalence rate of gonorrhoea was essentially the same for IUD and other contraceptive users. They concluded (page 704) that none of the methods of contraception (including IUDs) seemed to be associated with an increase in the prevalence of gonorrhoeal infection.
The increased prevalence of symptoms of PID in IUD users was a comparative observation to those women who were using methods of contraception which provided a chemical or physical barrier against the ascension of bacteria from the vagina. The authors hypothesized that the protective effect of other types of contraceptors may have given rise to the differing prevalence rates of PID.
Targum & Wright 1974
This was a case-control study which examined first-episode acute PID in 50 disadvantaged black and Hispanic patients in a major New York hospital:For a diagnosis of PID there needed to be a temperature in excess of 37.7 o C and 2 of the following 3 findings:
· lower abdominal pain and tenderness;
· cervical motion tenderness;
· adnexal tenderness.Two sets of controls were chosen, one from women attending the general medical clinic and the other from women appearing with sick children in the paediatric emergency room.
100 controls were selected and their current contraceptive practises were as follows:
· 9% IUDs;
· 50% OCs;
· 41% no contraception.The current contraceptive practises of the 50 cases were:
· 48% IUDs
· 22% OCs
· 36% no contraception.Of the 24 cases using an IUD, 11 women had Lippes Loops, 8 had the Majzlin Springs (ultimately recalled by the FDA) and 4 were using the Saf-T-Coil. One women did not know which IUD she was wearing.
The relative risk estimate for IUD users versus non-users was 9.3
(4.2 - 20.9).The relative risk for IUD users versus non-contraceptors was 7.3
(2.9 - 18.4):The relative risk estimates for IUD users in this study is one of the highest reported in any of the studies and may have been due in part to the fact that of the 50 subjects 40 had been hospitalised.
Of the 50 PID cases, there were 24 IUD users. Of the 100 controls, there were 9 IUD users. There were no copper devices in the study. The relative risk estimates were accompanied by very wide confidence intervals. Eschenbach said, as I have stated above that in his view those three studies were enough to lead to the conclusion that there was an increased risk of PID from IUD use.
Lippes 1975
This study was a preliminary report to answer the question “ Does the IUD increase the incidence of pelvic infection? ” It was prepared by Dr Jack Lippes and presented to the FDA at its hearing on IUDs in August 1974.
To prepare for his testimony (and therefore the study) the medical record library of the Buffalo General Hospital was asked to review all cases carrying a discharge diagnosis of PID, salpingitis or endometritis for the year 1973. 91 patients were identified as a result.
The diagnosis of PID was based on acceptance of a discharge diagnosis of PID, salpingitis or endometritis but was otherwise not specified.
Controls were identified by a review of the charts of all women of reproductive age with a diagnosis of acute appendicitis for the years 1972 and 1973. 38 controls were identified.
Among the cases identified, 13 of the 91 (14%) were wearing an IUD. Among controls, 3 of 38 (8%) were wearing an IUD. The difference on these very small numbers is not statistically significant.
One Dalkon Shield associated with a serious bacterial septicaemia was specifically identified, but otherwise, individual devices were not specified.
The relative risk estimate for IUD users versus non-users was 1.9 (0.5-7.2).
The relative risk estimate for IUD users versus non-contraceptors was not ascertained.
Dr Lippes concluded from the data collected that the incidence of PID was no higher in women wearing an IUD than in women who did not. However, he acknowledged that there was a potential for hospital bias as the PID cases who were admitted to the hospital were the more seriously ill patients with PID.
Phaosavasdi et al 1975
This was a retrospective case series in which evidence of PID was observed during 200 transvaginal tubal sterilisations carried out at the Chulalongkorn Hospital, Bangkok between October 1972 and June 1973.200 patients selected from the Family Planning Unit at the hospital underwent transvaginal tubal sterilisation. Among the cases, 101 patients were immediate past users of the Lippes Loop (the only device in the study), 73 were past users of other contraceptives and 26 were non-contraceptors. The authors state that no prostitutes or other unspecified categories particularly prone to PID were included among the cases.
PID was diagnosed if at least one of the following were present:
· adhesions
· congestion
· tubal occlusion with clubbing or oozing of pus from the fimbria.
No hysterological examination was performed.There were 33 patients with evidence of pelvic inflammatory change. Among those 28 (of 101) were Lippes Loop users, 3 (of 73) were other contraceptive users and 0 (of 26) were non-contraceptors. This calculated that 84% of those with evidence of pelvic inflammatory diseases were Lippes Loop users, 9% were other contraceptors and 0% were non-contraceptors.
The relative risk estimate for IUD users versus non-users was 9.2 (3.7 - 22.5) ( Senanayake & Kramer 1980 ) :
The relative risk estimate for IUD users versus non-contraceptors was not reported.
The authors stated that the observations contained in this study were very preliminary and there was a need for further studies on larger groups of women.
Senanayake & Kramer 1980 noted that this retrospective (self-selected) cohort involved a non-specific definition of previous or current PID which was diagnosed visually at sterilization. There was a reported prevalence rate of 15% of PID in the cohort whose method of selection (other than being sterilization patients) was not clear:
Combined Bangkok Hospital Group 1975
This study involved an examination of unsuspected pelvic infection in 1583 healthy, previously fertile Thai women undergoing interval sterilization over a 5 month period in 1974 at 4 Bangkok hospitals:Cases were taken consecutively. The diagnosis of PID depended on surgical putative evidence of past PID including adhesions, hydrosalpinx, adnexitis, uterine adhesions or enlarged ovaries.
Post-partum sterilization cases were excluded to ensure adequate visualization. The pathological examination of tissue was only undertaken in a minority of cases and there were no pre-operative cervical smears.
Associated demographic information was collected but no information about prior pelvic infection or sexual history, other than contraceptive use, was obtained nor was there any differentiation between the types of devices used.
60% of subjects were sterilized by laparoscopy, 19% by laparotomy and 16% by culdoscopy.
Among the 1583 subjects, 86.5% had used contraception and 71% were current contraceptors, 13.5% had never used any form of contraception and 29% were non-current contraceptors.
At sterilization, 10% had putative evidence of unsuspected past PID.
The authors then focussed on the largest group with putative evidence, namely, 120 cases involving tubal adhesions.
Despite absence of detail in the report, Edelman 1980 concluded that 8.9% of cases with tubal adhesions had never used a contraceptive method, 12.3% had used the IUD only and 4.9% had used the pill only.
The relative risk estimate for IUDs versus non-users was 2.1 (1.3-3.3).
The relative risk estimate for IUD users versus non-contraceptors was 1.4 (0.8-2.5) ( Senanayake & Kramer 1980 ) :
The authors proposed a risk gradient which showed that IUD only users had the highest chance of tubal adhesion, pill only users had the lowest chance and users of both methods had an intermediate level of chance of tubal adhesion.
The authors suggested that there was a high risk of developing adhesions or hydrosalpinx among women who did not wear or had not worn the IUDs very long flattening out to a low and reasonably constant risk over longer wearing periods.
The authors urged caution in relation to the results from this self-selected population because of the perceived difference between women seeking sterilization and women abandoning a non-permanent method. The reasons for their discontinuances were not dealt with.
The authors also noted that, given the population would never again test their fecundity, it was not possible to say whether the lesions would have compromised their ability to become pregnant.
Faulkner & Ory 1976
This was a case-control study which was designed to investigate the relationship between PID and IUDs.
Following Wright & Laemmle 1968 and Noonan & Adams 1974 , this study was also conducted at Grady Memorial Hospital, Atlanta, Georgia amongst a population of substantially black women.
Cases were selected from women seen in the emergency room of the Hospital.
50 febrile PID cases were identified. These were women who:
(a) were seen in the emergency room with a complaint of abdominal
pain for discomfort;
(b) were confirmed to have an oral temperature of 38C or more;
(c) had uterine or adnexal tenderness on pelvic examination; and
(d) had no diagnosis other than PID to explain this illness either at the time of the initial visit or within one month.100 cases of afebrile PID were also identified. These women did not have confirmed oral temperatures as high as 38C.
200 women were selected as controls from women who:
(a) were seen in the same emergency area;
(b) had an initial complaint other than abdominal pain or discomfort;
and
(c) did not have this complaint diagnosed as a gynaecologic problem
either at the time of the initial examination or within 1 month.Distribution of IUDs among cases and controls appear by type in Table 4 of the article. Only 1 Gravigard was used by 1 control. The balance of the IUDs were primarily Lippes Loops and Dalkon Shields. The Dalkon Shield was found in a higher proportion of febrile cases than the controls.
The relative risk estimates for IUD users versus non users was 5.1 (2.5 - 10.5).
The relative risk estimate for IUD users versus non contraceptors was 3.7 (1.7 - 8.2).
The authors concluded that there was no association between the length of IUD use and the risk of developing PID or febrile or afebrile cases.
The numbers in this study are comparatively small. Among the 50 febrile PID cases, 19 women were using an IUD. This can be compares with 17 women who were using no method of contraception and 10 women who were using an oral contraceptive.
Among the 100 afebrile PID cases, 24 women were using an IUD. This compares with 48 women who were using no method of contraception and 23 women who were using an oral contraceptive.
Westrom et al 1976
This was a case-control study which examined the frequency of IUD use among women with acute salpingitis who had been laparoscopically verified against the frequency of age matched controls selected from patients seen in the University Hospital, Lund, Sweden between January 1971 and June 1975:
In the study period, 515 cases of acute salpingitis which had been laparoscopically verified were treated in the Department of Obstetrics and Gynaecology at the University Hospital.
1545 potential controls were matched for age from the Lund regional register and contacted by postal questionnaire. Currently pregnant and women using no contraception were excluded leaving 741 contracepting controls.
The authors reported that in the period up to August 1972, IUDs available in Sweden consisted of Lippes Loops and Saf-T-Coils. After August 1972, the Copper-T 200 and Gravigard were used almost exclusively. There was no precise analysis of cases and controls by reference to the type of device used. The authors state that there had been a significant increase in the frequency of IUD use per se both among patients and controls in the period 1971 to 1975.
125 of the 515 patients were IUD users compared with 72 of 741 controls. The higher rate of usage of IUDs was most pronounced in women who were less than 20 years of age and who had never been pregnant.
The authors concluded that if the relative risk of acute salpingitis in women not using IUDs (but still contracepting) is 1, then the relative risk for IUD users was 3.1 for all women, 6.9 for those who had never been pregnant and 1.7 for those women who had previously been pregnant.
Senanayake & Kramer 1980 , calculated the relative risk estimate in IUD users versus all other contraceptors as 2.9 (2.1-4.0):
The relative risk estimate in IUD users versus non-contraceptors was not ascertained.
The authors considered that a higher proportion of sexually promiscuous women amongst the cases than among the controls could possibly explain the differences shown but there was no basis reported for asserting that there was in fact any such difference between the two groups.
Vessey et al 1976
This was the Interim Report of a long term follow-up prospective study of British women using different methods of contraception seen by the Oxford Family Planning Association which commenced in 1968:To be eligible for admission to the study a woman had to be married, aged 25-39, a white British subject, willing to participate, and either a current user of oral contraceptives of at least 5 months standing or a current user of the diaphragm or an IUD of at least 5 months standing without prior exposure to oral contraceptives.
By the end of July 1974, when recruiting ended, 17,032 subjects had been admitted. 56.6% were oral contraceptive users, 24.8% were diaphragm users and 18.6% were IUD users.
Cases of PID were classified as acute (surgically confirmed) definite but not acute (not surgically confirmed) and other (inconclusive clinical diagnosis without confirmation).
On admission to the study 0.28% of women reported a past occurrence of salpingitis or oophoritis. There were no statistically significant differences between oral, diaphragm and IUD users in this group.
The report included almost 10,000 woman-years observation of IUD users. 58% used the Lippes Loop; 29% used the Saf-T-Coil and 13% used an unknown IUD.
Table 19 in the Interim Report records results for salpingitis, PID and endometritis as a standardized first event rate (per 1000 woman-years) according to method of contraception actually used during follow-up among subjects with no past history of salpingitis on admission.
The relative risk estimate for IUD users versus all other contraceptors was 3.5 (2.1-6.1) ( Senanayake & Kramer 1980 p.855 ) :
The study did not include non-contraceptors. There was no relative risk estimate in IUD users versus non-contraceptors reported.
Further analysis of the Oxford Family Planning Study resulted in papers by Vessey et al in 1981 (not tendered) and a reanalysis by Buchan et al 1990 .
Kessel 1989 reported that the 1981 paper by Vessey et al summarized the findings of the Oxford Family Planning Study up to 1979. By that time data from 20,482 woman years of use for all devices had been obtained and the rate was 1.51 per 1000 woman-years. The data revealed a PID rate of 1.15 per 1000 woman-years in Lippes Loop users, 2.09 per 1000 women years in Saf-T-Coil users, 8.09 per women years in Dalkon Shield users and 0.90 per women years in Cu-7 users.
Although the Dalkon Shield results were significantly higher, this outcome was based on only 3 cases in a study where all non-users were current contraceptors.
The reanalysis by Buchan et al 1990 was published 14 years after the Interim Report. The authors noted the influence of the WHO Scientific Group Technical Report ( WHO 1987 ) which had identified a number of what were said to be significant epidemiological flaws in earlier studies, including previous reports from the Oxford Family Planning Study:
Using multiple regression analysis the authors subjected the data to control for age, parity, social class, cigarette smoking and age at first marriage.
Parity was excluded (there were no nullips).
Having examined the remaining associations without reference to the influence of contraceptive methods, it was decided to allow for the influence of age, social class, cigarette smoking and age at first marriage. These results are reported in Tables I to IV of the study.
First hospital referral rates for acute, definite and other pelvic inflammatory disease in relation to current use of various contraceptive methods are given in Tables V-VII.
IUD users were divided into non-medicated devices (Lippes Loop, Saf-T-Coils and Dalkon Shields) versus the more modern medicated devices (Cu-7s and Copper-Ts).
The reference group was taken as women who used any other method of contraception. This would have included a large number of oral contraceptive users.
Table VIII records the first hospital referral rates per 1000 woman years for all pelvic inflammatory disease by contraceptive method. Non-medicated IUDs gave rise to a relative risk versus non-users of 3.3 (2.3-5.0). Medicated IUD users had a relative risk estimate versus non-users of 1.8 (0.8-4.0), which was said to be not statistically significant.
Table VIII also confirmed that the protective effect of oral contraceptives is statistically significant. The authors indicated that there was then considerable evidence that barrier methods and female sterilisation are also protective.
Among current users of an IUD who have had previous exposure to a Dalkon Shield, the relative risk of hospital referral for PID was estimated as 4.7 (2.1-9.0) compared to current users of IUDs who had had no exposure to a Dalkon Shield. Use of the Dalkon Shield appeared to be a greater risk than other inert and medicated devices.
The authors aligned their finding in relation to the relatively lower rate of referral for PID in users of modern medicated devices by comparison with older inert devices with the results of the two leading studies on tubal infertility which had reported lower risks of tubal occlusion in users of copper containing devices than in users of unmedicated devices ( Daling et al 1985, Cramer et al 1985 ).
It was submitted on behalf of the defendants that the data set did not allow an effective study of the relationship between the onset of PID and the duration and use of IUDs as was possible in the other significant data set, the Women’s Health Study.
Eschenbach et al 1977
This study involved the case-control matching of patients with gonococcal and non-gonococcal PID against control subjects without PID who had uncomplicated gonorrhoea or no gonorrhoea in order to elucidate risk factors which contribute to infection of the Fallopian tubes and pelvic peritoneum by bacteria which normally remain confined to the cervix and vagina.204 women with a diagnosis of acute PID were recruited between July 1972 and February 1973 (102 consecutive women) and between January 1974 and July 1974 (102 consecutive women) at University of Washington affiliated hospitals, were matched against 308 consecutive sexually active women without PID who had entered a Seattle VD clinic between February 1972 and December 1972 and 399 consecutive sexually active women without PID or contraceptive related complaints who had attended the gynaecology clinics at the University of Washington Hospital and Harbourview Medical Centre between February 1973 and July 1974.
The clinical diagnosis of acute PID required the patient to have had:
· pain in the lower abdomen for less than or equal to 3 weeks
duration,
· abdominal tenderness and,
· on pelvic examination, moderate to severe tenderness maximally
localized to one or both adnexal areas with involuntary guarding
proportional to the degree of tenderness.Diagnosis of PID was only laparoscopically confirmed in 80 of 204 cases. Of these 80, salpingitis was diagnosed in 18 of 20 women who subsequently underwent laparoscopy or laparotomy because they presented with a clinical course which was atypical.
Each PID patient was matched with a patient in the concurrent comparison group by age, race, marital status, presence or absence of gonococcal infection, parity and number of sex partners within the past 6 months.
Among the cases (204 patients), 30% were using an IUD; 23% were using oral contraceptives; 4% were using mechanical (barrier) methods; 1.5% had undergone tubal ligation; 0.5% were using the rhythm method and 41% were non-contraceptors.
Among the combined controls (707 patients) 15% were using an IUD; 39% were using oral contraceptives; 8% were using mechanical (barrier) methods; 5% had undergone tubal ligation; 1% had had a hysterectomy; 1% were using the rhythm method and 30% were non-contraceptors.
Case-control matching for the analysis of contraceptive methods produced 173 matched pairs (Table II). In these pairs, 29% of cases versus 11% of controls used an IUD. Oral contraceptive use was lower amongst PID patients (24%) than among the matched controls (42%).
In these 173 matched pairs, there were 40 pairs in which the PID patient was using an IUD but the control was not. In 9 pairs, the PID patient was not using an IUD but the control was. This ratio, expressed as a relative risk, of 4.44 was statistically significant.
The analysis was then further limited by exclusion of women who had undergone tubal ligation or hysterectomy and all non-contraceptors. Among the balance, 50 of 99 PID patients (50%) and 19 of 114 controls (17%) were using IUDs.
At 95% confidence, the relative risk estimate for IUD users versus non-users was calculated at 4.4 (3.0-6.9).
The authors concluded that PID was attributable to use of an IUD in the case of 77% of IUD users.
The relative risk estimate for IUD users versus non-contraceptors calculated by Senanayake & Kramer 1980 as 4.0. (No confidence interval was available.) P124 page 855.
Among IUD users, 70% used Dalkon Shields, 18% used Lippes Loops, 11% used the Saf-T-Coil and there was only one case who used the Cu-7.
No type of IUD was specifically associated with a disproportionate risk of PID, although the number of IUDs other than the Dalkon Shield were probably too small to permit a statistical comparison.
It was submitted on behalf of the defendants that this paper is significantly at odds with other studies for the following reasons:
(a) The study included a disproportionate number of Dalkon Shield
users by comparison with only one identified Cu-7.
(b) The authors found no increased risk of PID with an increased frequency of intercourse or number of sexual partners.
(c) The highest relative risk was in the higher socio-economic group,
the lowest relative risk was in the lower socio-economic group.
(d) The IUD was usually in place several months before the development of PID.
(e) Fever was less common in IUD cases.Thaler 1978
Edelman 1980, Senanayake & Kramer 1980 and Kessel 1989 all refer to a paper by Thaler et al entitled " Intrauterine device and pelvic inflammatory disease " published in the International Journal of Fertility in 1978 (not tendered).Kessel 1989 indicates that the Thaler study took place in Haifa, Israel, between 1969 and 1970:
Edelman 1980 stated that the Thaler study was a case-control study in which 101 cases, not including women with recurrent PID, were matched against 101 controls for age, current marital status and interval since last pregnancy termination but that the type of IUDs were not stated.
The study was said to have reported a relative risk estimate for IUD users versus non-users of 2.3 (0.9-5.9).
Senanayake & Kramer 1980 point out that the population from which the controls for the study were drawn was unspecified and that the study was limited to first event acute PID only.
Flesh et al 1979
This was a case-control study conducted in a hospital clinic serving an indigent population in Central and East Los Angeles.The authors asserted that recent studies had used inappropriate control groups, (ie women using methods of contraception which reduced the risk of developing acute salpingitis). They asserted that it was therefore not known whether IUDs increase the chance of developing acute salpingitis compared to non contraception. However, they did not take any alleged protection effect of using certain contraceptors into account in their analysis.
The study initially involved 163 consecutive cases who were women hospitalized with a clinical admission diagnosis of acute salpingitis. These were ultimately reduced to 146.
There was no written diagnostic criteria for hospitalising patients with salpingitis but those hospitalised were generally those with physical signs of acute pelvic peritonitis. All of the study group had uterine or adnexal tenderness and a temperature greater than 99.8 on admission.
There were 222 controls who were ultimately reduced to 199. They were women presenting to the minor trauma section of the emergency room for treatment for minor lacerations, burns, fractures and other injuries..
Four variables were analysed:
· black versus non black;
· IUD versus no IUD;
· multiple sex partners versus one sex partner; and
· previous salpingitis versus no previous salpingitis.The authors indicated that the control group in the study was not perfect but concluded that their results demonstrated that the number of sexual partners in the previous 12 months was a major risk factor for PID. They also warned against women using IUDs who are from a low socio-economic status, who are black or who have had previous salpingitis on the basis that the study only dealt with women from a low socio-economic population and the author did not know what the results would be in other socio-economic populations.
There was no information given as to the type of IUDs used nor as to the duration of use. Relative risk estimates were not calculated in the paper but Senanayake & Kramer 1980 estimated the relative risk of IUDs versus non users in this study to be 2.2 (1.3 - 3.9):
Maqueo et al 1979
This is the report of a study of the histology of the fallopian tubes in asymptomatic women who had used an IUD for at least one and usually two years by comparison with asymptomatic post-partum women and normally menstruating women scheduled for surgical sterilisation.Among 225 patients, 75 had had an inert IUD in place for at least one year, 75 underwent a salpingectomy 2-3 days post-partum and 75 were normally menstruating women (controls) who had an elective salpingectomy at various stages of their cycles.
On histological examination, there was no difference in the frequency of oedema and congestion in the tissue of control subjects versus those using an IUD whereas there was a higher frequency of this alteration in post partum subjects which was statistically significant.
An analysis of the intensity of acute inflammatory infiltrate (patchy/moderate/marked) does not indicate any statistically significant difference between the distribution in control patients (non-contraceptors) and that observed in IUD or post-partum subjects.
The authors refer to the histological observation reported by the Combined Bangkok Hospital Group (1975 ) where only one of 39 subjects with external adhesive disease showed histological evidence of salpingitis.
The authors also refute the proposition by Smith and Soderstrom that salpingitis " similar to that found in the endometrium " and, by inference, possibly an extension thereof, is a concomitant of IUD use.
The IUD device types were inert but were not otherwise specified.
Senanayake & Kramer 1980 calculated a relative risk estimate for IUD users versus non-contraceptors of 1.5 (0.8-2.8). Senanayake & Kramer stated that the small number of subjects, the absence of any IUD users with a recently inserted device and a non-specific case definition for PID limited the use of this study:
Ryden et al 1979
This study involved an analysis of the influence of different contraceptive techniques on the incidence of PID in 672 female patients with gonorrhoea:The lowest frequency of PID was found in patients using hormonal contraceptors (8.8%). This compared with 23.5% of patients using IUDs and 15.1% of patients using neither technique.
Between 1973 and 1977, 744 women were treated for uro-genital gonorrhoea at the University Hospital in Linköping, Sweden. 72 women were excluded due to pregnancy, sterilisation, being post-menopausal, as non- contraceptors or because of doubtful diagnosis of PID.
Among the remaining 672 patients, PID was diagnosed in 87 (12.9%).
PID was confirmed by laparoscopy. In all cases the diagnosis required a reddened serosa of the tubes, oedema of the tubal wall or purulent discharge from the abdominal orifice of the tube.
Each case was then matched against 2 uncomplicated gonorrhoea patients.
All of the IUD users had used copper devices with the exception of some isolated instances of use of plastic devices. The authors determined that it was not possible to evaluate any difference in risk of PID in patients using copper or plastic devices due to the small number of plastic devices.
Among the PID cases, 30 used hormonal contraceptives, 20 used IUDs and 37 used neither hormonal contraception nor an IUD.
Among the controls, 90 used hormonal contraceptives, 20 used IUDs and 64 used neither.
Each group was examined by age, marital status and the number of sexual contacts (1, 2, 3, 4 or more).
By using multiple regression analysis, the authors concluded that, apart from birth control methods, no other parameter (age, marital status, number of sexual contacts) influenced the incidence of PID in the patients.
The relative risk estimate for IUD users versus non-users was calculated by Senanayake & Kramer 1980 at 2.3 (1.2-4.6) which difference is said by the authors to be statistically significant:
The relative risk for IUD users versus non-contraceptors was
1.7 (0.8-3.6).The authors found that a statistical analysis demonstrated a significantly lower incidence of PID amongst hormonal contraceptive users than IUD users and users of neither method. However, it was said there was no significant difference found between IUD users and patients using neither hormonal contraceptives nor IUDs. Cervical gonorrhoea was also more common in women with PID than in women with uncomplicated gonorrhoea.
The authors considered that the results obtained in the present study, with a higher incidence of PID in the IUD group; and a lower incidence in the hormonal contraceptive group compared to patients using neither method, were related to the contraceptive method per se.
Kaufman et al 1980
This was a hospital-based case-control study of the relationship between IUD use and first episodes of PID amongst sexually active women. The effects of duration of use of particular IUDs were also investigated.There were 44 patients and 259 hospital control subjects. All of the women used IUDs or oral contraceptives. The age adjusted relative risk for those using IUDs at the time of admission was 6.5 (95% confidence material 3.2 to 13.0).
The records of 185 pre-menopausal women with a discharge diagnosis of PID, salpingitis or pelvic abscess at hospitals in 7 metropolitan areas in the United States, one hospital in Canada and two in Israel were reviewed to identify cases adjudged to be first episodes of PID. 120 such cases were identified.
From 4891 pre-menopausal women available as potential controls, 2058 were selected as suitable on the basis of having a first diagnosis judged not to be related to contraceptive use. Women with gynaecological disorders were excluded as were women who had recently undergone an abortion or suffered an ectopic pregnancy. The analysis was then confined to women who were presumed to be sexually active, by virtue of their use of IUDs or other contraceptives. All cases and controls were less than 45 years of age.
The study reported on a final analysis of 44 subjects and 259 control subjects.
Among the subjects, half of all diagnoses were based on laparotomy, laparoscopy or observation of pus emerging from the cervix. The balance were diagnosed on the basis of clinical evidence such as fever and adnexal tenderness.
Table IV of the report sets out the relative risk of PID amongst IUD users by reference to specific devices. The referent group were oral contraceptive users rather than non-contraceptors.
The relative risk estimate for Cu-7 users versus non-users was 3.8 (9 of 44 cases versus 29 of 259 controls). The relative risk for Dalkon Shield users was 12.3 (5 of 11 versus 5 of 259). Lippes Loop users had a relative risk of 7.9 (11 of 44 versus 17 of 259). For users of the Saf-T-Coil, the relative risk was 9.2 (3 of 44 versus 4 of 259). The referent group consisted of 14 of 44 cases versus 172 of 259 controls.
Senanayake & Kramer 1980 calculated an overall relative risk estimate for IUD users versus non-users of 7.9 (3.7-16.9) and expressed their view that the limitation of the study population to only sexually active women who used either IUDs or oral contraceptives meant that no other comparison group was included, producing a spuriously low relative risk for oral contraceptive users and a spuriously elevated risk for IUD users.
In 1983 Kaufman et al published a further report (not tendered). According to Kessel 1989 , this further report updated the study period to March 1981 with a total subject population of 460. The relative risk for IUD users versus non-users reported was said to be 4.7 (3.1-7.1) which, when adjusted for the use of oral contraceptives, was 3.2 (1.6-6.6). However, the relative risk for users of the Dalkon Shield versus non-users was 78.7 (22.7-272).
Osser et al 1980
This was a study conducted in Malmo, Sweden to investigate whether PID was more common amongst IUD users than in a sexually active age matched control group and, if so, whether it is more common amongst those who have not previously been pregnant:The authors concluded that the risk of PID was raised twofold by IUD use and that there was no significant difference between women who had never been pregnant and those who had been pregnant.
The records of 720 patients admitted to the Department of Obstetrics and Gynaecology at the General Hospital in Malmo between 1973 and 1975 were traced and all relevant gynaecological and obstetric notes were extracted.
Age matched controls were then selected by questionnaire in relation to the month that the matched case was in hospital for PID. Apart from questions concerning pregnancy and health, each potential control was queried about her use of any method of contraception in that month, the number of sexual partners she had had and how often she had had intercourse during that month. Women who were not apparently sexually active were excluded. By this method, 690 of the 720 patients were matched with one control each and were included in the study..
A diagnosis of PID was confirmed by laparoscopy in 86.1% of cases.
The authors asserted that a complete statistical analysis of the frequency and distribution of PID in the balance of patients gave the same frequency of IUD use and the same risks in the different groups.
Among the patients, 31.9% were IUD users, 23.0% were users of oral contraceptives and 45.1% used other methods or none. Among the controls, 16.5% were IUD users, 47.1% used oral contraceptives and 36.5% used other or none. The relative risk estimate for IUD users versus non-users was said to be 2.1 (1.6-2.7). Senanayake & Kramer 1980 calculated a slightly different risk of 2.0 (1.6-2.6).
The type of IUDs included in the study were not specified.
These authors found a relative risk of PID for parous women of 2.6 compared with that for nulliparous women of 1.6. This difference between parous and nulliparous women contrasts with results reported in earlier studies but it does not appear to be statistically significant.
The authors suggested that whereas Professor Westrom had considered the mere use of a contraceptive as proof of sexual activity and excluded women not using contraceptives, this study had made greater efforts to ensure that all controls were sexually active.
As in Westrom et al 1976 , this study reported more oral contracepting controls than patients and that there appeared to be a significant protective effect which reduced the relative risk of PID by two thirds.
In reviewing this paper, Senanayake and Kramer 1980 calculated the relative risk for IUD use compared to no contraceptive use as 1.6 (1.2-2.1), which is the lowest statistically significant increase in risk shown in any of the studies reviewed by them.
Paavonen & Vesterinen 1980
This study sought to evaluate the contraceptive practices of hospitalized patients with clinically verified acute salpingitis, in particular the frequency of IUD use, compared with that of controls for asymptomatic partners of men suffering from non-gonococcal urethritis.The control group was chosen by the authors because of their concern that prior studies (cited in Table I of the study) may have included sexual exposure differences between cases and controls making it difficult to define the actual risk of salpingitis in IUD users.
The patient series consisted of 144 women treated for symptoms of acute PID at the University Central Hospital in Helsinki, Finland between October 1977 and August 1978.
Clinical diagnosis of acute salpingitis was based on:
· pelvic pain of short duration,
· increased vaginal discharge,
· tender adnexal masses,
· increased ESR (greater than 15) and usually fever.All patients were hospitalized because of severe symptoms.
The control group consisted of 240 consecutive women below 50 years of age who were examined at the same outpatient clinic during the same period of time and was limited to women whose partners were suffering from non-gonococcal urethritis diagnosed within the same hospital. Following the exclusion of pregnant women, 229 of these 240 women were included in the study.
The patients provided a contraceptive history by answering a standard questionnaire. They were divided between IUD use, oral contraceptive use and use of other methods or none at all. The third category was selected to represent the basal level of disease when calculating relative risks for salpingitis.
The types of IUD in the study were not specified however copper devices were included.
Among these patients, 58% were IUD users, 11% were oral contraceptive users and 31% used other methods or none at all. Among the controls, 32% were IUD users, 30% were oral contraceptive users and 38% used other methods or none at all.
The authors estimated the relative risk for salpingitis in IUD users versus the referent group, other methods or none at all was 2.06 (1.23-3.46). This appears to be attributable mainly to a higher relative risk 2.20, (1.09-4.43) in nulliparous women. Oral contraceptives were significantly protective with a relative risk of 0.37 (0.19 - 0.73).
Table IV of the article shows the relative risk in parous women as 1.92 but the confidence interval crossed unity. In nulliparous women, the relative risk was statistically significant but the lower confidence interval was only just above unity.
The referent group (other method or none at all) included users of barrier methods of contraception which, themselves, are protective against the transmission of STDs.
Burkman et al 1981
This is the first of 3 papers (Burkman et al 1981, Lee et al 1981 and Lee et al 1988 ) which analysed data collected through the Women’s Health Study. This was a concurrent case-control study conducted at 16 hospitals in 9 cities throughout the United States. It was designed to examine the relationship between the use of an IUD and PID severe enough to require hospitalization:Data collection for the study commenced in October 1976 and concluded in 1978.
In attempting to detect whether the use of an IUD resulted in serious complications which required hospitalisation, the Women’s Health Study examined uterine perforation, PID, vaginal haemorrhage, ectopic pregnancy, foetal loss, abruptio placentae and placenta previa. Burkman et al 1981 and the subsequent reports examined the findings of the Women’s Health Study regarding PID.
Subjects in the study were between 18 and 44 years of age, sexually active, not surgically sterile (and the partner not surgically sterile) and menstruating within the last year. Women who had been pregnant in the previous 6 weeks were excluded. In addition, women with a hospital discharge diagnosis of epilepsy, thromboembolism, cerebrovascular disease or cancer were excluded.
Cases were confined to patients who required hospitalization with a diagnosis of PID which diagnosis had been confirmed on discharge.
Patients were presumed to have an infection of the upper genital tract if they had:
· low abdominal pain, fever
· adnexal pain on examination
· adnexal masses and
· vaginal discharge.PID cases were then sub-categorized as:
· certain (visualization of the adnexa or the draining of an
abscessed cavity)
· moderately certain (hospitalization with acute symptoms but no
surgery), and
· less certain (discharge diagnosis of PID without surgery and not
acute).From 2761 potential PID cases, 1447 patients were finally classified as having PID and were included in the study.
Controls were drawn from women admitted to the same hospital during the same period but excluding women admitted for several diseases that might influence contraceptive practice. Both surgical and medical controls were admitted. However, in the absence of any discernible difference, the two groups were combined in the analysis. While controls were not matched to cases, they were parallelled by hospital and time of admission.
From 7199 potential controls, 3453 women were included in the study as controls.
Age, race, gravidity, parity, sexual activity and number of sexual partners were all considered potentially confounding factors and were analysed using a multi-variate analysis to determine whether IUD use had any effect in the presence of other risk factors.
In order to determine contraceptive method women who had used a method during the 30 days prior to hospital admission or, for IUD users, had the IUD present at the time of admission or had had it removed within 30 days prior to admission were all included as current users of the various contraceptive methods.
An analysis of past users and women who had never used contraception produced little discernible difference and accordingly those two groups were combined and labelled as non-users.
The study was designed to ensure that the non-contracepting group did not include women who very recently ceased using methods known to be relatively protective against PID, such as oral contraceptives.
Among cases, 22% were IUD users, 15% oral contraceptive users, 8% barrier users, 5% used other methods and 50% were non-contraceptors. Among controls, 15% were IUD users, 29% were oral contraceptive users, 13% were barrier users, 15% used other methods and 28% were non-contraceptors.
27% of cases versus 16% of controls had had more than one sex partner in the last six months.
The relative risk estimate for IUD users versus non-users was 1.6 (1.4-1.9). In patients with a PID history, this relative risk was 1.2 (1.0-1.6) whereas patients without a history of PID had a higher relative risk of 2.1 (1.8-2.5). It was submitted that this report of an increased relative risk for patients without a history of PID was anomalous given earlier findings and argues against the proposition that previous PID compromises the host tissue against the threat of future episodes.
Table VI on page 274 of the article provides a ranking of selected risk factors for PID among women without a prior PID history. The highest relative risks were associated with a woman’s number of sexual partners and potential for exposure to STDs. Other significant relevant risk factors included women who were black and under the age of 25. These relative risks were all greater than current use of an IUD.
There were no major differences between nulligravid and gravid women with or without adjustment for age, race, sexual activity and number of partners.
Lee et al 1983
In this analysis 657 cases and 2566 controls were identified in the Women’s Health Study and data was analysed to attempt to determine the PID risks associated with individual IUD types.
The risk of PID by different contraceptive methods compared with women who used no contraceptive method was analysed. The results appear in Table 1 of the article. Oral contraceptives and barrier methods were found to be protective against PID whereas IUDs were found to have a relative risk of 1.9.
The next matter analysed was the risk of PID associated with the various types of IUDs. This had not been done before. The results are set out in Table 2 of the article. Compared with users of other types of IUDs, Dalkon Shield users had 5 times the risk of pelvic inflammatory disease. Cu-7 users had a relative risk of 1.9.
The analysis continued by excluding the Dalkon Shield users. It was then found that the increased risk of PID was greater in the first four months after insertion than after that time. The highest PID risk was within the first month, a point estimate of 3.8. However, after the first four months, the relative risk was 1.1. These results are set out in Table 4 of the article.
The study then further analysed the effect of duration of use by specific IUD types. The relative risk for Cu-7 users in the first four months after insertion was 3.3 and after four months was 1.3. Both these figures are higher than for the Lippes Loop whose figures were 2.5 and 1.0.
The defendants submitted that the main finding of Dr Lee’s 1983 analysis is that the risk of PID from all IUDs was double that of using no contraception. However, part of this risk is due to the Dalkon Shield because it had five times greater risk than other IUD types. Once the Dalkon Shield users were excluded from the analysis, it was submitted the increased risk was confined to the first four months after insertion. This submission must be considered in the light of the figures set out above.
It was submitted that the duration of use results reported by this study represented a significant advance in state-of-art knowledge about iatrogenic PID. There had been previous reports of PID incidence rates over duration of use, but they had provided only limited information.
Witoonpanich et al 1984
This was the report of a multinational case-control study of PID in relation to contraceptive use initiated through the World Health Organisation Special Program of Research Development and Research Training in Human Reproduction and conducted concurrently in 12 centres located in 10 developed and developing countries worldwide.The objective was to compare characteristics of patients with acute febrile episodes of moderate or severe PID with the characteristics of controls and, in doing so, to take into account the extent of PID associated with contraception in developing countries:
Cases included women aged 15-44 with acute PID treated as inpatients or outpatients between March 1978 and December 1979.
A definitive diagnosis of PID required:
· a temperature of 38C or higher within 24 hours of presentation
suprapubic tenderness with guarding on physical examination
and,
· in addition to these mandatory signs, on vaginal examination
there should have been at least tenderness on movement of the
cervix.
· adnexal tenderness or a palpable adnexal mass.
· Operative diagnosis was acceptable and, where possible,
bacteriology and histopathology was performed.Controls were selected from non-gynaecological non-obstetrical hospital admissions or outpatients in a six month period before or after the identification of each case. Controls were then individually matched (2 for 1) by age band, parity, marital status and by in or out patient status. Controls were also matched by post-abortal or post-partum status.
Subjects were questioned as to this relevant medical history and use of current and previous contraceptive types. However, due to cultural sensitivity in some centres, the study did not obtain a history of sexual activity and matching for marital status was used as a proxy. It was submitted that in light of the findings from the Women's Health Study, the use of this proxy introduced a problematic bias.
There were 968 cases and 1936 controls identified. From this, 862 initial case-control sets included 608 sets of women who were either nulliparous or had a pregnancy which terminated more than 6 weeks prior to interview were studied.
Types of devices were not differentiated in the study.
The relative risk estimate of the first episode of PID in IUD users compared with non-contraceptors reported ranged from 11.5 (3.6-36.2) in current nulliparous IUD users in developed countries to 4.1 (1.1-15.1) in current parous users in developed countries. When differentiating between first episodes and recurrent PID and between nulliparous and parous women, the study produced increased relative risks associated with IUD use ranging between low and statistically insignificant risks to high risks with very wide confidence intervals.
The authors stated that the inability to properly assess sexual activity would introduce serious bias if a higher proportion of cases were generally active compared with the controls.
Cramer et al 1985
The study was directed by Dr Cramer of Harvard University. It was a case-control study conducted at 7 different centres across the North American continent designed to investigate tubal infertility and IUD use. Data was collected between 1981 and 1983. Over 2,000 potential cases were interviewed and over 4,000 controls were identified.
Eventually 388 women with tubal disease confirmed by HSG or laparoscopy were identified and used in the analysis. The profiles of the cases and controls are set out in Table 1 of the article.
The study concluded inter alia:
1. The adjusted risk of primary tubal infertility associated with any
IUD use before a first time birth was 2.0 (95% confidence limits
1.5 to 2.6) relative to non-use.
2. Users of copper IUDs had a risk of 1.6 (1.1 to 2.4).
3. Women who reported having only one sexual partner had no
increased rate of primary tubal infertility associated with IUD use.
4. Tubal infertility is associated with IUD use but less so with
copper IUDs.Cramer stated:
“In general, the use of an IUD by women with a history
of a genital or pelvic infection doubled their baseline risk. As important, even women without a history of infection at the time of IUD insertion had a significantly increased risk of tubal infertility from the IUD. Thus, absence of prior infection cannot be used as an indicator that the use of an IUD by a nulliparous woman will be safe.” (p.946, RHC, middle para).The plaintiffs submitted that the duration of IUD use (Table 3 of the article) showed an elevated risk for the first 3 months and a more gradual increase thereafter. They also submitted that these results are contrary to the defendants’ submission that the increased risk of PID is limited to iatrogenic risk.
The defendants submit that the plaintiffs’ submission is erroneous for the following reasons:
First, the duration of use estimates are for all IUDs including the Dalkon Shield which the authors acknowledge artificially elevated the relative risk estimates.Second, the study was designed to look at tubal infertility as opposed to episodes of PID. Accordingly, duration of use data cannot provide any insight concerning the impact of iatrogenic risk. The cases in this study were all women who were presenting for tubal infertility. Causal factors for their infertility had all occurred at some unknown time in their past. No data were collected on past episodes of PID or when it had occurred in relation to contraceptive use.
Therefore, it was submitted the duration of use results in Table 3 of the article demonstrate something in the nature of cumulative incidence rates of tubal infertility. It was submitted that tubal infertility is not PID and they are not the same as the relative risk estimates for PID of the kind presented in the Women's Health Study ( Lee et al 1983 Table 5):
It was submitted that if results for IUD users for less than 3 months are reported as opposed to IUD users of more than 3 years, it is inevitable that additional risk factors will be incorporated and additional cases of tubal infertility will be identified. More events happen as more time elapses. Multi-variate analysis attempts to control for some of these known confounding variables but it cannot control for all of them so it was submitted that there was a cumulative incidence effect.
In Table 4 of the article the authors set out their findings of the influence of certain subject characteristics on the risk of primary tubal infertility. They reported that women who had never used an IUD but reported having more than one sexual partner had a higher relative risk of tubal infertility than women who had only one partner and did not use IUDs. Further they reported women who had used a copper IUD and reported having only one sexual partner had no increased risk of tubal infertility. Finally women who reported suffering a minor infection in the past (genital herpes, venereal warts, trichomonas infection or vaginitis other than due to yeast) had an increased relative risk of tubal infertility even if they had not used an IUD. The relative risk estimates and confidence intervals are set out in the table below:
Confidence Interval
Relative Risk
Never used IUD but more than 1 partner1.1 - 2.1
1.5
Copper IUD and 1 partner0.5 - 2.7
1.1
Copper IUD and more than 1 partner1.7 - 4.5
2.8
No IUD use but suffered minor infection in past
1.5
1.0 - 2.3
In Table 3 of the article the relative risk of primary tubal infertility for women using a copper device only was 1.6 (1.1 - 2.4). Such an estimate is similar to the relative risk of primary tubal infertility for women who had never used an IUD before but had either more than one partner or suffered an episode of a minor infection in the past. The defendants submitted that when the relative risk estimate for copper IUD users with one partner (1.1) is compared to those other figures (1.5), the results of the study do not support the plaintiffs' submissions. The authors concluded that the risk of tubal infertility does not apply to women with a single partner at low risk for sexually transmitted diseases and that for women who have had one child copper IUDs may offer a relatively safe, if not risk-free, alternative to other methods.
In relation to the number of sexual partners which a woman had Cramer made the following observation:
“The risk of tubal infertility associated with IUD use was
confined to women with more than one sexual partner. The
usefulness of the number of partners as a predictor variable as defined by this study should not be overstated, however, because it is based on the life-time number of partners reported at interview, rather than on the number of partners by the time of the first use of an IUD. Furthermore, not only the number of partners, but obviously their likelihood of having sexually transmitted diseases, is important.”
(p.946, RHC, middle para).Daling et al 1985
Daling et al 1985 follows Cramer et al 1985 in the same edition of the New England Journal of Medicine. In their introductory remarks, the authors offered a summary of the state of the literature as at 1985 in relation to the increased risk of infertility for IUD users as a result of PID. They said that it was likely that users of an IUD had an increased risk of subsequent involuntary infertility (through their susceptibility to salpingitis) although the data fell short of being conclusive.In her evidence, Professor Daling said that this state-of-art conclusion related not only to the 12 references footnoted in her paper but was an accepted part of general knowledge at the time. ( T2889 line 53 - T2890 line 6).
Daling dealt with Cramer in Ex. P209, Tab 2, para 20:
“The selection out of one non-significant finding among
monogamous women does not accurately depict this
study’s conclusions. Cramer found a significant
increase in risk of infertility among women who use the
Copper IUD, RR=1.9.”Cramer dealt with criticism of his study in a letter to the New England Journal of Medicine on 5 September, 1985 when he said:
“Dr Hasson’s primary arguments may perhaps be
paraphrased as follows: There is only one cause of tubal
infertility, and that cause is infection with sexually transmitted diseases; therefore, IUDs cannot be a cause of tubal
infertility. Even if one assumes that the majority of cases of
tubal infertility are preceded by infection, this does not negate a role for the IUD in potentiating the damage from those
infections. This is clearly illustrated in Table 4 of our original report, which demonstrated that women who had minor
genital or major pelvic infections and who had used an IUD
had greater relative risks of infertility than women who had
had these infections but had not used the IUD”.
(p.637, LHC, second last para).Daling gave evidence consistent with Cramer’s view as to the potentiating effects of IUD use on infections. (T3034, L5-25):
“…A woman who uses an IUD, if she is exposed to those
pathogens it is more likely to exhibit an infection of the
magnitude that rises up into the tube. It is an effect modifier. Given a woman is exposed, her chance of getting an infection that leads to disease who wears an IUD is different than a woman who does not. It is clear from our data, clear from other people’s data, that is truly the case. We call that an effect modifier; something that modifies the effect of an exposure.”Senanayake and Kramer 1980 made similar observations.
(p.859, Ex. P124).
The study was a case-control study with data being collected between 1979 and 1981. It was not a multi-centre study and 159 matched pairs were identified and used in the analysis.
Like Cramer et al 1985 , this report looked at the characteristics of women with primary tubal infertility and matched controls. The results were similar to the Women's Health Study. More cases reported having 5 or more sexual partners. More cases were smokers and were married more often. The Dalkon Shield was associated with the highest relative risk for primary tubal infertility and the duration of use figures indicated a higher relative risk for the shortest period of use tabulated. ( P130 pages 939, 940).
The authors also reported that for those women who had only used a copper IUD the relative risk was 1.3 (0.6 - 3.0). That is not a statistically significant result. The authors reported that women who had used a copper IUD had very little excess risk and said that they regarded the result as reassuring.
Daling reported that use of any IUD in the past increased the risk of primary tubal infertility to 2-6 (confidence 95% CL 1.3 - 5.2). She found that the risk for women who had only used copper IUDs was 3.1 (95% CL 1.5 - 6.5) when she reanalysed her data in 1992.
Westrom 1980
This was in the nature of a review article which also incorporated a report of original data. Professor Westrom was continuing his original research which is referred to in his earlier papers.
This paper attempted an analysis of the incidence, prevalence and trends of acute PID in a range of industrialized countries. Professor Westrom stated that, in the vast majority of cases, salpingitis is caused by STD infections which ascend from the lower to the upper genital tract. He added that, from a public health perspective, any attempt to decrease the incidence of PID must start with attempts to decrease the incidence of STD in general.
The relationship between IUDs and PID was referred to on a number of occasions within this review. Professor Westrom referred to iatrogenic PID and identified IUD insertion as being one of a number of different procedures which involve instrumentation of the uterus with consequent risk of infection.
Professor Westrom stated that, as at 1980, epidemiological studies had revealed that IUDs were used significantly more often by women with PID than by control women. He also noted that age, race, socio-economic status and sexual activity were of importance in differences in the risks of PID in IUD users versus non-users.
Westrom 1980 also included original data from a study population of women in Lund [see Table II of the article]. Professor Westrom noted that this data appeared to show that the use of an IUD increased the risk of acute salpingitis. However, as he noted in his 1976 study, there appeared to be a protective effect provided by oral contraceptives and barrier methods. In Professor Westrom’s view, sexually active women who use no contraceptive method constitute the correct reference group to be used when studying the epidemiology of genital tract infections but he did not perform a further statistical analysis to take this into account.
The plaintiffs had IUDs inserted on the following dates:
(a) Mrs Lee 19 May 1975
(b) Mrs Orders 20 April 1977
(c) Mrs Robertson 1 August 1977
(d) Ms Ottaway 13 September 1977
(e) Mrs Southren 14 February 1978
(f) Mrs Moylan 10 July 1978
(g) Mrs Gentle 3 November 1978
(h) Mrs Denzin November 1978
December 1978
June 1980
November 1981
June 1983
June 1985
(i) Mrs Lane 22 August 1980
The plaintiffs sought the following finding of facts:
“At the time of each of the above insertions Searle was in
possession of strong evidence from epidemiological studies
which pointed to a causal association between IUD use and
PID.”The plaintiffs sought the following finding of fact and law:
“The association, as revealed in these studies was under a
legal obligation to act on the basis of that causal association.”With regard to the state-of-knowledge at that time, Dr Eschenbach said during his oral evidence (T1822, L46-57) that the three studies namely Wright and Laemmle 1968 , Noonan and Adams 1974 , and Targum and Wright 1974 were enough to lead to the conclusion that there was an increased risk of PID from IUD use.
Dr Ory from the Centre for Disease Control said of the six epidemiological studies he reviewed (viz Wright and Laemmle 1968 , Targum and Wright 1974 , Lippes 1975 , Faulkner and Ory 1976 , Westrom et al 1976 , and Vessey 1976 ):
“In summary, then, I think that these six studies point in the same direction, that IUD use is associated with the increased risk of acute PID. The best estimate suggests that IUD users have from three to five times the risk of developing PID compared to non-users, that teenagers appear as likely to be affected as older women, that copper IUDs do not offer any protection from gonococcal PID, and finally, there is the intriguing but as yet unconfirmed suggestion that nulliparous IUD users are more likely to develop acute PID than are multi-parous users.
As an epidemiologist, I kind of look at these six studies, you know, done by five different investigators at three different countries, all showing the same results, and that, to me, provides some evidence that the association is one of cause and effect. Each study had its strengths and its weaknesses. However, all together, they rule out virtually every major source of bias as a possible reason for observing this association.
In addition, when you couple it with Dick Soderstrom’s recent report of the histologic changes and the possibility that this is altering tubal defense mechanisms; also if you couple it with the fact that previously rare syndromes, or so they are said to be in the literature, such as unilateral tubo-ovarian abscess and infective agents like Actinomycoses are now being reported in association with IUD use, I think biologic plausibility is added to a causal interpretation of these six studies” (Exhibit P228, Tab 13, pp. 72-73).Before the second insertion (Mrs Orders 20 April 1977) there was a documented recognition of the risk within Searle itself. On 25 February 1977 Mr Peddie from Searle Australia sought from Dr Inglis in the US Searle's "official position" on the risk of PID and ectopic pregnancy associated with IUD use. Mr Peddie drew attention to then current literature which supported that causal association.
Before the third insertion (1 August 1977 Mrs Robertson) the FDA Panel proceedings took place. Dr O'Brien was the Searle US employee who had answered Mr Peddie's memorandum referred to above. There is no evidence of any corrective measures being taken to inform Mr Peddie of the information from the FDA Panel proceedings.
During the course of the FDA Panel proceedings, there was recognition among the participants that strong evidence supported a 3-5 fold increased risk of PID and its sequelae among IUD users. At several points during these proceedings, specific reference was made to copper devices (mainly the Cu-7) as being implicated. Dr Tyrer represented the Planned Parenthood Federation of America which was described during the proceedings as “undoubtedly the leader in the United States in family planning” (Exhibit P228, Tab 13, p. 60, line 23 ff) with the largest number of clients. She said the Cu-7 was by far the IUD of choice. She said there was sufficient data to document a definite decrease in fecundability after IUD use and the PPFA had prepared a memorandum “that there is a 3-5 fold increase risk of infection among IUD users” (ibid p. 5943, p. 61, L17). These remarks related to all IUD use but in context her remarks clearly implicated the Cu-7 because she wanted the Drug Panel (which was formally concerned with the copper devices) to issue the same warning (ibid p. 178, L12-17). In the period after the FDA Panel proceedings, all of the plaintiffs except Mrs Lee and Mrs Orders had a Copper 7 inserted.
In the period between the FDA Panel meeting on 6 June 1977 and Mrs Lane's insertion on 22 August 1980, Table 1 from Senanayake and Kramer shows publications by Eschenbach 1977 , Thaler 1978 , Flesh , Maqueo and Ryden 1979 , and Kaufman, and Osser 1980 .
This period of time covers all IUD insertions among those seven remaining plaintiffs except for Mrs Denzin's fourth, fifth and sixth insertions in November 1981, June 1983 and June 1985.
The plaintiffs sought a finding that the epidemiological studies showed that an IUD placed a woman at greater risk of PID.
The plaintiffs submitted that the studies referred to and the evidence of Professor Daling lead to the conclusion that the broad sweep of epidemiological evidence leads to the conclusion that use of the Cu-7 increases the risk of PID to the users.
The epidemiological material relied upon included many studies which do not include the Cu-7 and which relate to populations which are significantly different in a sociological sense to any of the plaintiffs.
The plaintiffs seek to draw a conclusion from this disparate material that use of the Cu-7 by the plaintiffs in these proceedings increased the risk of their contracting PID leading to a conclusion of probable causation of the injuries which they sustained.
The defendants on the other hand submitted that it was proper to analyse each study and to ascertain how they fitted into the pattern of knowledge after proper allowance was made for weaknesses which appeared in their methodology. In my view this is the correct approach.
In particular the defendants criticised studies on the bases that they:
(a) include Dalkon Shield users;
(b) the duration of use is not obtained so as to indicate iatrogenic
risk from any longer term risk;
(c) the control group contained women who used a method of
contraception which was protective against PID (e.g. barrier
methods, oral contraceptives);
(d) compounding factors for PID such as age, number of sexual
partners, marital and socio-economic status are not identified
and controlled using multivariate analysis;
(e) biases are not adequately addressed.These factors may, either individually or taken with others, elevate the relative risks for IUD users.
The defendants also submitted that the analyses of the material shows that women in the category of the plaintiffs do not have any statistically significantly higher risk of contracting PID when using copper IUDs.
In addition to lengthy analysis and criticism of the twenty-five publications referred to earlier the defendants relied upon extensive state-of-the-art literature supporting the use of IUDs.
I accept the evidence of Professor Roy as to the risk of infection based on his own experience and on the WHO Report ( Farley 1992 ) that the risk was greatest for the first twenty days after insertion and that thereafter the rate of infection closely paralleled the background rate of sexually transmitted organisms of the population under study. (Ex. D.301, para 41).
Professor Roy also drew attention to the 1983 Women’s Health Study which analysed different degrees of risk of PID witih specific types of IUDs ( Lee et al, 1983) and concluded that there were differences.
Professor Roy also pointed out that earlier studies had compared IUD users with women who used contraceptives which protected against PID and that this comparison falsely elevated the risk or PID for IUD users. Also earlier studies had in many cases included users of Dalkon Shield IUDs which increased the rate of infection shown.
I do not accept Professor Daling’s 1992 recalculation of her 1985 report as displacing the opinions of Professor Roy which I accept.
It follows that with all its manifest imperfection the epidemiological evidence does not establish in my view that there is any statistically significant increase in risk of contracting PID for users of the Cu-7 although there is an increased risk in the first twenty days after insertion which risk is associated with the insertion procedure.
In the light of this finding the findings sought by the plaintiffs on pages 181 and 182 are not relevant as the information which was then available was not in my view reflective of the true situation.
ECTOPIC PREGNANCY
There does not appear to be any issue between the parties as to the existence of a relationship between pelvic inflammatory disease and ectopic pregnancy. There is a dispute however between the parties as to the role of the IUD in this relationship.An ectopic pregnancy is a pregnancy which occurs at a site outside the uterus. Although they may occur elsewhere the vast majority of ectopic pregnancies occur in the fallopian tubes.
The plaintiffs tendered an editorial published by Dr Eschenbach and Dr Daling in the April 1983 edition of the Journal of the American Medical Association (Ex. 209, Tab C) in the following terms:
"The most widely held mechanism has been a partial failure of the ovum to pass down the fallopian tube in the carefully timed sequence that would allow proper uterine implantation. Partial tubal obstruction after tubal infection stands out as one of the most important known causes of ectopic pregnancy. Women with documented acute salpingitis have an ectopic pregnancy rate that is increased 7 fold over women who have had no known salpingitis. Seven percent of the first pregnancies that occur after salpingitis are ectopic. Evidence of prior salpingitis is demonstrated in one third of the women with an ectopic pregnancy at the time of operation" (Exhibit P209, Tab C).Dr Keeping gave the following evidence on behalf of the defendants. He said:
"Q. I am asking doctor about former IUD use?
A. Former now, now moving on to former IUD use and ectopic, whatever the literature says, there would have to be an increase in PID in a population that has had an IUD before. I have acknowledged that there is an insertion risk of PID and I think we got down to a figure that once the IUD is in, there might be a relative risk ratio of one point something to 1 for people who have an IUD in... Having said that, I have acknowledged that there will be some people that will get PID as a result of their IUD, particularly around, mainly around the insertion period. As such, that population must be more at risk of ectopics, so there is - there has to be an increased risk of ectopics in previous IUD users" (T6387 L51 - T6388 L9).The plaintiffs submitted that I should conclude that if an IUD use puts a woman at greater risk of PID, it will put her at greater risk of an ectopic pregnancy.
It was put that where an increased risk of ectopic pregnancy occurs among women who have used an IUD, it follows that they have been at a greater risk of contracting PID from IUD use. Similarly it was put that the tubal infertility studies were used as independent and objective confirmation of past PID, so the ectopic pregnancy studies operate in the same way. Dr Daling's evidence was that the studies of tubal infertility and ectopic pregnancy both avoid referral bias and diagnostic bias in the studies of clinical PID. The WHO Report 1987 makes the same point.
There have been a large number of studies referred to relating to the relationship between IUD use and ectopic pregnancies which included
review articles which summarise the results of studies over the years.In 1995, the medical journal Contraception published two such studies, Mol et al "Contraception and the risk of ectopic pregnancy: a meta-analysis" (Ex. P426) and Xiong et al "IUD use and the risk of ectopic pregnancy: A meta-analysis of case control studies"
(Ex. P209, Tab 2, Annexure B).The first meta-analysis by Mol et al reported the relationship between past IUD use and the increased risk of ectopic pregnancy in Table 1 of that review an extract from which is set out below:
TABLE 1: Detected studies and common odds ratios
Study (reference)
Control*
OR
95% CI‡
Ory (15)
IUCD (past use)
NP
1.6
1.3.2.0
Basuki (16)
NP
1.7
1.3-2.4
Holt (8)
NP
1.8
1.3-2.5
Chow (9)
P
1.4
1.0-2.1
Chow (19)
D
1.9
1.3-2.8
COR
1.6
1.4-1.8
0.1
1
-
-
Odds Radio and 95% CI*Control definition: NP = non-pregnant: P = pregnant: D = recently delivered: COR - Common Odds Ratio
:Odds ratio:
‡:95% confidence intervals
§Breslow Day Test Statistic Co homogeneityThe plaintiffs submitted that these results show that from 1981 five studies ( Ory 1981 , data from the Women’s Health Study; Chow, Daling & Ors 1986 ; Chow et al 1990 ; Hoult , Chu , Daling & Ors 1991 ; Basuki , Rossing and Daling 1994 ) demonstrated an increased risk of ectopic pregnancy among past IUD users using pregnant, non-pregnant and recently delivered controls. Pooling the data, the common odds ratio was 1.6. For practical purposes it was submitted this is equivalent to a relative risk.
Mol et al concluded:
“……After disontinuance of an ICDU, however, the risk
continues to be increased, suggesting a relation between
ICDU use and subsequent tubal damage.”In the same volume of Contraception in 1995, Xiong and others published their own meta analysis which reviewed the published literature from 1977 to 1994.
In the Xiong meta analysis, the pooled risk of ectopic pregnancy associated with past IUD use from 19 studies between 1977 and 1994 was 1.4.
Past IUD use and the risk of ectopic pregnancy
Study
Duration of IUD Use
95% CI
OR
95%CI
Time
(Year)
OR
Quality
Score
Pregnant controls:
Savolainen et al
Levin et al
WHO
Chow et al
0.33
1.4
0.7
1.7
0.09-1.13
0.7-2.5
0.4-1.2
1.1-2.6
<1
>1-3
>3
1.2
5.4
2.90.3-4.8§
0.63
1.5-19.9
0.8-11.5
0.67
0.90
0.90
Marchbanks et al
Chow et al
Coste et al
Kalandidi et al
Pooled result:
1.2
1.5
1.6
3.89
1.33
0.6-2.6
1.0-2.2
0.9-2.9
0.72-21.02
1.08-1.63
>2
6.19
not avail
0.93
0.73
0.87
0.73
Non-pregnant controls:
Savalainen et al
Ory et alWHO
Zhang et alRossing et al
Basuki et al
Total pooled result:
1.45‡
1.40.7
0.25*
0.33**
1.61.7
1.42#
0.3-7.05
1.1-1.70.1-3.9
0.03-1.87
0.02-1.92
1.1-2.21.1-2.5
1.25-1.62#
<2
<1
>2
1-3
>3
<1
1-3
<3
<1
1-3
>3
1.0
1.1
3.9
1.4
2.5
0.9
1.6
1.8
1.6
1.1
2.3referent
1.1-14.40.7-1.9
0.8-2.6
1.5-4.3
0.5-7.0§
0.8-3.4
0.9-3.5
0.9-2.9
0.5-2.2
1.3-4.00.63
0.870.90
0.73
0.73
0.930.83
*Hospital control; **Neighborhood control; Unadjusted OR; ‡Unadjusted OR and legal abortion control; §Dalkon Shield users; #Only including adjusted OR
This table is adapted from Xiong.The authors concluded:
“Past IUD use could slightly increase the risk of ectopic
pregnancy (Pooled O.R.:1.40, 95% C.I..:1.23-1.59).
This result challenges currently held opinions about the
absence of risk associated with past IUD use. Although this
Pooled O.R. is weak, it is firmly confirmed by:1) consistent results (independent of choice of control group;
2) no indication of compounding (O.R.’s with or without
adjusted studies are similar);
3) no effect of quality of study (Table 5); and
4) no indication of publication bias (Figure 3).
Furthermore, five studies explored the relationship between
the duration of past IUD use and the risk of ectopic
pregnancy showing that the risk of ectopic pregnancy
increased with the duration of past IUD use.”Dr Daling gave evidence that "the overall estimate is significantly elevated" (third witness statement para 2). The authors themselves, referring to the 40% increase in risk, said: "Past IUD use could slightly increase the risk of ectopic pregnancy.”
The plaintiffs also submitted that recent studies have shown that the risk of ectopic pregnancy increases with duration of use of the IUD Parazzini et al . (Ex. P209, Tab 3, para 1 which refers to Parazzini et al “ Past Contraceptive Method Use and Risk of Ectopic Pregnancy” Contraception 1995 , and Tab 2, para 29 Rossing et al 1993 , para 32, Basuki et al 1994.)
Dr Daling gave evidence of her own findings that amongst long-term users of an IUD, an elevated risk of tubal pregnancy was apparent for at least seven years after discontinuance of use. ((Ex. P209, Tab 1, para 10.)
The defendants criticise reliance upon these studies of the epidemiology of ectopic pregnancy on the basis that they do not break out the data for copper IUD users to permit findings of increased risk associated with Cu- 7 use. They also submitted that a paper by Sivin published in 1983 shows Dalkon Shield use explains the reported increase risk in studies to that time.
However, one of the studies Dr Daling relied upon, namely Rossing et al 1993 , reported that for women who had used IUDs in the past, compared with women who had never used IUDs, there was a relative risk of 1.7 of an ectopic pregnancy after cessation of IUD use. For Copper-T and Cu-7 users, the risk was 2.2. Moreover, Rossing et al 1993 stated:
"Although there was no evidence of increased risk of tubal pregnancy associated with up to one year of past IUD use, the risk rose steadily with increasing duration of use beyond one year. For women with three or more yeas of past IUD use, the risk relative to women who had never used an IUD was 2.5 whether for IUD use in general or copper IUD use only" (P247, RHC, first para).Second, there was no Dalkon Shield use to affect the results of later research ( Basuki et al 1994 ). It was submitted that these support a finding that the epidemiological evidence suggests past IUD use in general and past Cu-7 use in particular puts a woman at greater risk of an ectopic pregnancy.
The duration of use for the various plaintiffs in these proceedings is as follows (the number of months being rounded up):
1.
25 months
Mrs Lee
14 May 1975 to 28 Jun 1977
2.
22 months
Mrs Southren
5 Apr 1976 to 14 Feb 1978
14 Feb 1978 to 6 Jul 1991
41 months (ie continuous use for 63 months)3.
24-28 months approx.
Ms Ottaway
16 Jul 1974 to late 1976
13 Sep 1977 to 8 Dec 1977
3 months4.
Mrs Moylan
10 Jul 1978 to 19 Feb 1980
19 Feb 1981 to 2 Feb 1982
19 Feb 1980 to 16 Apr 1980
13 Apr 1984 to 14 Sep 198419 months
5 months. (In total approximately 38 months of use)
2 months adding up to 21 months of continual use
12 months5.
approx. 24 months
Mrs Lane
1975 to 1977
1977 to 2 Apr 1980
22 August 1980 to 13 December 1983
approx. 36 months
40 months
A total of approx. 100 months6.
less than one month
Mrs Robertson
1 Aug 1977 to 23 Aug 1977
7.
less than two months
Mrs Orders
20 Apr 1977 to 9 Jun 1977
8.
Mrs Denzin
#1 Nov 1978
#2 Dec 1978 to Jun 1980
#3 Jun 1980 to Nov 1981
#4 Nov 1981 to Jun 1983
#5 Jun 1983 to May 1985
#6 June 1985 to Sep 19861 month
18 months
19 months
19 months
23 months
15 months
A total of approx. 93 months of use.Among these plaintiffs, ectopic pregnancies were experienced by Denzin, Lane, Orders, Ottaway and Southren, after discontinuing IUD use. On behalf of these five plaintiffs it was submitted that the use of the IUD placed them at greater risk of ectopic pregnancy after cessation of IUD use and that they were not warned of this risk by the defendants.
The defendants submitted that there were other risk factors involved in the results for the plaintiffs.
Orders, Ottaway and Lane were " maternal smokers ". The defendants submitted that by smoking they were exposed to the risk of tubal pregnancy. The relative risks for past smokers was 1.3 (CI 1.0-1.8) and smokers at conception was 1.4 (CI 1.0 to 2.0).
On 25 February 1977 Mr Peddie from Searle Australia wrote a memorandum seeking advice as to Searle's official position with regard to the risks of PID and ectopic pregnancy associated with Cu-7 use. In that memorandum, Mr Peddie drew particular attention to the publication by Dr Hallatt of 15 July 1976.
In that publication Dr Hallatt made the following observations about the association between IUD use and ectopic pregnancy:
"This is the crux of the whole question: Does the IUD cause PID? If it does, then it increases the incidence of ectopic pregnancy and I think the message for us is that PID is actually a contraindication to the use of the intrauterine device" (p. 758, LHC, last para).It was submitted that this analysis by Dr Hallatt was in accord with the evidence given by Dr Keeping that IUD use increases the risk of ectopic pregnancy.
During the 6 June 1977 FDA Panel proceedings Dr Tyrer made a similar warning of a higher than normal incidence of ectopic pregnancy among women who became pregnant following IUD removal (see Reply to Chapter VII p. 53, para 58.9). Searle's own documents show it was aware of this: (see Reply to Chapter VIII, para 19.3, 19.4).
It was submitted on behalf of the plaintiffs that I should find that from February 1977, when the Peddie memorandum was written, Searle was on notice of the causal association between past IUD use and ectopic pregnancy. It was submitted that the association between past IUD use and ectopic pregnancy was related to the proposition that IUD use increases the risk of PID and thus tubal damage which inevitably increases the risk of ectopic pregnancy and this risk continues after IUD use has ceased.
The literature which Professor Daling referred to in support of her conclusions was in all but two cases published after the distribution of the defendants’ product ceased in Australia.
The defendants reviewed an extensive range of medical literature:
1. Hallatt 1976
This case series compared 70 patients who had used an IUD and suffered an ectopic pregnancy against 1,428 consecutive ectopic pregnancy patients who had not used an IUD. (Ex. D477).Dr Hallatt concluded that any causal link between IUD use and ectopic pregnancy had not been established and was doubtful.
The author commented on the then increasing incidence of ectopic pregnancies as reflecting the increasing incidence of PID.
2. Snowden 1977
Snowden concluded that there was no certainty that the Progestasert had a causal role in ectopic pregnancy and a detailed evaluation was necessary.3. Eschenbach and Daling 1983 (Ex. P.209, Tab C)
The defendants summarized the effect of this paper, in my view correctly, as follows:
(a) As at 1983 there had been a dramatic increase in the number of
ectopic pregnancies over the previous 2 decades.
(b) The pathological cause of ectopic pregnancy is tubal obstruction most often the result of previous infection, particularly salpingitis.
(c) The mechanics of ectopic pregnancy were then, as now, not properly understood.
(d) There are a myriad of risk factors for ectopic pregnancy, among which the method of contraception is but one possible factor. Prof Daling's extensive work with the population of the Puget Sound in relation to other risk factors for ectopic pregnancy is dealt with elsewhere.
(e) These risk factors are inter-related and constantly changing.
(f) Among the 4 cited controlled studies published prior to 1983, the authors note only 1 ( Ory et al 1981 ) which analysed IUD use but yet concluded that, in aggregate, women who had used an IUD had the same risk of ectopic pregnancy as women who had never used one.
(g) Clearly, the high prevalence of STDs (gonorrhea and chlamydia) was causing tubal obstruction (and continues to do so).
(h) Oral contraceptives appear to be protective against the risk of ectopic pregnancy.
(i) However, only well designed studies using proper controls and multivariate techniques would, in Profs Eschenbach's and Daling's opinions be able to clarify the cause and risk factors for ectopic pregnancies.4. Sivin 1983 (Ex. D.384, Tab 24)
Dr Sivin concluded that when past and present IUD users are
compared with women who have never used an IUD the risk was not
statistically different from unity. He also concluded:
(a) A large part of the engine driving the ectopic pregnancy rate up was the increase in PID, although that was not a complete explanation.
(b) Oral contraceptives and barrier methods are protective against the risk of ectopic pregnancy.
(c) However, contrary to the tentative queries raised by Beral and Tatum, there was no indication of a causal relationship between copper bearing IUDs and ectopic pregnancies.5. WHO 1985 (Ex. D.437, pages 131-141).
The effect of this study of 1,108 cases effectively concluded that
there was no increased risk of ectopic pregnancy for past IUD users but
that there were other risk factors such as smoking which could affect
incidence rates.6. Chow et al 1986 (Ex. D.209, Tab E)
Professor Daling was one of the authors.
The results of this study showed that once the Dalkon Shield was
excluded there was no material association between ectopic pregnancy and
use of other IUD’s.7. Edelman & Porter 1986 (Ex. D.384, Tab 6, pages 55-62).
This report showed that there was no statistically significant
result showing any increase in ectopic pregnancies from the use of Cu-7s.8. Chow et al 1987 (Ex. P.209, Tab M, pages 70-88).
Again Professor Daling was one of the authors.
The effect of that review of the epidemiology of ectopic
pregnancy was that no IUCD other than the Dalkon Shield gave
statistically significant increase in risk.9. Rossing et al 1993 (Ex. P.132).
This study referred to a small number of Cu-7 users and did
not differentiate between devices. It produced results at the bottom of the
statistically significant range indicating some risk. The degree of risk
varied to some extent depending on time. The study concluded that there
may be increased risk.These factors have varying degrees of impact on the rates of ectopic pregnancies. It is not necessary to deal with the other to indicate that they may be factors which increase the apparent rate in association with IUDs.
Experts called on behalf of the defendants expressed the opinion that past use of a Cu-7 did not result in an increased risk of ectopic pregnancy.
Professor Grimes T.5884-5
Professor Roy T6099-6101
Dr Keeping T6389-6390
Professor Newton T6643-6645
Professor Fraser T6750-6751 and 7691-679410. Basuki et al 1994 (Ex. P209, Tab 1).
This study showed a statistically significant increased risk for
women with 36 or more months use. Any shorter duration of use did not
produce a statistically significant result. The study did not determine the
type of IUD used. There was also a very small number of relevant articles
which gives an extremely wide confidence interval.In my view the material relied upon by the plaintiffs and the
evidence by Professor Daling as to the effect of past use of Cu-7 on
ectopic pregnancies does not establish that there was in fact any
statistically significant increase in the risk of ectopic pregnancies as a
result of past use of a Cu-7. An analysis of the literature not show any
support basis for the proposition propounded by the defendants.I accept the evidence of Grimes, Roy, Keeping, Newton and
Fraser referred to above and find that there was no increase in ectopic
pregnancies as alleged by the plaintiffs.I found Professor Daling’s selective assessments throughout her evidence unfairly pointing to IUD associated risk unsatisfactory and her general approach to the giving of her evidence is not to be ignored in the conclusions which I arrive at.
Further, in my view none of the material referred to led to the
conclusion that the state-of-the-art while the Cu-7 was sold showed a
view that use of a Cu-7 would increase the risk as alleged. Nor is there
any basis for concluding that the defendants were aware of such an
increase. The so-called Peddie Memorandum does not in my view
disclose any such knowledge.The plaintiffs also relied upon three 1995 publications.
1. Parazzini
This study involved a small number of IUD users and there was
no indication of the types of devices used. The authors concluded that
there may be an increased risk of ectopic pregnancy which increased with
prolonged direction of use but that the study does not confirm that.2. Mol et al 1995 (Ex. P426, pages 337-340)
This paper relied upon the papers referred to above and one other
paper.3. Ziong et al 1995 (Ex. P.209, Tab 3, pages 23-32)
This paper pooled existing material calculated a relative risk of
1.4 (1.23 - 1.59) which the authors described as a mild or slight increase.
The result was obtained without differentiating the Dalkon Shield results.In my view these three recent publications, when properly
The defendants drew attention to other risk factors for ectopic
analysed, do not alter the conclusion which I have expressed above. The
Mol and Xiong papers analyse previous data without what, in my view is
necessary, namely excluding Dalkon Shield results and differentiating
between types of IUDs. Parazzini is a small and inconclusive study.
pregnancies including prior induced abortions, previous abdominal surgery
including appendectomy, vaginal douching and maternal smoking.
4.2 THE SCIENTIFIC ANALYSISThis topic was dealt with in detail in Part 3 of the plaintiffs’ submissions.
It was an attempt to demonstrate ways in which the IUD could be shown scientifically to cause the introduction of pathogens.
The matters dealt with under this heading are:
1. Tailstring
It was said that bacteria which are pathogenic if introduced into the upper genital tract are able to adhere to the polymer of which the Cu-7 tailstring was made and that as that polymer was prone to fray more bacteria could adhere to it than would otherwise be the case so that in the event that the tailstring retracted into the uterus pathogens were introduced. Secondly it was said that bacteria could adhere to the tailstring and colonise it. The bacteria could then spread along the tailstring thereby finding a means of entry into the upper genital tract.In summary it was submitted that the Cu-7 may “ cause” the ingress of PID bacteria by
(a) retraction of the tailstring
(b) “ ascension” of the bacteria up the tailstring
(c) carriage upon insertion.The defendants submitted that neither (a) nor (b) have been shown to increase the chance of upper genital tract infection. The defendants admitted that (c) was a possibility but submitted that in relation to Lee, Moylan, Orders, Ottaway and Robertson whose cases were opened as cases of insertion infection they did not suffer PID as a result of the insertion of their Cu-7s.
They further submitted that insertion infection was a matter of which all doctors were aware and was not a risk which rendered the Cu-7 defective so that there was no causal link rendering Searle liable for insertion infection.
0
14
0