This article is a response to ‘Truth is the daughter of time, and not of authority’: Aspects of the Cartwright Affair by Martin Wallace, NZ Skeptic 96.
The Cartwright Inquiry1 was held after the publication of “An Unfortunate Experiment at National Women’s” in Metro magazine in June 1987. The events leading up to the publication of the article and the findings of the subsequent inquiry have been contested ever since.
The inquiry heard from 67 witnesses, many doctors, 84 patients and relatives, and four nurses. In addition, 1200 patient records were reviewed, with 226 used as exhibits. The final report released in August 1988 has had a long-lasting impact. It recommended many changes in the practice of medicine and research, including measures designed to protect patients’ rights and a national cervical screening programme. These have since been implemented. The Medical Council announced in 1990 that four doctors were to face disciplinary charges resulting from the inquiry’s findings of disgraceful conduct and conduct unbecoming a medical practitioner. Charges against Dr Herbert Green were dropped due to ill health.
The report of the Committee of Inquiry has withstood many challenges, including judicial reviews and many articles alleging its findings to be flawed. Yet there have been allegations of a miscarriage of justice, charges of a witch-hunt, even a feminist conspiracy.
Where does this leave Dr McIndoe and others who had mounting concerns for so many years? Why did so many women develop cancer? In this article I will explore the findings of the Cartwright Inquiry, its context, the research and the criticisms, and attempt to find a more nuanced understanding of the “unfortunate experiment” and its ongoing effects. Page numbers in parentheses refer to pages in the Cartwright Report. CIN3 and CIS are interchangeable terms for a lesion of the cervical epithelium which can be a precursor to invasive cancer.
The Findings of the Inquiry
The report found that Green, rather than developing a hypothesis, aimed to prove a point (p 21) that even at the time was known not to be the case. A 1961 compilation of studies from Paris, Copenhagen, Stockholm, Warsaw, and New York showed CIS progressed to invasive cancer in 28.3 percent of cases (p 23). As at 1958 the official policy was “… treatment of carcinoma of the cervix Stage 0, [CIS] should be adequate cone biopsy … provided the immediate follow-up is negative and … the pathologist is satisfied that the cone biopsy has included all the carcinomatous tissue” (p 26). Standard treatment of the time involved excising all affected tissue and the ‘conservative’ treatment of conisation was in use well prior to 1966.
Green’s initial proposal stated “… It is considered that the time has come to diagnose and treat by lesser procedures than hitherto, a selected group of patients with positive (A3-A5) smears. Including the four 1965 cases, there are at present under clinical, colposcopic, and cytological observation, 8 patients who have not had a cone or ring biopsy. All of these continue to have positive smears in which there is no clinical or colposcopic evidence of invasive cancer”… The minutes then record that “… Professor Green said his aim was to attempt to prove that carcinoma-in-situ (CIS) is not a premalignant disease”… (p 22). This appeared to come about because of concern about unnecessarily extensive surgery for CIS between 1949 and 1962. During this period, some centres were beginning to use cone biopsy as effective treatment; however there were limitations to its use (p 27).
There were some questions over whether the work was a research project. The inquiry concluded this was the case and that a research protocol, however flawed, was put in place (p 69). Green published in peer-reviewed journals on his hypothesis and findings. By 1969, three cases of invasive disease had occurred in patients with positive cytology monitored for more than a year, and this should have made it clear that following patients with persistent CIS was unsafe (p 52).
Green then explained those patients by concluding that they’d had invasive cancer that was missed at the outset. The report contends this was dangerous to the patients as it demonstrated that the proposal was incapable of testing the hypothesis. These patients were reclassified by Green and the patients removed from the study (p 55). In addition, patients over the age of 35 were included in the research in breach of the protocol vp 49).
There were many subsequent issues, including lack of patient consent (p 136). Patients also had to return for repeated tests and other invasive procedures, often receiving general anaesthetics in the process (p 42-49). A collection of cervices from foetuses and stillborn infants and another of baby uteri in wax were collected by Green for research which was later abandoned. This did not appear to comply with the Human Tissue Act (1964) as no consent was obtained from the parents of the stillborn infants (p 141).
As part of an earlier 1963 trial to test whether abnormal cytology in women later developing CIS or invasive cancer was present at birth (pp 34 & 140), 2,244 new-born babies had their vaginas swabbed without formal consent from the parents (there was a decision to abandon this trial soon after it started but this wasn’t communicated to nursing staff until 1966).
Procedures such as vaginal examinations and IUD insertions/removals on hysterectomy cases were performed by students without patient knowledge or consent while they were under anaesthetic (p 172). There was a further study on carcinoma of the cervix treatment, where patients either had radiotherapy alone or hysterectomy and radiation (p 170). The method of randomisation was by coin toss.
The idea that patients were divided into two experimental groups arose from McIndoe et al (1984)2. The patients were divided retrospectively into two groups which overlapped strongly but not completely with groups defined by Green, that he called “special series”. In his 1969 paper, cited in the report (p 40-41) he stated: “The only way to settle the question as to what happens to carcinoma in situ is to follow adequately diagnosed but untreated lesions indefinitely … it is being attempted at NWH by means of 2 series of cases. (I) A group of 27 women … are being followed, without ‘treatment’, by clinical, colposcopic, and cytologic examination after initial histological diagnosis of carcinoma in situ … has been established by punch biopsy … (II) A group of 25 women who have had a hysterectomy (4 for cervical carcinoma in situ) and who now have histologically-proven vaginal carcinoma in situ, has been accumulated …” This was done semi-randomly, with cases presenting themselves fortuitously.
The outcome for the group of 25 who were included in the punch biopsy “special series” was summarised in the McIndoe et al (1984) paper. Nine out of 10 women who were monitored with continuing positive smears developed invasive cancer. Only one out of 15 women who had normal follow-up cytology later developed invasive cancer. While Coney and Bunkle may have made a mistake, it’s clear the judge didn’t. The report states: “Green’s 1966 proposal was not a randomised control trial, but it was experimental research combined with patient care” (p 63).
Green’s interpretation of the data in his 1974 paper is suspect, having concluded that the progression rate was 7-10/750 (0.9 to 1.3 percent) or 6/96 (6.3 percent) of ‘incompletely treated’ lesions (p 54). These were explained by suggesting that either invasive cancer was missed at the start, or over-diagnosed at the end. Dr Jordan (expert witness) deemed this interpretation incorrect as of the 750 cases, 96 had continuing positive cytology, meaning that the other 654 patients could be considered free of disease. Of that 96, 52 patients had not been assessed further, making it impossible to know whether or not this group already had unsuspected invasion. Of the 44 patients remaining with ongoing carcinoma in situ who had more investigations, seven were found with invasive carcinoma. The incidence of known progression was therefore 7/44 (16 percent), which approximates McIndoe et al (1984) findings. This means that the proportion of invasive cancer cases in those inadequately treated was much higher compared with those who had returned to negative cytology, even before any cases where slides were re-read and excluded are considered.
McIndoe et al (1984) covered the follow-up data for 948 patients with a histological diagnosis of CIS patients who had been followed for a minimum of five years; there was a further paper in 1986 regarding CIS of the vulva. The same method used by Dr Green to group women by cytology after diagnosis and treatment was used, but using the correct denominators and the original diagnosis. Patients who were diagnosed with invasive cancer within one year were excluded to avoid the possibility the cancer had been missed initially. The management was cone biopsy or amputation of the cervix in 673 patients, with 250 managed by hysterectomy. The only biopsies in 25 women were punch biopsy (11), wedge preceded by punch biopsy (7) and wedge biopsy alone (7). Twelve out of 817 (1.5 percent) of group 1 patients developed invasive cancer. Given the lengthy follow-up with negative cytology for group 1 patients, the authors concluded these represented the development of new carcinoma. There were marked differences in the completeness of excision between the two groups and the second group shows markedly different results, with 29/131 (22 percent or 24.8-fold higher chance) with positive cytology developing invasive cancer. At 10 years this was 18 percent rising to 36 percent after 20 years, irrespective of the initial management or histologic completeness of excision. This needs to be explained, as those figures strongly suggest the progression of CIS to invasion when it is and was a totally curable lesion. The answer is that a prospective investigation, as done by Green, has to establish that invasive disease is not present, while conserving affected tissue that is required for later study. The argument has been posed that women in the second group did get cone biopsies and hysterectomies. This ignores the fact that while many women were treated with various procedures, there was evidence of continuing disease, demonstrating that the intervention was inadequate. This was not followed up, posing a high risk of development of invasive disease.
This differs from group 1 patients, who were successfully treated at the outset. It’s pertinent to point out that the Cartwright Report did not rely on this study (or the Metro article) to reach its conclusions, but on review of patient records.
There have been two follow-up studies. McCredie et al (2008)3 examined medical records, cytology and histopathology for all women diagnosed with CIN3 between 1955 and 1976, whose treatment was reviewed by judicial inquiry. This paper gave a direct estimate of the rate of progression from CIN3 to invasive cancer. For 143 women that were managed by only punch or wedge biopsy the cumulative incidence was 31.3 percent at 30 years and 50.3 percent in a subgroup who had persistent disease at 24 months.
The cancer risk for 593 women who received adequate treatment and who were treated conventionally for recurrent disease was 0.7 percent at 30 years. These findings support McIndoe et al (1984) and extend the period of follow-up.
McCredie et al (2010)4, described the management and outcomes for women during the period 1965-74 and makes comparisons with women diagnosed 1955-64 and 1975-76. This showed that women diagnosed with CIN3 in 1965-74 were less likely to have treatment with curative intent (51 percent vs 95 percent and 85 percent), had more follow-up biopsies, were more likely to have positive cytology during follow-up and positive smears that were not followed by curative treatment within six months, as well as a higher risk of cancer of the cervix or vaginal vault.
Those women initially managed by punch or wedge biopsy alone in the period 1965-74 had a cancer risk 10 times higher that women treated with intention to cure. This was despite the 1955-64 group being largely unscreened, which would have delayed diagnosis. This study is important as it shows the medical experience of the women, where they were subjected to many interventions that were not meant to treat but rather to monitor.
Scientific misconduct happens, and for those trying to address it the risks are high. Brian Martin5 looked at several cases, and stated: “In each case it was hard to mobilize institutions to take action against prestigious figures. Formal procedures, even when invoked, were slow and often indecisive.”
McIndoe and others encountered similar difficulties and ultimately failed to get Green’s proposal reviewed. The concept of “clinical freedom” (p 127), where the doctor was the arbiter of the best course of action for the patient, was one major issue to emerge from the report. Colleagues tended to be very reluctant to intrude upon this, and this meant that the proposal could continue with little oversight or intervention. McIndoe had mounting concerns, particularly after 1969, which were disregarded or treated lightly.
These concerns were shared by pathologist-in-charge Dr McLean, and were raised internally with Medical Superintendent Dr Warren, who consulted with the Superintendent-in-Chief, Dr Moody and an internal working party set up to look at the issue in 1975. Twenty-nine cases that had developed invasive disease were referred to it; however only 13 were examined, and having set up its own terms of reference it only considered whether the protocol had been adhered to and disregarded concerns about patient safety (p 83).
The 1966 proposal effectively ceased when McIndoe withdrew colposcopic services and Green reverted to cone biopsy in most new cases (p 88), but it was never formally terminated. While Green himself did not take any steps to prevent the review of records by McIndoe and colleagues, Bonham did, and wrote a letter to the Medical Superintendent (p 92).
There are some important lessons to be learned from this, including that those with the authority to deal with the situation should make the best effort to achieve a balanced view of the situation and assess it fairly to allow the claimant a fair hearing.
The potential risks of Green’s proposal outweighed any benefits such as avoiding hysterectomy or cone biopsy. Invasive cancer could not be ruled out because there were poor safeguards against the risk of progression. This was unethical from the outset, regardless of the issue of informed consent. In addition, patients that developed invasive disease had their slides reclassified and were removed by Dr Green from the study. This would be considered research misconduct then and now as it manipulated the data.
It does not matter if the initial motivations were sincere; they ultimately fail on these points. This proposal had a very human cost. Moreover Green’s views had long-term effects, including influence on undergraduate and postgraduate medical students, and support for the attitude that cervical screening was not worthwhile. This ‘atypical’ viewpoint was also promoted in the scientific literature and in the press, creating confusion within the medical scene and with the public.
It can be incredibly hard to admit our failings and let go of old loyalties. In the aftermath of the report many doctors objected to cervical screening, ‘unworkable’ consent forms and the intrusion of lay committees on practice6. It’s true this had negative effects on the perception of doctors overall, particularly in regard to practices that were widespread in hospitals at the time, and there were times that unfair criticisms were aired. This impacted on the nursing profession as well, for nurses are meant to be patient advocates.
This was also about power. The really unfortunate thing is that medical responsibilities to patients are almost totally ignored in the midst of the argument, when they should be brought to the forefront. Likewise respect, justice and beneficence were lacking for the patients involved. No doctor raised concerns about the lack of consent, even though from the 1950s there was the growing expectation that this be sought, particularly with participants in research.
The Medical Association working party that examined this stated that it was “regrettable that the trial deteriorated scientifically and ethically and did not change as scientific knowledge advanced or as adverse results were observed”7. They found it deplorable that patients involved did not know they were part of a trial, and that it took a magazine article for it to be investigated.
Unfortunately, instead of addressing this and examining whether Dr Green made any errors or misinterpretations himself, the findings in McIndoe et al (1984) and other papers were not accepted. There is the unfortunate implication that, rather than there being mounting and valid concerns over decades, that Green was unfairly toppled and the resulting inquiry was a whitewash.
The report couldn’t have been written without the assistance of the medical community as expert witnesses and advisors. It’s not surprising that there would be loyalty for a colleague, but perhaps instead of attempting to rehabilitate Green it’s time McIndoe and his colleagues were vindicated. Morality did not totally fail and attempts were made to prevent patients being harmed8.
Acknowledgements: many thanks to Dr. Margaret McCredie of Otago University who assisted me with my research.
- The Cartwright Report: www.nsu.govt.nz/current-nsu-programmes/3233.asp
- W.A. Mcindoe; M.R. McLean; R.W. Jones; P.R. Mullins 1984: J. Am. Coll. Obst. 64(4).
- M.R.E. McCredie; K.J. Sharples; C. Paul; J. Baranyai; G. Medley; R.W. Jones; D.C. Skegg 2008: The Lancet Oncology DOI:10.1016/S1470-2045(08)70103-7
- M.R.E. McCredie; C. Paul; K.J. Sharples; J. Baranyai; G. Medley; D.C. Skegg; R.W. Jones 2010: A&NZ J. Obst. Gyn. DOI:10.1111/j.1479-828X.2010.01170.x
- B. Martin 1989: Thought and Action 5(2), 95-102.
- J. Manning (Ed.) 2009: The Cartwright Papers: Essays on the Cervical Cancer Inquiry 1987-88. Bridget Williams Books Ltd.
- L. Bryder 2009: A History of the “Unfortunate Experiment” at National Women’s Hospital. Auckland University Press.
- C. Paul 2000: BMJ 320, 499-503.