We would like to congratulate the New Zealand Medical Journal on an excellent and thought-provoking issue, covering the controversy and reawakened interest in Dr Herbert Green's \"unfortunate experiment\" at the National Women's Hospital.Dr Graeme Overton1 offers a wonderfully clear explanation of the key problem with the 1984 paper by McIndoe and colleagues.2 He points out that, not only were the two groups retrospectively divided on the basis of persistent abnormal cervical cytology during follow-up and not prospectively as experimental groups for comparison of different treatment strategies, but that the paper reports absolutely no discernible differences in the initial treatments received by these two groups, with similar proportions having initially undergone hysterectomy (an unthinkably radical treatment by today's standards), cone biopsy and other procedures.Unfortunately, the same level of attention has not been paid to the similar problem in the 2008 paper by McCredie and colleagues.3 This paper states clearly that the authors divided their sample into adequately and inadequately treated groups. However, a major problem in their methods is that they use the outcome following treatment as part of the classification system: "Any procedure followed by a positive smear in the following 6-24 months was classified as inadequate treatment (Figure 1). Four women who developed cancer within 2 years of CIN3 diagnosis, but who had no follow-up cytology, were assumed to have had inadequate treatment." Given that they partially base the classification of adequacy of treatment on outcomes, it is totally unsurprising and uninformative that the authors should find that women who are classified as "inadequately treated" have poorer outcomes. It is difficult to follow exactly what this paper was trying to prove, but as a means to demonstrate that conservative treatment led to worse outcomes, the methods are wholly inadequate. We therefore suggest that the New Zealand Medical Journal editorial4 errs in citing this paper as evidence that women were harmed. It provides no such evidence. It is regrettable that Professors Paul and Jones, co-authors of the McCredie paper, should have turned down the opportunity to provide a scientific defence of their approach for this issue of the New Zealand Medical Journal and we applaud Professor Frizelle for his decision to continue with publication in the face of those difficulties.4 (We declare we have no conflicts of interests.) Ms Josie Sandercock DPhil Candidate Dr Amanda Burls Director of Postgraduate Programmes in Evidence-Based Health Care Department of Public Health & Primary Care, University of Oxford Oxford, UK

We would like to congratulate the New Zealand Medical Journal on an excellent and thought-provoking issue, covering the controversy and reawakened interest in Dr Herbert Green's \"unfortunate experiment\" at the National Women's Hospital.Dr Graeme Overton1 offers a wonderfully clear explanation of the key problem with the 1984 paper by McIndoe and colleagues.2 He points out that, not only were the two groups retrospectively divided on the basis of persistent abnormal cervical cytology during follow-up and not prospectively as experimental groups for comparison of different treatment strategies, but that the paper reports absolutely no discernible differences in the initial treatments received by these two groups, with similar proportions having initially undergone hysterectomy (an unthinkably radical treatment by today's standards), cone biopsy and other procedures.Unfortunately, the same level of attention has not been paid to the similar problem in the 2008 paper by McCredie and colleagues.3 This paper states clearly that the authors divided their sample into adequately and inadequately treated groups. However, a major problem in their methods is that they use the outcome following treatment as part of the classification system: "Any procedure followed by a positive smear in the following 6-24 months was classified as inadequate treatment (Figure 1). Four women who developed cancer within 2 years of CIN3 diagnosis, but who had no follow-up cytology, were assumed to have had inadequate treatment." Given that they partially base the classification of adequacy of treatment on outcomes, it is totally unsurprising and uninformative that the authors should find that women who are classified as "inadequately treated" have poorer outcomes. It is difficult to follow exactly what this paper was trying to prove, but as a means to demonstrate that conservative treatment led to worse outcomes, the methods are wholly inadequate. We therefore suggest that the New Zealand Medical Journal editorial4 errs in citing this paper as evidence that women were harmed. It provides no such evidence. It is regrettable that Professors Paul and Jones, co-authors of the McCredie paper, should have turned down the opportunity to provide a scientific defence of their approach for this issue of the New Zealand Medical Journal and we applaud Professor Frizelle for his decision to continue with publication in the face of those difficulties.4 (We declare we have no conflicts of interests.) Ms Josie Sandercock DPhil Candidate Dr Amanda Burls Director of Postgraduate Programmes in Evidence-Based Health Care Department of Public Health & Primary Care, University of Oxford Oxford, UK