Dr Robin Smart’s recent letter to the NZMJ (http://www.nzma.org.nz/journal/116-1181/593/)1 criticises several aspects of our report of a survey of New Zealand general practitioners’ views of screening for prostate cancer,2 and also comments on important issues in relation to the conflicting views about the benefits of screening for and treating prostate cancer. I will attempt to respond to these criticisms and comments in the order in which they appear in his letter.

Apart from our paper in the NZMJ, the results of the survey have been reported to the New Zealand Guidelines Group and to several meetings of health professionals. There has been no attempt to use the public media to disseminate the results or conclusions of the survey and I do not understand what Dr Smart means by his reference to ‘use of the public media’ concerning the early diagnosis and management of prostate cancer.

The aim of our study was to determine whether the rapid increase in the reported incidence of prostate cancer was consistent with GPs screening for prostate cancer with the prostate specific antigen (PSA) test. It is not possible for a survey of the views of a random sample of 575 GPs to prove this hypothesis, but I believe that the results provide a reasonable level of support for this theory. Dr Smart says that the response rate of 66.3% for our survey was low. Although we would like to have achieved a response rate of 70% or better, many researchers who are experienced in this type of research would except that this response rate is sufficient to justify generalisation of the results to the total population of GPs. The survey was about screening for prostate cancer and not solely about PSA testing so that the systematic selection bias suggested by Dr Smart, whilst possible, seems unlikely.

Dr Smart is wrong when he states that the use of the term ‘screening’ is inappropriate. It is also unclear what he means by his preferred term of ‘early diagnosis’. There are various definitions of screening but an essential feature of screening for any disease is that it attempts to identify asymptomatic individuals who have a high probability of having the disease at a stage when it is amenable to treatment.3 Early diagnosis is a term that has been used to include both population screening programmes and opportunistic screening.4 The essential difference between these two forms of screening is the lack of a quality control process in opportunistic screening.3 The important criterion that the individual participants are asymptomatic is the same in both forms of screening. Concern about prostate cancer is not a symptom of prostate cancer. The survey of GPs was about screening for prostate cancer in men with no symptoms of the disease and this was made clear in both the introduction to the survey and the case vignettes.

The distinction between screening and case finding in men with symptoms is a fundamental issue in the assessment of the value of tests such as the PSA test because of the lead-time and length biases that occur in a screening programme. There is also the important ethical difference between screening, which encourages healthy, asymptomatic individuals to undergo tests, and a doctor using the best available tests and treatment, despite defects in medical knowledge, to help a patient who already has a disease.

Statements concerning published reviews of prostate cancer screening and its lack of proven benefit are not merely ‘an opinion of the authors’, but are clearly referenced statements of fact. All published systematic reviews, including the review by Bunting,5 have concluded that there is no evidence that prostate cancer screening will significantly reduce mortality.6 It is unclear why Dr Smart choses 1997 to define an ‘early PSA era’. A rapid escalation in the reported incidence of prostate cancer due to PSA screening occurred in the USA between 1988 and 1992.7 In the absence of a definitive randomised controlled trial, lead-time bias is only one of many factors that need to be considered in evaluating whether or not screening is likely to be of benefit.

In the randomised trial comparing radical prostatectomy with watchful waiting reported by Holmberg, the primary end point for the study was mortality from prostate cancer.8 Secondary endpoints were metastasis-free survival and the risk of local tumour progression. Using mortality figures as the most useful measure of outcome does not reveal the ‘narrowness of our view’ but reports the same measure used in Holmberg’s study. Overall mortality rather than disease-specific mortality is a more realistic measure of the outcome of most interest to the patient. Although the study excluded poorly differentiated tumours, only 5.2% of the study subjects had screen-detected prostate cancer so that the lead-time bias present in any screening programme was not considered. Clearly, the morbidity resulting from prostate cancer and its effect on the quality of life are important factors that need to be considered when evaluating the costs and benefits of treatment. However, these need to be compared with the morbidity caused by the treatment as well as other measures of the quality of life in men not receiving treatment. Dr Smart fails to do this in the figures he quotes and these issues were not reported or considered in the paper by Holmberg.

Screening can be effective only if the disease is discovered at a stage when cure is possible. This is possible in prostate cancer if the disease is localised to the prostate. In a report summarising some of the results of the first screening round of the two large-scale prostate cancer screening trials in the USA and Europe, 70% of screen-detected cancers were clinically localised and of these approximately 6% were Gleason score 8–10, 76% were Gleason score 5–7, and 15% were Gleason score 2–4.9 As stated in our paper the best estimate for the 10-year, untreated survival of prostate cancer with a Gleason score of 2–7 is in the region of 90%. The studies quoted by Dr Smart largely support this and there are other studies that have all been summarised in several systematic reviews.6 The Australian Health Technology Review found a 90–92% disease-specific, 10-year survival for Gleason score 2–7 tumours.10 It is clear from these figures that discussion of the survival rates for untreated Gleason score 8–10 tumours is largely irrelevant to the consideration of prostate cancer screening and I am unsure why Dr Smart includes these figures.

Trials of treatment provide little useful information unless there is an appropriate comparison for the outcomes of treatment. If prostate cancer with a Gleason score of 2–7 has a 90% 10-year survival when untreated, the 10-year survival figures after treatment must be better than 90% if treatment is to be of any benefit. Other than the study by Holmberg8 there are no randomised controlled trials of the treatment of prostate cancer. There is no evidence that the treatment of screen-detected, localised prostate cancers with a Gleason score of 2–7, either by radical prostatectomy or radiotherapy, is of any benefit. The studies quoted by Dr Smart do not have any control groups, and do not suggest any benefit from treatment. I know of no evidence to support his statement that ‘there is clear evidence that early curative treatment prevents prostate cancer deaths’.

In the latter part of his letter Dr Smart discusses several issues that were not considered or mentioned in our report of the survey of GPs’ views of prostate cancer screening. It is understandable that he prefers to treat men with localised and well-differentiated prostate cancers who have a better than 90% 10-year survival in comparison with men presenting with symptoms caused by local spread or metastases. However, there is a significant morbidity associated with radical prostatectomy and an ethical requirement that there is a benefit from screening that clearly outweighs any potential risks or harmful effects. The argument against prostate cancer screening can be summarised as follows:

I agree with Dr Smart that we should learn to use the tools we have as best we can, but with our present state of knowledge using the PSA test for screening for prostate cancer is likely to cause more harm than good.