6 July 1999
Reviews of first submission
First review by "F"
Second review by Dr Richard Pincus
Comment by Ms Nicki Mollard
Statistical review by anon
Response from the editors by Dr Ruth Armstrong
Authors' response by A/Prof Jeanette Ward
Reviews of revised article
Comments from the editors - Ruth Armstrong
Further comments by "F"
Further comments by statistical reviewer
Authors' response by A/Prof Jeanette Ward
Editorial Committee decision - Dr Ruth Armstrong
| Date | Comment |
|---|---|
| Mon Dec 21 11:58:42 1998 |
First review
Title: The title is misleading as the issue of prostate cancer detection in the scenario posed relates to a 58 year-old man. Therefore, "screening" should be replaced by "case selection" in the title as well as the text in relation to the scenarios. Abstract: In the Abstract's Conclusion (par 12) the authors should qualify the statement that "GPs perceive limited medicolegal protection…". The GPs sampled were from NSW, were mostly males and were predominantly in full-time and rural practice (ref par 27 in the text). The last sentence's dictate is concerning since, before making such a statement so assertively, the authors need to be absolutely certain that case selection (as opposed to screening) is more harmful than not "screening" in such selected cases. In order to do that, they need to balance the disadvantages of screening1 against the potential and emerging evidence of benefit (albeit for subgroups)2,3 from detection and treatment of localised disease in relatively young men. General comments: The issue of "screening/not screening" is not quite as clear-cut as implied by the researchers although most would concede that an undisputed benefit from screening is not established for prostate cancer. However, absence of evidence must not be confused with evidence of absence. Consequently, by not informing the GPs that the issue is not clear-cut, the basis for the "informed decision" is flawed as the information provided is incomplete - with repercussions for the evidence-based approach espoused throughout the paper. Furthermore, at no part in the paper do the authors even mention that Australian and some overseas guidelines differ. This is an essential inclusion if a truly informed Discussion and reasoned Conclusions are sought. That there are 3 Australian guidelines is a statement in itself. Introduction: Because of the comments above, the sentence in par 15 "As described elsewhere15, neither PSA alone or in combination with DRE is defensible as a screening strategy." needs to be changed. Furthermore, "will" should be changed to "may" in the last sentence of that paragraph. Methods: As stated above, it is a pity that, for the first scenario, the questions asked in the second part were not following information about the 3 Australian guidelines as well as at least some overseas guidelines with different recommendations. At the top of page 4 (par 27), the reason that 64 (16%) were ineligible should be stated. Pars 31 and 33 could be written in a manner which is easier to understand and these should be revisited by the authors. The first sentence in par 34 should be changed to "After reading the three Australian guidelines…" consistent with the above comments. Discussion: The sentence at the end of paragraph 41 suggests a (thinly veiled) agenda which causes concern. This is especially so since the guidelines are only guidelines and the evidence is changing continuously. Furthermore, if prostate cancer screening can be motivated by medicolegal concerns (para 44), which appear to be increasing at an exponential rate in Australia, the proposal for greater resourcing of implementation of Australian guidelines is likely to adversely affect the quality of life of the perplexed and pressured GP. Therefore, such a strategy cannot be endorsed in this paper unless the authors qualify this statement in the discussion - consistent with the paper's title! Is there evidence that greater resourcing would result in closer support of guidelines? This point should be addressed one way or other if the last sentence in par 41 is to remain as it is. No doubt the authors will be aware that there is overseas evidence that patients informed about the lack of mature research evidence regarding effectiveness of the PSA screening test are far less interested in undergoing screening4. Therefore, placing the onus on the patient would seem a more appropriate approach, provided both sides of the debate are proffered, rather than potentially making life even more difficult for GPs. In par 43, comments should be made about evidence outside Medicine and the Biological Sciences regarding the "Protection" afforded practitioners close adherence to guidelines in terms of litigation, as experience in other arenas may have relevance in principle for Medicine. In par 46, the authors state that their response rate was modest (65%). However, this is not unusual with such surveys as Dr Ward's publication experience demonstrates. Tables: The total numbers for the questions in the tables vary so these (the totals) should be included. Conclusion: Despite the comments made above, this is an interesting study and, with the suggested changes would, in my opinion, be suitable for publication. The authors have studied a select group of GPs in relation to a specific situation(s) but, on the basis of this are making general pronouncements. Because of the natural history of prostate cancer and the fact that, until very recently, scientific clinical studies have been few, objective data regarding vital issues such as PSA screening are lacking. However, with respect to PSA screening, the converse to Ward et al's assertions is also true. At present there is no clear evidence that PSA screening, let alone case selection, is not beneficial. Therefore, in the interests of balance, a significant change of emphasis is required by these authors. References: 1. Stewart-Brown S, Farmer A. Screening could seriously damage your health. BMJ 1997; 338:533. 2. Perotti M, Rabbani F, Farkas A, Ward WS, Cummings KB. Trends in poorly differentiated prostate cancer 1973 to 1994: observations from the Surveillance, Epidemiology and End Results database. J Urol 1998; 160:811-5. 3. Lu-Yao GL, Yao SL. Population based study of long-term survival in patients with clinically localised prostate cancer. Lancet 1997; 349:906-910. 4. Wolf AM, Nasser JF, Schorling JB. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Intern Med 1996; 156:1333-6. F |
| Tue Dec 22 14:16:56 1998 | Second review
This article is one of the highest quality articles I have been asked to review, and is certainly publishable in its present form in my opinion.
Paragraph 14: I am concerned about the word "undue." I am not certain that references 4 to 7, if available to me, would support the proposition that fear of medical litigation has an "undue" influence on clinical behaviour. One of the justifications of tort law is that it does influence practitioners. It is only if the cases are wrongly decided that this influence would be adverse to optimum patient care. There is much literature on the subject. Paragraph 15: The words "as fully described elsewhere" might be replaced by "Acceptance that the guidelines are sound depends on acceptance that: (etc)" Reasoning: The authors state that "current tests do not distinguish innocuous from aggressive prostate cancer." Certainly the only tests mentioned to that point in the article are incapable of distinguishing aggressive from "innocuous" prostate cancer (does such an entity exist?). This sentence, with the next two sentences (note particularly reference 15) presents the case against PSA and DRE testing of asymptomatic men, which is the basis of the guidelines. In my view this section should be rewritten to indicate that and no more. ( I, like the authors I suspect, am a great believer in and advocate of clinical practice guidelines. The opponents of guidelines fear that they will be forced to accept an orthodoxy even in areas where controversy still rages. By asserting the orthodoxy and ignoring the controversy, the article is less persuasive.) Paragraph 16: The sentence referenced by reference number 21 is obscure. 1 think it means: "In an American study, physicians were given a theoretical scenario concerning a patient who developed prostate cancer and who had not been previously screened by his GP. Those physicians who thought the patient was likely to be successful in court in suing his GP were significantly more likely to say they would perform PSA tests than those who thought the patient would fail in court." Paragraph 31 : The word "only" beginning the last sentence portrays the expectations of the author, but contributes nothing to the article. Paragraph 33: Again the word "only" is inappropriate in what should be a data reporting section. I could not make much sense out of the second sentence in paragraph 33. Paras 34-35: These are data-dense, difficult, and contain double negatives. I suspect many readers would get lost in these paragraphs. On my count there are 12 conclusions. The statistician in me would like reassurance that these 12 conclusions represent the only research hypotheses tested. Accepting that the article is intended to be polemic, it might be simpler to present the data and extract only those conclusions from it which the authors depend on for their recommendations. The data tables can be annotated to indicate what correlation testing was done, and which of the tests reached levels of significance. Each thoughtful reader will extract his own most interesting conclusion, and form his own inferences from that conclusion. For example, the most surprising conclusion for me was that 88% of GPs agreed or strongly agreed that patient's decisions should be based on full disclosure of what is known. One suspects this represents a sea-change in thinking, not uninfluenced (one imagines) by legal decisions. Having recently been offered DRE and PSA (quite insistently) during the course of a routine medical examination I may be unduly sensitive about the topic. Paras 41 & ff. I start to lose contact with the authors during the discussion. The reasons are complex. The clear intent of the authors would be better served by gentle persuasion and use of fewer unspoken assumptions. The authors appear puzzled by what could be perfectly responsible and rational behaviour on the part of their experimental subjects. It would be unfortunate if they alienated what I imagine is one of their main targets. There will be no more discriminating and careful reader than the subjects themselves, and their GP colleagues. GPs are at times upset by cooperating with a research project then being defamed in the publication. I do not see any justification (for example) in the study data as presented for the statement that "awareness of national guidelines was low". The actual data shows that requests by asymptomatic men are common. The subjects can thus be assumed to have considered the issues as they know them and to have decided their approach. Most GPs would be aware at least that the guidelines suggest these tests should not be offered routinely. The guidelines have been extensively canvassed in the media. A presumption that this was taken into account by the GPs in their decision would seem proper. I note the scenario is of a patient who asks for the tests, not one who asks about them. Patient's right to know and choose is sufficient explanation, and quite rational explanation, for the lack of influence of the guideline in this scenario.* Paragraph 42: Does the word "risk" mean risk of suit? The propositions advanced are that GPs have little knowledge and poor recall of "national guidelines", and that there is little or no influence of the guidelines on GPs reported behaviours. Assuming these propositions were correct, these are arguments against the recommendations put forward by the authors, not for them. It would be pointless collectively for the community to mandate that little known or poorly recalled guidelines, which have no reported effect on behaviour, should determine the outcome of law suits. It would be unjust for the individual. *The main issue the guidelines addresses is not the scenario the authors test. If the evidence was that these tests should be offered the patients, the guidelines would so state. The behaviour of the GPs would be entirely different in that situation. The considerations which arise when a GP is requested by a patient to order a certain test are entirely different. Very properly 88% of GPs claim at least that they would respond to such a request by providing all the relevant information and allowing the patient to decide. The discussion as written shows little empathy with, or understanding of, the factors that actually influence the GPs in providing the answers to the surveyors. Paragraph 44: The authors appear to assume that if physician behaviour is influenced to any extent by fear of malpractice suits, this is a bad thing. I will confine myself with saying "it ain't necessarily so." Paragraph 45: I have no idea what "interventional trials to diffuse public and professional anxiety about prostrate cancer screening" would be. I would be fascinated to see a proposal for such a study at my ethics committee. Paragraph 57: I note the title but I do not have access to the paper. I had understood that there was such a review, commonly called the "Tito Review". I confess to have heard Fiona Tito speak about the review, but not to have read it. As I understood her conclusions, they were that there was little or no evidence that there is a medical negligence litigation crisis in Australia, and a great deal of evidence that medical negligence is common. Richard Pincus, FRACGP, Barrister-at-law |
| Tue Dec 29 12:14:13 1998 | I have reviewed the comments of reviewers 1 and 2 and consider the concerns / issues they raise are worth consideration by the authors.
I have no further concerns with the article. Nicki Mollard, Solicitor |
| Mon Feb 8 23:02:56 1999 | This is an extremely important topic - reasonably well presented and requiring manageable changes.
Statistical issues: The reporting of absolute numbers with percentages in brackets is tedious to read - and should be altered to report percentages only. For the more important results, however, the reported percentages must be accompanied by confidence limits in parentheses. Similarly all the tables contain meaningless numbers, with the most interpretable figure, the percentages, given in parentheses. It would suffice (and simplify) to present N in the title of the table and just report percentages in the body. Even more tedious is the reporting of triples such as "(chi-squared=0.81, 1df, p=0.4)". If the authors insist, the p-value alone would be sufficient. Par25, after all makes it clear that chi-squared on 1 df is the principle hypothesis testing tool. (Although the use of McNemar's test in Tables 1 and 2 does not appear to be described in the Methods). In par 28 we are told the sample over-represented male GPs, those in fulltime practice, and those in rural practice. This begs the question of 'by how much?' A partially contradictory statement in par46 suggests the preponderance of males was "consistent with the reference population". Which statement is correct? Some minor issues By and large the authors are careful in their choice of words describing the respondents' perceptions: eg the respondents "thought" or "considered", or "perceived". However the use of the word "indicated" in several places throughout the paper is not so successful - and the clarity of the paper would be improved by reverting to one of the more suitable choices for which they seem already to have a reasonable thesaurus. Some of the strategies in Table 4 need to be worded more clearly. In par12 and par47 the word "unfounded" might be seen as offensive and alienating by GPs reading this paper. There may be plenty of medical evidence on the efficacy of screening for prostate cancer, but there is clearly no legal evidence or precedent on which anyone can predict the real-life outcome of the second-case scenario. The Conclusions read: - "This lack of evidence must be addressed by government, medical defence organisations and opinion leaders before unfounded perceptions undermine evidence-based health care". However, despite the strong support for the first question in Table 4 there is virtually no discussion of the usefulness of government legislation in this paper. Legislation is surely just as important as "the need for interventional trials to diffuse public and professional anxiety about prostate cancer screening" (par45). None given |
| Fri Feb 12 10:16:23 1999 |
Dr Ruth Armstrong Editorial Registrar Thank-you referees for your comments and authors for your submission and your patience with a few technical hitches. Having read the manuscript with interest and considered all the referees' comments, we would like to invite submission of a revised paper. Our suggestions for the revised version are; Title: Please think about the use of the word "screening" here and throughout the paper. "Testing" or "case finding" may be more appropriate. Abstract: When was the study done? Introduction: In view of the referees' comments you may like to consider making some changes in your underlying assumptions. Results: Why were 64 GPS ineligible? You may like to present more information on the demographic representativeness of the sample. Tables: These need to be shortened. Table 4 is probably not necessary and table 5 should be abbreviated to include only the most interesting/ necessary responses. Discussion: Reconsider in the light of the referees' comments. This will also need to be shortened to comply with our limit of 2500 words for research papers. Please provide a "covering letter" responding in full to the referees' comments and highlight changes in the text. We would like to see a revised manuscript by Thursday April 1 but the floor is open to further comments/ ideas/ at any time. Thanks again everyone for your participation Ruth Armstrong |
| Fri Mar 26 14:57:59 1999 | Authors' reply
Thank you for your email dated 11 February 1999, inviting response to referees' and editorial comments about this manuscript. First, we note that Referee 1 (anonymous) states 'this is an interesting study and, with the suggested changes, would, in my opinion, be suitable for publication'; Referee 2 (Dr Pincus) states 'this article is one of the highest quality articles I have been asked to review, and is certainly publishable in its present form in my opinion' and Referee 3 (anonymous statistician) states 'this is an extremely important topic'. We respond to referees' comments as follows: Referee 1 (Anonymous) 1.1 Suggests the title ought to be changed because 'the issue of prostate cancer detection in the scenario posed relates to a 58-year-old man. Therefore, screening should be replaced by case selection' in the title' .. and throughout. This comment reflects a common misunderstanding among clinicians, perhaps because of their limited training in the principles and practice of disease screening (see Nicol et al. Urological training in Australasia: perceptions of Fellows and current trainees. Australian & New Zealand Journal of Surgery 1995; 65: 278-283). Although 'case selection' as stated by this referee is yet to be coined in the literature, we assume it equates with 'case finding'. 'Case finding' is taken to mean the initiation of screening in a doctor's surgery, perhaps in response to a man's request, as opposed to 'mass screening'. For some reason, these are considered ethically and clinically different. It is, however, rarely appreciated that the criteria for mass screening and case finding are identical. Further, an asymptomatic man's risk for prostate cancer and the evidence (or absence of evidence) of benefit of earlier detection as well as the evidence for harm are identical for him as an individual irrespective of the context or motivation for screening - whether because of a public campaign, prompting by his partner, self-presentation or prompting by his GP. The use of 'case finding' in this circumstance has been denounced by public health authorities as it serruptiously 'evades the need to demonstrate net benefit' (Wald N, Morris J. What is case-finding? J Med Screening 1996; 3: 1). Thus, we decline to alter this aspect of the title. Further, discussion of the referee's views and our own about this issue would properly and usefully take place in the correspondence columns. 1.2 Suggests we should qualify our statement 'GPs' in the Abstract conclusion (para 12). Incorporated. 1.3 Is concerned about the last sentence of the Abstract, suggesting we need to be 'absolutely certain' that case selection (as opposed to screening) is more harmful than 'not screening in such selected cases' and draws our attention to 'the potential and emerging evidence of benefit (albeit for subgroups)' in two published studies (Perotti et al 1998; Lu Yao et al 1997) The Conclusion of the Abstract has been re-written. 1.4 Suggests that the issue of 'screening/ not screening is 'not quite as clear-cut as implied by the researchers', restating that the evidence base is currently inadequate. We have modified this section of the Introduction. 1.5 States that 'at no part in the paper do the authors even mention that Australian and some overseas guidelines differ'. All recognised Australian guidelines are unanimous in their positions about the absence of compelling evidence for prostate cancer screening. We decline the suggestion to add to the reference list a series of international guidelines, as this would unnecessarily inflate the word count and reference list. 1.6 Requests changes to the sentence in para 15 , including 'will' to 'may'. Incorporated. 1.7 States 'it is a pity' that we didn't include overseas guidelines in our survey. Our study confined itself to the impact of current Australian guidelines. perhaps the reviewer could conduct such a study. 1.8 States the reasons for ineligibility need to be stated. Details about the 64 (16%) of the 400 considered ineligible are: Deceased 2 Retired 9 Not in general practice (eg specialists, Family Planning) 25 On extended leave 5 Moved overseas 8 Changed address and uncontactable despite further efforts 10 Returned to sender and uncontactable despite further efforts 5 TOTAL 64 Others have examined the accuracy of GP databases, concluding that all have some degree of inaccuracy, such that sometimes up to a third of names on GP lists may be ineligible for surveys (Saltman D, Mant A. General practitioner databases in Australia. MJA 1992; 156: 16-20; Dickinson J, Sanson-Fisher RW. General practitioner databases in Australia (letter). MJA 1992; 156: 580; Bridges-Webb C. General practitioner databases in Australia (letter). MJA 1992; 156: 580). We do not consider this information worth the word count it would consume to convey. We would happily add '.. for reasons obtainable from the authors' if required by the Editor. Please let us know. 1.9 Requests paras 31 and 33 be rewritten 'in a manner which is easier to understand' Incorporated. 1.10 Suggests specific changes to first sentence para 34 Incorporated as well as re-written. 1.11 States 'the guidelines are only guidelines and the evidence is changing continuously'. The ethics of population-based screening require compelling evidence of mortality reductions which, in turn, need to be communicated to the public. We argue that guidelines do summarise the evidence (and lack thereof) for professional and public consumption. Further, the reviewer is optimistic in stating 'the evidence is changing continuously'. We agree that, from time to time, studies are published examining the sensitivity and specificity of alternative screening regimes or clinical case series (almost always from overseas centres) describing outcomes of treatment for prostate cancer. To address questions about the value of population-based screening however, the definitive answer will come from the RCTs which are yet to be finished and published. Evidence of public health importance has hardly grown in the last five to ten years. 1.12 Argues that implementation of guidelines 'is likely to adversely affect the quality of life of the perplexed and pressured GP' and, thus, that our recommendations 'cannot be endorsed' The reviewer provides no evidence in support of this assertion. The reviewer is entitled to his or her opinion. We look forward to learning of his or her evidence in support of these assertions in correspondence but decline to incorporate this opinion in our own paper. 1.13 Requests evidence that greater resourcing of guidelines would enhance compliance with them. This sentence has been rewritten. 1.14 Suggests the 'onus' be placed on the patient. This option is not mutually exclusive of GP interventions. We draw attention to an expert survey identifying information considered crucial for patients to make an informed decision (now published). 1.15 Requests evidence about 'protection' from areas outside of medicine Changes to the Discussion have removed ambiguous reference to medicolegal protection. In any case, the value of citing legal precedents from other fields presupposes the contexts are comparable. By contrast, medical practice is one of the least regulated or externally scrutinised. The standing of guidelines in other fields such as occupational health and safety which are have no plausible relevance in our view. Again, however, a reply from the reviewer or an accompanying editorial might do justice to this complex issue. 1.16 Requests totals for rows in tables We now have added the N of respondents in each table title, removed N from tables themselves and, in text, have stated that, where rows do not add to 100%, that data were missing. 1.17 Requests 'a significant change of emphasis' as 'there is no clear evidence that PSA screening, let alone case selection, is not beneficial'. See 1.1 and 1.11. Reviewer 2 (Dr Pincus) 2.1 States his comments 'are based on the assumption the authors wish as much to persuade as to present the data'. We applaud the referee's frankness but reject his suggestion that persuasion motivated us to conduct the study and prepare our manuscript. We sought to achieve the aims as stated. 2.2 Suggests the strength of the assertion in the Conclusion of the Abstract would be increased if there were evidence to show that guidelines do not offer legal protection, offering three references as sources of evidence. These three articles are opinion-based. We have not added any we have not already cited. 2.3 Concerned about the use of the word 'undue' in Para 14 as 'it is only if the cases are wrongly decided that this influence would be adverse to optimum patient care' Deleted. 2.4 Suggests specific changes to fourth sentence of Para 15 These suggestions were similar to referee 1's comments, reinforcing the need to re-write this paragraph to assert the controversy rather than any orthodoxy. This section has been reworked. 2.5 Suggests specific rewording in para 16 to explain ref 21. Incorporated. 2.6 Suggests deleting 'only' from para 31 and 33, further inviting changes to the second sentence on para 33. Incorporated. 2.7 States the data in paras 34 and 35 are dense and requests reassurance 'that these 12 conclusions represent the only research hypotheses tested'. Yes, we confirm this is so as per analysis paragraph. We do not data-dredge. 2.8 Suggests we only present data and extract from it 'only those conclusions from it which the authors depend on for their recommendations'. We would argue that this has been our approach. We are confident the revisions make this more explicit. 2.9 States 'it would be unfortunate if they alienated what I imagine is one of their main targets'. We have carefully re-read our revisions and do not consider this a likely outcome. 2.10 Proffers an opinion that GPs would know about the guidelines because they have been 'extensively canvassed in the media'. As the data from the survey itself suggest otherwise, we decline to modify our discussion. 2.11 Asks whether 'risk' in para 42 means 'risk of suit'. Incorporated. 2.12 States 'it would be pointless collectively for the community to mandate that little known or poorly recalled guidelines, which have no reported effect on behaviour, should determine the outcome of lawsuits'. Commercial, professional and lay interest groups can make screening a 'standard of care' without evidence of effectiveness (Wasson 1995). That inadequate dissemination and implementation of guidelines would be another force in such a trend is very disturbing. 2.13 Suggests we have 'little empathy with, or understanding of, the factors that actually influence the GPs' in their responses. We present the usual caveats and would argue this requires a separate study. 2.14 Suggests that is is not always so that influencing behaviour by threat of malpractice is a 'bad thing'. If this threat is explained by the situation in which evidence-based guidelines are inadequately disseminated, compounded by a legal ruling which contradicts medical evidence (not implausible), then this arguably would be a 'bad thing' for an evidence-based paradigm in health care. 2.15 States he has 'no idea' what an interventional trial would be. We have deleted. 2.16 States that ref 7 refers to the Tito review. This is incorrect. Hancock's work stands independent of the Tito review. The referee's statement pertaining to Tito's work is irrelevant to our own paper. Referee 3 (statistician) 3.1 States it is 'tedious' to read numbers and percentages, instead inviting confidence intervals in parentheses. We prefer the style of numbers and percentages in the text and note that the Journal has accepted this approach in the past. We are happy to respond to specific editorial advice on this matter. 3.2 Requests tables include only the percentages with N in the title. This does contradict suggestions made by referee 1. 3.3 Suggests it is 'even more tedious' to present the chi square values, degrees of freedom and p values. We disagree, as epidemiologists find the degrees of freedom informative as well as reassuring while the size of the chi square indicates variation. 3.4 States the use of McNemars in tables 1 and 2 is not included in method. A sentence now has been added. 3.5 Requests replacement of the work 'indicate'. We argue that respondents do indicate on paper. It is less plausible to claim we know what they 'think' .. we have edited. 3.6 States 'some of the strategies in Table 4 need to be worded more clearly'. Modified without compromising the words used in the survey. 3.7 Suggests in paras 12 and 47 that the word 'unfounded' is unjustified because 'there is clearly no legal evidence or precedent on which anyone can predict the real-life outcome of the second-case scenario'. These paragraphs have been revised. 3.8 States concern about the term 'government', inferring we meant 'government legislation'. This is an unintended interpretation. We have modified the sentence to reduce misunderstanding. Editorial suggestions E.1 Asks us to 'think about' the use of the word 'screening' in the title See A.1 E.2 Requests inclusion of study timing in Abstract Included. E.3 Invites us 'to consider making some changes' in the introduction in response to referees' comments. See 1.4, 1.5, 2.3 and 2.4. E.4 Asks why 64 GPs were ineligible. See 1.8. E.5 Invites us to present more information on the demographic representativeness of the sample. We now have written as follows: 'Our sample comprised 161 male respondents (74%) and 55 female respondents (26%) compared with 68% and 32% respectively for the NSW reference sample23. Median age of our respondents was 47 years (range 28-70 years); half had been in general practice for more than 17 years (range 1-47 years) and 127 (58%) worked in the Sydney metropolitan area, somewhat comparable with the NSW reference sample (51% aged 45 years or over and 70% practising in Sydney)23. As only proportions for the NSW sample have been published23, formal statistical testing of response bias was precluded.' For the editor's information, we have attached to this letter a full table describing the demographic characteristics. We would be happy to include it as an additional table if you feel this necessary. We inadvertently implied that there were significant differences between our sample and NSW data. These analyses cannot be performed validly because of the size of the reference sample. We were however pleased to not have suffered the response bias from male GPs in our sample. Perhaps we ought to have said this more clearly. In any case, we hope it now is clear. E.6 Suggests Table 4 is 'probably not necessary' Now deleted and key findings presented in text. E.7 Suggests Table 5 should be 'abbreviated to include only the most interesting / necessary responses'. Incorporated. E.8 Requests revision of discussion and shortened to comply with 2500 word limit. The manuscript is now 2498 words. On behalf of my co-authors, I have enclosed a revised paper, with changes highlighted in bold in the text. I look forward to hearing from you soon. A/Prof Jeanette Ward |
| Thu Apr 1 15:47:25 1999 | Thank-you for submitting your revised article and for your patience in awaiting my reply which has been delayed by some deliberations here in the editorial office.
Your responses to the referees' comments have enhanced the paper but there are still a couple of issues which need attention. The definition of testing an asymptomatic patient for a disease, at their own specific request, as "screening" touches on what seems to us to be an ill-defined area. It is interesting to me that the guidelines on prostate cancer screening are not really helpful in this regard. On page 1 of the AHTAC guidelines (your ref 19) it does say" "Three approaches to screening were considered: Mass screening (comprehensive testing of the male population), screening of selected population groups (eg those in high risk categories) and case finding (screening as part of a routine medical checkup)". Therefore case finding as a type of screening is the closest they come to the scenario presented to the GPs you surveyed, but the guidelines are unhelpful as to what doctors should do when faced with individual patient requests for testing (as mentioned by Dr Pincus), apart form suggesting that such patients should be fully informed. You have examined GPs beliefs and intended behaviours with regard to one type of screening but what would their responses have been to a scenario of unsolicited or mass screening? You need to make it clear in your abstract, introduction and discussion that you looked at one particular type of screening. We would also like you to provide your working definition of screening. This could be presented in a box so everyone knows what we are talking about. I would be very interested in the referees' opinions on this issue. We agree with all the statistician's comments on the presentation of your results, and would very much like to see 95% confidence intervals instead of the chi-squared results We would like you to revise the manuscript along these lines and, as mentioned, invite comments from the reviewers. If possible, we would like to have your comments/ amendments within the next two weeks. Hope everyone has a refreshing Easter break Ruth Armstrong Ruth Armstrong |
| Mon Apr 12 11:33:44 1999 | Further comments from "F"
Title:
|
| Wed Apr 21 12:14:10 1999 | The authors have chosen to largely ignore all the recommendations of the statistical reviewer with respect to presentation. The annoying triplets (chi-squared=0.81, 1df, p=0.4) are all still included; not a single confidence limit has been calculated; and the tables are still unnecessarily dense because of the authors' insistence on reporting absolute numbers with percentages alongside in parentheses, when only the precentages are really meaningful. [The total N could be shown in the title of each table (or at the end of each row-descriptor of reponse items for which missing values are perceived to be a problem....- although inspection of the tables suggests that missing values are not a big problem).] While the authors have clarified their meaning of 'ineligible' (par27) in their response letter, they have not modified the paper - so many readers will ask the same question as this reviewer. The word 'ineligible'implies some returned questionnaires were actually excluded from the analysis - while the explanation in the cover-letter makes it clear that many potential respondents were simply uncontactable. The authors have refused to implement any of the recommendations made by this reviewer with respect to presentation. The decision on whether to enforce any changes must therefore lie with the Editorial Committee. None given |
| Mon May 3 13:08:54 1999 | Thank you for your further email dated 1 April in which you acknowledge that our revisions 'have enhanced the paper'. We also received comments from 'F' dated 12/04/99 by email and a subsequent email with further statistician's comments dated 21/04/99. As advised by telephone on 16 April, we required longer than the two weeks initially requested to respond to all of these comments.
We respond as follows: Editorial suggestions E.1 States 'the guidelines on prostate cancer screening are not really helpful in this regard', requesting we clarify our use of the term 'screening' in a Box That the AHTAC guidelines are not helpful in this regard is part of the problem. We have been quite specific in this and other publications about the concept 'screening'. Case finding equates with screening. In view of the widespread misunderstandings, we have added a Box which explicitly works through an evidence-based logic for screening. We believe MJA readers will find this invaluable as a commentary and thank you for suggesting it. E.2 Requests 'would very much like to see 95% confidence intervals instead of the chi-square results' All estimates now have 95% confidence intervals. The 'N' have been deleted throughout. All chi-squares in the text have been replaced with ORs and 95% CIs around the ORs. However, paired McNemar's chi square is not conventionally converted to ORs so we have retained their use in tables. Further comments from 'F' F.1 Suggests a new title We have added 'scenarios' but reject the rest of the reviewer's suggestion. No other expert reviewer seemed perturbed. F.2 Suggests 'testing' would be appropriate in Abstract. We reject this as inappropriate interference. See E.1 F.3 Suggests changes to Results. Results has been changed. The problematic data have been deleted. F.4 Suggests we retain the word 'screening' in the second sentence of Results section of the Abstract AND also retain the word screening in the first sentence of the Discussion of the Abstract. That this reviewer has made these endorsements strengthens the case for our use of the word 'screening'. F.5 Suggests replace 'screening' in paras 15 and 16 of the Introduction. We reject this as inappropriate interference. F.6 Insists we cite American guidelines pertinent to the US physician survey. We cannot mention or cite everything recommended by reviewers without exceeding our word count. Interested readers can obtain the physician survey article and follow-up their own interest in US guidelines. We have cited the US Preventive Services Task Force guidelines as an authority for the Box. F.7 Rejects our response 1.8 regarding ineligibility data. We have now included an additional 19 words (deceased 2; retired 9; not in general practice 25; on extended leave 5; moved overseas 8; uncontactable 15). F.8 Requests para 30 and 31 to be 'even clearer' These paragraphs are clear. F.9 Requests the words 'screen, screening and screen' need to be addressed in Results and Discussion. We reiterate our view that 'F's comments are opinion only and would be usefully exchanged publicly with us through the Letters column. We have used the term 'screening' consistent with our principles now explictly outlined in the Box. See E.1. Statistician's second set of comments S.1 States 'the annoying triplets (chi squares, df, p values) are all still included' See E.2. S.2 States 'the total N could be shown in the title of each table'. These were included in the titles of the tables in the revised manuscript. S.3 States 'they have not modified the paper' with regard to reasons for ineligibility See F.7. E.4 Claims 'the authors have refused to implement any of the recommendations made by this reviewer'. We deny our position is one of outright 'refusal' as perceived by this reviewer but considered our responses to have been thoughtful and open to negotiation with the editor. We enclose an electronic version of the revised manuscript. On behalf of my co-authors, I look forward to hearing from you soon. Yours sincerely CSAHS Needs Assessment & Health Outcomes Unit A/Prof Jeanette Ward PhD FAFPHM Director None given |
| Fri Jul 2 10:46:13 1999 | Editorial Committee decision re the revised manuscript Potential impact (using case scenarios) of guidelines discouraging prostate cancer screening on general practitioners' perceptions of medicolegal risk.
Thanks everyone for your contributions to this manuscript, and for your patience. Because of the unresolved disagreement between one of the reviewers and the authors, we asked our Content Review Committee to comment. The committee met last Thursday (June 24). The general feeling was that this is an important paper which should be published. The committee also believed that the case scenario outlined, constituted clear cut screening. I have also been waiting for the comments of another epidemiologist which I will forward to the authors when they arrive. However, rather than hold things up further I am happy to inform you all today that we are accepting the paper for publication. It will be published immediately on our website and will the edited and published in print in the next few months. I do apologise for the delay in our decision-making process. It is our hope that the "Internet forum" format will improve quality by allowing more author-reviewer interaction but in this case we felt we needed an expert adjudicator. Thanks again, everyone, for your participation. Ruth Armstrong (assistant editor, MJA) |