Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Further analysis would be interesting
- Frank Buntinx (25 August 2000)
-
Re: Further analysis would be interesting
- John P Heptonstall (28 August 2000)
-
Backsliding into Correlations
- Alexander M Walker (6 September 2000)
-
Author's reponse
- Richard Farmer (14 September 2000)
|
|
|||
|
Frank Buntinx
Send response to journal:
|
The relation between third generation combined oral contraceptives and the emergence of venous thromboembolism among women may have major economical consequences. In such situation it is always interesting to see the effect of the product on data that have been collected routinely and independently of the research question at hand and according to rules that were defined in tempero non suspecto. The study based on the General Practice Research Database provides such results (1). It therefore is helpful to the discussion. The authors describe no decrease of the age-stratified incidence of venous thromboembolism after the fall in use of third generation combined oral contraceptives in 1995. The increase that has been described, however, is especially related to the first months of intake of the new contraceptive (2). It therefore would be helpful if the authors would be able to relate the emergence of venous thromboembolism in women to the first months of use of third generation combined oral contraceptives. In order to additionally enlighten the effect of the duration of intake, such analysis could be presented consecutively using the first one, three, six and twelve months to define the exposure status. The General Practice Research Database seems large enough to permit the analysis. It therefore is my hope that the authors will be able to present these data. 1. Farmer RDT, Williams TJ, Simpson EL, Nightingale AL. Effect of 1995 pill scare on rates of venous thromboembolism among women taking combined oral contraceptive: analysis of General Practice Research Database. BMJ 2000; 321: 477-479. 2. Vandenbroucke JP, Bloemenkamp KWM, Rosendaal FR, Helmerhorst FM. Incidence of venous thromboembolism in users of combined oral contraceptives: Risk is particularly high with first use of oral contraceptives. BMJ 2000; 320: 57-58. |
|||
|
|
|||
|
John P Heptonstall, Director of the Morley Acupuncture Clinic and Complementary Therapy Centre West Yorkshire
Send response to journal:
|
Editor I find it rather worrying that such an eminent team as Farmer et al appear to have omitted extremely important observations and analyses from their study - surely it is common knowledge that 'pill' medications tend to affect recipients very early in use, and/or very late in use, and this perspective should have been accounted for in the research? I also find it reprehensive that such studies are funded by pharmaceutical companies whose main interests include protecting their drug culture and image, perhaps out of necessity when governments fail to respond on behalf of an unsuspecting public to protect that public by ensuring funding is provided from essentially 'untainted' sources. I also would welcome the additional perspective requested by Prof. Dr. Buntinx, as I think will every 'pill'-taker throughout the world. Regards John H. |
|||
|
|
|||
|
Alexander M Walker, Professor of Epidemiology Harvard School of Public Health
Send response to journal:
|
In the distant past public health analysts turned to time trend analyses because there was little else an epidemiologist could do quickly. Lack of specificity, a host of competing explanations for the results, and an inability to control for individual-level confounders were balanced against a short turnaround and the certainty that the results would be taken with a grain of salt. "Hypothesis generating" correlational studies did occasionally give rise to interesting ideas, but they were never held out as proof. Farmer and his colleagues, having access to individual-level data on exposure, outcome, and covariates in the GPRD, have nonetheless produced a two-period time series anlaysis, and they have ignored most of the information that they could have uncovered. They lumped together all combined oral contraceptive use, and they forced calendar time to stand as the only proxy for exposure. And they did this with a problem that is now half a decade old and already burdened with scores of learned and contradictory commentaries. Rather than remind the reader of the fragility of their method or the peculiarity of performing such a weak analysis on potentially such powerful data, Farmer et al. have drawn unambiguous conclusions. They maintained that their "findings are not compatible with the assertion that third generation oral contraceptives are associated with a twofold increase in risk of venous thromboembolism compared with older progestogens." The findings are indeed compatible with the cited assertion, and with many others. Simple analyses have rhetorical power the exceeds their scientific merit. BMJ readers should be careful not to confuse quick-and-dirty with short-and-sweet. |
|||
|
|
|||
|
Richard Farmer, Professor of Pharmacoepidemiology and Public Health Postgraduate Medical School, University of Surrey
Send response to journal:
|
Walker (Rapid response 6 September 200) refers to our recent study (1) as a "time trend analysis". This is a misrepresentation of the investigation. The study was designed to measure the effect of the CSM's intervention on the pattern of combined oral contraceptive (COC) prescribing. In October 1995 the CSM issued a general warning that so- called third generation COCs doubled the risk of venous thromboembolic disease (2). In order to reduce the incidence of VTE amongst users of COCs they recommended that their use should be restricted. McPherson reviewed the studies that led to the CSM's decision and concluded that the increased risk was real (3). Furthermore, he defended the CSM's decision to issue a warning before the publication of the pivotal studies (4,5), and before the completion of one of them (6). He said "delaying its announcement until all the studies had been completed would have incurred a further 500,000 uninformed women years of use of third generation oral contraceptive pills, which might have resulted in 80 new cases of venous thromboembolic disease". It is quite clear that the purpose of the recommendation to avoid the use of the suspect oral contraceptives was to reduce morbidity and mortality amongst women using COCs. The response to the CSM's recommendation by doctors in the UK was rapid and substantial. Under these circumstances it would be reasonable to expect that there would be a reduction in VTE amongst COC users - in effect the CSM's recommendation was a community intervention, as such it was worthy of investigation. Our study compared the rate of VTE amongst users while 'third-generation' products contributed 53% of total use with a period during which they contributed 14%. Walker is correct when he says that we have access to individual level data on exposure outcomes and co- variates. It was because we had individual data that we were able to ascertain the exposure of each individual case of VTE and, as we stated in our paper, we were able to exclude women not using a COC at the time of their VTE and non-idiopathic cases. It follows that Walker is wrong in stating that "they forced calendar time to stand as the only proxy for exposure". In fact, exposure was precisely identified for each case. We found no difference in the incidence of VTE amongst COC users between the two periods. It is possible that the absence of a reduction in VTE following the 1995 'Pill Scare' was for reasons other than there being no difference in risk between the two types of COC. It could have been due to a change in the characteristics of women being prescribed COCs. Were this to be the case then it would be necessary to hypothesise that after the 'Pill Scare' a greater proportion of women prescribed COCs were in high-risk groups than before the 'Pill Scare'. For example, a greater proportion had a high BMI. There is no evidence to support this, moreover the effect would have to have been very large to mask the hypothesised fall in VTE rates. It has been suggested that new starters are at greater risk than established users. Our findings could only have been affected by changes in the proportion of new starters if the proportion had increased in the post-scare period. In an earlier paper we report that the greatest fall in use of COCs was amongst young women, a group with the highest proportion of new starters (7). Contrary to Walker's assessment we do not believe that the method is fragile or that the analysis is weak. We believe that we were correct in our conclusions. It is essential that epidemiologists remain circumspect in the interpretation of the findings from population-based studies, as none is perfect. Special caution is required when considering low odds ratios of borderline statistical significance for rare outcomes. In this particular instance it is difficult to explain to the General Public the justification for disrupting the lives of several million women when the benefit, if any, is unmeasurable. Richard Farmer, Professor of Pharmacoepidemiology and Public Health Tim Williams, Research Fellow Emma Simpson, Research Officer Alison Nightingale, Research Officer 1. Farmer RDT, Williams TJ, Simpson EL, Nightingale AL. Effect of 1995 pill scare on rates of venous thromboembolism among women taking combined oral contraceptives: analysis of General Practice Research Database. BMJ 2000; 321: 477-479 2. Committee on the Safety of Medicines. 1995. Combined oral contraceptives and thromboembolism. CSM, London. 3. McPherson K. Third generation oral contraception and venous thromboembolism. BMJ 1996; 312: 68-69 4. Poulter NR, Chang CL, Farley TMM, Meirck O, Marmot OG. World Health Organisation collaborative study of cardiovascular disease and steroid hormone contraception. Venous thromboembolic disease and combined oral contraceptives: results of international multicentre case-control study. Lancet 1995; 346:1575-82 5. Jick H, Jick SS, Gurewich V, Myers MW, Vasilakis C. Risk of idiopathic cardiovascular death and non-fatal venous thromboembolism in women using oral contraceptives with differing progestogen components. Lancet 1995; 346: 1589-93 6. Spitzer WO, Lewis MA, Heinemann LA, Thorogood M, MacRae KD. Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case-control study. Transnational Research Group on Oral Contraceptives and the Health of Young Women. BMJ 1996; 312:83-8 7. Farmer RDT, Lawrenson RA, Todd J-C et al. Oral contraceptives and venous thromboembolic disease. Analysis of the UK General Practice Research Database and the UK MediPlus Database. Human Reproduction Update 1999; 5 (6): 688-706 |
|||