Authors’ replyBMJ 2010; 340 doi: https://doi.org/10.1136/bmj.c2525 (Published 11 May 2010) Cite this as: BMJ 2010;340:c2525
- Philip C Hannaford, Grampian Health Board chair of primary care1,
- Lisa Iversen, research fellow1,
- Tatiana V Macfarlane, senior research fellow2,
- Alison M Elliott, senior research fellow1,
- Valerie Angus, data manager3,
- Amanda J Lee, professor of medical statistics4
- 1Centre of Academic Primary Care, University of Aberdeen, Foresterhill Health Centre, Aberdeen AB25 2AY
- 2Division of Applied Medicine, University of Aberdeen, School of Medicine and Dentistry, Aberdeen AB25 2ZD
- 3College of Life Sciences and Medicine, Aberdeen AB25 2ZD
- 4Medical Statistics Team, Section of Population Health, University of Aberdeen, Aberdeen AB25 2ZD
Brind thinks that common sense suggests that we should use the smaller, better, dataset when statistical trends differ materially.1 We presented both datasets so that readers can decide which one(s) they wish to use. Neither dataset showed substantial evidence of an increased risk of all cause mortality among ever users.
A higher proportion of observation periods in ever users related to current and recent pill use in the general practice observation dataset compared to the main dataset. This explains the increased risk of all circulatory disease and of breast cancer in ever users who had stopped oral contraception 5-9 years previously in the general practice observation dataset. Similarly, in the age stratified analyses of the main dataset, ever users under 30 had much more current and recent use than older ever users. That ever users under 30 had nearly three times the risk of death of similarly aged never users is therefore not surprising.
Most of the adverse mortality effects of oral contraception occur in current and recent users, effects which diminish with time since stopping.2 Our latest findings are compatible with our previous publications, and broadly in line with the collective evidence from other studies. A substantial proportion of the observation periods in the ever user group in the main dataset relates to pill use many years in the past. Unlike Brind, we think that the information provided by this dataset is valuable. We continue to interpret the results as not suggesting a substantial increased overall risk of death among ever users, especially since many events occurred long after the pill was stopped. Thus any medical treatment and death certification is unlikely to have been influenced by the doctor’s knowledge of a woman’s pill use.
We have tried to be careful when interpreting our results. Thus, we have stated clearly that the reduced overall risk of death in ever users in the main dataset may be due to selection processes or residual confounding rather than a direct effect of oral contraception.
Cite this as: BMJ 2010;340:c2525
Competing interests: The Centre of Academic Primary Care has received payments from Schering Plough and Wyeth Pharmaceutical for lectures and advisory board work provide by PCH.