The role of prescribing and referral bias in studies of the association between third generation oral contraceptives and increased risk of thromboembolism

Recently published papers have shown a slightly increased risk for venous thromboembolism (VTE) among users of third generation oral contraceptive pills (OC) as compared with users of second generation OC. The extra risk is small (about twofold) and could be explained possibly by biases introduced into the original case–control studies. This paper examines the possible role of prescribing bias (i.e. that doctors would prescribe third generation OC preferentially to women whom they knew to be at risk of VTE) and referral bias (i.e. that doctors would preferentially refer women on third generation OC to hospital for investigations of symptoms suggestive of VTE). A cross‐sectional survey was carried out of 106 physicians, to record their attitude to the prescription of OC in certain risk situations. Each doctor then extracted data on OC prescription from approximately 12 patients under his care. A total of 1192 patients were included. Data collected related to known risk factors to which the patient was prone, and the type of OC that the woman was taking. The results from the doctors' attitudinal survey showed that third generation oral contraceptives seemed to be very popular among doctors in England and that they would prefer to prescribe third generation to second generation oral contraceptives in virtually all risk situations, and also in situations where there was no particular risk of VTE. The results from the patients' survey, however, did not correspond with the attitudinal survey, and there was no evidence of a systematic bias running through this sample of patients, associating third generation OC prescription with particular risk factors. The results from the doctors' attitudinal survey on referral did not show any evidence of potential bias towards referring patients on third generation OC to hospitals. Comparison of this survey with an equivalent survey done in Germany suggests that there may be different factors governing doctors' behaviour in Germany and England, with reference to choice of OC prescription. © 1998 John Wiley & Sons, Ltd.