Agreement between oral contraceptive users and prescribers: implications for case-control studies
Van Leeuwen, F.E.; van Duijn, C.M.; Camps, M.H.; Kempers, B.A.; Mentjens, M.F.; Mulder, H.B.; Schouten, E.G.; Zwijsen, R.M.; Rookus, M.A.
Contraception 45(5): 399-408
1992
ISSN/ISBN: 0010-7824 PMID: 1623713 DOI: 10.1016/0010-7824(92)90154-lDocument Number: 318861
Between January 1983-January 1986, women spoke to 151 breast cancer patients at the National Cancer Institute in Amsterdam, the Netherlands and 93 healthy controls to determine agreement between women's lifetime histories of oral contraceptive (OC) use and prescribers' records. Both cases and controls provided the name of all physicians who ever prescribed an OC to them. Even though 93.6% of prescribers responded to the researchers request for information, only 46.1% of the forms provided information on brand name or date(s) of usage. In fact, only 33.6% had complete information. Median duration of OC use was 8.1 years and 33.1% began taking OCs before July 1965. Patient-prescriber agreement on brand names and dosage stood at 70%. Responses of starting dates of about 50% of the women corresponded within 1 year with that of their prescriber. Breast cancer patients tended to have better agreement than controls, but the difference was not significant. Further 59.1% agreed with their prescriber within 2 years on duration of use. The difference in agreement on duration of OC use between cases and controls was insignificant. The multiple regression analysis showed that women of low socioeconomic status were less likely to agree with their prescribers on brand names and dates of usage than were women of medium and high socioeconomic status. Further healthy women had better agreement than breast cancer patients. Moreover the recall period had a significant effect on agreement on brand names (p=.07) and on stopping dates (p=.0015). Even though women tended to overstate duration of OC use, recall was accurate enough to explore the presence of a moderately strong duration response relationship. This study showed that prescriber data should not be the only data source of a case control study and that they indeed complement women's histories.