Agreement rates between oral contraceptive users and prescribers in relation to drug use histories

Abstract

A comparison between prescriber records and patient-reported oral contraceptive histories obtained during a case-control study of thromboembolism and oral contraceptive use served as the basis for evaluating the extent of agreement between these two sources of information. Agreement between oral contraceptive user and prescriber was highest on the name of the most recently-used product (89% agreement), and dropped to 62.5% on the name of the product taken before the more recent one. For total duration of oral contraceptive use (to within one month of hospital admission), agreement (defined as differences not exceeding one month) was 36%, while 39% showed the users reporting longer duration of use, and 25% showed users reporting shorter duration. Cases showed a higher rate of agreement with prescriber records than the controls on both the name of the product and the total duration of use. Disagreement from prescriber records in the direction of reporting longer duration of use was 45% for the controls compared to 27% for the cases. These results indicate that while user information in connection with the most recently used oral contraceptive can be acceptable in studies employing the case-control research strategy, user reports about previous oral contraceptive use may be less useful.

PIP: A case-control study of thromboembolism and oral contraceptive (OC) use provided information on prescriber records and patient recollection of previous OC use. These 2 information sources were compared to obtain agreement rates between prescriber- and patient-reported OC histories. Agreement between the 2 sources was best on the name of the most recently used preparation (89% agreement); this dropped to 62.5% when comparing for the name of the product taken before the more recent one. Agreement on total duration of OC use (to within 1 month) was 36%, whereas 39% reported longer and 25% shorter duration of use. Cases (women with thromboembolism) showed a higher agreement rate with prescriber records than the controls (admitted to hospital for other, nonrelated reasons) on both the product name and total use duration. Disagreement from prescriber records in the direction of reporting longer duration of use was 45% for controls compared with 27% for cases. These results were analyzed in terms of their effect on research methodology (specifically recall bias problems), and it was concluded that while user information in connection with the most recently used OC can be acceptable in studies using case-control strategies, user reports about previous OC use may be less useful and more suspect.