Comparison of patient recall of hormone therapy with physician records.

OBJECTIVE: The objective of this study was to compare patient recall of menopausal hormone therapy, obtained by personal interview, with that recorded in physicians' records.
DESIGN: Data were from a case-control study of women aged 55 to 64 years at diagnosis of breast cancer in Los Angeles County, CA, between 1987 and 1989. Medical record information on 1,115 women reporting hormone therapy use (652 case patients and 463 control participants) was sought from 2,734 physicians and obtained from 1,350. Physician records for an additional 126 case patients and 90 control participants not reporting hormone therapy but reporting oral contraceptive use or benign breast disease were also obtained.
RESULTS: Agreement between interview and physician records was 85% (kappa=0.59) for ever/never use of estrogen and 84% (kappa=0.59) for progestogen. The correlations for months of use were 0.43 and 0.34, respectively. For both drugs participants tended to report longer use, primarily because they reported earlier starting ages. Older and unmarried women recalled more unconfirmed estrogen use, whereas naturally menopausal women were more likely to report no estrogen use when the medical record was positive. We found no evidence of preferential recall bias by case patients.
CONCLUSIONS: Interviews provide a moderately reliable measure for ever use of postmenopausal estrogen and progestogen. Recalled details are less reliable and may be affected by age, type of menopause, and recall interval.