Background. There are several methodological problems when studying contrac
eptive efficacy. The present study focuses on efficacy issues from the use
of oral contraceptives (OC).
Methods. Through a computerized system of medical records kept at the Depar
tment of Gynecology, Regional Hospital of Trondheim, Norway 800 OC-failure
pregnancies were identified among 8192 women having pregnancy termination f
rom January 1, 1987 to December 31, 1996. Ascertainment of cases was done b
y validating hospital data on OC failures by data collected by the referrin
g practitioner. Inconsistencies in failure data were found among 257 of the
800 (32%) women who claimed at the hospital that an OC was in use at time
for conception. The medical record kept by their practitioners stated that
158 women had stopped using OC before last menstrual period, of the remaini
ng 99 women neither data on actual use nor data on prescription could be co
nfirmed.
Results. Over the 10-year study period the proportion of failures by brand
of OC was remarkably constant from year to year and reflected the user patt
ern of OCs in Norway. One-third of the total number of failures (n=523) wer
e claimed to be method failures. The annual overall OC-failure rate was est
imated to be 1% during the 10-year study.
Conclusion. Case-control studies focusing on efficacy of OCs must be carefu
lly designed and comprise both women carrying a pregnancy to term as well a
s women terminating their pregnancies. Our study focused on only one segmen
t of the population of women experiencing an OC failure: women having pregn
ancy termination. The present study has shown that exposure data collected
from interviews have to be validated against prescription data.