Completeness and accuracy of reporting induced abortions performed in Hawaii's hospitals, 1970--74.

This study examined the completeness and accuracy of reporting induced abortions to the official agency by following up documented cases of abortions performed in hospitals in Hawaii for the years 1970--74. The following two sets of records were initially linked by computer on the basis of first name, maiden name, date of the event, age, and ethnic background: (a) a total of 18,531 induced abortions ascertained in hospitals with complete linking information and (b) official fetal death and live birth records filed at the State health department. Overall, 96.1 percent of the hospital abortions were matched in the official certificates. Of 17,550 abortions classified as elective, 98.7 percent were so recorded in the certificate file, whereas the remainder were reported as mostly either therapeutic or unknown for cause of fetal death. When the probability of matches for those abortions without complete linking information and possible errors in the information were considered, the estimates of underreporting to the health department were a maximum of 3.9 percent and a minimum of 1.2 percent; the mean value was 2.6 percent. The extent of underreporting of induced abortions is becoming serious with the widespread practice of abortion procedures in nonhospital institutions in recent years. The results of the study also demonstrated the utility of the record-linkage process in checking data for completeness and accuracy or studying separate vital events on the same persons.