Selection bias in studies of spontaneous abortion among occupational groups.

The purpose of this investigation was to determine whether in a cross-sectional study the miscarriage rate among female employees might be overestimated if the study population included only those women who were employed at the time of the investigation. Data were derived from the experience of 556 female laboratory workers at the University of Gothenburg (UG) who had ever been pregnant. Seventy five percent of the women who had been delivered of a child while working continued to work during the next pregnancy, while 91% of those who had had a miscarriage continued to work during the next pregnancy (P = .002). Women who terminated their employment at UG before 1979 had a lower miscarriage rate (5.9%) than those who were still working at that time (11.4%). The miscarriage rate was lowest among those who had stopped working in connection with a successful pregnancy (4.4%). It is concluded that selection bias dependent on the outcome of previous pregnancies may influence the frequency of miscarriages observed in cross-sectional studies.