Selection bias in a study on how women experienced induced abortion.

OBJECTIVE: One third of the women who had an induced abortion at the department of gynecology and obstetrics in Malmö 1989 refused to participate in an interview a year later to explore their experience and the care they had received. Using data from the mandatory preoperative visit it was then possible to compare participants with non-participants with regard to socio-demographic characteristics. reproductive history and stated reason for abortion. By continued follow-up of medical records it has furthermore been possible to ascertain how many women conceived within a year, and how many of them applied for another abortion and how many elected to continue the pregnancy to term.
SETTING: Department of Obstetrics and Gynecology, Lund University, University Hospital, Malmö Sweden. This is the sole referral hospital serving the population of Malmö (approx. 230000).
SUBJECTS: All 1285 women who underwent induced abortion at the department in 1989. + STUDY
DESIGN: In each case information on socio-demographic characteristics, reproductive history and stated reasons for abortion was collected at the mandatory clinical visit prior to the abortion. This information was used for comparison of participants and non-participants in the planned 1-year follow-up interview.
RESULTS: Young, unmarried women of low educational status and without full-time employment or studying were overrepresented in the non-participant group. The proportion of women with children was however smaller in that group. Within 12 months after the abortion, 118 women 66 (7.7%) of the participants in the follow-up interview and 52 (12%) of the non-participants conceived again but elected to continue the pregnancy to term (P<0.05). whereas 124 of the women 80 (9.5%) and 44 (10.2%) of the respective subgroups again applied for abortion within 12 months.
CONCLUSION: One third of the women who underwent induced abortion did not wish to be interviewed about their emotional and somatic experience of the abortion 1 year later. Non-participation at the follow-up interview was associated with socio-demographic factors which in studies about other medical problems have been shown to be associated with increased vulnerability and morbidity. Non-participation was also associated with an increased childbirth rate during the following 2 years. The large proportion of non-participants is a matter for concern as to some extent it reflects inability of the health care system to establish trustful relationships with these women. How the participants in the follow-up study experienced this relationship will soon be published in our next paper. Support from the women's surroundings and the adviser the women met at the public health care system clearly influenced the women's psychological reactions after the abortion. It is therefore important to ascertain to what extent the management of abortion applicants in fact fulfils their individual needs and expectations.