[Estimating the frequency of induced abortion: a comparison of two methods].

Objective: To compare two methods-the "ballot box" method and the "indirect questioning" method-for estimating the frequency of induced abortions in population-based studies.

Methods: A cross-sectional population-based study was conducted with a representative sample of 3 002 women between 15 and 49 years of age living in the city of Pelotas, Rio Grande do Sul, Brazil. The women were selected through multistage sampling and randomly assigned to answer questions concerning induced abortion with one of the two methods, after they had answered a general questionnaire that collected socioeconomic and demographic information. With the "ballot box" method, women received a small piece of paper containing direct questions on abortion. Each woman marked her answers on the paper and then deposited it into a small "ballot box" carried by the interviewer, thus assuring the confidentiality of the responses. With the second method, the interviewer verbally asked the woman a series of questions that indirectly inquired about abortion.

Results: Among the women assigned to the ballot box method, 7.2% reported having induced at least one abortion, versus 3.8% of the women assigned to the indirect questioning method. The ratio between the two methods was 1.89 (95% confidence interval: 1.39 to 2.60; P < 0.001). Of the women surveyed who were 45-49 years old (at the end of their reproductive life) and answered using the ballot box method, 12% of them reported having had at least one induced abortion in their lifetime. Among the women who answered using the ballot box method, the main reasons they gave for having had an abortion were economic difficulties, being too young, and being single. Approximately half of the women responding to the ballot box method questions reported they had used unsafe procedures to induce abortion, with 13% of them using misoprostol.

Conclusions: The ballot box method was suitable for studying the frequency of induced abortion and for researching other topics that may lead to underreporting by the persons interviewed, especially in population-based samples. Rephrasing the ballot box method question about abortion (for example, eliminating the word "child") would probably increase the method's sensitivity, that is, its ability to detect induced abortions that had occurred.