The Impact of Abortion Underreporting on Pregnancy Data and Related Research

INTRODUCTION: The impact on research findings that use pregnancy data from surveys with underreported abortions is not well-established. We estimate the percent of all pregnancies missing from women’s self-reported pregnancy histories because of abortion underreporting. METHODS: We obtained abortion and fetal loss data from the 2006–2015 National Survey of Family Growth (NSFG), annual counts of births from US vital statistics, and external abortion counts from the Guttmacher Institute. We estimated the completeness of abortion reporting in the NSFG as compared to the external counts, the proportion of pregnancies resolving in abortion, and the proportion of pregnancies missing in the NSFG due to missing abortions. Each measure was examined overall and by age, race/ethnicity, union status, and survey period. RESULTS: Fewer than half of abortions (40%, 95% CI 36–44) that occurred in the five calendar years preceding respondents’ interviews were reported in the NSFG. In 2006–2015, 18% of pregnancies resolved in abortion, with significant variation across demographic groups. Nearly 11% of pregnancies (95% CI 10–11) were missing from the 2006–2015 NSFG due to abortion underreporting. The extent of missing pregnancies varied across demographic groups and was highest among Black women and unmarried women (18% each); differences reflect both the patterns of abortion underreporting and the share of pregnancies ending in abortion. DISCUSSION: Incomplete reporting of pregnancy remains a fundamental shortcoming to the study of US fertility-related experiences. Efforts to improve abortion reporting are needed to strengthen the quality of pregnancy data to support maternal, child, and reproductive health research.