Measuring miscarriages: a comparison of self-reported and healthcare register data in Finland

Abstract Background Approximately one in four women experiences a miscarriage, which may affect fertility intentions and health. However, there is a lack of research on the topic, in part due to poor quality of miscarriage data. Both surveys and healthcare records likely miss some miscarriages. We compare data from self-reports and healthcare records at the individual level to understand to what extent these data differ and according to which socio-demographic characteristics. Analyses of the level and patterns of miscarriage reporting can improve our understanding of miscarriage experiences and enable more reliable estimation of miscarriage risk factors in future studies. Methods We used healthcare records on miscarriages in Finland from 1969 to 2021 linked to self-reported miscarriage data in antenatal appointments from 1998 to 2021, as well as in a national epidemiological survey, FinHealth, collected in 2017. The antenatal data were available for parous women, whereas the survey also captured those without children, but had a smaller sample size. We compared the differences between the records overall and by socio-demographic characteristics using descriptive statistics and multinomial regressions. Results For over 80% of people, there were no differences between self-reports and healthcare records. This was mainly driven by those who had never had a miscarriage, as this share dropped to 40% when only those with at least one miscarriage in one data source were analysed. We observed some self-reporting fewer (1–4% among full sample, 7–13% among those with a miscarriage) and some more miscarriages than in healthcare records (8–12% among the full sample, 46–56% among those with a miscarriage). Differences in both directions were associated with older age, higher parity, and lower education. Conclusions Healthcare records and self-reports overlap partly in the miscarriages they capture. More miscarriages were observed using self-reports, most likely due to some miscarriages occurring outside the healthcare system, and differences in miscarriage definition between lay people and healthcare professionals, with the latter having a more restrictive definition. Future miscarriage studies should carefully consider the strengths and limitations of each type of source against their research aims.