Abortion laws are often discussed as questions of rights, ethics and public policy, but they also create natural experiments in human decision-making. A new analysis of Japan’s 1991 reduction in the legal gestational limit for abortion suggests that changing the deadline did more than prevent procedures after a particular week. It appears to have shifted the timing of abortions toward earlier stages of pregnancy and was associated with fewer abortions among women who had already reached the nineteenth week.
The study, published in Nature Human Behaviour, examined what happened after Japan reduced its legal gestational limit from 24 weeks to 22 weeks of pregnancy. Researchers Yanagawa, Piedvache, Okubo and colleagues analyzed national records covering abortions, stillbirths and live births at gestational ages of at least 12 weeks. The dataset included 7,364,783 pregnancy outcomes recorded between 1989 and 1994, allowing the researchers to compare patterns before and after the legislative change.
The question is difficult to answer using ordinary observational data because abortion decisions are influenced by many factors at once. Economic conditions, access to healthcare, changes in contraception, public attitudes and improvements in prenatal diagnosis can all affect when abortions occur. To isolate the relationship with the law, the researchers used two quasi-experimental methods: an interrupted time-series analysis and regression discontinuity analyses. These approaches do not create a randomized experiment, but they can identify abrupt changes in trends that coincide with a policy intervention while reducing the influence of gradual changes over time.
The most striking result appeared at the legal boundary itself. Following the change, abortion incidence fell sharply at 22 weeks, with an incidence rate ratio of 0.25 and a 95% confidence interval from 0.21 to 0.30. In practical terms, the rate at that gestational age was approximately one-quarter of the level expected from the earlier pattern. At 23 weeks, the reduction was even greater: the incidence rate ratio was 0.07, with a 95% confidence interval from 0.06 to 0.10. The statistical evidence for both changes was strong, with P values below 0.001.
Yet the decline at the new limit was accompanied by increases just beforehand. Abortion incidence rose at 20 weeks, where the incidence rate ratio reached 1.28, with a 95% confidence interval of 1.15 to 1.43. At 21 weeks, the ratio was 1.33, with a 95% confidence interval of 1.18 to 1.50. These figures indicate that abortions became more concentrated in the weeks immediately before the new legal boundary. The pattern is consistent with a behavioral response to the policy: some procedures that might previously have taken place at 22 or 23 weeks appear to have been brought forward to 20 or 21 weeks.
The timing of the effect is important. Researchers found no meaningful changes at 19 weeks or earlier, and no meaningful changes at 24 weeks or later. That concentration around 20 to 23 weeks makes a broad transformation in reproductive behavior less likely than a direct response to the new deadline. If the entire healthcare system or the population’s underlying preferences had changed substantially, the effects might have appeared across a wider range of gestational ages. Instead, the observed pattern clustered near the point at which the law changed the options available to women and clinicians.
The researchers also examined cumulative incidence between 12 and 27 weeks. This measure captures the total proportion of pregnancies ending in abortion by each gestational age, rather than counting procedures only within individual weekly intervals. Cumulative incidence decreased for women who had reached between 19 and 23 weeks of pregnancy. That finding suggests the legislation was associated not only with a redistribution of abortions across nearby weeks, but also with fewer abortions among people who remained pregnant into the period closest to the revised limit.
This distinction separates two possible effects of a legal deadline. The first is substitution: a procedure that would have occurred later is performed earlier, leaving the overall number of abortions largely unchanged. The second is prevention or deterrence: some abortions do not occur after the deadline becomes more difficult to meet. Japan’s data point to both processes. The increase at 20 and 21 weeks indicates substantial movement toward earlier procedures, while the lower cumulative incidence from 19 to 23 weeks indicates that some women who reached those stages ultimately did not undergo an abortion recorded in the data.
The results do not establish why individual decisions changed. Some women may have accelerated a decision after learning about the new law, while others may have faced difficulty arranging care within the shorter timeframe. Changes in referral pathways, clinical scheduling, travel requirements or delays in recognizing a pregnancy could also have shaped the outcome. The study identifies a population-level association rather than documenting the personal experiences behind each record. It also cannot determine whether abortions not observed at later gestational ages were abandoned, performed earlier than 12 weeks, carried out outside the formal system or classified differently.
Even with those limitations, the study offers rare empirical evidence about how reproductive policy can alter the timing of medical decisions. The large national dataset and the sharp contrast around the 1991 policy change strengthen the argument that the legislation played a central role in the shift. The findings also underline why legal limits should not be evaluated solely by counting procedures after the cutoff. A law may reduce abortions at one gestational age while increasing them immediately beforehand, creating a wave of accelerated decision-making that is invisible if researchers examine only the prohibited weeks.
Japan’s experience provides a detailed case study of how a seemingly narrow regulatory change can reshape behavior across a sensitive medical timeline. The evidence indicates that the 1991 reduction from 24 to 22 weeks moved abortions earlier, sharply reduced procedures at 22 and 23 weeks, and was linked to lower cumulative incidence among women who had reached the nineteenth week. As countries continue to debate gestational limits, the study suggests that the consequences of such laws may be measured not only in legal eligibility, but also in delays, access, scheduling and the rapidly closing window in which patients make decisions.
Subject of Research: The association between changes in abortion gestational limits and abortion timing and incidence among women in Japan.
Article Title: A quasi-experimental study of abortion rates following gestational age limit changes in Japan
Article References: Yanagawa, Y., Piedvache, A., Okubo, Y. et al. “A quasi-experimental study of abortion rates following gestational age limit changes in Japan.” Nature Human Behaviour (2026). https://doi.org/10.1038/s41562-026-02531-3
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41562-026-02531-3
Keywords: abortion policy, gestational limits, Japan, reproductive health, women’s decision-making, quasi-experimental study, interrupted time series, regression discontinuity, abortion incidence, pregnancy outcomes
