Time to second abortion or continued pregnancy following a first abortion: a retrospective cohort study.

STUDY QUESTION: What proportions of women have a second abortion or continued pregnancy within 12-46 months of a first abortion?

SUMMARY ANSWER: Estimated return rates for a second abortion were 5, 10.9 and 19.8% at 12, 24 and 46-months, respectively, and rates of continued pregnancy were 5.6, 12.9 and 24.3% at the same intervals.

WHAT IS KNOWN ALREADY: Studies attempting to identify women at risk for 'repeat abortion' for intervention purposes have described a range of demographic and behavioural characteristics associated with presentation for more than one abortion, but few have taken timing of abortions into account.

STUDY DESIGN, SIZE, DURATION: Retrospective cohort study involving women presenting for a first abortion at a public hospital abortion clinic in New Zealand (2007-2010).

PARTICIPANTS/MATERIALS, SETTING, METHODS: Electronically stored records were analysed for women discharged from a public hospital abortion clinic in New Zealand. Outcome measures were the proportion of women having a second abortion or continued pregnancy within 24 months of a first abortion, and characteristics associated with shorter time to subsequent pregnancy. Cox proportional hazards modelling was used to detect factors associated with time to a second abortion or continued pregnancy, and Kaplan-Meier survival analyses were used to estimate time to one of these two pregnancy outcomes.

MAIN RESULTS AND THE ROLE OF CHANCE: A total of 6767 women had a first abortion between 2007 and 2010. Some data were missing for 11 women so were excluded from the cohort and analyses. Return rates for a second abortion estimated from survival analyses were 5, 10.9 and 19.8% at 12, 24 and 46 months, respectively. Estimated rates of continued pregnancies were 5.6, 12.9 and 24.3% at 12, 24 and 46 months, respectively. Younger age, non-European ethnicity and greater parity were significantly associated with shorter time to a second abortion and to a subsequent continued pregnancy (P < 0.01 for all factor P-values). Hazard ratios (HR) for a second abortion were highest among those aged 16-19 years (HR 1.6, 95% confidence interval (CI) 1.3-1.9, Reference 20-24), of Pacific Island (HR 1.35, 95% CI 1.1-1.7) or Maori ethnicity (HR 1.26, 95% CI 1.1-1.5, Reference New Zealand European), and with 1 (HR 1.41, 95% CI 1.1-1.7) or 2 (HR 1.41, 95% CI 1.1-1.9, Reference nulliparous) children at the time of the first abortion. Both pregnancy outcomes were observed among 120 women (1.8%), with 60% of these women having a second abortion before the continued pregnancy.

LIMITATIONS, REASONS FOR CAUTION: This study was limited to analysis of routinely collected clinical and demographic data for women presenting for abortion over a 4-year period. Conclusions could not be drawn about a wider range of personal and situational factors influencing pregnancy and pregnancy outcomes. Data were drawn from only one clinic but characteristics of the study sample were broadly representative of those reported nationally. Loss to follow-up for women seeking a second abortion elsewhere in the country cannot be ruled out and would serve to underestimate return rates reported here.

WIDER IMPLICATIONS OF THE FINDINGS: To date, the most effective public health measure known to reduce abortion return rates within 24 months is the initiation of long-acting reversible contraception (LARC) at the time of an abortion. The high proportion of women seeking a second abortion <4 years after a first abortion (20%) could be significantly reduced by use of LARC, as could unintended pregnancies that are continued soon after a first abortion, particularly among teenaged and young women. Barrier-free access to a range of LARC methods should be prioritized to prevent unintended and mistimed pregnancies.

STUDY FUNDING/COMPETING INTERESTS: Funded by a Lottery Health Research Grant and a University of Otago Research Grant. The authors have no competing interests.

TRIAL REGISTRATION NUMBER: Not applicable.