Changes in association between previous therapeutic abortion and preterm birth in Scotland, 1980 to 2008: a historical cohort study

Abstract

Background: Numerous studies have demonstrated that therapeutic termination of pregnancy (abortion) is associated with an increased risk of subsequent preterm birth. However, the literature is inconsistent, and methods of abortion have changed dramatically over the last 30 years. We hypothesized that the association between previous abortion and the risk of preterm first birth changed in Scotland between 1 January 1980 and 31 December 2008.

Methods and findings: We studied linked Scottish national databases of births and perinatal deaths. We analysed the risk of preterm birth in relation to the number of previous abortions in 732,719 first births (≥24 wk), adjusting for maternal characteristics. The risk (adjusted odds ratio [95% CI]) of preterm birth was modelled using logistic regression, and associations were expressed for a one-unit increase in the number of previous abortions. Previous abortion was associated with an increased risk of preterm birth (1.12 [1.09-1.16]). When analysed by year of delivery, the association was strongest in 1980-1983 (1.32 [1.21-1.43]), progressively declined between 1984 and 1999, and was no longer apparent in 2000-2003 (0.98 [0.91-1.05]) or 2004-2008 (1.02 [0.95-1.09]). A statistical test for interaction between previous abortion and year was highly statistically significant (p<0.001). Analysis of data for abortions among nulliparous women in Scotland 1992-2008 demonstrated that the proportion that were surgical without use of cervical pre-treatment decreased from 31% to 0.4%, and that the proportion of medical abortions increased from 18% to 68%.

Conclusions: Previous abortion was a risk factor for spontaneous preterm birth in Scotland in the 1980s and 1990s, but the association progressively weakened and disappeared altogether by 2000. These changes were paralleled by increasing use of medical abortion and cervical pre-treatment prior to surgical abortion. Although it is plausible that the two trends were related, we could not test this directly as the data on the method of prior abortions were not linked to individuals in the cohort. However, we speculate that modernising abortion methods may be an effective long-term strategy to reduce global rates of preterm birth.

Figure 1. Selection of the study cohorts.

Figure 2. Cumulative incidence of preterm birth from 24 wk onwards in relation to number of previous abortions for 732,719 nulliparous women, Scotland 1980–2008.

The relative risk of preterm birth for women with zero, one, two, or three or more previous abortions significantly varied across the range 24 to 36 wk gestational age (global test of proportional hazards assumption: p = 0.02). The graph is confined to the risk prior to 34 wk to allow better visualisation of the differences in incidence of extreme preterm births.

Figure 3. Forest plots of odds ratios for preterm birth in Scotland by epoch.

(A) Unadjusted odds ratio for a one-unit increase in number of previous abortions (coded as 0, 1, 2, and 3 or more) in relation to risk of preterm first birth among 732,719 women for births from 1980 to 2008. (B) As in (A), but odds ratio adjusted for maternal characteristics (deprivation category, previous miscarriage, maternal age, height, and marital status). (C) Adjusted odds ratio for a one-unit increase in number of previous abortions in relation to risk of preterm first birth among 414,373 women for births from 1992 to 2008. Odds ratios adjusted for maternal characteristics as in (B), but also for smoking. (D) Adjusted odds ratio for a 10-cm decrease in maternal height in relation to the risk of preterm first birth among 732,719 women for births from 1980 to 2008. Odds ratios adjusted for deprivation category, maternal age, marital status, previous abortion, and previous miscarriage. The interaction p-value is for a Wald test of the null hypothesis that the odds ratios did not significantly differ across the period 1980 to 2008. Year is treated as a continuous variable in all the statistical tests of interaction.

Figure 4. Crude rates of spontaneous preterm birth for nulliparous women with (n = 63,428) and without (n = 669,291) a past history of abortion in Scotland, 1980–2008.

Figure 5. Annual numbers of abortions by method among nulliparous women in Scotland, 1992–2008.

These data were aggregated and were not linked to SMR02 data. (A) Observed data. (B) Sensitivity analysis where 5% of medical procedures are re-classified as surgical procedures with cervical pre-treatment. (C) Sensitivity analysis where 10% of medical procedures are re-classified as surgical procedures with cervical pre-treatment.