The effect of method and gestational age at termination of pregnancy on future obstetric and perinatal outcomes: a register-based cohort study in Aberdeen, Scotland

Authors

  • A Woolner,

    Corresponding author
    1. Obstetrics & Gynaecology, Division of Applied Health Sciences, School of Medicine and Dentistry, University of Aberdeen, Aberdeen, UK
    2. Department of Obstetrics & Gynaecology, Aberdeen Maternity Hospital and Aberdeen Royal Infirmary, NHS Grampian, Aberdeen, UK
    • Correspondence: Dr A Woolner, Obstetrics & Gynaecology, University of Aberdeen, Aberdeen Maternity Hospital, Cornhill Road, Foresterhill, Aberdeen, AB25 2ZD, UK. Email a.woolner@abdn.ac.uk

    Search for more papers by this author
  • S Bhattacharya,

    1. Obstetrics & Gynaecology, Division of Applied Health Sciences, School of Medicine and Dentistry, University of Aberdeen, Aberdeen, UK
    2. Department of Obstetrics & Gynaecology, Aberdeen Maternity Hospital and Aberdeen Royal Infirmary, NHS Grampian, Aberdeen, UK
    Search for more papers by this author
  • Soh Bhattacharya

    1. Obstetric Epidemiology, Dugald Baird Centre for Research on Women's Health, Aberdeen Maternity Hospital, Aberdeen, UK
    Search for more papers by this author

Abstract

Objective

To determine whether termination of pregnancy (TOP), including the method used or gestational age at termination, affects future obstetric and perinatal outcomes.

Design

Register-based cohort.

Setting

Aberdeen Maternity Hospital, Scotland, UK.

Population

From the Aberdeen Maternity and Neonatal Databank (AMND) we identified 3186 women who had terminated their first pregnancy and then had a second pregnancy of beyond 24 weeks of gestation between 1986 and 2010. We identified 42 446 women who had their first delivery in the same time period, for comparison.

Methods

Univariate and multivariate logistic regression was used to compare outcomes between groups. Complete case analysis with adjustment of confounding factors was carried out, and adjusted odds ratios (aORs) with 99% confidence intervals are presented.

Main outcome measures

The primary outcome was spontaneous preterm delivery (SPTD).

Results

No statistically significant association was found between TOP in the first pregnancy and SPTD in the next pregnancy (aOR 1.05; 99% CI 0.83–1.32). Neither medical (aOR 1.03; 99% CI 0.72–1.46) nor surgical (aOR 1.06; 99% CI 0.78–1.44) termination appeared to affect the risk of spontaneous preterm delivery in the subsequent pregnancy. Late termination (≥13 weeks of gestation) did not appear to increase the risk of spontaneous preterm delivery compared with early termination (<13 weeks of gestation) (aOR 1.65; 99% CI 0.94–2.92), nor compared with primigravid women (aOR 1.25; 99% CI 0.97–1.62). There was an associated increased risk of antepartum haemorrhage in the next pregnancy following TOP (< 0.01; aOR 1.26; 99% CI 1.10–1.45).

Conclusions

Evidence on obstetric and perinatal outcomes following TOP remains conflicting. This study suggests that TOP is not associated with an increased risk of spontaneous preterm delivery. Neither the method nor the gestational age of TOP has any effect on this lack of association.

Introduction

Legal termination of pregnancy (TOP) is a common procedure with low mortality rates when performed by skilled practitioners.[1] It has, however, been suggested that TOP is associated with potential maternal and perinatal risks in subsequent pregnancies.[1-8] Consequences of TOP are of interest to women, researchers, and clinicians alike, as most women who undergo TOP in their initial pregnancy are likely to become pregnant again.[7, 9-11]

Early observational studies failed to identify significant obstetric or perinatal risks following TOP;[12, 13] however, more recent research suggests an association with future infertility, ectopic pregnancy, miscarriage, placental disorders, low birthweight, and spontaneous preterm birth.[2-4, 7, 8, 11, 14, 15] Termination of pregnancy and, specifically repeat TOPs, have been associated with an increased risk of spontaneous preterm delivery in subsequent pregnancies, particularly preterm delivery before 32 weeks of gestation (termed ‘very’ preterm birth).[2-6, 8, 16] A recent systematic review and meta-analyses found an increased risk of preterm labour and low birthweight following TOP.[17] Preterm birth (birth at <37 weeks of gestation)[18] remains a significant cause of perinatal morbidity and mortality, with high health service costs as well as psychosocial implications for families.[18, 19]

In 2011 47% of terminations in England and Wales,[1] and 72.5% in Scotland,[9] were medically induced.[10] The majority of TOPs in the UK were performed at early gestations: 91% at <13 weeks of gestation in England and Wales,[10] and 65% at <9 weeks of gestation in Scotland.[9] It is plausible that trauma caused by cervical dilatation and uterine evacuation during surgical TOP could damage the cervix and it is possible that late gestational age terminations could lead to a greater risk of spontaneous preterm delivery in future pregnancies. Few studies have specifically investigated the effect of different methods of termination on future obstetric and perinatal outcomes, or the effect of the gestational age at which TOP is carried out.[2, 4, 11, 20, 21]

The Aberdeen Maternity Neonatal Databank (AMND) routinely collects data (including method and gestational age at TOP) for all pregnancies in a defined locality,[22] and this provides an opportunity to investigate not only the effect of TOP, but specifically the effects of method and gestational age at TOP on future pregnancies. This study aimed to investigate whether, in comparison with primigravid women, obstetric and perinatal outcomes were adversely affected in a second pregnancy in women who, in a first pregnancy, underwent: (1) any TOP; (2) surgical TOP or medical TOP; or (3) early TOP or late TOP.

Methods

Study design and conduct

A retrospective cohort study was conducted using an anonymised data set extracted from the AMND.[22] The AMND is a population-based database that holds obstetric- and fertility-related data from 1951 to present day for all deliveries and reproductive outcomes from the only maternity hospital for the geographical area: the Aberdeen Maternity Hospital.[22] Eligible records were identified and an anonymised data set was extracted from the database by the AMND data management team for women whose pregnancies occurred between 1986 and 2010. Exposure was defined as TOP in the first pregnancy. All women who underwent TOP in their first pregnancy, either medical or surgical termination, for any reason, who thereafter had a subsequent singleton pregnancy beyond 24 weeks of gestation within the period 1986–2010 were included in the exposed cohort. It was not possible to determine whether the pregnancy at TOP was singleton or multiple, as that information is not routinely collected. The unexposed comparison cohort included women whose first pregnancy resulted in a singleton pregnancy and delivery beyond 24 weeks of gestation (all primigravid women) between 1986 and 2010. Exclusions included: multiple pregnancies and women who had a miscarriage or ectopic pregnancy in the pregnancy following TOP, or in their first pregnancy in the unexposed cohort. All data were anonymised, no identifiable data was passed onto the researchers, and there was no prospective involvement of patients or medical records. We used the recommendations of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for undertaking observational research.[23]

Outcome measures

The TOP-related data included gestational age at which TOP was undertaken and method of TOP. Demographic details analysed were: age of women at delivery; Carstairs’ category of social deprivation,[24] or husband's social class (recorded using the Registrar General's occupation-based social classification); and self-reported smoking status as recorded at the antenatal booking visit in the subsequent pregnancy following TOP, or in the first pregnancy for primigravid women. Primary outcome was spontaneous preterm delivery (SPTD), defined as spontaneous onset of labour and subsequent delivery at <37 weeks of gestation. Women who underwent elective caesarean section or induction of labour prior to 37 weeks of gestation were excluded from that particular analysis. Secondary outcomes analysed included: all preterm births (birth at <37 weeks of gestation, including spontaneous, elective, or induced births); very preterm births (births at <33 weeks of gestation, including spontaneous, elective, or induced births); any hypertensive disease of pregnancy, including pre-eclampsia;[25] antepartum haemorrhage (APH); postpartum haemorrhage (PPH, defined as >500-ml blood loss at vaginal delivery, or >1000-ml blood loss at caesarean section); low birthweight (defined as birthweight of <2500 g); and neonatal unit admission.

Comparisons

All women who underwent TOP in their first pregnancy were compared with all primigravid women for each of the outcome measures to determine the effect, if any, of TOP on the primary outcome of SPTD, and thereafter the secondary outcomes. Women who had undergone previous TOP were then stratified according to method of TOP: medical or surgical. Women who had initially undergone a medical TOP but thereafter required a surgical TOP were included in the surgical group. Outcomes from the second pregnancy for women with exposure to either medical or surgical TOP in their first pregnancy were then compared with outcomes for primigravid women in their first pregnancy (the unexposed cohort): previous medical TOP versus primigravid women; previous surgical TOP versus primigravid women.

Women with a history of medical TOP were then compared with women who had undergone previous surgical TOP for each outcome measure: previous medical TOP versus previous surgical TOP.

Similarly, the exposed cohort (women with a history of TOP in their first pregnancy) was stratified according to the gestational age at TOP: early TOP (<13 weeks of gestation) or late TOP (≥13 weeks of gestation). Each of these groups were compared with all primigravid women, and thereafter with each other for the primary outcome of SPTD and all of the secondary outcomes: previous early TOP versus primigravid women; previous late TOP versus primigravid women; previous late TOP versus previous early TOP.

Power calculation

Preliminary interrogation of the AMND database revealed that there were 3187 women who underwent TOP in the first pregnancy, and who subsequently had a further continuing pregnancy between 1986 and 2010. Similarly, 42 446 primigravid women were identified whose first pregnancy progressed beyond 24 weeks of gestation during the same time period. Using a 5% level of significance and two sided test, a power calculation determined that the study would have 98% power to detect a difference of 2% in the risk of preterm delivery with the assumption that the unexposed cohort has 6% risk[26] of the complication.

Statistical analysis

IBM spss Statistics for Windows (version 20.0.); IBM Corp., Armonk, NY, USA was used for statistical analyses. Descriptive statistics and univariate analysis were used to compare the demographics of the groups. The data set provided independent and unpaired samples. Categorical outcomes were analysed using the chi-square test. Non-parametric variables were analysed using the Mann–Whitney U-test, and continuous, normally distributed variables were analysed using the independent Student's t-test. Two-tailed P-values were reported and a P-value of 0.01 was used to define statistical significance. Univariate binary logistic regression was employed to determine the probability of the occurrence of the aforementioned obstetric and perinatal outcomes within the samples, and to calculate crude odds ratios (cORs). Potential confounding factors were adjusted for in a multivariate binary logistic regression model. Both unadjusted and adjusted odds ratios (aORs) are presented. To help reduce the chance of erroneous results because of multiple testing (type-1 error) we used P ≤ 0.01 throughout, and therefore present ORs with 99% confidence intervals.

As some data were missing from covariates, but few data were missing for outcomes, and also because the data were not missing completely at random (MNCAR), we did not think multiple imputation was appropriate. Instead, we used other variables to provide the missing information, so that the proportion of missing data in each variable was reduced to <5%. For example, we used different measures of social class (Registrar General's occupation-based social class,[27] Carstairs’ deprivation category,[25] and Scottish Indices of Multiple Deprivation)[28] to derive a composite variable with few missing data for social class. Where such an approach could not be incorporated, we created a value for ‘missing-ness’ (for example, 999 for missing smoking status) and ran a complete case analysis using this value. We also did a sensitivity analysis excluding all cases with missing data, and the results were similar. Where data were missing related to exposure data such as gestational age at TOP, only cases that had data on the exposure stratification were included.

Results

Data for 45 632 women were included in the study. There were 3187 women who underwent a TOP in their first pregnancy and thereafter had a subsequent pregnancy beyond 24 weeks of gestation in the data set. One record was excluded for improbable values. Therefore, 3186 women were included as the exposed cohort and 42 446 women were included in the unexposed group (primigravid women whose first pregnancy continued beyond 24 weeks of gestation). Demographic characteristics at time of delivery (for pregnancy number 2 for women with a history of TOP, and for pregnancy number 1 for primigravid women in the comparison cohort) were each described and then compared as shown in Table 1. The majority of women (n = 2315, 72.7%) underwent TOP prior to 13 weeks of gestation, with only 431 (13.5%) terminations at gestations of 13 or more weeks, although data were missing (not at random) for gestational age at TOP for 440 women (13.8%). A total of 1385 (43.5%) women underwent medical TOP, and 1800 (56.5%) women underwent surgical TOP, with method of TOP missing from the records for one woman.

Table 1. Baseline characteristics during first pregnancy for primigravid women and during second pregnancy for women with a previous history of TOP
CharacteristicPrimigravid women = 42446Women with TOP in first pregnancy n = 3186 P
  1. a

    Represents a statistically significant result.

  2. b

    Using composite variable of Registrar General's occupation based social class, Scottish Indices of Multiple Deprivation and Carstairs' deprivation category.[24]

Mean age in years at delivery (SD)26.09 (5.39)25.79 (5.44)0.002a
Missing35 (0.08)0
Smoking status, n (%)
Never smoked28 875 (68.0)1696 (53.2)<0.001a
Ex-smoker3528 (8.3)457 (14.3)
Smoker9117 (21.5)993 (31.2)
Missing926 (2.2)40 (1.3)
Socio-economic status, n (%)b
Not deprived (1, 2)17 178 (40.5)1278 (40.1)<0.001a
Deprived (3–6)23 200 (54.6)1851 (58.1)
Missing2068 (4.9)57 (1.8)

Pregnancy outcomes of primigravid women and those who underwent a TOP in their first pregnancy are summarised in Table 2. Women who had undergone TOP in their first pregnancy appeared to be at no higher risk of SPTD when compared with primigravid women (aOR1.05; 99% CI 0.83–1.32), and both groups were similar for occurrence of any preterm (not specifically spontaneous preterm) delivery, very preterm delivery, and low birthweight. Women who had TOP in their first pregnancy appeared to have protection against hypertensive disease in their subsequent pregnancy, compared with women in their first pregnancy (aOR 0.69 (99%CI 0.61–0.78). Women with a history of TOP appeared to be more likely to suffer from antepartum haemorrhage (aOR 1.26; 99% CI 1.10–1.45).

Table 2. Odd ratios following binary logistic regression comparing obstetric and perinatal outcomes between primigravid women in their first pregnancy and second-pregnancy outcomes for women who underwent a previous TOP
OutcomeWomen with previous TOP in first pregnancy n (%)Primigravid women n (%)Previous TOP vs primigravid women
(= 3186, unless otherwise specified)(= 42446, unless otherwise specified)Unadjusted PUnadjusted ORs (99% CI)Adjusted PAdjusted ORs (99% CI)
  1. a

    Analysis restricted to women with a spontaneous onset of labour (n = 28 012 primigravid women; n = 2093 women with previous TOP).

  2. b

    Analysis restricted to women with a vaginal birth (n = 33 520 primigravid women; n = 2497 women with previous TOP).

  3. c

    Analysis restricted to women with a caesarean section delivery (n = 8916 primigravid women; n = 689 women with previous TOP).

Preterm birth240 (7.5)3056 (7.2)0.481.05 (0.88–1.26)0.501.05 (0.88–1.26)
Missing      
Spontaneous preterm delivery (SPTD)a(n = 2093) 141 (6.7)(n = 28012) 1806 (6.4)0.601.05 (0.83–1.32)0.621.05 (0.83–1.32)
Missing00    
Very preterm birth45 (1.4)733 (1.7)0.190.82 (0.55–1.21)0.170.81 (0.54–1.21)
Missing00    
Any hypertensive disease of pregnancy543 (17.0)9804 (23.1)<0.010.68 (0.60–0.78)<0.010.69 (0.61–0.78)
Missing16 (0.5)395 (0.9)    
Pre-eclampsia177 (5.6)2466 (5.8)0.550.95 (0.78–1.17)0.530.95 (0.77–1.17)
Missing16 (0.5)395 (0.9)    
Antepartum haemorrhage433 (13.6)4666 (11.0)<0.011.27 (1.11–1.46)<0.011.26 (1.10–1.45)
Missing00    
Postpartum haemorrhage at vaginal deliveryb(n = 2497) 340 (13.6)(n = 33520) 4150 (12.4)0.071.12 (0.96–1.31)0.041.14 (0.97–1.33)
Missing17 (0.7)182 (0.5)    
Postpartum haemorrhage at caesarean sectionc(n = 689) 91 (13.2)(n = 8916) 1179 (13.2)0.960.99 (0.74–1.34)0.961.01 (0.74–1.36)
Missing4 (0.6)87 (1.0)    
Low birthweight202 (6.3)3004 (7.1)0.121.13 (0.93–1.37)0.071.14 (0.94–1.39)
Missing09 (0.002)    
Neonatal unit admission475 (14.9)5823 (13.7)<0.010.85 (0.74–0.97)<0.010.83 (0.73–0.96)
Missing299 (9.4)11 564 (27.2)    

Second pregnancy outcomes for women with a previous termination (stratified according to method of TOP: medical or surgical) were compared with primigravid women, as summarised in Table 3. Women with a history of medical TOP were also compared with women with a history of surgical TOP in their first pregnancy for each of the outcomes (Table 3). Neither method of TOP appeared to significantly affect the risk of SPTD, any preterm, or very preterm delivery in the subsequent pregnancy, compared with primigravid women. Similarly, comparison of medical versus surgical TOP groups revealed no significant difference in associated risk of SPTD, any preterm, or very preterm birth in the subsequent pregnancy. The risk of any hypertensive disease of pregnancy was higher in the primigravid group compared with women in both the previous medical and the previous surgical TOP groups. Antepartum haemorrhage appeared to be increased following either a previous medical or surgical TOP, when compared with primigravid women; however, only surgical TOP remained statistically significant after adjustment. Women with a history of medical TOP appeared to be at a higher risk of postpartum haemorrhage at vaginal delivery in their second pregnancy, compared with primigravid women (aOR 1.49; 99% CI 1.21–1.85) and women who had undergone surgical TOP in their first pregnancy (aOR 1.69; 99% CI 1.25–2.29). There were no differences found on analyses of postpartum haemorrhage following caesarean section.

Table 3. Odds ratios following binary logistic regression comparing obstetric and perinatal outcomes between primigravid women in their first pregnancy and women in their second pregnancy who previously underwent: (a) medical termination of pregnancy (MTOP) and (b) surgical termination of pregnancy (STOP). Outcomes also compared between women with previous MTOP and STOP
OutcomeMTOP, n (%)STOP, n (%)Primigravid women, n (%)MTOP vs primigravidSTOP vs primigravidMTOP vs STOP
 (n = 1385 unless otherwise specified)(n = 1800 unless otherwise specified)(n = 42446 unless otherwise specified)Unadjusted ORs (99% CI) Adjusted ORs (99% CI)dUnadjusted ORs (99% CI) Adjusted ORs (99% CI)dUnadjusted ORs (99% CI) Adjusted ORs (99%CI)d
  1. a

    Analysis restricted to women with a spontaneous onset of labour (n = 28 012 primigravid women; n = 906 women with previous MTOP; n = 1187 women with previous STOP.

  2. b

    Analysis restricted to women with a vaginal birth (n = 33 520 primigravid women; n = 1064 women with previous MTOP; n = 1421 women with previous STOP).

  3. c

    Analysis restricted to women with a caesarean section delivery (n = 8916 primigravid women; n = 321 women with previous MTOP; n = 367 women with previous STOP).

  4. d

    All odds ratios are adjusted for maternal age at delivery, smoking, and social class.

Preterm birth98 (7.1)142 (7.9)3056 (7.2)0.98 (0.75–1.29)1.10 (0.88–1.39)0.89 (0.63–1.26)
Missing0000.98 (0.75–1.29)1.10 (0.88–1.39)0.89 (0.63–1.27)
Spontaneous preterm delivery (SPTD)a(n = 906) 60 (6.6)(n = 1187) 81 (6.8)(n = 28012) 1806 (6.4)1.03 (0.73–1.46)1.06 (0.79–1.44)0.97 (0.62–1.53)
Missing0001.03 (0.72–1.46)1.06 (0.78–1.44)0.97 (0.62–1.53)
Very preterm birth22 (1.60)23 (1.3)733 (1.7)0.92 (0.52–1.61)0.74 (0.43–1.28)1.25 (0.58–2.70)
Missing0000.91 (0.52–1.60)0.73 (0.42–1.26)1.25 (0.58–2.71)
Any hypertensive disease of pregnancy220 (15.9)322 (17.9)9804 (23.1)0.63 (0.52–0.76)0.73 (0.62–0.85)0.87 (0.68–1.11)
Missing3 (0.2)13 (0.7)395 (0.9)0.63 (0.52–0.76)0.73 (0.62–0.85)0.87 (0.68–1.11)
Pre-eclampsia78 (5.6)98 (5.4)2466 (5.8)0.97 (0.71–1.31)0.93 (0.71–1.23)1.04 (0.69–1.55)
Missing3 (0.2)13 (0.7)395 (0.9)0.97 (0.71–1.31)0.93 (0.71–1.22)1.04 (0.70–1.56)
Antepartum haemorrhage 177 (12.8) 256 (14.2) 4666 (11.0) 1.19 (0.96–1.46))1.34 (1.12–1.61) 10.88 (0.67–1.16)
Missing0001.18 (0.95–1.46.33 (1.11–1.59)0.88 (0.67–1.16)
Postpartum haemorrhage at vaginal delivery.b(n = 1064) 184 (17.2)(n = 1421) 156 (11.0)(n = 33 520) 4150 (12.4)1.48 (1.20–1.83)0.87 (0.69–1.08)1.71 (1.26–2.31)
Missing5 (0.5)12 (0.8)182 (0.5)1.49 (1.21–1.85)0.88 (0.71–1.11)1.69 (1.25–2.29)
Postpartum haemorrhage at caesarean sectionc(n = 321) 44 (13.7)(n = 367) 47 (12.8)(n = 8916) 1179 (13.2)1.04 (0.68–1.59)0.96 (0.64 1.45)1.08 (0.61–1.93)
Missing2 (0.6)2 (0.7)87 (1.0)1.06 (0.69–1.62)0.96 (0.64–1.45)1.10 (0.61–1.97)
Low birthweight81 (5.8)121 (6.7)3004 (7.1)1.23 (0.91–1.65)1.06 (0.83–1.35)1.16 (0.79–1.70)
Missing009 (0.002)1.23 (0.92–1.68)1.08 (0.84–1.38)1.15 (0.78–1.69)
Neonatal unit admission220 (15.9)255 (14.2)5823 (13.7)0.86 (0.71–1.04)0.84 (0.70–1.01)1.02 (0.78–1.32) 1
Missing58 (4.2)241 (13.4)11 564 (27.2)0.84 (0.69–1.02)0.83 (0.69–0.99).02 (0.79–1.33)

Obstetric and perinatal outcomes in women who had early (<13 weeks of gestation) or late (≥13 weeks of gestation) TOP and primigravid women are summarised in Table 4. There were no statistically significant differences in terms of SPTD, all preterm birth, and very preterm birth for both the early and the late TOP groups, compared with primigravid women. Notably, however, the proportion of women with SPTD and preterm birth (9%) was greater following late TOP, whereas the corresponding values in primigravid women and those with previous early TOP were around 6–7%. Antepartum haemorrhage was also more likely to occur in women following early TOP (but not late TOP), compared with primigravid women (aOR 1.29; 99% CI 1.10–1.51). The risk of hypertensive disease of pregnancy was lower in women with previous early (aOR 0.64; 99% CI 0.55–0.75) and late (aOR 0.83; 99% CI 0.71–0.98) TOP, in comparison with primigravidae.

Table 4. Odds ratios following binary logistic regression comparing obstetric and perinatal outcomes between primigravid women in their first pregnancy and women in their second pregnancy who previously underwent: (a) early TOP at <13 weeks of gestation and (b) late TOP at ≥13 weeks of gestation. Outcomes also compared between women with previous early TOP and late TOP
OutcomeEarly TOP, n (%)Late TOP, n (%)Primigravid women, n (%)Early TOP vs primigravidLate TOP vs primigravidLate TOP vs early TOP
(= 2315 unless otherwise specified)(n = 431 unless otherwise specified)(n = 42446 unless otherwise specified)Unadjusted ORs (99% CI) Adjusted ORs (99% CI)dUnadjusted ORs (99% CI) Adjusted ORs (99% CI)dUnadjusted ORs (99% CI) Adjusted ORs (99% CI)d
  1. a

    Analysis restricted to women with a spontaneous onset of labour (n = 28 012 primigravid women; n = 1504 women with early TOP; n = 304 women with late TOP.

  2. b

    Analysis restricted to women with a vaginal birth (n = 33 520 primigravid women; n = 1818 women with early TOP; n = 344 women with previous late TOP).

  3. c

    Analysis restricted to women with a caesarean section delivery (n = 8916 primigravid women; n = 497 women with early TOP; n = 87 women with late TOP).

  4. d

    All odds ratios are adjusted for maternal age at delivery, smoking, and social class.

Preterm birth171 (7.4)39 (9.0)3056 (7.2)1.03 (0.83–1.27)1.13 (0.91–1.41)1.25 (0.77–2.01)
Missing0001.03 (0.83–1.27)1.13 (0.91–1.40)1.24 (0.77–2.00)
Spontaneous preterm delivery (SPTD)a(n = 1504) 94 (6.3)(n = 304) 30 (9.9)(n = 28 012) 1806 (6.4)0.97 (0.73–1.28)1.26 (0.98–1.62)1.64 (0.93–2.89)
Missing0000.97 (0.73–1.28)1.25 (0.97–1.60)1.65 (0.94–2.92)
Very preterm birth34 (1.5)8 (1.9)733 (1.7)0.85 (0.54–1.34)1.04 (0.65–1.65)1.27 (0.46–3.52)
Missing0000.84 (0.53–1.33)1.02 (0.64–1.62)1.25 (0.45–3.50)
Any hypertensive disease of pregnancy374 (16.2)74 (17.2)9804 (23.1)0.64 (0.55–0.74)0.83 (0.70–0.98)1.08 (0.75–1.54)
Missing15 (0.6)1 (0.2)395 (0.9)0.64 (0.55–0.75)0.83 (0.71–0.98)1.07 (0.75–1.54)
Pre-eclampsia123 (5.3)19 (4.4)2466 (5.8)0.91 (0.71–1.16)0.87 (0.64–1.17)0.82 (0.43–1.57)
Missing15 (0.6)1 (0.2)395 (0.9)0.91 (0.71–1.16)0.82 (0.43–1.57)0.81 (0.42–1.56)
Antepartum haemorrhage320 (13.8)50 (11.6)4666 (11.0)1.30 (1.11–1.52)1.03 (0.85–1.25)0.82 (0.54–1.24)
Missing0001.29 (1.10–1.51)1.02 (0.84–1.25)0.80 (0.53–1.21)
Postpartum haemorrhage at vaginal deliveryb(n = 1818) 241 (13.3)(n = 344) 44 (12.8)(n = 33 520) 4150 (12.4)1.08 (0.90–1.30)1.02 (0.83–1.26)0.96 (0.61–1.51)
Missing10 (0.6)2 (0.6)182 (0.5)1.09 (0.90–1.31)1.05 (0.85–1.29)1.01 (0.64–1.60)
Postpartum haemorrhage at caesarean sectionc(n = 497) 62 (12.5)(n = 87) 10 (11.5)(n = 8916) 1179 (13.2)0.93 (0.65–1.39)0.92 (0.59–1.42)0.91 (0.36–2.30)
Missing3 (0.6)087 (1.0)0.94 (0.66–1.35)0.94 (0.60–1.45)0.92 (0.36–2.34)
Low birthweight149 (6.4)31 (7.2)3004 (7.1)1.11 (0.89–1.39)0.99 (0.78–1.26)0.89 (0.52–1.50)
Missing009 (0.002)1.13 (0.90–1.41)1.01 (0.79–1.29)0.92 (0.54–1.56)
Neonatal unit admission335 (14.5)71 (16.5)5823 (13.7)0.83 (0.71–0.98)0.99 (0.84–1.17)1.18 (0.81–1.71)
Missing679 (24.7)48 (11.1)11 564 (27.2)0.82 (0.70–0.96)0.98 (0.82–1.16)1.18 (0.81–1.71)

Discussion

Main findings

We found no association between TOP and risk of SPTD in a subsequent pregnancy. Neither method of termination nor gestational age at TOP appeared to affect the risk of SPTD in a future pregnancy.

Our findings do suggest that women with a history of either medical or surgical TOP in their first pregnancy are at greater risk of APH in a subsequent pregnancy, compared with primigravid women. TOP in the first pregnancy appeared protective against hypertensive disease of pregnancy in the next pregnancy when compared with primigravid women.

Strengths and weaknesses

Unlike previous studies that were unable to distinguish between spontaneous and iatrogenic preterm birth, a major strength of this study was our ability to identify our primary outcome. We were able to categorise method and gestational age at TOP as these data are routinely collected by AMND, and were available for 99.9 and 86.2% of the cohort, respectively. This is substantially higher than comparable figures in a study using national data.[7] The majority of women included underwent TOP prior to 13 weeks of gestation (84.3%); therefore, the results of the subgroup analyses for gestational age need to be interpreted with caution, as the numbers of women who had late TOP at ≥13 weeks of gestation are small enough to generate concerns about a potential type-II error. The available sample size allowed us to achieve 98% power to show a potential 2% difference in the rate of SPTD between primigravid women and all women who underwent a TOP in their first pregnancy. Given the possibility of type-1 error from the multiple analyses performed, we defined statistical significance as < 0.01.

As a result of the consistent record keeping within AMND, overall there were few missing data for outcomes and specifically for data related to potential confounding factors, i.e. over 97% of the smoking data and over 95% of the social class data were available. Confounding factors such as body mass index, intergestational interval, or year of delivery could have been included to improve our results.

Although there is no ideal comparison group in studies investigating outcomes after TOP,[1-8, 29] we felt that the decision to use women in their first continuing pregnancy beyond 24 weeks of gestation (gravida 1, para 0) was clinically relevant, and a potential strength of this study. The key question women contemplating a TOP in their first pregnancy need to address is whether or not to continue their pregnancy or have a TOP. Therefore, any differences in outcomes between these two groups are clinically relevant, and are likely to better inform the decision-making of both the women and their clinicians. Particularly as previous studies suggest an increased risk of low birthweight and SPTD following TOP in first pregnancies, when compared with women whose first pregnancy resulted in a term delivery.[2, 7, 11] Another potential comparison group is women with a history of miscarriage in their first pregnancy; however, as women with a history of miscarriage have been shown to have an increased risk of SPTD,[7] we were unsure as to whether this comparison would be helpful.

Results from a specific population could affect the generalisability of the results; however, the use of standardised protocols within the UK National Health Service (NHS) ensure results will be generalisable to other UK areas and countries where legal terminations are performed.

Clinical practice in regard to surgical and medical terminations has remained similar over the study period in Aberdeen. One potential source of bias is that most surgical terminations in Aberdeen have involved the use of medical priming. As women may have moved away from the area for their subsequent pregnancy, this is a potential source of bias; however, a recent tracing survey conducted by the AMND revealed that only 3.8% of women had migrated outside the area following a maternity record in AMND,[22] and therefore any effect is likely to be small.

Interpretation

Research on outcomes following TOP is characterised by a number of methodological challenges. Ascertainment bias can occur in countries where termination is illegal, randomised trials are not feasible, small sample sizes are common, and inadequate data collection makes it difficult to define variables such as SPTD, or exposures such as gestational age and method of termination. As a consequence, conclusions on risks following TOP remain unclear.[2, 7, 12, 20]

Our results do not concur with previous reports supporting an association between TOP and subsequent preterm birth.[2-6, 8, 16] They do agree with results reported by Virk and colleagues[30] who found that the risks of preterm birth, low birthweight, ectopic pregnancy, and spontaneous miscarriage in the Danish population were similar for both medical and surgical terminations prior to 9 weeks of gestation. A Finnish study by Raatikainen and colleagues[31] on mainly women who had undergone surgical TOP also failed to identify increased risks, including preterm delivery, in subsequent pregnancies following TOP. Bracken and colleagues[12] reported no adverse effects on birthweight in subsequent pregnancies related to gestational age at TOP. A systematic review of seven observational studies also found no associated increased risk of preterm delivery for medical versus surgical TOP.[21] In contrast, Atrash and colleagues[11] suggest that surgical TOP leads to an increased risk of spontaneous miscarriage and low birthweight. Given the paucity of adverse pregnancy outcomes found in earlier observational studies,[12, 13] the aforementioned studies,[21, 30, 31] and our results, questions still remain unanswered regarding the risk of SPTD in a subsequent pregnancy following TOP.[2-8, 17]

Our study demonstrates an association between TOP and antepartum haemorrhage in a subsequent pregnancy. This supports earlier work demonstrating an increased risk of antepartum haemorrhage and placenta praevia following TOP,[2, 15, 16, 32] and specifically sharp curettage surgical termination.[33] A study found that the rate of placental abruption was almost double in women with previous TOP compared with primigravid women, although the results were not statistically significant.[34] Conversely, another study suggested previous TOP was not associated with placental abruption.[35]

We found that primigravid women were at significantly higher risk of any hypertensive disease of pregnancy compared with women with a previous TOP, which agrees with previous reports.[35, 36, 38] It is widely accepted that the risk of hypertensive disease, and specifically pre-eclampsia, is greatest in the first pregnancy, and it is likely that a previous pregnancy, even a terminated one, provides some protection against subsequent hypertensive problems.[37, 38]

Our results do not provide clear guidance on whether medical or surgical termination should be preferred. A study by Henshaw and colleagues[39] found that women found both medical and vacuum aspiration methods acceptable. Given the generally similar outcomes found in our study and a similar study between methods of termination,[30] it seems reasonable that both options continue to be offered.

Some of Hill's criteria for causality are fulfilled,[40] such as plausibility of SPTD following TOP and a dose–response relationship for the reported effects of repeat TOP.[17] However given the conflicting body of information, lack of consistency of any effect, and lack of association found in our study, no clear causal relationship is currently demonstrated.

Conclusion

We found no increased risk of SPTD in a subsequent pregnancy following an initial TOP, in comparison with women in their first continuing pregnancy. Although not definitive, this study continues to inform the debate on future adverse outcomes following TOP, and adds to the literature surrounding this important reproductive issue. A larger prospective cohort study is needed to investigate the effect of second-trimester TOP on SPTD in a more definitive manner.

Disclosure of interests

The authors declare that they have no competing interests.

Contribution to authorship

SB and SohB conceived the idea for the study, and SB was the principal investigator. The study was designed by SB, SohB, and AW. Both SB and AW applied for permissions and data access. SB, SohB, and AW managed the project. AW and SohB cleaned the data and performed the initial analyses. AW performed statistical analyses and interpreted the results, with help from SB and SohB AW wrote the first and subsequent drafts of the article. All authors commented on and contributed to the revision of drafts and the final draft of the article.

Details of ethics approval

Ethical approval (CERB/2012/5/772) was granted by the College Ethics Review Board, College of Life Sciences and Medicine, at the University of Aberdeen on 14 June 2012. NHS Research & Development approval was sought and granted by NHS Grampian on 26 June 2012. Approval was granted from the Aberdeen Maternity Neonatal Databank (AMND) on 23 November 2011 for access to an extracted data set.

Funding

This study had no external funding.

Acknowledgements

The authors would like to thank the data management team at Aberdeen University for data extraction, Dr Jill Mollison for statistical advice, and Mrs Margery Heath for secretarial support.

Ancillary