Partner change and perinatal outcomes: a systematic review

Authors


Dr Jun Zhang, Epidemiology Branch, National Institute of Child Health and Human Development, NIH, Building 6100, Room 7B03, Bethesda, MD 20892, USA. E-mail: zhangj@mail.nih.gov

Summary

Epidemiological studies suggest that partner change may affect perinatal outcomes in subsequent pregnancies. We conducted a systematic review on the association between paternity change and perinatal outcomes. We searched the literature in MEDLINE using keywords ‘paternity’, ‘partner’, ‘pre-eclampsia’, ‘preterm birth’, ‘low birth weight’, and ‘birth defects’ from 1966 to 2005. We identified 19 studies that examined the association between partner change and specific perinatal outcomes: 12 on pre-eclampsia or hypertension in pregnancy, three on birth defects, three on preterm birth, and two on low birthweight.

Partner change was consistently associated with an increased risk of pre-eclampsia or hypertension in pregnancy in 11 of 12 studies (the unadjusted relative risk [RR] ranging from 1.2 to 8.3). However, after controlling for birth interval as a confounder in multivariate analysis, two studies using the same birth registry data showed a modestly reduced risk in relation to partner change (RR = 0.84 and 0.73, respectively), while two studies found a slightly increased risk (both RR = 1.3). Retrospective cohort studies presented inconsistent findings on the association between partner change and risk of preterm birth and low birthweight. Finally, three population-based cohort studies demonstrated that partner change significantly reduced the recurrence of the same or similar birth defects in subsequent births (RRs ranging from 0.1 to 0.76). We conclude that partner change reduces the risk of recurrent same birth defects. However, epidemiological evidence on the effect of partner change on pre-eclampsia, preterm birth and low birthweight is inconclusive. Whether birth interval should be controlled for in the association between partner change and pre-eclampsia warrants caution.

Introduction

Changing partners during a woman's reproductive age is not uncommon in industrialised societies. Partner change may have three potential effects on reproductive and perinatal outcomes. These include: (i) re-establishment of maternal immune adaptation and tolerance to new paternal antigens;1 (ii) genetic sharing between the new partners;2 and (iii) change in environment and social behaviour.3 However, the effect of partner change on perinatal outcomes in subsequent pregnancies has not been well studied. Although newer epidemiological studies have emerged in recent years, the literature concerning these effects remains scarce.

We conducted a systematic epidemiological review on the effects of partner change on adverse perinatal outcomes, including pre-eclampsia, birth defects, low birthweight and preterm birth. Although this review is intended to summarise the existing literature on partner change and perinatal outcomes, women who change partner differ from those who do not in many aspects. Thus, associations observed in retrospective observational studies may not necessarily be causal.

Methods

A MEDLINE search was performed using keywords ‘paternity’, ‘partner’, ‘pregnancy’, ‘hypertension’, ‘pre-eclampsia’, ‘preterm birth’, ‘fetal growth’, ‘birth defects’, ‘congenital malformation’, ‘low birth weight’ and ‘spontaneous abortion’. We identified five perinatal outcomes, namely pre-eclampsia or pregnancy-induced hypertension, preterm birth, birth defects, low birthweight and infant mortality, that have been associated with partner change in the literature. All articles were retrieved and the references cited in these articles were carefully examined. Relevant information was then reviewed and extracted from the papers by both study authors.

We found three studies that examined the associations of partner change with birth defects, three studies on partner change and preterm birth, two on partner change and low birthweight, and one on infant mortality. Thus, no attempt was made to synthesise the findings. We identified 12 studies that addressed the association between partner change and pre-eclampsia or pregnancy-induced hypertension; however, the study design, study population, definition of the outcome, and adjustment for potential confounders varied tremendously from study to study. A synthesised result would make no better sense than simply presenting individual findings. Therefore, this is a systematic review and not a formal meta-analysis.

Results

Partner change and pre-eclampsia

The first study on partner change and pre-eclampsia was a hospital-based matched case–control study published in 1980.4 Women with a multifetal pregnancy or a record of hypertension or proteinuria in previous pregnancies were excluded. A normotensive control was selected for each case by taking the nearest singleton delivery in the hospital records for the year that exactly matched for parity, race, sociodemographic group and for age to within 3 years. The authors found that 13 of 47 (28%) multiparous women with severe pre-eclampsia had changed partner, while 3 of 47 (6%) normotensive controls did so (OR = 6.3, [95% CI 1.5, 36.6]). In addition, the recorded evidence suggested the possibility of partner change in a further 5 (11%) cases, compared with 2 (4%) of the controls. Two other hospital-based case–control studies using either telephone interview or in-person interview, confirmed this finding (OR = 8.1 for pregnancy-induced hypertension, [95% CI 2.7, 24.2];5 OR = 8.6 for de novo pre-eclampsia, [95% CI 3.1, 23.5])6 respectively. Birth interval was not considered in any of these studies.

Trupin et al.7 examined the association between partner change and pre-eclampsia in the Child Health and Development Studies cohort of 5800 multiparous women. After having adjusted for potential confounders (excluding birth interval), the adjusted relative risks (RR) for pre-eclampsia and gestational hypertension were 1.4 [95% CI 0.8, 2.4] and 1.2 [95% CI 1.0, 1.5] respectively, compared with women who did not change partner. Saftlas et al.8 conducted a cohort analysis in a large clinical trial of inner city, young primiparous women. They found that women with a history of abortion who conceived again with the same partner had nearly half the risk of pre-eclampsia (adjusted OR = 0.54, [95% CI 0.31, 0.97]). In contrast, women with an abortion history who conceived with a new partner had the same risk of pre-eclampsia as women without a history of abortion (adjusted OR = 1.03, [95% CI 0.72, 1.47]).

Six retrospective cohort studies examined the association between partner change and pre-eclampsia while being mindful of birth interval, of which three studies were based on the same birth registry data. Lie et al.9 reported that partner change did not significantly alter the incidence of pre-eclampsia in a second pregnancy (1.6% vs. 1.3% with and without partner change respectively). Adjusting for maternal age and birth interval did not change the finding. Two later studies10,11 found that without adjusting for birth interval, partner change was associated with a modestly higher risk of pre-eclampsia among women who did not have pre-eclampsia in the first pregnancy (Table 1). However, after adjusting for birth interval and maternal age, both studies showed a 20–25% reduction in pre-eclampsia risk among women who changed partner. These findings were mostly consistent with an earlier report using a different birth registry data.12 However, due to a smaller sample size, none of the results in this last study were statistically significant.12

Table 1.  Literature review on partner change and risk of pre-eclampsia
Author group
(year)
Study designStudy populationCrude risk ratio (rr) for pre-eclampsia
[95% CI]
Adjusted risk ratio (rr) for pre-eclampsia
[95% CI]
Adjusted factorsConclusionStrength (+) and weakness (–)
  1. SGA, small-for-gestational-age.

Feeney and Scott4 (1980)Matched case–controlCases: 47 multiparous women who had severe pre-eclampsia after previous normotensive pregnancies. Controls: nearest singleton delivery in medical records, matching on parity, race, socio-economic group and age within 3 years. Partner information was based on medical recordsDefinite or strong evidence of partner change:
OR = 6.3 [1.5, 36.6] Possible evidence of partner change:
OR = 3.5 [0.6, 9.8]
  Pre-eclampsia in mulitparous women was associated with partner change+ clearly defined, carefully selected severe cases and matched normotensive controls;
− ‘evidence’ of changed paternity was not defined
Robillard et al.5 (1993)Case–controlCases: multiparous women who had hypertensive disorders in pregnancy and fetal complications and whose babies were admitted to the Neonatal Department (40 with chronic hypertension; 34 with pregnancy-induced hypertension). Controls: women who did not have hypertension during pregnancy and whose babies were also admitted to the Neonatal Department (n = 60). Partner information was based on in-person interviewProportion of women with different fathers:
Control: 10/60 = 17%
Pregnancy-induced hypertension: 21/34 = 62%;
OR = 8.1 [2.7, 24.2]
Chronic hypertension: 4/40 = 10%;
OR = 0.6 [0.1, 2.1]
  Changing partner was significantly associated with pregnancy-induced hypertension but not with pre-existing chronic hypertension− the study population was selected based on infant condition and no assessment of possible selection bias was made; information on paternity status was collected several weeks postpartum; the main outcome was pregnancy-induced hypertension overall; no control for potential confounders
Trupin et al.7 (1996)Prospective cohort5800 multiparous women from the Child Health and Development Studies cohort from Oakland, CA, 1959–67; 573 had changed partnersrr for pre-eclampsia with a new partner: 1.7 [1.0, 2.9]
rr for gestational hypertension with a new partner: 1.3 [1.1, 1.6]
Adjusted rr for pre-eclampsia: 1.4 [0.8, 2.4] Adjusted rr for gestational hypertension: 1.2 [1.0, 1.5]Race, body mass index, time of first prenatal visit, parity, history of pre-eclampsiaChange of partner is associated with pre-eclampsia and gestational hypertension+ a large, prospective cohort study
Lie et al.9 (1998)Retrospective cohortAll births in the Medical Birth Registry of Norway since 1967–92 with more than 16 weeks of gestation;
14 266 pairs of 1st and 2nd pregnancies with a different father and 363 758 pairs of 1st and 2nd pregnancies with the same father
Incidence of pre-eclampsia in 2nd pregnancy for: Women without pre-eclampsia in 1st pregnancy: same partner: 1.3%, new partner: 1.6%;
Women with pre-eclampsia in 1st pregnancy : same partner: 13.1%, new partner: 11.8%
  Partner change was not associated with pre-eclampsia+ population-based; more reliable information on partner status;
− classification of pre-eclampsia cases was not validated; smoking not controlled
Tubbergen et al.6 (1999)Case–controlCases: 333 multiparous women with a diastolic blood pressure ≥100 mmHg from hospital obstetric database. Controls: 182 women with normotensive pregnancies and normal grown fetus, selected from an outpatient midwife clinic with complete antenatal careOR for de novo pre-eclampsia: 8.6 [3.1, 23.5]  A change of partner was associated with pre-eclampsia in subsequent pregnancies+ clearly defined cases and controls;
− retrospective information on paternity; small sample size; no control for potential confounders
Li and Wi13 (2000)Retrospective cohortWomen with 2 consecutive births during 1989–91 from the California birth certificate data (n = 140 147); 13 372 changed partners Without pre-eclampsia in 1st birth and change of partner: 1.3 [1.1, 1.6] With pre-eclampsia in 1st birth and change of partner: 0.7 [0.4, 1.2]Race, age, education, smoking, time of first prenatal visit, parity and birth intervalThe effect of changing partner depends on the history of pre-eclampsia with the previous partner. Parental human leucocyte antigen may play a role in the aetiology of pre-eclampsia+ population-based; very short birth interval (max. 3 years);
− paternity status was based on father's name on birth certificate; classification of pre-eclampsia was not validated
Trogstad et al.10 (2001)Retrospective cohort547 238 women with a 1st and 2nd pregnancy registered in the Medical Birth Registry of Norway, 1967–98. Only women with singleton pregnancies both times were includedWith no previous pre-eclampsia: 1.21 [1.09, 1.35]; With previous pre-eclampsia: 1.02 [0.85, 1.24]With no previous pre-eclampsia: 0.80 [0.72, 0.90] With previous pre-eclampsia: 0.97 [0.79, 1.18]Birth interval, maternal age at 2nd delivery and year at 2nd deliveryThe protective impact of a new father for the 2nd pregnancy challenges the hypothesis of primipartner+ population-based;
− classification of pre-eclampsia cases was not validated; smoking was not controlled
Basso et al.12 (2001)Retrospective cohortWomen recorded with pre-eclampsia/eclampsia between 1980 and 1994 in Danish hospital records were divided into 2 cohorts:PE+cohort 0.92 [0.74, 1.13] PE–cohort 1.23 [1.08, 1.40]PE+cohort
Adjusting for age: 0.91 [0.71, 1.12] Social variables: 0.89 [0.72, 1.11] Birth interval: 0.82 [0.66, 1.03] PE–cohort
Maternal age, social variables, smoking and birth intervalsAfter adjusting for birth interval, partner change was associated with a reduced risk of pre-eclampsia+ population-based; smoking and previous abortion as potential confounders were examined;
− classification of pre-eclampsia cases was not validated
(1) women with pre-eclampsia in 1st birth (n = 8401) (PE+cohort), Adjusting for age: 1.23 [1.08, 1.41] Social variables: 1.13 [0.98, 1.29]   
(2) woman without pre-eclampsia in 1st birth (n = 26 596)
(PE–cohort)
 Birth interval: 0.88 [0.76, 1.02]   
Saftlas et al.8 (2002)Prospective cohort4589 healthy primiparous women 11–21 weeks pregnant were enrolled at 5 centres in the US for the Calcium for Pre-eclampsia Prevention Trial, from 1992 to 1995Incidence of pre-eclampsia: primigravidae: 8.0%; history of abortion with a different partner: 7.9%; history of abortion with the same partner: 4.3%Women with an abortion history who conceived with a new partner had the same risk of pre-eclampsia as women without a history of abortion: 1.03 [0.72, 1.47] Women with the same partner: 0.54 [0.31, 0.87]Body mass index, race, health insurance status, smoking and clinical centreProlonged exposure to fetal antigens from a previous pregnancy protects against pre-eclampsia in a subsequent pregnancy with the same father+ a large, prospective study; well-defined pre-eclampsia cases; smoking was controlled;
− no information on inter-pregnancy interval
Skjærvan et al.11 (2002)Retrospective cohort551 478 sets of 1st and 2nd singleton deliveries in Medical Birth Registry of Norway from 1967 to 1998. 509 548 had the same father and 31 683 had different fathers.From 1st to 2nd pregnancies with a different partner: 1.14 [1.04, 1.26]; From 2nd to 3rd pregnancies with a different partner: 1.42 [1.25, 1.62]From 1st to 2nd pregnancies with a different partner: 0.75 [0.68, 0.83] From 2nd to 3rd pregnancies with a different partner: 0.93 [0.81, 1.07]Maternal age, year of delivery, and birth intervalAfter adjustment for the interval between births, a change of partner is associated with a reduced risk of pre-eclampsia+ population-based;
− classification of pre-eclampsia cases was not validated; smoking was not controlled
  209 423 sets of 1st, 2nd and 3rd singleton deliveries; 158 284 had the same father, 24 252 sets the partner had changed either between the 1st and 2nd or between the 2nd and 3rd deliveries     
Mostello et al.14 (2002)Population-based case–controlLinked birth certificate data from Missouri, 1989–97. Cases: 2332 women with hypertension in 2nd pregnancy. Controls: 2370 women without hypertension in 2nd pregnancy1.43 [1.25, 1.67]1.25 [1.0, 1.43]Previous pre-eclampsia, chronic hyper-tension, diabetes, renal disease, gestational age at 1st delivery, previous SGA, birth interval, inadequate prenatal careHaving the same father for the 2nd birth was protective+ population-based;
− outcome was hypertension in pregnancy overall, not just pre-eclampsia; classification of cases was not validated; paternity status was based on father's name on birth certificate
Hjartardottir et al.15 (2004)Retrospective cohortWomen from deliveries at Landspitalinn University Hospital in Reykjavik, Iceland, from 1984 to 1999.Only women who had 2 pregnancies resulting in birth were included. The study included 614 women with confirmed de novo hypertension in 1st pregnancy and a family history of pregnancy-induced hypertension. 121 women changed partnersGestational hypertension occurred in 66% of women with the same partner and 66% of women with different partners. Pre-eclampsia occurred in 58% of women with the same partner and 53% who had changed partnersOR for developing recurrent hypertensive disorder in pregnancy was 1.15 [0.96, 1.37] for every year in woman with a new partner.
OR for recurrence was 1.15 [1.05, 1.27] for every year in woman with the same partner
Maternal ageIn women with a positive family history and previous hypertension in pregnancy, change of partner does not influence the risk of recurrence− subjects probably had high susceptibility to chronic hypertension, a less optimal study population for pre-eclampsia research

Li and Wi13 examined this issue using the linked California birth certificate data from 1989 to 1991. They found that, among women who did not have hypertension in pregnancy in the first delivery, partner change was associated with a 30% increase in risk of pre-eclampsia after adjusting for several factors, including age and birth interval. It should be noted that birth interval in this study (maximum of 3 years) was very short compared with all other studies.

In a large population-based, case–control study using the Missouri maternally linked cohort from 1989 to 1997, Mostello et al.14 controlled for birth interval as well as other factors, such as previous hypertension in pregnancy, chronic hypertension, diabetes, renal disease and previous small-for-gestational-age (SGA) newborn. The increased risk for overall hypertension in pregnancy persisted with a change in partner (crude RR = 1.43, [95% CI 1.25, 1.67]; adjusted RR = 1.25, [95% CI 1.0, 1.43]).

In an Iceland population, Hjartardottir et al.15 found that, in women with a family history of hypertensive disorders in pregnancy who also had de novo hypertension in the first pregnancy, partner change did not alter the risk of recurrent hypertension in the second pregnancy (RR = 1.02, [95% CI 0.94, 1.11]). One could probably argue that hypertension in these women might reflect an underlying mechanism that differs from pre-eclampsia.

Partner change and birth defects

Studies have shown that certain birth defects may recur in subsequent pregnancies. Lie et al.16 used data from the Medical Birth Registry of Norway to examine whether change in residence (a surrogate for environment) and partner (a surrogate for certain genetic factors) affects recurrence of similar birth defects (Table 2). Among women who had had a child with birth defects and stayed in the same municipality, the RRs of recurrence of similar birth defects in the following pregnancy were 11.6 [95% CI 9.3, 14.0] for women who had the same partner and 7.3 [95% CI 3.5, 11.1] for women who changed to a new partner respectively. This was a 37% reduction in RR attributable to partner change.

Table 2.  Literature review on partner change and risk of recurrent birth defects in the following pregnancy
Author group
(year)
Study designStudy populationCrude risk ratio (RR)
[95% CI]
Adjusted risk ratio (RR)
[95% CI]
Adjusted factorsConclusion
Lie et al.16 (1994)Retrospective cohortThe 1st and 2nd singleton infants delivered by 371 933 women in the Medical Registry of Norway from 1967 to 1989. The reference population was women whose first infant had no defectRR of recurrent similar birth defects in 2nd birth: Same municipality: 0.59 [0.35, 1.00]. Different municipality: 0.76 [0.38, 1.51]  The relative risk of recurrent birth defect is somewhat lower in women who changed partner
Basso et al.3 (1999)Retrospective cohort8761 and 9635 women who had a singleton birth with and without a birth defect, respectively, in Danish Medical Birth Registry, 1980–92RR of recurrent similar birth defects in 2nd birth: 0.26 [0.11, 0.65]  Partner change is associated with a lower recurrence risk of similar defects
Christensen et al.17 (1995)Retrospective cohort4189 women who had infants with facial cleft defects in the Central Person Registry in Denmark between 1952 and 1987. 1902 mothers had additional children after the 1st with a facial cleft defect. 236 infants had mothers who changed partners RR of recurrent facial-cleft defects 0.1 [0.02, 0.7]MunicipalityHaving a different partner reduced a woman's recurrence risk of facial-defect

A more striking reduction was later reported by Christensen et al.17 Using data from the Danish Medical Registry, the authors examined whether changes in municipality and partner affected recurrence of facial-cleft defects. They found that among 236 infants of mothers who changed partners, 1 (0.4%) had a facial-cleft defect, in comparison with 77 (3.3%) of 2350 infants of mothers who did not change partners (RR = 0.1, [95% CI 0.02, 0.9]).

From the same data source, Basso et al.3 identified 8761 women who had a singleton birth with a birth defect between 1980 and 1992 and another delivery after the index pregnancy during the same period. They also randomly selected 9635 women who had a normal singleton birth that was followed by another birth during the same study period. Partner change was associated with a reduced recurrence of similar defects (OR = 0.26, [95% CI 0.11, 0.65]).

Partner change, preterm birth and low birthweight

In two retrospective cohort studies using the Danish Medical Birth Registry, Basso et al.18,19 showed that partner change was not associated with preterm birth or low birthweight regardless of whether or not women had had a preterm birth or low-birthweight infant in a previous pregnancy (Table 3). For women who had a term birth in the previous pregnancy, the adjusted RR for preterm birth in the subsequent pregnancy was 1.18 [95% CI 0.80, 1.72], while for women who had a preterm birth, the RR was 1.02 [95% CI 0.86, 1.21]. For women who had a normal-birthweight infant in the previous pregnancy, the adjusted RR for low birthweight after partner change was 1.26 [95% CI 0.76, 2.08], while for women who had a preterm birth, the RR was 1.06 [95% CI 0.86, 1.31]. The authors adjusted for birth interval but not maternal smoking.

Table 3.  Literature review on partner change and risk of preterm birth and low birthweight in the following pregnancy
Author group (year)Study designStudy populationAdjusted risk ratio (RR) [95% CI]Adjusted factorsConclusion
  1. HLA, human leucocyte antigen.

Basso et al.18 (1997)Retrospective cohortFrom Danish Medical Birth Registry, 1069 women who had a low- birthweight infant and a subsequent liveborn infant between 1980 and 1992 (exposed) and a random sample of 10 211 who gave birth to a normal-weight infant and to a subsequent liveborn infant between 1980 and 1992 (unexposed)Partner change and low birthweight: Among exposed: 1.06 [0.86, 1.31] Among unexposed: 1.26 [0.76, 2.08]Social status of the couple at birth of index child, age of mother at birth of current child, parity, birth interval, and gestational age of the current childChange of male partner did not modify the risk of low birthweight in subsequent pregnancy
Basso et al.19 (1999)Retrospective cohort10 455 women in Danish Medical Birth Registry who gave birth to a preterm child and had a subsequent live birth between 1980 and 1992 (exposed). 9849 women gave birth to a child after 37 weeks of gestation and had a subsequent liveborn child between 1980 and 1992 (unexposed).Partner change and preterm birth: Among exposed: 1.02 [0.86, 1.21] Among unexposed: 1.18 [0.80, 1.72]Social status of the couple when the index child was born, age of mother at birth of current child, parity of mother at index child, and birth intervalThere was no sign of paternal genetic effect on preterm delivery
Li20 (1999)Retrospective cohort128 329 California residents who had 2 consecutive births from 1989 to 1991. Three cohorts were selected based on the gestational age of the 1st delivery: <34 weeks, 34–36 weeks, >36 weeksPartner change and preterm birth by gestational age in 1st birth <34 weeks: 2nd birth <34 weeks: 0.67 [0.52, 0.88]; 2nd birth at 34–36 weeks: 0.99 [0.78, 1.25]; 2nd birth ≥37 weeks: 1.16 [1.04, 1.30]
Gestational age in 1st birth at 34–36 weeks: 2nd birth <34 weeks: 0.80 [0.62, 1.02]; 2nd birth at 34–36 weeks: 0.91 [0.77, 1.08]; 2nd birth ≥37 weeks: 1.09 [1.01, 1.17]
Maternal race, age, educational level, prenatal smoking, gestational month of initial prenatal visit, parity, and interval between previous birth and conception of the index pregnancyThe effect of partner change depends on the pregnancy outcome with the previous partner and supports the hypothesis that parental HLA sharing may be related to preterm delivery
Vatten and Skjaerven21 (2003)Retrospective cohortNorway medical birth registry. 31 683 women who changed partner between 1st and 2nd births and 456 458 women with the same partner for both births in 1967–98Partner change and preterm birth by mother's education: ≤10 years 1.8 [1.6, 2.0]; 11–13 years 2.0 [1.9, 2.1]; ≥14 years 2.6 [2.2, 3.0] Partner change and low birthweight by mother's education: ≤10 years 2.0 [1.7, 2.3]; 11–13 years 2.5 [2.3, 2.6]; ≥14 years 3.8 [3.3, 4.5]Maternal age, birth interval (2-year categories), and year of birthWomen who change partner are at a higher risk of having preterm birth and low-birthweight infant

Li20 identified 128 239 women who had two consecutive births during 1989–91 through data linkage of the California birth certificates. Women were categorised by the gestational age of the first delivery: <34 weeks, 34–36 weeks, and ≥37 weeks. Partner change appeared to reduce the risk of very preterm birth among women who had had a preterm birth in the previous pregnancy. The author hypothesised that parental human leucocyte antigen (HLA) sharing may be related to preterm delivery. Changing partner in women with a previous preterm birth may result in the likelihood that the new partner has a different genetic profile. This may reduce parental HLA sharing and, in turn, reduce the risk of preterm birth in subsequent pregnancies.

In contrast, Vatten and Skjærven21 identified a large number of women who had first and second births in the Norway medical birth registry from 1967 to 1998. After adjustments for maternal age, education, birth interval and decade of birth, the risk of preterm birth, low birthweight and infant mortality in the second birth was significantly higher for women who changed partner between the first two births compared with those who had the same partner for both births. The authors suggested that women who change partner may be a selected group with characteristics that increase risk or they may change their lifestyle or behaviour in ways that are unfavourable to pregnancy outcomes. Indeed, women who changed partner were three times more likely to smoke than women who did not change partner.21 However, smoking was not controlled for in that study.

Discussion

Pre-eclampsia

Pre-eclampsia occurs mostly in the first pregnancy. The prevailing hypothesis is that after the first pregnancy, the maternal immune system has ‘recognised’ the paternal antigens and built a greater immune tolerance towards the same antigens in subsequent pregnancies.1 Changing partner introduces new paternal antigens. The maternal immune system, therefore, has to re-establish an immune tolerance. Indeed, most, but not all, previous studies have shown that partner change increased the risk of pre-eclampsia or hypertension in pregnancy in general.

However, women who change partners often have a longer birth interval, and a longer interval is associated with a higher incidence of pre-eclampsia.11,12 Some authors have suggested that the association between partner change and pre-eclampsia may be due to confounding by birth interval.11,12 When birth interval was controlled as a confounder, partner change was associated with a reduced risk of pre-eclampsia.10,11 In a separate paper, it has been argued that birth interval is a collider.22 If birth interval is controlled or stratified, factors that may both affect fertility, which, in turn, prolong birth interval and increase the risk of pre-eclampsia, should also be controlled. These factors include, but are not limited to, autoimmune diseases, insulin resistance, the metabolic syndrome and endocrine disorders.23 Otherwise, collider-stratification may distort the relationship between partner change and the risk of pre-eclampsia.24 This may explain why the RR changed from above 1.0 to below 1.0 after having controlled for birth interval in previous studies.11,12 Mostello et al.,14 for instance, showed that when both birth interval and other factors, such as previous pre-eclampsia, chronic hypertension, diabetes, renal disease and previous SGA newborn, were controlled simultaneously, partner change was still associated with an increased risk of hypertension in pregnancy.

It is also noteworthy that the type of hypertensive disorders in pregnancy (de novo pre-eclampsia only or hypertension in pregnancy in general) and accuracy of classification of pre-eclampsia varied substantially from study to study. In general, studies that had more focused cases and reliable diagnosis4,8 tended to yield higher RRs than studies based on population registry data.9–12 Case–control studies4–6 also showed much higher RRs than retrospective cohort studies,9–14 probably because the former were more selective for both cases and normotensive controls. Furthermore, because smokers are two to three times more likely to change partner12,21 and smoking is associated with a reduced risk of pre-eclampsia, controlling for smoking appears essential. However, only three out of 12 studies controlled for smoking.8,12,13 Finally, in addition to smoking habit, women who change partner may also differ in some other aspects from those who do not. Part of the differences between the findings of the various papers may depend on what type of confounder was taken into account.

Birth defects

The relationship between genetic factors and birth defects is fairly well described. A fetus who inherits recessive genes from both parents may develop certain defects.2 For a more common mutation, for example, a fetus who carries two copies of T variant of 6,12-methylenetetrahydrofolate reductase (C677T, TT genotype) gene has an increased risk of neural tube defects.25 Therefore, it is quite plausible that for women who give birth to a fetus with congenital anomalies, partner change may avoid a defective fetal genotype in subsequent births and, therefore, reduce the risk of recurrence of the same birth defects. However, such a reduction may not be observed for dissimilar defects.

Preterm birth and low birthweight

Four large population-based retrospective cohort studies using birth registry data found inconsistent results regarding the association between partner change and preterm birth and low birthweight.18–21 Several factors may be attributable to the inconsistency. Women who changed partners often have different characteristics from women who have not changed partner.12,21 Some of the unfavourable characteristics, such as smoking, have a direct impact on perinatal outcomes, including low birthweight. Unfortunately, birth registry data often lack accurate and in-depth information on potential confounders. Inadequate control of important confounders may result in bias. Various degrees of residual confounding may have in part caused the inconsistency.

Another major challenge that confronted these studies was the classification of preterm birth. All studies relied on the last menstrual period to calculate gestational age, which was subject to error. More importantly, preterm birth has a diverse aetiology, most of which remains to be elucidated.26 Medical or obstetric complications that warrant preterm delivery and intrauterine infection that results in preterm labour are commonly recognised causes. Genetic factors may also contribute to preterm birth. For example, maternal inflammatory hyper-responsiveness may be more important than any fetal/paternal contribution to preterm birth.27 Unfortunately, birth registry data do not allow for accurate classification. Further studies with better classification of preterm birth are needed to clarify this association.

In summary, our systematic review indicates that in women who had a birth with a congenital anomaly, partner change may result in a reduced risk of recurrence of the same birth defects in subsequent pregnancies. For a multiparous woman who changed partner, her risk of pre-eclampsia may be increased. However, whether birth interval should be adjusted for in the association between partner change and pre-eclampsia warrants further discussion. Literature on the association between partner change and preterm birth and low birthweight is limited and inconsistent, which may be attributable to the suboptimal assessment of the outcome and inadequate control of confounding.

Ancillary