Recurrent preterm delivery following twin versus singleton preterm delivery: A retrospective cohort

The main risk factor for preterm delivery (PTD; <37 gestational weeks) is having a history of PTD. The aim of this research was to compare the risk for recurrent PTD following twin versus singleton gestation PTD.

singleton gestation following twin PTD. 12 Some studies found no increased risk for recurrent PTD in singleton gestation, in cases where the twin deliveries occurred between 34 and 37 gestational weeks 13,14 or among spontaneous PTD in twins. 15However, when comparing women with preterm and term twin deliveries, the risk for PTD in the subsequent singleton pregnancy is greater among women with preterm twins. 16,17In a meta-analysis based on 3376 women from eight cohort studies, having a history of preterm twin delivery was associated with up to 9.73 (95% confidence interval [CI]   4.99-18.98)higher odds of subsequent preterm singleton delivery if the twin delivery was at <30 gestational weeks compared with term twin delivery. 18 having a history of PTD is a main risk factor for recurrence of PTD, it is expected that PTD as compared with term in twin delivery is associated with an increased risk for PTD recurrence in subsequent pregnancies.In women with a history of PTD in singleton pregnancies, the follow-up plan in subsequent pregnancy includes prevention strategies, such as administration of progesterone. 19wever, controversy exists regarding the management of women who had a PTD in twin gestation.Would the same recommendations be relevant?Should prevention be recommended for women with a history of twin gestation PTD?Should these be recommended for PTDs that occurred at any gestational ages, or should prevention strategies be recommended only if the PTD in twins was at earlier gestational ages?
The aim of the current study was to evaluate whether a previous PTD in twin versus PTD in singleton gestation is associated with a similar risk of recurrent PTD, and to find possible gestational age and IPI cut-off points for predicting the increased risk of PTD recurrence following twin gestation PTD.The association was also evaluated separately by mode of conception.

| ME THODS
A population-based retrospective cohort study was conducted, including women who had two consecutive pregnancies, the first of which, defined as the index pregnancy, ended with PTD.Eligible women were included once in the study (if a woman had recurrent PTDs with a subsequent singleton pregnancy following each PTD, only the first PTD per women was included in the study).
The index pregnancy was either twin or singleton gestation.
All pregnancies occurred between January 1, 1991 and January 1, 2021 at the Soroka University Medical Center (SUMC), the single tertiary medical center located in southern Israel.Only women who had both consecutive deliveries at SUMC were included and women with triplets or higher birth order were excluded.
All data, including information on gestational duration, were obtained from the electronic medical records, which include all pregnancy follow-up information and delivery hospitalization characteristics.In Israel, all pregnancy follow-up and delivery expenses are covered by national health insurance law for all citizens, and pregnancy dating is based on self-report of first day of the last menstrual cycle, clinical evaluation, and first-trimester ultrasound evaluation.Ultrasound imaging is performed several times as part of the routine pregnancy follow-up.

| Statistical analyses
The independent variables were the twinning status of the index pregnancy (twin versus singleton gestation), gestational age at delivery of the index pregnancy and IPI.
The dependent variables were gestational duration and PTD incidence.
Covariables included maternal characteristics, such as age, infertility treatments, insufficient prenatal care (sufficient prenatal care was defined as having at least three prenatal care visits, the first during the first trimester), ethnicity (Bedouin-Arabs or Jewish based on medical records, used as a marker of socioeconomic position); pregnancy complications, including gestational diabetes mellitus; pregnancy-related hypertensive disorders, including pre-eclampsia and eclampsia; and progesterone treatment.
In order to identify a possible trend or a cut-point for a change in the risk for recurrent PTD, a comparison of recurrent PTD incidence was performed between singletons and twins based on gestational ages at delivery of the index pregnancy, and by centiles of IPI.
In order to identify possible confounding variables, covariables, including maternal and subsequent pregnancy characteristics, were compared between women with singletons and twins at the index pregnancy.Additionally, maternal and pregnancy (both index and the recurrent) characteristics were compared between cases of recurrent PTD and women without recurrent PTD.Variables associated with PTD recurrence and twinning status were included in multivariable models to test their confounding effect on the studied association.
Multivariable logistic models were used to study the association between twinning (the independent variables) and the risk for PTD recurrence (the dependent variables).
The risk for recurrent PTD was also studied by gestational ages <34 and 34-36 + 6 at delivery of the index pregnancy; by short (≤6 months) versus longer (>6 months) IPI; and by mode of conception in the subsequent pregnancy (spontaneous vs following fertility treatments).An interaction term was created between singleton delivery at the index pregnancy and any of the following variables: delivery at<34 gestational weeks at the index pregnancy, short IPI, and pregnancy following fertility treatments.The terms of each of the interactions were tested in multivariable logistic models to study the possible interaction effect of exposure to both factors.None of the interaction terms were significantly associated with the risk for recurrent PTD and were not included in the final models.All analyses were two-sided, with α ≤ 0.05 and β = 0.2.The analysis was done using SPSS software, version 27.0 (IBM, Armonk, NY, USA).
The study protocol was approved by the SUMC institutional review board committee (0357-19-SOR) and was exempt from the requirement for informed consent.

| RE SULTS
The study population included 15 590 women, of whom 1680 (10.8%) had twins in their index pregnancy and 13 910 (89.2%) had singletons.Maternal and pregnancy characteristics by twinning status in the index pregnancy are presented in Table 1.As compared with mothers of singletons in the index PTD, mothers of twins were older, more likely to conceive following infertility treatments and to TA B L E 1 Maternal and pregnancy characteristics by twinning status in the index pregnancy.women following twin versus singleton PTD (2.6% vs 6.7%, OR 0.37, 95% CI 0.27-0.50,P < 0.001).
Table 2 presents characteristics associated with PTD recurrence.
Women with recurrent PTD were less likely to have twins in the index pregnancy (5.5% vs 12.2%, OR 0.42, 95% CI 0.36-0.49),delivered at an earlier gestational week, and were more likely to smoke, have a history of recurrent pregnancy loss, and pregnancy complications in either the index or subsequent pregnancy.They were also more likely to have had infertility treatments in the subsequent pregnancy.
Several multivariable logistic models are presented in OR 0.50, 95% CI 0.32-0.76,P < 0.001).Exclusion of women with twins in the subsequent pregnancy had no effect on the results of this model.
Figure 1 presents PTD rates in subsequent pregnancy following index pregnancy PTD in twin and singleton deliveries, by gestational age at index pregnancy.The incidence of recurrent PTD varied between 2.4% and 26.7% following twin PTD and between 18.5% and 31.6%following singleton PTD.In all but the 30th gestational week, the rates of recurrent PTD were higher in singletons than in twins in the index gestation.
Additional multivariable logistic models are presented in Table 3

| DISCUSS ION
In this large, population-based, retrospective cohort study of PTD births, women with PTD in twin gestations were at lower risk for recurrent PTD than were women with a singleton PTD.This association was robust and remained significant while adjusting for maternal and pregnancy characteristics, and when stratifying the population by gestational age categories in the index pregnancy or by short or longer IPI.TA B L E 3 Adjusted odds ratios and 95% confidence interval (CI) for the association between preterm delivery (PTD) recurrence among twins versus singletons at the index birth.Abbreviations: CI, confidence interval; adj.OR, adjusted odds ratio.

Inter-pregnancy interval (IPI)
a In subsequent pregnancy.
The rates of recurrent PTD in the study were higher than PTD rates in the general population, which are estimated to be 7.6% in Israel; 20 however, this was expected as all women in this study are considered at high risk for PTD due to their history of PTD.Still, all women had a history of a single PTD, as only the first PTD per woman was included in the analysis.
Multiple etiologies are suspected to be involved in PTD recurrence, and it is likely that persistent factors, such as genetics and environmental exposures, contribute to the risk for PTD recurrence. 21e findings in this study support the benign nature of PTD in twin gestations as a risk factor for subsequent PTD compared with singleton gestation.Length of pregnancy is shorter in twin pregnancies, possibly due to the over-distension of the uterus which affects birth timing. 22This explanation, however, may account for only a portion of the earlier deliveries in twin gestations, as twin gestations are also at an increased risk for very early PTDs. 9In the present study, twins versus singletons were less likely to be delivered late preterm (34-36 gestational weeks) and more likely to be delivered earlier Preterm delivery (PTD) rates by gestational age at index pregnancy.
F I G U R E 2 Preterm delivery (PTD) rates by inter-pregnancy centiles.
(i.e., before 28 or 34 gestational weeks), when the uterus has not yet over-distended.Alternatively, PTD in twins may also be a result of pathological processes such as placental implantation pathologies, which are more prevalent among twins than among singleton gestations. 23,24Twin gestations are at increased risk for complications, including pregnancy-related hypertensive disorders and gestational diabetes mellitus, which may lead to iatrogenic PTD, 10 although labor induction in our study was significantly lower in twin gestation.
Singleton gestation PTD is necessarily a result of a pathological process, such as infections or placental pathologies, leading to the delivery or labor induction due to maternal or fetal jeopardy. 24The differences in the nature of PTD between singletons and twins may also explain the better survival of twins versus singletons delivered at similar early gestational weeks. 25Having a history of a prenatal pathological process, as found in singleton PTD, is a risk factor for recurrence, 1,26 leading to higher risk for PTD recurrence in singletons.
As earlier gestational age is a known risk factor for PTD recurrence and as gestational age was significantly lower among twins, the multivariable analysis accounted for gestational week at the index pregnancy delivery.Additionally, a stratified analysis was conducted by gestational age group in the index pregnancy.In both analyses, twins were at a lower risk for PTD recurrence.In studies involving twins only, the risk for (recurrent) PTD in subsequent pregnancy was significantly higher with earlier gestational week of the first PTD, 12 some studies suggesting that this risk was higher only when gestational age was <34 weeks. 14In the current study, twins delivered at either before or after 34 gestational weeks were at lower risk for PTD recurrence as compared with singleton PTDs.
The finding regarding progesterone treatment was surprising, as women receiving progesterone were at higher risk for PTD recurrence.This suggests that more severe cases of PTD or women with additional markers or risk factors for PTD have received the treatment, leading to a higher incidence of PTD among these women.
One of the strengths of the present study is the large, populationbased design, which includes a relatively large sample of women with PTD in twin gestations, a population that is under-studied and frequently excluded from perinatal research.The study design, comparing women with a history of twin versus singleton PTD, can translate into clinical decision-making when considering treatment plans based on obstetric history.Another strength is the sub-analysis by gestational age at the index pregnancy PTD, by IPI, and by mode of conception.These characteristics have been identified as confounding variables or variables with clinical significance, and stratification analysis has shown consistent and similar results.However, as this is a retrospective study, some limitations are nested in this design and the use of data not created or collected to answer a specific research question, including possible unmeasured confounding, such as information on maternal diet and changes in body mass index between pregnancies, as well as environmental exposures, which have been shown to be associated with PTD recurrence. 21Another limitation is the unexpected, relatively low incidence of progesterone treatment in the subsequent pregnancies.Possible explanations for this may be the wide range of years included in the study, including before progesterone treatment was recommended, the fact that approximately 60% of the study members were Bedouin Arab women, who are less likely to confer with the pregnancy follow-up program, 27 and there is a possibility that some medications were not identified in our database, leading to a misclassification of this documentation.
Unfortunately, in the study setting, there was no option to distinguish between spontaneous or iatrogenic PTDs, which is another study limitation.Although the possible effect and differences between the two types of PTD could not be addressed, it has been shown that having a history of either type of PTD is a risk factor for recurrent PTD of either type. 28eterm delivery is a major cause of offspring mortality and a significant cause of long-term morbidities and disabilities.Lowering the rate of this major, relatively prevalent, pregnancy complication has been declared an urgent priority by WHO. 29 While risk factors for PTD have been widely studied, the risk of recurrence following twin gestation has rarely been addressed, although it is clear that PTD etiology differs between singleton and twin gestations.
Close monitoring of the cervical length and prophylactic treatment is recommended in women who have experienced PTD, 19 in either twin or singleton gestations.While current research indicates a notably reduced risk of recurrence following twin gestation, it remains higher than that observed in the general population, and therefore a prevention strategy should be implemented in these women.

| CON CLUS IONS
Based on current findings, although women with a history of PTD in twins gestation are at lower risk for recurrence as compared to a women with a history of singleton PTD, the risk is still higher than women without such history, regardless of gestational age of the PTD, IPI or mode of conception.

Twins (n = 1680 [10.8%]) Singleton (n = 13 910 [89.2%]) OR; 95% CI P-value
Note: Data are n (%) unless noted otherwise.Abbreviations: CI, confidence interval; OR, odds ratio; SGA, small for gestational age.havesufficient prenatal care.Mothers with twin gestations were also less likely to undergo labor induction and more likely to have cesarean deliveries.Mean gestation length was shorter in twin pregnancies, and the incidence of late preterm (34-36 gestational weeks) was significantly higher among singletons, while among twins the incidence of deliveries at <34 gestational weeks was significantly higher.IPI was significantly longer following twin gestations.Several

Table 3
Maternal and pregnancy characteristics associated with preterm delivery (PTD) recurrence.