Recurrence Risk of Pregnancy Complications in Twin and Singleton Deliveries.

The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Consecutive Pregnancy Study collected retrospective electronic data from 20 hospitals in Utah, listed in Supplemental Table 1 , which included birthing individuals with at least 2 pregnancies (live births or stillbirths at ≥ 20 weeks’ gestation) who delivered between 2002 and 2010. Birthing individuals were included if they had ≥ 2 singleton or twin deliveries (n = 51,079) and were grouped into three groups based on the birth number of their first (index) and second (subsequent) pregnancies ((index–subsequent) singleton-singleton, n = 49,868, twin-singleton, n = 448, or singleton-twin, n = 723). Demographic, clinical, and recurrent risk characteristics were compared across singleton-singleton, twin-singleton, and singleton-twin subsequent pregnancy groups. All participating hospitals received approval and informed consent waivers from their institutional review boards. Extensive demographic, clinical history, labor and delivery, and outcomes data were analyzed, and the recurrence risk of PTB, GDM, GH, and PE & E were assessed. PTB was defined as obstetrically estimated gestational age < 37 completed weeks. GDM, GH, and PE & E were defined using electronic medical records and ICD-9 codes with birthing individuals considered to have the condition if a diagnosis was coded in either source. We first conducted descriptive analyses to compare the distribution of sociodemographic and clinical characteristics among birthing individuals who had singleton-singleton, twin-singleton, or singleton-twin subsequent pregnancies. Then we examined and compared the recurrence rate of each of the outcomes of interest (PTB, GDM, GH, and PE & E) by index-subsequent singleton-twin birth status. We subsequently calculated the recurrence risk for each outcome of interest using unadjusted and adjusted relative risks (aRR) with 95% confidence intervals (CI) which were estimated using Poisson regression models with robust variance estimators. 57 Unadjusted models controlled for the occurrence of the outcome of interest in the index pregnancy with no occurrence in the index pregnancy as referent. The adjusted models for all outcomes were controlled for second pregnancy pre-pregnancy body mass index (BMI), which was either categorized as obese or not obese (PE & E and GH models) or as underweight, normal weight, overweight, or obese (PTB and GDM models). The GDM model was also adjusted for maternal race/ethnicity, and the PTB model included additional adjustments for maternal race/ethnicity, interpregnancy interval, maternal age as a continuous variable, marital status, insurance, and smoking status during the second pregnancy. Different adjustments were made to different models due to sample size limitations. Additionally, birthing individuals with pre-existing diabetes and pre-existing hypertension were excluded from the GDM and GH models, respectively. All analyses were conducted in SAS 9.4, with tables finalized in Microsoft Word.

Of the 51,079 birthing individuals in the study, 723 (1.4%) had a twin pregnancy that followed a singleton pregnancy (singleton-twin), and 448 (0.9%) had a singleton pregnancy that followed a twin pregnancy (twin-singleton) ( Table 1 ). Compared to birthing individuals who had singletons in both pregnancies (singleton-singleton), those who had twins in either the first or second pregnancy were more likely to be white (88.1–89.7% vs 86.3%), had higher prevalence of second pregnancy cesarean deliveries (44.4–59.9% vs 18.5%), and were more likely to have private insurance (75.9–76.8% vs 73.7%). Birthing individuals who had a twin-singleton subsequent pregnancies had the greatest prevalence of second pregnancy pre-pregnancy obesity (31.3% vs 15.8–17.1%), whereas being overweight before pregnancy was most common in singleton-twin subsequent pregnancies (28.1% vs 20.4–23.2%). Birthing individuals who had twins in either the first or second pregnancy had greater rates of PTB (9.2%-62.8% vs 7.6%), PE & E (2.5–6.2% vs 2.0%), and GDM (4.4–5.4% vs 3.7%) than consecutive pregnancies of singletons. Birthing individuals who had singleton-twin subsequent pregnancies had the greatest prevalence of GH (4.4% vs 2.2–2.4%). The prevalence of second pregnancy PTB, GDM, GH, and PE & E among birthing individuals with or without the corresponding condition in the index pregnancy is described in Table 2 . The prevalence of PTB in birthing individuals with previous PTB was greater than in those without previous PTB regardless of plurality in the first or second pregnancy (singleton-singleton 30.7% vs 5.7%; twin-singleton 12.1% vs 3.8%; singleton-twin 86.5% vs 60.7%). This pattern of recurrence persisted for GDM (singleton-singleton 63.7% vs 2.2%; twin-singleton 36.8% vs 4.0%; singleton-twin 70.0% vs 2.6%), GH (singleton-singleton 14.3% vs 1.9%; twin-singleton 12.9% vs 1.5%; singleton-twin 14.8% vs 3.8%), and PE & E (singleton-singleton 15.0% vs 1.6%; twin-singleton 11.1% vs 1.3%; singleton-twin 19.1% vs 5.8%). The adjusted recurrence relative risks (aRR) for PTB, GDM, GH, and PE & E among singleton-singleton, twin-singleton, and singleton-twin pregnancies are depicted in Figure 1 , with unadjusted values shown in Appendix Figure 1 . Recurrence risks of PTB, GDM, GH, and PE & E were significantly increased in both unadjusted and adjusted relative risk models regardless of plurality in the first or second pregnancy, with the exception of the adjusted GH model for singleton-twin pregnancies. The PTB recurrence risk was aRR=5.1 (95% CI: 4.8 – 5.5) in singleton-singleton pregnancies, aRR=2.5 (95% CI: 1.1 – 5.9) in twin-singleton pregnancies, and aRR=1.4 (95% CI: 1.3 – 1.6) in singleton-twin pregnancies. The GDM recurrence risk was aRR=22.7 (95% CI: 20.8 – 24.8) in singleton-singleton pregnancies, aRR=5.2 (95% CI: 2.3 – 11.8) in twin-singleton pregnancies, and aRR=20.8 (95% CI: 11.6 – 37.2) in singleton-twin pregnancies. The GH recurrence risk was aRR=5.8 (95% CI: 5.1 – 6.6) in singleton-singleton pregnancies, aRR=7.6 (95% CI: 2.8 – 20.5) in twin-singleton pregnancies, and aRR=2.8 (95% CI: 1.0 – 7.7) in singleton-twin pregnancies. Finally, the PE & E recurrence risk was aRR=8.1 (95% CI: 7.0 – 9.3) in singleton-singleton pregnancies, aRR=9.2 (95% CI: 2.9 – 28.6) in twin-singleton pregnancies, and aRR=3.2 (95% CI: 1.2 – 8.2) in singleton-twin pregnancies.

In this study of singleton-singleton, twin-singleton, and singleton-twin subsequent pregnancy groups, we observed increased recurrence risk of PTB, GDM, GH, and PE & E, which was persistent regardless of plurality in the first and second pregnancy. The magnitude of recurrence risk varied between each of the subsequent pregnancy groups, with the greatest aRR of PTB and GDM in singletons that followed a singleton pregnancy (aRR=5.1 [95% CI: 4.8 – 5.5] and aRR=22.7 [95% CI: 20.8 – 24.8], respectively), and the greatest aRR of GH and PE & E in singletons that followed a twin pregnancy (aRR=7.6 [95% CI: 2.8 – 20.5] and aRR=9.2 [95% CI: 2.9 – 28.6], respectively). The etiology of singleton pregnancies is inherently different from that of multiple gestations. 53 – 55 Pregnancies of multiples have been associated with PTB, anemia, urinary tract infections, puerperal endometriosis, cesarean delivery, and gestational diabetes, in addition to hypertensive disorders of pregnancy. 55 , 56 In some cases, the occurrence of these outcomes may be due to the increased physiological burden of carrying multiple gestations, however, little is known about the risk of recurrence of these conditions in singletons that follow a pregnancy of multiples, or if birthing individuals are more likely to have these conditions during a twin pregnancy after an affected singleton pregnancy. Our PTB recurrence rates were 30.7% (aRR=5.1 (95% CI: 4.8 – 5.5)) in singleton-singleton pregnancies, 12.1% (aRR=2.5 (95% CI: 1.1 – 5.9)) for twin-singleton pregnancies, and 86.4% (aRR=1.4 (95% CI: 1.3 – 1.6)) in singleton-twin pregnancies. Our results for PTB in singleton-singleton pregnancies was similar to the odds ratio (OR) from a single cite study in Texas where PTB was assessed as birth at ≤ 35 weeks (OR=5.6 (95% CI: 4.5 – 7.0), 11 however, unlike this study we found that singleton-twin pregnancies were also at increased risk of PTB recurrence (aRR=1.4 (95% CI: 1.3 – 1.6)) vs (OR=1.9 (95% CI: 0.46 – 8.14)). Our aRR estimate for twin-singleton pregnancies were also lower than the OR in a recent meta-analysis of 8 cohort studies conducted in South Carolina, Sweden, Texas, Illinois, Delaware, the Netherlands, Korea, and Canada (aRR=2.5 (95% CI: 1.1 – 5.9) vs (OR=4.34 (95% CI: 2.83 – 6.65)), 1 as well as the OR estimates from a Chicago based single cite cohort study (OR=5.0 (95% CI: 1.1 – 22.9)) 13 and a nationwide study in the Netherlands (OR=6.9 (95% CI: 3.1 – 15.2)), 9 but were similar to estimates from an older South Carolina study (RR=2.87 (95% CI: 1.02 – 8.09)). 12 Despite these differences in estimates, many of the confidence intervals from these studies were wide and overlapped with our observed estimates. Number of previous PTBs, gestational age at birth, and whether the PTB is spontaneous, indicated, or elective should also be considered, although outside of the scope of this analysis. 1 , 15 To our knowledge, no studies have been conducted to compare the recurrence of GDM in subsequent pregnancies involving multiple gestations where the birth number in the index and subsequent pregnancy is specified. In our study, the recurrence rate of GDM was 63.7% (aRR=22.7 (95% CI: 20.8 – 24.8) in singleton-singleton pregnancies, 36.8% (aRR=5.2 (95% CI: 2.3 – 11.8)) in twin-singleton pregnancies, and 70.0% (aRR=20.8 (95% CI: 11.6 – 37.2)) in singleton-twin pregnancies. Our singleton-singleton GDM recurrence rate (63.7%) is consistent with a systematic review that cited a range of 30–85% recurrence in consecutive singleton pregnancies, 18 and lies between two recent international studies on recurrent GDM in singleton-singleton consecutive pregnancies (72.3% for a single site in Germany 17 and 47.2% for two perinatal medical centers in Japan 20 ). We observed a GH recurrence rate of 14.3% (aRR=5.8 (95% CI: 5.1 – 6.6)) in singleton-singleton pregnancies, 12.9% (aRR=7.6 (95% CI: 2.8 – 20.5) in twin-singleton pregnancies, and 14.8% (aRR=2.8 (95% CI: 1.0 – 7.7)) in singleton-twin pregnancies. These findings all exceeded the overall recurrence rate of GH reported in 2015 in a meta-analysis of 64 studies, which varied in the inclusion and assessment of multiples (8.6%). 51 Our estimates more closely resembled results from a nation-wide study in France that reported 11.3% GH recurrence with multiples included in the study but not examined as an effect modifier for GH. 28 Our recurrence rates were markedly lower than those observed in a multi-hospital network in Australia (26%), 25 and a single cite in Iceland (46.8%) 24 which were both assessed without regard for plurality. These differences may reflect differences in the populations being studied, such as obesity, activity levels, nutrition, and access to health care. Previous studies have noted that PE during a pregnancy of multiples may be more related to multiplicity than other risk factors, 29 a theory that has been supported across multiple studies. 26 , 48 – 50 What is less demonstrated in the literature is whether singleton pregnancies that occur after an affected pregnancy of multiples are at increased recurrence risk of hypertensive disorders. Our estimates for PE & E recurrence included 15.0% (aRR=8.1 (95% CI: 7.0 – 9.3)) for singleton-singletons, 11.1% (aRR=9.2 (95% CI: 2.9 – 28.6)) in twin-singletons, and 19.1% (aRR=3.2 (95% CI: 1.2 – 8.2)) in singleton-twins. These findings were similar to studies which reported PE recurrence rates of 11.6% in consecutive singleton pregnancies in Utah 41 and 14–15% in subsequent pregnancies of singletons or multiples in Australia 25 as well as a Swedish study that assessed recurrent PE in 2–3 consecutive pregnancies including multiple gestations, where they found 14.7% recurrence in second pregnancies after PE in the index pregnancy. 29 This study also reported recurrence of PE of 6.3% in singletons after a PE affected twin pregnancy, which was notably lower than our estimate (11.1%). 29 This study was strengthened by its large sample size as we were able to assess plurality as well as several pregnancy conditions. Due to the objective of this study, our analysis was focused primarily on recurrence and differences in recurrence risk in the presence of plurality, rather than predictors and mechanisms of the conditions themselves. While there is considerable evidence of associations between each of these outcomes with various demographic factors, maternal health markers, and mechanisms during pregnancy and delivery, analyzing these factors was outside of the scope of this analysis. Based on variables availability and sample size, we were able to adjust for several known risk factors for these conditions. We adjusted for pre-pregnancy BMI 7 , 10 , 17 , 24 , 30 , 32 – 41 (at the time of the second pregnancy) for each outcome of interest, however due to limited sample size, we were unable to adjust for additional confounding factors in the GH and PE & E models. This study was also limited geographically, as the data were drawn from a single state with a low representation of non-White racial groups, which may limit generalizability to other states within the US. In conclusion, we have demonstrated that PTB, GDM, PE & E, and GH in an index pregnancy of either a singleton or twins result in increased recurrence risk in a subsequent pregnancy of either a singleton or twins, with the greatest recurrence risk of PTB and GDM in singleton-singleton consecutive pregnancies, and the greatest recurrence risk of PE & E and GH in singletons that followed a twin pregnancy. These findings on the risk of recurrence are important for prognostic counseling. Future studies with broader geographical reach and larger sample sizes of consecutive pregnancies with plurality in the index or subsequent pregnancy are needed.

Recurrence risk of preterm birth (PTB), 1 – 15 gestational diabetes mellitus (GDM), 16 – 23 and hypertensive disorders including gestational hypertension (GH), preeclampsia, superimposed preeclampsia, and eclampsia (PE & E) 24 – 30 have been observed in previous studies. The recurrence of these conditions has been associated with maternal race/ethnicity, 3 , 4 , 12 , 14 , 18 , 23 , 26 , 31 maternal age, 2 , 7 , 10 , 14 , 23 maternal weight, 7 , 10 , 17 , 24 , 30 , 32 – 41 gestational weight gain, 7 , 22 , 24 , 42 reduced number of prenatal care visits, 2 short interpregnancy intervals, 2 , 10 , 22 , 43 family history, 17 , 22 , 44 – 47 and marital status. 2 , 12 Most studies examining the recurrence of PTB, GDM, GH, and PE & E have focused on an individual condition in consecutive pregnancies of singletons, 2 , 3 , 6 – 8 , 15 , 20 , 27 , 28 , 36 , 38 , 40 – 43 , 48 – 50 with fewer studies assessing the recurrence of these conditions in subsequent pregnancies including multiple gestations, 5 , 9 – 13 , 23 , 25 , 29 , 51 , 52 particularly for GDM, 23 GH, 25 , 51 and PE & E. 25 , 29 , 51 The etiology of multiple gestations differs from that of singleton pregnancies, 53 – 55 with additional physiological stressors 55 , 56 that may modify the recurrence risk of these conditions. The timing and birth number of pregnancies of multiples, whether in the index or subsequent pregnancy, should also be considered. Studies examining the risk of PTB recurrence have shown that plurality plays a key role, especially in singletons that follow a twin pregnancy. 1 , 9 , 10 , 13 A systematic review of PTB recurrence found a four-fold (95% CI 2.83 – 6.65) increased risk of PTB in singletons following an index pregnancy of preterm twins. 1 Findings also suggest that GH and PE & E have decreased recurrence risk in subsequent pregnancies after an index pregnancy of multiples, 29 , 51 with as much as half the risk of recurrence, 29 however, the birth number of these subsequent pregnancies is not specified. Additionally, a study on GDM recurrence found increased risk in subsequent pregnancies of multiples (adjusted relative risk: 2.13 (95% CI: 1.34 – 3.38), 23 still without birth number specification, which further demonstrates the critical need for understanding unique risk profiles across conditions based on the presence of multiples in either the index or subsequent pregnancy. To add to the above body of literature and to expand on previous findings, we compared the recurrence risk of PTB, GDM, GH, and PE & E in subsequent pregnancies of singletons following a singleton pregnancy, singletons following a twin pregnancy, and twins following a singleton pregnancy. This study will be the first to assess these four conditions using the same population of birthing individuals and will improve current understanding by comparing subsequent pregnancies including multiples, with birth number specified in the index and subsequent pregnancy. The findings of this study can be used for prognostic counseling.