The impact of prepregnancy body mass index on pregnancy and neonatal outcomes

Methods

This is a Geisinger IRB-approved (#2017-0520) retrospective cohort study of pregnant women who delivered singleton births within all five delivery hospitals in the Geisinger Health System in Pennsylvania between October 16, 2007 and December 3, 2023. We queried the electronic health records (EHRs) and included all pregnancies in which the maternal prepregnancy BMI was greater than or equal to 18.5 kg/m². BMI was assigned based on measurements taken between three months prior to conception and one month after their last menstrual period. We excluded pregnancies complicated by multifetal gestations and where a pre-gravid BMI was not recorded or classified as underweight. Neonates were only included if they were linked in the EHR to the mother. The cohort was stratified by the primary predictor of interest, i.e., the prepregnancy BMI category: normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), and class I (30.0–34.9 kg/m²), class II (35.0–39.9 kg/m²), and class III (40.0 kg/m²) obese [23].

Demographic variables were assessed within each category of BMI, including age at delivery, self-reported race/ethnicity, marital status, insurance, parity, education, tobacco use, pregestational diabetes mellitus, and chronic hypertension (cHTN). The primary outcome was composite maternal morbidity (i.e., gestational hypertension [GHTN], preeclampsia, and GDM), adjusted for diabetes mellitus and cHTN. Secondary pregnancy outcomes included preterm prelabor rupture of membranes (PPROM), preterm delivery (PTD<37 and <32 weeks), induction of labor (IOL), CD, and postpartum hemorrhage (PPH). Neonatal outcomes included a composite adverse outcome (that included stillbirth, intraventricular hemorrhage [IVH], hypoglycemia, RDS, APGAR<7 at 5 min [in cases of live births], shoulder dystocia, and neonatal intensive care unit [NICU] admission), birthweight, fetal growth restriction (FGR), and macrosomia (birthweight>4,000 g). IVH and RDS were assessed within the first six months of life. Missing data were excluded.

For several of the outcomes, algorithms were created to improve demarcation of these outcomes, some of which are delineated in a separate publication [20]. Since that publication, we have updated the definition of PPH to reflect ≥1,000 cc of blood loss at delivery, regardless of mode. For some of the outcomes assessed, billing codes (i.e., International Classification of Diseases [ICD-9 and ICD-10] codes) were utilized as a basis, i.e., PPROM, IVH, and RDS. Some outcomes were discrete fields that were able to be abstracted from the EHR, i.e., birthweight and APGAR scores.

Data are reported as frequency and percentage for categorical variables and mean and standard deviation (SD) for continuous variables. To model the binary outcomes of interest, a generalized mixed linear regression (log-binomial) model for the binary outcomes with a random effect to account for repeated pregnancies was performed. For the continuous outcome of birthweight, a mixed linear model including a random effect to account for the repeated pregnancies was performed. We controlled for maternal age at delivery, race/ethnicity, marital status, medical assistance, parity, prenatal smoking status, diabetes mellitus, and cHTN. The following outcomes were controlled for history of the same: GHTN, preeclampsia, GDM, PPROM, PTD, CD, PPH, stillbirth, macrosomia, and shoulder dystocia. The odds ratios (ORs) or applicable estimate and 95 % CIs are reported for each BMI class. Those with normal prepregnancy BMI were used as the reference group. A test of trend was performed for each outcome utilizing the described modeling techniques for binary or continuous outcomes as appropriate, where the variable BMI class is included in the model as continuous. We did a stratified analysis by parity: nulliparity vs. multiparity. A p value less than 0.05 was considered significant. All analyses were performed using SAS 9.4v (SAS Institute, Inc., Cary, NC).

Results

A total of 40,256 patients with 55,202 singleton pregnancies and available information on 53,664 neonates were included. Among the pregnancies, 21,102 (38.2 %) of the patients were normal weight, 14,388 (26.1 %) were overweight, 9,268 (16.8 %) were class I obese, 5,675 (10.3 %) were class II obese, and 4,769 (8.6 %) were class III obese. The average maternal age at delivery was 28.5 years (SD 5.6). The majority of women were non-Hispanic White (79.7 %), married (58.7 %), privately insured (51.8 %), and parous (61.0 %); 16.8 % smoked prenatally, 2.7 % had pregestational diabetes, and 5.0 % had pre-existing hypertension. Demographic characteristics varied among BMI categories (Table 1). Table 2 shows the association between each BMI classification and maternal and neonatal adverse outcomes of interest. After controlling, we observed statistically significant increasing ORs across BMI classes for several adverse outcomes such that as BMI increased from one class to the next, so did the odds of certain adverse outcomes including IOL and CD (Table 2). Higher BMI classes had a protective effect on the incidence of FGR, most significantly in overweight and class I patients as compared to those who were of normal weight (OR 0.73, 95 % CI 0.66–0.81; OR 0.70, 95 % CI 0.62–0.80, respectively). There was a significantly increased risk of a PTD<32 weeks and a significant test of trend for increased odds of macrosomia as compared to normal-weight patients as BMI class increased. Stillbirth (n=24) and IVH (n=200) cases were small in numbers, and therefore, comparison across groups may not have clinical relevance. There is an increase in composite maternal morbidity with increasing BMI class: those who are overweight, class I, class II, and class III obese were 1.72, 2.66, 3.35, and 4.40 times as likely to have maternal morbidity as compared to those with normal-pregnancy BMI (Figure 1), respectively. These findings are maintained in the subanalysis of nulliparous patients vs. multiparous patients (data not shown). An increased risk was also seen for composite neonatal morbidity, although the increase was not as strong as maternal morbidity.

Figure 1:

Odds of composite maternal and neonatal morbidity for various BMI classes. Composite maternal morbidity includes gestational hypertension, preeclampsia, and gestational diabetes mellitus. Composite neonatal morbidity includes stillbirth, intraventricular hemorrhage, hypoglycemia, respiratory distress syndrome, and APGAR score<7 at 5 min and neonatal intensive care unit admission.

Discussion

As BMI class increased, the odds of composite pregnancy morbidity increased to as much as 4.4 for a patient with class III obesity as compared to one with normal prepregnancy BMI. Composite neonatal morbidity also increased. While it seems intuitive that patients with class III obesity would be at the highest risk for adverse outcomes, this is one of the larger studies in the United States assessing these outcomes for the individual classes based on prepregnancy BMI. Patients with any class of obesity were at a significantly higher risk of developing hypertensive disorders of pregnancy or gestational diabetes as compared to those of normal weight. In fact, those with class III obesity were at significantly increased odds of developing GHTN (4.55), preeclampsia (2.35), and GDM (1.71), compared to their counterparts with normal prepregnancy BMI. As the obesity class increased, so did the odds of IOL, CD, PTD<32 weeks, and PPH. Obesity had a significant protective effect for FGR. The increase in macrosomia is clinically relevant, as those with class II obesity had 1.63 increased odds of having a macrosomic neonate as a patient with normal weight. Parity did not significantly affect these results.

Previous studies have demonstrated that pregnant women with obesity are at a higher risk of morbidity compared to women who are not obese [23], [24], [25], [26], [27], [28]. Studies have shown similar results of increasing risk of GHTN, preeclampsia, gestational diabetes, and CD for patients with obesity and morbid obesity [17, 26]. A similar study in the United Kingdom showed that individuals with BMI>30 kg/m² have a 3.6 increased odds of developing gestational diabetes compared to patients with a normal BMI [16]. We were able to confirm these findings and additionally were able to include women with BMI>40 kg/m² and demonstrate that those patients are at highest risk of complications compared to patients with a normal BMI. Increased risk of GDM supports routine early screening for gestational diabetes (with a 1 h glucose challenge test) [14], [15], [16, 26, 29] for women with obesity. Multiple studies have also shown that obesity is an independent risk factor for CD, with an OR of 3.0 for patients with BMI≥35 kg/m² [26]. Our study showed increased odds for CD with each successive BMI class when compared to the normal BMI class, with patients with class III obesity having an OR of 2.47 (95 % CI 2.17–2.81). Although our institution recommends IOL for patients with class III obesity at 39 weeks, the increased CD rate does not appear to be related to this practice based on a prior study conducted at our institution [30].

Our study further demonstrates that as BMI increases from obesity class I to II to III, there is a significantly increased risk of developing GDM. Although some studies have stratified classes of obesity, they utilized ICD codes to identify outcomes, but ICD codes are considered unreliable overall [17, 18, 21]. Our data support the findings in Kim et al. [17], yet one-third of their cohort did not have recorded BMIs and the data were abstracted from ICD-9 codes alone.

One strength of our study is the large number of patients analyzed utilizing an electronic medical record system. Analyses were adjusted not only for potential confounders, such as demographic characteristics, but also for history of the outcome in prior pregnancy because this is generally the biggest risk factor for adverse outcomes in subsequent pregnancies. Additionally, previous studies utilized BMI at the time of delivery or combined all classes of obesity into one group, whereas our study assessed the individual BMI classes prepregnancy to assess how the risks for each specific obesity class.

Limitations include that our population is mostly non-Hispanic White, which may limit generalizability to other populations. Additionally, some adverse outcomes such as preterm birth, hemorrhage, and CD occur more frequently in racial/ethnic minority groups and therefore may not have occurred frequently in our population [31], [32], [33]. Despite our large numbers, certain outcomes (e.g., stillbirth and neonatal IVH) occurred at too small a frequency to be able to assess for significant differences between the classes. Neonatal charts were excluded if they were not linked to their mother, but because this represents only 1,538 neonates, we feel that it is unlikely that this would have significantly impacted the results. Finally, definitions of some of the variables changed throughout the study period, but we were consistent with our definitions as outlined in the Methods section.

Conclusions

In conclusion, this study showed an increasing trend in the risk of composite maternal and composite neonatal adverse outcomes as prepregnancy BMI increased. This information can be applied to facilitate antenatal counseling of pregnant women by providing prepregnancy BMI class-specific pregnancy outcomes.