Association of Gestational Weight Gain With Infant Morbidity and Mortality in the United States

Key Points Question Is there an association between gestational weight gain (GWG) and adverse infant outcomes among women across body mass index (BMI) categories? Findings In this cohort study of 15.8 million mother-infant dyads, infant morbidity and mortality risks showed U-shaped associations with GWG. Inadequate GWG and weight loss were associated with increased risks across BMI categories, but the effect sizes were smaller among women with obesity. Meaning In this study, extreme GWG ranges were associated with increased risks of infant morbidity and mortality across BMI categories, suggesting that prenatal weight loss or weight maintenance should not be encouraged.


Introduction
Adverse pregnancy outcomes are associated with inappropriate gestational weight gain (GWG).
Inadequate GWG increases the risks of preterm birth and being born small for gestational age. 1,2 Excessive GWG is linked to pregnancy complications such as gestational hypertension and diabetes, cesarean delivery, postpartum weight retention and obesity in later life, and adverse outcomes to the offspring such as being born large for gestational age, macrosomia, and childhood obesity. [1][2][3] The existing recommendations for GWG were issued by the US National Academy of Medicine (NAM) in 2009. 4 The recommendations were based on previous studies on potential associations of GWG with maternal and child health, and provided a single range of GWG (5.0-9.0 kg) for women with obesity, owing to insufficient evidence stratified by the severity of obesity. In 2019, the Maternal Obesity and Childhood Outcomes Study Group in the LifeCycle Project (henceforth referred to as "LifeCycle Project") conducted a meta-analysis of individual participant data of 196 670 women and suggested respective GWG ranges for class 1 (2.0-6.0 kg), class 2 (Յ4.0 kg), and class 3 (Յ6.0 kg) obesity. 2 Apart from the LifeCycle Project, growing evidence also suggested lower levels of GWG than the NAM recommendations, or even weight loss, for obese women. [5][6][7][8] In examining the associations between GWG and infant outcomes, NAM, the LifeCycle Project, and many other studies used intermediate outcomes, such as preterm and postterm birth, born small for gestational age, and born large for gestational age, as proxy markers of ultimate health outcomes-infant morbidity and mortality. [1][2][3][4] By contrast, directly addressing infant morbidity and mortality offers more informative and reliable estimates, especially if the total effects are not fully reflected by intermediate outcomes. 4 However, ultimate health outcomes of the infant associated with GWG are understudied, to our knowledge. Inconsistent results have been reported on the benefits and risks of low GWG and weight loss associated with infant mortality among women with obesity. [9][10][11][12][13][14] Studies on infant morbidity were fewer and showed that GWG outside the recommended ranges was associated with increased risks of neonatal intensive care unit (NICU) admission, assisted ventilation, and seizures. [15][16][17] Moreover, this limited evidence on the associations of GWG with infant morbidity and mortality was rarely stratified by maternal body mass index (BMI), 9,[11][12][13][14][15] especially by the severity of obesity. 11 Infant morbidity and mortality are important public health issues. The infant mortality rate is used widely as an overall indicator of reproductive health for a population, and neonatal anomalies can have permanent and devastating effects on the offspring and the family. The risks of infant morbidity and mortality should be accounted for when recommending GWG ranges for optimizing pregnancy outcomes. Based on analysis of nationwide birth and infant death data from 2011 to 2015, this study aims to assess the optimal GWG ranges associated with reduced risks of infant morbidity and mortality across BMI categories.

Study Participants
We used birth cohort linked birth and infant death data sets from the National Center for Health Statistics (NCHS) database, registered in all 50 states in the US and Washington, DC. 18  weight gain of 30 kg or more, to assess the optimal ranges of GWG.

Assessment of Adverse Infant Outcomes
We adopted 2 composite outcome measures: (1) significant morbidity of the newborn infant (derived from medical records), 22 defined as any presence of assisted ventilation, admission to the NICU, surfactant therapy, antibiotic therapy, or seizures; and (2) infant mortality younger than 1 year of age (reported by medical examiners or coroners), 25 consisting of death in less than 1 hour, 1 to 23 hours, 1 to 6 days, 7 to 27 days, and 28 to 365 days after birth, excluding nonnatural death (accident, suicide, homicide and self-inflicted death, and death from pending or undetermined causes). The definition of each type of morbidity was provided in the specifications of birth certificate. 22

Statistical Analysis
Statistical analysis was performed from February 11 to October 14, 2021. Given the effect modification of BMI, 2,5-9 all statistical analyses were stratified by the 6 BMI categories. We calculated the prevalence of composite and individual outcomes across GWG groups. Adjusted odds ratios (AORs) for composite outcomes were estimated by comparing the odds of each GWG group with the odds of all the rest of the groups, 2 using logistic regression models adjusted for putative confounders that were associated with GWG (eTable 1 in the Supplement), including maternal age (<25, 25-34, and Ն35 years; categorical), race and ethnicity (as BMI classifications vary by race and ethnicity), educational level, marital status, smoking before or during pregnancy, parity, sex of the infant, place of birth, and type of health insurance. 26,27 We did not adjust for other pregnancy outcomes that could act as mediators between GWG and infant outcomes, such as maternal complications during pregnancy. 2,28 We calculated 95% CIs of AORs based on the normal distribution. The optimal GWG range was defined as all GWG groups with significant risk estimates (AOR) less than 1, and GWG groups with nonsignificant AORs less than 1 but between 2 significant AORs less than 1. The overall optimal GWG range was defined as the overlapping GWG ranges for reduced risks of infant morbidity and mortality (eMethods in the Supplement).
We performed 5 sensitivity analyses. First, we reanalyzed the associations in infants born in 2015 to reduce the risk of selection bias, given the lowest rate of missing values in GWG or BMI (5.5%; eFigure in the Supplement). Second, we corrected P values to control type I error of multiple testing, using the Benjamini-Hochberg approach that was statistically powerful for large numbers of comparisons. 29 Third, we reanalyzed the associations in a subset restricted to Hispanic, non-Hispanic Black, and non-Hispanic White races and ethnicities, given that a different BMI classification is used for evaluating body fat percentages in Asian populations. 30 Fourth, neural tube defects could be associated with low birth weight and insufficient GWG, but were less likely to result from inappropriate GWG 4 ; hence, they were excluded to address potential reverse causality. We did not exclude other congenital anomalies that could be associated with inappropriate GWG, such as congenital heart disease. 31 Fifth, infant morbidity regarding specific disorders was examined by removing NICU admission from the composite variable of morbidity. Missing data of variables used in this study were described in eFigure and eTable 2 in the Supplement. All statistical tests were 2-sided and P < .05 denoted statistical significance. All statistical analyses were performed in R, version 3.6.2 (R Group for Statistical Computing).

Results
Of the 15  eTable 4 in the Supplement shows that 8.8% of the newborns experienced significant morbidity, with the lowest prevalence among infants delivered by normal weight women (8.0%) and the highest among infants delivered by women with class 3 obesity (12.4%); 3.4 per 1000 infants (0.34%) died within 1 year of birth, with the lowest prevalence among infants delivered by normal weight women (0.28%) and the highest among infants delivered by women with class 3 obesity (0.58%).
The absolute risks of infant morbidity and mortality were lowest in the middle groups of GWG, with U-shaped patterns (Figure 1 and Figure 2) (numerical results in eTables 5 and 6 in the Supplement). Infant morbidity (Figure 1) was most prevalent in the lowest GWG groups among underweight (<8 kg GWG, 14.7%) and normal weight women (<0 kg GWG, 16.6%) and most prevalent in the highest GWG groups among women with obesity (obesity class 1: Ն30 kg GWG, 15.2%; obesity class 2: Ն20 kg GWG, 13.7%; obesity class 3: Ն20 kg GWG, 17.4%) (eTable 5 in the Supplement).
Infant mortality ( Figure 2) was most prevalent in the lowest GWG groups and remained at low levels in middle and high GWG groups across BMI categories.
The associations of GWG with infant morbidity (Figure 3) and mortality (Figure 4) also showed U-shaped patterns. Middle groups of GWG were associated with AORs less than 1, and extremely low and high GWG groups were associated with AORs greater than 1 (numerical results in eTables 7 and 8 in the Supplement). The curvilinear associations interacted with BMI categories. Infant mortality was associated with GWG less than 6.0 kg for women with class 1 obesity (4.0 to <6.0 kg GWG: AOR, The overall optimal GWG ranges for reduced infant morbidity and mortality were 12.0 to less than 24.0 kg for women with underweight and normal weight BMIs, 10.0 to less than 20.0 kg for women with overweight BMIs, 8.0 to less than 16.0 kg for women with class 1 obesity, 6.0 to less than 16.0 kg for women with class 2 obesity, and 6.0 to less than 10.0 kg for women with class 3 obesity  Figure 5). Comparing NAM recommendations, the LifeCycle Project, and the present study, the results showed that the optimal ranges from the present study were wider, with higher upper bounds. With regard to the lower bounds, our results approximated to NAM recommendations for women with underweight (12.0 vs 12.5 kg) and normal weight (12.0 vs 11.5 kg) BMIs, but for women with overweight BMIs and women with class 1, 2, or 3 obesity, our results (10.0, 8.0, 6.0, and 6.0 kg, respectively) appeared to be higher than those from the LifeCycle Project (2.0, 2.0, <0, and 0 kg, respectively) and NAM recommendations (7.0, 5.0, 5.0, and 5.0 kg, respectively).

JAMA Network Open | Pediatrics
In the first sensitivity analysis, the AORs among births in 2015 were comparable to those from the main analysis, while the optimal GWG ranges became slightly narrower, possibly owing to the smaller sample size and fewer significant 95% CIs (eTables 9 and 10 in the Supplement). The other

Discussion
We analyzed data from 2011 to 2015 from a nationwide database and explored optimal GWG ranges for reducing adverse infant outcomes across BMI categories. Associations between GWG and adverse pregnancy outcomes were modified by BMI. It has been reported that the associations between inadequate GWG and infant mortality were stronger among women without obesity, 9 and our results also showed that the ranges of optimal GWG generally decreased with BMI categories and the severity of obesity. Recommending a single GWG range (5.0-9.0 kg) for women with obesity may be of concern, especially with increasing obesity in pregnant women (increased by 8% from 2011 to 2015 32 ). Research based on maternal outcomes and intermediate infant outcomes has suggested lower levels of GWG or even weight loss for women with obesity. 2,5-8 However, Bodnar et al 11 analyzed Pennsylvania birth and infant death data from 2003 to 2011, showing that weight loss and low GWG (z score <−1 SD) were associated with increased risks of infant mortality in 219 000 women with class 1 or 2 obesity. We categorized GWG into 2.0-kg groups and found that increased risks of infant mortality were significantly associated with GWG less than 6.0 kg for women with class 1 Adjusted odds ratios (AORs) were adjusted for age (<25, 25-34, and Ն35 years; categorical), race and ethnicity, educational level, marital status, smoking before or during pregnancy, parity, sex of infant, place of birth, and type of health insurance. The numerical results are shown in eTable 7 in the Supplement. obesity, less than 4.0 kg for women with class 2 obesity, and less than 2.0 kg for women with class 3 obesity. Our study corroborates the finding by Bodnar et al 11 that weight maintenance and weight loss during pregnancy could increase adverse infant outcomes even for women with obesity.

JAMA Network Open | Pediatrics
Our results added direct evidence to the increased health risk to offspring of inadequate GWG.
There are several causal explanations. First, insufficient GWG is associated with being born small for gestational age, preterm birth, and decreased head circumference, 1,2,33 and in turn linked to higher risks of hypoglycemia, hypocalcemia, immaturity of the respiratory system, serious infections, and cerebral palsy due to hypoxia. 34-37 Second, low GWG can reflect a severe negative energy balance and a possible decrease in body mass of the mother, given that approximately 6 to 7 kg in the total GWG is composed of the products of conception (fetus, placenta, and amniotic fluid) and expansion Adjusted odds ratios (AORs) were adjusted for age (<25, 25-34, and Ն35 years; categorical), race and ethnicity, educational level, marital status, smoking before or during pregnancy, parity, sex of infant, place of birth, and type of health insurance. The numerical results are shown in eTable 8 in the Supplement. of blood and other extracellular fluid. 38 Such a negative energy balance could be associated with the viability of the fetus. 39 Alternatively, low GWG may be a physiological sign of suboptimal health conditions, such as depression and anxiety, prediabetes, and suboptimal fetal and placental development. 34,40 Excessive GWG has been associated with increased risks of adverse maternal outcomes and infants born large for gestational age for women with an overweight BMI and women with obesity, 2,4 and our results showed similar associations with infant morbidity and mortality. The mechanisms for adverse outcomes associated with excessive GWG remain understudied, but researchers have explored the pathologic characteristics of pregnancy among obese women. Obesity-related insulin resistance and hyperinsulinemia is associated with placental functions and neonatal hypoglycemia, which can cause severe brain damage if untreated. 41,42 This could partially explain the increased risks from excessive GWG, considering that neonatal hypoglycemia was also observed among women who gained weight above the recommendations. 17 A higher inflammatory state in placental and fetal tissue is another potential explanation, which has been reported to be associated with maternal obesity and excessive GWG. 43,44 The associations of inappropriate GWG with adverse infant outcomes could be different from those with maternal outcomes. Excessive GWG can be associated with larger adverse effects on the mother than the fetus, and is linked to markedly increased risks of pregnancy complications, cesarean delivery, and postpartum weight retention. 2,[45][46][47] Conversely, reducing GWG among women with obesity may trade better maternal outcomes with higher risks for infants. This is not a simple balancing between risks and benefits for the mother and the offspring. For women with obesity without chronic diseases and pregnancy complications related to obesity, minimal GWG that is safe for the offspring is recommended. For women with extreme obesity (BMI Ն40) or those with obesity-related diseases, prenatal care clinicians may adopt individualized GWG recommendations after comprehensive evaluations, and weight maintenance and weight loss should be recommended with great caution. Optimizing nutrient intake and appropriate physical exercise could be a possible way to minimize weight gain and reduce risks for both mothers and infants.

Strengths and Limitations
This study has some strengths, including directly addressing the ultimate health outcomes of the offspring, including morbidity and mortality; the large sample that allowed for cross-stratification by BMI and GWG; and a relatively short time span of 5 years to avoid effect modification of social transition. In addition, we corrected GWG by gestation length. Given that methods of standardizing GWG by gestation length have not yet been established, most previous studies used total GWG,2,27,48 which could overestimate the associations owing to the confounder of gestation length. Some studies adjusted for it in regression analyses, 2,49,50 which also would be problematic, as preterm and postterm were usually considered as intermediate outcomes in the pathway from risk factors to adverse infant outcomes. Therefore, we used GWG equivalent at 40 weeks to reduce misclassification between appropriate and inappropriate GWG for different gestation lengths, and to facilitate comparisons with the existing recommendations. 23 In addition, we separately examined infant morbidity and mortality, instead of creating a composite variable of any adverse infant outcomes, given the large disparities in the prevalence. The overall optimal GWG range was defined as the overlapping part to avoid automatic weighting by the prevalence when merging the 2 outcomes.
This study also has some limitations. First, prepregnancy weight and height from birth certificates were reported by mothers. 22 Studies showed that they were generally representative and consistent with medical records 51,52 but could be less reliable at the extremes. 53 Second, the sensitivity and detection rates of some adverse conditions were low, 54 but the associations were unlikely to be biased considering that sensitivity would not vary with GWG. Third, information on spontaneous abortion and stillbirth was unavailable, which could be associated with underestimation of the effect sizes (AORs) of inappropriate GWG. However, these severe events were associated with more extreme GWG ranges, 55 and less likely to change the results of optimal GWG ranges. Fourth, various types of morbidity were considered as equally important when merged into the composite variable of "infant morbidity," although this process automatically weighted each morbidity by its prevalence. Fifth, standardizing GWG to 40 weeks based on a linear pattern could be inaccurate, because the rate of GWG is lower in the first and late third trimesters in an S-shaped pattern.
However, this method attenuated misclassification and facilitated comparisons with existing recommendations. 23 Sixth, the optimal ranges from our results were relatively wide, as we used severe outcomes that were more likely to occur in extreme intrauterine environments. Maternal outcomes should also be accounted for to avoid maternal adverse effects associated with excessive GWG. Seventh, the remaining confounders, such as alcohol drinking, physical activity, and diet, could bias the associations. Further studies are warranted to examine the associations of GWG with maternal and infant outcomes independent of other health risk behaviors.

Conclusions
This cohort study of more than 15.8 million mother-infant dyads found that the extremes of GWG were associated with increased risks of adverse infant outcomes across BMI categories, suggesting that weight maintenance and weight loss should not be used as routine guidelines, even for women with obesity.