Gestational Diabetes and Subsequent Metabolic Dysfunction: An National Health and Nutrition Examination Survey Analysis (2011–2018)

Introduction

Gestational diabetes mellitus (GDM) is defined as glucose intolerance of varying degree first identified during pregnancy. ^1,2 GDM is one of the most common pregnancy complications with prevalence estimates varying but increasing in the United States. ^1,2 National birth certificate data suggest that the prevalence of GDM increased by 78% from 4.6% in 2006 to 8.2% in 2016. ^{3 –5} Meanwhile, Pregnancy Risk Assessment Monitoring System data suggest that the rate of GDM increased by 30% from 2016 (6%) to 2020 (7.8%). ⁵

GDM prevalence is consistently higher among racialized minority populations, with Asian and Pacific Islander populations having the highest rates compared with other racial/ethnic groups. ^{3 –5} In 2010, Asian/Pacific Islanders (16.3%) and Hispanics (12.1%) had the highest rates of GDM. ⁵ Data from 2020 suggest that the highest rates of GDM were in non-Hispanic Native Hawaiian or Pacific Islander women (10.6%), non-Hispanic American Indian/Alaska Native women (11.8%), and non-Hispanic Asian women (14.9%). ⁶ In addition, compared with non-Hispanic White women, Asian women have approximately two times the odds of having GDM. ⁷ As GDM rates continue to rise in the United States, research to further understand risk factors for this increasingly common condition, especially in high-risk populations, is warranted to inform future prevention efforts.

GDM is associated with an increased risk for subsequent metabolic dysfunction, and women previously diagnosed with GDM have a twofold to threefold greater risk for developing metabolic syndrome (MetS). ^{7 –10} MetS is a cluster of four metabolic dysregulations, including insulin resistance, atherogenic dyslipidemia, central obesity, and hypertension. ¹¹ Identification of MetS is based on clinical measurements of waist circumference, triglycerides, high-density lipoprotein (HDL) cholesterol, blood pressure, and fasting glucose. ¹² These measures align with the increased risks of central obesity, high blood pressure, increased triglycerides, and insulin resistance associated with GDM. ⁸

The cluster of metabolic dysregulations defined by MetS increases the risk for future chronic health conditions, including cardiovascular disease (CVD), diabetes, and cancer. ^{13 –15} Individuals with MetS have approximately twofold increased risk of developing CVD and a fivefold increased risk of developing diabetes. ^13,14 In addition, MetS is associated with an increased risk for some cancers, including endometrial, pancreatic, breast, rectal, and colorectal cancers. ¹⁵ The prevalence of MetS in the United States ranges between 36.2% and 38.3%. ¹⁶ In 2016, the prevalence of MetS was highest among Hispanic (40.4%) followed by White (37.6%), Black (30%), and Asian (26.1%). ¹⁶ Between 2011 and 2018, rates remained stable among all racial groups except for Asian people, which increased from 21.8% in 2011 to 31.2% in 2018. ¹⁷

Although the link between GDM and MetS has been well researched, there are limited data regarding this association among Asian women, among whom GDM risk is the highest. The common risk factors for GDM and MetS present differently in Asian populations compared to other racial/ethnic groups. For example, Asian women are at an increased risk for GDM at lower body mass index (BMI), which may be linked to underlying genetic risk factors in Asian populations. ¹⁸ Specifically, gene loci have been found to be associated with adiposity and glucose metabolism in a South Asian population. ¹⁹ Glucose metabolism is an important risk factor for MetS and may contribute to Asian women being more susceptible to MetS. Given the current lack of data regarding GDM and MetS specific to Asian women, and a risk profile for GDM that appears to be unique to Asian women, a greater focus on postpartum metabolic dysregulation among Asian women is warranted.

The objective of this study is to better understand racial/ethnic differences in the association between GDM and MetS, especially among Asian women. Based on the high prevalence of GDM in Asian women and prior research on the link between GDM and MetS, we hypothesize that Asian women with a prior diagnosis of GDM during pregnancy have an increased odds of subsequent metabolic dysregulations, and this association is stronger compared to women with a prior diagnosis of GDM of other racial/ethnic backgrounds. ^6,17

Materials and Methods

Data were drawn from the National Health and Nutrition Examination Survey (NHANES) 2011–2018, a repeated U.S. cross-sectional survey since 1999. ²⁰ NHANES data were collected from a nationally representative, noninstitutionalized sample of ∼5000 individuals every year, and are available in 2-year cycles. A total of 39,156 individuals were available in the dataset during the study period. Included in the analysis were 8355 women 20 years of age and older who reported ever been pregnant. We excluded those who were currently pregnant (n = 204), those with a diagnosis of diabetes before GDM diagnosis (n = 38), and those who had missing values for the cardiometabolic variables (n = 918), leading to a final analytic sample of 7195 women.

Metabolic dysfunction was identified using metabolic characteristics informed by four clinical measurements of MetS, including systolic blood pressure (SBP), waist circumference, HDL cholesterol, and glycosylated hemoglobin (HbA1c). ²¹ Clinically significant cutoff points were based on pre-existing criterion of the National Cholesterol Education Program Adult Treatment Panel III for MetS, which include a waist circumference ≥88 cm in women, HDL cholesterol <40 mg/dL, and SBP ≥130 mmHg. ²² In addition, glucose metabolism was defined using glycohemoglobin with a clinically significant cutoff point of ≥6.5%. ²³ All measures were collected in mobile examination centers by trained medical personnel. ²⁰ SBP was defined using the average of four measurements collected during mobile examination. Each variable is dichotomous indicating either dysfunction or nondysfunction. For descriptive purposes we summed the 4 dichotomous metabolic dysfunction variables to create a metabolic dysfunction index (range 0–4).

A previous diagnosis of GDM was self-reported by respondents during interviews at the mobile examination center with trained interviewers. ²⁰ Women 20 years of age and older were asked “During pregnancy, were you ever told by a doctor or other health professional that you had diabetes, sugar diabetes, or gestational diabetes?.” Respondents who answered “yes” were identified as having prior GDM. To retain sample size, those that answered “no,” “borderline,” “refused,” or “don't know” were identified as not having prior GDM.

Covariates included age in years (20–30, 31–40, 41–50, 51–60, 61+), race [non-Hispanic White (White), non-Hispanic Black (Black), Mexican American, Non-Hispanic Asian (Asian), Other], poverty income ratio (PIR) (<1, 1–2.9, 3–4.9, >5), marital status (married/living with partner, other), education (high school or lower, some college or higher), BMI (continuous), physical activity level (participates in moderate to vigorous physical activity/does not participate in moderate to vigorous physical activity), alcohol consumption of at least 12 drinks per year (yes, no, don't know), number of days smoking cigarettes in past 30 days (continuous), family history of heart attack (yes/no), family history of diabetes (yes/no), current diabetes status (yes/no), and current hypertension status (yes/no). These potential confounders were identified through previous literature. ^{2,7,23 –25}

Weighted mean ± 95% confidence interval (CI) was calculated for the cardiometabolic variables by key demographic characteristics. The weighted frequencies and percentages were obtained for the key demographic characteristics. A series of logistic regression models were fit to estimate odds ratios (ORs) and 95% CI for the relationship between prior GDM status and each of the four metabolic characteristics. An additional model was fitted for the relationship between prior GDM and current diabetes status. Model 1 included GDM and NHANES survey cycle. Model 2 added race, age, income, marital status, and education. Model 3 added activity level, alcohol consumption, smoking, family history of heart attack, family history of diabetes, current diabetes, and current hypertension. Model 3 for HbA1c excluded diabetes status. Models were run overall and by race category. Sample weights were used in all models to account for oversampling, noncoverage, and nonresponse to provide representative estimates for the U.S. population.

We conducted sensitivity analyses in which we excluded women that reported “refused” or “don't know” to having prior GDM and fit fully adjusted regression models overall and by race/ethnicity. In addition, we conducted a sensitivity analysis regarding waist circumference among Asian women, identifying those with waist circumference ≥80 cm as having “high waist circumference,” as literature suggests 80 cm may be a population-specific measure for Asian women. ²⁶ All analyses were performed using SAS OnDemand for Academics. This project was considered exempt by the University of North Dakota Institutional Review Board.

Results

Mean cardiometabolic variables, overall and by covariates, are included in Table 1. Overall, 8.2% of women reported being previously diagnosed with GDM. The sample was 37.9% White, 22.8% Black, 14.6% Mexican American, 10.0% Asian, and 14.7% Other racial identities. A majority of women were ≥61 years of age or older (35.2%) followed by age categories of 51–60 years (19.7%), 41–50 years (18.9%), 31–40 years (16.0%), and 20–30 years (10.2%). In addition, the sample included 20.6% of women in the lowest family PIR category (0–0.99) with the largest percentage of women within the 1–2.99 PIR (39.3%). Women with prior GDM had a higher mean waist circumference and glycohemoglobin and lower HDL cholesterol and SBP compared to women without prior GDM. Black women had the highest mean SBP and waist circumference (104.0 cm) compared to the other racial groups. In addition, Asian women had the smallest mean waist circumference (86.9 cm) and lowest average metabolic index (1.32).

Logistic regression results for the association between prior GDM and cardiometabolic outcomes are included in Table 2. Overall, women with a previous diagnosis of GDM had nonsignificant increased odds (OR: 1.45; 95% CI: 0.9–2.3) of having a high waist circumference compared to women without prior GDM (Table 2). No statistically significant observations for the association between prior GDM and high waist circumference were observed in race-specific analyses.

There was no overall association between SBP and history of GDM (OR: 0.77; 95% CI: 0.6–1.1). However, in race-specific analyses, among Mexican American women, those with prior GDM had 98% significantly increased odds (OR: 1.98; 95% CI: 1.1–3.6) of a high SBP compared to women without prior GDM. Among Asian (OR: 0.62) and White (OR: 0.57) women, there were nonsignificant decreased odds of high SBP among those with prior GDM.

Overall, there was a nonsignificant increased odds of low HDL cholesterol (OR: 1.25; 95% CI: 0.95–1.6) among those with prior GDM. In race-specific analyses, Black women with prior GDM had 65% significant increased odds (OR: 1.65; 95% CI: 1.0–2.6) of low HDL cholesterol compared to Black women without prior GDM. In addition, there were nonsignificant increased odds of low HDL cholesterol among Asian (OR: 1.21), Mexican American (OR: 1.38), and White (OR: 1.07) women with prior GDM.

We observed an association between prior GDM and high levels of glycohemoglobin in the overall sample (OR: 2.9; 95% CI: 2.0–4.2). In race-specific analyses, there were greater odds of an increased glycohemoglobin among all racial groups. The odds of having increased glycohemoglobin were 620% significantly greater in Asian women with prior GDM (OR: 7.2; 95% CI: 2.0–18.4) compared to Asian women without prior GDM. For women with prior GDM, the odds of an increased glycohemoglobin were 208% significantly greater (OR: 3.08; 95% CI: 1.7–5.6) among Black women, 186% significantly greater (OR: 2.86; 95% CI: 1.5–5.6) among Mexican American women, and 151% significantly greater (OR: 2.51; 95% CI: 1.4–4.7) among White women compared to women without prior GDM. Similar results were found for the relationship between current diabetes status and prior GDM (Table 3).

Results of the sensitivity analysis (Table 4) in which those that responded “refused” or “don't know” to having prior GDM that were excluded from analysis were similar to the main analysis in Table 2 (results not shown). Results of the sensitivity analysis regarding waist circumference among Asian women using 80 cm as the cut-point (OR: 1.96; 95% CI: 0.77–5.04) were similar to the results using 88 cm as the cut-point (OR: 1.60; 95% CI: 0.60–4.26), in that we observed no statistically significant association between GDM and high waist circumference.

Discussion

This study aimed to investigate racial disparities in the association between GDM and metabolic dysfunction. Overall, we observed that prior GDM increased risk of high HbA1c for all racial/ethnic groups. Race-specific increases in risk were observed for SBP among Mexican American women and HDL cholesterol among Black women. These findings partially supported the hypothesis that Asian women with prior GDM have increased odds of metabolic dysregulation, and this elevated risk is more substantial compared to women of other racial/ethnic groups. Metabolic dysfunction among Asian women was primarily seen through dysregulation of glucose metabolism, as measured through glycohemoglobin.

Our findings align with previous literature. A sample of 1968 women from China were followed for an average of 3.5 years; prior GDM was associated with 164% increased odds of central obesity, 314% increased odds of hypertriglyceridemia, 260% increased odds of high blood pressure, and 62% increased odds of hyperglycemia. ⁸ Another study from Denmark investigated the association between GDM and metabolic outcomes in a sample of 198 women. ⁹ The Denmark study found that women with prior GDM had increased risk of high blood pressure (51.6% vs. 25.7%), raised triglycerides (25% vs. 5.7%), impaired glucose metabolism (64.8% vs. 31.4%), and reduced HDL cholesterol (50% vs. 24.3%) compared to women without prior GDM. ⁹ Finally, a study from Finland with a total sample of 240 participants found that women with prior GDM had higher rates of increased waist circumference (74.2% vs. 60.8%) and increased fasting glucose (15% vs. 3.3%). ¹⁰ Overall, these studies suggest that central obesity and glucose metabolism are significantly associated with prior GDM.

Our observations contribute to the growing body of evidence of race-specific determinants of GDM and race-specific consequences of GDM postpregnancy. ^{27 –29} It is important to note that specific to Asian women, these race-specific findings regarding GDM are independent of BMI, as Asian women may develop GDM at lower BMI than women of other racial/ethnic groups. ¹⁸ Given the evidence which suggests that Asian women have higher risks of GDM and potentially higher risks of postpartum metabolic dysfunction than other racial/ethnic groups, determinants of GDM specific to Asian populations should be more fully investigated to inform prevention efforts.

Both genetic and socioeconomic factors may be linked to metabolic dysfunction among Asian women. Specifically, the increased odds of hyperglycemia in this group may be associated with underlying genetic factors that impact pancreatic function and insulin sensitivity. One study using a genome-wide association meta-analysis evaluated the genetic effects on type 2 diabetes. ¹⁹ This study found that there were genes linked with adiposity and glucose metabolism in a South Asian population. ¹⁹ An additional study indicated that there are genetic differences between Asian and White populations when looking at the etiology of type 2 diabetes. ³⁰ Specifically, this study noted that Asian populations have less β cell function and that genetic variants in Asian populations are involved with insulin secretion, β cell development, and β cell growth. ³⁰

Metabolic function may also be linked to lack of preventive care among Asian populations in the United States, which may be due to financial, physical, communication, and cultural factors. ³¹ Financial barriers, such as high out-of-pocket cost, may influence an individual's decision to seek care. In addition, Asian in the United States women may experience language and health literacy barriers. ³¹ Individuals who are less familiar with the English language may have greater difficulty expressing their needs and understanding physician recommendations due to technical terminology. Finally, Asian communities in the United States have different cultural norms. The use of health care may be for curative purposes when people find themselves ill rather than preventive care. ³¹ In addition, increased acculturation has been found to be a protective factor for GDM in Asian populations. ¹ Higher acculturation may contribute to diet or other lifestyle changes that may impact an individual's risk for developing GDM. ¹

The findings from this study can help guide health promotion efforts targeting patients with prior GDM. There are several factors that impact metabolic outcomes. Understanding the factors that are most significant for each racial/ethnic group may provide guidance on important preventive care strategies. Strategies to improve access to diabetes prevention resources through language-specific and culturally relevant community services have improved diabetes-related knowledge among South Asian communities in Canada. ³² Community-based approaches that combined policy, systems, and environmental approaches have been successful in addressing key factors that contribute to poor cardiometabolic health among racially minoritized communities. ³³ Implementation of improved preventive care and screenings may help reduce chronic conditions and help to prevent poor health outcomes in the future. Finally, further investigation regarding modifying factors or pathway analyses for the relationship between GDM and metabolic dysfunction is warranted, such as the role of birth weight and lactation. ^34,35

There are some limitations to consider in this study. First, the NHANES data do not have information regarding prepregnancy BMI, gestational age, parity, or weight gain in pregnancy, which have been identified in previous literature as potential confounders. Second, we were unable to evaluate the specific MetS index due to missing data. More specifically, to identify MetS, measures for triglycerides and fasting glucose are required. However, as only a subset of NHANES participants complete the mobile examination center module during which blood draw occurs, these variables had a high number of missing values and could not be used for analysis. In addition, the missing values for key variables resulted in a smaller sample size for analysis. Due to the smaller sample size, we were unable to include the years after GDM diagnosis in the analysis. However, we used sample weights specific for the mobile examination center sample; thus, our observations are representative.

Third, this was a cross-sectional study which does not allow us to assess temporality. Therefore, we cannot be certain that the metabolic dysfunctions were not present before a GDM diagnosis. Finally, as GDM was assessed retrospectively, recall bias may limit accuracy of the exposure. However, multiple studies suggest that recall regarding events in the perinatal period is accurate, even up to 20 years postdelivery. ^{36 –38}

Regarding strengths, the results from this study are generalizable to women in the United States due to the representativeness of NHANES data. In addition, glycohemoglobin is less susceptible to recent dietary changes or illness than fasting glucose, making glycohemoglobin a more reliable long-term assessment of glucose metabolism. ³⁹ To further investigate the association of GDM with diabetes related outcomes, this study also analyzed future diabetes status as an outcome, which indicated similar results as glycohemoglobin. Finally, the NHANES data allow analysis of racial disparities present in the GDM and MetS relationship, specifically in the Asian population. Therefore, the study adds to current understanding about the association between GDM and MetS and identifies racial disparities present in this relationship. These findings can help inform future guidelines for clinical screenings for MetS.

Conclusions

The purpose of this study was to investigate the association between GDM and metabolic dysfunctions and how these associations differ by racial/ethnic category. Specifically, we found that prior GDM was consistently associated with higher levels of HbA1c across racial/ethnic groups. Yet, the association between prior GDM and HbA1c is more pronounced among Asian women. Understanding the differences between populations has important clinical implications. These findings may suggest the need for improved screening guidelines and preventive strategies to address the increased odds of poor metabolic outcomes, which have been linked to chronic health conditions.