The role of social determinants of health in post-diagnosis gestational weight gain trajectories among women with gestational diabetes mellitus: a multicenter cohort using latent class linear mixed models

Abstract

Background

Excessive gestational weight gain (GWG) is common among women with gestational diabetes mellitus (GDM) and is associated with adverse maternal and neonatal outcomes. However, the influence of social determinants of health (SDOH) on GWG trajectories in this population remains poorly understood. This study aimed to characterize GWG trajectories and evaluate how SDOH influence these patterns.

Methods

We conducted a multicenter cohort study of women with GDM who delivered at four hospitals in Wuhan, China. GWG trajectories were based on at least three weight measurements from GDM diagnosis to delivery, with gain defined as current weight minus pre-pregnancy or early-pregnancy weight measured before 8 weeks, and were modeled using latent class linear mixed models. SDOH was assessed using 19 indicators across five dimensions, summarized into a composite index, and further divided into three tertiles. Associations between SDOH classes and GWG trajectories were examined with multinomial logistic regression adjusted for maternal age, pre-pregnancy body mass index (BMI), education, employment, and parity. Subgroup analyses were conducted according to pre-pregnancy BMI (< 24.0 vs. ≥ 24.0 kg/m²).

Results

Among 1,916 women, three GWG trajectories were identified: low weight gain (33.40%), moderate weight gain (50.42%), and high weight gain (16.28%). SDOH were categorized into three tertile-based classes: Class 1 (most advantaged), Class 2 (intermediate), and Class 3 (most disadvantaged). In the fully adjusted multinomial logistic regression model G, compared with women in Class 1, those in Class 3 had higher odds of belonging to the high weight gain trajectory rather than the moderate trajectory (aOR = 2.18, 95% CI 1.49–3.20). In subgroup analyses, among women with pre-pregnancy BMI < 24.0 kg/m², both Class 2 and Class 3 were associated with higher odds of the high weight gain trajectory (Class 2: aOR = 1.56, 95% CI 1.03–2.37; Class 3: aOR = 3.16, 95% CI 1.99–5.01), whereas no significant association with the high weight gain trajectory was observed among women with pre-pregnancy BMI ≥ 24.0 kg/m².

Conclusions

Among women with GDM, GWG follows three distinct trajectories. Greater social disadvantage is associated with a higher likelihood of high GWG. These findings support the early identification of socially disadvantaged women and the development of BMI-specific, socially informed prenatal weight management strategies.

Trial registration

ChiCTR2200063189, registered 01 September 2022.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27471-5.

Background

Gestational diabetes mellitus (GDM) affects an estimated 14% of pregnancies worldwide [1, 2] and approximately 15% of pregnancies in mainland China, making it one of the most common obstetric complications [3]. Clinical management of GDM primarily focuses on glycemic control through diet, physical activity, glucose monitoring, and pharmacotherapy when needed [4]. However, gestational weight gain (GWG), which refers to the changes in a pregnant woman’s weight from before pregnancy or during the early stages of pregnancy to the late stages, is a crucial and closely related modifiable factor [5]. Nearly half of pregnant women gain weight above Institute of Medicine (IOM) recommendations, and such excessive GWG is consistently associated with higher risks of large-for-gestational-age birth, macrosomia, cesarean delivery, hypertensive disorders, and neonatal complications, including NICU admission [6]. Therefore, weight management is an essential complement to glycemic management in women with GDM [7].

Social Determinants of Health (SDOH) are non-medical factors that influence health outcomes and these factors include socioeconomic status, education, race or ethnicity, healthcare access, and neighborhood conditions [8, 9]. Addressing differences in SDOH accelerates progress toward health equity, a state in which every person especially pregnant people has the opportunity to attain their highest level of health [10]. While pre-pregnancy body mass index (BMI) and maternal age are well-established predictors of GDM and GWG, a growing body of evidence demonstrates that broader SDOH also critically shape GDM risk and weight gain trajectories during pregnancy [11]. Therefore, understanding how SDOH influence GWG patterns in women with GDM is essential for developing targeted interventions to mitigate risks and improve maternal health.

However, most national GWG curves are derived from perinatal databases that are rich in clinical data but lack upstream social determinants like neighborhood disadvantage or healthcare access [12]. This limitation persists despite guidance from the National Academy of Medicine and the National Academies, which calls for explicitly modeling structural SDOH, such as education, income, and access to care, rather than merely adjusting for them as simple covariates [13, 14]. Treating these complex factors as mere variables risks overattributing GWG differences to ethnicity, thereby masking the true culturally and structurally driven patterns. Consequently, even the newest GWG curves remain stratified only by BMI, and few studies adopt a comprehensive, life-course approach that captures the cumulative and synergistic impact of multiple SDOH domains on GWG trajectories.

The purpose of this study was to address this research gap by comprehensively assessing the impact of a composite SDOH score on GWG trajectories in women with GDM. The study first sought to identify distinct GWG trajectories and SDOH score in this population. We then quantified the associations between the composite SDOH score and these trajectories, and further examined how economic and community deprivation domains contributed to or modified the relationship between the composite SDOH score and GWG trajectories.

Materials and methods

Study design and data source

This multicenter retrospective cohort study utilized data from four hospital-based perinatal registry, including women diagnosed with GDM and their neonates who underwent prenatal care and delivery at four clinical centers in Wuhan, China, between January 2022 and December 2023. A complete list of participating investigators is provided in Supplementary File S1. The study design and reporting adhered to the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guidelines for cohort studies and provided in the Supplementary File S2 [15].

Participants

Eligible participants were women diagnosed with GDM between 24 and 28 weeks of gestation according to the International Association of Diabetes and Pregnancy Study Groups (IADPSG) one-step 75 g Oral Glucose Tolerance Test (OGTT), meeting any of the following thresholds: fasting plasma glucose ≥ 5.1 mmol/L, 1-hour ≥ 10.0 mmol/L, or 2-hour ≥ 8.5 mmol/L [16, 17]. Additional inclusion criteria were singleton live birth and availability of both pre-pregnancy BMI and at least three gestational weight measurements. Exclusion criteria included pre-existing type 1 or type 2 diabetes, multiple gestations, major fetal malformations, missing pre-pregnancy BMI, fewer than two gestational weight measurements, and missing SDOH data due to non-completion of the questionnaire or non-participation in the SDOH telephone follow-up. Following the diagnosis of GDM, all women received standardized routine care. This included dietary counseling, physical activity guidance, blood glucose monitoring, and repeated weight monitoring during antenatal follow-up. When glycemic control was inadequate, insulin therapy was initiated when clinically indicated.

Exposure assessment of social determinants of health

SDOH were assessed using 19 binary indicators across five key domains: Financial Circumstances, Education Access and Quality, Health Care Access and Quality, Neighborhood and Built Environment, and Social and Community Context [18]. These indicators were collected using structured questionnaires administered face-to-face during the delivery hospitalization or, when in-person contact was not feasible, by telephone within 1 month postpartum. This time frame was chosen because core socioeconomic conditions were unlikely to change over this short interval, while information on family and social support throughout pregnancy could be more comprehensively captured after delivery. Each indicator was coded as 1 (favorable) or 0 (unfavorable) according to its relevance to maternal and metabolic health and its alignment with the goals of Healthy China 2030. A multidomain SDOH index was constructed from these 19 baseline indicators, with domains and scoring criteria summarized in Supplementary Table S1. A composite SDOH index was then derived using factor analysis, as described in the "Statistical analysis" section.

Definition of gestational weight gain trajectories

The primary outcome was gestational weight gain (GWG) trajectory from GDM diagnosis to delivery. GWG was calculated from at least three antenatal maternal weight measurements and defined as current weight minus pre-pregnancy weight or early-pregnancy weight measured before 8 gestational weeks. By incorporating repeated measurements over time, this longitudinal approach captures both the direction and rate of weight gain and is less sensitive to short-term fluctuations than conventional cross-sectional GWG categories. GWG trajectories were identified using latent class linear mixed models, as detailed in the "Statistical analysis" section [19].

Statistical analysis

All statistical analyses were conducted using R software (version 4.5.1). Statistical significance was defined as a two-sided P < 0.05. Continuous variables were presented as mean (SD), and categorical variables as n (%). Baseline characteristics were compared across the three SDOH classes using one-way analysis of variance (ANOVA) for continuous variables and the chi-square test for categorical variables.

To derive a single quantitative measure of social conditions, a composite SDOH index was constructed from 19 binary baseline indicators using factor analysis (FA). The suitability of the data for FA was first assessed using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity. The overall KMO value was 0.626, indicating acceptable sampling adequacy, and Bartlett’s test was significant (χ² = 2259.28, P < 0.001), indicating that the correlation matrix was suitable for factor analysis [20]. Principal axis factoring with varimax rotation was then performed. Factor retention was based on eigenvalues greater than 1, inspection of the scree plot, and interpretability of the rotated factor structure. Five factors were retained in the final solution, with eigenvalues of 2.19, 1.51, 1.42, 1.19, and 1.14, respectively, jointly accounting for 39.23% of the total variance. Based on the rotated factor loading matrix, these five factors broadly reflected socioeconomic position, health-related behaviors, dietary patterns, healthcare and social support, and the physical living environment. Factor scores were estimated using the regression method, and the composite SDOH index was calculated as the sum of standardized factor scores, with higher values indicating more favorable social conditions. In the present dataset, the composite index ranged from − 9.71 to 2.37. Participants were subsequently categorized into three groups according to tertiles of the composite SDOH index, with the highest tertile defined as Class 1, the middle tertile as Class 2, and the lowest tertile as Class 3.

GWG trajectories were identified using latent class linear mixed models (LCLMMs) based on repeated antenatal weight measurements obtained from GDM diagnosis to delivery. At each visit, gestational weight gain was defined as current weight minus pre-pregnancy weight or early-pregnancy weight measured before 8 gestational weeks. Gestational week and its quadratic term were specified as fixed effects to capture potential non-linear patterns of weight gain, while random intercepts and random slopes were included to account for within-subject correlation. Models with two to four latent classes were fitted sequentially using maximum likelihood estimation. Model selection was based on BIC, entropy, class size (≥ 5% of the sample), and clinical interpretability [21]. The three-class model was retained as the final solution because it provided the best balance between parsimony and clinical interpretability, identifying low, moderate, and high GWG trajectories.

The primary exposure in the regression analyses was SDOH class, and the primary outcome was GWG trajectory class. Associations between SDOH classes and GWG trajectories were examined using multinomial logistic regression, with the moderate GWG trajectory (Trajectory 2) as the reference outcome. Models were fitted sequentially: Model E was unadjusted; Model F adjusted for maternal age; and Model G further adjusted for pre-pregnancy BMI and parity. Because education and employment were included as components of the SDOH indicators, they were not additionally adjusted for in the main models to avoid potential overadjustment. Given the limited sample size in some pre-pregnancy BMI categories, participants were stratified by pre-pregnancy BMI according to Chinese criteria (< 24 vs. ≥24 kg/m²), and stratified analyses were conducted to explore potential effect modification by baseline weight status.

Results

Demographics and baseline characteristics

A total of 1 916 pregnant women with GDM were included in the analytic sample after applying eligibility criteria (Fig. 1). The mean maternal age was 32.65 ± 3.98 years, and 30.01% were aged ≥ 35years. The mean pre-pregnancy BMI was 22.75 ± 3.54 kg/m²; 7.20% of women were underweight (BMI < 18.5 kg/m²), 62.53% had normal weight (18.5–23.9 kg/m²), 22.34% were overweight (24.0–27.9 kg/m²), and 7.93% were obese (BMI ≥ 28 kg/m²). Most women were nulliparous (71.92%), while 28.08% were multiparous. Overall, 78.08% had a university degree or above and 58.82% were employed during pregnancy. The overall mean SDOH score was 13.73 ± 1.75, and participants were distributed almost equally across the three tertile-based SDOH groups: Class 1 (most advantaged; mean score 14.87 ± 1.09), Class 2 (moderate; mean score 13.96 ± 1.37), and Class 3 (most disadvantaged; mean score 12.34 ± 1.69) (Table 1). The baseline characteristics of women included in the final analytic sample and those excluded were compared to assess possible selection bias (Supplementary Table S4). Compared with excluded women, included women were generally older, more likely to be aged ≥ 35 years, better educated, and differed in parity and pre-pregnancy BMI distribution. Complete SDOH data were unavailable for excluded women.

Fig. 1.

Flowchart of subject selection. Abbreviations: GDM, gestational diabetes mellitus; GWG, gestational weight gain; BMI, body mass index

Table 1.

Basic characteristics of the study population by the trajectory of gestational weight gain among women with GDM

Characteristic	Overall (n = 1,916)	Trajectory 1 (640, 33.40%)	Trajectory 2 (966, 50.42%)	Trajectory 3 (310, 16.18%)	P Value
Physical characteristics
Maternal age, years, mean (SD)	32.65 (3.90)	32.09 (3.90)	32.69 (3.99)	32.02 (4.04)	0.006
Age ≥ 35, n (%)	575 (30.01)	204 (31.88)	282 (29.19)	89 (28.71)	0.446
Educational level, n (%)
College or higher	1496 (78.08)	510 (79.69)	758 (78.47)	228 (73.55)	0.092
Employment, n (%)
Employed	1127 (58.82)	402 (62.81)	563 (58.28)	162 (52.26)	0.007
Pre-pregnancy BMI, mean (SD)	22.75 (3.54)	23.62 (4.00)	22.37 (3.29)	22.15 (2.86)	< 0.001
Pre-pregnancy BMI group, n (%)					< 0.001
< 18.5	138 (7.20)	36 (5.63)	83 (8.59)	19 (6.13)
18.5—23.9	1198 (62.53)	351 (54.84)	632 (65.42)	215 (69.35)
24.0—27.9	428 (22.34)	171 (26.72)	191 (19.78)	66 (21.29)
≥ 28	152 (7.93)	82 (12.81)	60 (6.21)	10 (3.23)
Parity, n (%)					0.775
Nulliparous	1378 (71.92)	454 (70.94)	698 (72.26)	226 (72.90)
Multiparous	538 (28.08)	186 (29.06)	268 (27.74)	84 (27.10)
SDOH score, mean (SD)	13.73 (1.75)	13.68 (1.74)	13.77 (1.69)	13.68 (1.96)	0.530
SDOH class, n (%)					< 0.001
SDOH class1	639 (33.40)	243 (37.97)	323 (33.44)	73 (23.55)
SDOH class2	638 (33.30)	202 (31.56)	343 (35.51)	93 (30.00)
SDOH class3	639 (33.40)	195 (30.47)	300 (31.06)	144 (46.45)

Data are presented as mean (SD) or n (%), as appropriate

GWG Gestational weight gain, GDM Gestational diabetes mellitus, BMI Body mass index, SDOH Social determinants of health, SD Standard deviation

Trajectory 1, low weight gain trajectory; Trajectory 2, moderate weight gain trajectory; Trajectory 3, high weight gain trajectory

P values are for comparisons across gestational weight gain trajectories (ANOVA or χ² test, as appropriate)

Bold P values indicate statistical significance (P < 0.05)

Gestational weight gain trajectories

Three distinct GWG trajectories were identified (Fig. 2A). The model demonstrated good overall fit and excellent classification accuracy, with BIC = 49,530.2 and an average posterior probability of correct classification (OCC) = 0.9544. Participants were distributed as follows: 33.40% in the low weight gain trajectory, 50.42% in the moderate weight gain trajectory, and 16.18% in the high weight gain trajectory. All three trajectories showed an overall upward trend across gestation, with different rates of weight gain: the low weight gain trajectory exhibited a gradual increase from approximately 24 to 40 weeks, the moderate weight gain trajectory showed a steady intermediate rise, and the high weight gain trajectory had the steepest slope. Between-group comparisons of baseline characteristics were presented in Table 1. Maternal age differed significantly across trajectories (p = 0.006), with women in the high weight gain trajectory being younger on average. Pre-pregnancy BMI differed significantly among groups (p < 0.001), being highest in the low weight gain trajectory and lowest in the high weight gain trajectory. Employment status differed significantly across trajectories (p = 0.007), with fewer employed women in the high weight gain trajectory. In contrast, educational level (p = 0.092) and parity (p = 0.775) did not differ significantly among the three trajectories.

Fig. 2.

Gestational weight gain trajectories and social determinants of health in women with gestational diabetes mellitus. A Latent class linear mixed model–derived trajectories of gestational weight gain (GWG) from gestational diabetes mellitus (GDM) diagnosis to delivery, with predicted mean curves (solid lines) and 95% confidence bands (shaded areas) for Trajectory 1 (low–weight gain trajectory), Trajectory 2 (moderate–weight gain trajectory), and Trajectory 3 (high–weight gain trajectory). B Violin plots of composite social determinants of health (SDOH) score by GWG trajectory (ANOVA: F(2,1913) = 0.63; p = 0.530). C Proportions of women in each SDOH latent class within GWG trajectories (χ²(4) = 35.11; p < 0.001). Abbreviations: Class 1: most advantaged SDOH; Class 2: intermediate; Class 3: most disadvantaged. SDOH, social determinants of health; GWG, gestational weight gain; GDM, gestational diabetes mellitus; OR, odds ratio; CI, confidence interval

We compared solutions with 2–4 trajectory classes (Supplementary Table S2). The 2-class model yielded the lowest AIC/BIC (AIC = 40,177; BIC = 40,238) but showed poor class separation (entropy = 0.523; OCC = 0.874) and collapsed clinically distinct patterns. The 4-class model modestly improved AIC/BIC relative to the 3-class solution (AIC = 41,486; BIC = 41,591) but produced an unstable class without additional clinical meaning. The 3-class solution provided the best balance of fit and classification quality (entropy = 0.902; OCC = 0.958), with all classes ≥ 5% and clear interpretability; therefore, we retained K = 3 as the final model.

Characteristics of tertile-based SDOH groups

Across the study cohort, the composite SDOH score ranged from 5 to 18, with a mean of 13.73 ± 1.75. The score distribution was slightly skewed toward higher values, suggesting relatively favorable social conditions among included participants. Participants were subsequently categorized into three groups according to tertiles of the composite SDOH score. Class 1 represented the most advantaged profile, characterized by higher education levels, higher employment rates, and overall higher SDOH scores (mean 14.87 ± 1.09, n = 639, 33.40%). Class 2 reflected an intermediate socioeconomic position with moderate access to resources and social support (mean 13.96 ± 1.37, n = 638, 33.30%). Class 3 comprised the most disadvantaged group, typified by lower educational attainment, lower employment rates, and reduced access to supportive social environments (mean 12.34 ± 1.69, n = 639, 33.40%) (Table 1).

Association between SDOH class and gestational weight gain trajectories

Between-group comparisons of SDOH measures across the three GWG trajectories are summarized in Fig. 2B and C. The mean continuous SDOH score did not differ significantly among the trajectories (F (2,1913) = 0.63, p = 0.530). In contrast, the distribution of tertile-based SDOH classes varied markedly across trajectories (χ²(4) = 35.11, p < 0.001), with women in the most disadvantaged SDOH group more frequently belonging to Trajectory 3. Univariate comparisons of baseline characteristics showed that maternal pre-pregnancy BMI (p < 0.001), age (p = 0.006), and employment status (p = 0.007) differed significantly across trajectories, whereas educational level (p = 0.092) and parity (p = 0.775) did not. These significant variables were subsequently included in the multivariable multinomial logistic regression models.

Multinomial logistic regression models (Models E–G) were used to examine the associations between SDOH classes and GWG trajectories, with the moderate GWG trajectory (Trajectory 2) as the reference outcome (Table 2). In the unadjusted Model E, compared with women in the most advantaged SDOH Class 1, those in SDOH Class 2 had lower odds of belonging to the low GWG trajectory rather than the moderate trajectory (Trajectory 1 vs. 2; OR = 0.78, 95% CI 0.62–1.00, p = 0.046), whereas no significant association was observed for SDOH Class 3 (OR = 0.86, 95% CI 0.68–1.10, p = 0.243). For the high GWG trajectory, women in the most disadvantaged SDOH Class 3 had significantly higher odds of belonging to Trajectory 3 rather than Trajectory 2 (OR = 2.12, 95% CI 1.54–2.93, p < 0.001), while the association for SDOH Class 2 was not statistically significant (OR = 1.20, 95% CI 0.85–1.69, p = 0.297). After adjustment for maternal age in Model F, the results were materially unchanged. In the fully adjusted Model G, the association between SDOH Class 3 and the high GWG trajectory remained robust (Trajectory 3 vs. 2; aOR = 2.18, 95% CI 1.49–3.20, p < 0.001), whereas the associations for SDOH Class 2 and for the low GWG trajectory were not statistically significant. Overall, these findings suggest that greater social disadvantage was associated with higher odds of belonging to the high GWG trajectory relative to the moderate trajectory, whereas associations with the low GWG trajectory were weaker and attenuated after full adjustment. The original multivariable multinomial logistic regression results with Trajectory 1 as the reference outcome are presented in Supplementary Table S5.

Table 2.

Multivariable multinomial logistic regression of gestational weight gain trajectories among women with gestational diabetes mellitus

Comparison	Variable	Model E OR (95%CI)	P	Model F OR (95%CI)	P	Model G OR (95%CI)	P
Trajectory1 VS 2	SDOH class 1	Reference		Reference		Reference
	SDOH class 2	0.78 (0.62-1.00)	0.046	0.78 (0.61–0.99)	0.042	0.78 (0.60–1.01)	0.062
	SDOH class 3	0.86 (0.68–1.10)	0.243	0.87 (0.68–1.11)	0.254	0.86 (0.64–1.16)	0.317
Trajectory3 VS 2	SDOH class 1	Reference		Reference		Reference
	SDOH class 2	1.20 (0.85–1.69)	0.297	1.21 (0.86–1.71)	0.265	1.24 (0.87–1.78)	0.235
	SDOH class 3	2.12 (1.54–2.93)	< 0.001	2.10 (1.52–2.90)	< 0.001	2.18 (1.49–3.20)	< 0.001

Odds ratios (ORs) and 95% confidence intervals (CIs) are presented for comparisons of Trajectory 1 vs. Trajectory 2 and Trajectory 3 vs. Trajectory 2, with Trajectory 2 as the reference outcome. SDOH class 1 (most advantaged) is the reference group

GWG Gestational weight gain, SDOH Social determinants of health, OR Odds ratio, CI Confidence interval 

Model E: unadjusted; Model F: adjusted for maternal age; Model G: further adjusted for pre-pregnancy BMI and parity

Bold P values indicate statistical significance (P < 0.05)

Trajectory 1, low weight gain trajectory; Trajectory 2, moderate weight gain trajectory; Trajectory 3, high weight gain trajectory

Subgroup analysis

After stratification by pre-pregnancy BMI (< 24.0 vs. ≥24.0 kg/m²), three GWG trajectories with similar shapes and ordering to the main analysis were identified within each stratum (Supplementary Figure S1). Detailed distributions of trajectory membership according to pre-pregnancy BMI categories in the overall cohort and within the BMI-stratified models are presented in Table 3. In the fully adjusted multinomial logistic regression models with Trajectory 2 as the reference outcome, among women with pre-pregnancy BMI < 24.0 kg/m², compared with SDOH Class 1, women in Class 2 and Class 3 had higher odds of belonging to the high GWG trajectory rather than the moderate trajectory (Trajectory 3 vs. 2; Class 2: aOR = 1.56, 95% CI 1.03–2.37, p = 0.037; Class 3: aOR = 3.16, 95% CI 1.99–5.01, p < 0.001). In addition, Class 2 was associated with lower odds of belonging to the low GWG trajectory rather than the moderate trajectory (Trajectory 1 vs. 2; aOR = 0.74, 95% CI 0.55–1.00, p = 0.047). In contrast, among women with pre-pregnancy BMI ≥ 24.0 kg/m², SDOH Class 3 was associated with lower odds of belonging to the low GWG trajectory rather than the moderate trajectory (Trajectory 1 vs. 2; aOR = 0.45, 95% CI 0.25–0.81, p = 0.007), whereas no significant associations were observed for the high GWG trajectory (Trajectory 3 vs. 2). Taken together, these findings suggest that the association between social disadvantage and high GWG trajectories was evident primarily among women with pre-pregnancy BMI < 24.0 kg/m², whereas among those with pre-pregnancy BMI ≥ 24.0 kg/m², social disadvantage was associated mainly with a lower likelihood of belonging to the low rather than the moderate GWG trajectory.

Table 3.

Distribution of gestational weight gain trajectory membership according to pre-pregnancy BMI categories in the overall cohort and BMI-stratified models

Pre-pregnancy BMI category	Overall (n, %)	Trajectory 1 (n, %)	Trajectory 2 (n, %)	Trajectory 3 (n, %)	P Value
Panel A: Overall cohort	n = 1,916	640 (33.40)	966 (50.42)	310 (16.18)	< 0.001
< 18.5 kg/m²	138 (7.20)	36 (26.09)	83 (60.14)	19 (13.77)
18.5–23.9 kg/m²	1198 (62.53)	351 (29.30)	632 (52.75)	215 (17.95)
24.0–27.9 kg/m²	428 (22.34)	171 (39.95)	191 (44.63)	66 (15.42)
≥ 28 kg/m²	152 (7.93)	82 (53.95)	60 (39.47)	10 (6.58)
Panel B: BMI ≥ 24.0 kg/m²	n = 580	163 (28.10)	323 (55.69)	94 (16.21)	0.007
24.0–27.9 kg/m²	428 (73.79)	108 (25.23)	241 (56.31)	79 (18.46)
≥ 28 kg/m²	152 (26.21)	55 (36.18)	82 (53.95)	15 (9.87)
Panel C: BMI < 24.0 kg/m²	n = 1336	470 (35.18)	645 (48.28)	221 (16.54)	0.123
< 18.5 kg/m²	138 (10.33)	41 (29.71)	78 (56.52)	19 (13.77)
18.5–23.9 kg/m²	1198 (89.67)	429 (35.81)	567 (47.33)	202 (16.86)

Data are presented as n (%). Percentages in the Overall column were calculated using the total number of women within each panel as the denominator, whereas percentages in the Trajectory 1–3 columns were calculated using the total number of women within each prepregnancy BMI category or subgroup as the denominator

Panel A shows results from the overall cohort model, while Panels B and C show results from models refitted separately within the prepregnancy BMI ≥ 24.0 kg/m² and < 24.0 kg/m² strata, respectively

P values were calculated using the chi-square test

Bold P values indicate statistical significance (P < 0.05)

Discussion

In this study, we identified a clear social gradient in GWG trajectories among women with GDM: socially disadvantaged women were more likely to follow high weight gain trajectories, whereas those in more advantaged conditions tended toward moderate weight gain trajectory. These patterns remained after adjustment for demographic and anthropometric factors and in BMI-stratified analyses, suggesting that social conditions independently shape weight dynamics during pregnancy.

We identified three distinct GWG trajectories, broadly consistent with prior pregnancy-cohort studies [22]. Variation in the number and shape of trajectories reported across the literature likely reflects differences in cohort composition, baseline BMI distribution, and the frequency of antenatal weight measurements [23]. Excessive GWG carries particular significance for women with GDM because it is linked to poorer glycemic control, hypertensive disorders, cesarean delivery, fetal overgrowth, and postpartum weight retention, all of which contribute to long-term metabolic risk for both mother and child [24]. These observations underscore the importance of understanding upstream determinants of gestational weight patterns [25].

Importantly, the clinical implications of a given GWG trajectory are unlikely to be the same across pre-pregnancy BMI groups. For women with pre-pregnancy overweight or obesity, lower or more moderate GWG may be clinically preferable, whereas among women with lower pre-pregnancy BMI, insufficient GWG may have different implications. Therefore, the trajectory labels in this study should be interpreted as relative patterns within this cohort rather than as universally optimal weight-gain targets. This is also consistent with our BMI-stratified analyses, in which the association between social disadvantage and higher GWG trajectories was more evident among women with pre-pregnancy BMI < 24.0 kg/m², whereas among those with pre-pregnancy BMI ≥ 24.0 kg/m², social disadvantage was associated mainly with a lower likelihood of belonging to the low rather than the moderate GWG trajectory. Together, these findings suggest that pre-pregnancy BMI should be considered when interpreting GWG trajectories and designing weight-management strategies for women with GDM.

Using a multidomain composite index, we characterized SDOH across educational, occupational, household and neighborhood domains. Prior GDM research has typically examined isolated social indicators and has focused primarily on GDM incidence or glucose outcomes [26]. Our findings suggest that cumulative social disadvantage meaningfully shapes GWG, plausibly through constrained access to high-quality food, safe activity environments, stable employment and consistent prenatal care. Mechanistically, structural disadvantage may act through limited resources and chronic psychosocial stress, influencing dietary quality, physical activity, and stress-related metabolic pathways, including hypothalamic–pituitary–adrenal axis activity, thereby contributing to steeper GWG trajectories [27, 28]. The persistence of associations after adjustment for pre-pregnancy BMI and age supports effects beyond baseline adiposity and suggests potential amplification among women with early glycemic dysregulation [29].

The observed association between SDOH disadvantage and less favorable GWG trajectories among women with GDM has important clinical implications. Medical centers should adopt targeted and structured strategies to support weight management, particularly for women in disadvantaged SDOH groups [30]. In routine care, clinicians may consider administering a brief SDOH questionnaire soon after the diagnosis of GDM to identify social barriers directly relevant to weight management, such as financial difficulty, limited access to healthy foods, low health literacy, inadequate family support, and limited opportunities for safe physical activity. Based on this early assessment, women at higher social risk could be prioritized for more intensive follow-up and tailored support. Practical strategies may include simpler dietary advice based on affordable foods, individualized physical activity recommendations, more frequent review of weight gain and glucose records, and earlier follow-up after diagnosis. Where possible, referral pathways may also be established for dietitians, diabetes educators, social workers, or community support services In addition, group education, or telephone follow-up, and risk-stratified care plans may help improve continuity of support for women facing greater social challenges [31]. Because the clinical implications of GWG may differ by pre-pregnancy BMI, counseling and monitoring should also be individualized according to baseline weight status rather than applying a uniform approach to all women with GDM. Future research should further examine how SDOH-related GWG patterns relate to maternal and neonatal outcomes [32].

Strengths of this study include the relatively large cohort of women with GDM, the use of LCLMM to characterize dynamic GWG trajectories, and the construction of a comprehensive multidomain SDOH index. Several limitations should also be acknowledged. First, the study was conducted in a relatively advantaged urban setting, which may limit the generalizability of the findings to populations with less favorable social conditions. Second, women included in the final analytic sample differed from excluded women in several baseline characteristics, and complete SDOH data were unavailable for all excluded participants; therefore, selection bias cannot be ruled out. Third, SDOH measures were based on self-report and may be subject to reporting error. Fourth, detailed time-varying behavioral data, such as changes in diet and physical activity during pregnancy, were unavailable, which may have resulted in residual confounding. Finally, although posterior probabilities indicated acceptable classification quality, some misclassification of trajectory membership remains possible [33]. Further multicenter studies in more socioeconomically diverse populations are warranted to link distinct SDOH–GWG patterns with adverse maternal and neonatal outcomes and to integrate biologic markers and device-based assessments of lifestyle behaviors to clarify the underlying pathways [34].

Conclusions

Among pregnancies complicated by GDM, three distinct gestational weight gain trajectories were identified, reflecting meaningful heterogeneity in weight gain patterns across pregnancy. This association was most evident among women with pre-pregnancy BMI < 24.0 kg/m², whereas among those with pre-pregnancy BMI ≥ 24.0 kg/m², no significant association was observed for the high GWG trajectory. These findings suggest potential heterogeneity in the relationship between social disadvantage and GWG trajectories by baseline BMI status. These findings support the integration of SDOH assessment into routine antenatal care for women with GDM and highlight the potential value of targeted, socially informed strategies to prevent excessive gestational weight gain and improve maternal health.

Abbreviations

GWG

Gestational weight gain

GDM

Gestational diabetes mellitus

SDOH

Social determinants of health

IOM

Institute of Medicine

OR

Odds ratio

aOR

Adjusted Odds Ratio

CI

Confidence interva

NICU

Neonatal Intensive Care Unit

BMI

Body mass index

IADPSG

International Association of Diabetes and Pregnancy Study Groups

OGTT

75 g Oral Glucose Tolerance Test

LCLMM

Latent Class Mixed Modeling

BIC

Bayesian Information Criterion

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

Authors’ contributions

DL. and X.L. contributed to the conception and design of the study, data acquisition, statistical analysis, and interpretation of the results, and D.L. wrote the first draft of the manuscript with input from X.L. H.B.K.H. contributed substantially to data interpretation and critically revised the manuscript for important intellectual content. D.L.J., D.L.K., and C.W.X. participated in the critical review of the manuscript and contributed to its revision. J.M. and J.H. supervised the data collection process, provided overall supervision and guidance throughout the study, and critically reviewed the final version of the manuscript. All authors edited, reviewed, and approved the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. J.M. and J.H. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Funding

This work was supported by the Key Research Project of Science and Technology of Sichuan Province (Grant No. 2023YFS0039).

Data availability

The datasets generated and analyzed during the current study are not publicly available because they are derived from routine clinical records across multiple hospitals and are subject to institutional policies and privacy regulations; de-identified data may be made available from the corresponding author upon reasonable request, contingent on approval from the participating hospitals and the relevant ethics committee.

Declarations

Ethics approval and consent to participate

This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (approval number: TJ-IRB20220611). As the study used anonymized retrospective data, the requirement for informed consent was waived.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.