Self‐management in gestational diabetes mellitus in Bangladesh: Determinants from women and caregivers' perspectives employed a cross‐sectional study

ABSTRACT

Background

Although clinical guideline of gestational diabetes mellitus (GDM) is well‐defined, still influence of the household environment and caregiver support on patient's adherence to self‐management remains under‐explored.

Objective

This study aimed to identify the patient, household, and caregiver‐level determinants of GDM self‐management in Dhaka, Bangladesh.

Methods

This cross‐sectional study was conducted among 251 dyads of GDM patients including their caregivers across four tertiary hospitals in Dhaka. Diabetes self‐management (DSM) was evaluated using the validated Diabetes Self‐Management Questionnaire (DSMQ) (Cronbach's alpha = 0.74), supplemented by a pretested, semi‐structured instrument assessing caregiver knowledge and support. Predictors of suboptimal DSM were identified using multivariable logistic regression with backward elimination in STATA. Model discrimination was validated using the area under the receiver operating characteristic curve (AUC‐ROC).

Results

Suboptimal DSM was observed in 68.5% of participants. Domain‐specific analysis revealed critical gaps in glycemic monitoring (82.5%), dietary control (75.7%), and physical activity (73.3%), while healthcare utilization (56.2%) showed higher adherence. Multivariable modeling demonstrated that household architecture was a dominant predictor; living in nuclear or extended families significantly increased the odds of suboptimal DSM compared to living alone. Further determinants included clinical complexity (comorbidities, family history) and reproductive history (parity ≥2). Notably, caregiver‐level factors, specifically poor knowledge and inadequate support (prevalent in >50% of caregivers), were robustly associated with deficient glycemic monitoring and healthcare utilization (AUC = 0.78).

Conclusion

The findings suggest that the household unit, rather than the individual patient, is the functional unit of GDM care. Transitioning toward family‐centered therapeutic models is essential to improve maternal and neonatal outcomes in GDM cases.

This prevalence study evaluated 251 patient‐caregiver dyads across four tertiary hospitals in Dhaka, Bangladesh to identify the determinants of gestational diabetes mellitus (GDM) self‐management practice. Suboptimal self‐management was found highly prevalent at 68.5% of patients, driven by critical gaps in glycemic monitoring, dietary control, and physical activity, alongside poor caregiver knowledge and support. The findings demonstrate that the supportive household environment, rather than the individual patients, is the functional unit of care, recommending a transition toward family‐centered therapeutic models.

INTRODUCTION

Gestational diabetes mellitus (GDM) is a major contributor to the growing burden of noncommunicable diseases (NCDs) worldwide. Globally, an estimated 14–20% of live births are affected by hyperglycemia in pregnancy, with approximately 79–85% attributable to GDM, and the highest prevalence observed in low‐ and middle‐income countries (LMICs) ¹ , ² . South Asia represents a particular hotspot, where women from South Asia are at the highest risk of GDM, with prevalence in South‐East Asia reaching 31.8% and urban prevalence ranging from 10% to over 25% in the region, reflecting rapid epidemiological transition, rising maternal age, and increasing adiposity ² , ³ , ⁴ . Beyond immediate obstetric risks, GDM is now recognized as a powerful early marker of future metabolic disease: women with prior GDM have a 7–10‐fold higher risk of developing type 2 diabetes mellitus (T2DM) compared with women with normoglycemic pregnancies, with cumulative incidence reaching approximately 16–20% within 5–10 years postpartum and a 50–60% lifetime risk ⁵ , ⁶ , ⁷ , ⁸ . At the population level, GDM therefore represents a critical entry point into the intergenerational transmission of cardiometabolic risk, contributing substantially to the accelerating NCD epidemic.

Effective management of GDM during pregnancy relies predominantly on diabetes self‐management (DSM) behaviors, including regular blood glucose monitoring, dietary regulation, physical activity, and appropriate healthcare utilization. Evidence suggests that treatment of GDM can reduce the risk of macrosomia by approximately 40–60%, lower rates of preeclampsia (relative risk 0.46), and decrease the need for pharmacological treatment ¹ , ⁹ , ¹⁰ . However, real‐world adherence remains suboptimal: studies across diverse settings report that only 61.5% of women with GDM performed ≥80% of required blood glucose tests, and adherence to dietary and lifestyle recommendations varies considerably, with multiple barriers identified including lack of knowledge, social support deficits, and cultural constraints ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ . While individual knowledge and health literacy are important, growing literature indicates that self‐management during pregnancy is deeply embedded within social and household contexts. In many LMICs, pregnant women's autonomy over diet, mobility, and healthcare decisions is constrained by family norms, caregiving roles, and economic dependence, yet these influences remain underrepresented in conventional models of GDM care ¹⁰ , ¹³ , ¹⁵ . Caregivers, often spouses or senior family members, frequently act as gatekeepers to food choices, healthcare access, and adherence to medical advice, with both supportive and inhibiting effects.

In Bangladesh, where NCDs now account for over 70% of total adult mortality and noncommunicable diseases represented 14 of the top 20 causes of death in 2019, diabetes prevalence has increased substantially, with age‐standardized mortality rates decreasing from 1509.3 to 714.4 deaths per 100,000 population between 1990 and 2019, yet the burden of diabetes and hypertension continues to rise ¹⁶ , ¹⁷ , ¹⁸ . Despite this, empirical evidence on how women with GDM manage their condition in everyday settings, and how household and caregiver factors shape this process, remains limited. Existing studies in Bangladesh have focused primarily on prevalence, screening practices, or individual knowledge, offering little insight into the social architecture of diabetes care during pregnancy. To address this gap, the present study examines the status of DSM among women with GDM attending tertiary‐level hospitals in Dhaka, Bangladesh. We hypothesized that GDM self‐management is shaped by multilevel determinants, with household structure and caregiver support exerting independent effects beyond individual patient characteristics. Accordingly, the objective of this study is to identify factors associated with suboptimal GDM self‐management by integrating perspectives from women with GDM and their caregivers, and to delineate patient‐, household‐, and caregiver‐level determinants of self‐management practices to inform family‐centered GDM care strategies in resource‐constrained settings.

METHODS

Study design

This analytical cross‐sectional study utilized a quantitative approach to collect structured data from 01/05/2025 to 30/07/2025. The study explored the status of DSM and the factors associated with suboptimal care among women with GDM in selected urban hospitals of Bangladesh.

Study participants, sample size, and sampling

This study included 251 adult women with GDM who had been receiving specialized care for a minimum of 8 weeks and their respective caregivers who were attended in the four tertiary‐level hospitals of Dhaka city, Bangladesh. This specific service receiving duration was established as a clinical threshold to ensure respondents had sufficient opportunity to integrate self‐management protocols into their domestic routines. The study utilized a stratified random sampling approach to select four tertiary‐level facilities from the Bangladesh Health Facility Registry, ensuring a representative mix of public and private sectors where regular diabetes treatment is available for pregnant women. The hospitals were: Bangladesh Institute of Health Sciences General Hospital, Mirpur; Shaheed Suhrawardy Medical College Hospital, Sher‐E‐Bangla Nagar; Bangabandhu Sheikh Mujib Medical University Hospital, Shahbag; Azimpur Maternity Hospital, Azimpur, Dhaka. Quantitative information for this study was collected from respondents, signified adult women diagnosed with GDM who met the following inclusion criteria: being at any trimester of pregnancy; having received treatment for at least 8 weeks prior to data collection; being accompanied by a caregiver or family member; and expressing willingness to participate in the study.

Initially, the required sample size was estimated to be 228 using Cochran's formula ‘n = ‘Z²pq/d²’, where Z (the standard normal deviate) was set at 1.96; p (the estimated proportion of nonadherence to self‐management) was taken as 18.1% based on a previous study ¹⁹ and expressed as 0.181; and the margin of error (d) was set at 0.05. However, the final sample size was purposively powered up to 251 by accounting for a 10% nonresponse rate and to enhance the stability of the multivariable logistic regression models.

Dhaka district, under the Dhaka division of Bangladesh, was selected based on a greater number of GDM treatment facilities. The study sites were selected randomly from the list of hospitals in Dhaka city, Bangladesh, that offer regular GDM treatment ²⁰ . GDM women from the selected hospitals who met the inclusion criteria were included in the study sample. Caregivers who accompanied patients during their hospital visit and voluntarily consented to participate were included in the study.

Data collection

A pretested, semi‐structured questionnaire was used to collect quantitative data from adult women with GDM and their caregivers, using an interviewer‐administered method. Respondents were recruited and interviewed in this study from May to July 2025. Each interview required approximately 10–15 min for the interviewer to complete. All authors had access to the collection and preservation of participants' information during or after data collection. The survey was administered in the Bengali language with the utmost support of the local authority.

Ethical considerations

This study was approved by the Institutional Review Board of Northern University Bangladesh (NUB/ DPH/EC/2025/37‐a) and conformed to the Declaration of Helsinki. The respondents' participation was anonymous and voluntary. A written informed consent was obtained from each respondent at the beginning of the survey, and participants could withdraw at any time. To protect participant privacy, data were de‐identified and pseudonymized during the analysis phase.

Questionnaire design

The questionnaire was prevalidated by two independent reviewers and pretested among 10 respondents. Feedback from the pretest was incorporated into the final version. The questionnaire consisted of several sections: (i) DSM assessment using Diabetes Self‐Management Questionnaire (DSMQ) among GDM women: glucose management, dietary control, physical activity, and healthcare use related characteristics; (ii) Sociodemographic and hospital access‐related characteristics of the GDM women: age, education, occupation, number of children, family size, family type, monthly family income, distance to health facility, transport cost; (iii) Disease and pregnancy related characteristics: blood glucose at enrollment, comorbidity, pregnancy trimester, previous conception; (iv) Sociodemographic, knowledge and support services related characteristics of caregivers of GDM women: age, education, occupation, marital status, knowledge and support services on GDM management (preconception care, weight gain, drug intake, frequency of food intake, avoided carbohydrate and fat foods, physical exercise, blood glucose test, follow‐up visit and postpartum plan). Healthcare utilization is conceptualized as the engagement of the GDM patients with healthcare facilities in Bangladesh. Within the DSMQ framework, the healthcare use subscale specifically assessed the regular maintenance of clinical check‐ups and the patient's adherence to consult healthcare professionals for diabetes‐related complications. Additionally, we recorded physical and economic barriers to utilization, including the distance to the nearest health facility (measured in kilometers) and the transport costs incurred for healthcare service utilization. To ensure construct and linguistic validity, the whole questionnaire underwent a rigorous forward and back‐translation process into Bengali by independent linguistic experts. The instrument's reliability in the local context was confirmed with a Cronbach's alpha of 0.74, indicating robust internal consistency. To minimize ascertainment bias, the questionnaire was pretested among a pilot cohort (n = 10), and feedback was synthesized to refine the clarity of the sociodemographic and caregiver support modules prior to formal administration.

Data analysis

The collected data were checked and analyzed using Stata software. Descriptive statistics (frequencies and percentages) were used to summarize the study characteristics. Continuous variables such as age, family size, and monthly income were categorized using cut‐offs at the midpoints of the percentage distributions ²¹ . A scoring system was developed to determine the caregiver's knowledge and support toward GDM women. Each correct response, including multiple answer items, was assigned a score of 1, and each incorrect response a score of 0. The total score was then converted to a percentage. Due to the lack of a standardized threshold for this specific population, the mean score was used as the cut‐off point. Caregivers scoring below the mean were classified as ‘poor,’ and those scoring at or above the mean were classified as ‘sufficient.’ This approach facilitated a relative assessment of support levels within the study subjects ²² . The DSMQ scale consists of 16 items rated on a 4‐point Likert scale. A ‘Sum Scale’ was calculated and transformed into a score ranging from 0 to 10, where higher scores indicate more effective self‐management. For the data analysis of this study, self‐management outcome was dichotomized as a score of ≥6 which was defined as ‘optimal’ (indicating adequate adherence to self‐care), and a score of <6 which was defined as ‘suboptimal’ ²³ . Univariate and multivariable logistic regression analyses were performed, followed by backward elimination modeling that eliminated prespecified confounders. Odds Ratios (ORs) with 95% confidence intervals (CIs) concerning DSM (optimal and suboptimal) were calculated for the specified exposures. The predictive accuracy and discriminatory performance of the final models were rigorously assessed using the area under the receiver operating characteristic curve (AUC‐ROC), with values categorized from moderate to good.

RESULTS

Characteristics of women with gestational diabetes mellitus (GDM) and their caregivers

A total of 251 women with GDM and their corresponding caregivers were included in the analysis (Tables 1 and 2). Most participants were aged 25–34 years, with nearly two‐fifths aged 25–29 years. Although one‐third had completed tertiary education, the majority had secondary or lower educational attainment, and over four‐fifths were housewives. Family structures were predominantly nuclear, though nearly one‐third lived in extended households. More than half of the participants reported a family history of diabetes, and approximately one‐third had at least one comorbid condition. Most women were recruited during the third trimester, and over half had experienced two or more previous conceptions (Table 1).

Table 1.

Sociodemographic and clinical characteristics of patients (n = 251)

Variables	Category	n (%)
Age (years)	<25	50 (19.9)
	25–29	100 (39.8)
	30–34	66 (26.3)
	≥35	35 (13.9)
Education	≤SSC	109 (43.4)
	HSC	56 (22.3)
	Tertiary (≥graduate)	86 (34.3)
Occupation	Housewife	208 (82.9)
	Others	43 (17.1)
Number of children	None	83 (33.1)
	One	94 (37.5)
	≥2	74 (29.5)
Family size	<4	112 (44.6)
	4–5	103 (41.0)
	>5	36 (14.3)
Family type	Living alone	44 (17.5)
	Nuclear	132 (52.6)
	Extended	75 (29.9)
Monthly family income	<245 USD	95 (37.9)
	245–327 USD	95 (37.9)
	>327 USD	61 (24.3)
Distance to health facility	<14 km	222 (88.5)
	≥14 km	29 (11.5)
Transport cost	<2.27 USD	212 (84.5)
	≥2.27 USD	39 (15.5)
Blood glucose at enrollment	<6.7 mmol/L	30 (12.0)
	≥6.7 mmol/L	221 (88.0)
Family history of diabetes	No	104 (41.4)
	Yes	147 (58.6)
Comorbidity	No	173 (68.9)
	Yes	78 (31.1)
Pregnancy trimester	1st	59 (23.5)
	2nd	75 (29.9)
	3rd	117 (46.6)
Previous conception	<2	115 (45.8)
	≥2	136 (54.2)

Data are presented as frequency (n) and percentage (%); HSC, Higher Secondary Certificate; SSC, School Secondary Certificate.

Table 2.

Sociodemographic, knowledge and support services related characteristics of caregivers (n = 251)

Variable	Category	n (%)
Age (years)	<30	41 (16.3)
	30–39	93 (37.1)
	40–49	81 (32.3)
	≥50	36 (14.3)
Education	Illiterate	47 (18.7)
	<SSC	87 (34.7)
	HSC	30 (12.0)
	Tertiary (≥graduate)	87 (34.7)
Occupation	Service	85 (33.9)
	Business	59 (23.5)
	Housewife	85 (33.9)
	Others	22 (8.8)
Marital status	Married	237 (94.4)
	Separated	14 (5.6)
Number of children	None	58 (23.1)
	1–2	130 (51.8)
	≥3	63 (25.1)
Caregiver knowledge	Sufficient	118 (47.0)
	Poor	133 (53.0)
Caregiver support	Sufficient	94 (37.5)
	Poor	157 (62.5)

Data are presented as frequency (n) and percentage (%); Caregiver knowledge and support were categorized using a mean‐split method. HSC, Higher Secondary Certificate; SSC, School Secondary Certificate.

Caregivers were typically middle‐aged, with more than two‐thirds aged 30–49 years. Educational attainment was heterogeneous, with roughly one‐third completing tertiary education, while a substantial proportion had primary or no formal schooling. Caregiver roles were diverse, spanning service, business, and homemaking occupations. Notably, over half of caregivers demonstrated poor diabetes‐related knowledge, and nearly two‐thirds reported poor caregiving support, highlighting a potentially important contextual influence on diabetes management (Table 2).

Status of diabetes self‐management (DSM) among women with GDM

Suboptimal care predominated across most care domains among women with GDM. In the glycemic monitoring (GM) domain, 82.5% of participants had suboptimal care, while only 17.5% achieved optimal care. Similarly, suboptimal care was observed in dietary control (75.7%) and physical activity (73.3%). In contrast, healthcare utilization (HU) demonstrated comparatively better outcomes, with 56.2% of participants receiving optimal care. Overall, DSM was largely suboptimal, affecting 68.5% of participants, whereas only 31.5% demonstrated optimal self‐management practices (Figure 1).

Figure 1.

Status of diabetes self‐management (DSM) among women with GDM (n = 251). DC, dietary control; GM, glucose management; HU, healthcare use; PA, physical activity.

Univariate associations with suboptimal diabetes self‐management

In univariate analyses, household structure emerged as a dominant correlate of suboptimal DSM. Women living in nuclear or extended families had substantially higher odds of suboptimal DSM compared with those living alone. A family history of diabetes, presence of comorbidities, and having two or more prior conceptions were also positively associated with poorer self‐management. In contrast, individual sociodemographic factors such as age, education, occupation, and access‐related variables (distance to facility, transport cost) were not independently associated with DSM at the univariate level. Among caregiver characteristics, tertiary caregiver education was associated with higher odds of suboptimal DSM, whereas caregiver support showed no clear univariate association (Table 3).

Table 3.

Univariate logistic regression analysis of factors associated with suboptimal diabetes self‐management (DSM) (n = 251). Outcome: Suboptimal DSM (reference = Optimal DSM)

Variable	Category	OR	95% CI	P‐value
Age (years)	25–29	0.75	0.36–1.59	0.458
	30–34	0.96	0.43–2.17	0.926
	≥35	0.75	0.29–1.89	0.536
Education	HSC	1.81	0.88–3.71	0.106
	Tertiary	1.56	0.85–2.87	0.155
Occupation	Others	1.23	0.59–2.54	0.580
Number of children	One	0.66	0.34–1.26	0.206
	≥2	0.63	0.31–1.24	0.181
Family size	4–5	1.05	0.59–1.87	0.866
	>5	1.08	0.48–2.43	0.859
Family type	Nuclear	4.71	2.28–9.70	<0.001
	Extended	3.48	1.60–7.59	0.002
Monthly income	245–327 USD	1.15	0.63–2.10	0.646
	>327 USD	1.87	0.90–3.88	0.092
Distance to facility	≥14 km	1.51	0.62–3.69	0.368
Transport cost	≥2.27 USD	1.40	0.65–3.03	0.395
RBS at enrollment	≥6.7 mmol/L	0.92	0.40–2.12	0.853
Family history of DM	Yes	1.73	1.01–2.97	0.046
Comorbidity	Yes	2.47	1.30–4.69	0.006
Pregnancy trimester	2nd	0.74	0.35–1.57	0.439
	3rd	0.77	0.39–1.55	0.467
Previous conception	≥2	2.08	1.21–3.57	0.008
Diabetes complications	Yes	1.48	0.81–2.72	0.202
Patient knowledge	Poor	1.40	0.82–2.38	0.222
Caregiver education	Tertiary	2.54	1.19–5.41	0.016
Caregiver support	Poor	1.41	0.82–2.43	0.216

Reference categories: hospital (reference site), age < 25 years, ≤SSC education, housewife, no children, family size <4, living alone, income < 30,000 BDT, distance < 14 km, transport cost < 278 Tk, RBS < 6.7 mmol/L, no family history of diabetes, no comorbidity, first trimester, <2 conceptions, sufficient knowledge. CI, confidence interval; OR, odds ratio.

Multivariable models of diabetes care outcomes

Multivariable analyses identified distinct yet overlapping predictors across DSM, GM, dietary control (DC), physical activity (PA), and HU (Table 4).

Table 4.

Multivariable logistic regression models for diabetes care outcomes. Adjusted odds ratios (AOR) with 95% confidence intervals shown for all predictors retained in final models

Predictor	DSM	GM	DC	PA	HU
Family type: Nuclear	4.48 (2.08–9.64)***			4.71 (2.05–10.82)***
Family type: Extended	4.01 (1.72–9.34)**	0.16 (0.04–0.57)**		3.41 (1.24–9.38)*
Family history of diabetes	1.91 (1.06–3.46)*	3.09 (1.45–6.58)**	1.94 (1.08–3.50)*
Comorbidity	2.17 (1.09–4.33)*
Previous conception ≥2	2.78 (1.41–5.48)**			2.47 (1.15–5.29)*
Number of children: ≥2		0.06 (0.01–0.30)***		0.29 (0.11–0.79)*
Family size >5		9.83 (1.47–65.72)*		0.43 (0.16–1.13)†
Patient occupation: Others		0.31 (0.12–0.78)*	0.54 (0.27–1.11)†
Caregiver occupation: Housewife		0.35 (0.15–0.79)*			2.19 (1.16–4.17)*
Caregiver occupation: Others				6.78 (1.33–34.60)*
Caregiver support: Poor		0.37 (0.15–0.89)*			1.64 (0.92–2.92)†
Caregiver age ≥ 50 years		3.29 (1.04–10.37)*			2.26 (0.96–5.30)†
Patient transport cost >278 Tk					0.46 (0.21–1.01)†
Patient education: Tertiary					0.57 (0.30–1.09)†
Caregiver education: Tertiary					1.97 (0.98–3.96)†
Model constant (_cons)	0.28 (0.12–0.66)**	14.82 (4.29–51.16)***	2.44 (1.56–3.81)***	0.46 (0.13–1.67)	0.41 (0.22–0.77)**
AUC	0.73	0.78	0.61	0.76	0.68

AUC, area under the ROC curve. Significance: ***P < 0.001; **P < 0.01; *P < 0.05; ^† P < 0.10.

For DSM, household context remained central. Living in nuclear or extended families was independently associated with markedly higher odds of suboptimal DSM, even after adjustment. Clinical complexity also played a role, with comorbidities, family history of diabetes, and multiple prior conceptions showing independent positive associations. The DSM model demonstrated acceptable discrimination (AUC = 0.73).

The GM model revealed a more nuanced interplay between household, caregiver, and patient factors. Extended family living and poor caregiver support were associated with lower odds of optimal GM, whereas family history of diabetes and larger family size increased the likelihood of suboptimal GM. Caregiver characteristics were prominent: caregiver occupation, age, and support status all contributed independently. This model showed the strongest discriminatory performance among all outcomes (AUC = 0.78).

For DC, fewer predictors were retained. Family history of diabetes remained positively associated with suboptimal dietary control, while patient occupation showed a borderline inverse association. Overall discrimination was modest (AUC = 0.61), suggesting limited explanatory capacity for this domain.

The PA model highlighted reproductive history and family structure as key determinants. Living in nuclear or extended families and having two or more prior conceptions were associated with poorer physical activity practices, while having children was inversely associated with optimal PA. The PA model demonstrated good discrimination (AUC = 0.76).

In the HU model, caregiver‐related factors were particularly influential. Caregiver occupation, age, and educational attainment, along with caregiver support, were retained in the final model. Higher transport costs and tertiary patient education were inversely associated with healthcare utilization, albeit at borderline significance. The HU model showed moderate discrimination (AUC = 0.68).

Model performance and robustness

Across outcomes, model constants varied substantially, reflecting differences in baseline prevalence of suboptimal behaviors. Discriminatory performance ranged from moderate to good, with strongest performance observed for GM and PA, and weakest for DC. Collectively, the findings underscore the salience of household structure and caregiver context, often outweighing individual sociodemographic characteristics, in shaping multiple dimensions of diabetes care among women with GDM.

DISCUSSIONS

The present study identified that self‐management practices among pregnant women with diabetes in urban Bangladesh are predominantly suboptimal, with over two‐thirds of participants exhibiting inadequate overall DSM. The most significant deficiencies were noted in GM, dietary control, and PA, whereas healthcare utilization demonstrated relatively better performance. These findings suggest that although women may be engaging with formal health services, the translation of clinical advice into daily self‐care behaviors remains limited. However, several previous studies have shown that effective self‐management practices, particularly adherence to a healthy diet, regular PA, and consistent blood glucose monitoring, can substantially reduce hyperglycemia and improve both maternal and neonatal pregnancy outcomes ²⁴ , ²⁵ , ²⁶ . Evidence from a scoping review in LMICs suggests that self‐care approaches incorporating lifestyle counseling, family involvement, and regular follow‐up are frequently associated with improved glycemic control and enhanced overall well‐being among women with gestational diabetes ²⁷ .

In our adjusted regression model, we demonstrate that effective diabetes care during pregnancy extends beyond individual‐level clinical management and is deeply influenced by family structure and caregiving environments. While World Health Organization's global guidelines for GDM emphasize glycemic targets, dietary modification, and PA ⁴ , our findings indicate that the feasibility of implementing these recommendations is strongly shaped by household dynamics and caregiver engagement. In particular, the elevated likelihood of suboptimal self‐management and lifestyle practices among women living in nuclear or extended family settings suggests that shared household decision‐making, competing domestic responsibilities, and sociocultural norms may constrain adherence to recommended behaviors. From an implementation perspective, these results highlight a critical gap between guideline intent and real‐world delivery, indicating that clinic‐based counseling alone may be insufficient without family‐centered approaches embedded within antenatal care. Our findings align with and extend prior evidence showing that GDM management is influenced by social and familial contexts, particularly in settings where women's autonomy over diet, PA, and healthcare‐seeking may be constrained ²⁸ , ²⁹ , ³⁰ . Previous qualitative research has similarly indicated that conflicting household food preferences, entrenched family routines, and shared decision‐making processes often impede adherence to recommended dietary and lifestyle practices among women with GDM ²⁹ , ³⁰ . In line with earlier evidence, family members may simultaneously function as facilitators and barriers, where misconceptions about the disease, unequal dietary freedoms within the household, and competing caregiving responsibilities reduce women's ability to prioritize their own self‐care. In contrast, previous studies have primarily focused on individual knowledge and self‐efficacy; however ²⁶ , ³¹ , ³² , our results suggest that household structure and caregiver attributes exert an independent and often stronger influence across several domains of diabetes care. These parallels indicate that, as observed in broader diabetes populations, effective GDM self‐management requires not only individual counseling but also active engagement of family members to minimize obstructive behaviors and strengthen supportive household environments. Importantly, the differential predictors observed across diabetes care domains further highlight that GDM management is multidimensional, with distinct mechanisms shaping self‐management, lifestyle behaviors, and healthcare engagement. This underscores the need for tailored intervention strategies rather than a uniform approach to GDM care.

Another important finding of this study is the significant association between caregiver characteristics and domain‐specific diabetes care outcomes. Caregiver age ≥ 50 years was significantly associated with higher odds of suboptimal GM, indicating that older caregivers may face challenges in supporting regular glucose testing or adapting to contemporary monitoring practices. This finding is supported by qualitative evidence from previous studies showing that older caregivers often experience physical health limitations, psychological exhaustion, and economic burdens, which can reduce their capacity to consistently assist with diabetes management care ³³ . Such constraints may hinder timely glucose monitoring and sustained engagement in routine self‐care supervision. In contrast, having a caregiver whose occupation was a housewife was associated with significantly lower odds of suboptimal GM, indicating that greater household presence and time availability may facilitate routine glucose checking and day‐to‐day supervision of monitoring behaviors. This finding is supported by previous studies indicating that the presence of a full‐time homemaker caregiver is associated with improved diabetes management outcomes, including better glycemic control and lower HbA1c levels, likely due to increased caregiving time and closer health supervision ³⁴ . However, the same caregiver characteristic was linked to higher odds of suboptimal healthcare utilization, implying that while housewife caregivers may effectively support home‐based self‐care activities, this involvement does not necessarily extend to encouraging consistent engagement with formal health services such as scheduled clinic visits or follow‐up consultations. Collectively, this contrasting pattern highlights that caregiver roles exert dual and domain‐specific influences, where availability may strengthen daily management practices but does not automatically translate into improved interaction with the healthcare system.

Our study revealed a significant association between family history of diabetes and suboptimal diabetes care behaviors. Women with familial exposure to diabetes had higher odds of suboptimal overall self‐management, dietary control, and PA compared with those without such history. Although this finding may appear counterintuitive, similar patterns have been reported in previous studies, indicating that familial exposure to diabetes is often associated with less healthy lifestyle behaviors, weaker behavioral regulation, and poor glycemic control ³⁵ , ³⁶ . Prolonged exposure to diabetes within the household may normalize the condition, reduce perceived disease severity, and consequently diminish the urgency to adhere strictly to recommended lifestyle and monitoring practices. However, studies from South Asian and African American contexts present a different picture, reporting that individuals with a diabetic family history may demonstrate better dietary adherence, adopt healthier behavioral changes, and maintain more optimal self‐care practices ³⁶ , ³⁷ . This inconsistency suggests that the influence of familial exposure is context‐dependent and may vary according to cultural beliefs, levels of health literacy, and access to supportive information and healthcare resources. In addition, the presence of comorbid conditions in the current study was also associated with higher odds of suboptimal self‐management, likely reflecting increased treatment burden ³⁸ , competing health priorities ³⁹ , and psychological stress ⁴⁰ that may limit sustained behavioral adherence. Together, these findings indicate that both familial and clinical factors interact to shape DSM, underscoring the importance of personalized counseling that addresses risk perception, belief systems, and coexisting health demands.

This study has several strengths. First, it simultaneously examined multiple components of diabetes care, allowing a more comprehensive understanding of GDM management than studies focusing on a single outcome. Second, the inclusion of both patient and caregiver characteristics enabled assessment of household‐level influences that are often overlooked in clinical research. Third, the use of multivariable models with consistent variable selection criteria across outcomes enhanced internal comparability and reduced selective reporting. Finally, the moderate to good discriminatory performance of the models supports the robustness of the identified associations and their potential relevance for clinical and public health practice.

Several limitations should be acknowledged. The cross‐sectional design precludes causal inference, and observed associations may reflect bidirectional or unmeasured relationships. Self‐reported measures of behaviors such as PA and self‐management may be subject to recall or social desirability bias. Although a broad range of patient and caregiver variables were included, residual confounding from unmeasured factors—such as intra‐household power dynamics, cultural norms, or mental health—cannot be excluded. Additionally, the study was conducted within a specific healthcare and sociocultural context, which may limit generalizability to settings with different family structures or healthcare systems. Finally, while stepwise modeling was used to identify relevant predictors, this approach may yield models that are sensitive to sample‐specific variation.

The findings have important implications for both clinical practice and implementation science. Clinically, they suggest that effective GDM management requires moving beyond an exclusive focus on the pregnant individual to include systematic assessment of household and caregiver contexts during antenatal visits. Incorporating caregivers into education sessions, counseling, and follow‐up may enhance adherence to recommended practices, particularly for GM and lifestyle modification. From an implementation perspective, the results highlight the need for family‐centered models of care that are adaptable to diverse household structures and sociocultural settings. Interventions that leverage caregivers as partners in care, rather than assuming uniform support, may improve uptake and sustainability of guideline‐recommended behaviors. At the policy level, integrating family‐focused strategies into routine antenatal services could strengthen the real‐world effectiveness of GDM programs, particularly in resource‐constrained settings.

CONCLUSION & RECOMMENDATIONS

This study revealed that gestational DSM in urban Bangladesh is predominantly suboptimal, with the domestic architecture and household support capacity exerting a more profound influence on outcomes than individual clinical or sociodemographic profiles. The persistent failure in GM and dietary adherence is significantly determined by a widespread lack of caregivers' knowledge. The study also identified the home environment as a primary barrier to metabolic control. Consequently, clinical management must undergo a paradigm shift from a patient‐centric approach to a family‐centered and support‐mediated dyadic care model. We recommend that health policy interventions formally integrate primary caregivers into clinical counseling and leverage culturally tailored platforms to foster shared household accountability. Addressing these socioecological determinants is essential to improve long‐term maternal–neonatal outcomes, especially for GDM cases in resource‐constrained South Asian settings.

DISCLOSURE

The authors have declared that no conflict of interest exists.

Approval of the research protocol: This study was approved by the Ethical Review Committee of Northern University Bangladesh (NUB/ DPH/EC/2025/37‐a).

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

FUNDING

Bangladesh Medical Research Council (BMRC).

AUTHOR CONTRIBUTIONS

Bilkis Banu, Nasrin Akter, Fatema Ashraf: conceptualization. Bilkis Banu, Nasrin Akter: data curation. Md Nazmul Karim, Bilkis Banu: formal analysis. Bilkis Banu, Fatema Ashraf, Nasrin Akter: funding acquisition. Bilkis Banu, Fatema Ashraf, Nasrin Akter: investigation. Bilkis Banu: methodology. Bilkis Banu, Fatema Ashraf: resources. Md Nazmul Karim, Bilkis Banu: software. Bilkis Banu, Fatema Ashraf: supervision. Bilkis Banu, Fatema Ashraf, Md Nazmul Karim, Nasrin Akter: validation. Bilkis Banu, Nasrin Akter, Md Jiaur Rahman, Moomtahina Fatima, Fatema Ashraf, Md Nazmul Karim: visualization. Bilkis Banu, Nasrin Akter, Md Jiaur Rahman, Moomtahina Fatima, Fatema Ashraf, Md Nazmul Karim: writing – original draft. Bilkis Banu, Nasrin Akter, Md Jiaur Rahman, Moomtahina Fatima, Fatema Ashraf, Md Nazmul Karim: Writing – review and editing.

DATA AVAILABILITY STATEMENT

The data supporting the findings of this study are available from the corresponding author upon reasonable request.