This sub-analysis of the BHARAT study evaluated the association between anthropometric indices and T2DM in adults with obesity. Higher age increased the likelihood of diabetes, while measures of central and regional adiposity showed variable associations. NHR emerged as a consistent marker associated with T2DM in both sexes, whereas WHR showed an association only in men. ROC analysis demonstrated modest discriminatory ability of NHR for identifying T2DM.

The mean age of the study population was 48.16 years. Women constituted 63.6% of the participants, which may reflect differences in health-seeking behavior. The distribution across obesity classes was 43.2% in class I, 36.8% in class II, and 20.0% in class III obesity.

In this cohort, most anthropometric measures increased with higher obesity classes in both sexes, although some differences were noted. NC and NHR showed a progressive rise across obesity classes in women, whereas this pattern was not evident in men. Prior studies have reported that NC tends to show a stronger relationship with metabolic risk in women, while the association may plateau in men at higher BMI levels [32, 33]. Men generally have larger baseline neck measurements because of greater skeletal and muscular frame size, which may reduce the sensitivity of NC to incremental fat accumulation [13, 34]. Another possible explanation could be the greater accumulation of visceral and truncal fat rather than cervical fat in men [35]. Consistent with this, earlier studies have shown that obesity in men is largely characterized by visceral fat deposition, whereas women tend to demonstrate greater expansion of subcutaneous adipose tissue [36, 37].

The proportion of participants with T2DM in this cohort was high (73%). This is likely related to the recruitment setting, as participants were enrolled from endocrinology outpatient clinics and therefore may not reflect the general population. Diabetes was more frequent in men (81.6%) than in women (67%). This pattern may reflect greater metabolic susceptibility among men, while women may seek medical attention earlier during the course of obesity. The prevalence of diabetes did not increase with higher obesity classes, and the lowest proportion of T2DM was observed among individuals with class III obesity. This pattern may also reflect the clinic-based nature of the cohort, where individuals with severe obesity may present earlier for weight management.

Among men, multivariate analysis revealed that age above 50 years, higher pulse rate, and increased WHR and NHR were independent predictors of T2DM. In women, age above 40 years, married status, higher systolic blood pressure, and elevated NHR remained independently associated with T2DM. ROC analysis showed that the NHR had modest ability to discriminate T2DM, with an AUC of 0.63 and an optimal cutoff of 0.24. WHR was a determinant of T2DM in men but not in women. The difference may be attributed to distinct fat distribution patterns in men, who are more predisposed to visceral fat accumulation [37].

Our findings align with global and regional evidence supporting NC and NHR as clinically relevant markers. A 2023 meta-analysis comprising over 30,000 participants reported that each 1-cm increase in NC increased the odds of T2DM by 16%, independent of BMI and WC [38]. In an Indian cohort, both NC and NHR showed strong associations with metabolic syndrome and cardiovascular risk factors, with NHR demonstrating slightly better predictive accuracy than NC [39].

An Indian community-based study, the Indian Diabetes Risk Score (IDRS), demonstrated a strong positive correlation between NC and WC (r = 0.837, p < 0.0001) and proposed sex-specific cutoff values for identifying abnormal NC (≥ 37 cm in men and ≥ 34 cm in women). Although NC correlated well with central adiposity, it could not fully replace WC for predicting diabetes risk, and substituting WC with NC in the IDRS yielded comparable risk stratification [40]. The proposed cutoff values for NC in IDRS were close to thresholds observed in our cohort.

CC and WrC, though were not independently associated with T2DM, are emerging metrics that require further validation. In a large analysis, inclusion of CC in defining metabolic syndrome improved the prediction of adverse outcomes [16]. A longitudinal study in Iran demonstrated that higher WrC was associated with incident diabetes in women but not in men, a trend that may indicate sex differences in skeletal frame size and insulin-related bone mass accrual [41].

WCR increased significantly across obesity classes in both sexes in our study; however, it did not independently predict T2DM. As an index combining central adiposity and peripheral muscle mass, WCR has shown predictive value for T2DM in other studies [42]. In a population-based Indian cohort, WHR and WC were stronger predictors of T2DM than BMI or overall body fat measures [43]. These inter-study variations likely reflect differences in population structure, ethnicity, sex distribution, and sample size.

Our study has several limitations. The cross-sectional design does not allow conclusions about causality between anthropometric indices and T2DM. Participants were recruited from endocrinology outpatient clinics, which may introduce selection bias and limit the generalizability of the findings. Although anthropometric measurements were obtained using standardized procedures, some degree of inter-observer variation cannot be excluded. Lifestyle factors such as alcohol intake and smoking were recorded only as binary variables, and the absence of quantitative information limits evaluation of potential dose–response relationships.

Direct assessment of body composition, for example using dual-energy X-ray absorptiometry, was not performed and therefore anthropometric measures could not be validated against more precise measures of fat distribution. Diabetes status was based on ADA criteria or current use of glucose-lowering medication or prior physician diagnosis, which introduces the possibility of misclassification. However, GLP-1-based therapies used for weight management, such as semaglutide and tirzepatide, were approved in India only in 2025, after completion of the study period, and are therefore unlikely to have influenced inclusion.

Information on glucose-lowering medications was not included in the analysis. Different classes of glucose-lowering drugs may influence body weight and fat distribution. GLP-1 receptor agonists and SGLT2i are associated with weight reduction, whereas insulin and sulfonylureas may promote weight gain, thereby influencing anthropometric measurements. In addition, the study population likely included heterogeneous forms of type 2 diabetes, and factors such as disease duration and the presence of complications were not evaluated. Reliance on physician diagnosis may also have influenced the ascertainment and classification of diabetes type.

Despite these limitations, simple anthropometric measures such as NC and NHR have practical value in routine clinical settings. They are easy to obtain, inexpensive, noninvasive, and are less influenced by factors such as posture or recent food intake. These features make them suitable for use in primary care and community-based screening programs, particularly in low- and middle-income settings. Incorporating such indices into existing risk assessment approaches may improve identification of individuals at increased metabolic risk. Future studies with longitudinal follow-up will be important to determine whether these measures predict incident T2DM and related metabolic outcomes. Establishing sex-specific and ethnicity-appropriate cutoffs will also be necessary before wider clinical implementation.