Wrist circumference (WrC) in early gestation: A novel anthropometric predictor of gestational diabetes mellitus (GDM) in Asian Indian women

ABSTRACT

Objective:

To investigate the association between early gestational Wrist circumference (WrC) and risk of gestational diabetes mellitus (GDM) in Indian pregnant women.

Materials and Methods:

In this prospective case-control study, pregnant women of 18–40 years at 6–14 weeks of gestation, were subjected to anthropometric and biochemical evaluations including wrist circumference, waist and hip circumference, waist: hip ratio, fasting plasma glucose, fasting insulin, lipid profiles, HbA1c. Insulin sensitivity was calculated by HOMA IR, QUICKI and McAuley’s index. Patients were followed up to 24–28 weeks of gestation with 75 g-OGTT to diagnose GDM.

Results:

After initial screening of 232 patients, 173 patients were included in the study. Based on the oral glucose tolerance test (OGTT) at 24–28 weeks of gestation, patients were divided into GDM group (n = 32) and NGT (n = 141). Mean WrC in the dominant hand was higher in GDM patients compared to NGT group (15.86 ± 1.99 cm vs. 15.33 ± 2.18, P = 0.006) with similar observations in the non-dominant hand (16.1 ± 2.26 cm vs. 15.6 ± 1.98, P = 0.001). On regression analysis, non-dominant WrC, WHR and WC emerged to be independent predictors for GDM in the entire cohort [OR: 1.86 (1.62–2.06), OR: 1.48 (1.19–1.82) and OR: 1.23 (1.01–1.45) for WrC, WHR and WC respectively, (P_all < 0.001). However, dominant WrC failed to prove an independent predictor of GDM in this cohort (P = 0.112). The non-dominant WrC of >16.1 cm demonstrated 79% sensitivity and 68% specificity to predict onset of GDM (AUC 0.765, 95% CI 0.62–0.89, P = 0.001), which was superior to established measures of WC and WHR.

Conclusion:

Wrist circumference in non-dominant hand measured at early gestation could potentially predict the onset of GDM in Asian Indian pregnant women.

Introduction

Gestational diabetes mellitus (GDM), a frequent metabolic complication, is considered as any degree of hyperglycaemia during pregnancy.[1] The complexity of GDM lies in its pathophysiology, as well as in variable diagnostic criteria. Characteristically, both insulin resistance and beta cell defects have been documented in GDM, possibly due to altered metabolic derangements during pregnancy.[2] Nevertheless, GDM poses significant risk to both mother and fetus including development of hypertensive disorders during pregnancy, excessive fetal adiposity, and growth. Moreover, development of GDM is associated with increased risk of diabetes, obesity and early cardiovascular diseases in both mother and offspring. Globally, the prevalence of GDM is widely variable, between 1% to >30%.[3] Reportedly, in India, the prevalence ranges from 7% to 16%.[4,5] There are multiple established risk factors for GDM including genetic factors, pre-pregnancy obesity, physical inactivity, unhealthy diet, maternal age, family history of diabetes and endocrine disruptors.[3] Considering the adverse outcome associated with GDM, patients often need frequent antenatal visits and intensive blood glucose monitoring, which put significant stress on resource limited health care systems, particularly in India. Hence, it is imperative to detect at-risk patients for GDM during early gestations.

Over the years, several studies have documented significant correlations between various anthropometric parameters and the risk of metabolic syndrome, thus, highlighting the importance of adiposity, specifically the distribution of adipose tissue, in the development of adverse cardio-metabolic outcome.[6,7,8] Considering the inherent limitations of traditional measures like body mass index (BMI), novel anthropometric measures have gained increased importance in recent times. These parameters are simple, easy to perform in the out-patient clinic and have shown less observer-variability. One such measure is the wrist circumference (WrC), which has been reported as a surrogate marker of adipocyte dysfunctions and in turn associated with risk of metabolic syndrome.[9,10] GDM shares multiple risk factors with metabolic syndrome and demonstrates diverse metabolic dysfunctions. Thus, we hypothesized that novel anthropometric measures like WrC could be useful as early markers of GDM. Given the paucity of literature in this regard, we conducted a prospective case-control study to investigate investigated whether the first trimester WrC Asian Indian pregnant women was associated with subsequent development of GDM in the second trimester (24^th–28^th week).

Materials and Methods

Study design

We conducted a prospective case–control study over a period of 2 years amongst pregnant women presenting to the outpatient departments for their antenatal visit during 6–14 weeks of their gestation [Figure 1]. The study was pursued in accordance with the Declaration of Helsinki and appropriate ethical clearance was obtained from the Institutional Ethics Committee of the Chittaranjan Health Institute, West Bengal (IEC No: CHI74/11/2022 After consideration of predetermined inclusion and exclusion criteria and obtaining informed consent, n = 232 pregnant women, aged between 18–40 years, were initially screened for the study and subjected to estimation of fasting and post prandial plasma glucose levels and HbA1c estimations. Patients who were detected with GDM or overt diabetes in the first trimester based on the IADPSG criteria[11] were excluded from the study (n = 39). The remaining women with normal glucose tolerance (n = 193) were included in the study and followed up at 24–28 weeks of gestation with a 2-hour, 75 g Oral glucose tolerance test (OGTT). During the follow-up period twenty patients were lost and remaining patients (n = 173) were included for the final evaluation. Diagnosis of GDM, based on the IADPSG criteria, was confirmed if any one of either fasting plasma glucose ≥92 mg/dL, post-OGTT 1-hour glucose ≥180 mg/dL or 2-hour glucose ≥153 mg/dL were present.[11] If a participant received a diagnosis of GDM at 24–28 weeks of gestation by OGTT, she was assigned to the GDM group (n = 32; 18%). Otherwise, she was considered part of the normal glucose tolerance group (NGT) (n = 141).

Figure 1.

Flowchart describing the study design and recruitment

Study participants

Pregnant women who presented between 6–14 weeks of gestation were included in the study. The inclusion criteria were as follows: (1) Natural conception, singleton pregnancy, (2) met the diagnostic criteria for GDM and no other pregnancy-related diseases, and (3) not taking any medications for lipid/glycaemic/thyroid-related diseases before and during pregnancy. Exclusion criteria were: (1) Multiple pregnancies, (2) history of thyroid -related disorders, (3) history of pre-diabetes or diabetes mellitus, (4) history of chronic diseases including cardiovascular, hepatic or renal dysfunction, psychiatric illness, alcoholism or HIV[4] use of drugs that could affect thyroid function tests, glycaemic or lipid parameters, and (5) miscarriage[6] any deformity of wrist joints. We used the same exclusion criteria to identify patients without GDM as the control group.

Study parameters

Baseline characteristics and obstetrical history were collected at first visit, as also anthropometric measurements were obtained. Waist and hip circumference were measured according to WHO STEPS protocol.[12] Waist circumference was measured at the midpoint between lower margin of palpable rib and the top of iliac crest in parallel with the floor at the end of the expiration. Hip circumference was measured at the widest part of the buttocks.[13] Wrist circumference (WrC) was measured in both dominant and non-dominant hands in a seated position using a tape measure positioned over Lister’s tubercle of the distal radius and over the distal ulna.[14] Blood pressure was measured in seated position after 10 min rest. McAuley’s insulin sensitivity index and Quantitative insulin sensitivity check index (QUICKI index) were calculated according to established formula.[15,16] The homeostasis model assessment of insulin resistance (HOMA-IR) was carried out using the following formula: HOMA-IR = [ Fasting glucose (nmol/L) × Fasting insulin (μU/mL)/22.5].[17] Biochemical measurements performed at first visit included a fasting blood test to measure fasting plasma glucose, fasting insulin, lipid profiles (total cholesterol, HDL and LDL cholesterol, triglycerides) and HbA1c.[17] HbA1c was measured by high performance liquid chromatography in Bio-Rad D10^® (CV <10%). Fasting plasma glucose was measured by glucose oxidase-peroxidase (GOD-POD) method and fasting insulin was measured in IMMULITE 1000^®, Siemens, Germany (CV <5%). Lipid parameters were measured in Cobas Elecsys^®, Rosche. Between 24-28 weeks of gestation, all subjects underwent a 2-hour, 75 g OGTT. Patients were asked to attend the clinic with at least 8 hour fasting condition and allowed to drink 75 g anhydrous glucose (G75^®). Subsequently blood samples were drawn at 0, 60, and 120 min.

Statistical analysis

Groups were compared using unpaired t-test or Mann-Whitney’s U test for quantitative variables and Chi-square test, with Fisher’s correction where appropriate, for categorical variables. A P value ≤ 0.05 was considered significant. Correlation co-efficient is expressed using Spearman’s rho for parameters with significant linear correlation. Binomial logistic regression was done to identify the independent predictors of gestational diabetes mellitus outcomes in two steps—univariate analysis (step 1) followed by multivariate analysis using independent variables that were found to be significant (P < 0.05) or suggestive of significance (P < 0.100) in step 1. Multicollinearity was tested between the multiple predictors and parameters with VIF more than five were not included in the same logistic regression model. All the logistic models were validated using the Hosmer-Lemeshow test ROC curves were constructed and analysed for differences in AUROC for the different parameters which could predict risk of gestational diabetes mellitus. Statistical analysis was performed using SPSS version 22.0 (SPSS Inc., Chicago, Illinois).

Results

In this prospective study, no significant differences were observed in relation to mean age and period of gestations at first visit between GDM and NGT groups (P = 0.61 and P = 0.15 respectively) [Table 1]. Patients with GDM had significantly higher pregestational BMI compared to NGT patients (P = 0.33). The proportion of primipara in GDM and NGT groups were 53.5% and 41.2% respectively, while no significant difference was observed in proportions of multipara (P > 0.05). Family history of diabetes and history of GDM were significantly higher in GDM patients compared to NGT patients (all P < 0.05). In addition, no significant differences were observed in relation to FPG, triglyceride, LDL cholesterol and HDL cholesterol levels. Fasting insulin and HbA1c at baseline were significantly higher in GDM group (all P < 0.05) compared to NGT group. Though calculated HOMA2 IR, Macaulay’s Index and QUICKI Index were significantly higher amongst patients with GDM, no significant difference was observed in relation to the presence of acanthosis nigricans (37.5% vs. 12%, P = 0.06). Regarding the anthropometric parameters, WC and WHR were significantly higher in GDM group, though hip circumference was not significantly different between the two groups (P = 0.09). The mean WrC in the dominant hand (15.86 ± 1.99 cm vs. 15.33 ± 2.18, P = 0.006) was higher in GDM patients compared to NGT group., with the non-dominant hand WrC too showing similar findings (16.1 ± 2.26 cm vs. 15.6 ± 1.98, P = 0.001).

Table 1.

Baseline characteristics of the study population (n=179)

Variables	GDM (n=32) mean±SD	NGT (n=141) mean±SD	P
Age (years)	30.4±6.5	29.3±8.2	0.61
Period of gestation at first visit (weeks)	13.8±3.5	12.9±2.7	0.15
Pregestational body mass index (BMI) (kg/m2)	26.8±5.5	24.1±4.3	0.04*
Gravida [n (%)]	Primi-17 (53.5)	Primi-58 (41.2)	0.33
	Multi-15 (46.5)	Multi-83 (58.8)
Family h/o of diabetes [n (%)]	21 (66)	41 (29)	0.04*
Past h/o of GDM [n (%)]	7 (21.4)	12 (8.8)	0.04*
FPG at diagnosis (mg/dL)	101.6±8.6	81.8±6.7	0.001*
Fasting insulin (μU/L)	13.88±7.12	7.12±3.6	0.03*
HbA1c at diagnosis (%)	5.4±0.5	5.1±0.3	0.02*
Haemoglobin (gm/dL)	12.1±0.9	11.4±1.7	0.07
LDL-C (mg/dL)	143.5±50.8	134.2±49.82	0.11
HDL-C (mg/dL)	45.66±13.8	48.82±19.5	0.21
Triglyceride (mg/dL)	179.6±65.8	171.1±62.2	0.09
HOMA-IR	1.56±0.79	1.18±0.55	0.02*
QUICKI	0.35±0.14	0.30±0.11	0.01*
McAulay’s index (Median, IQR)	5.15 (4.35-5.98)	5.83 (5.06-6.52)	0.02*
Serum creatinine (mg/dL)	0.51±0.14	0.48±0.07	0.64
Acanthosis nigricans [n (%)]	12 (37.5%)	9 (12%)	0.06
Waist circumference (cm)	94.8±9.5	81.7±11.9	0.007*
Hip circumference (cm)	96.1±10.76	93.4±10.8	0.09
Waist-Hip ratio	0.95±0.16	0.81±0.10	0.004*
Dominant wrist circumference (cm) (WrC)	15.86±1.99	15.33±2.18	0.006*
Non-dominant wrist circumference (cm) (WrC)	16.1±2.26	15.6±1.98	0.001*

*P<0.05-considered as statistically significant. HbA1c=glycosylated haemoglobin, FPG=fasting plasma glucose, HOMA IR=homeostatic model assessment for insulin resistance, LDL-C=low density lipoprotein cholesterol, HDL-C=high density lipoprotein cholesterol

Both dominant and non-dominant WrC showed significant positive correlations with SBP, DBP fasting plasma glucose (FPG) and plasma insulin [Table 2]. Though HOMA-IR and QUICKI showed positive correlations with WrC, no significant correlation was observed with Macaulay’s index (r = 0.254, 0.178, P > 0.05). Similarly, no significant correlations were observed between WrC (dominant and non-dominant) and lipid parameters and HbA1c. However, WrC showed significant positive correlations with other anthropometric measures including WC, WHR and hip circumference (P < 0.05).

Table 2.

Correlation analysis between different variables and wrist circumference (WrC)

Variables	Non-dominant WrC		Dominant WrC
	r	P	r	P
Age (years)	0.236	0.04*	0.243	0.05
Systolic BP (mm Hg)	0.289	0.04*	0.211	0.04*
Diastolic BP (mm Hg)	0.265	0.04*	0.183	0.10
Fasting plasma glucose (mg/dL)	0.302	0.03*	0.211	0.04*
Post-OGTT 1 hour glucose (mg/dl)	0.344	0.02*	0.213	0.03*
Post-OGTT 2-hour glucose (mg/dl)	0.399	0.01*	0.289	0.02*
HbA1c (%)	0.086	0.29	0.075	0.34
Waist circumference (cm)	0.396	0.001*	0.312	0.02*
Hip circumference (cm)	0.283	0.007*	0.256	0.02*
Waist-hip ratio	0.412	0.001*	0.321	0.01*
Fasting insulin (pmol/L)	0.311	0.02*	0.205	0.04*
HOMA-IR	0.456	0.001*	0.308	0.01*
QUICKI	0.378	0.01*	0.277	0.03*
McAaulay’s index	0.254	0.03*	0.178	0.32
Total cholesterol	0.079	0.18	0.101	0.11
LDL-cholesterol	0.098	0.15	0.055	0.26
HDL-cholesterol	–0.199	0.11	–0.187	0.24
Triglycerides	0.202	0.04*	0.198	0.08

*P<0.05-considered as statistically significant. LDL=low density cholesterol, HDL=high density cholesterol

Due to strong collinearity between themselves, dominant and non-dominant WrC, WHR, WC and pre-pregnancy BMI could not be used as continuous independent variables in the same model. We thus analysed them separately in the same model for predicting GDM adjusted for maternal age, parity, gestational age, family history of diabetes and pre-pregnancy BMI as other independent variables. On regression analysis, non-dominant WrC, WHR and WC emerged to be independent predictors for GDM in the entire cohort [OR: 1.86 (1.62–2.06), OR: 1.48 (1.19–1.82) and OR: 1.23 (1.01–1.45) for WrC, WHR and WC respectively, (P_all < 0.001). However, dominant WrC failed to prove to be an independent predictor of GDM in this cohort (P = 0.112). Subsequently, we designed AUROC curves for the three significant predictors and compared them. The non-dominant WrC of > 16.1 cm demonstrated 79% sensitivity and 68% specificity to predict onset of GDM (AUC 0.765, 95% CI 0.62-0.89, P = 0.001) [Table 3, Supplementary Figure 1 ^{(55.9KB, tif)} ], whereas WHR of 0.84 showed 75% sensitivity and 66% specificity to predict the onset of GDM (AUC 0.74, 95% CI 0.58, 0.90, P = 0.01) [Table 3, Supplementary Figure 2 ^{(59.8KB, tif)} ]. The non-dominant WrC had the greater AUROC in comparison to WHR (P = 0.04) and WC (AUC 0.55, 95% CI 0.33, 0.67) P = 0.01) [Table 3].

Table 3.

ROC analysis of WrC, WHR and WC as early marker of GDM

Anthropometric indices	Area under curve	95% CI	SE	Cut-off value	Sensitivity (%)	Specificity (%)	P
WrC (non-dominant)	0.765	0.62, 0.89	0.05	16.1	79%	68%	0.001
WHR	0.740	0.58, 0.90	0.07	0.848	75%	66%	0.01
WC	0.551	0.33, 0.67	0.12	84.7	67%	60%	0.04

*P<0.05 considered as statistically significant. WrC=wrist circumference, WHR=waist-hip ratio, WC=waist circumference

Discussion

In this prospective study of Indian women, we investigated whether WrC in early pregnancy might predict GDM, with the greater aim of enabling interventions to reduce the incidence and consequences of GDM. Including n = 179 gestational women, aged 18–40 years, we observed that wrist circumference (WrC) in non-dominant hand measured at early gestation could predict the onset of GDM, even after adjustment of other covariates like maternal age, parity, gestational age, family history of diabetes and pre-pregnancy BMI. Though not specifically designed to look at WrC as a marker of GDM, a previous study done by White et al.[18] found WrC in early gestation to be significantly higher in GDM patients than non-GDM patients (P = 0.02). Interestingly, another study done by the same group of authors but with obese women (median BMI 35 kg/m², IQR 32.8–38.5) and measurement of WrC at second trimester (mean 17 weeks of gestation) did not observe significant differences in WrC between GDM and non-GDM groups.[19]

Evidences suggest that WrC is significantly associated with insulin resistance, blood pressure and lipid parameters.[9,20,21] In a prospective study, Jahangiri Noudeh et al.[22] observed that 1-SD increase in subject’s WrC was associated with 17% and 31% increase in diabetes incidence in males and females respectively. However, predictability of WrC was lost when BMI and WC were considered, particularly in males. Though, we observed significant correlations with blood pressure (SBP, DBP) and insulin resistance indices (HOMA IR, QUICKI index), no correlation was demonstrated between WrC and lipid parameters like triglyceride, LDL cholesterol and HDL cholesterol levels. This is possibly because most of the earlier studies were done in non-pregnant population, while for our study several regulatory factors including hormones play important role in the lipid metabolism in pregnant women.

As discussed earlier, higher BMI (>25 kg/m²) is associated with significant risk of development of GDM.[2] However, BMI per se has several limitations and more so in gestation. Firstly, it does not fully capture the cardiometabolic risk. Second, BMI has low sensitivity (~50%) and high specificity (~90%) to detect excess adiposity which can often misclassify patients with excess fat mass as non-obese. Thirdly, BMI fails to differentiate between lean and fat mass and fails to point towards the distribution of body fat, the latter seeming to be more important for development of atherosclerosis.[23] Furthermore, due to substantial variability of visceral adipose tissue, frequent discordance is noted between BMI and WC.[24] Other anthropometric parameters including higher WC and WHR have traditionally shown significant correlations with risk of metabolic syndrome and insulin resistance in non-pregnant subjects.[25,26] With progressive change in uterine volumes during pregnancy, waist and hip circumferences will get altered leading to erroneous interpretations. Moreover, measurements of these parameters often need removal of clothing and are frequently affected by respiration and postprandial state. In contrast, the wrist circumference is a simple, easy-to-do measurement of skeletal frame size.[27] It likely manifests deposition of subcutaneous fat and weakly correlated with total body fat in women.[28] Importantly, WrC showed positive correlations with metabolic syndrome irrespective of methods of measurements.[29] It is possible that increased in girth of wrist may be due to effects of hyperinsulinemia.[30] Furthermore, observed gender differences between risk of diabetes and WrC could possibly be due to role of sex-steroid hormones and their influence on bone metabolism and glucose homeostasis.[31,32]

To the best of our knowledge, this is the first study to evaluate the association of first-trimester wrist circumference with risk of GDM in Indian pregnant women. The prospective nature of our study and application of uniform criteria to diagnose GDM are added strengths. In addition, the cut-offs derived for non-dominant hand WrC can prove to be clinically useful in early identification of pregnant women at risk of GDM, particularly in resource-limited settings like India.

However, there are some limitations to our study. A larger sample size and multicentric study design would have further enhanced the findings. Moreover, mechanistic links of WrC and risk of GDM has not been established in this study and needs further evaluation.

Conclusion

WrC of the non-dominant hand, measured at early gestation (6–14 weeks), could potentially predict the onset of GDM at 24–28 weeks in Asian Indian pregnant women. Achieving a sensitivity of nearly 80%, WrC>16.1 cm could prove to be a simpler, cost-effective yet superior surrogate for risk of GDM than the traditionally established anthropometric measures.

Data availability statement

The data underlying this article cannot be shared publicly due to the privacy of individuals that participated in the study. The data will be shared on reasonable request to the corresponding author

Relevance of the Study

• What is already known on this topic: Gestational diabetes mellitus (GDM) is a globally relevant public health problem which puts significant stress on the pregnant mother, as also the healthcare system. Early prediction of risk of developing GDM is imperative, and there is a paucity of suitable anthropometric biomarkers in this respect.

• What this study adds: Wrist circumference has been increasingly linked to adiposity and metabolic syndrome, though it is exact association with GDM has remained unexplored. Our study posits wrist circumference of the non-dominant hand measured in early pregnancy as a potential biomarker of development of GDM, and has superior predictive ability to existing anthropometric measures.

• How this study might affect research, practice, or policy: Given the significant burden of GDM in the community, early prediction would help triaging medical resources and instituting appropriate therapeutic interventions to prevent development of GDM. This could be hugely significant in resource-limited settings, with wrist circumference being an easily reproducible, cost-effective measurement.

Conflict of interest

The authors here by declare no conflicts of interest.

Supplementary File

Supplementary Figure 1

ROC curve for non-dominant wrist circumference (WrC) in first trimester of pregnancy as predictor of GDM (24–28 weeks)

Supplementary Figure 2

ROC curve for waist-hip ratio (WHR) in first trimester of pregnancy as predictor of GDM (24–28 weeks)

Funding Statement

Nil.