Risk assessment of type 2 diabetes mellitus using Indian diabetes risk score among females aged 30 years and above in urban Delhi

INTRODUCTION

Type 2 diabetes mellitus (T2DM) is highly prevalent in India and is rising alarmingly. With most cases being concealed (undiagnosed), diabetes demonstrates the best example of the Iceberg phenomenon. The clinical, social, and economic impact of the condition can be lessened by detecting diabetes early with the right screening techniques, especially in people with higher risk.^[1]

India had a 9.3% prevalence of diabetes.^[2] In India, 10.2% of females between the ages of 18 and 69 had diabetes.^[3] East Delhi had a diabetes prevalence of 18.3% (known as 10.8% and recently discovered as 7.5%).^[4]

In the 21^st century, non-communicable diseases have grown in importance as a major public health issue in India due to epidemiological changes. Diabetes being a crucial disease, considered a “disease of urbanization.” While T2DM is becoming more frequent among urban Indian adults, it is important to remember that undiagnosed diabetes is still common.^[5-7]

Women playing various tasks at home and in the community are more likely to have more specific risk factors, such as physical inactivity and central obesity, which increase the chance of developing diabetes. Gestational diabetes mellitus (GDM) represents high blood glucose levels in pregnant women. GDM is a potential risk factor for poor perinatal consequences, and long-term danger to children of developing glucose intolerance and obesity. GDM is strongly linked to hypertensive adversities during pregnancy and a high risk of T2DM afterward.^[8]

Studies specifically focusing on diabetes risk among females in urban areas using the Indian diabetes risk score (IDRS) are scarce. Most of the studies were concerned with the urban adult population. Data collection was done during the forenoon when most of the adult males were not present in the house probably due to occupation. Therefore, an effort was made to conduct this study, particularly among urban Delhi women.

IDRS was created at the Madras Diabetes Research Foundation by Mohan et al. It is a verified tool for locating people with a high risk of acquiring T2DM. It consists of two non-modifiable risk factors, age, and family history, and two modifiable risk factors, abdominal obesity, and physical activity.^[9] Details of IDRS are shown in [Table 1].

Low, moderate, and high risk of diabetes are determined by IDRS scores of <30, 30–59, and >60, respectively.

According to the population-based survey, the adult population ≥30 years is considered for screening for diabetes.^[10] By assessing the risk of T2DM among females aged ≥30 years, proper intervention can be done on time related to lifestyle. Thus, it is important to detect this large number of participants with undiagnosed T2DM in India and start early initiation of treatment.

The objectives of the study were to:

1. To determine the risk of T2DM among females aged 30 years and more using IDRS in an urban resettlement colony, Dakshinpuri, Delhi

2. To study selected factors associated with the risk of T2DM among females aged 30 years and more among the study participants.

MATERIALS AND METHODS

This was a community-based cross-sectional study. Due to logistic reasons, two blocks of Dakshinpuri extension, New Delhi were purposively selected. All females aged 30 years and more without diagnosed T2DM, residing in that area at least for the past 1 year considered to be included. Those who were already diagnosed with diabetes were excluded from the study.

RESULTS

A total of 626 women aged 30 years and more without diagnosed T2DM were included in the study. IDRS categorization revealed 99 (15.8%), 279 (44.6%), and 248 (39.6%) participants in low-, moderate-, and high-risk categories, respectively. [Table 2] is showing the baseline features of the participants.

More than half (104; 56.5%) of the illiterate study participants were at high risk and 27 (65.9%) graduate participants were at moderate risk. There was a high statistically significant association between low education status with high-risk status (P < 0.0001).

The mean (standard deviation [SD]) age of the participants was 43.9 (12.2) years. More the three-quarter (143; 76.1%) of participants aged >50 years were at high risk. More than half and two-thirds (125; 69.1%) of the participants (136; 52.9%) aged 35–49 years and 30–34 years were at moderate risk, respectively. The association between risk status and age group was highly significant (P < 0.0001).

Among the participants with a history of one parent diabetic, half (52, 50.0%) of them considered as high risk. Of participants with both diabetic parents, the majority (4, 80.0%) of them were considered as high risk. The link between diabetic family history was statistically significant with risk status (P = 0.037).

More than half (129; 58.6%) of participants with a history of no exercise and sedentary activities at home/work were considered as high risk. Physical there was a significant association between physical activity and risk status (P < 0.0001).

More than 2/3 (114, 67.8%) of participants with waist circumference (>90 cm) were considered as high risk. Statistically significance (P < 0.0001) was found between risk status and waist circumference.

Almost half (74; 44.9%) of the participants with any comorbidities were at high risk which was significantly associated (P < 0.0001). Almost half (18.46.2%) of the participants with tobacco usage history were considered as high risk, where statistical significance (P < 0.0001) association was seen. 40% of the study participants with a history of alcohol consumption were considered as high risk, where statistical significance (P = 0.867) was not seen.

Univariate logistic regression for the high-risk participants [Table 3] showed that minimum education up to graduation had 87% less chance of having high-risk status (odds ratio [OR]: 0.13; 95% confidence interval [CI]: 0.05–0.33), where statistical significance (P < 0.0001) was seen. Participants aged ≥50 years had 10.07 times more odds of having high-risk status with respect to the non-high-risk group, where statistical significance (P < 0.0001) was seen. Diabetic family history in at least one parent had 1.79 times more odds of having high-risk status considered statistical significance (P < 0.0001). Participants performing regular exercise and/or strenuous work had 71% less chance of having high-risk status, where statistical significance (P < 0.0001) was seen. Participants with waist circumference >80 cm had 8.18 times more odds of having high-risk status where statistical significance (P < 0.0001) was seen. Participants with any comorbidity had 34% more chance of having high-risk status which was not significant statistically (P = 0.111).

On multivariable logistic regression for high-risk group [Table 3], age 50 years or more (OR: 13.2; 95% CI: 7.57–23.02; P < 0.0001), family history of at least one parent diabetic (OR: 5.5; 95% CI: 3.03–9.98; P < 0.0001), participants performing regular exercise and/or strenuous work (OR: 0.22; 95% CI: 0.14–0.36: P < 0.0001), and participants having waist circumference >80 cm (OR:12.56, 95% CI: 6.59–23.91; P < 0.0001) had a highly statistically significant association. Significant statistical associations (P < 0.05) were seen between minimum education (high school; OR 0.45; 95% CI: 0.22–0.94: P = 0.033 and graduate; OR: 0.16; 95% CI: 0.05– 0.48; P = 0.001) and high-risk group.

FBS was collected from consenting participants from high-risk group, non-high-risk group using simple random sampling. Overall, the prevalence of T2DM was 22.6% (16.5– 28.7) among all the participants.

The sensitivity and specificity among study participants by dividing the IRDS score into 2 categories is shown in [Table 4].

[Table 5] provides the sensitivity and specificity of different cutoffs for IDRS. IDRS >60 had optimum sensitivity (67.5%) and specificity (41.6%) for identifying diabetes. The receiver operating characteristic (ROC) curve, made for validation of IDRS to detect diabetes using comparison against FBS values, provides an area (area under the curve [AUC]) of 0.6 (95% confidence interval [CI]: 0.47–0.68) under the curve (P < 0.001, denoting the sufficient level of accuracy) [Figure 1].

Close

Figure 1:

ROC curve showing performance of IDRS. AUC (95% CI) of ROC= 0.6 (0.47–0.68). ROC: Receiver operating characteristic, AUC: area under the curve, IDRS: Indian diabetic risk score.

Export to PPT

DISCUSSION

This study comprised 626 female participants residing in an urban resettlement area, Dakshinpuri near the urban health center, AIIMS, Delhi. This consists of around two-fifths (41.1%) of the participants aged 35–49 years followed by aged >50 years (30.0%) and <30 years (28.9%) with a mean (SD) age of 43.9 (12.2) years. The majority were married (87.1%) and residing in a nuclear family (53.9%). Almost one-third were illiterate (29.5%) followed by educated up to primary (20.5%).

Bala et al.,^[11] in their study, conducted among 150 females from the industrial area in Hyderabad in 2019 found the mean (SD) age to be 35.39 (13.3) years and the majority of females aged group 31–35 years came up with 57.4%. More than 50% were married (52%) residing in a nuclear family (78 almost one-third were educated up to intermediate (30.7%).

A study conducted by Raghavendra et al.^[17] in urban East Delhi (Gazipur) found the majority of women (42.6%) aged 31–40 years with illiteracy among 50%.

The proportion of participants with a high risk of T2DM was 39.6%. Patil and Gothankar published findings that were comparable^[18] at Pune in 2016 (36.55% high-risk group), Mohan et al.^[9] in urban Chennai (43% high risk), Nagarathna et al.^[19] in multiple sites in India in 2020 (40.9% high risk), Bala et al. at^[11] Hyderabad in 2019 (38% high risk), and Sengupta and Bhattacharjya^[20] in Tripura (34.2% high risk).

A relatively lesser proportion of high-risk status was obtained by Gupta et al.^[21] at urban Puducherry (31.2% high risk), Singh et al.^[22] in the assessment of risk among north Indian young medical students (high risk 0.6%), Sahai and Ahuja at.^[23] Gwalior (0% high risk), and Ashok et al.^[24] at multiple sites in India (7% high risk). These variations are probably due to variations in sample size and study settings, the inclusion of younger age groups, the inclusion of both male and female participants, higher literacy rates, increased physical activity, etc.

A relatively higher proportion of high-risk status was found by Sankar et al.^[25] in a semi-urban hospital in southern India (48.5% high risk), Acharya et al.^[26] at Delhi (51.8% high risk), and Nittoori and Wilson^[27] in North Telangana (74.3% high risk). These variations are probably due to variations in sample size and study settings, the inclusion of the elder age group, the inclusion of both male and female participants, higher illiteracy rates, decreased physical activity, etc.

Our study coined that, with the progression of age, the risk for diabetes increases. Studies conducted by Mohan et al.,^[9] Patil and Gothankar,^[18] Singh et al.,^[28] and Menon et al.^[29] found similar results. A high risk of diabetes was observed among participants with at least one diabetic parent in this study. Similar results were found in several studies.^[15,26]

Over the previous years, a sizable section of the working population transitioned from physically demanding agricultural manual labor to less strenuous office labor. Rapid urbanization in India is accompanied by rising obesity rates and a decline in physical activity, which have changed people’s lifestyles, and diets, and transitioned them from manual labor to less physically demanding jobs.^[28] Increasing physical activity has a beneficial effect with a lesser risk of diabetes^.[7,15,18] Waist circumference is an important determinant of the risk of T2DM; various studies have found that waist circumference and undiagnosed diabetes had a significant association, which was similar to the present study results.^[15,18]

In the present study, participants with any comorbidity had a 60% less chance of having a high risk of diabetes, probably due to chance alone.

In this study, IDRS more than equal to 60 had optimum sensitivity (67.5%) and specificity (41.6%) for determining diabetes. A study conducted by Bala et al. produced almost equal findings,^[11] (sensitivity 59.4% and specificity 37.3%), Mohan et al.^[9] (sensitivity 72.5% and specificity 60.1%), Adhikari et al.^[30] (sensitivity 62% and specificity 73%), Sharma et al.^[31] (Sensitivity 72.5% and specificity 60.1%).

At 60 cutoff value, different results were found in the study conducted by Khan et al.^[32] (sensitivity 29.9% and specificity 98.1%), Agarwal et al.^[33] (sensitivity 45.5% and specificity 88%), Taksande et al.^[34] (Sensitivity 97.5% and specificity 81.9%), Dudeja et al.^[35] (sensitivity 95% and specificity 29%), and Sengupta and Bhattacharjya^[20] (sensitivity 83.1% and specificity 82.6%).

Bhadoria et al. found optimum sensitivity and specificity at a level of ≥40, which was unlike from our study.^[36] Kaushal et al. in Shimla, reported optimum specificity and sensitivity as 56.14% and 61.33%, respectively, at IDRS cutoff point ≥70.^[37]

This difference could be described by the variation in eligibility criteria, sample size, training of the investigator, and study setting in various study designs. Our study included only women. There was a difference in the physical activity, denoting the variations in sensitivity and specificity.

The present study reported an AUC (95% CI) of 0.6 (0.47– 0.68) at the IDRS cutoff point ≥60. This value is lower than the study by Mohan et al.^[9] (AUC 0.69: 95% CI 0.66–0.73), Adhikari et al.^[30] (AUC 0.66), Sengupta and Bhattacharjya^[20] (AUC 0.83; 95% CI 0.77–0.88), and Patel et al.^[38] (AUC 0.838). These variations occurred as freshly diagnosed diabetics were included in the above studies except in the study by Sengupta and Bhattacharjya, where both freshly diagnosed diabetics and pre-diabetics were included in the study. Other causes might be differences in inclusion criteria, study settings, presence of trained data collectors, etc. In a study conducted by Barjatya et al.,^[39] at the IDRS cutoff point ≥35, AUC was 0.704 (95% CI 0.52–0.89).

CONCLUSION

For community-based research to identify people at high risk for diabetes, IDRS is a straightforward, non-invasive method. Non-modifiable risk factors, for example, increment in age and family history of diabetes, and modifiable risk factors, for example, lack of physical activity and abdominal obesity found to be the most common factors associated with high diabetes risk. This study also validates that IDRS is an accurate, simple, and efficient method to screen undiagnosed diabetes in the community with public health importance.