Incidence and risk factors for dysglycaemia in Asian-Indians: a 10-year population-based prospective cohort study

Abstract

Background

Diabetes prevalence estimates suggest an increasing trend in South-East Asia region, but studies on its incidence are limited. The current study aims to estimate the incidence of type 2 diabetes and pre-diabetes in a population-based cohort from India.

Methods

A subset of Chandigarh Urban Diabetes Study cohort (n=1878) with normoglycaemia or pre-diabetes at baseline was prospectively followed after a median of 11 (0.5–11) years. Diabetes and pre-diabetes were diagnosed as per WHO guidelines. The incidence with 95% CI was calculated in 1000 person-years and Cox proportional hazard model was used to find the association between the risk factors and progression to pre-diabetes and diabetes.

Results

The incidence of diabetes, pre-diabetes and dysglycaemia (either pre-diabetes or diabetes) was 21.6 (17.8–26.1), 18.8 (14.8–23.4) and 31.7 (26.5–37.6) per 1000 person-years, respectively. Age (HR 1.02, 95% CI 1.01 to 1.04), family history of diabetes (HR 1.56, 95% CI 1.09 to 2.25) and sedentary lifestyle (HR 1.51, 95% CI 1.05 to 2.17) predicted conversion from normoglycaemia to dysglycaemia, while obesity (HR 2.43, 95% CI 1.21 to 4.89) predicted conversion from pre-diabetes to diabetes.

Conclusion

A high incidence of diabetes and pre-diabetes in Asian-Indians suggests a faster conversion rate to dysglycaemia, which is partly explained by sedentary lifestyle and consequent obesity in these individuals. The high incidence rates call for a pressing need for public health interventions targeting modifiable risk factors.

Introduction

The prevalence of type 2 diabetes mellitus (T2DM) continues to grow at an alarming rate and according to International Diabetes Federation (IDF), there will be over 700 million people with diabetes in the world by 2045. [1] Of these, 153 million will be in South East Asian region (SEAR) alone with maximum contribution by India. [1] While prevalence estimates are important, it is equally important to know the incidence of diabetes to understand its secular trend and to assess the effect of public health measures. Very few studies from SEAR have focused on the incidence of diabetes and tried to ascertain the predictors of conversion from normal glycaemic tolerance (NGT) to dysglycaemia (ie, pre-diabetes and diabetes). Among those available, majority are from southern India and the estimated incidence of diabetes in them has varied from 20.2 to 33.1 per 1000 person-years. [2–6] The current study aimed to estimate the incidence of diabetes and pre-diabetes and the predictors of conversion from NGT state to a dysglycaemic state in a representative follow-up cohort of population of Chandigarh.

Methods

Setting

India is one of the member countries of WHO SEAR and belongs to a low to middle income country with a heterogeneous income and literacy structure. Its administrative structure comprises of 28 states and 8 union territories.

Study design and participants

Chandigarh Urban Diabetes Study (CUDS) was a community-based cross-sectional study carried out by house to house visit from April 2008 to August 2009, to estimate the prevalence of diabetes. [7, 8] In brief, a multistage cluster random sampling was used to enrol a total of 2227 individuals.

The flow diagram of study participants is shown in figure 1. After excluding individuals who had diabetes at baseline (n=349), the remaining ones constituted the cohort for follow-up (n=1878), and were assessed after a mean of 10.27 years from April 2019 to March 2020. Inclusion criteria for the current follow-up were: (1) participants in the CUDS; (2) those who had NGT or pre-diabetes at baseline survey and (3) availability of informed consent. Exclusion criteria for the current follow-up were: (1) person who were sick and cannot undergo OGTT, (2) pregnant women and (3) person on steroid medications or any other medication likely to cause dysglycaemia. In these eligible individuals, 981 (52.23%) were not accessible as they have either migrated, were not reachable after three attempts or were not traceable. Among the remaining 897 individuals that were accessible, 424 refused to give consent and 26 had died. The reasons for refusal were unwillingness for intravenous puncture, to undergo OGTT, apprehension of contracting SARS-CoV-2 and non-availability in early morning. Of the 26 cases that had died, verbal autopsy was available in sixteen. Finally, a total of 473 individuals completed the study, giving a response rate of 52.7% (figure 1).

Survey procedure

The survey was conducted by house-to-house visit. Individuals were approached a day prior to the main visit and their eligibility was assessed. The next day information was collected pertaining to sociodemographic parameters, non-communicable disease risk factors, physical measurements, glycaemic status ascertainment and blood samples were collected for biochemical testing.

Assessment of sociodemographic parameters and non-communicable disease risk factors

Two interviewers were trained and collected data using a standard case record form. The sociodemographic parameters collected were age, sex, marital status and socioeconomic status using modified Kuppuswamy scale. [9] The non-communicable disease risk factors assessed were: (1) family history of diabetes, hypertension (HTN), coronary artery disease and cerebrovascular accident; (2) physical activity; (3) smoking and (4) alcohol intake. Physical activity was assessed by using Global Physical Activity Questionnaire. [10] Total metabolic equivalents/week (MET/week) were calculated and individuals grouped as sedentary (<600 MET/week) and active (≥600 MET/week). Smokers were defined as those who are currently consuming or have consumed tobacco in the last ten years and alcohol use was defined as current alcohol consumption.

Physical measurements

Physical measurements included anthropometry and blood pressure measurement and was performed by two trained personnel (PK and PH). Height was measured in centimetres using standard stadiometer to the nearest 0.1 cm. Weight was measured in kilogram using a digital weighing machine (Omron HN 283, Omron, Kyoto, Japan) to the nearest of 0.1 kg. Waist circumference (WC) was measured at the midpoint between lower border of rib cage and upper border of iliac crest using a metallic tape. Blood pressure was measured using a digital apparatus (Omron HEM-8712, Omron). A mean of three readings was taken for all the physical measurement. Obesity was defined as a body mass index (BMI) of ≥25 kg/m² and overweight as 23–24.9 kg/m². [11] Central obesity was defined as WC of ≥90 cm in males and ≥80 cm in females. [11] HTN was defined as current use of antihypertensive medications, and/or a systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg. [12]

Glycaemic status

Glycaemic status was assessed by a 75 g oral glucose tolerance test (OGTT). Optium Free Style Neo (Abbott, Chicago, USA) glucometer, based on glucose dehydrogenase method was used to measure capillary blood glucose. In every 10th case, plasma glucose was estimated by using the glucose oxidase method (Autoanlayzer 902; Hitachi, Tokyo, Japan). In case of death of a person, verbal autopsy was performed by analysing medical records and death certificates, when available to assess the presence of diabetes prior to death. In individuals already on medications for treatment of diabetes, the duration of the use of same was ascertained by personal interview and reviewing medical records. Diabetes and pre-diabetes were defined as per WHO guidelines. [13]

Biochemical measurements

Biochemical measurements included glycosylated haemoglobin (HbA1c) and fasting lipid profile. HbA1c was estimated by Bio-Rad 10 system (Bio Rad, Hercules, California, USA), functioning on High Performance Liquid Chromatography (HPLC)-based ion exchange chromatography (Diabetes Control and Complications Trial standardised) and lipid profile by Beckman Coulter auto-analyser 5800. Elevated triglyceride (TG) was defined as ≥150 mg/dL and low high density lipoprotein-cholesterol (HDL-C) as <40 mg/dL in males and <50 mg/dL in females. [11]

Statistical analysis

Statistical analyses were performed using SPSS V.23 (IBM) and WHO Epi Info (V.7.2.0.1). The student’s t- test, one-way Analysis of Variance (ANOVA), Mann-Whitney or Kruskal Wallis was used to compare difference between continuous variables, as appropriate. χ² test was used to examine difference between proportions. Person-years for diabetes or pre-diabetes were calculated from the baseline examination until the event developed, death occurred or until the last examination, whichever was observed first. Incidence of diabetes with 95% CI was calculated in per 1000 person-years. Cox proportional hazards model was used to find the association between risk factors and progression to dysglycaemia (ie, pre-diabetes or diabetes) and diabetes. The variables to be included in the model were decided by a combination of their clinical relevance and a p<0.2 in univariate analysis. All comparisons were done at a level of significance of 0.05.

Results

The results are from 473 cases contributing a total of 4861 person-years, followed for a mean duration of 10.27 years. The mean age was 51.9±13.9 years and 71.4% of the individuals were <60 years of age. The male to female ratio was 0.9:1% and 56.9% belonged to either upper or upper middle socioeconomic class (table 1). As compared with baseline they had a higher BMI (24.8±4.7 vs 26.8±5.1, p<0.01), WC (87.3±12.6 vs . 88.5±11.6, p<0.01), fasting plasma glucose (95.4±15.9 vs 102.9±27.6, p<0.01), 2-hour post prandial glucose (125.9±31.8 vs 151.5±59.4, p<0.01) and HbA1c (5.6±0.6 vs 6.1±1.1, p<0.010).

Incidence rate for dysglycaemia

Of the 473 individuals, 105 progressed to diabetes and 74 progressed to pre-diabetes. Of these 179 individuals, 118 (44 diabetes and 74 pre-diabetes) were unaware of diabetes, while 61 developed diabetes prior to the follow-up (table 2). The incidence of diabetes, pre-diabetes and dysglycaemia (either pre-diabetes or diabetes) was 21.6,18.7 and 31.7 per 1000 person-years, respectively. Among 376 individuals with normoglycaemia at baseline, 51 progressed to diabetes with an incidence rate for diabetes of 12.9 per 1000 person-years. While in 97 individuals with pre-diabetes at baseline, 54 progressed to diabetes giving an incidence rate of 59.8 per 1000 person-years. Among those with pre-diabetes, highest incidence for diabetes was seen in individuals with baseline isolated impaired glucose tolerance (74 per 1000 person-years), followed by individuals with combined impaired fasting glucose plus impaired glucose tolerance (67.6 per 1000 person-years) and isolated impaired fasting glucose (43.2 per 1000 person-years). Overall, among those with normoglycaemia at baseline, 13.6% converted to diabetes and 19.7% to pre-diabetes giving an overall rate of 33.2% for conversion to dysglycaemia over a period of 10.27 years. Among those with pre-diabetes at baseline 55.7% converted to diabetes. There was no significant difference between the two genders regarding conversion to diabetes, pre-diabetes or dysglycaemia (p>0.05).

Risk factors for conversion to dysglycaemic states

There were significant differences between individuals who remained normoglycaemic compared with those who converted to pre-diabetes or diabetes. Among the risk factors an increasing age, presence of family history of diabetes, sedentary life style, obesity, abdominal obesity, elevated TG and a higher baseline HbA1c, fasting plasma glucose and 2 hours post prandial glucose showed significant association for conversion from normoglycaemia to dysglycaemia (table 3). The risk factors associated with conversion from pre-diabetes to diabetes were sedentary lifestyle, obesity, elevated TG and higher post prandial plasma glucose (table 3). There was no significant association of gender, socioeconomic status, smoking and alcohol for conversion from normoglycaemia to either diabetes or pre-diabetes and from pre-diabetes to diabetes (table 3). In multivariate analysis, significant predictors for conversion from normoglycaemia to dysglycaemia were age (HR 1.02; 95% CI 1.01 to 1.04), family history of diabetes mellitus (HR 1.56; 95% CI 1.09 to 2.25) and sedentary lifestyle (HR 1.51; 95% CI 1.05 to 2.17) while for conversion from pre-diabetes to diabetes it was obesity (HR 2.43; 95% CI 1.21 to 4.89) (table 4).

Comparison of respondents and non-respondents

There were significant differences between respondents and non-respondents (online supplemental table 1). As compared with non-responders, responders had a significantly lower proportion of individuals belonging to high socioeconomic status (88.6% vs 69.6%), lower HDL-C levels (46.3±6.4 vs 45.4±6.8 mg/dL) and a higher SBP (127.9±16.1 vs 130.3±17.3 mmHg), DBP (84.1±9.1 vs 85.6±9.3 mmHg) and TG (139.4±28.7 vs 148.4±34.5 mg/dL). Though we did not find a significant association of socioeconomic status, blood pressure and lipid levels with incident dysglycaemia in the multivariate analysis, whether these differences will affect estimation of incident rates cannot be ruled out with certainty.

Discussion

The current study provides the population-based data for incidence of diabetes and pre-diabetes for Asian-Indians living in north India. The incidence rate of diabetes and pre-diabetes was 21.6 per 1000 person-years and 18.7 per 1000 person-years, respectively. It was 59.8 per 1000 person-years in individuals having pre-diabetes and 12.9 per 1000 person-years in individuals with normoglycaemia. Of all the incident diabetes cases, 41.9% were undiagnosed. Advancing age, family history of diabetes mellitus and physical inactivity were associated with an increased risk of conversion to a dysglycaemic state, while obesity was associated with conversion from pre-diabetes to diabetes.

The incidence rates from the current study need to be seen in the light of similar studies from SEAR and that from the remaining six IDF regions. From the Indian subcontinent there have been five studies pertaining to diabetes incidence, the majority being from southern India. [2–6] Type 2 diabetes incidence reported in these studies varied from 20.2 to 33.1 per 1000 person-years. [2–6] The current study found a similar incidence rate suggesting that both ethnicity (Dravidian in south India and Indo European in north India) are similarly affected by the current diabetes epidemic. [14] For a country of more than 1.35 billion people the present incident rate will translate into a substantial number of individuals developing diabetes in future. On the contrary, the incidence in other IDF regions is much less (5.5/1000 person-years in USA, 6.8–9.5/1000 person-years in China, and 10.6/1000 person-years in Iran). This high incidence rate in the current study was seen despite a similar proportion of individuals having a BMI ≥30 kg/m² in comparison to other IDF regions, suggesting that Asian-Indians have a higher predisposition to diabetes at a lower BMI. [15–17]

In addition, individuals with pre-diabetes also had a high incidence rate of diabetes that is, 59.8 per 1000 person-years and it was similar to that of diabetes that is, 18.7 and 21.6 per 1000 person-years, respectively. Both of these findings suggest that Asian-Indians have a faster conversion rate from pre-diabetes to diabetes. This phenomenon is partly explained by the earlier reported findings of simultaneous presence of insulin resistance and rapidly declining ß-cells function at the disease onset itself in Asian-Indians. [18] Furthermore, this high conversion rate is not only confined to Indo-European ethnicity but is also seen in Dravidians where the incidence of diabetes in individuals having pre-diabetes has been variably reported to be between 64.8 and 78.9 per 1000 person-years. [2, 4] Since the total estimated number of individuals with pre-diabetes is expected to increase to 49.8 million by 2045 in SEAR, a targeted intervention measures in this high risk subgroup individuals can significantly help in reducing the number of new diabetes cases. [1] Furthermore, the conversion rate from pre-diabetes to diabetes in Asian-Indians is much higher than that in Caucasians. In study from Spain by Valdés et al, incidence rate were 21 per 1000 person-years in individuals with baseline IFG and 34.7 per 1000 person-years in those with IGT. [19] Similarly in the Baltimore Longitudinal Study of Aging and Rancho Bernardo study incidence rate of diabetes in individuals with IGT was 35.8 and 40.0 per 1000 person-years, respectively. [20] These differences were seen despite the current study cohort having a lower BMI and WC suggesting a higher predisposition for conversion from pre-diabetes to diabetes at a lower BMI in Asian-Indians.

The predictors of progression to diabetes have been variably reported to be obesity, fasting and 2 hours post glucose challenge glucose, HbA1c, family history of diabetes mellitus, advancing age, socioeconomic status and dietary pattern. [15, 20–23] In the current cohort, obesity was found to be the most important factor for progression from pre-diabetes to diabetes. Furthermore, it was present in 44.5% individuals at baseline survey and increased to 63.7% at follow-up. This denotes that acquisition of adiposity is contributing to declining β-cells function and targeting it can potentially prevent it. Furthermore, we did not find socioeconomic status to be significantly associated with incident dysglycaemia. This reflects that diabetes in Indian subcontinent is no more a disease of affluent class. The same is also reflected in a similar diabetes incidence rate in Indian urban and rural area. Lastly increasing age was found to be an important non-modifiable risk factor for incident dysglycaemia. For Indian subcontinent where the proportion of individuals more than 45 years of age is on a rise, this will mean that the absolute number of diabetes cases will continue to increase even if other modifiable risk factors stay the same. Lastly, the association of family history with incident dysglycaemia suggests a hereditary predisposition, meaning that Asian-Indians living abroad will continue to have a high risk for diabetes.

Another important finding from the current study is that 41.9% cases of diabetes were undiagnosed. This is similar to the IDF diabetes atlas estimate of undiagnosed T2DM cases of 56.7% in the SEAR, suggesting an urgent need of population-based screening strategy for dysglycaemia. [1]

The findings from the current study have significant public health implications. A high incidence rate of diabetes in normal individuals and a high conversion rate of pre-diabetes to diabetes in a country of more than 1.35 billion people with an ageing population will mean that Indian subcontinent is likely to stay diabetes capital of the world in near future. Moreover, the existence of an Asian-Indian phenotype has health implications for Indians living abroad, who will continue to have a higher risk for diabetes and might require a different approach pertaining to diabetes prevention strategies. In this regard, as compared with the native population, a lower cut-off for BMI (≥25 kg/m²) and WC (≥90 cm in males and ≥80 cm in females) is required to identify obesity in Asian-Indians. Lastly with half of the diabetes cases undiagnosed, there is an urgent need for intensification of the screening programmes in SEAR for timely identification and subsequent treatment, in order to avoid associated long term morbidity and mortality.

The strengths of the study are over a decadal follow-up in a representative sample, use of OGTT for estimating glycaemic status and identification of modifiable risk factors for conversion from pre-diabetes to diabetes. The limitations are a high migration rate with significant differences between responders and non-responder, lack of annual follow-up and inclusion of only north Indian population.

Conclusion

The incidence rate of diabetes and pre-diabetes in urban Asian-Indians is 21.6 and 18.8 per 1000 person-years, respectively. Advancing age, family history of diabetes mellitus and physical inactivity were associated with an increased risk of conversion from a normoglycaemic to a dysglycaemic state (pre-diabetes or diabetes) while obesity was associated with conversion from pre-diabetes to diabetes. The high incidence rates call for a pressing need for public health interventions targeting modifiable risk factors to steady the diabetes epidemic in this population.

Acknowledgements

The authors acknowledge the efforts of Aman Puri (AP), Rashi Goel (RaG), Persis Khalkho (PK) and Priya Hiteshi (PH) in conducting the study.

Contributors

RG and SSJ were involved in conducting house to house survey, data collection, data analysis and manuscript preparation. AR, SKB and AB conceived the idea, assisted in data interpretation, manuscript preparation and editing. NS and SR provided assistance in performing laboratory analysis and manuscript editing. AR is the guarantor of the study and accepts full responsibility for the workand/or the conduct of the study, had access to the data, and controlled decision to publish.

Funding

The study received grant from Research Society for Diabetes in India (grant number RSSDI/HQ/Grants/2019/878).

Competing interests

None declared.

Provenance and peer review

Not commissioned; externally peer reviewed.

Data availability statement

Data are available upon reasonable request. The dataset generated and/or analysed during the current study is not publiciliy available but are available from the corrosponding author on a reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by Post Gradute Institute of Medical Education and Research (INT/IEC/2019/001399).

References

Author notes