Incidence of type 2 diabetes mellitus and prediabetes in Kerala, India: results from a 10-year prospective cohort

Globally, around 450 million people are suffering from diabetes mellitus. The age-standardized global prevalence of diabetes mellitus among adult population has nearly doubled since year 1980, rising from 4.7 to 8.5% [1]. The greatest increase in the prevalence of diabetes mellitus is reported from low and middle-income countries [1, 2]. Asia, being the epicenter for the epidemics of diabetes, is responsible for more than 60% of the global burden of diabetes mellitus [3, 4]. Type 2 diabetes mellitus (T2DM) is the most common form of diabetes. One of the reasons cited for this increasing trend in the prevalence of T2DM in low and middle income countries is Asia paradox [5]. Asia paradox refers to the rapid socio-economic and demographic changes in the Asian population towards that of a developed economy. This can be evident in its economic development, urbanization, and nutritional transition [6]. India, with 69.2 million people with T2DM, is the country with 2^nd highest number of people living with diabetes mellitus worldwide next to China [7]. The federal state of Kerala in India is unique in that the health indicators of Kerala are on par with that of developed countries [8‐10]. However, T2DM is now highly visible across all sections of society within Kerala and that implies the existence of Asia paradox in Kerala State. Majority of the data on burden of T2DM in India is derived from the prevalence estimates [11]. Though incidence studies could provide more valid estimates of disease trend, such studies on T2DM are limited. A cohort study on incidence of T2DM in Kerala State thus aptly bridges the lacuna of incidence estimate of T2DM from a population at risk.

Results are based on the 10-year follow-up data obtained from the 869 participants of the cohort. Mean age of the participants was 54.50 ± 14.47 years. Table 1 shows demographic and behavioral characteristics of the study participants. Among current users of tobacco, mean ages for initiating smoking and smokeless tobacco are 21.83 ± 10.57 years and 37.52 ± 17 years respectively. The average duration of alcohol consumption among current alcohol users is 22.04 ± 11.5 years.

Our study describes the incidence of T2DM and IFG during 10 years of follow-up. Our finding that crude incidence of T2DM is 21.9% or 24.5 per 1000 person years re-emphasizes the high burden of T2DM in India. Similar to this finding, high incidence of T2DM was reported in south India by Mohan et al. in 2008 [22], Ghorpade et al. in 2013 [23] and Anjana et al. in 2015 [24]. The incidence of T2DM from those studies were 20.2, 21.5, and 22.2 per 1000 person years respectively. Our cohort attracts special attention in that inspite of being a semi-urban cohort, the incidence is still high in comparison to the two studies from urban south India [22, 24] and the one from rural south India [23]. It is to be noted that high incidence of T2DM is reported from other low and middle income countries too. Latifi et al. in 2016 reported a T2DM incidence rate of 21.9 per 1000 person year from Iran [25]. These observations confirm that the epidemic trends of T2DM is widespread across India as well as across nations.

With the additional information that nearly 60% of participants with baseline IFG converted to T2DM in 10-year follow-up, it is justifiable to conclude that we are facing an epidemic trend of T2DM. Again, this conversion rate (60% or 86.41 per 1000 person years) is much more than the conversion rates from other studies in India. Mohan et al. reported that 40.5% of baseline prediabetes participants converted to T2DM within 8 years of follow-up [22]. Another recent study from South India reported incidence rate of 58.9% or 78.9 per 1000 person-years [24]. Matching these findings, studies from high income country also reported very high incidence of diabetes among participants with baseline IFG [26, 27]. The high incidence rate of prediabetes as compared to T2DM and the high conversion rate of prediabetes to T2DM signal uncontrolled progressive increase in prevalence of T2DM.

Another important finding from our study is the high crude incidence of IFG (36.7%). The high incidence of IFG as compared to incidence of T2DM (crude incidence of 36.7% vs 21.9%) indicates high incidence of prediabetes in the community. In India, a recent multi centric cross- sectional study done in 15 states documented high prevalence of prediabetes as compared to T2DM (24.7% vs. 7.3%) [28]. Similar findings on high incidence of prediabetes were reported from other Asian countries. A study conducted by Latifi et al. in Iran reported the incidence rate of prediabetes as 40.8 per 1000 person years [25]. A study conducted by Vaidya et al. in China also reported high incidence rate of prediabetes (62.6 per 1000 person year) as compared to incidence rate of T2DM, which is 10.0 per 1000 person years [29]. Another 9-year follow-up study from Iran also reported high incidence of prediabetes, 46.1 per 1000 person-years in men and 36.8 per 1000 person-years in women [30]. Thus prediabetes too is at an epidemic trend in our cohort as elsewhere.

Among all the factors studied, obesity emerged as the single-most modifiable risk factor for T2DM. Prevalence of overweight or obesity is quite high (46%) in this cohort. 10% increase in the prevalence of overweight or obesity within 10 year was observed in this cohort. Also, nearly 70% of participants have central obesity based on Asian Indian cutoff. High prevalence of overweight or obesity (25%) was documented in a study conducted on non-communicable risk factors among adults in Kerala in 2009 [31]. In our study, we found that BMI ≥ 25 and central obesity were the significant risk factors for the incidence of T2DM. A person with BMI ≥25 has nearly 1.9 times higher risk than those whose BMI is below 25 (Table 5). The population attributable risk for BMI ≥ 25 is nearly 27% which shows that reduction in overweight or obesity can significantly reduce the incidence of T2DM. Stratified analysis also showed that BMI is independently associated with T2DM as significant risk factor for both men and women. In contrast to our finding, a study done in Iranian population reported BMI as an independent risk factor only in men [32]. Several cohort study as well as cross-sectional study from all around the world reported overweight or obesity and central obesity as important risk factors for T2DM [22, 23, 25, 28, 29, 33‐43]. The drastic increase in the prevalence of central obesity (47.6% vs 70.2%) and overweight or obesity (36.2% vs. 46.1%) in this cohort over 10 years amplifies the grave situation. We believe that reduction in obesity or over weight and abdominal obesity can produce substantial reduction in incidence of T2DM. Our study thus emphasizes the importance of addressing obesity in the primary prevention of T2DM. In our study, we measured only total cholesterol among lipid profile. Since we did not measure triglyceride and HDL-cholesterol, burden of metabolic vascular syndrome [44] was not estimated in our cohort.

We observed high crude incidence of T2DM among those aged 60 and above (27.1%). This is consistent with the study done in Canada by Lipscombe et al. in 2007 where they observed high incidence of T2DM over the age group 50 years [45]. Similar association of age as significant risk factor was documented in several studies [25, 28, 35, 36, 42, 46].

In our study, participants with age higher than 45 years has nearly 1.6 times higher risk of having T2DM. The proportion of those aged 60 and above is 37.9% and that could partly explain the high incidence of T2DM in our study population. Age-adjusted incidence of T2DM in our study came to be 14.56 per 1000 person years. But this would not minimize the relevance of our finding since global demographic trend in modern era is towards increasing life expectancy and ageing of population. Hence the burden of T2DM will ever be continually increasing unless preventive measures are adopted.

Another important non-modifiable risk factors observed in this study is family history of T2DM. Nearly 28% of participants with T2DM have a family history of T2DM and this association is significant. This is consistent to several other studies finding that family history of T2DM is an important risk factor for T2DM [25, 27, 28, 34, 45]. High incidence of T2DM in the present cohort ultimately gets reflected as high rates of family history of T2DM in successive generations, thereby spiraling the T2DM burden.

In the current study we didn’t find genderwise difference in the incidence of T2DM which is in contrast to some studies in Asia that reported high incidence of T2DM among males [23, 25, 29]. Whereas, similar to our study, absence of genderwise difference in the incidence of diabetes was noted in a study done in Iran [32]. Over 10 years, prevalence of hypertension (30.3% vs 59.8%) and hypercholesterolemia (35.7% vs. 70.2%) have doubled. Though we had observed high incidence of T2DM among those with hypertension and hypercholesterolemia, the association is not significant.

The high rate of conversion from IFG at baseline into T2DM at 10-year follow-up identifies people with IFG as the target population for preventive measures. From our study, it is also evident that obese individuals are the single most important target population while envisaging preventive strategies against incident T2DM. Moreover, in resource deficit countries, interventions targeting at-risk population are considered resource efficient. Hence, for states/countries with demographics matched to Kerala, public health preventive interventions targeting people with IFG and/or obesity will be the most effective and resource efficient step in preventing T2DM.

Among Indian states, the state of Kerala is known for its demographic indicators and health indicators progressing towards those of developed countries [10]. Thus the state of Kerala represents a transitional zone between developing countries and developed countries. It is likely that any state/country in transition from developing economy to developed economy might face similar trends in non-communicable diseases. This provides an opportunity for such states/countries to anticipate health outcomes and thereby take precautionary measures. A major strength of our study is the good follow-up rate and response rate. We could follow 68.9% of participants from the baseline survey, which is much higher than any other Indian studies.

The current prospective study estimates that the cumulative incidence of T2DM in Kerala is 21.9% and the incidence of prediabetes is 36.7%. Nearly 60% of the participants who had impaired plasma glucose at baseline converted to T2DM at present, shows an epidemic trend for T2DM in Kerala. Overweight/obesity and central obesity emerged as the most important modifiable risk factors for developing T2DM in Kerala. The findings of this study provide conclusive evidence that incidence of T2DM and prediabetes is increasing rapidly in developing countries.