Response to Comment on Nanditha et al. Secular TRends in DiabEtes in India (STRiDE–I): Change in Prevalence in 10 Years Among Urban and Rural Populations in Tamil Nadu. Diabetes Care 2019;42:476–485

We appreciate the interest shown by Tönnies et al. (1) in their comment on the prevalence data published by us and for having calculated the incidence based on the data provided in our article (2). In order to properly understand the causes of changes in the prevalence of diabetes, it is essential to assess incidence, as this, not prevalence, is the key index of population risk. Tönnies et al. (1) calculated the mean incidence across the 2006–2016 time period based on mathematical assumptions using the prevalence, temporal changes, mortality rate, etc. This calculation included a number of assumptions, since not all the actual data were available (1).

Taking a different analytical approach to the same problem, we have now used the data from the cross-sectional studies conducted in 2006 (3) and 2016 (2) to directly estimate incidence in each of the two study years, using methods adopted for national estimates of diabetes incidence in the U.S. (4). In this approach, incidence was calculated as the percentage of survey participants with a self-reported history of known diabetes (KDM) with ≤1 year of duration (numerator) among the total survey population minus the number of KDM cases with >1 year of duration (denominator). Diabetes duration was verified from medical records. We thus assessed the change in incidence of diabetes between 2006 and 2016 in a city, town, and periurban villages (PUVs) in Tamil Nadu, southern India, among adults aged ≥20 years. Statistical comparisons were performed with the t test and z test as relevant. As shown in Table 1, the incidence of diabetes increased in 2016 compared with 2006 in all locations. Although the incidence increased in the urban and rural areas, a sharper increase was noted in the city and town when compared with the villages. The ratio of known to new cases showed a reduction in all regions, a probable indicator that the rise in incidence was not due to higher detection or screening rates in the latter years. The proportion of incident diabetes to total KDM rose significantly only in the city and the town. This probably suggests that the higher incidence may be a recent phenomenon in the urban populations. Thus, in a decade the incidence of diabetes rose sharply in urban areas but only moderately in the villages. The increase in incidence has contributed to the higher prevalence reported in the Secular Trends in DiabEtes in India (STRiDE-I) study in all the populations (2).

While our incidence data and the data presented by Tönnies et al. (1) are not directly comparable (different methods and time frames), it is interesting to note that the modeled incidence values by Tönnies et al. for the period 2006–2016 differ by less than 20% from the mean of our 2006 and 2016 estimates shown in Table 1 (Tönnies et al. vs. the present analysis: 2.34 vs. 2.65 for city, 2.19 vs. 2.65 for town, and 1.48 vs. 1.55 per 100 person-years for PUV).

Because a sharp increase in the incidence has been noted, especially in the urban areas, a further increase in prevalence is likely to occur. Primary prevention strategies at the societal level would be an effective tool to curb the burden of diabetes in the urban and rural populations of India.