Forecasting the burden of type 2 diabetes in Singapore using a demographic epidemiological model of Singapore

Abstract

Objective: Singapore is a microcosm of Asia as a whole, and its rapidly ageing, increasingly sedentary population heralds the chronic health problems other Asian countries are starting to face and will likely face in the decades ahead. Forecasting the changing burden of chronic diseases such as type 2 diabetes in Singapore is vital to plan the resources needed and motivate preventive efforts.

Methods: This paper describes an individual-level simulation model that uses evidence synthesis from multiple data streams-national statistics, national health surveys, and four cohort studies, and known risk factors-aging, obesity, ethnicity, and genetics-to forecast the prevalence of type 2 diabetes in Singapore. This comprises submodels for mortality, fertility, migration, body mass index trajectories, genetics, and workforce participation, parameterized using Markov chain Monte Carlo methods, and permits forecasts by ethnicity and employment status.

Results: We forecast that the obesity prevalence will quadruple from 4.3% in 1990 to 15.9% in 2050, while the prevalence of type 2 diabetes (diagnosed and undiagnosed) among Singapore adults aged 18-69 will double from 7.3% in 1990 to 15% in 2050, that ethnic Indians and Malays will bear a disproportionate burden compared with the Chinese majority, and that the number of patients with diabetes in the workforce will grow markedly.

Conclusions: If the recent rise in obesity prevalence continues, the lifetime risk of type 2 diabetes in Singapore will be one in two by 2050 with concomitant implications for greater healthcare expenditure, productivity losses, and the targeting of health promotion programmes.

Keywords: Adult Diabetes; Demographics; Simulation; Statistical Methods.

Figure 1

Overview of model structure. Boxes represent submodels; arrows indicate direction of information flow between submodels. BMI, body mass index; T2DM, type 2 diabetes mellitus.

Figure 2

Age-specific, gender-specific, and ethnicity-specific prevalence estimates and forecasts of (diagnosed and undiagnosed) type 2 diabetes. Model forecasts are presented as bars with 95% prediction intervals. Data are indicated by dots with 95% empirical CIs.

Figure 3

Obesity and type 2 diabetes forecasts. Top: forecast prevalence of obesity and overweight in adults (A), forecast prevalence of type 2 diabetes among working age adults (B) and number of patients with type 2 diabetes in the workforce (C). Means and 95% prediction intervals are plotted. For prevalence, point estimates from the National Health Surveys are overlaid. Bottom (D–G): modeled age pyramids with patients with type 2 diabetes and diabetic workers overlaid. Red and blue bars indicate women and men, respectively; black bars indicate patients with type 2 diabetes (not in the workforce) of both genders; and green bars indicate working diabetics. The + symbol indicates data from the censuses of 2000 and 2010.