Several studies have suggested a fall or stabilisation of diabetes incidence rates in the last decade. We therefore conducted a multi-country analysis of trends in diabetes incidence over time. Data from 16 countries comprising 17 administrative sources and one set of annual surveys were analysed according to a standard protocol mainly from the period 1995-2015. We modelled incidence rates using age and calendar time as quantitative variables (scored as the midpoint of each interval), and using restricted cubic splines with 6 knots for age and one knot per 4 years of observation for calendar time. The estimated rates from the models were presented as age- and sex-standardized rates (standardized to the 2010 EU population) for each centre to provide an overview of general trends. To formally assess temporal changes in incidence rates, we also fitted joinpoint models with joinpoints at one of the dates 2009, 2010, 2011, and 2012, estimating the average trend before and after the joinpoint. Among the 18 populations we assembled, 4 of the studies were from predominantly non-Europid populations (Taiwan, Korea, Singapore, Hong Kong) and only 2 of the studies were from middle-income countries (Ukraine and Latvia). In general, from 2000 (or when data were available) until 2010 we saw increasing trends in incidence rates in populations from the U.S., Canada, Australia, Denmark, UK, Spain, Scotland and Latvia, while we saw decreasing incidence rates in Israel, Italy, UK, Hong Kong, Taiwan and Korea. In contrast, after 2010, incidence rates appeared stable or decreasing in 14/18 centres, with U.S. (Kaiser Permanente Healthcare), Ukraine, Singapore, and Taiwan showing increasing trend of diabetes incidence.
In conclusion, incidence rates of diabetes have shown a tendency to stabilise or even fall across some high-income populations in Europe, Asia, Canada and Australia. Patterns in low to middle income countries may be different, but little information is currently available from these.