Main findings
Assuming a constant incidence rate, we found that women and men at age 40 years in 2015 will live approximately 38 years and 33 years free of diagnosed T2D, respectively. Up to the year 2040, these numbers are projected to increase by 1.0 years and 1.3 years. However, we also found that small annual changes in future incidence rates strongly influence these trends. For instance, an annual increase in the incidence rate of 0.5% would result in decreases of T2D-free life expectancy by − 0.5 years and − 0.1 years among women and men, respectively, whilst an analogous decrease results in 2.2 years and 2.7 years increase in T2D-free life expectancy. Trends in \({\mathrm{MRR}}\) had no relevant impact on T2D-free life expectancy.
Assuming a continued decrease in \({\mathrm{MRR}}\) of 2% per year, we found that women and men aged 40 years with diagnosed T2D in 2015 lose 1.6 and 2.7 years of life, respectively, compared to a same aged person without T2D. In 2040, these numbers would reduce by approximately 0.9 years and 1.6 years. These reductions will be much smaller, if excess mortality does not improve after 2015. Trends in T2D incidence had no relevant impact on individual \({\mathrm{YLL}}\) associated with T2D. In the whole German population aged ≥ 40 years, \({\mathrm{YLL}}\) associated with T2D amounted to 10.8 million years in 2015 in case of an annual decrease in \({\mathrm{MRR}}\) of 2%. In 2040, this number will decrease by 4.4 million years. Depending on annual increases or decreases in the incidence rate of 0.5%, population-wide \({\mathrm{YLL}}\) will decrease by 3.9 million years or 4.8 million years, respectively. In case of no further improvements in \({\mathrm{MRR}}\), population-wide \({\mathrm{YLL}}\) would strongly increase to 18.6 million years in 2015 and further increase by 4.1 million years until 2040. In this scenario, \({\mathrm{YLL}}\) were already higher in 2015, because people with T2D would experience the higher \({\mathrm{MRR}}\) beyond 2015, in contrast to decreasing \({\mathrm{MRR}}\) in the other scenarios.
Comparison to previous studies
In general, some previous studies estimated lower T2D-free life expectancy than we did. For instance, a regional German study estimated T2D-free life expectancy at age 40 years in 2014 with 34 years and 29 years for women and men, respectively [12], compared to 38 years and 33 years in our study. Cunningham et al. [13] reported T2D-free life expectancy of 30 years and 33 years at age 40 years in 2004 for men and women in the U.S. In Australia, men and women aged 45 years between 2000 and 2005 were estimated to live 30 and 34 years without T2D [27]. However, direct comparisons of these results are problematic, since most studies relied on cross-sectional mortality rates and (implicitly) assumed that these remain constant beyond the study period. In contrast, we used projected prevalence and simultaneously accounted for projected decreases in general mortality and \({\mathrm{MRR}}\). Besides different study populations and time periods, these different methodological approaches might explain why we estimated higher T2D-free life expectancy. Accordingly, Cunningham et al. [13] found that fixing general mortality rates at levels observed between 1980 and 1989, results in much lower T2D-free life expectancy between 2000 and 2005, compared to using observed mortality rates between 2000 and 2005. They concluded that potential reductions in T2D-free life expectancy due to increases in incidence rates were partly offset by decreases in general mortality rates. This is in line with our finding that T2D-free life expectancy will increase up to the year 2040, even if the incidence rate remains constant.
A similar reasoning holds for the comparison of previous studies estimating \({\mathrm{YLL}}\) associated with T2D. In general, we found lower \({\mathrm{YLL}}\) on the individual level compared to studies from other countries [9, 25,26,27,28,29]. Assuming continuing decreases in T2D-associated excess mortality, we estimated that women and men aged 40 years in 2015 will lose 1.6 years and 2.7 years, respectively. Other studies estimated \({\mathrm{YLL}}\)s between 3.5 years among Swedish men aged 40 years in 2013 [29] and 8.5 years among Danish men aged 40 between 1995 and 2008 [9]. This range covers our results in the scenario assuming constant \({\mathrm{MRR}}\) (scenario A). Hence, the lower \({\mathrm{YLL}}\) in our base case scenario are probably mainly caused by assuming continuing decreases in general as well as excess mortality.
Due to different population sizes, our population-wide estimates of \({\mathrm{YLL}}\) cannot be compared to other countries. The GBD provides country-specific \({\mathrm{YLL}}\)s on the population level. However, the method is based on T2D-related deaths documented in death certificates and does not consider prevalent cases. Using this method, it was estimated that 256,217 years were lost among deaths with T2D as the documented cause of death in 2015 [10]. As expected, this is far below our estimate of 10.8 million \({\mathrm{YLL}}\) associated with T2D, since our approach summarizes all individual differences in life expectancies between people with prevalent T2D compared to same-aged persons without T2D.
Implications for public health
Given assumed trends in general mortality and excess mortality associated with T2D, the results suggest a substantial improvement of \({\mathrm{YLL}}\) associated with T2D and T2D-free life expectancy in the German population between 2015 and 2040. This should motivate further efforts to lower the incidence and excess mortality of T2D. This is particularly important, because we also found that increases in incidence and a sustained high T2D-associated excess mortality would lead to substantial increases in disease burden.
The results may be used to inform about the impact of future efforts in treating and preventing T2D. In this regard, two different mechanisms could be the target of interventions. Measures generally known as ‘primary prevention’ aim to prevent disease and thus target the incidence rate [30]. Given an effective measure of primary prevention, this would impact T2D-free life expectancy and \({\mathrm{YLL}}\) on the population level, but would not improve \({\mathrm{YLL}}\) in individuals with T2D. Examples for primary prevention include taxes on unhealthy products (e.g. sugar-sweetened beverages), food labeling and setting based approaches that support healthy food choices [31, 32]. With regard to food labeling, the nutriscore was recently introduced to Germany [33].
Contrary, tertiary prevention strategies aim at reducing the risk of complications among those with the disease [30], for instance by optimizing glucose control or screening for early stages of complications. One example for tertiary prevention in Germany are disease management programs for diabetes, which are structured models of diabetes care provided by health care institutions in cooperation with health insurances. This type of prevention may improve excess mortality of T2D and thus mainly improve \({\mathrm{YLL}}\) on the individual and population level, but not the T2D-free life expectancy.
In light of past decreases in excess mortality observed in other countries, health care systems were rather successful with regard to tertiary prevention [6, 7]. In contrast, heterogeneous trends in the incidence rate across countries do not suggest consistent improvements with regard to primary prevention [23, 34,35,36,37]. One might conclude that enhanced efforts of primary prevention are warranted, both to increase T2D-free life expectancy and decrease population-wide \({\mathrm{YLL}}\) associated with T2D.
Another important implication is that the methods used to estimate \({\mathrm{YLL}}\) and T2D-free life expectancy strongly influence the results. Here, we argue that these measures of disease burden involve assumption on future trends by definition, since they rely on life expectancy. Hence, incorporating best available evidence on future trends of mortality and incidence rates may yield more valuable estimates to inform policy than assuming currently observed rates [8].
Strengths and limitations
This is the first study that projects the future burden of T2D in terms of YLL associated with T2D and T2D-free life expectancy based on data comprising approximately 90% of the population in Germany. Previous studies in the German context reported only one of these measures and did not project future trends in the disease burden associated with T2D. The estimation of \({\mathrm{YLL}}\) and T2D-free life expectancy in this study was based on projected incidence and mortality rates. Compared to assuming currently observed rates, this may be a more valid approximation because these measures are a function of life expectancy which inherently involve future mortality rates. This is particularly relevant when assessing time trends in \({\mathrm{YLL}}\) and T2D-free life expectancy. For instance, Muschik et al. [12] found a decrease in T2D-free life expectancy between 2005 and 2014, using period life tables. It is not clear if these decreases are due to increases in incidence or due to ignoring future trends in mortality. Our results suggest, that T2D-free life expectancy increases even if the incidence rate remains constant, because of decreasing overall mortality. Hence, in the case of Muschik et al., one might conclude that the population will have shorter life time without T2D, without knowing if this would also be concluded if future trends in mortality were incorporated.
As a drawback, we had to rely on more or less speculative assumptions on future trends. This may be particularly problematic given the long projection over 85 years up to the year 2100. While future mortality rates of the general population are based on sound data dating back to 1871, assumptions on trends in the incidence rate were rather arbitrary. In fact, the input data for the incidence originates from 2010, which is quite dated. Given heterogeneous trends in Europe [23, 34,35,36,37], we were not able to establish a most plausible trend in incidence. Hence, we assumed constant incidence rates from 2010 onwards in the base case scenario and additionally included scenarios with varying time trends to address the lack of input data. With regard to trends in \({\mathrm{MRR}}\), we mostly relied on data outside of Germany. In contrast to trends in the incidence rate, there is consistent evidence from several high income countries that the \({\mathrm{MRR}}\) decreased over the last decades. Hence, we assumed a continued decrease in the base case scenario and compared the results to a scenario with constant \({\mathrm{MRR}}\). Some studies suggest that the trends in \({\mathrm{MRR}}\) differ between age groups [38]. For instance, in Australia, it was estimated that decreases in the \({\mathrm{MRR}}\) only occurred among people aged ≥ 80 years [39]. However, due to the lack of data in Germany, we assumed the same trend in \({\mathrm{MRR}}\) for all age groups. Technically, age-specific trends in \({\mathrm{MRR}}\) could be incorporated into the projection model.
Another debatable issue in our analysis is that in order to calculate \({\mathrm{YLL}}\), we assumed that a person without T2D at a given age and in a given year will not develop T2D after that year. Of course, this is an unrealistic assumption. Nevertheless, it provides valuable information, since \({\mathrm{YLL}}\) can then be interpreted as the potential life years that would be gained, if excess mortality associated with T2D was non-existent. The model we used would allow to incorporate the more realistic setting, in which persons could develop T2D after a given year. However, the resulting \({\mathrm{YLL}}\) estimate would be lower in scenarios with increasing incidence rates, because persons without T2D in a given year would be more likely to develop T2D and subsequently experience higher mortality rates. Measures of disease burden that indicate lower disease burden when the incidence rate increases are not useful to inform public health.
Finally, our study only considers diagnosed T2D, since we used data from statutory health insurance. Hence, our results do not consider people with undiagnosed T2D and those who did not seek health care in a given year. Furthermore, people in private health insurance are not included, which can be considered a minor issue, given that approximately 90% of the population in Germany is in statutory health insurance.