Projection of Diabetes Burden Through 2050: Impact of changing demography and disease prevalence in the U.S.

The model is similar to those used in other studies (7,8,9) and involves combining prevalence data with population projections. Specifically, to project the number of people with diagnosed diabetes (henceforth, diabetes) from 2000 to 2050, we multiplied the U.S. Census Bureau’s age-, sex-, and race-specific population projections by predicted diabetes prevalence rates for each age, sex, and race subgroup.

Data on the age-, sex-, and race-specific prevalence of diabetes from the U.S. representative National Health Interview Survey (NHIS) were used. The NHIS is an annual survey that asks a subsample of respondents in each year whether any household member has diabetes. Diabetes prevalence data based on the NHIS are available by race (whites, including whites of Hispanic origin; blacks, including black Hispanic; and other, comprised of Asians, Pacific Islanders, and Native Americans) for the years 1980–1998, and these data are reported for male and female respondents and the following four age-groups: 0–44, 45–64, 65–74, and ≥75 years (1). These prevalence rates for whites and blacks are listed in Tables 1 and 2. Predicted prevalence rates were determined by regressing the 19 years (1980–1998) of age-, sex-, and race-specific prevalence rates of diabetes against time. In all cases, the coefficient on time was positive and significant, indicating that prevalence rates increased during the 1980–1998 period for all subgroups, consistent with previous reports (2,3).

The U.S. Census Bureau has developed, as the most likely scenario, a middle series of population projections through 2050 on an intermediate set of assumptions regarding fertility rates, life expectancies, and net immigration (10). For each year from 2000 to 2050, population projections are available by age, sex, and race/ethnicity. We aggregated these population projections to get the projected population for each age, sex, and race subgroup for which diabetes prevalence rate data were available. Population projections based on the middle series suggest that the U.S. population is likely to grow from ∼275 million people in 2000 to almost 404 million in 2050 (+47%). The Census Bureau has also created a lowest and a highest series by varying assumptions regarding fertility rates, life expectancies, and net immigration (10).

We conducted sensitivity analyses by varying the following two inputs: the population projections and the projected prevalence rates of diabetes. The Census Bureau’s three population projection scenarios (low, middle, and high) were combined with two scenarios for projected prevalence rates (constant 1998 prevalence rates over the entire period of 2000–2050 and the set of predicted increasing prevalence rates) to yield six scenarios for the future burden of diabetes.

Projections of the number of people with diabetes based on the most likely scenario are shown in Table 3. The total number of people with diabetes will rise from ∼11 million in 2000 to almost 20 million in 2025. By 2050, this is projected at >29 million people—a 165% increase over the 2000 level. Note that these projections imply a steady increase in the overall prevalence of diabetes, from 3.99% in 2000 to 7.21% in 2050. The largest increases in the number of people with diabetes are likely to occur in the oldest age category. The number of women ≥75 years of age with diabetes will rise from ∼1.2 million in 2000 to >4.4 million in 2050 (+271%). The number of men with diabetes in this same age-group will rise from ∼0.8 million in 2000 to >4.2 million in 2050 (+437%).

It is important to test the effects of increasing prevalence rates and projected demographic changes in the U.S. population from 2000 to 2050 on the projected numbers of people with diabetes. To do this, we compared three projection scenarios. The first projections—11.8, 13.3, and 15.9 million people with diabetes in 2010, 2025, and 2050, respectively—are based on constant prevalence rates and the middle population series adjusted to 2000 demographics, i.e., the year 2000 age, sex, and race distribution for all years from 2000 to 2050. The second—12.9, 16.1, and 22.4 million in 2010, 2025, and 2050, respectively—uses increasing prevalence rates and the middle series, again adjusted to 2000 demographics. And the third is the most likely total projection scenario in Table 3—14.1, 19.9, and 29.1 million people in 2010, 2025, and 2050, respectively. Thus, by 2010 an additional 3.1 (14.1–11) million people are projected to have diabetes, and 0.8 (11.8–11) million of this increase is attributable to population growth, 1.1 (12.9–11.8) million is due to increasing prevalence rates, and 1.2 (14.1–12.9) million is due to changing demographics. As seen in Fig. 1, it is demographic changes that account for the largest share of the increases.

The largest increases in race-specific projections of diabetes also occur in the oldest age-groups. The number of white men ≥75 years of age with diabetes is likely to increase, from ∼686,000 in 2000 to ∼3.1 million in 2050 (+352%). For black men in the same age-group, a >10-fold increase in the number with diabetes (from 84,000 to 932,000) is projected between 2000 and 2050. The projected number of white women ≥75 years of age with diabetes is likely to increase by 210% between 2000 and 2050 (from 1.0 million to 3.1 million), whereas the number of black women with diabetes in this age-group will rise by 555% (from 168,000 to 1.1 million). The number of whites (males/females) with diabetes increases from 4.0/4.6 million in 2000 to 9.9/9.5 million in 2050, with the numbers projected to be equal at 7.8 million in 2030. The number of blacks (male/female) with diabetes increases from 0.8/1.2 million to 3.7/3.8 million from 2000 to 2050. Although whites account for the majority of the projected people with diabetes, the largest percentage increases occur among blacks, with ∼4.6 times as many black males with diabetes in 2050 as in 2000. Black females with diabetes are the next fastest growing group, with white males and females following.

The results are shown in Table 4 for selected years. Assuming constant 1998 prevalence rates, projections based on the low, middle, and high population series are quite similar through 2010, differing by only ∼500,000 at the extremes (12.8 million and 13.3 million). At 2025 these projections are more divergent, and by 2050 they differ by ∼8 million (17.0–25.0 million). Under the assumption of increasing prevalence rates, the projected numbers are larger, and the differences in the three scenarios are more pronounced, with a spread of ∼12 million by 2050 (24.7–36.5 million). Note that in all six projection scenarios, overall prevalences of diabetes increase. It is also interesting to note that with the middle population projections, if we project prevalences to increase to 2010 but remain constant thereafter, there are 22.2 million people predicted to have diabetes in 2050, still more than twice the number in 2000.

Based on the most likely scenario, we project that in 50 years the number of people with diagnosed diabetes in the U.S. will increase by 165%, from 11 million in 2000 to 29 million in 2050. The biggest percent increases will be among people aged ≥75 years (336%) and among blacks (275%).

The projections we report suffer from several limitations. Of concern is the omission of Hispanic-specific projections. To the extent that Hispanic prevalence rates are overtaking those of whites or blacks, our projections may understate the future burden of diabetes. Also, our projections do not assume advances in diabetes prevention or the possibility of a cure for diabetes in the next 50 years, which may reduce the numbers with diabetes. On the other hand, unforeseen increases in life expectancy or higher propensity for diabetes screening (11) may increase the numbers with diagnosed diabetes. Also, changes in diagnostic criteria (e.g., cut points) could affect prevalences, as well as increased or decreased access to medical care. There is also evidence of increasing prevalence of childhood diabetes (12), which may also increase the future number of people with diabetes. We have assumed a linear increase in diabetes prevalence. However, diabetes prevalences increased by 16% between 1980 and 1994 (1) and 33% between 1990 and 1998 (3). Should the rise in diabetes prevalence rates be nonlinear, our projections may be underestimates. Also, prevalences may rise at rate that is less than linear, and thus our projections may be overestimates. The reported projections seem reasonable if things continue as in the past, with no major changes occurring.

Limitations aside, our projections indicate that previous predictions concerning future diabetes burden in the U.S. are too low. With worldwide changes in lifestyle and demography, this situation concerning future diabetes burden is likely to be true for several countries. Therefore, the estimates of the global burden of diabetes reported by the World Health Organization (8) are likely to be underestimates if the anticipated changes in demographic composition and diabetes prevalence rates (which may capture obesity and lifestyle changes) are accounted for.

Diabetes is already recognized as a public health problem of pandemic proportions (13). Our projection of diabetes burden in the U.S. indicate that the situation may be more alarming than previously believed. Advances in primary prevention (14) may help reduce the number of people with diabetes. However, those who already have the disease deserve better translation of available treatment (15). The economic cost of diabetes is already staggering, and future increases in the number of people with the disease and future advances in technology will likely increase these costs further. Given the reality of finite resources, the optimal management of diabetes will be an increasingly tougher challenge, and something to which we need to respond.

By 2050, there will be an additional 18 million people with diagnosed diabetes in the U.S. alone, and 37% of this increase will be due to changes in the age, sex, and race composition of the population, 27% will be due to population growth, and 36% will be due to changes in prevalence rates. The elderly and minority groups in the U.S., already disproportionately affected by diabetes, will experience the most rapid growth in the number of people with diabetes. These numbers speak for the challenges ahead.