JAMA publishes paper on US diabetes prevalence and incidence trends from 1980 to 2012 – October 1, 2014

JAMA published a paper late last month entitled, “Prevalence and Incidence Trends for Diagnosed Diabetes Among Adults Aged 20 to 79 Years, United States, 1980-2012” (see the press release here). The study analyzed cross-sectional data from the CDC’s National Health Interview Survey (NHIS) for 664,969 adults (ages 20-79 years) between 1980 and 2012. The results suggested a doubling of the incidence and prevalence of diabetes from 1990-2008 (prevalence increased from 3.5% to 7.9% and incidence increased from 0.32% to 0.88%), followed by a plateauing between 2008 and 2012 (prevalence increased from 7.9% to 8.3% and incidence decreased from 0.88% to 0.71%).

That said, incidence and prevalence rates continued to rise rapidly among certain subgroups: the rate of increase of diabetes incidence was significantly higher for Hispanic (3.3% per year) and non-Hispanic black (1.5% per year) adults compared to non-Hispanic white adults (a decline of 8.0% per year from 2008-2012 following an increase of 5.5% per year from 1997-2008). The rate of increase of incidence was also significantly higher for young adults (age 20-44) compared to those age 45-64 (3.2% per year vs. 0.4% per year from 2002-2012). The rate of increase of diabetes prevalence was significantly higher for adults with a high school education (4.7% per year from 1988-2012) or less (7.8% per year from 1994-2003 and 2.3% per year from 2003-2012) than for those with more education, as well as for Hispanic adults (3.1% per year from 1997-2012) compared to non-Hispanic black adults (2.2% per year during the same time period); rates among whites were substantially lower (1.9% per year) from 2005-2012, but statistical comparison was complicated by several changes in the trends. The rate of increase in prevalence was also significantly higher for young adults (4.3% per year from 1984-2012) compared to older cohorts (1.3% per year from 2008-2012 for adults age 45-64; 2.4% per year from 2003-2012 for those age 65-79).

The authors propose several potential explanations for the recent plateau; in particular, slower growth in obesity rates in the US during the past decade (see our coverage of the CDC’s 2013 obesity data) may have been a factor, along with concurrent declines in overall caloric intake, food purchases, and energy intake – these rates, of course, are from a higher base so may not be that reassuring in fact. They also suggest that this leveling off could be partially a result of the recent adoption of the A1c test threshold for diagnosis, as prior studies have implied that this test may identify fewer cases of hyperglycemia compared to the fasting plasma glucose (FPG) test – still, most with whom we speak are major fans of using A1c. Regarding the overall doubling of prevalence and incidence between 1990 and 2008, the authors cite the aging population, improved survival rates, and increased risk factors such as obesity and sedentary lifestyle as likely contributing factors.

While a potential slowdown in the diabetes epidemic is an encouraging sign of sorts (it’s better than continued increases from high bases), this difference among subpopulations clearly exacerbates existing racial/ethnic and socioeconomic disparities in diabetes. Such findings illustrate the urgent need for further investment in Medicaid and a focus on education and advocacy for minority groups. It is also essential that any decline in overall growth rates does not cause the healthcare community to approach the problem as anything less than an urgent public health issue – according to the latest CDC statistics, 29 million Americans (9.3% of the US population) had diabetes as of 2012, along with a staggering 86 million with prediabetes, up from 26 million (8.3% of the population) with diabetes and 79 million with prediabetes in 2011.

• Despite the overall strength of the analysis (based on over three decades of data collected from a fairly representative sample), there are several limitations of the study that the authors point out:
  
  • NHIS does not identify undiagnosed disease, which may underestimate diabetes incidence and prevalence rates;
  • The data for some population subgroups may not have been adequate or sufficiently powered to detect trend changes;
  • NHIS does not include data on the institutionalized population;
  • There were certain changes in the conduct of NHIS during the time period analyzed, including changes to the sample design, use of proxy respondents, and the questionnaire;
  • The NHIS household response rates slightly declined in later years; and
  • There was no distinction between type 1 and type 2 diabetes, although we feel comfortable putting this data in the context of type 2 diabetes, as it accounts for 95% of all diabetes cases.   

-- by Melissa An, Emily Regier, and Kelly Close