INTRODUCTION

The prevalence of diagnosed and undiagnosed diabetes reached 9.4% of US adults in 2015.1 Obesity is a strong risk factor for diabetes and appears to be occurring at younger ages.1 Consequently, diabetes may also occur earlier. The American Diabetes Association (ADA) recommends screening for diabetes at 45 years in asymptomatic patients and at any age in overweight or obese adults with risk factors.2 Earlier screening might shift the age at diagnosis, but reports of this shift are largely anecdotal. We aimed to describe population-based trends in age at diagnosis of type 2 diabetes (T2D) among US adults.

METHODS

We conducted a cross-sectional, retrospective analysis using the 2001–2016 National Health and Nutrition Examination Survey (NHANES) data. We included people aged ≥ 18 years who had an HbA1C > 6.4%, fasting plasma glucose > 125 mg/dL, or who were previously told they had diabetes. Pregnant women, patients with probable type 1 diabetes (T1D) (aged < 20 years and receiving only insulin),3 and patients with missing data were excluded. Age at diagnosis was self-reported. For patients with previously undiagnosed diabetes, we assigned their age at the interview as age at diagnosis, but also conducted sensitivity analysis including only previously diagnosed patients. Duration of diabetes was the difference between age at interview and age at diagnosis. Mean age at diagnosis and duration were estimated for each 2-year period. Trends were assessed using linear regression and mid-points of survey cycles as independent variable. We also calculated the number of patients by age at diagnosis for 2001–2006, 2007–2012, and 2013–2016, respectively. A 2-sided p value < 0.05 was considered significant. We used Stata/MP 14.2, accounting for the complex survey design of NHANES.

RESULTS

The sample included 7016 T2D patients (representing 26.2 million US adults), 5308 (75%) of whom were previously diagnosed. Mean age was 59 years, 52% were male, and mean BMI was 32.8 (Table 1). Mean age at diagnosis was 50.5 years. Mean duration was 8.4 years. There was no significant change in either mean age or duration from 2001 to 2016 (Fig. 1a). Results were the same when restricted to either previously diagnosed or undiagnosed patients. Compared with 2001–2006, the number of patients diagnosed at ages 30–34, 50–54, 60–64, and 70–74 years in 2013–2016 was 1.7, 2.0, 1.7, and 1.1 times higher (p value < 0.001), respectively (Fig. 1b).

Table 1 Patient Characteristics of US Adults with Type 2 Diabetes, NHANES 2001–2016
Full size table
Figure 1
figure 1

a Trends in mean age at diagnosis and duration of type 2 diabetes in US adults (p value for trends > 0.05), and b number of US adults with type 2 diabetes by age at diagnosis, 2001–2016. Error bars in a represent 95% confidence intervals.

Full size image

DISCUSSION

After decreasing throughout the 1990s,4 the mean age at diagnosis of T2D in the USA appears to have stabilized. Mean age of diagnosis and duration did not change from 2001 to 2016. Overall, there was an increase of 50% in prevalence of T2D. The number of younger patients—those diagnosed between 30 and 34 years—almost doubled, but so did the number of patients diagnosed between 50 and 54 years, with a much larger absolute increase occurring in the latter group. Interestingly, there was almost no increase among patients aged ≥ 70 years. Therefore, although a casual observer might be struck by the increase in diabetes among younger adults, in fact, there were simply more patients with T2D at all ages. Given this finding, the ADA recommendation to begin screening at age 45 appears appropriate.

Our study has limitations. Because age at diagnosis was self-reported, findings were subject to recall bias. Also, since time from diabetes onset to diagnosis is unknown,5 we could not confirm when diabetes developed. The NHANES does not distinguish among diabetes types. Since > 95% of diabetics are type 2, and we excluded probable T1D patients, our estimates generally represent T2D.

In conclusion, we found that age at diagnosis of T2D among US adults did not change over 15 years. Instead, the doubling of diabetes prevalence at all ages may have created the appearance of an age shift.