The proportion of diabetes cases that go undiagnosed in the U.S. may be just 11%, much lower than previous, widely accepted estimates of one-quarter to one-third, a new study suggested.
The previous estimates were based on an analysis of national survey data that used only a single test to identify undiagnosed diabetes, but the new study used a second confirmatory test, as per guidelines from the American Diabetes Association (ADA), explained Elizabeth Selvin, PhD, of Johns Hopkins University in Baltimore, and colleagues.
The proportion of undiagnosed diabetes cases has also decreased over time, from 16% in 1988-1994 to the 11% figure in 2011-2014, they reported online in the Annals of Internal Medicine.
“Establishing the burden of undiagnosed diabetes is critical to monitoring public health efforts related to screening and diagnosis,” Selvin’s group wrote. “When a confirmatory definition is used, undiagnosed diabetes is a relatively small fraction of the total diabetes population; most U.S. adults with diabetes (about 90%) have received a diagnosis of the condition.”
“If we’re thinking about screening programs, these findings suggest that healthcare providers are doing a good job at diagnosing people when they’re coming in contact with the healthcare system,” Selvin said in a statement. “It’s those people who are not coming in contact with the healthcare system that need to be a focus of our efforts to ensure cases of diabetes are not missed.”
Selvin’s group analyzed the same data that was used for the previous estimates, which were published by the CDC: the National Health and Nutrition Examination Survey (NHANES) results for 1988-1994 and 1999-2014. The earlier data set included 7,385 U.S. adults, and the more recent set included 17,045.
The previous estimates relied on just one measurement of fasting plasma glucose, 2-hour glucose, or glycated hemoglobin (HbA1c) to identify undiagnosed diabetes. Selvin’s group defined undiagnosed diabetes as both elevated fasting glucose (≥7.0 mmol/L or ≥126 mg/dL) and an HbA1c of 6.5% or higher in people without a diagnosis of diabetes. Laboratory methods have changed since 1988, so the authors calibrated HbA1c values, and applied regression equations to the plasma glucose values, to take those changes into account.
The prevalence of total diabetes increased from 5.5% in 1988-1994 (9.7 million adults) to 10.8% in 2011-2014 (25.5 million adults), the study found. Confirmed undiagnosed diabetes also increased on an absolute scale, from 0.89% of the total study population in 1988-1994 to 1.17% in 2011-2014, but decreased as a proportion of total diabetes cases, the investigators said.
One of the strongest risk factors for undiagnosed diabetes was BMI. Individuals with a BMI of 30 or higher had more than seven times the risk compared to those with a BMI less than 25 (prevalence ratio 7.40; 95% CI 3.39-11.40).
Sex and race were also significant risk factors. Men were at higher risk than women (prevalence ratio 1.85, 95% CI 1.29-2.42). Compared with whites, Mexican Americans had more than three times the risk (PR 3.66, 95% CI 2.37-4.95), as did Asian Americans (PR 3.49, 95% CI 1.20-5.78). Blacks had more than twice the risk (PR 2.24, 95% CI 1.40-3.07), the investigators reported.
Individuals who lacked access to healthcare were also at greater risk. Not having had a recent healthcare visit (PR 2.43, 95% CI 1.61-3.26) and lack of health insurance (PR 1.63, 95% CI 1.02-2.25) were both significant risk factors for confirmed undiagnosed diabetes.
In an accompanying editorial, Anne Peters, MD, of the University of Southern California in Los Angeles, said, “Overall, these findings are encouraging. They show that there are not as many patients with undiagnosed diabetes as previously thought and that the proportion of cases is decreasing. However, diabetes is still going undiagnosed, particularly in persons who may not have regular access to healthcare or who have several risk factors. Thus, public health efforts to screen at-risk patients and help them receive care may be beneficial.”
The study underscores the importance of a second, confirmatory test to make a diagnosis in patients without symptoms, Peters added. “One test alone may be inaccurate, and making a diabetes diagnosis may have substantial implications for a patient. It may be psychologically distressing and affect a person’s eligibility for several types of insurance (including disability, long-term care, and life insurance policies).”
Peters pointed out some study limitations, including its cross-sectional design and the fact that fasting plasma glucose and HbA1c were measured from the same blood sample, not from separate samples drawn on different days, “which generally is the case during screening in a non-research setting,” she explained.
“Understanding the proportion of diabetes cases that are actually undiagnosed, and who those patient groups are, is really critical to allocation of public health resources,” the authors concluded. “Our results suggest that targeted screening in these populations and increasing health coverage could help make sure that persons who have diabetes receive a diagnosis and get the appropriate treatment that they need.”
The study was supported by the NIH.
Selvin disclosed no relevant relationships with industry. Co-authors disclosed relevant relationships with Abbott Diabetes Care, Becton Dickinson, Boehringer Ingelheim, Bristol-Myers Squibb/AstraZeneca,Calibra, Eli Lilly, Hygieia, Johnson & Johnson, Medtronic, Merck, Novo Nordisk, Roche, Sanoﬁ, Takeda, Dexcom, the NIH, and the National Kidney Foundation.
Peters disclosed no relevant relationships with industry.