The prevalence of type 2 diabetes (T2D) has almost doubled since 1980 with targeted glycated hemoglobin (A1C) <7% being recommended to prevent complications (1). A recent study reported that 41% of a cohort of 35,304 patients with T2D from the U.S. and Sweden had suboptimal glycemic control (2). Therapeutic inertia, defined as “the failure to advance or deintensify treatment” when appropriate to do so (3), is a key driver of uncontrolled hyperglycemia. The current evidence-based clinical guidelines (1) recommend early initiation of insulin in cases of very high glycemia (VHG) (i.e., A1C >10% or blood glucose levels ≥300 mg/dL [16.7 mmol/L]). Although therapeutic inertia can occur at any time during the disease process, it is more likely at insulin initiation due to fear of hypoglycemia or weight gain (3). Epidemiological data on therapeutic inertia to insulin initiation are limited, and factors associated with noninitiation of insulin treatment despite VHG (NIIT) remain unexplored at the population level.
In U.S. adults with T2D, we aimed to identify factors associated with both VHG and NIIT and to report the trends from 1999 to 2018.
The National Health and Nutrition Examination Survey (NHANES) surveys were approved by the National Center for Health Statistics Research Ethics Review Board (4), and all participants provided written informed consent.
We included nonpregnant individuals aged ≥20 years with known diabetes and pooled NHANES waves into five time intervals (1999–2002, 2003–2006, 2007–2010, 2011–2014, and 2015–2018) to improve the precision of our estimates. We used binomial regression to estimate the age-adjusted prevalence for both outcomes, and prevalence ratios across NHANES waves. To identify factors associated to outcomes, we estimated adjusted odds ratio (aORs) and 95% CI using multiple logistic regression. All statistical tests were two-sided, with a type I error probability of 5%.
Table 1 shows sex-stratified characteristics of 8,816 adults with diabetes and factors associated with outcomes. Non-Hispanic Caucasian individuals represented 39% of the population, 49.7% were aged ≥60 years, and 22.5% had an income-to-poverty ratio ≤1. In both sexes, non-Caucasian ethnicity was associated with higher odds ratio of both VHG and NIIT specifically in non-Hispanic African/African American (24.3%) and Hispanic female (13.3%) individuals. Hispanic male individuals had higher odds of VHG (aOR 5.11, 95% CI 2.17–12.01) with a trend toward higher odds of NITT (aOR 2.26, 95% CI, 0.99–5.17). Higher odds of both VHG and NIIT were also found in normotensive female patients (37.2%), patients with BMI ≥30 (33.9%), and male patients aged 20–40 years (6.6%). Poverty was associated with higher odds of VHG but not NIIT in both male (aOR 2.40, 95% CI 1.10–5.26) and female (aOR 2.85, 95% CI 1.04–7.80) individuals. Marital status, education level, and acculturation score were not associated with VHG or NIIT. In addition, smoking and chronic kidney disease did not increase the risk of VHG or NIIT.
Factors associated with VHG* and noninitiation of insulin treatment† in U.S. patients with T2D
Across the five NHANES waves, the overall prevalence of VHG dropped from 9.9% (1999–2002) to 5.6% (2015–2018), with a decrease in the age-adjusted prevalence for both VHG (17% to 10%) and NIIT (13% to 4%). In both sexes, accounting for the increasing prevalence of T2D, there was a significant linear trend of decreasing age-adjusted prevalence of NIIT from 1999–2002 (71 patients; 8.6%, 95% CI 6.8–10.8) to 2015–2018 (105 patients; 2.2%, 95% CI 1.8–2.6). After adjusting for age, sex, and race, we found that mean A1C decreased over time in participants who initiated insulin therapy (P for trend < 0.01) but not in those who never used insulin.
Among U.S. adults with T2D, 1) the age-adjusted prevalence of both VHG and NIIT significantly decreased from 1999–2002 to 2015–2018; 2) non-Caucasian ethnicity (i.e., being Black or Latino) is a major determinant of both VHG and NIIT, independent of education level; 3) male but not female individuals diagnosed with T2D at young adulthood represent a high-risk group for both issues; and 4) sex difference in the distribution of risk factors for both VHG and NIIT is evident. Poverty was associated with high odds of VHG but not with NIIT. Recent findings suggest that use of high-cost antidiabetes drugs in U.S. adults was determined by insurance status rather than by income (5). The distribution of insured versus uninsured patients across strata of income-to-poverty ratio may explain the null association between poverty and NIIT. A complete drug profile could not be established based on the NHANES data, so we could not account for treatments used by those who did not initiate insulin. The temporal changes in age-adjusted prevalences are partly due to current clinical guidelines and in keeping with previous studies (3). Targeted interventions accounting for those findings and emphasizing shared decision-making may help address NIIT in patients with T2D.
Article Information
Acknowledgments. We are grateful to all the study participants and acknowledge the contributions of all their collaborators for their valuable help.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. G.N. conceptualized and designed the study and performed statistical analysis. G.N., B.N., and A.P.K. interpreted the results and critically revised the manuscript for intellectual content. G.N. and B.N. wrote the manuscript. G.N. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
- Received June 14, 2021.
- Accepted September 27, 2021.
- © 2021 by the American Diabetes Association