Serious hypoglycemic reactions are one of the most common adverse drug events and are now recognized as a major public health issue. From the mid-2000s to mid-2012, non-VA guidelines emphasized tight control of glycemia for most patients. Our recent publication reported that as many as 70% of all Veterans on hypoglycemic agents are at high risk, and about half of those may have been over treated in 2009 based on a proposed overtreatment measure (Hemoglobin A1c <7%). This measure has been accepted as a concept measure for future development by CMS, following endorsement by the HHS Office of the National Coordinator for Health Information Technology.
Among various subgroups of high risk patients receiving insulin and/or secretagogues treatment during the period 2004-2016, Aim 1: assess the changes in rates and trends of potential overtreatment; Aim 2: evaluate annual rates, trends, and correlates of serious hypoglycemic events (resulting in emergency department visits or hospitalizations); Aim 3: Study rates and trends, and risk factors of mortality among high risk patients; and Aim 4: Explore de-intensification of glycemic control among potentially overtreated patients.
We utilized both serial cross-sectional (Aims 1, 2, 3) and longitudinal (Aims 2, 3, 4) analyses of Veteran patients based on linked VHA and CMS data files from years 2002 to 2016. We identified patients receiving insulin and/or secretagogues, dully enrolled in Medicare fee for service and removed those with end-stage renal disease (ESRD), in a hospice program or long-term care facilities. Analyses began with inspection of frequency distributions, bivariate analyses of rates among subgroups defined by our independent variables, and use of graphical methods for exploratory data analysis. Advanced statistical modeling will include marginal-Generalized Estimating Equations models with a logit link, Cox proportional hazards models, and the recursive partition classification tree method.
Study populations younger than 65 years in years 2004-2012 ranged from 22,420-30,822: 60-65% were White, 15-19% Black, 7-8% Hispanic, 3-5% women. Insulin use increased over the years, from 40% in 2004 to 57% in 2012. There were 37-43%, 9-12%, and 43-49% with mental health conditions (M), substance use (SU), and cardiovascular conditions (C), respectively. Higher percentages of Blacks (16%) and Hispanics (16%) than Whites (9%) had SU. Patients with SU had higher serious hypoglycemia rates than others regardless of race; in 2012, Blacks (55 events per 1,000 patients) and Hispanics (48) had more than 1.5 fold higher risks than Whites (31). Those with only M or only C had similar rates to those with none of the three conditions. Rates decreased markedly in most patient groups with SU starting around year 2007.
In 3,465 Veterans >= 65 years receiving only basal insulin in 2009-2011, 457 (13%) died or incurred serious hypoglycemic events. Of the 200 (5.8%) that incurred hypoglycemic events, 16%, 24.5%, and 23% had the event within 1, 2-3, and 4-6 months, respectively; of the 271 (7.8%) deaths, 10.7%, 12.5%, and 25.8% occurred in the same time intervals. Of the 2,789 (80%) patients having some A1c values within a year post-prandial, 10% had an A1c <6.5%. Compared to Whites (n=2,708; 78%), Blacks (336; 10%) had a higher hypoglycemia rate (8.0% vs. 5.6%, p=0.08) and a comparable death rate (8.6% vs. 8.4%, p=0.88). Both races had similar rates of having an A1c <6.5% (11.6% vs. 13.1%, p=0.48).
In January 2015 VA launched a Hypoglycemic Safety Initiative in the context of the Choosing Wisely Campaign, emphasizing the importance of avoiding overtreatment in high risk populations defined by the overtreatment measure. This study will pioneer among federal agencies in the development of an administrative data base approach to tracking prevention and surveillance of serious hypoglycemic events that could be used to evaluate success of Federal initiatives.
External Links for this Project
Grant Number: I01HX001601-01A2
None at this time.