Achieving glycemic control is the most difficult element of management for many of the estimated 1.3 million VA patients with diabetes. The VA compares favorably with the private sector based upon performance measures. However, the use of aggregate data masks inadequate glycemic control among veterans who are younger, of minority status, and/or have mental health conditions. Despite wide-spread recognition of the importance of chronic complex illness (CCI) in diabetes management, it is not known how they inter-relate with demographic factors (age, sex, race) to impact medication adherence and clinical inertia that are the major mediators of glycemic control.
Our specific objectives are: (1) To evaluate how differences in CCI are related to differences in HbA1c trends for individuals with prevalent and recent-onset diabetes. (2) To evaluate how differences in CCI are related to differences in antiglycemic treatment, including both clinical inertia (lack of medication intensification or step up in medication dose and/or new prescription when indicated) and diabetes medication non-adherence. (3) To evaluate the extent to which variation in diabetes care among racial/ethnic groups is related to differences in the presence of CCI. (4) To evaluate anti-hyperglycemic management and longitudinal HbA1c trends and variability in patients with dominant conditions (overall and subpopulations) and recent-onset diabetes.
This project was a retrospective analysis of veterans with diabetes identified from FY1998-2010, Veterans were classified into 5 chronic comorbid illness groups (CCIG) using a competing demands framework by Piette and Kerr: none, only concordant (DM related), only discordant (non-DM related), both, or dominant (short life expectancy). The first step was to determine if the comorbidity categories had an impact on quality of diabetes care among veterans with recent-onset DM. Second, we compared HbA1c trends for about 10 years across different comorbidity groups among those who recently initiated a mono anti-hyperglycemic oral agent. We used growth curve models (random effects) on repeated HbA1c measurements and adjusted for socio-demographics, visits and seasonality to evaluate individual HbA1c trend (i.e., slope). The analysis was stratified by age category to evaluate the interaction of CCI and age on the HbA1c trend. We also used Cox and logistic regression models to evaluate the relationship of CCI with clinical inertia and medication adherence and persistence.
Aim 1: We empirically evaluated the framework in a cross sectional study by comparing diabetes-care related 3 process measures and 2 treatment goal measures in cohort of 42,826 veterans with new onset diabetes in FY2003 (Pentakota, 2012). Compared to none, those belonging to discordant and dominant CCIs were associated with poorer care process and outcomes.
Aim 1 and 4: We examined the longitudinal impact of comorbid conditions on long-term glycemic control in veterans who recently initiated anti-diabetes treatment (n=79,249), using all outpatient HbA1c values in FY 2000-10, at least 30 days apart (manuscript submitted). Compared to those with only DM and no other illnesses, diabetes patients with concordant illnesses have shown better glycemic control and those with discordant illnesses have comparable or worse control. Those with dominant illnesses tend to have lower HbA1c trend. Age-stratified analysis showed that, compared to younger veterans, the older veterans had lower baseline HbA1c [under 55 year to 75 years and above; 8.60% to7.58%]; had a gradual initial descent in mean HbA1c values following anti-diabetic medication initiation [3.126 to 1.562% units/year]; reached lower HbA1c mean value at end of 6 months from initiating treatment [7.031 to 6.684%]; had a moderate rate of rise in mean HbA1c values following the initial drop [0.115 to 0.042% units/yr].
Aim 2: Overall, 50% of veterans had medication adherence greater than 80%. Presence of co-morbidities, except concordant CCIs, reduced odds for medication adherence by 12-32%. Among the 48,472 veterans who failed their initial anti-diabetes treatment, close to 40% remained un-intensified for a year or more indicating presence of treatment inertia. Concordant and dominant illness presence was associated with marginally lower treatment intensification odds (OR: 0.90 (0.82-0.99)).
Aim3: Compared to Non-Hispanic Whites, African-Americans were more likely to have no comorbidities (39.5% vs. 45.9%) or have discordant comorbidities (27.1% vs. 31.8%); were younger (Mean age (S.D.): 65.8 (11.0) vs. 59.8 (11.9)); had lower mortality (% deceased: 31.9 vs. 22.9) and longer duration of follow-up (median in years 8.4 vs. 8.6). African-Americans had significantly higher HbA1c values (8.6 %( 2.4)) at time of initiation of diabetes treatment than Non-Hispanic Whites (7.9 % (1.8)). For the first six months, following treatment initiation, the mean HbA1c values amongst African-Americans dropped more precipitously than those for Whites (2.67 vs. 2.17%/year). However, their mean HbA1c values at end of 6-months were higher than those for Whites (7.38% vs. 7.16%). Following the initial drop, we observed slightly more moderate rate of rise, of mean HbA1c values, amongst African-Americans (0.046 vs. 0.068%/year) compared to Whites, while persistently maintaining higher mean HbA1c values for the remainder of the study follow-up. The racial differences in pattern of glycemic control were similar across all age and comorbidity sub-groups. African-Americans were found to be at greater risk (34%) to becoming non-persistent (or discontinuity in treatment) with their diabetes treatment within first 2 years following treatment initiation, compared to Whites (HR: 1.34 (1.31-1.38)). Throughout the study follow-up, compliance (measured using proportion of days covered (PDC) with diabetes medications, was significantly lower among African-Americans and ~65% had either poor (PDC <0.70) or moderate (PDC 0.70 - <0.80) compliance. The overall mean PDC for African-Americans was 0.68 compared to 0.77 for Whites. African-Americans were also less likely to receive diabetes treatment intensification following failure of initial therapy. The odds for intensification within 1-year of failure of index treatment among African-Americans were 24% lower compared to those for Whites (OR: 0.76 (0.70-0.81)). African-Americans were more likely to be associated with comorbidity groups (None and discordant) that were associated with poorer glycemic control and lower adherence levels. Interventions aimed at improving adherence to diabetes medications and overcoming barriers to timely treatment modulations in response to failed glycemic control among African-Americans might help reduce some of the observed racial disparities in glycemic control, along with care coordination to help them cope with diabetes and other comorbidities.
Conclusion: Glycemic outcomes and medication utilization differed by co-morbid status. Veterans who are younger with no co-morbidities or discordant CCIs, had worse longitudinal glycemic control. Analysis of clinical inertia among patients who failed mono therapy shows that those with concordant or dominant CCI's had lower rates of treatment intensification. Finally, adherence to DM medications was lower in those with discordant and dominant CCIs.
These results suggest opportunities for targeting specific groups of Veterans - younger and healthier - who would benefit from tighter control.
- Pentakota SR, Rajan M, Fincke BG, Tseng CL, Miller DR, Christiansen CL, Kerr EA, Pogach LM. Does diabetes care differ by type of chronic comorbidity?: An evaluation of the Piette and Kerr framework. Diabetes Care. 2012 Jun 1; 35(6):1285-92.
- Pentakota S, Rajan M, Tseng C, Fincke G, Miller DR, Christiansen C. Interaction between Type of Comorbidity and Visit Frequency on Diabetes Care. Poster session presented at: AcademyHealth Annual Research Meeting; 2012 Jun 25; Orlando, FL.
- Pentakota S, Tseng C, Rajan M, Pogach LM. Impact of comorbid conditions on long-term glycemic control in Veterans with newly identified diabetes mellitus. Poster session presented at: AcademyHealth Annual Research Meeting; 2012 Jun 25; Orlando, FL.