Individualized, Non–Age-Based Glycemic Control in Elderly Veterans With Diabetes DREW A. HELMER, MD 1,2 USHA SAMBAMOORTHI, PHD 2,3 MANGALA RAJAN, MBA 1 CHIN-LIN TSENG, DPH 1,2 LEONARD M. POGACH, MD 1,2 OBJECTIVE — To examine the role of age and endocrinology care in glycemic testing and control in elderly veterans with diabetes. RESEARCH DESIGN AND METHODS — In this retrospective study of Veterans Health Administration clinic users aged 65 years with diabetes, we compared glycemic testing and poor glycemic control (A1C 9%) between older (75 years) and younger (65–74 years) veterans in the year 2000. RESULTS — Without adjustment, rates for glycemic testing were 70.2% in older and 71.1% in younger veterans, and those for poor control were 9.4% in older and 12.8% in younger veterans. After adjustment, older veterans had 1.8% lower probability of glycemic testing and 2.9% lower probability of poor control than younger veterans. Endocrinology care was associ- ated with a higher probability of both glycemic testing (9.7%) and poor control (1.0%), regard- less of age. CONCLUSIONS — Glycemic testing and control and effect of endocrinology care were comparable in older and younger veterans with diabetes. Diabetes Care 31:728–731, 2008 D iabetes affects 20% of individuals aged 65 years (1). While glycemic testing and control using A1C are indicated for all people with diabetes (2), glycemic management should be individ- ually tailored, especially in the geriatric population (3). Using age rather than rel- evant clinical and functional consider- ations to guide management decisions may lead to a systematic bias (i.e., age dis- parity). Under optimal care, however, one should not detect differences in rates of glycemic testing and poor control due to age. This article compares glycemic test- ing and poor glycemic control (A1C 9%) rates between younger (aged 65–74 years) and older (aged 75 years) veterans. RESEARCH DESIGN AND METHODS Data We used Veterans Health Administration (VHA) and Medicare data from the Diabetes Epidemiology Cohort of diabetic individu- als who used the VHA for health care (4). Our inclusion criteria were as follows: that subjects be aged 65 years, have had one or more VHA primary care visits in 1999, have continuous Medicare fee-for-service enroll- ment, and be alive on 30 September 2000. The Veterans Affairs New Jersey Health Care System approved the study. Dependent variables A1C testing in fiscal year (FY) 2000 was identified using Current Procedural Ter- minology codes in VHA or Medicare data (5). For individuals with A1C values available (Medicare data do not include laboratory values), we dichotomized indi- viduals’ last FY 2000 A1C value to high- light poor glycemic control (A1C 9%). Despite controversy (6), comparisons based on poor glycemic control minimize the confounding problems of comorbid- ity and patient preferences. Experts agree that poor glycemic control should be ad- dressed in all diabetic patients to mini- mize symptoms (7,8). Key independent variable Subjects were categorized as young-old (65–74 years) or old-old (75 years) in FY 1999, based on differences in life ex- pectancies. Americans live an average of 18.4 years after reaching 65 years of age and 11.8 years after reaching 75 years of age (9). Independent variables (from FY 1999) Variables consisted of demographics (sex, race/ethnicity, and marital status), socio- economic status (VHA priority group sta- tus), access to health care (Medicaid or Medicare Part B enrollment), endocrinol- ogy care, medical comorbidity, and men- tal illness based on ICD-9-Clinical Modification codes (10). Medical comor- bidity was based on individuals’ DxCG relative risk score (RRS) using ICD-9 codes from VHA and Medicare inpatient and outpatient records. The RRS is nor- malized to a mean of 1.0 (range 0.8 – 146.0) (DxCG Risk Smart, Revision A; DxCG, Boston, MA). Endocrinologist care was determined from clinic stop codes in VHA data and physician spe- cialty codes listed on Medicare claims. Analytic procedures We used probit regressions to examine the association between age and diabetes care after controlling for other indepen- dent variables. We calculated marginal ef- fects by transforming parameter estimates to probabilities of an outcome. Marginal effect is interpreted as the change in prob- ability of experiencing the dependent variable in response to a 1-unit change in the independent variable (e.g., from RRS [comorbidity score] of 1.0 to 2.0, and ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● From the 1 Center for Healthcare Knowledge Management, Veterans Affairs New Jersey Health Care System, East Orange, New Jersey; the 2 New Jersey Medical School, University of Medicine and Dentistry New Jersey, Newark, New Jersey; and the 3 Department of Community Health and Preventive Medicine, Morehouse School of Medicine, Atlanta, Georgia. Address correspondence and reprint requests to Drew Helmer, MD, MEDVAMC, 2002 Holcombe Blvd. (111PC), Houston, TX 77030. E-mail: drew.helmer@va.gov. Received for publication 24 July 2007 and accepted in revised form 7 January 2008. Published ahead of print at http://care.diabetesjournals.org on 17 January 2008. DOI: 10.2337/dc07- 1431. The views expressed are the authors’ and do not necessarily reflect the position or policy of the Depart- ment of Veterans Affairs. Abbreviations: FY, fiscal year; RRS, relative risk score; VHA, Veterans Health Administration. © 2008 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Epidemiology/Health Services Research B R I E F R E P O R T S 728 DIABETES CARE, VOLUME 31, NUMBER 4, APRIL 2008