OBJECTIVE: Metabolic decompensations (MD) are hospitalizations considered preventable with appropriate ambulatory care. We tested for associations between diabetes care and MD. RESEARCH DESIGN: We retrospectively compared care between cases (MD; n = 2714) and controls (without MD; n = 10,856) using merged Veterans Health Administration and Medicare data. Logistic regression tested for associations between MD and diabetes care controlling for patient characteristics. SUBJECTS: Veterans Health Administration users with diabetes stratified into high [hemoglobin A1c (HA1c) > or =9%; n = 2532] and low (HA1c <9%; n = 6176) risk groups. MEASURES: The outcome was hospitalization for MD. Care was defined as quarterly or semiannual diabetes visits and HA1c testing during individualized 12-month baseline periods. RESULTS: : Cases averaged more diabetes visits and HA1c tests than controls (P < 0.001 for both) in the 12-month baseline period. Among the high-risk, 29.8% of cases made 4 quarterly visits compared with 29.6% of controls (P = 0.004); among the low-risk, there was no difference in semiannual visits. Among the high-risk, models showed having no visit was associated with higher odds of MD (adjusted odds ratio: 3.05; 95% confidence interval: 1.69-5.49) compared with 4 visits; individuals with 1-4 visits had similar odds of MD. More HA1c testing was weakly associated with higher odds of MD. CONCLUSIONS: MD was associated with more diabetes care, even controlling for patient characteristics. This inconsistency with the theoretical association between appropriate ambulatory care and lower MD rates indicates that MD rates may not accurately reflect diabetes care quality.
OBJECTIVE: Metabolic decompensations (MD) are hospitalizations considered preventable with appropriate ambulatory care. We tested for associations between diabetes care and MD. RESEARCH DESIGN: We retrospectively compared care between cases (MD; n = 2714) and controls (without MD; n = 10,856) using merged Veterans Health Administration and Medicare data. Logistic regression tested for associations between MD and diabetes care controlling for patient characteristics. SUBJECTS: Veterans Health Administration users with diabetes stratified into high [hemoglobin A1c (HA1c) > or =9%; n = 2532] and low (HA1c <9%; n = 6176) risk groups. MEASURES: The outcome was hospitalization for MD. Care was defined as quarterly or semiannual diabetes visits and HA1c testing during individualized 12-month baseline periods. RESULTS: : Cases averaged more diabetes visits and HA1c tests than controls (P < 0.001 for both) in the 12-month baseline period. Among the high-risk, 29.8% of cases made 4 quarterly visits compared with 29.6% of controls (P = 0.004); among the low-risk, there was no difference in semiannual visits. Among the high-risk, models showed having no visit was associated with higher odds of MD (adjusted odds ratio: 3.05; 95% confidence interval: 1.69-5.49) compared with 4 visits; individuals with 1-4 visits had similar odds of MD. More HA1c testing was weakly associated with higher odds of MD. CONCLUSIONS: MD was associated with more diabetes care, even controlling for patient characteristics. This inconsistency with the theoretical association between appropriate ambulatory care and lower MD rates indicates that MD rates may not accurately reflect diabetes care quality.