Abstract
OBJECTIVE: We conducted a case-control study to examine if short-term glucose control is related to healthcare-associated bloodstream infections (BSI), urinary tract infections (UTI), and pneumonia in hospitalized adults with diabetes. SETTING AND PATIENTS: We analyzed 205 BSI, 510 UTI, and 109 pneumonia cases and 989, 2463, and 543 controls matched by age, sex and hospital stay seen at a large healthcare system in Manhattan from 2006 to 2008. METHODS: We examined whether infection risk was associated with serum glucose measured at admission and within 2 days to infection, using conditional logistic regression. Co-morbidities, immunosuppressive medications, prior hospitalizations, and insertion of indwelling devices were considered as potential confounders. RESULTS: Admission glucose level was not associated with infection. Glucose levels of ≥ 110 mg/dL measured within 2 days to infection were associated with BSI (Odds ratios from 2.04 to 2.67). Glucose level of ≥ 180 mg/dL was associated with pneumonia (Odds ratio=2.30). Decrease in glucose levels from admission to the infection was greater for controls than for infected cases. CONCLUSION: Healthcare-associated BSI and pneumonia were associated with glucose levels prior to infection diagnosis, but not with glucose levels at admission. Persistently high glucose level could be an indication of an underlying undiagnosed infection.