Healthcare-acquired infections in rehabilitation units of the Lombardy Region, Italy

Little data are available on the frequency and risk factors for infection in patients in rehabilitation units. This was a 2-year retrospective cohort study conducted in 131 rehabilitation units (RUs) of the Lombardy Region, including those for patients requiring musculoskeletal, cardiac, respiratory, neurological and general geriatric rehabilitation. RUs were stratified into three groups by infection rate calculated from administrative data, and a random sample of RUs in each group was selected for analysis. Discharges from these RUs were randomly selected for chart review, and healthcare-acquired infection was confirmed using CDC/NHSN definitions. A logistic regression analysis explored the association among demographic variables of age, sex, type of rehabilitation unit, Charlson comorbidity score, and location prior to RU admission for selected infections. For the 3,028 discharges from 28 RUs, hospital administrative data had a sensitivity of 0.45 and a positive predictive value of 0.89 to identify infections in the chart review. At least one infection occurred in 14.9% of patient discharges, with 71% of infections being urinary, 8.0% respiratory, and 5% skin and soft tissue. Urinary infection was associated with female sex [odds ratio (OR) 1.48, 95% confidence interval (CI) 1.13-1.93], age 75-85 years (OR 2.21, 95% CI 1.12-4.34), Charlson comorbidity score of a parts per thousand yen3 (OR 1.54, 95% CI 1.10-2.17), and the transfer from acute care (OR 1.45, 95% CI 1.04-2.02). For respiratory infection, male sex (OR 3.06, 95% CI 1.51-6.18), comorbidity score of 1 or 2 (OR 2.16, 95% CI 1.08-4.36), and transfer from a healthcare setting other than an acute care hospital were independent risks (OR 3.14, 95% CI 1.15-8.53). Infections are common in residents of these rehabilitation units, and risk factors may differ with type of infection. The proportion of infections which may be prevented and effective prevention strategies need to be determined.