|| Checking for direct PDF access through Ovid
Preventing rehospitalizations for patients with serious chronic illnesses is a focus of national quality initiatives. Although 8 million people are admitted yearly to an ICU, the frequency of rehospitalizations (readmissions to the hospital after discharge) is unknown. We sought to determine the frequency of rehospitalization after an ICU stay, outcomes for rehospitalized patients, and factors associated with rehospitalization.Retrospective cohort study using the New York Statewide Planning and Research Cooperative System, an administrative database of all hospital discharges in New York State.ICUs in New York State.ICU patients who survived to hospital discharge in 2008–2010.None.Primary outcome was the cumulative incidence of first early rehospitalization (within 30 days of discharge), and secondary outcome was the cumulative incidence of late rehospitalization (between 31 and 180 d). Factors associated with rehospitalization within both time periods were identified using competing risk regression models. Of 492,653 ICU patients, 79,960 had a first early rehospitalization (cumulative incidence, 16.2%) and an additional 73,250 late rehospitalizations (cumulative incidence, 18.9%). Over one quarter of all rehospitalizations (28.6% for early; 26.7% for late) involved ICU admission. Overall hospital mortality for rehospitalized patients was 7.6% for early and 4.6% for late rehospitalizations. Longer index hospitalization (adjusted hazard ratio, 1.61; 95% CI, 1.57–1.66 for 7–13 d vs < 3 d), discharge to a skilled nursing facility versus home (adjusted hazard ratio, 1.54; 95% CI, 1.51–1.58), and having metastatic cancer (adjusted hazard ratio, 1.46; 95% CI, 1.41–1.51) were associated with the greatest hazard of early rehospitalization.Approximately 16% of ICU survivors were rehospitalized within 30 days of hospital discharge; rehospitalized patients had high rates of ICU admission and hospital mortality. Few characteristics were strongly associated with rehospitalization, suggesting that identifying high-risk individuals for intervention may require additional predictors beyond what is available in administrative databases.