The Utility of ICU Readmission as a Quality Indicator and the Effect of Selection.

Objectives: Intensive care readmission rates are used to signal quality, yet it is unclear whether they represent poor quality in the transition of care from the ICU to the ward, patient factors, or differences in survival of the initial admission. This study aims to measure the selection effect of surviving the initial ICU admission on readmission rates.

Design: Retrospective cohort study of adult patients admitted to ICUs participating in the Case Mix Program database from the Intensive Care National Audit Research Centre.

Settings: The study includes 262 ICUs in the United Kingdom.

Patients: The study includes 682,975 patients admitted to ICUs between 2010 and 2014.

Interventions: None.

Measurements And Main Results: The study includes 682,975 patients admitted to ICUs in the United Kingdom. There were 591,710 patients discharged alive, of which 9,093 (1.53%) were readmitted within the first 2 days of ICU discharge. Post-ICU admission hospital mortality and ICU readmission were poorly correlated (r = 0.130). The addition of a selection model resulted in a weaker correlation (r = 0.082).

Conclusions: ICU readmission performed poorly as a performance metric. The selection process by which only patients who survive their index admission are eligible for readmission has a significant effect on ICU readmission rankings, particularly the higher ranked ICUs. Failure to consider this selection bias gives misleading signals about ICU performance and leads to faulty design of incentive schemes.