Original Articles

Journal of General Internal Medicine

, Volume 2, Issue 6, pp 400-405

First online:

Predicting emergency readmissions for patients discharged from the medical service of a teaching hospital

  • Russell S. PhillipsAffiliated withDivision of General Medicine, Beth Israel Hospital
  • , Charles Safran
  • , Paul D. Cleary
  • , Thomas L. Delbanco

Rent the article at a discount

Rent now

* Final gross prices may vary according to local VAT.

Get Access


Emergency readmissions among patients discharged from the medical service of an acute-care teaching hospital were analyzed. Using the multivariate technique of recursive partitioning, the authors developed and validated a model to predict readmission based on diagnoses and other clinical factors. Of the 4,769 patients in the validation series, 19% were readmitted within 90 days. Twenty-six per cent of the readmissions occurred within ten days of discharge, and 57% within 30 days. Readmitted patients were older, had longer hospitalizations, and had greater hospital charges (p<0.01). The discharge diagnoses of AIDS, renal disease, and cancer were associated with increased risks of read-mission regardless of patients’ demographics or test results. The relative risks (95% confidence interval) associated with these diagnoses were: AIDS, 3.3 (1.4–7.8); renal disease, 2.3 (1.7–3.0); cancer, 2.8 (2.4–3.4). Other patients at increased risk were those with diabetes, anemia, and elevated creatinine (2.1; 1.6–2.8) and those with heart failure and elevated anion gaps (2.2; 1.7–2.8). For patients without one of these diagnoses, a normal albumin and no prior admission within 60 days identified patients at reduced risk for readmission (0.4; 0.3–0.4). Thus, commonly available clinical data identify patients at increased risk for emergency readmission. Risk factor profiles should alert physicians to these patients, as intensive intervention may be appropriate. Future studies should test the impacts of clinical interventions designed to reduce emergency readmissions.

Key words

readmission risk factors recursive partitioning