Background: Financial readmission penalties hold hospitals responsible for 30-day readmissions regardless of whether the patient was readmitted to another hospital.  Emerging evidence suggests patients readmitted to non-index hospitals may experience higher mortality rates.  Hospitals with large proportions of patients readmitted elsewhere may need to alter their approach to these readmissions, but how often patients are readmitted to their index hospital versus another hospital is not well-described.   We sought to describe how often patients are readmitted to non-index hospitals, and characteristics of patients associated with readmission to a non-index hospital.Methods: This is a secondary analysis of the 2013 National Readmissions Database, a nationally representative sample of discharges from hospitals including all payers in 21 states.  We identified adults over age 18 who survived their initial hospitalization, were not transferred to another hospital, and had a readmission within 30 days of their index hospitalization.  We then conducted multivariable logistic regression to identify risk factors for readmission to a non-index facility.

Results: Of 1.56 million readmissions in the database representing 3.5 million readmissions nationally, 363,590 (23.3%) occurred at a non-index hospital.  The patient factors most strongly associated with readmission to a non-index hospital included bring a rural resident of the same state as the index hospital, undergoing prior treatment for alcohol or substance abuse, having Medicare or Medicaid as a primary payor, having a long hospital length of stay, and discharging to a skilled nursing facility for rehabilitation after the hospitalization.  Patient factors most strongly associated with being readmitted to the index hospital included being young, severely ill during the index hospitalization or being treated for COPD, and being discharged with home health care (Table 1).

Conclusions:   Nearly 1/4 of 30-day readmissions in the United States are readmissions to a non-index hospital.   This novel finding in a nationally-representative sample of payers, hospitals, and discharges is surprising and concerning.   Hospitals are increasingly held responsible for outcomes and costs of these readmissions, and preliminary evidence indicates non-index readmissions are associated with longer length of stay and increased mortality.  However, there is no national system currently for hospitals to identify non-index readmissions.  Our results suggest clinicians and hospitals may need to identify the “network” of readmission patterns in their area, and consider multi-site interventions, particularly in the patient sub-populations we identified who are most likely to have non-index readmissions.