B. P. Smith1, C. E. Fick2, D. N. Holena1 1University Of Pennsylvania,Surgery,Philadelphia, PA, USA 2Georgetown University Medical Center,Washington, DC, USA
Introduction: Hospital readmissions data is a critical aspect of patient care as it is directly related to patient outcomes and healthcare expenditures. Data suggest some patients experience improved outcomes with readmission to their index hospital versus another hospital. Much of the information known about re-admission for injured patients is based on single center or state level data sets, which report unplanned 30 day readmission rates between 2 and 7%. The purpose of this study was to define national estimates for trauma readmissions and compare outcomes of patients re-admitted to index hospitals compared to non-index hospitals.
Methods: We performed a retrospective cohort study using the 2013 National Readmission Database. Inclusion criteria were primary ICD-9CM diagnosis codes indicating injury (800.00-959.9, excluding 905-909, 910-924, 930-939), age≥18years, maximum Abbreviated Injury Score≥3, and non-elective admission. The index hospital was defined as the center of first injury admission, and readmission was defined as any non-elective re-admission within 30 days of discharge. The proportion of discharges readmitted to the index vs. non-index hospital were tabulated, and discharge weights and hospital strata were used to generate national estimates.
Results: After weighting, there were 350,102 trauma admissions meeting inclusion criteria (60% male, mean age 58 (SD 22) years, median Injury Severity Score 13 (IQR9-17)). Median index length of stay was 4 (IQR2-9) days. Of these, 31,558 (9%) had ≥ 1 30-day readmission of which only 22,372 (71%) were ever readmitted to the index hospital. Complications related to intracranial injuries, septicemia, and procedural outcomes accounted for nearly 25% of re-admission diagnoses. After adjusting for age, index length of stay, mechanism of injury, and injury severity score, diagnoses most strongly associated with re-admission to the index hospital included complications of surgery (OR 3.3, 95% CI 2.6-4.1), complications of devices (1.9, 1.5-2.4) and acute cerebrovascular accident (1.7, 1.4-2.1). Overall mortality for patients readmitted in 30 days was 4.83% and did not vary between those readmitted to index vs. other centers (4.81% vs. 4.89%, p =0.84). The most common admitting diagnosis resulting mortality in readmitted patients was sepsis, accounting for 36% of 30 day-readmission deaths.
Conclusion: Using a nationally representative dataset, we show the unplanned 30 day re-admission rate for injured patients is 9.0%, which far exceeds most single center and state reports. Most striking is that nearly 1/3 of injured patients are re-admitted to hospitals that differ from the index hospital. Although we are unable to demonstrate a mortality difference between re-admission locations, we do add crucial data to patient centered outcomes such as recovery from brain injury, and outcomes related to procedural and operative intervention for injury.