C. E. Rogers1, P. Baliga1, K. Chavin1, D. Taber1 1Medical University Of South Carolina,Transplant Surgery,Charleston, Sc, USA
Introduction: There is an increasing pressure for hospitals to reduce early hospital readmission (EHR) rates for high-cost, high-risk surgical procedures. Studies have also shown EHRs to be a measure of inpatient quality of care.
Methods: The aims of this study were to determine the predominant risk factors associated with EHR, to develop a risk model and to determine the etiologies, timing and preventability of readmissions in liver transplant (LTX) patients. All patients who received a LTX between Jan 2011 – May 2014 were included. Patients who experienced graft loss within one month after LTX were excluded.
Results: A total of 207 LTX recipients were included, 48% (n=67) were readmitted within 30 days (EHR). Risk factors for EHR included African American race (13% vs. 38%, p=0.006), primary diagnosis of biliary atresia (1% vs. 8%, p=0.025) and donor history of stroke (33% vs. 54%, p=0.007). Although not statistically significant, diagnosis of hepatitis C (34% vs. 45%, p=0.119), an increase in pre-transplant hemoglobin (10.8 vs. 11.3 gm/dL, p=0.119) and a decrease in serum albumin (3.1 ± 3.1 vs. 2.6 ± 0.8 gm/dL, p=0.188) also correlated with an increased risk for EHR. A history of a previous liver transplant (10% vs. 0%, p=0.007) and dialysis within a week prior to transplant (8% vs. 2%, p=0.067) appear to be protective against EHR. These 8 factors were then used as variables in a logistic regression analysis to develop a risk model that demonstrated a negative predictive value of 71.4%, a positive predictive value 67.8%, and an overall predictive value of 70.3%. The secondary analysis revealed that of the patients readmitted within 7 days (n=39), 25% were due to known or ongoing medical problems, which were identified, on average, 2 days (range1-4) prior to the EHR and potentially preventable. Graft loss was significantly higher in LTX with EHR (Figure 1).
Conclusion: This analysis identified specific factors for EHR that can potentially predict which patients are at high-risk for readmission. Future analyses should attempt to prospectively validate this model and target the high-risk patients through interventions designed to minimize EHR and improve overall quality of patient care.
