The Burden of Hospital Readmissions for Patients with Sickle Cell Disease in California and North Carolina: 2004 to 2007.

Hospital readmissions are viewed within health services and policy circles as an indicator of poor quality care. Little is known about what patient characteristics may also contribute to readmissions among patients with sickle cell disease (SCD). We used hospital administrative data from two states to examine patient and hospital discharge-level characteristics associated with hospital readmissions in this patient population.

The State Inpatient Databases (SID) provided by the Healthcare Cost and Utilization Project contain patient-level discharge data from 100% of hospitals in participating states. We used SID data from North Carolina and California for the years 2004 through 2007 to examine patient and discharge-level characteristics associated with hospital readmissions among patients with SCD. We identified SCD patients in the dataset using the following ICD9 codes: 28241-28242, and 28260-28269. Encrypted patient identifiers were availabel for the majority of patients in the dataset. For each individual SCD patient in the dataset, we defined an index admission as the patient's first admission in the dataset during the study period. We defined a readmission as any subsequent non-elective admission appearing in the dataset occurring within 1-month of any prior admission. We used ICD9 codes and Clinical Classification Software codes to identify comorbidities associated with SCD: vaso-occlusive crisis, pulmonary hypertension, cerebrovascular accidents, leg ulcers, gall bladder disease, retinopathies, avascular necrosis, chronic renal failure, acute chest syndrome, blood infections, and pneumonia. We also identified patients with diagnostic codes of HIV or other viral infections. We used discharge disposition codes provided in the dataset to identify patients who discharged themselves against medical advice (AMA). We used oneway ANOVAs, t-tests, chi-square tests, and multiple linear and logistic regressions to examine the association of patient-level and discharge-level characteristics with readmission history. A generalized estimating equation was used to account for within-patient clustering as appropriate.

We identified 9966 SCD patients who together accounted for 38363 hospital discharges during the study period. Sixty-five percent of the discharges were from California. Thirty-five percent of the discharges were identified as readmissions given our study definitions. The average length of stay for readmissions was 1.4 days longer than that of other hospitalizations (6.7 vs. 5.3, p < 0.001). Readmissions remained 0.76 days longer on average than other hospitalizations after adjustment for patient age, sex, insurance type, and sickle-related comorbidities (p < 0.001). The average billed charges for readmissions (unadjusted) were $4427 higher than that of other hospitalizations (p < 0.001). After adjustment for age, sex, and sickle-comorbidities, readmissions remained $1923 more expensive than other hospitalizations. The addition of length of stay to the model reversed the readmission effect on average charges. In a patient-level analysis, only 24% of the 9966 individuals in the dataset ever had a readmission during the study period. Nevertheless, this 24% of patients accounted for 71% of the total discharges, 76% of the total days of care, and 74% of the billed charges observed during the study period. In adjusted models, ever experiencing a readmission during the study period was significantly associated with a patient being female, being younger, leaving the hospital AMA, and ever experiencing vaso-occlusive crisis, pulmonary hypertension, cerebrovascular accidents, leg ulcers, gall bladder disease, retinopathies, avascular necrosis, chronic renal failure, acute chest syndrome, blood infections, pneumonia, or a viral infection (non-HIV) during the study period.

A relatively small percentage of SCD patients, who appear to have more severe illness, are responsible for an overwhelming majority of hospital readmissions, which in turn are associated with significantly increased days of hospital care and total charges. The development of interventions to reduce readmissions among the small percentage of patients who disproportionately account for these costly admissions may be an effective and equitable focus of policy efforts to reduce the readmission rate for patients with SCD.

No relevant conflicts of interest to declare.