Using administrative data sets to describe clinical care in sickle cell disease-Implications for public policy

Background: Sickle Cell Disease (SCD) affects 100,000 individuals in the United States. Advances in care have resulted in a growing population of adults with SCD. Without a parallel increase in the capacity to care for this group of patients, inequities have emerged in access to quality care and health outcomes. Most adult patients are hospitalized outside of SCD centers. Increased mortality in young adults following transition from pediatrics is often due to Acute Chest Syndrome (ACS), a life-threatening pulmonary process usually requiring blood transfusion. Recognition and appropriate treatment of ACS could represent a key indicator of care and promote wellness for adult SCD patients. Our recent query of the California Office of Statewide Health Planning and Development (OSPHD) database found that one-fifth of the hospital inpatient visits associated with the diagnosis of SCD between 2005 and 2008 were for ACS or a related pulmonary process. Despite NIH standard of care guidelines suggesting that transfusion should be used to treat ACS, we found that only 46% of those visits were associated with a transfusion, implying many patients are not receiving appropriate care. Administrative data allows researchers to access large populations but has not been validated for SCD. In light of recognized concerns regarding the relationship of coded diagnoses in administrative data to final clinical diagnoses, we compared OSHPD visit-level discharge data to three hospital databases. Methods: Hospital billing data from 2009-10 identified patients as having ACS. Equal numbers of cases were reported to OSHPD during the same time period. Primary and all secondary diagnoses were reviewed in administrative data. Chart reviews were conducted of identified cases. Results: Chart review identified a higher number of ACS visits in hospital data in light of cases in which ACS was investigated but not proven. More transfusions were identified in hospital data than OSHPD. The net effect was a lower transfusion rate in OSHPD (52.3% Hospital 1; 12.9% Hospital 2; 22.7% Hospital 3) than actual transfusion rate (77% Hospital 1; 50% Hospital 2; 50% Hospital 3). Conclusions: Administrative data is widely used to inform health policy, with OSPHD in particular widely used in California. However, these results suggest that using administrative data to assess clinical care for SCD may lead to inaccurate assumptions about quality of care, thus leading to continuous disparities in health care services.