Variation in Inpatient Resource Use in the Treatment of HIV

OBJECTIVE To estimate the impact of insurance status on inpatient resource use after adjusting for health upon admission and site of care. DESIGN Detailed patient information linked to billing records from the AIDS Cost and Service Utilization Survey (ACSUS), a longitudinal analysis of inpatient and outpatient care between March 1991 and August 1992. SETTING Hospitalizations of human immunodeficiency virus (HIV) patients from 10 US cities with high incidence of AIDS. PATIENTS One thousand, nine hundred and forty nine adolescents and adults at various stages of HIV. MAIN OUTCOME MEASURES We estimate inpatient charges, payments and length of stay as a function of patient, and provider and reimbursement characteristics for more than 1,500 hospitalizations to HIV patients. We control for patient characteristics and underlying risk factors including disease stage, CD4 percentage, mode of transmission, discharge status, type of admission, and region. We use hospital-fixed effects to control for unmeasured differences across facilities. RESULTS Unadjusted means indicate that uninsured patients or patients covered by public insurance have significantly lower charges and payments than privately insured patients with similar medical conditions. We find that those differences are substantially reduced after controlling for the hospital in which care is received. Further, we find little evidence that "underinsured" patients are discharged sooner on average. CONCLUSIONS Inpatient resource use is affected by both the hospital in which care is received and the type of patient admitted. Failure to control for unmeasured differences across hospitals is likely to overstate the impact of insurance substantially.