Inpatient costs of specific cerebrovascular events at five academic medical centers.

We estimated the hospital costs for patients with different cerebrovascular events and applied patient and administrative variables to explain the variance of the cost estimates with particular attention to the relationship between patient age and cost. The study sample was drawn from an administrative data set of all hospital discharges from five academic medical centers for the 1992 calendar year. Using International Classification of Diseases (ICD-9-CM) primary diagnosis codes, cases were classified into cerebrovascular subgroups: subarachnoid hemorrhage (SAH), intracerebral hemorrhage (ICH), ischemic cerebral infarction (ICI), and transient ischemic attack (TIA). The ICD-9-driven data file was supplemented with billing data containing inpatient charges reported in UB-82 format. Costs were imputed by applying Medicare charge-to-cost ratios and regional wage adjustments to the billing data. We estimated relationships between inpatient costs and a number of demographic and administrative variables. A statistically significant difference was found between cerebrovascular subgroups for both the mean cost per discharge (p<0.01) and the mean cost of an inpatient day (p<0.01). The mean cost per discharge for each subgroup was as follows: SAH, $39,994 (n=218); ICH, $21,535 (n=258); ICI, $9,882 (n=908); TIA, $4,653 (n=303). Likewise, the mean cost per inpatient day was as follows: SAH, $2,215; ICH, $1,396; ICI, $1,036; TIA, $1,117. Length of stay as a measure of resource use was strongly predictive of inpatient cost, explaining 72 to 82% of the variation in cost. Demographic variables (i.e., age, gender, race, insurance status), however, revealed virtually no predictive power, accounting for less than 10% of the variance in each of the four subgroups. There are substantial differences in the patient-level cost of hospital services for stroke-related events. After controlling for the type of cerebrovascular event, basic demographic variables and insurance status (including Medicare) contribute little to the total cost of inpatient care. More important factor include stroke severity, social factors, and clinical practice variations.

Duke Scholars

Author Gregory P. Samsa Biostatistics & Bioinformatics, Division of Biostatistics