Medicaid inpatient costs and nested structural analysis using a hierarchical linear modeling (HLM) approach

Abstract

This research investigates the factors affecting Medicaid inpatient costs using the nested relationship between hospital and patient levels. Using the 2005 Hospital Quarterly Financial and Utilization Data for the hospital level data and the 2005 Inpatient Discharge Data for the patient level data in California, we derive Medicaid inpatient costs by calculating a ratio of costs to charges at the hospital level, and then multiplying the ratio by each inpatient charge. Based on the selected factors of hospital (i.e., Hirschman–Herfindahl index, case mix index, number of BEDS, ownership, and teaching status) and patient (i.e., AGE, length of stay [LOS], diagnosis-related group weights, number of secondary diagnoses, race, and gender) levels, this study tests not only the cause and effect between factors and Medicaid inpatient costs but also the structural effects in terms of the hierarchical linear model (HLM). We confirm the theoretical arguments from previous literature but we have explored and provided more advanced causalities that the previous literature had not explored. Within the nested structure, the effects of LOS and number of secondary diagnoses are positively or negatively influenced by hospital characteristics such as hospital competition, for-profit status, and teaching hospital. We conclude that the HLM can examine both hospital and patient information and observe more accurate statistical relationships that the previous research had not investigated.