Understanding Variations In Hospital Length Of Stay And Cost

Author : Thomas L. Walsh
Publisher :
ISBN 13 :
Total Pages : 258 pages
Book Rating : 4.:/5 (86 download)

DOWNLOAD NOW!

Download or read book Explaining Hospital Costs and Utilization in the Nationwide Inpatient Sample written by Thomas L. Walsh and published by . This book was released on 2012 with total page 258 pages. Available in PDF, EPUB and Kindle. Book excerpt: Results. Study One. Overall, there were 5.6 million discharges in 684 hospitals. Discharge variables explained 59% of variation in average cost per discharge within hospitals. Contextual variables added just 0.16%. Hospital structural variables added 3% and attributes of the hospital that are unaccounted for by differences in discharge variables, context, or structure (the random effect) explained an additional 7%, for a total R2 value of 69%. Discharge variables explained 39% of the variation in average LOS per discharge within hospitals. Contextual variables just 0.66%, structural variables 0.79%, and previously unmeasured aspects of the hospitals explained 2% more. The total R2 for the LOS model was 42%. After adjusting for disease status, illness-severity, and individual poverty (through discharge variables), contextual factors, and structural variables the interquartile range for adjusted costs was $7966 ($7625 - $15591) and 3.4 days (3.3 - 6.7) for LOS. Study Two. Hospital-level differences in discharge variables explained 35% of the variation in mean costs per across hospitals. Hospital context variables added 5-21% more, depending on the order variables were entered into the model. Structural variables increased the variation explained by another 3%, to 59%. Hospital-level differences in discharge variables diagnostic groups explained 58% of the variation in mean LOS across hospitals. The addition of hospital context variables increased the variation explained 6%. Structural variables added 13% to a total explained of 77%. Study Three. The mean proportion of patients within hospitals with Medicare as the expected payer was 35% (0-98%). The Spearman's correlation of mean adjusted costs per hospital for discharges with Medicare and mean adjusted costs per hospital for discharges without Medicare was 0.77, p=0.0000. The correlation for LOS was 0.42, p=0.0000. Conclusions. The overarching aim of this dissertation was to examine the amount of variation in discharge level costs and LOS within and between hospitals. Furthermore, an additional aim was to address three criticism of prior variation research. Criticism one was the lack of contextual variables in studies of regional variation of expenditures. When examing costs and LOS, contextual variables were found to explain less than 1% of the variation within and between hospitals. This is reassuring because it would be very troubling if a hospital were found to modify its delivery of care solely based upon the proportion of patients with Medicare, Medicaid, private payer, or uninsured status, the proportion of patients who are female, or the proportion of patients from high, medium, or low income ZIP-codes. Criticism two pertained to the regional nature of prior variation research. These data revealed substantial unexplained variation in both costs and LOS within and between hospitals. The amount of variation in costs at the discharge level explained by covariates was 69% and 41% for LOS at the discharge level. Of variation at the hospital level 59% was due to variation between hospitals in covariates. Of the variation in mean LOS per hospital, 77% was due to hospitals. Criticism three concerned the generalizability of research findings using Medicare expenditures. Study three addressed this concern by testing the correlation between similar patients with and without Medicare as the expected payer for both costs and LOS. There was a high correlation in costs and a moderate correlation in LOS for patients with and without Medicare as the expected payer. It could be argued that a higher correlation should be expected. After all, a 63-year-old patient with a hip fracture and private insurance would be expected to receive similar care as a 67-year-old patient with the same diagnosis and similar illness severity. In total, the three studies address common criticisms of prior research into regional variation in health care expenditures. Expenditures are the product of prices and utilization. Prices are the result of a hospital's costs of care and market forces. This dissertation contains the first known attempts to examine variation in hospital costs. I have shown there to be substantial unexplained variation in adjusted costs and LOS within and between hospitals, which suggests it should be possible for hospitals to reduce costs and LOS in the face of declining reimbursement while still maintain operating margins.