An investigation of biases in Patient Safety Indicator score distribution among hospital cohorts
Loading...
Date
2016-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
The Ohio State University
Abstract
The Centers for Medicare and Medicaid Services (CMS) have implemented a hospital reimbursement system that incentivizes payment proportional to the quality of care delivered and performance on certain metrics. One such metric is the Agency for Healthcare Research and Quality’s Patient Safety Indicator 90 (PSI-90). It is composed of eight individual indicators designed to flag adverse patient events that are potentially preventable, such as post-operative wound dehiscence and accidental lacerations. CMS publicly reports four of these individual PSI scores (6, 12, 14 and 15) in addition to the composite PSI-90. Previous studies question the PSIs’ validity beyond screening purposes and furthermore question the underlying administrative data’s ability to accurately and reliably flag such events. This study looks to analyze biases in PSI score distribution for hospitals depending on teaching status, differences in patient demographics and lastly, interactions between teaching status and patient demographic factors and their ability to account for differences in PSI rates. Significant differences were found between teaching and non-teaching hospitals for PSIs 6, 12, 15 and 90 (p<0.01). Inpatient volume and patient severity (p<0.01) were found to be significantly different between teaching status cohorts. Lastly, significant differences in PSI scores were found between patient severity quartiles for PSI 6, 15 and 90 (p<0.05) and between socio-economic quartiles for PSI 6, 12, 15 and 90 (p<0.05); but interaction between patient severity and teaching status was only significant for PSI 90 (p<0.05) and between socioeconomic and teaching statuses for PSI 6 (p<0.05). These results indicate current PSI score distributions may be biased against teaching hospitals for 4 out of 5 PSI measures. Further studies will involve assessing the adequacy of risk-adjustment methodology for PSI metrics. Until then, use of PSI metrics to determine federal reimbursement can lead to bias against teaching hospitals.
Description
Denman Research Forum- 2nd Place, Health Professions-Clinical
Keywords
Patient Safety, PSI, Quality, Value-Based Purchasing, Teaching Hospitals