Abstract W MP109: Variation In Utilization Of Do-not-resuscitate Orders Between Hospitals In Patients With Intracerebral Hemorrhage: A National Perspective

Abstract

Background: An unfavorable clinical course and limited treatment for intracerebral hemorrhage (ICH) leads to frequent do not resuscitate (DNR) orders. Data on inter-hospital variability of DNR utilization in patients (pts) with ICH is limited. Hypothesis: To analyze the between hospital variation in the utilization of DNR in ICH among low volume and high volume hospitals. Methods: We reviewed the Healthcare Cost and Utilization Project's Nationwide Inpatient Sample (NIS) database from 2011-2012 for ICH admissions in patients >18 years using the ICD 9-CM code 431. We defined DNR status with ICD9CM code V49.86 entered during the same admission as a secondary diagnosis and estimated severity of illness by the 3M™ All Patient Refined DRG (APR DRG) classification System. We calculated ICH annual hospital volume and ranked them into four quartiles. 1st quartile hospitals were defined as low volume and rest as high volume. We generated hierarchical two level multivariate models and calculated Inter class correlation (ICC) and Median Odds Ratio (MOR). Results: We analyzed 25768 patients (weighted estimate126254) with ICH out of which 4620 (18%) patients (weighted estimate 22668) had DNR orders placed. Low volume hospitals accounted for total of 6575 pts (weighted: 32849), out of which 1504 pts (weighted: 7508) had DNR orders. Of the variation in DNR utilization between low volume hospitals, 63.0 % (ICC of 0.63; C statistics: 0.9) was attributable to individual hospital behavior of as compared to 29% in high volume (ICC=0.29; C statistics: 0.8), Also, the MOR was 9.3 for low volume hospitals indicating that a randomly selected patient at a particular hospital would have approximately nine times the chance of having a DNR order placed than an identically matched patient at another. This was three times as high as that for high volume hospitals (MOR=3). All models were adjusted for patient level, hospital level characteristics, APR DRG severity scale and other clinical characteristics. Conclusion: In conclusion, there is significant individual variation between hospitals in DNR utilization. Variation is three times higher in low volume hospitals as compared to high volume hospitals. Further attempts should be made to understand this inter-hospital difference.