Mortality Prognostication in Long-Term Care Residents: The MDS-CHESS Scale

Quality end of life (EOL) care for the elderly continues to be a challenge, in part due to late referral for palliative and/or hospice services. The purpose of this study was to determine the accuracy of the MDS-CHESS scale (Hirdes, Frijters & Teare, 2003) in predicting 6-month mortality in a stable American nursing home (NH) population. A secondary aim was to determine any differences in mortality between the 2 sites. A sample of 191 residents of two NHs was analyzed in a retrospective, correlational cohort study, using data collected from the federal MDS database. Correlational statistics and logistic regression were utilized to determine relationships between predictive variables, and to examine the overall quality of the regression model. The items that comprise the MDS-CHESS scale (DNR status, daily pain, IV site, IV medications, oxygen, suctioning, Physician\u27s visits, Physician\u27s orders and abnormal labs) demonstrated weak relationships among some of variables, although p-values were significant. The predictive model was analyzed on a final sample of 81 subjects who died during the evaluation period. Regression coefficients were generally low (range 0.09-0.46) and none demonstrated significance in the likelihood ratio test. Odds ratios were uniformly low (range 0.52-2.26) as well. Differences between the 2 NH were negligible in the correlational analysis, as well as the logistic regression statistics and overall mortality. This study examined the accuracy of the MDS-CHESS scale in predicting death in a cohort of NH residents, and determined differences between the 2 NH. Early identification of elderly NH residents at risk of dying can improve the quality of care by promoting palliative care and hospice referral in a timely manner. Reduction of suffering through determination of goals of care and appropriate therapeutic interventions remains an important priority in all settings