Use of the MMRI-R prognostic tool for older patients discharged to nursing homes from hospital: a prospective cohort study.

Background: the Minimum Dataset Mortality Risk Index-Revised (MMRI-R) is a prognostic score predicting 6-month mortality in US nursing homes. It has not been validated in the UK nor at the hospital-nursing home interface.

Methods: prospective cohort study of consecutive patients discharged from hospital or intermediate care to nursing homes from January 2012 to January 2014. MMRI-R scoring was done prior to discharge and subsequent deaths were ascertained. Calibration plots, receiver operative characteristic curves with area under the curve (AUC) and an optimal cutpoint were obtained. Kaplan-Meier curves were plotted with scores stratified by the cutpoint.

Results: a total of 183 patients were followed up for a median of 230 days. Median age was 87 years and 55.7% were female. Median MMRI-R score was 55. By the end of follow-up, 99 patients (54.1%) were dead. The Hosmer-Lemeshow test showed P-values of 0.4406 for 3-month and 0.8904 for 6-month mortality. The AUC was 0.70 (95% CI: 0.622-0.777) for 3-month death prediction and 0.723 (95% CI: 0.649-0.797) for death at 6 months. Of patients with MMRI-R scores >48 (the cutpoint), 43.6% were dead at 3 months and 53.6% by 6 months. The corresponding figures for scores <48 were 9.6 and 16.4% (P < 0.001, log-rank test).

Conclusion: the MMRI-R can be used at the acute hospital/nursing home interface, and can help predict 3-month and 6-month mortality. The finding of an MMRI-R score of ≥48 should trigger end-of-life discussions.