Preliminary derivation of a nursing home confusion assessment method based on data from the minimum data set

OBJECTIVES: To develop a Nursing Home Confusion Assessment Method (NH-CAM) for diagnosing delirium using items found on the Minimum Data Set (MDS) and to compare its performance with that of the delirium Resident Assessment Protocol (RAP) trigger and to an additive score of six of the RAP items. DESIGN: Retrospective cohort study using MDS and Medicare claims data. SETTING: Free-standing NHs in urban markets in the 48 contiguous U.S. states. PARTICIPANTS: Long-stay residents who returned to their NHs after acute hospitalizations between April and September 2000 (N=35,721). MEASUREMENTS: Mortality and rehospitalization rates within 90 days of readmission to the NH from the hospital. RESULTS: Almost one-third (31.8%) of the residents were identified as having delirium according to the RAP; 1.4% had full delirium, 13.2% had Subsyndromal II delirium, and 17.2% had Subsyndromal I delirium. More-severe NH-CAM scores were associated with greater risks of mortality and rehospitalization. NH-CAM levels were strong independent risk factors for survival and rehospitalization in a Cox model (hazard ratios ranging from 1.5 to 1.9 for mortality and 1.1 to 1.3 for rehospitalization) adjusting for cognitive and physical function, diagnoses, inpatient care parameters, care preferences, and sociodemographic factors. CONCLUSION: The NH-CAM successfully stratified NH residents' risk of mortality and rehospitalization. If validated clinically, the NH-CAM may be useful in care planning and in further research on the determinants and consequences of delirium in the NH.