Abstract
Better prediction of hospitalisations may allow preventive interventions directed to appropriate target groups. We compared the predictive accuracy of eight existing risk scores to hospitalisations among older care dependent home dwelling adults across six countries from the IBenC study.
We assessed 2884 persons aged 65 or older, who received professional homecare in six different countries in Europe and followed them for 6 months. Eight existing index risk scores were computed with baseline data: (1) The Changes in Health, End-stage Disease, Signs, and Symptoms Scale (CHESS); (2)Detection of Indicators and Vulnerabilities for Emergency Room Trips (DIVERT); (3); (4) Identification Seniors At Risk Primary Care (ISAR PC); (5) Emergency admission risk likelihood index (EARLI); (6) Sherbrooke Postal Questionnaire (SPQ); (7) the Elders Risk Assessent (ERA), and (8) Community Assessment Risk Screen (CARS). Their accuracy to predict one or more reported hospitalisations or Emergency Department visits was expressed in the area under the ROC curve (AUC).
194 older adults were admitted at the ED and/or hospital ward during the six-month study period. The highest AUC value was found for the EARLI AUC=0.75, followed by DIVERT (AUC =0.69), CHESS (AUC =0.66), CARS (AUC =0.64), ERA (AUC=0.60), ISAR-PC (AUC =0.47) and SPQ (AUC =0.41). Significantly better AUC value were found in persons without a recent admission at baseline for DIVERT,EARLI and CARS risk scores.
Our results reveal two promising risk scores: EARLI, and DIVERT. Identification by these risk scores may help to target preventive intervention in high-risk groups.