External validation of the detection of indicators and vulnerabilities for emergency room trips (DIVERT) scale: a retrospective cohort study Journal Articles uri icon

  • Overview
  • Research
  • Identity
  • Additional Document Info
  • View All


  • Abstract Background The Detection of Indicators and Vulnerabilities of Emergency Room Trips (DIVERT) scale was developed to classify and estimate the risk of emergency department (ED) use among home care clients. The objective of this study was to externally validate the DIVERT scale in a secondary population of home care clients. Methods We conducted a retrospective cohort study, linking data from the Home Care Reporting System and the National Ambulatory Care Reporting System. Data were collected on older long-stay home care clients who received a RAI Home Care (RAI-HC) assessment. Data were collected for home care clients in the Canadian provinces of Ontario and Alberta, as well as in the cities of Winnipeg, Manitoba and Whitehorse, Yukon Territories between April 1, 2011 and September 30, 2014. The DIVERT scale was originally derived from the items of the RAI-HC through the use of recursive partitioning informed by a multinational clinical panel. This scale is currently implemented alongside the RAI-HC in provinces across Canada. The primary outcome of this study was ED visitation within 6 months of a RAI-HC assessment. Results The cohort contained 1,001,133 home care clients. The vast majority of cases received services in Ontario (88%), followed by Alberta (8%), Winnipeg (4%), and Whitehorse (< 1%). Across the four cohorts, the DIVERT scale demonstrated similar discriminative ability to the original validation work for all outcomes during the six-month follow-up: ED visitation (AUC = 0.617–0.647), two or more ED visits (AUC = 0.628–0.634) and hospital admission (AUC = 0.617–0.664). Conclusions The findings of this study support the external validity of the DIVERT scale. More specifically, the predictive accuracy of the DIVERT scale from the original work was similar to the accuracy demonstrated within a new cohort, created from different geographical regions and time periods.


  • Mowbray, Fabrice I
  • Jones, Aaron
  • Schumacher, Connie
  • Hirdes, John
  • Costa, Andrew

publication date

  • December 2020