Use of hospital-related health care among Health Links enrollees in the Central Ontario health region: a propensity-matched difference-in-differences study.

BACKGROUND:Health Links are a new model of providing care coordination for high-cost, high-needs patients in Ontario. We evaluated use of hospital-related health care services among Health Links patients in the Central Local Health Integration Network (LHIN) of Ontario in the year before versus after program enrolment and compared rates of use with those among similar patients with complex needs not enrolled in the program (comparator group). METHODS:We identified all patients who received a Health Links coordinated care plan before Jan. 1, 2015, using linked registry and health administrative data. We used propensity scores to match (1:1) enrollees (registry) with comparator patients (administrative data). Using a difference-in-differences approach with generalized estimating equations, we evaluated 5 measures of Health Link performance: rates of hospital admission, emergency department visits, days in acute care, 30-day readmissions and 7-day postdischarge primary care follow-up. RESULTS:Of the 344 enrollees in the registry, we matched 313 [91.0%] to comparator patients. All measured sociodemographic, comorbidity and health care use characteristics were balanced between the 2 groups (all standardized differences < 0.10). For enrollees, the rate of days in acute care per person-year increased by 35% (incidence rate ratio 1.35 [confidence interval 1.11-1.65]) after versus before the index date, but differences were nonsignificant for all other measures. Difference-in-differences analyses revealed greater reductions in hospital admissions, emergency department visits and acute care days after the index date in the comparator group than among enrollees. INTERPRETATION:Initial implementation of the Health Link program in the Central LHIN did not reduce selected indicators of Health Link performance among enrollees. As the Health Link program evolves and standardization is implemented, future research may reveal effects from the initiative in other outcomes or with longer follow-up.