Interventions to improve outcomes for high-need, high-cost patients: A systematic review and meta-analysis


Advocate Aurora Research Institute


Background: Chronic disease patients who are the greatest users of healthcare services are often referred to as high-need, high-cost (HNHC). Payers, providers, and policymakers in the United States are interested in identifying interventions that can modify or reduce preventable healthcare use among these patients, without adversely impacting their quality of care and health. We systematically reviewed the evidence on the effectiveness of complex interventions designed to change the healthcare of HNHC patients, modifying cost and utilization, as well as clinical/functional, and social risk factor outcomes.

Methods: We searched 8 electronic databases (January 2000 to March 2021) and selected non-profit organization and government agency websites for randomized controlled trials and observational studies with comparison groups that targeted HNHC patients. Two investigators independently screened each study and abstracted data into structured forms. Study quality was assessed using standard risk of bias tools. Random-effects meta-analysis was conducted for outcomes reported by at least 3 comparable samples.

Results: Forty studies met our inclusion criteria. Interventions were heterogenous and classified into 7 categories, reflecting the predominant service location/modality (home, primary care, ambulatory intensive caring unit [aICU], emergency department [ED], community, telephonic/mail, and system-level). Home-, primary care-, and ED-based interventions resulted in reductions in high-cost healthcare services (ED and hospital use). ED-based interventions also resulted in greater use of primary care. Primary care- and ED-based interventions reduced costs. System-level transformation interventions did not reduce costs.

Discussion: We found limited evidence of intervention effectiveness in relation to cost and use, and additional evidence is needed to strengthen our confidence in the findings. Few studies reported patient clinical/functional or social risk factor outcomes (e.g., homelessness) or sufficient details for determining why individual interventions work, for whom, and when. Future evaluations could provide additional insights, by including intermediate process outcomes and patients' experiences, in assessing the impact of these complex interventions.

Prospero registration number: CRD42020161179.

Document Type


PubMed ID