163 Study Design Influences Effect Size: Systematic Review and Meta-Analysis of Mobile Integrated Health Community Paramedicine Programs’ Effect on Emergency Department Visits

Abstract

Mobile Integrated Health Community Paramedicine (MIH-CP) programs are designed to reduce emergency department (ED) usage and increase access to care for vulnerable, underserved populations. Multiple MIH-CP systematic reviews have reported wide ranges of interventions, effect sizes, and a high prevalence of biased methods. The objectives of this study were to review MIH-CP interventions and perform a meta-analysis with a focus on the impact of study design on the magnitude of the observed effect. We hypothesized study design would influence effect size. We queried Pubmed, Embase, CINAHL, and Scopus databases for peer-reviewed MIH-CP literature from January 1, 2000, to July 31, 2020. We included all full-text English studies whose program met the National Association of Emergency Medical Technicians definition, had an MIH-CP related intervention and outcome, and was implemented through Emergency Medical Services. Our primary outcome was ED visits. We interpreted data to establish risk ratios for each study and performed a random-effects meta-analysis. We further performed a cumulative meta-analysis, tests of heterogeneity, and a moderator analysis to assess for factors influencing the magnitude of the observed effect. We included eight studies in our meta-analysis. All studies were observational; two used matched controls, two pre-post, and four no controls. Five studies’ interventions were diversion/triage: three provided acute care visits upon EMS activation and transported patients as indicated, one provided acute care “after- hours,” and one performed health clearance for mental health calls with direct transport to a mental health facility when appropriate. Three studies intervened with health education/home primary care services: two provided resources and regular follow-up for post-discharge patients, one provided similar service to patients with high ED utilization. Pooled risk ratio (RR) was 0.46 (95% confidence interval [CI] 0.25-0.83, I-square = 99%). Cumulative meta-analysis revealed MIH-CP programs began to show reductions in ED visits as of 2018. Moderator analysis revealed difference in effect size as a function of study design: matched controls RR 0.45 (95% CI 0.12-1.7), pre-post RR 1.03 (95% CI 0.27-4.0), uncontrolled RR 0.29 (95% 0.11-0.77). Study design has a major influence on the magnitude of observed effect in MIH-CP studies. While our pooled meta-analysis appeared to suggest a reduced risk of ED visits for patients engaged in MIH-CP programs, there was high heterogeneity among studies, and moderator analysis revealed a significant effect was observed only in uncontrolled studies. Non-significance may be due in part to small sample sizes. Future studies can focus on using appropriate controls to identify interventions’ effects more accurately.