Natural Experiment Outcomes Studies in Rotor Wing Air Medical Transport: Systematic Review and Meta-Analysis of Before-and-After and Helicopter-Unavailable Publications From 1970 to 2022

Abstract

Objective: Helicopter emergency medical services (HEMS) is widely used for prehospital and interfacility transport, but there is a paucity of HEMS outcomes data from studies using randomized controlled trial designs. In the absence of robust randomized controlled trial evidence, judgments regarding HEMS potential benefit must be informed by observational data. Within the study design set of observational analyses, the natural experiment (NE) is notable for its high potential methodologic quality; NE designs are occasionally denoted "quasi-experimental." The aim of this study is to examine all NE outcomes studies in the HEMS literature and to discern what lessons can be learned from these potentially high-quality observational data.

Methods: HEMS NE studies were identified during the development of a new HEMS Outcomes Assessment Research Database (HOARD). HOARD was constructed using a broad-ranging search of published and gray literature resources (eg, PubMed, Embase, and Google Scholar) that used variations of the terms "helicopter EMS," "air ambulance," and "air medical transport." Among the 221 studies ultimately included in HOARD, 16 NE publications describing 13 sets of observational data comprising myriad diagnostic groups were identified. Of these 16 HEMS NEs, 4 HEMS NE studies assessing trauma outcomes were used in a meta-analysis. A meta-analysis was also performed of 4 HEMS NE studies.

Results: Although the disparity of studies (in terms of both case mix and end points) precluded the generation of a pooled effect estimate of an adjusted mortality benefit of HEMs versus ground emergency medical services, HEMS was found to be associated with outcomes improvement in 8 of the 13 cohorts.

Conclusion: The weight of the NE evidence supports a conclusion of some form of HEMS-mediated outcomes improvement in a variety of patient types. Meta-analysis of 4 HEMS NE studies assessing trauma outcomes generated a model with acceptable heterogeneity (I² = 43%, Q test: P = .16), which significantly (P < .01) favored HEMS use with a pooled HEMS survival odd ratio estimate of 1.66 (95% confidence interval, 1.23-2.22).