Dispatcher and volunteer-app package raised bystander CPR in French trial but not outcomes

A cluster-randomised trial newly indexed in PubMed from The Lancet Public Health reports that a bundle of dispatcher training, volunteer smartphone alerts and agency feedback was associated with more bystander CPR after out-of-hospital cardiac arrest in France. The authors also reported that the gain "did not translate into improved clinical outcomes" ^[1].

The study, by Debaty and co-authors, used a stepped-wedge, cluster-randomised, controlled design across French emergency medical service agencies. Enrollment ran from Aug 27, 2018 to Feb 20, 2022, according to the abstract. The primary outcome was bystander-initiated CPR before EMS arrival ^[1].

Bystander CPR occurred in 61.4% of control patients and 70.9% of intervention patients. The adjusted odds ratio was 1.54, with a 95% confidence interval of 1.09 to 2.19 and a p value of 0.017. The authors wrote that the estimate "was sensitive to the modelling of secular trends" ^[1].

What the records show

The abstract describes the intervention as three parts delivered together. Dispatchers were trained to recognise cardiac arrest early and to give telephone-assisted CPR. Trained volunteers were activated through a geolocation smartphone application. Each EMS agency received structured audit and feedback ^[1].

A dedicated protocol told dispatchers to identify cardiac arrest from unresponsiveness and abnormal breathing, then to give scripted compression-only instructions and activate volunteers. The protocol set a target of recognition within 60 s of call pick-up and first compression within a further 60 s ^[1].

The trial randomised 15 French EMS agencies, stratified by tercile of the cardiac arrest volume each agency managed in 2015. Each centre had a 1-month transition period before its intervention period began. Two agencies withdrew before recruitment started and were replaced, and one enrolled no patients. In the end, 14 agencies recruited 2160 patients ^[1].

The intention-to-treat analysis was limited to the 12 originally randomised agencies and included 1684 patients, 686 in the control group and 998 in the intervention group. Patients were adults aged 18 or older with non-traumatic out-of-hospital cardiac arrest. Mean age was 67 years, with a standard deviation of 15, and 1205 patients, or 71.6%, were male ^[1].

The authors ran six prespecified sensitivity analyses. The odds ratios ranged from 1.38 to 1.55 and stayed statistically significant in four of them. The result was not significant after adjustment for cluster-specific secular trends, where the odds ratio was 1.38 with a confidence interval of 0.98 to 1.94. It was also not significant after adjustment for baseline characteristics, at 1.39 with a confidence interval of 0.97 to 1.98 ^[1].

The abstract states that "No intervention-related adverse events were reported." The trial was prospectively registered on ClinicalTrials.gov as NCT03633370 and is listed as completed. The funder named in the abstract is the Direction Générale de l'Offre de Soins ^[1].

Why it matters for EMS

The trial tested the dispatch centre and the community, not the responding ambulance crew. The unit of change was the EMS agency, and the levers were a recognition script, compression-only telephone instructions, a volunteer app and feedback to the agency itself ^[1].

The authors describe bystander CPR as "a key determinant of survival" after cardiac arrest, and they note that rates "remain suboptimal worldwide" ^[1]. Their own data show a higher bystander CPR rate under the intervention, yet the abstract says the change did not improve clinical outcomes. That gap between process and outcome is the central finding for agencies weighing similar programs ^[1].

Because the three components were bundled, the record does not separate the contribution of dispatcher training from the volunteer app or the audit-and-feedback loop. The stepped-wedge design also means every agency eventually received the intervention, which is why the authors tested whether background trends over time could explain the result ^[1].

What the records do not tell us

• The abstract says clinical outcomes did not improve but does not list which outcomes were measured or report their numbers. No survival or neurological figures appear in the record ^[1].
• The record does not report how often dispatchers met the 60 s recognition and compression targets, or how many volunteers were alerted and reached a patient ^[1].
• The abstract does not explain why the two agencies withdrew before recruitment, or why the one agency enrolled no patients ^[1].
• The intention-to-treat analysis covers the 12 originally randomised agencies rather than all 14 that recruited patients. The abstract does not describe results for the replacement agencies ^[1].
• The trial was conducted in France. The record contains no US data and does not address how the French dispatch model compares with other systems ^[1].
• Only the abstract is in the records. The full methods, tables and secondary outcomes were not reviewed ^[1].

What to watch

The authors wrote that the findings "support further investigation of coordinated community-based strategies" aimed at the early links in the chain of survival. The abstract names no follow-up study, date or decision ^[1].

This is news, not medical advice. EMS clinicians: follow your local protocols and medical direction.