Keith Couper, Chen Ji, Charles D Deakin, Rachael T Fothergill, Rebecca Kandiyali, Jerry P Nolan, Helen Pocock, Michael A Smyth, Jason V Wiles, Ranjit Lall, Gavin D Perkins
Based on non-informative priors which represent minimal prior knowledge of the treatment effect, an intraosseous-first strategy in adult out-of-hospital cardiac arrest is more likely to be harmful, rather than beneficial. However, the low probability of a moderate or large treatment effect suggests that the overall treatment effect is likely to be modest.
OBJECTIVE: To evaluate the clinical-effectiveness of an intraosseous-first strategy compared with an intravenous-first strategy in adult out-of-hospital cardiac arrest in a pre-planned Bayesian secondary analysis of the PARAMEDIC-3 trial.
METHODS: In PARAMEDIC-3, adults with out-of-hospital cardiac arrest requiring vascular access were randomised to either an intraosseous-first strategy or an intravenous-first strategy. In this analysis, we describe the probability that an intraosseous-first strategy was superior or inferior to an intravenous-first strategy, and the probability that this exceeds specific odds ratio thresholds. Our outcomes were 30-day survival (primary), favourable neurological outcome at hospital discharge and return of spontaneous circulation at hospital handover. Our primary analysis used non-informative priors, adjusted for key baseline characteristics.
RESULTS: In 6,082 randomised participants, the posterior probability that an intraosseous-first strategy improved 30-day survival (odds ratio (OR) > 1.0) was 18.4%. The probability that an intraosseous-first strategy was inferior to an intravenous-first strategy was 81.6% for any effect (OR<1.0), 55.0% for a small effect (OR<0.9), and 13.4% for a moderate effect (OR<0.75). For favourable neurological outcome, findings were comparable to 30-day survival. For return of spontaneous circulation, the posterior probability that an intraosseous-first strategy was inferior to an intravenous-first strategy (OR < 1.0) was 98.0%.
CONCLUSION: Based on non-informative priors which represent minimal prior knowledge of the treatment effect, an intraosseous-first strategy in adult out-of-hospital cardiac arrest is more likely to be harmful, rather than beneficial. However, the low probability of a moderate or large treatment effect suggests that the overall treatment effect is likely to be modest.
TRIAL REGISTRATION: ISRCTN14223494.