Trauma is one of the top ten causes of disability worldwide, accounting for approximately 109 million disability-adjusted life-years (DALYs) globally in 2021. In England and Wales, it is the leading cause of death in children and adults under 40, costing the U.K. an estimated £3.3–3.7 billion annually. Despite widespread recognition of the importance of prompt intervention in severe injury, the survival benefit of prehospital emergency anaesthesia with intubation has remained uncertain.
Randomised controlled trials are extremely difficult to conduct in the prehospital trauma setting due to the near-impossibility of obtaining patient consent and achieving clinical equipoise. The only existing RCT of prehospital intubation was limited to severe traumatic brain injury in 312 patients and found improved neurological outcomes but no mortality benefit, and its conditions lacked generalisability. Prior observational studies have been rated as very low quality, with high risk of bias and conflicting results.
A new study addressed this evidence gap by estimating the causal effect of prehospital intubation on 30-day survival using advanced causal modelling methods applied to routinely collected observational data.
The study analysed prehospital clinical data from 6,467 patients admitted to Southmead Hospital, Bristol, a major trauma centre in the U.K., across two time periods: February 2012 to March 2017 and April 2017 to November 2019. Data were drawn from the UK Trauma Audit and Research Network (TARN) database, the largest European trauma registry. Inpatient trauma cases, patients transferred from other hospitals, and those with no prehospital data were excluded.
From an initial 3,301 variables, 210 prehospital features were retained after filtering for data completeness and relevance. Machine learning models were trained to predict two outcomes: early intubation (prehospital or in the emergency department) and 30-day mortality.
The model predicted early intubation and 30-day mortality with excellent accuracy on prospective, unseen test data. The top predictor of early intubation was total Glasgow Coma Scale score at scene; the top predictor of 30-day mortality was patient age.
Patients predicted by the model to require early intubation but who did not receive prehospital intubation had substantially lower 30-day survival than those not predicted to require it (71.4% vs 94.4%). This difference persisted even when accounting for predicted 30-day mortality and regardless of whether a doctor or helicopter emergency medical service (HEMS) team was present at scene.
In the high-risk group identified by Intub-8 (n=229), the conditional average treatment effect (CATE) of prehospital intubation was −0.103, representing a 10.3% reduction in 30-day mortality, equating to 28 lives saved among 229 high-risk patients. In the low-risk group, the effect was much smaller (0.7% reduction). Naïve intubation of all patients without risk stratification produced a far smaller average treatment effect, confirming that targeted deployment is essential to maximising benefit.
Scaled to UK-wide trauma incidence, the risk-stratifying model is projected to prevent 170 deaths per year, approximately half the estimated annual survival benefit of major trauma centres. The annual QALY return of 3,395 translates, at NICE cost-effectiveness thresholds, to a value of £101 million per year. The estimated cost per QALY was £89.32 based on prehospital costs alone, rising to £1,110 when post-intubation care is included. The analysis concludes that the £62.5 million annual cost of U.K. HEMS services is economically justified by targeted prehospital intubation alone.
The authors present this as, to their knowledge, the highest level of evidence to date on the survival benefit of prehospital intubation in major trauma. The study argues for a re-evaluation of prehospital service provision and makes a strong health economic case for central funding of specialist critical care teams with prehospital emergency anaesthesia capability.
Source: The Lancet
Image Credit: iStock
