Automated Identification of Adults at Risk for In-Hospital Clinical Deterioration

Escobar, G. J.

doi:10.1056/NEJMsa2001090

Escobar Advance Alert Monitor

Automated Identification of Adults at Risk for In-Hospital Clinical Deterioration

Patient / Population Intervention / Exposure Comparison Outcome

In 43,949 non-ICU hospitalisations reaching the alert threshold, automated deterioration alerts reviewed by remote nurses and escalated reduced 30-day mortality after an alert compared with usual care where the system was not yet deployed.

N

43,949patients

Design

Staggered deployment, non-randomised (19 hospitals)

Endpoint

Mortality within 30 days after an alert

Relevance

3Supporting — context rather than must-know.

Result30-day mortality after an alert lower in the intervention cohort (adjusted RR 0.84, 95% CI 0.78–0.90; P<0.001); 15,487 intervention vs 28,462 comparison hospitalisations, drawn from 548,838 non-ICU hospitalisations.

Escobar GJ, et al. N Engl J Med. 2020;383(20):1951-1960. 10.1056/NEJMsa2001090

Discussion & critique

The first large-scale report that an automated, EHR-based deterioration model wired into a clinical workflow — remote nurse review plus rapid-response teams across Kaiser Permanente hospitals — was associated with lower mortality, and the template for later randomised tests such as CONCERN. Deployment was staggered but not randomised, so the comparison with not-yet-deployed hospitals rests on multivariable adjustment and is vulnerable to secular trends and site differences.