Objective: To examine NFR orders in relation to adverse events and emergency team calls in hospitals with or without a Medical Emergency Team (MET) system during the MERIT study.
Method: Within a cluster randomized controlled trial (the MERIT study), examining the effect of introducing a MET system, we recorded NFR orders in relation to adverse events and emergency team calls. We compared the proportion and rate of NFR orders issued in relation to "adverse events" and "adverse event-free emergency team calls" in hospitals with or without a MET system.
Results: Information on NFR orders was available for 3650 patients who died, 1466 patients who had an unplanned ICU admission, 574 patients who suffered a cardiac arrest and 1529 patients who had a adverse event-free emergency team call. Close to 90% of deaths occurred in patients with a previously documented NFR order. Only approximately 4% of cardiac arrests had a previously documented NFR order. In patients with unplanned ICU admission, NFR orders were present in approximately 3% of cases. An NFR order was issued at the time of an "event" in 3.85% of cases in MET hospitals compared with 1.72% in control hospitals (OR=2.29; 95% CI: 1.31-4.01; p=0.005). This difference was mostly due to a greater proportion of patients being made NFR in MET hospitals at the time of a "adverse event-free" emergency team call (7.96% vs. 3.05%; OR=2.75; 95% CI: 0.97-7.80; p=0.048). The number of NFR orders issued at the time of a serious adverse event-free emergency team call was 10 times higher in MET hospitals (0.398 vs. 0.041 per 1000 admissions; weighted absolute risk difference: 0.49 (95% CI: 0.20-0.78; p=0.002). Multivariate models could only account for less than 50% of the variance in the issuing of NFR orders.
Conclusions: In a cohort of Australian hospitals, most deaths occurred in patients with a previously documented NFR order but NFR orders were uncommon before cardiac arrest calls or unplanned ICU admissions. During the conduct of a cluster randomised controlled trial, more NFR orders were issued by emergency teams in those hospitals that implemented a MET system than in control hospitals. MET allocation, teaching hospital status, number of hospital beds and metropolitan location could only explain less than 50% of variance in NFR orders.