Interobserver agreement rate of the spontaneous breathing trial

Purpose: During the mechanical ventilation weaning process, the spontaneous breathing trial (SBT) is the confirmatory test of patients' capability to breathe unassisted. However, the SBT interobserver agreement rate (its reliability) is unknown, and our objective was to evaluate it.

Materials and methods: This is a prospective, multicentric and observational study. Patients were included when the SBT criteria were fulfilled. Two physicians and 2 respiratory therapists (RTs) rated each SBT. The SBT interobserver agreement was measured using κ statistic and also the percentage of agreement with its 95% credible interval (CrI) calculated by a Bayesian inference.

Results: Ninety-three distinct physicians and 91 distinct RTs rated 130 SBTs. The κ coefficient was 0.46 for physicians and 0.57 for RT, indicating a moderate interobserver agreement rate. The percentage of agreement was 87.7% between physicians (95% CrI, 81.0%-92.3%) and 86.2% between RT (95% CrI, 79.2%-91.1%). The physicians' and RT' percentage of agreement were not statistically different (P = .71).

Conclusions: The SBT interobserver agreement rate is only moderate for physicians and RT. The percentage of agreement between 2 different SBT observers is 79.2% to 92.3%. Therefore, a relevant percentage of patients will have different extubation decisions depending on the SBT observer.