Determinants of time to decannulation and predictors of early weaning from tracheostomy: a multicenter, retrospective Italian cohort study.

Background: Weaning from tracheostomy is largely left to experts' opinion. Shared and validated protocols for decannulation are lacking, and procedures during the weaning process depend upon clinical judgement. Determinants of tracheostomy decannulation in real life are largely unknown. Methods: This was a retrospective, observational, multicenter study. Patients that underwent endotracheal intubation, percutaneous or surgical tracheostomy and at least a decannulation trial between 2017 and 2023 were recruited from five academic hospitals in Italy. Clinical characteristics, procedures, pre-decannulation respiratory and biochemistry parameters, respiratory support, Quantitative semi-Quantitative clinical score (QsQ), and in-hospital outcomes were collected. Patients were operationally divided in early (<20), average (20-40), and late (>40 days) decannulation. The aim was to assess predictors of faster decannulation. Secondary outcomes included: weaning failure, instrumental procedures during weaning, accuracy of QsQ criteria for decannulation failure. Results: The final analysis included 191 patients (26.7% males, median age 63 years), of which 79.6% had at least one comorbidity and 61.2% were intubated for Coronavirus Disease 2019 pneumonia. Decannulation was successful in 183 patients (95.8%) and failed in 8 (4.2%). Early, average and late decannulation was observed in 23.3, 31.7 and 45% of patients, respectively. Weaning was faster if patients underwent bronchoscopy (Log-rank P = 0.044), and longer if swallowing efficiency was assessed (Log-rank P = 0.001). Reduction of cannula caliber (OR 4.224, 95%CI: 1.037-17.207; P = 0.044) predicted earlier decannulation, while swallowing assessment predicted slower decannulation (OR 0.161, 95%CI: 0.037-0.694; P = 0.014). Patients' baseline clinical characteristics and instrumental procedures didn't differ in patients that failed and that succeeded weaning. Successes had a lower bronchial secretion burden (P = 0.012) and tended to develop less frequently tracheal stenosis (P = 0.072). Sufficient data availability for QsQ score calculation were limited to some of the major and minor criteria. Conclusions: Specific procedures were associated with reduced time to decannulation, however weaning rarely failed likely because of late weaning initiation. Results generalizability could be limited by younger age, low comorbidity burden and high prevalence of COVID-19 pneumonia. Trials investigating per-protocol weaning to detect the most performant screening procedures are required. Trial registration: The study protocol has been registered and approved by ClinicalTrials.gov February the 28th 2022 (NCT05271786).