OBJECTIVE: To evaluate the diagnostic accuracy of 2 threshold values (105 breaths per minute [bpm]/L and 130 bpm/L) of the rapid shallow breathing index (RSBI) to predict a successful weaning trial outcome in a homogenous group of patients with chronic obstructive pulmonary disease (COPD).METHODS: A consecutive population of patients with COPD who were intubated for hypercapnic respiratory failure during a 2-year period were studied prospectively. RSBI was measured by 2 investigators at minute 5 of the T-piece trial, whereas 2 other physicians evaluated the 30 minute T-piece trial as successful or unsuccessful, according to clinical criteria.RESULTS: Of 64 patients with COPD (53 male, 11 female) who constituted the study population, 42 patients (35 male, 7 female; aged 70 ± 9.2 years) completed the spontaneous breathing trial (SBT) and remained clinically stable (group 1). The remaining 22 patients (18 male, 4 female; aged 71.9 ± 4.7 years) had to return to ventilatory support by the end of the SBT because of clinical deterioration (group 2). The 2 threshold values that were evaluated had low specificity (38.1% for < 105 bpm/L and 66.7% for < 130 bpm/L), low sensitivity (63.6% for < 105 bpm/L and 54.5% for < 130 bpm/L), and low diagnostic accuracy (46.8% for < 105 bpm/L and 65.6% for < 130 bpm/L) in predicting a successful T-piece trial outcome.CONCLUSION: RSBI measured early during an SBT cannot accurately predict the successful outcome of a T-piece trial in a homogenous population of patients with COPD.