Haberthur 2002.
| Methods | Cross‐over controlled trial Setting: Medical intensive care unit of the university hospital of Basel, Switzerland Sample size: 38 patients were needed in each group to detect, at a power of 80%, a difference in extubation outcome of 30%, with a two‐tailed alpha error of 0.05 |
|
| Participants | ‐ 90 patients with improved or resolved the underlying cause of acute respiratory failure receiving ventilatory support for more than 24 h ‐ Mean age: 57 years ‐ Male: 58; Female: 32 ‐ Mechanical ventilation time: mean of 6.3 days ‐ Mean APACHE II score: 20 ‐ There was no significant difference between comparison groups at baseline |
|
| Interventions | 1‐ T‐tube (n=30) 2‐ Pressure support of 5 cmH2O and PEEP of 5 cmH2O (n=30) 3‐ Automatic tube compensation, PEEP of 5 cmH2O (n= 30) |
|
| Outcomes | 1‐ Successful spontaneous breathing trial 2‐ Successful extubation 3‐ Extubation failure 4‐ Respiratory rate (RR) 5‐ Tidal volume (VT) 6‐ Rapid shallow breathing index (RR/VT) 7‐ Minute ventilation (VE) 8‐ Inspired oxygen fraction (FiO2) 9‐ Partial arterial oxygen pressure (PaO2) 10‐ CO2 pressure arterial (PaCO2) |
|
| Notes | Group 3 was not considered for analysis in this review Part I of the study: In patients who tolerated the 2h breathing trial well, the extubation was performed. ( we used the outcomes of part 1 these study) Part II of the study: In patients who did not tolerate the spontaneous breathing trial, another or other two modes were tried thereafter in random order Successful extubation was considered if there was no need for ventilatory assistance and/or reintubation within 48h following extubation We excluded the outcomes 7, 8, 9 and 10 the review |
|
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | Before the start of the study all (future) patients were allocated to one out of the three “ventilator modes” based on a computer‐generated random list |
| Allocation concealment (selection bias) | Low risk | Patients were randomly assigned in a blinded fashion using opaque, sealed envelopes to undergo a trial of spontaneous breathing |
| Selective reporting (reporting bias) | Low risk | Outcomes were clinically relevant. There was neither explicit nor implicit preference from authors to one or another method that could force then to elect specific outcomes |
| Other bias | Unclear risk | The study design can to be one potential risk of bias (cross‐over) |
| Free of detection bias ? | Low risk | Outcome data were assessed from the medical records by another person, not involved in the study (i.e. Rolf Bingisser). Data reported by author |
| Free of attrition bias? | Low risk | The patient flow was clear and there were no reported dropouts |