To the Editor:
We read with interest the article by Miu et al1 on a very important issue regarding weaning from mechanical ventilation: the identification of predictors of re-intubation in subjects who have successfully completed a spontaneous breathing trial. We would like to comment on the selection of periods at risk for re-intubation used in this study. Two related aspects should be considered when selecting these periods for the prediction to be the most useful.
First, a prediction of risk for re-intubation assumes a near mechanistic relationship between the condition of the patient on the day of prediction and of extubation failure. How long does this relationship last? Does it include most patients at risk? Does it yield a sufficiently accurate prediction? Although more patients may require re-intubation within the first 24 h than in any subsequent day, the risk is likely to extend beyond this period. The authors indicate a median time to re-intubation of 22 h, but the results suggest that only 41% of all re-intubated patients (155 of 379) were re-intubated in the first 24 h. Time periods for risk of extubation failure of 48 and 72 h have been more commonly studied.2–5 In contrast, after a number of days, unforeseen conditions can develop in critically ill patients and be responsible for a new respiratory failure.3 Can an extubation failure 6 days after extubation be reasonably predicted? We suspect that, after a period of time postextubation, the relationship between the condition on extubation day and extubation failure weakens, the associated risk decreases, and a prediction of high risk for late re-intubation becomes less accurate.
The second aspect to consider is the clinical implication of the prediction. A prediction of high risk for early re-intubation would allow the clinician either to delay extubation until the patient's condition further improves or to apply immediate, more intensive measures after extubation, as suggested by the authors, to minimize such risk. A prediction of high risk for late re-intubation would, however, have less clear implications. Options such as delaying extubation for several days, tracheostomy, and immediate extubation with prolonged intensive respiratory care could affect a sizable number of patients, and resources. Knowing the accuracy of this latter prediction would be very important to assist with these decisions.
The period at risk for early re-intubation of 24 h selected in this study may not include a substantial proportion of re-intubations related to the patient condition on the prediction day. Analyzing the frequency of distribution of time to re-intubation could be informative to select a more inclusive early period at risk. Depending on its accuracy, a prediction for a longer early period could enhance its clinical implication.
The alternative period at risk selected, any time after extubation, may include re-intubations not directly related to the patient condition on prediction day and may be heavily influenced by a majority of re-intubations in the first few days, including the first 24 h. Selecting mutually exclusive early and late periods at risk may better inform whether late re-intubations are predictable.
Footnotes
The authors have disclosed no conflicts of interest.
- Copyright © 2014 by Daedalus Enterprises
References
- 1.
- 2.
- 3.
- 4.
- 5.