To the Editor:
I read with great interest the article by Östberg et al.1 which provides clinically relevant information. The authors and the accompanying editorial2 adequately mention the main limitation of the study (i.e., findings are restricted to a healthy population undergoing low-risk surgery of relatively short duration). My concern relates to the data presentation and interpretation.
Outliers may affect results, assumptions, and conclusions. Labeling an observation as an outlier implicitly suggests that it is an unlikely observation, is not the result of natural variability, and that a similar observation is unlikely to occur in the future under comparable conditions. The authors provide no explanation for designating nine data points from eight patients (27% of all patients!) as outliers. In the absence of a possible explanation for why the outliers might have occurred, such an approach seems questionable. The fact that the outliers lie outside the ranges (fig. 2) would suggest that they were not included in the overall data analysis. Especially in studies with a rather small number of patients, it would be preferable to treat the outliers like all other data because those “extremes” may well reflect natural variability. In this context, the lack of the upper range whisker of the baseline values in the Zero PEEP [positive end-expiratory pressure] group (fig. 2) prevents an adequate assessment of how “extreme” the outliers really were. Overall, the rationale for designating certain observations as outliers remains unclear. At a minimum, the method of defining an outlier needs to be described, labeling of the observation as an outlier be justified, and the implication of using outliers be spelled out.
The authors interpret their findings as showing that postoperative atelectasis was small without effect on oxygenation. This interpretation is based on median values. However, the ranges of areas of atelectasis and Pao2/Fio2 ratios were large, reflecting considerable individual variability. The area of atelectasis increased by 7 cm2 or more in two patients of each group. It would clinically be relevant to know the oxygenation status of these particular patients.
The statement that most patients had very small areas of atelectasis at baseline is not fully supported by the data. Two patients in the PEEP and three in the Zero PEEP group had baseline atelectasis areas of approximately 8.5 to 9.5 cm2 (PEEP group) and 5.5 to 6.5 cm2 (Zero PEEP group). This constitutes 17% of the entire population.
The overall evidence suggests a detrimental effect of an Fio2 1.0 before extubation on the development of postoperative atelectasis. Why not abandoning administration of 100% oxygen before extubation altogether? Best practice may well consist of administrating an Fio2 of maximally 0.8 before emergence from anesthesia.3–5 A patient who requires an Fio2 greater than 0.8 before extubation should probably not be extubated. “The knowledge is there! Yet, the tradition is long lived, and we still frequently practice 100% oxygen before extubation.”3
The author declares no competing interests.