Reply to: Zero-heat flux thermometry.
Editor, We appreciate the interest and the comments by Sessler and Mascha1 on our study ‘Intraoperative temperature monitoring with zero heat flux technology (3 M SpotOn sensor) in comparison with sublingual and nasopharyngeal temperature’.2 Indeed, in our study, a relatively large number of patients were excluded from the final analysis because of a calibration failure of the sublingual probe. For this study, we bought a new probe with optimised accuracy. Unfortunately, the initial temperature data obtained were fluctuating and unreliable. A technical check-up of the device revealed a ‘calibration failure’ malfunction. Therefore, we decided to exclude these 19 patients from the final analysis, which is stated clearly in the article. We studied elective patients undergoing orthopaedic surgery (mainly arthroscopic surgery of the shoulder, hip or knee), or gynaecological surgery (mainly laparoscopic interventions). Among these American Society of Anesthesiologists (ASA) status 1 and 2 patients included in the final analysis, none had to be excluded because of postoperative need for ventilation, unexpected blood loss (need for blood transfusion) or haemodynamic instability defined by the need for continuous norephinephrine infusion. This result is comparable with our clinical experience regarding this type of surgery and patient cohort. Regarding the statistical methodology, we confirm that the Bland-Altman plot was the primary analysis in our study. However, additional analyses are reasonable and provide supplementary information to interpret the results. The primary intention of the study was to assess agreement of temperatures at three different time points reflecting different clinical situations because ‘agreement and accuracy may have varied over time’ as Sessler and Mascha1 state in their letter. Therefore, we analysed the data from each time point separately using the ‘conventional’ Bland-Altman analysis.3 In fact, results were presented in the text including the mean bias ± standard deviations, and the 95% limits of agreement as required. Owing to limited space, we presented a condensed plot which included data for all three time points together instead of six separate plots (Fig. 1). However, we were surprised that in a previous study by Kimberger et al. published in 2009, evaluating accuracy and precision of a novel noninvasive core thermometer,4 the ‘conventional’ Bland-Altman analysis3 instead of the repeated-measures version5 as requested by Sessler and Mascha was used to analyse 1287 temperature measurement pairs obtained at 5 min intervals from 36 patients.Fig. 1: Bland-Altman analyses for different time points.Paired t tests were not the ‘wrong’ analysis tool but allowed a comparison of the means of the different methods of temperature assessment. Spearman correlation coefficient is the more popular and frequently applied analysis of correlation in comparison to Lin's concordance correlation coefficient. However, we agree that Lin's concordance correlation coefficient may have been a valid alternative in this setting as, in contrast to the Pearson and Spearman correlation coefficient, it measures only the correlation with respect to the 45° line through the origin and not to an arbitrary line. A Bonferroni correction as a consequence of performing two comparisons of interest was applied as described in the text and the reported P values were adjusted accordingly. Indeed, the ‘coefficient of determination’ is R2 not r which is a mistake in the abstract. However, in the text, the correct term ‘correlation coefficient’ r was used consistently. In the absence of a genuine gold standard temperature in our study, as was discussed in the manuscript, a calculation of the ‘slope and intercept of the regression line of gold standard’ is not possible. Finally, all potential conflicts of interest including the 3 M advisory board membership of Berthold Bein were declared to the Journal on separate forms. However, we regret that we have overlooked these declarations in the Acknowledgements section of the published article and have requested an erratum in the Journal to correct the omission. Acknowledgements related to this article Assistance with the reply: none. Financial support and sponsorship: none. Conflicts of interest: BB has received consulting fees and/or travel support from CSL Behring, Air Liquide, 3 M, GE Healthcare and Deltex. He is a member of the 3 M, Pulsion Medical Systems, Orion Pharma and Ratiopharm advisory boards. JH has received payments for lectures from 3 M, The Surgical Company, Moelnlycke and Abbvie.
Read more
