Comment & Response November 5, 2019

Spontaneous Breathing Trials and Successful Extubation—Reply

Carles Subirà, Rafael Fernández

JAMA. 2019;322(17):1717. doi:10.1001/jama.2019.14244

In Reply

We agree with Drs Friedrich and Burns that prophylaxis against postextubation respiratory failure with NIV or HFNC was more common in the PSV group than in the T-piece group (24.7% vs 18.7%).¹ However, the decision to apply prophylaxis was made before randomization, so this difference can only be attributed to chance. We would like to clarify that patients could receive prophylactic NIV or HFNC but not both.

我们同意Friedrich 和 Burns 医生的意见，即NIV或HFNC预防拔管后呼吸功能衰竭在PSV组较T管组更多 (24.7% vs 18.7%)。然而，是否使用上述预防措施是在随机化之前作出的决定，因此，这一差异仅因偶然造成。我们还想澄清，患者可以使用预防性NIV或HFNC，但不能两者均使用。

In response to their question about the rate of NIV and HFNC to treat postextubation respiratory failure in each group, postextubation respiratory failure occurred in 213 patients (20.9%): 103 in the T-piece group (21.2%) and 110 in the PSV group (20.7%) (P = .84). In patients with postextubation respiratory failure, NIV was applied in 43 patients in the T-piece group (41.7%) and in 48 patients in the PSV group (43.6%) (P = .78), and HFNC was applied in 28 patients in the T-piece group (27.2%) and in 19 patients (17.3%) in the PSV group (P = .08). Four patients in each group received both treatments (3.9% in the T-piece group and 3.6% in the PSV group; P = .92). These results reinforce the idea that the higher rate of successful extubation in PSV was due to better tolerance of the SBT, with similar rates of respiratory failure and reintubation in the 2 groups. Prophylactic treatment did not seem to influence the main result of the study.

有关各组使用NIV和HFNC治疗拔管后呼吸功能衰竭的比例这一问题，拔管后共213例患者发生呼吸功能衰竭(20.9%)：T管组103例 (21.2%)，PSV组110例 (20.7%) (P = .84)。在拔管后发生呼吸衰竭的患者中，T管组和PSV组分别有43例 (41.7%) 和 48例 (43.6%) 使用了NIV (P = .78)，分别有28 例 (27.2%) 和 19 例 (17.3%) 使用了HFNC (P = .08)。每组各有4名患者接受NIV和HFNC (T管组3.9%，PSV组3.6%；P = .92)。这些结果强调一下观点，即PSV组成功拔管率较高是由于SBT更好的耐受性，而两组间呼吸衰竭和再插管率相似。预防性治疗措施并未影响研究的主要结果。

Dr Mancebo and colleagues point out that our results would be expected based on Bayes’ theorem; however, our study demonstrated that these predictions hold true in real-life routine clinical scenarios. The high percentage of patients who tolerated the SBT merely reflects that most physicians routinely wait for patients to fully recover before subjecting them to an SBT; importantly, even in these conditions, 30 minutes of PSV seems better than a 2-hour T-piece SBT. Whether physicians should perform SBTs earlier remains open to investigation.

Mancebo医生及其同事指出，根据Bayes定理可以预见到我们的结果；然而，我们的研究显示，这些预测在真实生活中常规临床状况下也是正确的。耐受SBT患者比例很高，这一点仅仅反映了多数医生常规等待患者完全恢复，再进行SBT；重要的是，即使在这种情况下，30分钟PSV仍然优于2小时T管SBT。医生是否应当更早进行SBT有待研究。

Their comments about the possible effects of the interim analysis on the meaning of our findings are based on a misunderstanding. Our sentence “…a prespecified interim analysis was performed…showed a nonsignificant difference in primary outcome between groups…” only referred to the fact that due to the sample size, the observed difference between groups (SBT tolerance, 82.5% vs 74.2%; P = .02) did not reach the prespecified statistical significance level for stopping the trial. To dispel suspicion about possible behavioral changes after the interim analysis, we would like to clarify that the size of the effect was similar in the periods before and after the analysis.

他们有关中期分析对研究结果可能产生的影响是因误解导致的。我们的原话“...进行了预先确定的中期分析...结果显示，两组间主要预后指标没有显著差异...研究者决定完成预计的样本量”仅仅指以下事实，即由于样本量的原因，观察到的组间差异(SBT 耐受性, 82.5% vs 74.2%; P = .02) 没有达到预先确定的终止试验所需的统计学显著差异水平。为消除中期分析后研究行为发生改变的可能疑虑，我们希望指出，在中期分析前后阶段的疗效是相似的。