现在的位置: 首页研究点评, 进展交流>正文
[JAMA Intern Med读者来信]:皮质类固醇在治疗COVID-19中的不确定作用
2020年08月26日 研究点评, 进展交流 暂无评论


Comment & Response August 3, 2020

The Uncertain Role of Corticosteroids in the Treatment of COVID-19

Theodore G. Liou, Frederick R. Adler, Nathan D. Hatton

JAMA Intern Med. Published online August 3, 2020. doi:10.1001/jamainternmed.2020.2438

To the Editor Infection with coronavirus disease 2019 (COVID-19) causes exuberant lung inflammation leading to respiratory failure, acute respiratory distress syndrome (ARDS), and death. Wu et al1 present early experience and retrospective analysis highlighting potential mortality reduction of COVID-19–associated ARDS using corticosteroids to reduce inflammation. However, despite a novel cause, the clinical syndrome resembles that of older diseases, and the analysis faces statistical challenges that have been encountered previously.


Prior studies in ARDS reveal variable steroid effects potentially related to different causes and resulting pathophysiologies invisible at the bedside.2 Different studies have found corticosteroid effects ranging from harmful to beneficial. Within 3 cohort studies of influenza A (H1N1) during the 2009 pandemic, as cited,2 steroid use appeared either ineffective or harmful. Other cohort studies and randomized clinical trials for treatment of ARDS wrestled with artifacts due to indication and survivor bias. The former bias is a familiar issue3 created when unblinded clinicians treat individuals with more serious illness more aggressively, in this case using steroids to prevent or mitigate ARDS. The latter bias arises in 2 ways, either by missing patients unable to survive long enough to receive steroids or failing to follow up with patients long enough to record late deaths due to secondary infections or other steroid-associated complications.


Wu et al1 found that steroid therapy had a low hazard ratio for death for patients receiving steroids for ARDS. However, the result is at odds with results suggesting harm caused by steroids used to prevent ARDS1 and is at odds with another recent report4 using a potentially overlapping patient cohort that found no steroid association with mortality. Wu et al1 suggest that because indication bias usually erroneously suggests harm from a therapy, a beneficial hazard ratio for steroid treatment of ARDS should be believed. However, this assumes that other biases are inconsequential, such as survivor bias due to rapid disease progression compounded by health care resource exhaustion. We note that the Kaplan-Meier curves presented in the original article1 show that substantial numbers of patients were censored, follow-up was substantially shorter than needed to observe steroid adverse reactions, the last observed Kaplan-Meier survival data points of the 2 groups were not statistically different, and, finally, use of steroids was not statistically different between survivors and nonsurvivors of ARDS (Table 3).1


Thus, we urge caution before using steroids for ARDS due to COVID-19. Meticulous observation as performed by Wu et al1 should continue; however, a rigorous blinded randomized clinical trial is needed to discover the benefit or harm of this therapy with confidence.


Comment & Response August 3, 2020

The Uncertain Role of Corticosteroids in the Treatment of COVID-19

Grant B. Ellsworth, Marshall J. Glesby, Roy M. Gulick

JAMA Intern Med. Published online August 3, 2020. doi:10.1001/jamainternmed.2020.2444

To the Editor We question the conclusion of Wu et al1 about the benefit of corticosteroid use in the treatment of coronavirus disease 2019 (COVID-19). The World Health Organization and the US Centers for Disease Control and Prevention have issued clinical guidance against the use of corticosteroids in the treatment of COVID-19 unless another indication is present, such as a chronic obstructive pulmonary disease exacerbation, or as adjunct treatment of septic shock. This guidance was made on the basis of systematic reviews of observational studies of corticosteroids that demonstrated an association with increased mortality and secondary infections in influenza2 and no benefit with possible harms in severe acute respiratory syndrome.3 A retrospective cohort study of Middle East respiratory syndrome showed that corticosteroids were associated with no difference in mortality and prolonged respiratory viral shedding.4


It is unclear that the proportional hazards analysis in the study modeled methylprednisolone administration as a time-dependent covariate, which is necessary to mitigate what has been termed survivor treatment selection bias.5 Stated simply, patients who died rapidly may have been less likely to receive methylprednisolone, leading to an observed difference in mortality that was incorrectly associated with this intervention. Mortality was 23 of 50 (46%) among those with acute respiratory distress syndrome who received methylprednisolone vs 21 of 34 (62%) among those who did not. Even if there were no bias present and methylprednisolone provided some short-term survival benefit, the end point difference of 16% less mortality in those who received corticosteroids is not significant (95% CI, −40% to 8%; P = .23). From the Kaplan-Meier curves, the ultimate mortality rate was roughly similar in both groups—around 60%.

目前尚不清楚该研究模型中甲强龙作为时间相关变量的风险比例分析,因此有必要减少生存者治疗选择偏差。简单地说,快速死亡的患者很少应用甲强龙治疗,导致观察到的死亡率差异与这项干预错误相关。ARDS患者中应用甲强龙治疗的死亡率为23/50 (46%),未应用甲强龙治疗的死亡率为21/34 (62%)。即使没有偏倚存在,甲强龙提供了一定的短期生存获益,但接受皮质类固醇的患者死亡率降低16%的终点差异并不显著(95%CI, -40%to8%; P=0.23)。从K-M曲线来看,两组的最终死亡率大致相似,约60%左右。


您必须 [ 登录 ] 才能发表留言!