著者

Nobuyuki Horita Satoru Hashimoto Naoki Miyazawa Hiroyuki Fujita Ryota Kojima Miyo Inoue Atsuhisa Ueda Yoshiaki Ishigatsubo Takeshi Kaneko

出版者

一般社団法人 日本内科学会

雑誌

Internal Medicine (ISSN:09182918)

巻号頁・発行日

vol.54, no.12, pp.1473-1479, 2015 (Released:2015-06-15)

参考文献数

39

被引用文献数

8 28

---

Objective The impact of corticosteroids on acute respiratory distress syndrome (ARDS) mortality remains controversial following the publication of numerous trials, observational studies and meta-analyses. An updated meta-analysis is warranted, as a few original studies on this topic have been published since the last meta-analysis. Methods We searched for eligible articles using four databases. In particular, we included full-length original articles providing sufficient data for evaluating the impact of corticosteroid treatment on adult ARDS mortality in the form of odds ratios. A fixed model with the confidence interval method was used. An assessment of publication bias and sensitivity analyses were also conducted. Results We included 11 of 185 articles. The pooled odds ratio for corticosteroids with respect to all-cause mortality involving 949 patients was 0.77 [95% confidence interval (CI): 0.58-1.03, p=0.079] with strong heterogeneity (I2=70%, p<0.001). The results of the sensitivity analysis, Begg-Kendall test (τ=0.53, p=0.024) and funnel plot consistently suggested the existence of strong publication bias. After six potentially unpublished cohorts were filled using Duval's trim and fill method, the pooled odds ratio shifted to 1.11 (95% CI 0.86-1.44, p=0.427). In addition, the sensitivity analyses suggested that corticosteroid treatment has a different impact on mortality depending on the comorbidities and trigger events. Conclusion We were unable to confirm, based on the data of published studies, the favorable impact of corticosteroid therapy on mortality in overall ARDS cases. Published articles exhibit strong publication bias, and previous meta-analyses may be affected by this publication bias. Further research focusing on pathophysiology- or trigger event-specific ARDS is anticipated.