Table 2.

Summary of findings for outcomes comparing corticosteroids vs. no corticosteroids

No. of studies	Quality assessment					Relative effect (95% CI)	Absolute effect	Quality
	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias
All-cause mortality								⊕⊕⊝⊝ LOW
7	Seriousa	None	None	None	Suspectedb	RR 0.85 (0.73–0.99)	47 fewer per 1000 (from 3 fewer to 84 fewer)
A composite disease progression								⊕⊕⊝⊝ LOW
4	Seriousa	None	None	None	Suspectedb	RR 0.85 (0.77–0.93)	50 fewer per 1000 (from 23 fewer to 77 fewer)
Serious adverse events								⊕⊕⊝⊝ LOW
4	Seriousa	None	None	None	Suspectedb	RR 1.13 (0.54–2.38)	4 more per 1000 (from 16 fewer to 47 more)

^aSome included studies have high risk of bias according to risk of bias results

^bDue to small number of included trials, publication bias cannot be excluded