Table 2.

Assessment sheet of overall quality of evidence by Grading of Recommendation, Assessment, Development, and Evaluations (GRADE) guidelines

Certainty assessment							No of patients		Effect		Certainty	Importance
No of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Surgery	Control	Relative (95% CI)	Absolute (95% CI)
All-cause mortality
9	Randomised trials	Seriousa	Not serious	Not serious	Seriousb	Publication bias strongly suspectedc	7/423 (1.7%)	18/439 (4.1%)	RR 0.53 (0.21 to 1.38)	19 fewer per 1,000 (from 32 fewer to 16 more)	⨁◯◯◯ Very low
Pneumonia
8	Randomised trials	Seriousa	Not serious	Not serious	Not serious	None	51/362 (14.1%)	93/376 (24.7%)	RR 0.57 (0.35 to 0.92)	106 fewer per 1,000 (from 161 to 20 fewer)	⨁⨁⨁◯ Moderate
Tracheostomy
6	Randomised trials	Seriousa	Not serious	Not serious	Seriousb,d	None	42/291 (14.4%)	63/297 (21.2%)	RR 0.70 (0.38 to 1.30)	64 fewer per 1,000 (from 132 fewer to 64 more)	⨁⨁◯◯ Low
Length of ICU stay
8	Randomised trials	Seriousa	Seriouse	Not serious	Not serious	Publication bias strongly suspectedc	372	380	–	MD 3.05 days lower (5.87 lower to 0.22 lower)	⨁◯◯◯ Very low
Length of hospital stay
7	Randomised trials	Seriousa	Seriouse	Not serious	Seriousd	None	354	361	–	MD 3.79 days lower (9.33 lower to 1.75 higher)	⨁◯◯◯ Very low
Duration of mechanical ventilation
7	Randomised trials	Seriousa	Seriouse	Not serious	Not serious	None	300	304	–	MD 4.62 days lower (7.64 lower to 1.6 lower)	⨁⨁◯◯ Low

^aSeven randomised clinical trials had some concerns of bias

^bInsufficient optimal information size

^cFunnel plot indicates evidence of significant publication bias

^dLarge confidence interval

^eHigh heterogeneity (I2 > 75%)

CI, confidence interval; RR, risk ratio; ICU, intensive care unit; MD, mean difference