Table 2.

GRADE Evaluation of the Certainty of Evidence (Low Risk of Bias Trials Only)

Certainty Assessment
No. of Studies	Study Design	RoB	Certainty	Importance	Imprecision	Other Considerations	Lower Fluid Volumes	Higher Fluid Volumes	RR Effect (97% or 99% CI)	Absolute Effect (97% CI or 99% CI)
All-cause mortality
8	Randomized trials	Not serious	Not serious	Not serious	Seriousa	None	556/1,802 (30.9)	568/1,824 (31.1)	0.99 (0.89-1.10)	3 fewer per 1.000 (from 34 fewer to 31 more)	Moderate	Critical
SAEs (highest proportion) - no low RoB trials
6	Randomized trials	Seriousb	Not serious	Not serious	Seriousc	None	338/1,817 (18.6)	358/1,783 (20.1)	0.95 (0.83-1.07)	10 fewer per 1.000 (from 34 fewer to 14 more)	Low	Critical
Health-related quality of life
0	Randomized trials	...	...	...	...	...	0/0 (0)	0/0 (0)	Not estimable	...	...	Critical
Duration of MV, d
4	Randomized trials	Not serious	Not serious	Not serious	Not seriousd	None	1,653	1,672	...	Mean difference, 0.11 days lower (0.5 lower to 0.28 higher)	High	Important
Ventilator-free days
5	Randomized trials	Not serious	Not serious	Not serious	Seriouse	None	1,667	1,687	...	Mean difference, 0.11 days lower (1.89 lower to 1.67 higher)	Moderate	Important
Duration of vasopressor or inotropes, h
3	Randomized trials	Not serious	Not serious	Not serious	Not seriousf	None	1,583	1,603	...	Mean difference, 0 h (0.33 lower to 0.33 higher)	High	Important
Vasopressor-free days
3	Randomized trials	Not serious	Not serious	Not serious	Seriousg	None	1,582	1,603	...	Mean difference, 0.43 d higher (0.68 lower to 1.53 higher)	Moderate	Important
Use of RRT
6h	Randomized trials	Not serious	Not serious	Not serious	Seriousi	None	203/1,662 (12.2)	206/1,681 (12.3)	1.01 (0.80-1.26)	1 more per 1.000 (from 25 fewer to 32 more)	Moderate	Important
Duration of RRT, d
3j	Randomized trials	Not serious	Not serious	Not serious	Not seriousk	None	866	887	...	Mean difference, 0.17 d lower (0.7 lower to 0.37 higher)	High	Important
RRT-free days
5	Randomized trials	Not serious	Not serious	Not serious	Seriousl	None	1,631	1,654	...	Mean difference, 0.46 d higher (0.68 lower to 1.61 higher)	Moderate	Important
Incidence of AKI
2	Randomized trials	Not serious	Not serious	Not serious	Seriousm	None	182/811 (22.4)	199/834 (23.9)	0.94 (0.75-1.19)	14 fewer per 1.000 (from 60 fewer to 45 more)	Moderate	Important
Use of blood products, units
3	Randomized trials	Not serious	Not serious	Not serious	Not seriousn	None	891	914	...	Mean difference, 0 units (0.1 lower to 0.1 higher)	High	Important
ICU length of stay, d
6	Randomized trials	Not serious	Not serious	Not serious	Not seriouso	None	965	982	...	Mean difference, 0.33 d lower (0.99 lower to 0.33 higher)	High	Important
Hospital length of stay, d
3	Randomized trials	Not serious	Not serious	Not serious	Seriousp	None	866	888	...	Mean difference, 0.78 d higher (0.73 lower to 2.28 higher)	Moderate	Important

Values are No. of patients, No. of patients (%), or as otherwise indicated. AKI = acute kidney injury; MV = mechanical ventilation; RIS = required information size; RoB = risk of bias; RR = risk ratio; RRT = renal replacement therapy; SAE = serious adverse event; TSA = trial sequential analysis.

^aTSA highlighted that 83% of RIS was reached. The area of futility to detect a predefined relative risk reduction of 15% was reached; however, the CI overlaps no effect, and we cannot exclude important benefit or harm. Adjusted CI was 0.89 to 1.11 (from 34 fewer to 34 more).

^bROBs were adjudicated as some concerns for all six trials based measurement of the outcome, deviations from intended intervention, or selection of the reported result.

^cTSA highlighted that 46% of RIS was reached. Adjusted CI was 0.78 to 1.15 (from 44 fewer to 30 more).

^dTSA highlighted that the boundary for futility crossed and RIS was reached, thus the adjusted CI is identical to the unadjusted. Thus, we can exclude a predefined mean difference of 1 day.

^eTSA highlighted that 34% of RIS was reached. Adjusted CI was −1.86 to 2.27.

^fTSA highlighted that the boundary for futility crossed and RIS was reached; therefore, the adjusted CI is identical to the unadjusted. Therefore, we can exclude a predefined mean difference of 24 h.

^gTSA highlighted that 38% of RIS was reached. Adjusted CI was −1.41 to 2.27.

^hJessen et al had zero events and was not included in meta-analysis or TSA.

ⁱTSA highlighted that 19% of RIS was reached. Adjusted CI was 0.49 to 2.04 (from 62 fewer to 127 more).

^jJessen et al had zero events and was not included in meta-analysis or TSA. The remaining three trials are low RoB trials.

^kTSA highlighted that the boundary for futility was crossed and RIS was reached; therefore, the adjusted CI is identical to the unadjusted. Therefore, we can exclude a predefined mean difference of 1 d.

^lTSA highlighted that 35% of RIS was reached. Adjusted CI was −1.50 to 2.43.

^mTSA highlighted that 22% of RIS was reached. Adjusted CI was 0.46 to 1.93.

ⁿTSA highlighted that the boundary for futility was crossed and RIS was reached; therefore, the adjusted CI is identical to the unadjusted. Therefore, we can exclude a predefined mean difference of 1 unit of blood product.

^oTSA highlighted 87% of RIS was reached, and the boundary for futility was crossed based on a predefined MD of 1 d. Adjusted CI was −1.07 to 0.40.

^pTSA highlighted that 18% of RIS was reached. Adjusted CI was −4.60 to 6.15.