Summary of findings 1. High versus low blood pressure target for cardiac surgery with cardiopulmonary bypass.

High versus low blood pressure target for cardiac surgery with cardiopulmonary bypass
Patient or population: adults undergoing cardiac surgery with cardiopulmonary bypass Setting: hospital Experimental: high blood pressure target with mean arterial pressure ≥ 65 mmHg Comparison: low blood pressure target with mean arterial pressure < 65 mmHg
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with low blood pressure target	Risk with high blood pressure target
Acute kidney injury Follow‐up: until discharge from the surgical department or 6 months after the surgery	Study population		RR 1.30 (0.81 to 2.08)	487 (2 RCTs)	⊕⊝⊝⊝ Very lowa	—
	107 per 1000	139 per 1000 (86 to 222)
Cognitive deterioration Follow‐up: 90 days to 6 months	Study population		RR 0.82 (0.45 to 1.50)	389 (2 RCTs)	⊕⊕⊝⊝ Lowb	Definition of cognitive deterioration of each study was: Vedel 2018: change from baseline neuropsychological test performance; ISPOCD test (Moller 1998) 90 days after surgery. Gold 1995: deterioration on ≥ 3 cognitive tests at 6 months after surgery defined as a cognitive complication. For each test, assessment was based on within‐patient change in test performance from preoperative baseline. Since we could not obtain the study protocol, details of each test were unclear. Minimally important difference of cognitive deterioration was defined as a minimal difference in frequency of cognitive deterioration required to have a clinical significance.
	104 per 1000	89 per 1000 (48 to 162)
All‐cause mortality Follow‐up: 30 days to 6 months	Study population		RR 1.33 (0.30 to 5.90)	737 (3 RCTs)	⊕⊝⊝⊝ Very lowc	—
	22 per 1000	29 per 1000 (7 to 128)
Quality of life Follow‐up: 6 months	Study population		RR 0.78 (0.30 to 2.01)	218 (1 RCT)	⊕⊝⊝⊝ Very lowd	Since Gold 1995 counted quality of life as a dichotomous outcome defined as a decline of > 5 points on the Physical Component Summary score of the SF‐36 (Stewart 1989), we used this definition in this review.
	83 per 1000	64 per 1000 (24 to 161)
Acute ischaemic stroke Follow‐up: 30 days to 6 months	Study population		RR 1.29 (0.07 to 23.63)	426 (2 RCTs)	⊕⊝⊝⊝ Very lowe	—
	46 per 1000	43 per 1000 (18 to 100)
Length of hospital stay Follow‐up: 6 months	The mean length of stay in the low blood pressure target group was 12 days	MD 1.25 days longer (0.78 longer to 1.73 longer)	—	540 (2 RCTs)	⊕⊝⊝⊝ Very lowf	—
Perioperative transfusion of blood products Follow‐up: not reported	The mean perioperative transfusion of blood products was 2.0 units.	MD 0.1 units higher (0.13 lower to 0.34 higher)	—	540 (2 RCTs)	⊕⊝⊝⊝ Very lowg	—
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; ISPOCD: International Study of Post‐Operative Cognitive Dysfunction; MD: mean difference; OIS: optimal information size; RCT: randomised controlled trial; RR: risk ratio; SF‐36: 36‐item Short Form.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded one level for imprecision because the OIS of 33,954 was over 10 times larger than the number of participants; one level for indirectness because the definitions used in the included studies were inconsistent with each other, causing quite different occurrences; one level for risk of bias since Azau 2014 was not of overall low risk of bias; and one level for inconsistency with large heterogeneity (I² = 72%).
^bDowngraded one level for risk of bias because the number of follow‐ups was not balanced in one study; and one level for imprecision because the OIS of 910 was not met.
^cDowngraded one level for risk of bias because the two studies were not at overall low risk of bias; one level for inconsistency with large heterogeneity I² = 49%; and one level for imprecision because the OIS of 5948 was not met.
^dDowngraded one level for risk of bias because the study was not at overall low risk of bias; and two levels for imprecision because the sample size was small and the CIs around the RR included 1.0.
^eDowngraded one level for risk of bias because the number of follow‐ups was not balanced in one study; one level for imprecision because the OIS of 4670 was not met; one level because the CI spanned potential benefit, no benefit, and possible harm; and one level for inconsistency for large heterogeneity (I² = 82%).
^fDowngraded one level for risk of bias because it was unclear whether physicians decided the date of discharge could know the allocation of the patients in one study; one level for imprecision because the OIS (1540 or 4906 depending on standard deviation used) was not met; and one level for inconsistency with large heterogeneity (I² = 76%).
^gDowngraded one level for risk of bias because the study was not at overall low risk of bias; one level for performance bias because the allocation of the participants could affect the strategy of transfusion; and one level for imprecision because the CI spanned potential benefit, no benefit, and possible harm. The OIS was met (174 or 230 depending on standard deviation used).