Proton pump inhibitors for gastrointestinal bleeding prevention after percutaneous coronary intervention: Comparing evidence in dual antiplatelet therapy

Table 3 Certainty of evidence assessment using grading of recommendations, assessment, development, and evaluation

Certainty assessment							Number of patients		Effect		Certainty	Importance
Number of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	GI bleed-RCTs only	Placebo	Relative (95%CI)	Absolute (95%CI)
GI bleed-RCTs only
6	Randomized trials	Not serious	Not serious	Not serious	Not serious	None	30/2564 (1.2%)	100/4093 (2.4%)	RR 0.37 (0.20-0.68)	15 fewer per 1000 (from 20 fewer to 8 fewer)	High	Critical
GI bleed-RCTs and omeprazole only
4	Randomized trials	Not serious	Not serious	Not serious	Not serious	None	17/1113 (1.5%)	53/2058 (2.6%)	RR 0.45 (0.21-0.96)	14 fewer per 1000 (from 20 fewer to 1 fewer)	High	Critical
GI bleed- RCTs and pantoprazole
2	Randomized trials	Not serious	Not serious	Not serious	Serious3	None	12/1120 (1.1%)	30/1096 (2.7%)	RR 0.28 (0.05-1.55)	20 fewer per 1000 (from 26 fewer to 15 more)	Moderate1,2	Critical
GI bleed-observational only
8	Non-randomized studies	Serious1	Not serious	Not serious	Serious2	All plausible residual confounding would suggest spurious effect, while no effect was observed	737/49743 (1.5%)	965/51554 (1.9%)	RR 0.00 (-0.01 to 0.00)	5 fewer per 1000 (from 10 fewer to 1 more)	Very low1,2	Important
Upper GI bleed-RCTs only
3	Randomized trials	Not serious	Not serious	Not serious	Not serious	None	16/1981 (0.8%)	37/2941 (1.3%)	RR 0.50 (0.28-0.89)	6 fewer per 1000 (from 9 fewer to 1 fewer)	High	Critical
2	Randomized trials	Not serious	Not serious	Not serious	Not serious	None	8/984 (0.8%)	24/1929 (1.2%)	RR 0.42 (0.19-0.91)	7 fewer per 1000 (from 10 fewer to 1 fewer)	High	Critical
Upper GI bleed-observational studies only
2	Non-randomized studies	Serious1	Not serious	Not serious	Serious2	All plausible residual confounding would suggest spurious effect, while no effect was observed	17/190 (8.9%)	77/830 (9.3%)	RR 0.89 (0.16-5.02)	10 fewer per 1000 (from 78 fewer to 373 more)	Very low1,2	Critical
Lower GI bleed-observational studies only
3	Non-randomized studies	Serious1	Not serious	Not serious	Serious2	All plausible residual confounding would suggest spurious effect, while no effect was observed	170/11392 (1.5%)	146/14191 (1.0%)	RR 1.65 (1.31-2.06)	7 more per 1000 (from 3 more to 11 more)	Moderate1,2	Critical
Major bleed
2	Non-randomized studies	Serious1	Not serious	Not serious	Not serious	All plausible residual confounding would suggest spurious effect, while no effect was observed	141/35663 (0.4%)	240/35759 (0.7%)	RR 0.59 (0.48-0.73)	3 fewer per 1000 (from 3 fewer to 2 fewer)	Low1	Critical
All-cause mortality-observational studies only
3	Non-randomized studies	Serious1	Not serious	Not serious	Serious2	All plausible residual confounding would suggest spurious effect, while no effect was observed	70/2791 (2.5%)	43/1803 (2.4%)	RR 1.10 (0.75-1.63)	2 more per 1000 (from 6 fewer to 15 more)	Very low1,2	Important

¹Observational studies are prone to confounding, selection bias, and misclassification even if adjusted.

²95% confidence interval crosses the line of no effect (1.0), suggesting uncertainty.

³Wide confidence interval, almost crossing the line of non-effect.

CI: Confidence interval; RR: Relative risk; RCT: Randomized controlled trial; GI: Gastrointestinal.