Skip to main content

Table 1 The conventional meta-analysis and trial sequential analysis using random-effects and fixed-effects for all outcomes

From: Stress ulcer prophylaxis with proton pump inhibitors or histamine 2 receptor antagonists in critically ill adults - a meta-analysis of randomized controlled trials with trial sequential analysis

 Conventional meta-analysisTrial sequential analysis (TSA)
Random-effectsFixed-effects Incidence in control groupRequired information size
RR or WMD
(95% CI)
PRR or WMD
(95% CI)
PI2Random-effect TSA-adjusted 95% CIFixed-effect TSA-adjusted 95% CIActual diversity (D2)
Primary endpoints
Clinically important GI bleeding
  All trials (11 trials)0.58 (0.42–0.81)0.0010.56 (0.41–0.76)< 0.0010%0.23–1.510.22–1.390%4.6%22,114
  Low risk of bias (4 trials)0.64 (0.45–0.92)0.0170.64 (0.45–0.92)0.0160%0.15–2.800.15–2.770%4.1%24,928
  High risk of bias (7 trials)0.38 (0.17–0.84)0.0170.35 (0.18–0.67)0.00211.8%0.01–9.930.03–4.9319%6.7%13,827
  Received PPI (6 trials)0.61 (0.43–0.88)0.0080.61 (0.43–0.88)0.0080%0.14–2.700.14–2.680%3.8%26,970
  Received H2RA (6 trials)0.45 (0.17–1.22)0.1160.42 (0.23–0.78)0.00642.6%50%10.4%
  Received EN (8 trials)0.61 (0.44–0.85)0.0040.61 (0.44–0.84)0.0030%0.16–2.380.16–2.330%4.2%24,310
  Did not receive EN (3 trials)0.27 (0.04–2.09)0.2110.27 (0.10–0.73)0.01062.3%65%13.0%
Overt GI bleeding
  All trials (27 trials)0.48 (0.36–0.63)< 0.0010.50 (0.42–0.59)< 0.00134.6%0.31–0.750.38–0.6562%12.1%15,468
  Low risk of bias (4 trials)0.62 (0.48–0.79)< 0.0010.62 (0.49–0.79)< 0.0010%0.37–1.040.37–1.040%8.6%11,384
  High risk of bias (23 trials)0.42 (0.30–0.60)< 0.0010.40 (0.32–0.51)< 0.00137.4%0.25–0.720.28–0.5752%18.1%7595
  Received PPI (8 trials)0.57 (0.45–0.71)< 0.0010.57 (0.45–0.71)< 0.0010%0.36–0.880.36–0.880%8.9%10,969
  Received H2RA (22 trials)0.45 (0.30–0.67)< 0.0010.43 (0.33–0.54)< 0.00145.7%0.26–0.790.30–0.6059%19.3%8258
  Received EN (13 trials)0.64 (0.42–0.96)0.0290.60 (0.48–0.74)< 0.00125.7%0.12–3.350.25–1.4271%9.2%27,681
  Did not receive EN (14 trials)0.37 (0.25–0.55)< 0.0010.36 (0.27–0.48)< 0.00138.4%0.22–0.630.25–0.5252%21.9%6029
Secondary endpoints
All-cause mortality
  All trials (24 trials)1.01 (0.93–1.09)0.8421.02 (0.93–1.11)0.6920%0.90–1.130.91–1.140%26.8%3230
  Low risk of bias (4 trials)1.01 (0.92–1.12)0.8131.01 (0.91–1.12)0.8580%0.67–1.520.67–1.520%29.8%2801
  High risk of bias (20 trials)1.00 (0.87–1.16)0.9941.04 (0.89–1.22)0.6430%0.80–1.240.82–1.320%21.1%4380
  Received PPI (9 trials)1.02 (0.93–1.13)0.6471.02 (0.93–1.13)0.6190%0.82–1.280.82–1.280%28.5%2976
  Received H2RA (18 trials)0.97 (0.83–1.14)0.7110.99 (0.83–1.18)0.9210%0.74–1.280.73–1.350%22.3%4089
Received EN (12 trials)1.05 (0.96–1.16)0.2641.05 (0.96–1.16)0.2540%0.88–1.270.88–1.270%28%3048
  Did not receive EN (12 trials)0.85 (0.71–1.02)0.0770.82 (0.66–1.02)0.0750%0.63–1.150.58–1.170%21.7%4230
Pneumonia
  All trials (12 trials)1.09 (0.95–1.24)0.2211.07 (0.94–1.22)0.3210%0.90–1.310.89–1.290%14.8%6681
  Low risk of bias (3 trials)1.00 (0.86–1.17)0.9591.00 (0.86–1.17)0.9540%0.79–1.270.79–1.270%16.3%5972
  High risk of bias (9 trials)1.47 (1.10–1.97)0.0101.40 (1.03–1.89)0.0320%0.44–4.870.40–4.830%9.9%10,496
  Received PPI (6 trials)1.03 (0.89–1.20)0.6681.04 (0.89–1.20)0.6420%0.82–1.290.83–1.300%15.2%6478
  Received H2RA (8 trials)1.39 (1.01–1.91)0.0461.29 (0.93–1.80)0.1300%0.37–5.130.33–5.030%12.3%8246
  Received EN (8 trials)1.09 (0.95–1.25)0.2361.07 (0.93–1.24)0.3120%0.88–1.340.87–1.320%15.1%6528
  Did not receive EN (4 trials)1.08 (0.65–1.80)0.7571.02 (0.63–1.67)0.9350%0.14–8.580.14–7.530%11.8%8640
Clostridium difficile infection
All trials (4 trials)0.78 (0.45–1.35)0.3770.78 (0.46–1.34)0.3700%0.08–7.320.09–7.010%1.6%70,261
Low risk of bias (3 trials)0.75 (0.43–1.31)0.3090.74 (0.43–1.29)0.2900%0.08–7.260.08–7.010%1.7%66,068
High risk of bias (1 trial)3.06 (0.13–74.19)0.4923.06 (0.1–74.19)0.492 0%
Received PPI (4 trials)0.78 (0.45–1.35)0.3770.78 (0.46–1.34)0.3700%0.08–7.320.09–7.010%1.6%70,261
Received EN (4 trials)0.78 (0.45–1.35)0.3770.78 (0.46–1.34)0.3700%0.08–7.320.09–7.01%1.6%70,261
Tertiary endpoints
Duration of ICU stay
  All trials (11 trials)−0.23 (−1.19–0.73)0.635−0.13 (−0.90–0.65)0.75118.4%  
  Low risk of bias (3 trials)0.11 (−3.20–3.43)0.947−0.42 (−2.02–1.18)0.60754.3%  
  High risk of bias (8 trials)−0.16 (−1.18–0.86)0.757−0.03 (− 0.92–0.85)0.93810.5%  
  Received PPI (5 trials)− 0.54 (− 1.66–0.59)0.349− 0.57 (− 1.63–0.49)0.2947.0%  
  Received H2RA (7 trials)−0.20 (−1.94–1.54)0.8200.38 (−0.76–1.52)0.51022.6%  
  Received EN (7 trials)−0.00 (−1.15–1.15)0.9980.05 (−0.79–0.88)0.91530.5%  
  Did not receive EN (4 trials)−1.18 (−3.26–0.90)0.266−1.18 (− 3.26–0.90)0.2660%  
Duration of MV
  All trials (7 trials)−0.41 (−1.42–0.61)0.434−0.41 (−1.42–0.61)0.4340%  
  Low risk of bias (3 trials)−0.01 (−2.56–2.53)0.991−0.52 (−1.80–0.76)0.42346.6%  
  High risk of bias (4 trials)−0.21 (−1.87–1.46)0.809−0.21 (−1.87–1.46)0.8090%  
  Received PPI (4 trials)−0.04 (−1.63–1.55)0.965−0.35 (−1.55–0.85)0.56720.8%  
  Received H2RA (4 trials)−0.57 (−2.57–1.43)0.574−0.55 (− 2.47–1.37)0.5765.6%  
  Received EN (6 trials)−0.25 (−1.43–0.93)0.676−0.37(−1.40–0.66)0.48111.5%  
  Did not receive EN (1 trial)−2.00 (−8.83–4.83)0.566−2.00 (−8.83–4.83)0.566  
  1. ICU intensive care unit, PPI proton pump inhibitors, H2RA histamine2 receptor antagonists, EN enteral nutrition, MV mechanical ventilation, GI gastrointestinal TSA trial sequential analysis, WMD weighted mean difference, RR relative risk, CI confidence interval
  2. “– “means unavailable data due to too little information used
  3. TSA was conducted with an adjusted type I error of 3.3% for the primary endpoints and 2.5% for the secondary endpoints, power of 80%, D2 suggested by the included trials, relative risk reduction of 20%, two-tailed. If the actual measured D2 was zero, a D2 of 25% was used, because in this case heterogeneity would most likely increase when further studies are included