The following keywords, proton pump inhibitors, PPI, H₂ receptor antagonists, H₂RA, endoscopic, bleeding, randomized controlled trial and clinical trial, were used as search terms in the Cochrane library, MEDLINE, EMBASE and PubMed until July 2014.

The inclusion criteria for this study were: (1) all patients in the experimental group were diagnosed with any type of upper gastrointestinal bleeding after successful endoscopic therapy; (2) comparison therapies of proton pump inhibitors or H₂ receptor antagonists in similar baseline level patients; (3) the end-points included recurrent bleeding rate; (4) randomization, controls, and measurable outcomes were reported; and (5) the articles were written in English.

The articles were extracted independently by two investigators and any disagreements were resolved by discussion or by asking the third investigator. The first author’s name, publication year, sample size, participants’ age, participants’ gender, smoking (%), alcohol abuse (%), positive H. pylori infection (%), NSAID user (%), drug type, intervention measure and outcome assessment time were extracted. The main outcomes included were: (1) mortality (n); (2) re-bleeding (n); (3) received surgery (n); (4) blood transfused (unit/500 mL); and (5) hospitalization stay time (d).

We used the Cochrane Collaboration’s tool[12] to evaluate the quality of the articles. The following seven items of risk of bias in the tool were assessed: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and Other bias, all studies were classified as low risk, high risk and unclear risk. The assessment was performed independently by two investigators; disagreements were be resolved by discussion or by involving the third investigator.

At the end of treatment for the individual trials, the odds ratio (OR) and mean difference (MD) with their 95% confidence interval (CIs) were calculated. The OR and weighted mean difference (WMD) were used as summary estimators. A fixed-effect model weighted by the inverse variance method was used following a homogeneity test. We performed the homogeneity test using a χ² test on N-1 degrees of freedom. A P value of 0.05 was regarded as statistically significant, rejecting the assumption of homogeneity (P < 0.05), and the random-effect model was then performed using the inverse variance method. Publication biases were evaluated using a Funnel plot, Begg’s test and Egger’s test; P≥ 0.05 indicated there was no publication bias.

All statistical analyses were performed using Review Manager 5.3.3 statistical software (Cochrane Collaboration, Oxford, United Kingdom) and Stata 12.0 statistical software (Stata Co. College Station, TX, United States) for the meta-analysis.

One-thousand and eighty-two publish articles were identified, and we initially excluded 923 duplicate articles. Then, after reading the title and abstract, 53 articles remained. Finally, by reading the full text of each article, 10 RCTs[13-22] were included in this meta-analysis (Figure 1). The characteristics of all included studies are shown in Table 1. This meta-analysis included 1283 patients, with 678 subjects in the PPI group and the remaining 605 subjects in the H₂RA group. The maximum sample size of the included studies was 200 cases[16]; the minimum was 77 cases[20]. For the study of Lin et al[16], which used two randomized controlled programs, the programs were divided into a and b programs as two separate studies and included in the meta analysis.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Flow chart of study selection. RCTs: Randomized controlled trials.

The number of recurrent bleeding subjects reported after treatment was investigated in all included studies[13-22]. According to the different routes of administration, the studies could be divided into three subgroups, intravenous followed by oral, simple intravenous and simple oral. There was not any significant heterogeneity between the included trials (P = 0.13, I² = 35%) and subgroups (P = 0.90, I² = 0%); therefore, we used the fixed effects model for the analysis. Seven of the studies[13,16,17,19-22], including 922 subjects, used the first intravenous and then oral administration method; the results of these studies did not statistically significantly reduce the re-bleeding rate (OR = 0.35; 95%CI: 0.24-0.52; Figure 2). However, one trial[15], which included 149 subjects, reported the re-bleeding number after simple intravenous treatment and did observe a statistically significant difference (OR = 0.45; 95%CI: 0.15-1.36; Figure 2). Meanwhile, no statistical difference was observed in two trials[14,18] that reported the re-bleeding number after simple oral treatment (OR = 0.31; 95%CI: 0.09-1.03; Figure 2). The meta-analysis results for re-bleeding rate revealed that after successful endoscopic therapy, compared with the H₂RA therapy group, the PPI therapy group had significantly reduced re-bleeding rates (OR = 0.36; 95%CI: 0.25-0.51; Figure 2). Figure 2 shows the quality assessments of these ten articles, as evaluated by the Cochrane Collaboration’s tool.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Forest plot and risk of bias summary from the Cochrane risk of bias tool comparing proton pump inhibitors therapy vs H₂ receptor antagonists therapy efficacy for the recurrent bleeding rate. PPI: Proton pump inhibitors; H₂R: H₂ receptor antagonists.

Inconsistencies within the end-point time, nationality and intervention drugs, may have resulted in significant and substantial heterogeneity in our analysis (Table 2). The first subgroup analysis used the time of recurrent bleeding occurrence after successful endoscopic therapy (d). Only one trial[17] reported the re-bleeding number within 24 h and two trials[15,22] reported the re-bleeding number within 3 d. These trials were significantly heterogeneous in reducing the re-bleeding rate (OR = 0.23; 95%CI: 0.07-0.76; Table 2). An additional two trials[15,17] reported the re-bleeding number within 7 d after successful endoscopic therapy and observed a statistically significant reduction in the re-bleeding rate (OR = 0.32; 95%CI: 0.13-0.79; Table 2). Furthermore, heterogeneity was observed in the three trials[16,17,22] that reported the re-bleeding number within 14 d (OR = 0.26; 95%CI: 0.16-0.43; Table 2). Meanwhile, a statistical difference was observed in two trials[14,15] that reported the re-bleeding number within 28-30 days (OR = 0.45; 95%CI: 0.15-1.36; Table 2). Finally, there were five trials[13,18-21] that reported re-bleeding number within 6 wk or more and no statistical difference was observed (OR = 0.53; 95%CI: 0.30-0.94; Table 2). In summary, this subgroup analysis demonstrated that the end-point time after successful endoscopic therapy was not significantly different between studies (OR = 0.33; 95%CI: 0.20-0.45; Table 2) and subgroups (P = 0.42; Table 2).

The second subgroup analysis used the nationality of origin for each study. Four Japanese trials[18-21] had an OR = 0.63 (95%CI: 0.33-1.17; Table 2) and did not demonstrate a statistically significantly reduced re-bleeding rate. Three Chinese trials[13,16,22] showed heterogeneity for reducing the re-bleeding rate (OR = 0.25; 95%CI: 0.14-0.42; Table 2). Two South Korea trials[14,17] and one United States trial[15] had an OR of 0.25 (95%CI: 0.07-0.95; Table 2) and 0.45 (95%CI: 0.15-1.36; Table 2), respectively. In summary, in this subgroup analysis for re-bleeding, the nationality of the study did not significant effect drug use after successful endoscopic therapy between the different studies (OR = 0.36; 95%CI: 0.25-0.51; Table 2) and subgroups (P = 0.42; Table 2).

The third subgroup analysis used groupings by the PPI intervention drug type. Five trials[14,16,19,20,22] used omeprazole as the intervention PPI drug and had significant heterogeneity, with an OR of 0.32 (95%CI: 0.20-0.52; Table 2). These five trials were not statistically significantly different compared with the three pantoprazole studies[13,15,17] (OR = 0.31, 95%CI: 0.15-0.65; Table 2). The fourth subgroup used groupings by the H₂RA intervention drug type. Three trials used cimetidine[16,18,22] as the H₂RA intervention drug and had an OR of 0.27 (95%CI: 0.16-0.44; Table 2) with significant heterogeneity. Three trials used famotidine[14,20,21] as the H₂RA intervention drug (OR = 0.85; 95%CI: 0.28-2.61; Table 2) and two trials[13,15] used ranitidine as the H₂RA intervention drug (OR = 0.34; 95%CI: 0.14-0.85; Table 2). In summary, the H₂RA intervention drug subgroup analysis for the re-bleeding rate revealed that the H₂RA drug type after successful endoscopic therapy drug use had no significant effect in reducing the re-bleeding rate (OR = 0.33; 95%CI: 0.22-0.48; Table 2) or between subgroups (P = 0.18; Table 2).

In general, the subgroup analyses results did not reveal statistically significant differences in patients with upper gastrointestinal bleeding after successful endoscopic therapy (Table 2).

Five studies[13,15-17,22], involving 715 patients, reported mortality after treatment. According to the different routes of administration, the studies could also be divided into three subgroups, intravenous followed by oral, simple intravenous and simple oral. There was no significant heterogeneity between these trials (P = 0.68, I² = 0%) or subgroups (P = 0.22, I² = 33.3%); therefore, we used a fixed effects model for the analysis. Four of the studies[13,16,17,22], including 588 subjects, used the first method, intravenous and then oral administration, and the mortality was not statistically significantly reduced (OR = 0.29; 95%CI: 0.08-0.17; Figure 3). Only one trial[15], including 149 subjects, reported the dead number after simple intravenous treatment and no statistically significant difference was observed (OR = 1.07; 95%CI: 0.21-5.49; Figure 3). In summary, this meta-analysis of mortality revealed that after successful endoscopic therapy, there were no significant differences in mortality between the two drugs (OR = 0.46; 95%CI: 0.17-1.23; Figure 3).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Forest plot comparing the efficacy of proton pump inhibitors therapy vs H₂ receptor antagonists therapy for the recurrent bleeding rate mortality. PPI: Proton pump inhibitors; H₂R: H₂ receptor antagonists.

Four studies[13,16,17,22], involving 566 patients, reported the number of patients who received surgery after treatment. There was no significant heterogeneity between these trials (P = 0.77, I² = 0%); therefore, a fixed effects model was used for the analysis. The meta-analysis revealed that after successful endoscopic therapy, compared with H₂RA therapy, PPI therapy significantly decreased the number of patients that received surgery (OR = 0.29; 95%CI: 0.09-0.96; Figure 4).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Forest plot comparing the efficacy of proton pump inhibitors therapy vs H₂ receptor antagonists therapy for the required surgery rate. PPI: Proton pump inhibitors; H₂R: H₂ receptor antagonists.

Three studies[13,16,22], involving 402 patients, reported the units (mL; 500 mL per unit) for patients who required blood transfusions after treatment. There was significant heterogeneity between these trials (P = 0.0001, I² = 89%); therefore, a random effects model was used for the analysis. The meta-analysis revealed that after successful endoscopic therapy, PPI therapy was more effective in decreasing blood transfusion units (WMD: -0.70 unit; 95%CI:-1.64-0.25; Figure 5).

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Forest plot comparing the efficacy of proton pump inhibitors therapy vs H₂ receptor antagonists therapy for blood transfusion units. PPI: Proton pump inhibitors; H₂R: H₂ receptor antagonists.

Four studies[13,16,20,22], involving 479 patients, reported the hospitalization time of patients. There was significant heterogeneity between these trials (P < 0.00001, I² = 97%); therefore, a random effects model was used for the analysis. The meta-analysis revealed that after successful endoscopic therapy, PPI therapy was more effective in decreasing hospitalization time (WMD: -0.77d; 95%CI: -1.87-0.34; Figure 6).

Open in New Tab   Full Size Figure   Download Figure

Figure 6 Forest plot comparing the efficacy of proton pump inhibitors therapy vs H₂ receptor antagonists therapy for hospitalization time. PPI: Proton pump inhibitors; H₂R: H₂ receptor antagonists.

Publication bias was evaluated using a Funnel plot (Figure 7). The asymmetrical scatter plot revealed little publication bias in our meta-analysis. In addition, we also performed the Begg’s test and Egger’s test; both tests demonstrated that there was no publication bias in our meta-analysis (Begg’s test: P = 0.283; Egger’s test: P = 0.339).

Open in New Tab   Full Size Figure   Download Figure

Figure 7 Funnel plot evaluating the publication bias for recurrent bleeding rate.

In patients with upper gastrointestinal bleeding, recurrent bleeding is the most important adverse effect, resulting in morbidity and mortality. Both proton pump inhibitors and H₂ receptor antagonists are commonly administered to upper gastrointestinal bleeding patients after successful endoscopic therapy.

A series of randomized controlled trials were conducted to compare proton pump inhibitor (PPI) therapy and H₂ receptor antagonists (H₂RA) therapy after successful endoscopy to determine the appropriate first-line treatment drug for upper gastrointestinal bleeding.

This is the first meta-analysis to evaluate the efficacy and safety of PPI therapy and H₂RA therapy after successful endoscopy in patients with upper gastrointestinal bleeding, using several important endpoints, such as re-bleeding rate, mortality and receiving surgery rate.

The results suggested that PPI therapy is superior to H₂RA therapy in upper gastrointestinal bleeding patients after successful endoscopic therapy. This result may provide valuable information to clinicians.

PPI is a type of H⁺/K⁺ ATPase inhibitor. Its acid-suppressing activity is potent, highly specific and has a long duration. H₂RA can selectively block wall H₂ receptors on the cell membrane, reducing gastric acid secretion.

This is a well-performed meta-analysis that aimed to determine whether PPI therapy is more efficacious and safe than H₂RA therapy in upper gastrointestinal bleeding patients after successful endoscopic therapy that provides definitive results.