Meta-analysis: comparison of oral vs. intravenous proton pump inhibitors in patients with peptic ulcer bleeding


  • K. K. F. Tsoi,

    1. The Institute of Digestive Disease, The Chinese University of Hong Kong, Shatin, Hong Kong
    2. The Hong Kong Branch of the Chinese Cochrane Centre, The Chinese University of Hong Kong, Shatin, Hong Kong
    3. School of Public Health and Primary Care, The Chinese University of Hong Kong, Shatin, Hong Kong
    Search for more papers by this author
  • H. W. Hirai,

    1. The Institute of Digestive Disease, The Chinese University of Hong Kong, Shatin, Hong Kong
    Search for more papers by this author
  • J. J. Y. Sung

    Corresponding author
    • The Institute of Digestive Disease, The Chinese University of Hong Kong, Shatin, Hong Kong
    Search for more papers by this author

  • As part of AP&T's peer-review process, a technical check of this meta-analysis was performed by Dr P. Collins.

Correspondence to:

Prof. J. J. Y. Sung, Department of Medicine and Therapeutics, Prince of Wales Hospital, Shatin, NT, Hong Kong.




The efficacy of adjuvant use of intravenous proton pump inhibitors (PPIs) after endoscopic therapy has been proved in peptic ulcer bleeding patients, but the efficacy of oral PPIs is uncertain.


To compare the clinical outcomes of oral PPIs vs. intravenous PPIs in patients with peptic ulcer bleeding.


Prospective randomised controlled trials were systematically searched from OVID databases until June 2012. Trials comparing oral and intravenous PPIs among patients with peptic ulcer bleeding were included. Recurrent bleeding, length of hospitalisation, blood transfusion, requirement of surgery and mortality were measured. The risk of bias, study quality and heterogeneity were also evaluated.


Six randomised trials from 2006 to 2011 were included. A total of 615 patients were randomly assigned to receive oral PPIs (n = 302) or intravenous PPIs (n = 313). The mean age was 60 years and 71.1% was male. No significant difference between oral and intravenous PPIs was observed regarding recurrent bleeding (RR: 0.92, 95% CI: 0.56–1.50), mean volume of blood transfused (−0.02 unit, 95% CI: −0.29–0.24 unit), requirement of surgery (RR: 0.82, 95% CI: 0.19–3.61) and all-cause mortality (RR: 0.88, 95% CI: 0.29–2.71). The duration of hospital stay in days was significantly shortened in those using oral PPIs (−0.74 day, 95% CI: −1.10 day to −0.39 day).


Oral PPIs demonstrate a similar effectiveness to intravenous PPIs among patients with peptic ulcer bleeding, but the results were combined from open-labelled trials with limited sample size. A large double-blind non-inferiority trial is required to better assess the role of oral PPIs.


Acute upper gastrointestinal bleeding (UGIB) remains a common medical problem associated with significant morbidity and mortality. The annual incidence of acute UGIB is approximated as 50–170 per 100 000 population,[1-5] and for around 50% of adult patients, UGIB is due to peptic ulcer.[5, 6] Endoscopic therapy, such as thermocoagulation or haemoclipping, has been generally recommended as the first-line of treatment for peptic ulcer bleeding, as it has been shown to reduce recurrent bleeding, requirement for surgery and mortality.[7-9] However, gastric acidity inhibits clot formation on bleeding ulcers and interferes with haemostasis, and therefore the neutralisation of intragastric pH level is crucial to consolidate clot formation. It also reduces rebleeding rate.[10]

Proton pump inhibitors (PPIs) are potent gastric acid suppression agent inhibiting the enzyme pumping action of acidic hydrogen ion into stomach to reduce acid secretion. The efficacy of adjuvant use of intravenous PPIs after endoscopic therapy has been proved in peptic ulcer bleeding.[11-13] Nonetheless, the optimal dose and the best administration methods remain uncertain.[14, 15] The oral and intravenous PPIs have shown similar effectiveness on rising intragastric pH level, and the intragastric pH differs only at the first hours of drug administration.[16] In the 24-h intragastric pH monitoring, oral PPIs may be able to replace intravenous PPIs in patients with bleeding ulcers. Moreover, the same dose of various PPIs given through either intravenously or orally can also raise intragastric pH level over 6 for 3 days after successful endoscopic haemostasis.[17]

In recent years, several randomised controlled trials have head-to-head compared the efficacy of oral and intravenous PPIs in peptic ulcer bleeding, but most of the trials have been limited by relatively small sample sizes.[18-23] There are also variations in entry criteria on bleeding stigmata, treatment responses between oral and intravenous PPIs, and time of outcome measures. We perform this meta-analysis based on the published data to compare the overall benefits of oral PPIs with intravenous PPIs in the patients with peptic ulcer bleeding.


Search strategy

Prospective randomised controlled trials that recruited endoscopically proved peptic ulcer bleeding patients to compare the performance of oral PPIs vs. intravenous PPIs were included in this meta-analysis. Literature search was performed in OVID databases to identify the trials in full publication with English abstracts. Keywords for the search were ‘proton pump inhibitors’, ‘PPI’, ‘omeprazole’, ‘lansoprazole’, ‘pantoprazole’, ‘rabeprazole’, ‘esomeprazole’, ‘dexlansoprazole’, ‘oral’, ‘intravenous’ and ‘ulcer bleeding’. Computerised databases, including MEDLINE, EMBASE, EBM Reviews, Cochrane Centre Register of Controlled Trials, British nursing index and archive, Wan Fang Data and Google Scholar were searched on 15th December 2011. Literature search was performed to check for the updates on 15th June 2012.

Inclusion & exclusion criteria

Randomised controlled trials were included if they met the following criteria: (i) all patients admitted with symptoms of ulcer bleeding as evidenced by haematemesis or melaena; (ii) endoscopic therapy was performed to stop the bleeding; (iii) patients were randomised to receive either oral or intravenous PPIs; and (iv) at least one of the following outcomes was reported after endoscopic therapy: recurrent bleeding, surgical intervention or mortality. Trials were excluded if they studied patients with bleeding from malignancy, or patients already on PPIs treatment.

Data extraction

Two investigators (KKT, HWH) independently assessed the titles and abstracts of all generated papers for relevancy. We reviewed each identified trial and determined inclusion. Investigators also independently abstracted the data into a standardised data extraction form. Decisions regarding inclusion of studies and data extraction were reached by consensus between the two reviewers. When discrepancies were found, the third investigator (JJS) would make the definitive decision for trial eligibility and data extraction.

Study outcomes

Primary outcome of this meta-analysis was recurrent bleeding, which is defined as the unsuccessful control of bleeding after the first endoscopic therapy. Secondary outcomes included volume of blood transfused in 500 mL per unit, duration of hospital stay in days, requirement of surgical intervention and all-cause mortality after endoscopy therapy.

Risk of bias and study quality

Potential sources of bias were evaluated by Cochrane Risk of tool.[24] The Risk of Bias basically evaluated the adequate sequence generation, subject allocation and concealment, blinding of patients and outcome assessment, outcome data completely addressed, selective outcome reporting and other potential bias. Quality of each trial was also assessed by Oxford quality scoring system, Jadad criteria[25]: (i) study described as randomised, (ii) method used to generate the sequence of randomisation described and appropriate, (iii) study described as double-blind, (iv) method of double-blinding described and appropriate, and (v) description of withdrawals and dropouts. Some of these quality parameters have been included in the Risk of Bias evaluation such as adequate sequence generation, subject allocation and concealment, and blinding methods.

Statistical analyses

Meta-analyses were performed with Review Manager.[26] Relative risk (RR) and mean difference (MD) with 95% confidence interval (CI) were used to evaluate the discrete and continuous variables respectively. When the standard deviation of the mean is not reported, equal standard deviations were assumed between oral and intravenous PPIs groups, and were estimated with reference to the reported P-value for statistical significance. Statistical heterogeneity among the trials was assessed, and P-value < 0.1 was considered statistically significant. We assessed heterogeneity with I2, which describes the percentage of total variation across studies due to the heterogeneity rather than chance alone. High values of I2 would show a large heterogeneity. We used a Mantel-Haenszel fixed-effects model for significant homogeneous trials, and otherwise, random-effects model was applied.[27] Subgroup analyses were performed on the trials with high doses of intravenous PPIs, the trials included only patients with major stigmata (i.e. spurting, oozing or nonbleeding vessel), and the trials not at the high level risk of bias.


Literature search

The initial search identified 1031 abstracts (Figure 1). All abstracts were evaluated and 25 studies were relevant to the comparison between oral and intravenous PPI. Nineteen studies were excluded for the following reasons: two studies recruiting healthy subjects without ulcer bleeding, four studies including patients on post-ESD gastric ulcers or acid reflux, eleven studies only using intragastric pH level as the primary outcome and two studies with nonrandomised design. The definitive analysis in this meta-analysis included 6 randomised controlled trials with patients from United States, Turkey, Iran, Taiwan and Korea and published in the period between 2006 and 2012 (Table S1).[18-23] One trial was published in Korean, but the major findings were reported in English in the abstract and the Tables presented the clinical outcomes.[23]

Figure 1.

Flow chart of literature search in OVID databases.

Characteristic of individual trials

Among the 6 randomised controlled trials, a total of 615 subjects were randomly allocated to receive oral PPIs (n = 302) vs. intravenous PPIs (n = 313) (Table S1). Most of the trials were scored as 2 out of 5 due to lack of blinding,[18-21, 23] and one trial was graded as 0[22] with reference to the Jadad quality score. This trial was also classified as the high level Risk of Bias with unsatisfying randomisation, allocation and blinding procedure according to the Cochrane's definition. The number of patients in the individual trials ranged from 25 to 211. These trials included patients with mean age of 60 years and 437 male (71.1%) (Table S1). In the pooled data, ulcer bleeding patients showed a wide range of bleeding stigmata, including clean base ulcer (26%), flat/red spot (8%), adherent clot (19%), nonbleeding vessel (14%), visible vessel (10%), and active bleeding, including spurting and oozing vessel (23%). Intravenous PPIs were given in the first 3 days and then PPIs were orally given for around 1–2 months. Four trials used high-dose PPIs i.e. 80 mg bolus PPIs[18-20, 22], whereas two used 40 mg twice daily infusion.[21, 23] Different PPIs were used in these studies, including Pantoprazole,[18, 20, 22] Omeprazole,[19, 21-23] Rabeprazole[21] and Lansoprazole.[23] Endoscopic procedures were performed within 24 h of all admissions. All trials offered standard treatment regimens for the patients with Helicobacter pylori infection.

Primary outcome

All trials reported recurrent bleeding. Heterogeneity among the trials was not statistically significant (P = 0.90) and the combined result showed no significant difference on the recurrent bleeding between the oral (8.6%) and intravenous (9.3%) PPIs groups (RR: 0.92, 95% CI: 0.56–1.50) (Figure 2). Four trials[18-20, 22] reported 30-day recurrent bleeding and oral vs. intravenous PPIs showed no significant difference in outcome (RR 0.77, 95% CI: 0.37–1.61). Two trials[21, 23] reported recurrent bleeding within 15 days. Again, no statistically significant difference was found (RR 1.07, CI: 0.55–2.10).

Figure 2.

Forrest plot for recurrent bleeding.

Secondary outcomes

The length of hospital stay among patients received oral PPIs was significantly reduced compared with those received intravenous PPIs (mean difference on hospitalisation: −0.74 day, 95% CI: −1.10 day to −0.39 day) (Figure 3). However, there was no statistically significant difference in terms of mean volume of blood transfused (−0.02 unit, 95% CI: −0.29–0.24 unit) (Figure 4), requirement of surgical intervention (1.1% vs. 1.4%, RR: 0.82, 95% CI: 0.19–3.61) and all-cause mortality (1.8% vs. 2.1%, RR: 0.88, 95% CI: 0.29–2.71) (Figure 5).

Figure 3.

Forrest plot for length of hospital stay (days).

Figure 4.

Forrest plot for mean volume unit of blood transfusion (500 mL per unit).

Figure 5.

Forrest plot for overall mortality.

Subgroup analyses

In the subgroup of trials with high doses of intravenous PPIs, i.e. 80 mg intravenous bolus PPIs and 8 mg/h infusion,[18-20, 22] no statistically significant difference was shown in terms of recurrent bleeding (RR: 0.77, 95% CI: 0.37–1.61) (Figure 2, the same subgroup with rebleeding within 30 days), volume of blood transfused (mean difference: 0.07 unit, 95% CI: −0.22–0.36 unit) and all-cause mortality (RR: 0.67, 95% CI: 0.18–2.52). However, the length of hospital stay became statistically insignificant (mean difference: −0.31 day, 95% CI: −0.74 day to 0.12 day). Data were insufficient to perform meta-analysis for the requirement of surgery.

Three trials reported the outcomes of 244 patients with high-risk stigmata,[18, 21, 23] i.e. spurting, oozing, or nonbleeding vessel. The recurrent bleeding of using oral PPIs and intravenous PPIs was similar between the oral (11.7%) and intravenous (12.1%) PPIs groups (RR:0.99, 95% CI 0.51–1.92). With very limited sample size, no statistically significant difference was observed among the other outcomes.

One trial was excluded due to the high risk of bias according to the Cochrane classification.[24] The meta-analysis showed the same conclusions. No significant difference was observed on the recurrent bleeding (RR: 0.87, 95% CI: 0.51–1.50), volume of blood transfused (mean difference: 0.01 unit, 95% CI: −0.29 unit to 0.32 unit), requirement of surgery (RR: 0.82, 95% CI: 0.19–3.61) and all-cause mortality (RR: 0.87, 95% CI: 0.25–2.99). The length of hospital stay remained statistically significant (mean difference: −0.85 day, 95% CI: −1.29 day to −0.42 day).


This meta-analysis of 6 randomised controlled trials compared the treatment efficacy between oral and intravenous PPIs for patients with peptic ulcer bleeding. Intravenous PPIs showed no distinct advantage over the oral PPIs in terms of recurrent bleeding, volume of blood transfused, requirement of surgery, and overall mortality but extended hospital stay for intravenous infusion of medication.

A recent randomised controlled trial showed that high dose of intravenous PPIs (80 mg bolus followed by a continuous infusion of 8 mg/h for 72 h) was more effective than standard dose of intravenous PPIs (40 mg bolus twice a day for 72 h) in reducing rate of recurrent bleeding, requirement of blood transfusion and length of hospital stay.[28] In this meta-analysis, high-dose intravenous PPIs showed similar effectiveness to that of oral PPIs. One of the reasons is that intragastric pH for 24-h period of intravenous or oral PPI administration was comparable and maintained greater than 6.[16] The results are also similar to some of previous studies that the inhibitions of gastric acid secretion of 40 mg oral pantoprazole with a similar intravenous dose are equipotent, and the rebleeding rates are comparable in patients with peptic ulcer bleeding.[29, 30]

The risk of recurrent bleeding from a peptic ulcer is highly correlated with the bleeding stigmata seen at endoscopic examinations.[31, 32] This meta-analysis included some low-risk patients with forest classification of IIc or III. These patients would have lower risk of recurrent bleeding and therefore this would be one of the explanations for the comparable effectiveness between oral and intravenous PPIs. In theoretical approach, analysis of each subgroup of bleeding stigmata would help better understanding of the treatment efficacy between oral and intravenous PPIs, but data were not available in the original publications and this subgroup analysis is largely limited by the insufficient sample size. In this meta-analysis, we can only select subgroup trials on high-risk stigmata patients for analysis,[18, 21, 23] and the recurrent bleeding of using oral and intravenous PPIs was similar.

There are some limitations in this meta-analysis. First, this study is likely to be underpowered with only 615 patients for a rare rebleeding rate as the primary outcome. The 95% CI for the recurrent bleeding ranges from 0.56 to 1.50 and therefore it is possible that the effectiveness of oral PPIs may be up to 50% worse than that of IV PPIs. If the sample size is larger, a non-inferiority analysis could demonstrate that oral PPIs are not much worse than IV PPIs under a reasonable non-inferiority threshold. Secondly, lack of blinding is also a key limitation of this meta-analysis, although most of the trials reported procedures of randomisation with clear definitions on outcomes. All included trials were open-labelled with short follow-up period that might induce a significant level of observer bias among physicians and patients. As a result, the overall quality scores of included trials were fair. A double-blinded randomised trial with dummy placebo is required to better assess the role of intravenous and oral PPIs in peptic ulcer bleeding patients. Thirdly, patients with different bleeding stigmata were included in the individual trials. The results may not be totally applicable to patients with bleeding stigmata strategy who are at high risk of rebleeding. However, the individual data are not available in the published literature and therefore a meta-analysis for individual patients by different stigmata is not applicable in this study. Besides, endoscopic therapies were heterogeneously applied among the trials. Patients were selectively treated by epinephrine injection, thermocoagulation, or haemoclips. A meta-analysis showed that the application of haemoclips is superior to injection alone, but comparable to thermocoagulation in producing definitive haemostasis[9] and therefore the clinical outcomes among patients with only epinephrine injection may have more negative outcomes. Moreover, the eligible trials used a variety of PPIs with different pharmacokinetics, including Pantoprazole, Omeprazole, Rabeprazole and Lansoprazole. The pharmaceutical formulas of these PPIs are not exactly the same and their treatment performances are potentially distinct. Subgroup analysis based on different PPIs can be investigated, but the available sample size is too small for such analysis. Lastly, patients who received intravenous PPIs required continuous infusion in hospital for 3 days. It may be unfair to compare oral PPIs in terms of the length of hospitalisations.

Intravenous PPIs have been proved to be an effective treatment to reduce recurrent bleeding after endoscopic therapy of bleeding peptic ulcers.[11, 13] This study demonstrates that oral PPIs have similar clinical effectiveness, but the conclusion is limited by insufficient sample size. Before we have a larger double-blinded controlled trial to confirm the non-inferior effectiveness of oral PPIs, it is worth considering a shorter term of IV PPIs with early conversion oral PPIs for those without the high risk of bleeding stigmata, so that patients can be discharged earlier after endoscopy. With limited healthcare resource for an expanding human population, the cost of treatment is important for hospital administration. The cost of PPIs is directly associated with the pharmacotherapy. Intravenous Pantoprazole is much more expensive than orally administered forms.[33] The intravenous PPIs require extra equipment cost on dedicated intravenous lines that cannot be repeatedly used for blood or crystalloid and colloid replenishment. Patients who received intravenous PPIs also require nursing continuous monitoring on the infusion site reactions, and therefore, a longer hospitalisation is needed. This increases administrative cost of intravenous PPIs. When patients can be early discharged with oral PPIs, the cost of therapy for peptic ulcer bleeding towards the healthcare system is inevitably reduced.

In conclusion, this meta-analysis shows that oral PPIs are clinically similar to intravenous PPIs in controlling of recurrent bleeding, mean volume of blood transfused, requirement of surgery, and overall mortality in patients with peptic ulcer bleedings from different bleeding stigmata. Oral PPIs can significantly reduce the length of hospital stay and probably be a cost-saving approach in hospital administration. However, the results of this meta-analysis are combined from open-labelled trials with limited sample size. A large double-blind non-inferiority trial is required to better assess the role of oral PPIs.


Guarantor of the article: Kelvin KF Tsoi.

Author contributions: Kelvin KF Tsoi: study design, drafting of manuscript; Hoyee W Hirai: data search, extraction and analysis; Joseph JY Sung: study supervision and revision of the manuscript. All authors have approved the final version of the manuscript.


Declaration of personal and funding interests: None.