Abstract
Purpose
Most intensive care unit (ICU) patients receive stress ulcer prophylaxis. We present updated evidence on the effects of prophylactic proton pump inhibitors (PPIs) or histamine 2 receptor antagonists (H2RAs) versus placebo/no prophylaxis on patient-important outcomes in adult ICU patients.
Methods
We conducted a systematic review with meta-analysis and trial sequential analysis (TSA) of randomised clinical trials assessing the effects of PPI/H2RA versus placebo/no prophylaxis on mortality, gastrointestinal (GI) bleeding, serious adverse events (SAEs), health-related quality of life (HRQoL), myocardial ischemia, pneumonia, and Clostridium (Cl.) difficile enteritis in ICU patients.
Results
We identified 42 trials randomising 6899 ICU patients; 3 had overall low risk of bias. We did not find an effect of stress ulcer prophylaxis on mortality [relative risk 1.03, 95% confidence interval (CI) 0.94–1.14; TSA-adjusted CI 0.94–1.14], but the occurrence of any GI bleeding was reduced as compared with placebo/no prophylaxis (0.60, 95% CI 0.47–0.77; TSA-adjusted CI 0.36–1.00). The conventional meta-analysis indicated that clinically important GI bleeding was reduced (RR 0.63, 95% CI 0.48–0.81), but the TSA-adjusted CI 0.35–1.13 indicated lack of firm evidence. The effects of stress ulcer prophylaxis on SAEs, HRQoL, pneumonia, myocardial ischemia and Cl. difficile enteritis are uncertain.
Conclusions
In this updated systematic review, we were able to refute a relative change of 20% of mortality. The occurrence of GI bleeding was reduced, but we lack firm evidence for a reduction in clinically important GI bleeding. The effects on SAEs, HRQoL, pneumonia, myocardial ischemia and Cl. difficile enteritis remain inconclusive.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Stress ulcer prophylaxis with PPI or H2RA did not seem to affect mortality, but likely reduced the occurrence of gastrointestinal bleeding. |
Introduction
Patients admitted to the intensive care unit (ICU) are at risk of stress-related gastrointestinal (GI) mucosal damage that may evolve to ulceration and bleeding [1]. The reported prevalence of GI bleeding ranges from 5 to 10% in recent reports, and GI bleeding is associated with an increased risk of death and length of stay in the ICU [2,3,4,5]. Stress ulcer prophylaxis is routinely used in the ICU, even though recommendations in international guidelines are conflicting [6, 7]. However, the quantity and quality of evidence supporting use of stress ulcer prophylaxis in adult ICU patients is low with no firm evidence for benefit or harm [8, 9]. Importantly, increased rates of myocardial ischaemia, Clostridium (Cl.) difficile enteritis and hospital-acquired pneumonia with the use of stress ulcer prophylaxis have been suggested [1, 8, 10, 11]. Several randomised clinical trials (RCT) and systematic reviews have compared the effects of proton pump inhibitors (PPIs) and histamine-2-receptor antagonist (H2RAs), but neither PPIs nor H2RAs have demonstrated superiority as compared with placebo or no prophylaxis [10, 12,13,14,15].
Recently, new relevant trials, including the SUP-ICU trial, have been published [3, 5, 16,17,18]. Consequently, we performed an updated systematic review on stress ulcer prophylaxis with PPI or H2RA versus placebo or no prophylaxis in adult ICU patients. We hypothesised an absence of effect on mortality, a reduction of GI bleeding, and an increase of infectious adverse events and myocardial ischemia.
Methods
We conducted this systematic review according to the preplanned statistical analysis plan of the published protocol [19]. We registered the protocol in the international prospective register of systematic reviews database (PROSPERO) (CRD42018089151) and used the methodology of the Cochrane Collaboration [20], the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) [S1, Electronic Supplementary Material, (ESM)] [21], Keus et al. [22], Jakobsen et al. [23], and Grading of Recommendations Assessment, Development, and Evaluation (GRADE) [24].
Eligibility criteria
We included any RCT comparing stress ulcer prophylaxis with either PPI or H2RA versus placebo or no prophylaxis in adult ICU patients. We accepted any dose, formulation and duration of intervention [19].
Search methods for identification of studies
We did not restrict the search by language, date, publication status or any other trial characteristics. MB searched the following electronic databases: Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library; Ovid MEDLINE; Ovid Embase; Science Citation Index Expanded (Web of Science); Biosis Previews (Web of Science); and PubMed. The systematic search included the following keywords: peptic ulcer; gastrointestinal haemorrhage; proton pumps; histamine h2 receptor antagonists; critical illness; critical care; intensive care units; artificial respiration; craniocerebral trauma; heart arrest; myocardial infarction; sepsis; and surgery. The full search is available in the ESM. The literature search was updated on 11 October 2018. We manually identified additional potential eligible trials by screening the reference lists of the included studies, other relevant systematic reviews, and searched trial registries.
Selection of studies
At least two authors (MB, SM, AG or CTA) independently screened each title and abstract. Reports deemed potentially relevant were obtained in full-text and assessed for inclusion in accordance with the inclusion criteria. Disagreements were resolved by consensus and MHM/JW were consulted when agreement could not be met.
Data extraction and management
Two review authors (MB and SM) independently extracted predefined data of the included trials using a predefined data collection form (S2, ESM). The following data were collected: (1) Trial: country, duration of the trial, date of publication, and type of trial (single versus multi centre); (2) Participants: numbers randomised, numbers analysed, numbers lost to follow-up/withdrawn, type of population, mean/median age, sex, inclusion criteria, and exclusion criteria; (3) Interventions: intervention, comparator, and concomitant interventions; (4) Outcomes: predefined primary and secondary outcomes [19].
Outcomes
Predefined co-primary outcomes were all-cause mortality and the proportion of participants with any GI bleeding (overt and clinically important bleeding defined by trialists). Co-secondary outcomes were: the proportion of participants with one or more serious adverse events (SAEs) (as defined by trialists using the term ‘serious adverse event’, ‘severe adverse event’, ‘serious adverse reaction’, ‘serious complication’, ‘severe complication’ or similar terms fulfilling the criteria of the Good Clinical Practice Guideline of the International Conference on Harmonization (ICH-GCP) definition [25]); health-related quality of life (HRQoL) (any valid scale used by trialists); proportion of participants with myocardial ischemia (as defined by trialists); proportion of participants with hospital-acquired pneumonia (as defined by trialists); proportion of participants with CI. difficile enteritis (as defined by trialists).
For all outcomes, we used the trial results reported at time-points closest to 90 days.
Risk of bias
MB and SM independently assessed the risk of systematic errors (bias) in the included trials using the Cochrane Collaboration’s risk of bias tool [20], with additional prespecified criteria (ESM) [19]. Two review contributors not involved in the SUP-ICU trial [3] assessed risk of bias and extracted data from this trial. We specifically assessed the following domains: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incomplete outcome data; (6) selective reporting; and (7) other biases, including baseline imbalance, early stopping and bias due to vested financial interest or academic bias. The included trials were judged as ‘overall low risk of bias’ when all bias domains were judged as low risk of bias. Conversely, trials were judged as ‘overall high risk of bias’ when unclear or high risk of bias was judged in one or more domains [26].
We assessed publication bias by inspecting funnel plots for signs of asymmetry when ten or more trials were included in an analysis [20, 23]. We tested asymmetry with the Harbord test [27].
Data synthesis
Summary measures
We calculated relative risks (RRs) with 95% confidence intervals (CIs) and trial sequential analysis (TSA)-adjusted CIs [28] for all outcomes. We hypothesised an absence of effect on mortality, a reduction of GI bleeding, and an increase of infectious adverse events and myocardial ischemia, assuming a required information size corresponding to a relative risk reduction (RRR) or a relative risk increase (RRI) of 20% [19, 29].
Meta-analyses
The primary analysis included trials with overall low risk of bias. We calculated pooled effect estimates using Review Manager [30]. We considered a P value of 0.05/[(2 + 1)/2] = 0.033 or less as statistically significant in the analyses of each primary outcome, and we considered a P value of 0.05/[(5 + 1)/2] = 0.017 or less as statistically significant in the analyses of each secondary outcome, in order to restrict the family-wise error rates (FWER) to 0.05 [23]. We calculated Bayes factor to assess if the summary effect estimates fitted better with the null hypothesis than alternative hypotheses of the anticipated intervention effects [23].
Dealing with missing data
Corresponding authors were contacted to clarify important missing data related to the methods, data reporting, or if further trial details were needed (S4, ESM).
We conducted a predefined sensitivity analysis by imputing missing outcome data in a best-/worst-case scenario and a worst-/best-case scenario to assess the potential impact of loss to follow-up. In the best-/worst-case scenario analysis, it was assumed that all participants lost to follow-up in the experimental group did not experience the event, and that all those with missing outcomes in the control group did experience the event. In the worst-/best-case scenario analysis, it was assumed that all participants lost to follow-up in the experimental group did not experience the event, and that all those with missing outcomes in the control group did experience the event [19, 23].
Assessment of heterogeneity
We assessed heterogeneity by visual inspection of the forest plots, the inconsistency statistics (I2) and the estimates of diversity (D2) [31]. When I2 = 0, we used a fixed effects model [32, 33], and when I2 was above zero, we used both fixed and random effects models [32, 34, 35], and reported the most conservative estimate being the point estimate closest to no effect or the estimate with the widest CI.
Subgroup analyses
We planned to conduct the following predefined subgroup analyses: high versus low risk of bias; medical versus surgical versus mixed ICU setting; shock versus no shock; renal replacement therapy (RRT) versus no RRT; invasive mechanical ventilation versus no invasive mechanical ventilation versus unknown status; PPI versus H2RA; and placebo versus no prophylaxis [19]. In addition, we conducted post hoc subgroup analyses on the co-primary outcomes: one according to a dose of PPI (max 40 mg daily versus > 40 mg daily) and one according to publication year (median publication date 1993/1994). We accepted the definitions used in the included trials, and only trials defining subgroups on a trial level were included. Presence of statistical heterogeneity was assessed by the χ2 test with significance set at P < 0.10 [19].
Sensitivity analyses
We conducted a sensitivity analysis to assess the potential impact of reporting bias by excluding trials not reporting on clinically important bleeding [19].
In two post hoc sensitivity analyses, we estimated the number of patients with one or more SAEs: (1) highest proportion of reported SAEs in each trial, and (2) all reported SAEs cumulated in each trial (information available in the ESM).
Trial sequential analysis
TSA is a sequential meta-analysis considering how much information (randomised patients) is needed to conclude on a specific a priori anticipated intervention effect in updated, repetitive testing meta-analyses. If information size is smaller than required in the meta-analysis, the TSA-adjusted CI becomes wider than the conventional naïve, meta-analytic 95% CI, and the threshold for statistical significance becomes more restrictive. However, if the required information size is reached, the TSA-adjusted CI and the naïve CI, anticipating a specific intervention effect, becomes identical.
We used TSA to assess the risk of random errors due to sparse data and multiple testing of accumulating data [36,37,38,39,40,41,42,43,44], and to calculate the required information size [31]. The calculated required information size takes into account the control event proportion, the anticipated heterogeneity variance (D2) [22] of the meta-analysis, and the assumption of a plausible RRR or RRI.
We used a FWER of 5% [23] leading to a statistical significance level of 3.3% and 96.7% CIs for each of the two co-primary outcomes and 1.7% and 98.3% CIs, respectively, for each of the five co-secondary outcomes [19]. We used a beta of 10%, and a D2 [31] as suggested by the trials in the meta-analysis [23], or a D2 of 20% if the measured heterogeneity was zero [45]. As anticipated intervention effects for the primary and secondary outcomes in the TSA, we used a realistic a priori RRR or RRI of 20%. Furthermore, we used an RRR or RRI based on the 95% confidence limit closest to a null effect in the traditional meta-analysis [19]. In addition, we have made a TSA anticipating a 15% RRR of mortality on the meta-analysis of new trials published after our first review [34].
We present 95% CIs and TSA-adjusted CIs, adjusted for multiplicity of outcomes, sparse data, and repetitive testing for all estimates. For a more detailed description of the statistical analysis plan and TSA, we refer to the published protocol [19].
Grading quality of evidence
We used the GRADE approach [24] to assess the overall certainty of evidence for all outcomes. We appraised the certainty of evidence and our confidence in the effect estimates based on risk of bias, inconsistency, indirectness, imprecision and publication bias. Thus, we rated the overall certainty of evidence as high, moderate, low or very low.
Results
Study selection
We identified 10,054 references (Fig. 1) and included 41 RCTs [3,4,5, 12, 16,17,18, 46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79] with a total of 6790 participants. Some 37 trials were in English, 2 in German [75, 78], 1 in Portuguese [54], and 1 in French [61].
Characteristics of the included studies
The included trials were published between 1977 and 2018. Some 35 trials were published as full trial reports and 6 as conference abstracts. The 41 included trials covered 44 trial comparisons; 32 trials assessed H2RAs and 12 assessed PPIs. The control group was placebo in 31 trials and no prophylaxis in 13 trials. Details and additional information of the included trials are presented in S3 and S4, ESM. Characteristics of the excluded studies and ongoing trials are summarised in S5, ESM.
Risk of bias assessment
Three trials were judged as having overall low risk of bias [3,4,5]; the remaining 38 all had overall high risk of bias (Figs. 2 and S4 in the ESM) [12, 16,17,18, 46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79].
Outcomes
Mortality
A total of 28 trials with 5656 participants reported data on all-cause mortality, including the 3 trials with overall low risk of bias with 3587 participants.
The meta-analysis of the three trials with overall low risk of bias did not show any difference in all-cause mortality between stress ulcer prophylaxis and placebo/no prophylaxis: RR 1.03 (95% CI 0.94, 1.14; P = 0.52; I2 = 0%; TSA-adjusted CI 0.94, 1.14; Bayes factor 239,649) (Fig. 3) (S6–S9, ESM). TSA showed that the boundary for futility was crossed, indicating firm evidence for no difference in mortality between the groups. The certainty of evidence, using the GRADE approach, was high (Table 1).
The corresponding summary estimate of all 28 trials (n = 5656) regardless of risk of bias was RR 1.01 (95% CI 0.93, 1.10; P = 0.75; I2 = 0%; TSA-adjusted CI was 0.93, 1.10; Bayes factor 941,833) (Fig. 3).
The sensitivity analyses on missing data were consistent with the primary analysis (S10–S11, ESM), and Harbord’s test did not indicate asymmetry [P = 0.83 (S12, ESM)]. The certainty of evidence was moderate due to risk of bias (Table 1).
The subgroup analyses of PPI versus H2RA and placebo versus no prophylaxis showed no interaction (Table S6, ESM). We observed an interaction in the subgroup analysis of ICU setting (test-of-interaction P = 0.08), suggesting that surgical ICU patients had lower risk of mortality with stress ulcer prophylaxis, compared with medical ICU patients (S6, ESM). Additional subgroup analyses were consistent with the primary analysis (Table S6, ESM). The subgroup analyses of RRT versus no RRT and shock versus no shock could not be performed as no trials (nor stratified subgroups) were eligible for inclusion in these analyses. In the post hoc subgroup analyses of dosing of PPI and publication year, there was no interaction (Table S6, ESM). TSA anticipating a 15% RRR showed that the boundary for futility was crossed, indicating firm evidence for no difference in mortality between the groups (S8, ESM).
GI bleeding
A total of 39 trials with 6627 participants reported on GI bleeding, including the three trials with overall low risk of bias with 3596 participants.
The meta-analysis of the three trials with overall low risk of bias showed a reduction in GI bleeding with stress ulcer prophylaxis versus placebo/no prophylaxis: RR 0.60 (95% CI 0.47, 0.77; P < 0.0001; I2 = 0%; TSA-adjusted CI 0.36, 1.00; Bayes factor 0.004) (Fig. 4), and TSA showed that the required information size to detect a 20% relative difference had been reached (S13, ESM). The certainty of evidence was high (Table 1).
The corresponding summary effect estimate of all 39 trials (n = 6627) regardless of risk of bias was RR 0.52 (95% CI 0.45, 0.61; P < 0.00,001; I2 = 43%; TSA-adjusted CI 0.39, 0.68; Bayes factor 9 × 10−9) and TSA showed that the required information size to detect a 20% relative difference had been reached (Fig. 4).
The sensitivity analyses on missing data were consistent with the primary analysis (S10 and S11, ESM), and Harbord’s test did not indicate asymmetry [P = 0.33 (S16, ESM)]. The certainty of evidence was low due to risk of bias and inconsistency (Table 1).
The subgroup analyses of PPI versus H2RA and placebo versus no prophylaxis showed no interaction (Table S6, ESM). Additional subgroup analyses were consistent with the primary analysis (S6, S14 and S15, ESM). In the post hoc subgroup analyses of dosing of PPI and publication year, there was no interaction (Table S6, ESM).
A total of 14 trials (n = 4833) reported on clinically important GI bleeding. The meta-analysis showed a reduction in clinically important GI bleeding with stress ulcer prophylaxis versus placebo/no prophylaxis: RR 0.63 (95% CI 0.48, 0.81; P = 0.0005; I2 = 1%, Bayes factor 0.017) (S17, ESM). However, this was not confirmed by TSA (TSA-adjusted CI 0.35, 1.13), indicating that the required information size to detect or reject a 20% relative difference had not been reached (S18, ESM).
Serious adverse events
Four trials (three with overall low risk of bias, n = 3587 participants) reported on SAEs [3, 12, 52, 64], although not defining the adverse events according to ICH-GCP. All four trials reported zero events in each group despite reporting mortality and GI bleeding.
A total of 42 trials reported on outcomes categorised by us as SAEs according to the ICH-GCP definition [25] (S19 and S24, ESM).
The two post hoc analyses estimating the number of patients with one or more SAEs were inconclusive. Details of the analyses are available in S19–S29, ESM . The certainty of evidence was judged to be low/very low due to risk of bias, inconsistency, imprecision, very serious indirectness and strongly suspected publication bias (Table 1).
Health-related quality of life
No trials reported data on HRQoL.
Myocardial ischaemia
We identified one trial (low risk of bias, 3291 participants) which reported on myocardial ischaemia [3]; RR 1.07 (95% CI 0.85, 1.61). TSA highlighted that only 11% of the required information size had been reached. The certainty of evidence was judged to be low due to very serious imprecision (Table 1).
Hospital-acquired pneumonia
A total of 16 trials with 4951 participants reported data on pneumonia, including the three trials with overall low risk of bias with 3596 participants.
The meta-analysis of the three trials with overall low risk of bias showed no difference in hospital-acquired pneumonia between stress ulcer prophylaxis and placebo/no prophylaxis: RR 1.01 (95% CI 0.87, 1.18; P = 0.64; I2 = 0%; TSA-adjusted CI 0.77, 1.33; Bayes factor 82) (S30 and S31, ESM), and TSA showed that only 52% of the required information size had been reached. The certainty of evidence was moderate due to imprecision (Table 1).
The corresponding summary estimate of all 16 trials (n = 4951) regardless of risk of bias was RR 1.07 (95% CI 0.94, 1.21; P = 0.34; I2 = 0%; TSA-adjusted CI 0.89, 1.27; Bayes factor 7465) (S32 and S33, ESM), and TSA showed that only 70% of the required information size had been reached. The sensitivity analyses of missing data were consistent with the primary analysis (S34 and S35, ESM). Harbord’s test did not indicate asymmetry [P = 0.17 (S36, ESM)]. The certainty of evidence was low due to risk of bias and imprecision (Table 1).
The subgroup analyses of PPI versus H2RA and placebo versus no prophylaxis showed no interaction (Table S6, ESM). Additional subgroup analyses were consistent with the primary analysis; however, there was interaction in the analysis of ICU setting (test-of-interaction P = 0.06), suggesting that medical ICU patients had higher risk of hospital-acquired pneumonia, compared with surgical or mixed ICU patients (S6, ESM).
Cl. difficile enteritis
A total of four trials with 3698 participants reported data on Cl. difficile enteritis, including the three trials with overall low risk of bias with 3596 participants.
The meta-analyses of trials with overall low risk of bias and trials regardless of risk of bias were both inconclusive (S37, ESM). TSA highlighted that less than 5% of the required information size had been reached. The certainty of evidence was low/very low due to very serious imprecision and risk of bias (Table 1).
Subgroup analyses of PPI versus H2RA and placebo versus no prophylaxis were not applicable. The sensitivity analyses of missing data and subgroup analyses were consistent with the primary analysis (S38 and S39, ESM).
Discussion
In this updated systematic review, we did not find a difference in mortality between adult ICU patients receiving PPI or H2RA versus placebo/no prophylaxis, and TSA highlighted that the required information size to detect a 20% (and even a 15%) relative difference in mortality had been reached, indicating firm evidence. Furthermore, we found a reduction in the occurrence of any GI bleeding and clinically important GI bleeding, and TSA highlighted that firm evidence for such a reduction in any GI bleeding had been reached; however, this was not the case for clinically important GI bleeding. The effects on the other outcomes, including SAEs, HRQoL, myocardial infarction, pneumonia, and CI. difficile enteritis, were inconclusive.
Strengths and limitations
Strengths of this review include the systematic, transparent and robust methodology used, including the use of the Cochrane Handbook [20], the PRISMA statement [21], a prespecified protocol [19], an up-to-date comprehensive literature search, and the independent study selection, data extraction, and risk of bias assessment by two authors. Also, we used TSA to assess the overall risk of random error to increase the reliability of the results of the meta-analysis, and to identify the required information size. Finally, we assessed the certainty of evidence using GRADE.
Limitations of our review include a risk of clinical heterogeneity between trials. Furthermore, statistical heterogeneity was present in the analyses of GI bleeding and SAEs. To account for systematic errors and missing data in the included trials, we conducted subgroup analyses comparing trials of overall high risk of bias with trials of overall low risk of bias, and sensitivity analyses to account for missing data. We cannot exclude a biased effect estimate of the trials of overall high risk of bias; hence, the certainty of evidence for all trials irrespective of risk of bias was downgraded one level for risk of bias. We were unable to include the losses to follow-up from four trials (n = 81) in the sensitivity analyses exploring uncertainty due to missing data, as the trial reports did not specify to which intervention group these patients belonged. The uncertainty due to loss to follow-up is therefore higher. None of the included trials reported detailed data on SAEs according to the ICH-GCP recommendation [25]; however, four trials reported zero SAEs in both groups, although mortality, clinically important GI bleeding and hospital-acquired pneumonia were reported [12, 52, 64]. Accordingly, SAEs are likely considerably underreported. To estimate the effect on SAEs actually reported in the included trials we conducted two post hoc analyses aiming to estimate the effect on the proportion of patients having one or more SAEs expected to lie between these two extremes. Analysing SAEs according to ICH-GCP may not be optimal in ICU patients who may experience numerous SAEs each day, making it difficult to register them all; thus, a composite outcome as defined by ICH-GCP may be inappropriate. Although we had two co-authors not involved in the SUP-ICU trial to assess the risk of bias in this trial, we acknowledge the potential for indirect conflicts of interests from review authors being involved in the SUP-ICU trial. Finally, limited data on SAEs, HRQoL, myocardial ischemia, pneumonia, and CI. difficile enteritis resulted in no firm evidence on the balance between the benefits and harms for these outcomes.
Our results in relation to previous systematic reviews
Previous systematic reviews have not observed a difference in mortality between PPI/H2RA and placebo/no prophylaxis [80,81,82,83], which our results, including TSA, confirm. Previous reviews have shown conflicting results regarding the effects of stress ulcer prophylaxis on any GI bleeding [80, 82, 83]. Our results show an absolute difference in any GI bleeding of 3.4%, corresponding to a number needed to treat of 35 (CI from 46 fewer to 20 fewer) in trials with overall low risk of bias. Previous reviews have also shown inconsistent results in clinically important GI bleeding [81, 83]. In accordance with previous reviews, we did not observe a statistically significant difference in hospital-acquired pneumonia, indicating no firm evidence for benefit or harm [80,81,82,83]. A recently published systematic review did not report a difference in CI. difficile enteritis which is supported by our results [82]. SAEs, HRQoL, and myocardial ischemia have not been assessed in previous reviews.
Clinical implications and perspectives
Nowadays, GI bleeding, including clinically important GI bleeding, is an important but rare event in adult ICUs. Yet, stress ulcer prophylaxis is used in three out of four acutely admitted adult ICU patients [2], and recommendation on its use is conflicting [6, 7].
Our results indicate that, although we did not find an effect of stress ulcer prophylaxis on mortality, GI bleeding is reduced by almost 50% and clinically important bleeding a little less, which could be used as an argument for using PPI/H2RA as a prophylactic intervention in intensive care patients. Conversely, GI bleeding occurs in 12% of intensive care patients and clinically important GI bleeding in only 5% of the patients with placebo or no intervention. Furthermore, as mortality does not seem to be reduced using PPI/H2RA, it could be argued that the prophylactic use is unnecessary and that treatment with antacids should be reserved for patients developing active GI bleeding. Moreover, a pre-planned subgroup analysis in the recently published SUP-ICU trial suggested excess mortality among patients with a Simplified Acute Physiology Score II greater than 53 allocated to PPI compared with placebo, indicating that the most severely ill patients may be harmed from prophylactic PPI [3]. On the other hand, prophylactic PPI does not appear to substantially increase the number of SAEs, including nosocomial infections and myocardial ischemia. Accordingly, additional data on the importance of disease severity on the overall effects of stress ulcer prophylaxis are needed, along with data on long-term outcomes, HRQoL, and an economic analysis [84].
Conclusions
In this updated systematic review, we were able to refute a relative change of 20% of mortality when prophylactic PPI or H2RA were compared with placebo or no prophylaxis in adult ICU patients. GI bleeding was reduced with PPI or H2RA, but firm evidence for a reduction in clinically important GI bleeding was not found. The effects on SAEs, HRQoL, myocardial ischemia, pneumonia, and CI. difficile enteritis remain inconclusive.
Discrepancy between protocol and review
We used a power of 90%, and not 80% as reported in the protocol [19], as meta-analyses should use a higher (or the same) power as its included trials to be able to communicate the best available evidence.
We choose to report two post hoc analyses of the effect of PPI/H2RA on SAEs as none of the trials reported these according to the ICH-GCP criteria. Furthermore, we conducted two post hoc subgroup analyses according to dose of PPI and publication year. In addition, we have made a TSA anticipating a 15% RRR of mortality on the meta-analysis of new trials published after our first review [34].
References
Marik PE, Vasu T, Hirani A, Pachinburavan M (2010) Stress ulcer prophylaxis in the new millennium: a systematic review and meta-analysis. Crit Care Med 38(11):2222–2228
Krag M, Perner A, Wetterslev J, Wise M, Borthwick M, Bendel S et al (2015) Prevalence and outcome of gastrointestinal bleeding and use of acid suppressants in acutely ill adult intensive care patients. Intensive Care Med 41(5):833–845
Krag M, Marker S, Perner A, Wetterslev J, Wise MP, Schefold JC et al (2018) Pantoprazole in patients at risk for gastrointestinal bleeding in the ICU. N Eng J Med. https://doi.org/10.1056/nejmoa1714919
Alhazzani W, Guyatt G, Alshahrani M, Deane AM, Marshall JC, Hall R et al (2017) Withholding pantoprazole for stress ulcer prophylaxis in critically ill patients: a pilot randomized clinical trial and meta-analysis. Crit Care Med 45(7):1121–1129
Selvanderan SP, Summers MJ, Finnis ME, Plummer MP, Ali Abdelhamid Y, Anderson MB et al (2016) Pantoprazole or placebo for stress ulcer prophylaxis (pop-up): randomized double-blind exploratory study. Crit Care Med 44(10):1842–1850
Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R et al (2017) Surviving sepsis campaign: international guidelines for management of sepsis and septic shock: 2016. Intensive Care Med 43(3):304–377
Madsen KR, Lorentzen K, Clausen N, Oberg E, Kirkegaard PR, Maymann-Holler N et al (2014) Guideline for stress ulcer prophylaxis in the intensive care unit. Dan Med J 61(3):C4811
Krag M, Perner A, Wetterslev J, Moller MH (2013) Stress ulcer prophylaxis in the intensive care unit: is it indicated? A topical systematic review. Acta Anaesthesiol Scand 57(7):835–847
Marker S, Krag M, Moller MH (2017) What’s new with stress ulcer prophylaxis in the ICU? Intensive Care Med 43(8):1132–1134
Alhazzani W, Alenezi F, Jaeschke RZ, Moayyedi P, Cook DJ (2013) Proton pump inhibitors versus histamine 2 receptor antagonists for stress ulcer prophylaxis in critically ill patients: a systematic review and meta-analysis. Crit Care Med 41(3):693–705
Lin PC, Chang CH, Hsu PI, Tseng PL, Huang YB (2010) The efficacy and safety of proton pump inhibitors vs histamine-2 receptor antagonists for stress ulcer bleeding prophylaxis among critical care patients: a meta-analysis. Crit Care Med 38(4):1197–1205
Kantorova I, Svoboda P, Scheer P, Doubek J, Rehorkova D, Bosakova H et al (2004) Stress ulcer prophylaxis in critically ill patients: a randomized controlled trial. Hepatogastroenterology 51(57):757–761
Pongprasobchai S, Kridkratoke S, Nopmaneejumruslers C (2009) Proton pump inhibitors for the prevention of stress-related mucosal disease in critically-ill patients: a meta-analysis. J Med Assoc Thai 92(5):632–637
Barkun AN, Bardou M, Pham CQ, Martel M (2012) Proton pump inhibitors vs. histamine 2 receptor antagonists for stress-related mucosal bleeding prophylaxis in critically ill patients: a meta analysis. Am J Gastroenterol 107(4):507–520
Alshamsi F, Belley-Cote E, Cook D, Almenawer SA, Alqahtani Z, Perri D et al (2016) Efficacy and safety of proton pump inhibitors for stress ulcer prophylaxis in critically ill patients: a systematic review and meta-analysis of randomized trials. Crit Care 20(1):120
Liu BL, Li B, Zhang X, Fei Z, Hu SJ, Lin W et al (2013) A randomized controlled study comparing omeprazole and cimetidine for the prophylaxis of stress-related upper gastrointestinal bleeding in patients with intracerebral hemorrhage. J Neurosurg 118(1):115–120
Lin CC, Hsu YL, Chung CS, Lee TH (2016) Stress ulcer prophylaxis in patients being weaned from the ventilator in a respiratory care center: a randomized control trial. J Formos Med Assoc 115(1):19–24
El-Kersh K, Jalil B, Mcclave SA, Cavallazzi R, Guardiola J, Guilkey K et al (2018) Enteral nutrition as stress ulcer prophylaxis in critically ill patients: a randomized controlled exploratory study. J Crit Care 43:108–113
Barbateskovic M, Marker S, Jakobsen JC, Krag M (2018) Stress ulcer prophylaxis in adult intensive care unit patients—a protocol for a systematic review. Acta Anaesthesiol Scand 62(6):744–755
Higgins JPT, Green S (eds) (2011) Cochrane handbook for systematic reviews of interventions version 5.1.0. Updated March 2011. The cochrane collaboration, 2011. Available from www.handbook.cochrane.org
Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP et al (2009) The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ 339:b2700
Keus F, Wetterslev J, Gluud C, Van Laarhoven CJ (2010) Evidence at a glance: error matrix approach for overviewing available evidence. BMC Med Res Methodol 10:90
Jakobsen JC, Wetterslev J, Winkel P, Lange T, Gluud C (2014) Thresholds for statistical and clinical significance in systematic reviews with meta-analytic methods. BMC Med Res Methodol 14:120
Atkins D, Best D, Briss PA, Eccles M, Falck-Ytter Y, Flottorp S et al (2004) Grading quality of evidence and strength of recommendations. BMJ 328(7454):1490
International conference on harmonisation of technical requirements for registration of pharmaceuticals for human use (1997). ICH harmonised tripartite guideline. Guideline for good clinical practice. Updated July 2002
Savovic J, Turner RM, Mawdsley D, Jones HE, Beynon R, Higgins JPT et al (2018) Association between risk-of-bias assessments and results of randomized trials in cochrane reviews: the ROBES meta-epidemiologic study. Am J Epidemiol 187(5):1113–1122
Harbord RM, Egger M, Sterne JA (2006) A modified test for small-study effects in meta-analyses of controlled trials with binary endpoints. Stat Med 25(20):3443–3457
Thorlund K EJ, Wetterslev J, Brok J, Imberger G, Gluud C. User manual for trial sequential analysis (TSA). Www.Ctu.Dk/Tsa/Files/Tsa_Manual.Pdf. Accessed 15 Oct 2018
Wetterslev J, Jakobsen JC, Gluud C (2017) Trial sequential analysis in systematic reviews with meta-analysis. BMC Med Res Methodol 17(1):39
Revman (2014) Review manager (Revman) (computer program), version 5.3. The Nordic Cochrane Center, the Cochrane Collaboration, Copenhagen. https://Community.Cochrane.Org/Help/Tools-and-Software/Revman-5/Revman-5-Download. Accessed 15 Oct 2018
Wetterslev J, Thorlund K, Brok J, Gluud C (2009) Estimating required information size by quantifying diversity in random-effects model meta-analyses. BMC Med Res Methodol 9:86
Mantel N, Haenszel W (1959) Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 22(4):719–748
Demets DL (1987) Methods for combining randomized clinical trials: strengths and limitations. Stat Med 6(3):341–350
Dersimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7(3):177–188
Deeks JJ, Higgins JPT (2010) Statistical algorithms in review manager 5. 2010: RevMan 5.3
Brok J, Thorlund K, Gluud C, Wetterslev J (2008) Trial sequential analysis reveals insufficient information size and potentially false positive results in many meta-analyses. J Clin Epidemiol 61(8):763–769
Brok J, Thorlund K, Wetterslev J, Gluud C (2009) Apparently conclusive meta-analyses may be inconclusive—trial sequential analysis adjustment of random error risk due to repetitive testing of accumulating data in apparently conclusive neonatal meta-analyses. Int J Epidemiol 38(1):287–298
Higgins JP, Whitehead A, Simmonds M (2011) Sequential methods for random-effects meta-analysis. Stat Med 30(9):903–921
Imberger G, Gluud C, Boylan J, Wetterslev J (2015) Systematic reviews of anesthesiologic interventions reported as statistically significant: problems with power, precision, and type 1 error protection. Anesth Analg 121(6):1611–1622
Mascha EJ (2015) Alpha, beta, meta: guidelines for assessing power and type i error in meta-analyses. Anesth Analg 121(6):1430–1433
Pogue JM, Yusuf S (1997) Cumulating evidence from randomized trials: utilizing sequential monitoring boundaries for cumulative meta-analysis. Control Clin Trials 18(6):580–593
Terkawi AS, Mavridis D, Flood P, Wetterslev J, Terkawi RS, Bin Abdulhak AA et al (2016) Does ondansetron modify sympathectomy due to subarachnoid anesthesia?: Meta-analysis, meta-regression, and trial sequential analysis. Anesthesiology 124(4):846–869
Thorlund K, Devereaux PJ, Wetterslev J, Guyatt G, Ioannidis JP, Thabane L et al (2009) Can trial sequential monitoring boundaries reduce spurious inferences from meta-analyses? Int J Epidemiol 38(1):276–286
Wetterslev J, Thorlund K, Brok J, Gluud C (2008) Trial sequential analysis may establish when firm evidence is reached in cumulative meta-analysis. J Clin Epidemiol 61(1):64–75
Thorlund K, Imberger G, Johnston BC, Walsh M, Awad T, Thabane L et al (2012) Evolution of heterogeneity (I2) estimates and their 95% confidence intervals in large meta-analyses. PLoS ONE 7(7):e39471
Apte NM, Karnad DR, Medhekar TP, Tilve GH, Morye S, Bhave GG (1992) Gastric colonization and pneumonia in intubated critically ill patients receiving stress ulcer prophylaxis: a randomized controlled trial. Crit Care Med 20(5):590–593
Basso N, Bagarani M, Materia A, Fiorani S, Lunardi P, Speranza V (1981) Cimetidine and antacid prophylaxis of acute upper gastrointestinal bleeding in high risk patients: controlled, randomized trial. Am J Surg 141(3):339–341
Ben-Menachem T, Fogel R, Patel RV, Touchette M, Zarowitz BJ, Hadzijahic N et al (1994) Prophylaxis for stress-related gastric hemorrhage in the medical intensive care unit: a randomized, controlled, single-blind study. Ann Intern Med 121(8):568–575
Van Den Berg B, Van Blankenstein M (1985) Prevention of stress-induced upper gastrointestinal bleeding by cimetidine in patients on assisted ventilation. Digestion 31(1):1–8
Burgess P, Larson GM, Davidson P, Brown J, Metz CA (1995) Effect of ranitidine on intragastric ph and stress-related upper gastrointestinal bleeding in patients with severe head injury. Dig Dis Sci 40(3):645–650
Cartier F, Gauthier-Lafaye P, Lareng L, Mottin J, Cara M, Passelecq J et al (1980) Cimetideine in patients at risk of stress ulcers: a multi-centre controlled trial. Intensive Care Med 6:54
Chan KH, Lai EC, Tuen H, Ngan JH, Mok F, Fan YW et al (1995) Prospective double-blind placebo-controlled randomized trial on the use of ranitidine in preventing postoperative gastroduodenal complications in high-risk neurosurgical patients. J Neurosurg 82(3):413–417
Darlong V, Jayalakhsmi TS, Kaul HL, Tandon R (2003) Stress ulcer prophylaxis in patients on ventilator. Trop Gastroenterol 24(3):124–128
Domingues SHS, Stoeber GH, Stoeber AC (1985) Ranitidina Injetável Em Pacientes De Alto Risco. Folha Med 91(3):225–228
Friedman CJ, Oblinger MJ, Suratt PM, Bowers J, Goldberg SK, Sperling MH et al (1982) Prophylaxis of upper gastrointestinal hemorrhage in patients requiring mechanical ventilation. Crit Care Med 10(5):316–319
Groll A, Simon JB, Wigle RD, Taguchi K, Todd RJ, Depew WT (1986) Cimetidine prophylaxis for gastrointestinal bleeding in an intensive care unit. Gut 27(2):135–140
Gundogan K, Karakoc E, Teke T, Zerman A, Coruh A, Sungur M (2017) Effects of enteral nutrition on stress ulcer hemorrhage in critically ill patients: multicenter randomized controlled trial. Intensive Care Med Exp 5(2):44
Gursoy O, Memis D, Sut N (2008) Effect of proton pump inhibitors on gastric juice volume, gastric ph and gastric intramucosal pH in critically ill patients: a randomized, double-blind placebo-controlled study. Clin Drug Investig 28(12):777–782
Halloran LG, Zfass AM, Gayle WE, Wheeler CB, Miller JD (1980) Prevention of acute gastrointestinal complications after severe head injury: a controlled trial of cimetidine prophylaxis. Am J Surg 139(1):44–48
Hanisch EW, Encke A, Naujoks F, Windolf J (1998) A randomized, double-blind trial for stress ulcer prophylaxis shows no evidence of increased pneumonia. Am J Surg 176(5):453–457
Hummer-Sigiel M, Jacquier A, Girard A, Garric J, Laxenaire MC, Mandorla JY (1986) Ranitidine Pour La Prophylaxie De L’ulcére De Stress Chexz Les Traumatisés Crâniens Graves. Ann Med Nancy l’Est 25:101–103
Jakob SM, Parviainen I, Ruokonen E, Uusaro A, Takala J (2005) Lack of effect of ranitidine on gastric luminal Ph and mucosal PCO2 during the first day in the ICU. Acta Anaesthesiol Scand 49(3):390–396
Kam J, Modi C, Doraiswamy V, Abdul-Jawad S, Dixit D, Spira T et al (2011) Role of gastrointestinal ulcer prophylaxis in critically ill patients. Am J Gastroenterol 106(suppl. 2):s420
Karlstadt RG, Iberti TJ, Silverstein J, Lindenberg L, Bright-Asare P, Rockhold F et al (1990) Comparison of cimetidine and placebo for the prophylaxis of upper gastrointestinal bleeding due to stress-related gastric mucosal damage in the intensive care unit. J Intensive Care Med 5:26–32
Koelz HR, Aeberhard P, Hassler H, Kunz H, Wagner HE, Roth F et al (1987) Prophylactic treatment of acute gastroduodenal stress ulceration: low-dose antacid treatment without and with additional ranitidine. Scand J Gastroenterol 22(9):1147–1152
Larson GM, Davidson P, Brown J, Wilson T, Bishop A (1989) Comparison of ranitidine versus placebo on 24-hour gastric Ph and upper gastrointestinal (UGI) bleeding in head injury patients. Abstr Am J Gastroenterol 84:1165
Luk GD, Summer WR, Messersmith JF (1982) Cimetidine and antacid in prophylaxis of acute gastrointestinal bleeding: a randomized, double-blind, controlled study. Gastroenterology 82:1121
Macdougall BR, Bailey RJ, Williams R (1977) H2-receptor antagonists and antacids in the prevention of acute gastrointestinal haemorrhage in fulminant hepatic failure: two controlled trials. Lancet 1(8012):617–619
Martin LF, Booth FV, Karlstadt RG, Silverstein JH, Jacobs DM, Hampsey J et al (1993) Continuous intravenous cimetidine decreases stress-related upper gastrointestinal hemorrhage without promoting pneumonia. Crit Care Med 21(1):19–30
Metz CA, Livingston DH, Smith JS, Larson GM, Wilson TH (1993) Impact of multiple risk factors and ranitidine prophylaxis on the development of stress-related upper gastrointestinal bleeding: a prospective, multicenter, double-blind, randomized trial. The Ranitidine Head Injury Study Group. Crit Care Med 21(12):1844–1849
Nielsen HJ, Witt K, Moesgaard F, Kehlet H (1989) Ranitidine for improvement of delayed hypersensitivity response in patients with sepsis. Acta Chir Scand 155(9):445–449
Peura DA, Johnson LF (1985) Cimetidine for prevention and treatment of gastroduodenal mucosal lesions in patients in an intensive care unit. Ann Intern Med 103(2):173–177
Powell H, Morgan M, Li SK, Baron JH (1993) Inhibition of gastric acid secretion in the intensive care unit after coronary artery bypass graft. Theor Surg 8:125–130
Rigaud D, Accary JP, Chastre J, Mignon M, Laigneau JP, Reinberg A et al (1988) Persistence of circadian rhythms in gastric acid, gastrin, and pancreatic polypeptide secretions despite loss of cortisol and body temperature rhythms in man under stress. Gastroenterol Clin Biol 12(1):12–18
Rohde H, Lorenz W, Fischer M (1980) Eine Randomisierte Klinische Studie Zur Stressulkusprophylaxe Mit Cimetidin Beim Schweren Polytrauma. Z Gastroenterol 18(6):328–329
Ruiz-Santana S, Ortiz E, Gonzalez B, Bolanos J, Ruiz-Santana AJ, Manzano JL (1991) Stress-induced gastroduodenal lesions and total parenteral nutrition in critically ill patients: frequency, complications, and the value of prophylactic treatment: a prospective, randomized study. Crit Care Med 19(7):887–891
Spapen H, Diltoer M, Nguyen DN, Ingels G, Ramet J, Huyghens L (1995) One week treatment with cimetidine does not attenuate the cortisol response to a short corticotropin test in stable intensive care patients: a prospective, randomized, and controlled study. Acta Anaesthesiol Belg 46(3–4):133–140
Vlatten A, Wiedeck H, Reinelt H, Stanescu A, Georgieff M (1998) Stressulkus-Prophylaxe Bei Hoch-Risiko-Intensivpatienten. Vergleich Von Omeprazol, Pirenzepin Und Plazebo. Wien Klin Wochenschr Suppl 110(suppl. 1):38
Zinner MJ, Zuidema GD, Smith P, Mignosa M (1981) The prevention of upper gastrointestinal tract bleeding in patients in an intensive care unit. Surg Gynecol Obstet 153(2):214–220
Krag M, Perner A, Wetterslev J, Wise MP, Hylander Moller M (2014) Stress ulcer prophylaxis versus placebo or no prophylaxis in critically ill patients: a systematic review of randomised clinical trials with meta-analysis and trial sequential analysis. Intensive Care Med 40(1):11–22
Toews I, George AT, Peter JV, Kirubakaran R, Fontes LES, Ezekiel JPB et al (2018) Interventions for preventing upper gastrointestinal bleeding in people admitted to intensive care units. Cochrane Database Syst Rev 6:Cd008687
Huang HB, Jiang W, Wang CY, Qin HY, Du B (2018) Stress ulcer prophylaxis in intensive care unit patients receiving enteral nutrition: a systematic review and meta-analysis. Crit Care 22(1):20
Alhazzani W, Alshamsi F, Belley-Cote E, Heels-Ansdell D, Brignardello-Petersen R, Alquraini M et al (2018) Efficacy and safety of stress ulcer prophylaxis in critically ill patients: a network meta-analysis of randomized trials. Intensive Care Med 44(1):1–11
Macias WL, Nelson DR, Williams M, Garg R, Janes J, Sashegyi A (2005) Lack of evidence for qualitative treatment by disease severity interactions in clinical studies of severe sepsis. Crit Care 9(6):R607–R622
Acknowledgements
MB, SM, JCJ, AP and JW were supported by the public Innovation Fund Denmark (4108-00011B), which did not have any role in study design, data collection, data analysis, data interpretation, or writing of the report. No other sources of financial support were obtained for this review. The authors thank Sanam Safi and Kiran Kumar Katakam, who were not involved in any aspects of the SUP-ICU trial, for extracting data and evaluating risk of bias of this trial. We also wish to thank Maria Hernandez Sierra, Aleksandra Mazur, Ning Liang and Dezhao Kong for translating papers.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflicts of interest
Marija Barbateskovic: PhD student at the Copenhagen Trial Unit and the Centre for Research in Intensive Care. Søren Marker: PhD student at the Department of Intensive Care at Rigshospitalet and the Centre for Research in Intensive Care. Coordinating investigator of the randomised clinical trial ‘Stress Ulcer Prophylaxis in the Intensive Care Unit’ (SUP-ICU). Anders Granholm: Coordinating investigator of the SUP-ICU trial. Carl Thomas Anthon: Coordinating investigator of the SUP-ICU trial. Mette Krag: Coordinating investigator of the SUP-ICU trial. Janus Christian Jakobsen: Director of Research, Chief Physician, Department of Cardiology, Holbæk Sygehus, Holbæk, Denmark. Anders Perner: Head of Research at the Department of Intensive Care at Rigshospitalet. The intensive care unit receives support for research from CSL Behring, Fresenius Kabi, Ferring Pharmaceuticals and the Novo Nordisk Foundation. Dr Perner is initiator of the SUP-ICU trial. Jørn Wetterslev: Member of the Copenhagen Trial Unit task force for developing Trial Sequential Analysis theory, manual and software which is presently free-ware at www.ctu.dk/tsa. Dr Wetterslev is member of the SUP-ICU trial steering group. Morten Hylander Møller: Sponsor and initiator of the SUP-ICU trial.
Ethical approval
An approval by an ethics committee was not applicable.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Barbateskovic, M., Marker, S., Granholm, A. et al. Stress ulcer prophylaxis with proton pump inhibitors or histamin-2 receptor antagonists in adult intensive care patients: a systematic review with meta-analysis and trial sequential analysis. Intensive Care Med 45, 143–158 (2019). https://doi.org/10.1007/s00134-019-05526-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00134-019-05526-z