Critically ill patients are at risk of stress-related gastrointestinal (GI) mucosal damage, ulceration and bleeding [1]. Endoscopic studies have shown that gastric erosions are present in up to 90 % of patients by the third day in the intensive care unit (ICU) [2, 3]. These lesions are, in the vast majority of patients, superficial and asymptomatic, but some can progress and result in overt and clinically important GI bleeding [4]. Clinically important GI bleeding in the ICU is a serious condition, with an estimated one- to four-fold increased risk of death and excess length of ICU stay of 4–8 days [1, 5]. It has been suggested that prophylaxis with acid suppressants reduces the risk of GI bleeding and hence the risk of death [6]. In this context, stress ulcer prophylaxis (SUP) was introduced and is recommended in international guidelines [710] and regarded as standard of care in the ICU [5, 11]. However, clinical research has not been able to confirm that SUP improves outcome [12]. A recent meta-analysis comprising 20 randomised clinical trials (RCTs) comparing proton pump inhibitors (PPIs) and/or histamine-2-receptor antagonists (H2RAs) versus placebo or no prophylaxis did not find any differences in patient important outcome measures between the SUP and the placebo/no prophylaxis groups [12]. Furthermore, concern has been expressed about potentially increased risks of side effects in patients receiving prophylactic treatment with acid suppressants [1316]. The higher gastric pH in these patients may compromise host immunity and increase the risk of pneumonia and Clostridium difficile infection (CDI) [15, 17]. However, no meta-analyses of randomised trials have shown a significantly increased risk of nosocomial pneumonia when using SUP compared to placebo/no prophylaxis [12, 18]. Additionally, no trials have assessed the incidence of CDI in an ICU setting, but a recently published large cohort study found a 2–4 fold increased risk of CDI in adult mechanically ventilated patients receiving PPIs compared to H2RAs [19]. Studies conducted outside the ICU have demonstrated similar findings [20, 21]. Also, an association between the use of PPIs and an increased risk of cardiovascular events has been suggested [18, 22, 23].

Taken together, the balance between benefits and harms of SUP is unclear in critically ill patients in the ICU. The aim of the SUP-ICU trial is to assess the benefits versus harms of PPI (pantoprazole) in acutely ill adults in the ICU. We hypothesise that a PPI reduces the rates of GI bleeding, but increases the rates of nosocomial infections and myocardial ischaemia. The effect on overall mortality is, therefore, unpredictable.


Trial design

The SUP-ICU trial is an investigator-initiated, pragmatic, international, multicentre, randomised, blinded, parallel-group trial of SUP with a PPI versus placebo.


The trial is approved by the Danish Health and Medicine Agency (2015030166), the Committees on Health Research Ethics in the Capital Region of Denmark (H-15003141) and the Danish Data Protection Agency (RH-2015-3203695) and registered at (Identifier: NCT02467621).


European ICUs admitting adult patients.


Inclusion criteria

All adult (18 years or older) patients who are acutely admitted to the ICU with one or more risk factors for GI bleeding [5]:

  • Shock (continuous infusion with vasopressors or inotropes, systolic blood pressure below 90 mmHg, mean arterial blood pressure below 70 mmHg or plasma lactate level 4 mmol/l or above)

  • Acute or chronic intermittent or continuous renal replacement therapy (RRT)

  • Invasive mechanical ventilation which is expected to last more than 24 hours

  • Coagulopathy (platelets below 50 × 109/l, or international normalised ratio (INR) above 1.5, or prothrombin time (PT) above 20 s) documented within the last 24 hours

  • Ongoing treatment with anticoagulant drugs (prophylactic doses excluded)

  • History of coagulopathy (platelets below 50 × 109/l or INR above 1.5 or PT above 20 s within the 6 months prior to hospital admission)

  • History of chronic liver disease (portal hypertension, cirrhosis proven by biopsy, computed tomography (CT) scan or ultrasound or history of variceal bleeding or hepatic encephalopathy)

Exclusion criteria

  • Contraindications to PPIs (including intolerance of PPIs and treatment with atazanavir (anti-human immunodeficiency virus (HIV) medication))

  • Current daily treatment with a PPI and/or a H2RA

  • GI bleeding of any origin during current hospital admission

  • Diagnosed with peptic ulcer during current hospital admission

  • Organ transplant during current hospital admission

  • Withdrawal from active therapy or brain death

  • Fertile woman with positive test for urinary or plasma human chorionic gonadotropin (hCG)

  • Consent according to national regulations not obtainable

Trial medication

Enrolled patients will be randomised to receive either pantoprazole 40 mg (pantoprazole, Actavis, Gentofte, Denmark) or placebo, given once daily intravenously, from randomisation until ICU discharge or death for a maximum of 90 days. Identical vials with and without pantoprazole powder will be masked with a full covering label. The nurse caring for the patient will have access to an electronic medication distribution system, which allows the allocation of the appropriate vial to the patient. The nurse will add 10 ml of sodium chloride to the vial, shake it, and administer the contents intravenously to the patient. As the powder immediately dissolves to a colourless fluid it will not be possible to distinguish dissolved pantoprazole in sodium chloride from sodium chloride alone.

Outcome measures

Primary outcome measure

All-cause mortality 90 days after randomisation

Secondary outcome measures

  • Proportion of patients with one or more of the following adverse events during ICU stay: clinically important GI bleeding, pneumonia, CDI, or acute myocardial ischaemia

  • Proportion of patients with clinically important GI bleeding during ICU stay

  • Proportion of patients with one or more infectious adverse events (pneumonia or CDI) during ICU stay

  • Days alive without use of mechanical ventilation, RRT or circulatory support in the 90-day trial period

  • Number of serious adverse reactions (SARs) during ICU stay

  • Mortality 1 year after randomisation

  • A health economic analysis will be performed. The analytic details will be based on the results of the trial and specified at that time (cost-benefit versus cost-minimisation analyses)

The specific elements of the composite outcomes will be reported in the primary publication.


See Appendix 1.


All patients referred to a participating clinical trial site will be considered for participation (screened). Patients will be eligible if they fulfil all of the inclusion criteria and none of the exclusion criteria listed. Inclusion and exclusion of patients (including reasons for exclusion) will be reported according to the Consolidated Standards of Reporting Trials (CONSORT) statement [24].


Staff at trial sites will have 24-hour access to web-based central randomisation allowing immediate and concealed allocation of trial medication. Randomisation will be performed in blocks with varying block sizes according to the generation of the allocation sequence by the Copenhagen Trial Unit (CTU) [25]. A unique patient identification number will be entered into the system to ensure that the patient is not randomised twice. In addition, each patient will be allocated a unique patient number (screening number).


The allocated trial medication will be blinded to the patient, the clinical staff caring for the patient, the investigators, the outcome assessors, the data manager, the statistician conducting the analyses, and the writing committee when drafting the abstract for the primary publication.

An independent company (Nomeco Clinical Trial Supply Management (CTSM) [26]) will handle masking, coding and distribution of the vials containing the investigational medicinal product (IMP)/placebo. A computer programme will generate the coding list (CTU) with numbers for the vials. Each trial site will have a sufficient number of vials to be allocated to participating patients. This will ensure that the patient only receives the trial intervention they are randomised to receive.


Patients can be withdrawn from the trial if:

  • A clinical indication for treatment with a PPI/H2RA arises (GI bleeding and/or ulcer/gastritis/varices verified endoscopically). Patients will receive treatment for GI bleeding according to local standards

  • Another clinical indication for withdrawal than the above mentioned (judged by responsible clinician or local investigator)

  • A SAR/suspected unexpected serious adverse reaction (SUSAR) occurs (see below)

  • The patient or next of kin withdraws consent

The independent Data Monitoring and Safety Committee (DMSC) can recommend pausing or stopping the trial. Details are provided in Appendix 2.

Serious adverse reactions

Adverse reactions are specified in the product characteristics of pantoprazole. The following conditions related to the intervention will be considered SARs:

  • Anaphylactic reactions

  • Agranulocytosis

  • Pancytopenia

  • Acute hepatic failure

  • Stevens-Johnson syndrome and toxic epidermal necrolysis

  • Interstitial nephritis

  • Angioedema (Quincke’s oedema)

The occurrence of SARs will be recorded daily in the electronic case report form (eCRF) during ICU stay and the distribution of SARs in the two groups will be compared by the DMSC at the interim analyses. During the trial the sponsor will send a yearly report to the ethics committees and medicine agencies.

SUSARs are defined as serious adverse events (SAEs) not described in the product characteristics for pantoprazole. SUSARs will be reported by the trial site investigators to the sponsor within 24 hours. The sponsor will report any SUSAR to the medicine agency within 7 days.

SAEs will not be recorded as an entity because the majority of ICU patients will experience a number of SAEs during their critical illness. SAEs will be captured in the secondary outcome measures.

Patient withdrawal

Patients who are withdrawn from the trial intervention will be followed-up and included in the intention-to-treat analysis. Patients may be withdrawn from the trial according to national consent regulations. In order to limit the amount of missing data, as much data as possible from each patient will be collected. All randomised patients will be reported, and all data available with consent will be used [27].

Patients who are transferred to another ICU will be regarded as discharged from the ICU unless the new ICU is an active SUP-ICU trial site. If so, the allocated trial intervention will be continued. All patients transferred to another ICU will be followed-up for the primary outcome measure.


A predefined analysis plan will be prepared and published before data analysis.

The primary analysis will include the intention-to-treat population comparing mortality 90 days after randomisation in the two groups by binary logistic regression analysis with adjustment for stratification variables: site and active haematological cancer. A secondary analysis will be performed adjusting for stratification variables together with other known major prognostic co-variates: age, baseline Sequential Organ Failure Assessment (SOFA) score, and type of admission (medical, elective surgery or emergency surgery). A sensitivity analysis will be conducted including the per-protocol population, excluding patients with a major protocol violation (patients who did not receive the allocated trial intervention at all, patients who did not receive the trial intervention for at least 2 days in a row, treatment with a PPI or a H2RA without clinical indication and withdrawal from trial intervention). The prevalence and pattern of missing values will be collected and analysed according to the predefined statistical analysis plan. If missingness exceeds 5 % and data is not missing completely at random (Little’s test <0.05) multiple imputation with at least 10 imputations will be performed, and the primary result of the analysis will be from the aggregated intervention effects from the imputed datasets. All statistical tests will be two-tailed and P < 0.05 will be considered statistically significant.

Sample size estimation

Assuming a baseline 90-day mortality of 25 % [5] (see Appendix 3), α = 0.05 (two-sided), and β = 0.1, 3350 patients (2 × 1675) will be needed to show a 20 % relative risk reduction (RRR) or increase (RRI) corresponding to a 5 % absolute risk reduction or risk increase in the primary outcome measure.

Interim analyses

Interim analyses will be performed after 1650 and 2500 patients. The DMSC may recommend pausing or stopping the trial if the group difference in the primary outcome measure, SARs or SUSARs is found at the interim analyses with statistical significance levels adjusted according to the LanDeMets group sequential monitoring boundaries based on the O’Brien-Fleming alpha-spending function, or otherwise finds that the continued conduct of the trial clearly compromises patient safety.

Data registration

Data will be entered into a web-based eCRF (CTU) by trial or clinical personnel. From the eCRF the trial database will be established. Paper case report forms (CRFs) will be used in case of technical difficulties with the eCRF. Details on data collection are shown in Appendix 1.

Data handling and retention

Data will be handled according to the national data protection agencies. All original records (including consent forms, CRFs, SUSAR reports and relevant correspondences) will be retained at trial sites or the CTU for 15 years to allow inspection by the Good Clinical Practice (GCP) Unit or local authorities. The trial database will be maintained for 15 years and anonymised if requested by the authorities.


The trial will be externally monitored according to a monitoring plan developed in collaboration with the GCP Unit in Copenhagen, which will coordinate the monitoring done by local GCP Units and/or monitors in all countries. Trial site investigators will give access to source data. A centralised day-to-day monitoring of the eCRF will be done by the coordinating investigator or her delegates.

Ethical justification

The trial will adhere to the latest version of the Helsinki Declaration [28] and the national laws in the participating countries. Inclusion will start after approval by the ethical committees, medicines agencies and data protection agencies.

Stress ulceration is a condition often seen in critically ill patients in the ICU [1]. The majority of patients will be temporarily incompetent because of severe illness or as a consequence of the treatment, including sedation. We cannot perform the trial in competent patients because less sick (and thus competent) patients do not suffer from stress ulcers. Patients requiring acute treatment in the ICU, e.g. mechanical ventilation, are in an acute life-threatening condition and it would expose the patient to great risk not to initiate the necessary treatment in order to obtain informed consent. To conduct clinical trials with the goal of improving the outcome for ICU patients at risk of stress-related GI bleeding, it is necessary to randomise and enrol patients before obtaining their informed consent. Informed consent will be obtained from all participants or representatives according to the national regulations. The process leading to the achievement of consent may differ in the participating countries, but will be described and be in compliance with all applicable local regulations.

No biological material will be collected for the trial; thus, no bio-bank will be formed.


Patients from Denmark, Finland, Italy, The Netherlands, Norway, Switzerland and the United Kingdom are expected to participate in the trial. The trial will be initiated in Denmark in January 2016 followed by the other countries when national approvals are obtained. The trial is expected to recruit patients during a 2-year period.

Trial management and organisation

The trial is part of the SUP-ICU research programme [29] and is supported by the Centre for Research in Intensive Care (CRIC) and the CTU.

A Steering Committee has been formed consisting of all national principal investigators and a Management Committee (see Appendix 4). The Steering Committee will manage and coordinate the trial centrally.

A local research team consisting of a principal investigator and a trial coordinator will manage and coordinate the trial locally. The principal investigator has the responsibility for data collection and maintenance of trial documentation.

Co-enrolment of participants in other interventional trials has to be approved by the SUP-ICU Steering Committee, but is generally allowed.


Upon trial completion the main manuscript with trial results, whether positive, negative or neutral, will be submitted for peer-review to one of the major clinical journals. Furthermore, the results will be published at the SUP-ICU web page [29].

The Steering Committee will grant authorship depending on personal input according to the Vancouver Principles. The DMSC and investigators not qualifying for authorship will be acknowledged with their names under the ‘SUP-ICU trial investigators’ in an appendix to the final manuscript.

Data sharing

According to the recommendations from the Institute of Medicine and the Scandinavian Trial Alliance a clean file dataset used for final analysis of the main results of the trial, the statistical analysis plan, a variable explanation, and the protocol will be made publicly accessible in an anonymised form 2 years after the last follow-up of the last patients [30].


2014–2015: applications for funding, ethical committees and medicine agencies, development of an eCRF, development of monitoring plan and education of clinical staff

2016–2017: inclusion of patients

2018: data analyses, writing and submission of the main manuscript for publication

2021: data sharing according to the CRIC contract between partners [31]


The trial has been developed and conducted in collaboration with the Scandinavian Critical Care Trial Group (SCCTG). The trial is administered by the CRIC [31]. The CTU has developed the eCRF in close collaboration with the Steering Committee. The web-based randomisation system and the system for allocation of trial medication have been developed and administered by the CTU. Pharma-Skan ApS produces the placebo vials and Nomeco CTSM masks and distributes trial medication to all sites.


The trial is funded by the Innovation Fund Denmark and supported by the Aase and Ejnar Danielsens Foundation, the Ehrenreichs Foundation, the Scandinavian Society of Anaesthesia and Intensive Care Medicine (SSAI), the Danish Society of Anaesthesiology and Intensive Care Medicine (DASAIM), the Danish Medical Association, and the European Society of Intensive Care Medicine. Patient insurances will be sought financed from public and private funds. The funding sources will have no influence on trial design, trial conduct, data handling, data analysis or publication.


Trial rationale

Clinical trials have suggested that there is a reduction in the incidence of GI bleeding among ICU patients receiving SUP compared with ICU patients receiving placebo or no prophylaxis [3, 3238]. Based on this research conducted 15–20 years ago, and because of potentially increased mortality and morbidity in patients with clinically important bleeding, SUP is recommended as a standard of care in critically ill patients [7]. Around 75 % of critically ill patients in the ICU receive an acid suppressant during their ICU stay and PPIs are the most frequently used agents [5]. However, the quantity and quality of evidence supporting a reduction in clinically important GI bleeding and mortality with these agents is low [12]. Importantly, it has been suggested that PPIs may increase the risk of pneumonia, CDI, and acute myocardial ischaemia, and SUP may, in the worst case scenarios, increase mortality [1316]. Taken together, SUP with a PPI is standard of care in ICUs worldwide but has never been tested in large high-quality clinically placebo-controlled trials. As a consequence, PPIs have been used as SUP for several years without convincing evidence of improved outcome.


The population in this trial constitutes adult patients acutely admitted to the ICU with one or more risk factors for GI bleeding [5].


In recent years a PPI has been considered the drug of choice in the management of most acid-related GI disorders [39]. The superior efficacy of PPIs over H2RAs has been demonstrated in various GI disorders, including peptic ulcer disease [39], and randomised trials and meta-analyses have assessed PPIs compared to H2RAs as SUP in the ICU. A recently published meta-analysis by Alhazzani et al. (14 trials, 1720 patients) compared a PPI and a H2RA [40], and found that a PPI was more efficient in reducing clinical important and overt GI bleeding, but no differences were shown regarding mortality, length of stay or incidence of pneumonia [40].

In most countries PPIs are more frequently used as SUP than H2RAs [5]. Since PPIs are considered equally effective, and pantoprazole is the most frequently used PPI [5], we chose this as the intervention.


As described in the previous section, it has been suggested that a PPI is superior to a H2RA in the prevention of clinically important and overt GI bleeding. However, before comparing different SUP agents we need firm evidence of SUP being superior to placebo. This information is currently not available [12].


Assessing mortality as the primary outcome has a number of advantages. First, mortality has not been the primary outcome of previous trials and we are sceptical that previous trials have collected high-quality data on mortality other than short-term mortality (ICU/hospital) [12]. Second, nearly all previous trials assessing PPIs or H2RAs as SUP have high risks of bias [12]. We know that trials with high risks of bias tend to overestimate benefit and underestimate harm [41]. Accordingly, previous trial results might be biased and even though they seem to find a neutral effect on mortality this may be a biased estimate actually concealing excess mortality in the SUP group. Third, meta-analysis of previous trials did not reach a realistic information size so even neutral mortality estimates may be misleading [12]. Fourth, as a consequence of the 6S trial [42], where we found that bleeding was associated with death and that death was partly mediated by bleeding (and renal insufficiency), it appears less likely that there should be a clinically significant reduction in GI bleeding (if PPIs do prevent GI bleeding) without any effect on mortality [43]. Consequently, assessing mortality as the primary outcome measure gives the opportunity to weigh up potential benefits and harms.

Sample size

It is difficult to produce reliable sample size estimations according to anticipated effects on GI bleeding because we have no reliable control group data due to the widespread use of PPIs [5]. As a consequence, it has been necessary to calculate sample size estimations given that something may change if we stop/avoid using PPIs until GI bleeding actually happens (see Appendix 3). The chosen intervention effect of 20 % RRR or RRI of the primary outcome may seem high, but in a population with septic shock or in, e.g. patients after cardiac arrest, a 20 % hazard ratio reduction corresponds to 1 month of extra median survival in patients with a median survival time of approximately 5 months. Furthermore, 3350 patients included in a low-risk-of-bias trial would make a huge contribution to existing evidence, more than doubling the number of randomised patients and providing trial results with low risk of bias on mortality and SAEs. Additionally, trial sequential analysis (TSA) [44, 45] of existing trials (n = 16) has shown that 34 % (1584 patients) of the required information size to detect or reject a 20 % RRR has been accrued; corresponding to a required information size of 4575 patients [12] (see Appendix 5). Consequently, there is an information gap of around 3000 patients assuming a 20 % RRR in mortality. With the inclusion of an additional 3350 patients it is expected that the pooled effect will cross the boundary for benefit/harm or the boundary for futility.

However, no single trial, whether large or well-conducted, gives the final answer and the SUP-ICU trial will not be an exception. Thus, existing meta-analyses of SUP should be updated with the SUP-IUC trial results.


The SUP-ICU trial is a large multicentre clinical trial designed to provide high-quality data with low risk of bias. The trial is monitored according to GCP standards, and before data analyses a statistical analysis plan will be available. Furthermore, the strengths include concealed group assignment, blinding of the patient, the clinical staff caring for the patient, the investigators, the outcome assessors, the data manager, and the trial statistician. The trial design is pragmatic with routine practice maintained except from prescription of SUP; with resulting high generalisability.

Prior to designing the trial we have thoroughly described the available evidence in systematic reviews and a meta-analysis with TSA [12, 46]. Determining the incidence of GI bleeding in critically ill patients in the ICU is complicated by varying definitions of the outcome, difficulties in measuring the outcome, and differences in case mix. To make sure the available data on GI bleeding and risk factors were valid and up-to-date we conducted a large international observational study assessing the incidence of GI bleeding, risk factors for GI bleeding, and the use of SUP in more than 1000 adult critically ill patients in the ICU [5].


As already described in previous sections the sample size estimation is based on estimates, as we do not have valid data describing mortality among patients with risk factors for GI bleeding not treated with a PPI due to the widespread use of acid suppressants. The power for even major effects on each of the possible side effects (pneumonia, CDI and acute myocardial ischaemia) may be reduced, but it will still make a large contribution to our knowledge on these outcomes that may seriously question, overthrow or confirm what we know so far. Furthermore, assessing the potential side effects as a composite outcome measure will increase the power. Additionally, there is a risk of excluding high-risk patients as patients already receiving daily treatment with a PPI or a H2RA cannot be enrolled in the trial due to the risk of discontinuing a therapy for another indication, e.g. history of peptic ulcer. The definition of overt GI bleeding includes haematochezia which might occur from a lower GI bleeding source not affected by PPI, e.g. colonic bleeding. Finally, we do not assess the use of a H2RA or other SUP agents and will not be able to draw conclusions about these drugs.


The SUP-ICU trial will provide important high-quality data and the results will inform clinicians, guideline committee members and policy-makers on the use of SUP in ICU patients. Together with existing data the trial will establish a more solid evidence base for the use of a prophylactic PPI in critically ill patients in the ICU.

Trial status

Recruiting. First patient planned for inclusion in January 2016.