Stress ulcer, gastritis, and gastrointestinal bleeding prophylaxis in critically ill pediatric patients: A systematic review

Review Articles Stress ulcer, gastritis, and gastrointestinal bleeding prophylaxis in critically ill pediatric patients: A systematic review Ludovic Reveiz, MD, MSc; Rafael Guerrero-Lozano, MD; Angela Camacho, MD; Lina Yara, MD; Paola Andrea Mosquera, Psi, MSc Objective: To identify and evaluate the quality of evidence supporting prophylactic use of treatments for stress ulcers and upper gastrointestinal bleeding. Stress ulcers, erosions of the stomach and duodenum, and upper gastrointestinal bleeding are well-known complications of critical illness in children admitted to the pediatric intensive care unit. Data Sources: Studies were identified from the Cochrane Central Register of Controlled Trials, PUBMED; LILACS; Scirus. We also scanned bibliographies of relevant studies. Study Selection: This systematic review of randomized controlled trials assessed the effects of drugs for stress-related ulcers, gastritis, and upper gastrointestinal bleeding in critically ill children admitted to the pediatric intensive care unit. Data Extraction and Synthesis: Two reviewers independently extracted the relevant data. Most randomized s were judged as having unclear risk of bias. When pooling two randomized s, treatment was significantly more effective in preventing upper gastrointestinal bleeding (macroscopic or important bleeding) compared with no treatment (two studies 300 participants; relative risk, 0.41; 95% confidence interval, 0.19 0.91; I 2 12%). Meta-analysis of two studies found no significant difference in death rates among groups (two randomized s 132 participants; relative risk, 1.39; 95% confidence interval, 0.70 2.79; I 2 4%). The rate of pneumonia was not significantly different when comparing treatment and no treatment in one study. When comparing ranitidine with no treatment, significant differences were found in the proportion of mechanically ventilated children with normal gastric mucosal endoscopic findings by histologic specimens (one randomized 48 participants; relative risk, 3.53; 95% confidence interval, 1.34 9.29). No significant differences were found when comparing different drugs (omeprazole, ranitidine, sucralfate, famotidine, amalgate), doses, or regimens for main outcomes (deaths, endoscopic findings of erosion or ulcers, upper gastrointestinal bleeding, or pneumonia). Conclusions: Although pooled data of two studies suggested that critically ill pediatric patients may benefit from receiving prophylactic treatment to prevent upper gastrointestinal bleeding, we found that high-quality evidence to guide clinical practice is still limited. (Pediatr Crit Care Med 2010; 11:124 132) KEY WORDS: children; stress ulcer; gastrointestinal bleeding; intensive care unit; critical care; prophylaxis; systematic review Stress ulcers of the stomach and duodenum as well as upper gastrointestinal (UGI) bleeding are well-known complications of critical illness in children admitted to a pediatric intensive care unit (ICU). The prevalence of stress ulceration in critically ill adults and children may vary depending on the severity of the illness and methods used for diagnosis. A cohort of 1006 consecutive admissions enrolled in a pediatric ICU reported that 10.2% of pediatric participants had UGI bleeding From the Research Institute (LR, PAM), Grupo de Evaluación de Tecnologías, and the Department of Pediatrics (RG-L, AC, LY), National University of Colombia, Bogotá, Colombia. The authors have not disclosed any potential conflicts of interest. For information regarding this article, E-mail: mmreveiz@hotmail.com Copyright 2010 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: 10.1097/PCC.0b013e3181b80e70 and 1.6% had clinically significant UGI bleeding (1). Clinically important UGI bleeding has an important attributable morbidity and mortality in adults, associated with a significant risk of death (relative risk [RR], 4.1; 95% confidence interval [CI], 2.6 6.5) and an excess length of ICU stay of approximately 4 to 8 days (2). Clinically important UGI bleeding is defined as macroscopic bleeding that results in hemodynamic instability and the need for red blood cell transfusion and may lead to complications, such as gastrointestinal perforations and surgery (3, 4). Prophylaxis against stress ulcers has been recommended for the prevention of UGI bleeding in critically ill adults patients. A systematic review published more than one decade ago found that prophylaxis with histamine2-receptor antagonists decreases the occurrence of overt gastrointestinal bleeding (odds ratio [OR], 0.58; 95% CI, 0.42 0.79) and clinically important bleeding (OR, 0.44; 95% CI, 0.22 0.88) (5). Another study found that, among critically ill adult patients requiring mechanical ventilation, those receiving ranitidine had a significantly lower rate of clinically important gastrointestinal bleeding than those treated with sucralfate and no significant differences were found in the rates of ventilator-associated pneumonia, the duration of the stay in the ICU, or mortality (6). However, a more recent integrative study found that ranitidine was ineffective in the prevention of UGI bleeding in patients in intensive care compared with placebo (OR, 0.72; 95% CI, 0.30 1.70) and might increase the risk of pneumonia when compared with sucralfate (OR, 1.35; 95% CI, 1.07 1.70) and that studies on sucralfate do not provide conclusive positive results (7). A guideline on stress ulcer prophylaxis published in 2006 recommended pharmacologic intervention in adults admitted to the ICU who have coagulopathy, require mechanical ventilation for 48 hrs, have a his- 124 Pediatr Crit Care Med 2010 Vol. 11, No. 1

tory of gastrointestinal ulceration or bleeding within 1 yr before admission, or have at least two of the following risk factors: sepsis, ICU stay of 1 wk, occult bleeding lasting 6 days, and use of 250 mg of hydrocortisone or the equivalent (8). Unfortunately, there is still conflicting evidence concerning prophylaxis for stress ulcers in children and we did not find any systematic review on this topic. The aims of the systematic review presented here are to assess the best evidence on the effects of interventions for stress ulcer in children, to identify gaps in the literature, and to suggest further clinical investigation. METHODS Literature Search Relevant randomized s (RCTs) were identified from the Cochrane Central Register of Controlled Trials (The Cochrane Library 2008, Issue 3), PubMed (1966 to June 2008), LILACS (1982 to June 2008), and Scirus (June 2008). A search strategy to locate studies on stress ulcer, UGI in children was structured and adapted according to each electronic database (Appendix A). The International Clinical Trials Registry Platform search portal (http://www.who.int/trialsearch/default.aspx), the metaregister of s (www.controlled-trials.com) and http:// clinicaltrials.gov/ were searched for ongoing trials. Eligible RCTs were included regardless of the language of publication. We also scanned bibliographies of relevant studies for possible references to additional RCTs. Study Selection Two authors independently decided which trials fit the inclusion criteria. Any disagreements were resolved by discussion between the reviewers, with referral to a third author if necessary. Only RCTs of interventions for stress ulcer in hospitalized children (studies that included participants aged 18 yrs) were considered in this systematic review. We considered any hospitalized children including critically ill pediatric patients having or not having mechanical ventilation, children admitted to pediatric ICUs or who underwent surgery, preterm and full-term newborns among others. Quasirandomized and nonrandomized controlled studies were not discussed further. We considered all doses and regimen treatments as single or combined therapy used as prophylaxis against stress ulcers. The comparators were placebo, no treatment, or another active compound. Data Extraction At least two reviewers (L.R., L.Y., P.A.M.) independently extracted the relevant data, using a predesigned data extraction form; disagreement was resolved by consensus with all authors. We extracted year of publication, patient population, number of patients (by intention to treat), aspects of study quality, sociodemographic, interventions (drug, dose, duration of treatment), clinical, endoscopic, and histologic outcomes and adverse effects. Risk of Bias Assessment/GRADE System The Cochrane Collaboration recently proposed a new tool having six domains to assess the risk of bias of RCTs (namely, sequence generation of randomization, allocation concealment to prevent foreknowledge of group assignment in an RCT, blinding, incomplete outcome data, selective outcome reporting and other issues ). Therefore, a risk of bias evaluation of each RCT was done for the assessment of these features (9). Additional quality information reported included also information of withdrawals, inclusion and exclusion criteria, sample size calculation, and baseline comparability of age, gender, relevant clinical characteristics and diagnoses and duration of complaint. The tool for assessing risk of bias in each RCT comprises a description and a judgment for each entry in a risk of bias table. The judgment for each entry involves answering a question, with answers Yes indicating low risk of bias, No indicating high risk of bias, and Unclear indicating either lack of information or uncertainty over the potential for bias. A study should be considered as having low risk of bias if all key domains were judged as Yes and with unclear risk if the reviewers judged Unclear risk of bias for one or more key domains (9). We also used the GRADE system for grading the quality of evidence and the strength of recommendations by two reviewers. A systematic approach for grading the strength of management recommendations can minimize bias and aid interpretation of clinical practice guidelines. This system takes into account study design, study quality, consistency, and directness in judging the quality of evidence for each important outcome (10). Definitions and Outcomes The main outcomes considered were death; presence of ulcers, gastritis, or UGI bleeding; bleeding that requires transfusion; bleeding associated with hemodynamic instability; gastrointestinal perforations; and pneumonia. In addition, endoscopic findings categorized according to the severity of UGI tract lesions were also considered primary outcome. We anticipated that diverse classifications could have been used by trialists. Other outcomes were: delayed gastric emptying; bacteremia during the follow-up; mean patient s gastric aspirates ph during 24-hr monitoring, pediatric ICU stay; duration of mechanical ventilation hematologic test; cultures of gastric and tracheal secretion, gastric and tracheal colonization; and complicated postoperative course. We only included data on adverse events from RCTs; no further searches for other types of studies were done (11). Statistical Analysis Statistical analyses were done with Review Manager version 5.0 (Cochrane Collaboration) software. The results expressed as RR and 95% CI for dichotomous primary outcomes were calculated by the Mantel-Haenszel fixedeffects model. Weighted mean difference with 95% CI was used for continuous outcomes. For the pooled analysis, we calculated the I 2 statistic, which describes the percentage of total variation across studies caused by heterogeneity (9). Low, moderate, and high levels of heterogeneity approximately correspond to I 2 values of 25%, 50%, and 75%, respectively. RESULTS Description of Studies A total of 294 citations were identified from the diverse sources of information. Of the 74 potentially RCTs screened, we excluded 52 references because they were guidelines, observational studies, or case series reports. We identified 22 studies assessing the effects of different therapeutic interventions for stress ulcers in children (Fig. 1). However, we excluded 13 studies because they were nonrandomized or noncontrolled or focused on pharmacokinetics of drugs (12 24). One study in Hungarian is pending for evaluation (25). In total, we included and analyzed eight RCTs evaluating the effects of drugs for preventing stress ulcers in children (26 33). The main characteristics of the eight included studies are detailed in Table 1. Table 1: We could not assess publication bias (e.g., funnel plot or Egger regression test) because we found less than nine RCTs and we could not pool outcomes for more than three studies. Risk of Bias Seven studies were open and we found no RCT with low risk of bias. Overall six Pediatr Crit Care Med 2010 Vol. 11, No. 1 125

Figure 1. Flow diagram of the process of identifying and including references. treated in a neonatal ICU showed that rates of normal gastric mucosal endoscopic findings by visual inspection (one RCT 48 participants; RR, 3.04; 95% CI, 1.30 7.12) and histologic specimens (one RCT 48 participants; RR, 3.53; 95% CI, 1.34 9.29) were significantly higher in the ranitidine group compared with no treatment groups. However, no significant differences among groups were found in the rates of patients with erosions or ulceration, gastrointestinal problems (bleeding from the gastrointestinal tract and/or vomiting or delayed gastric emptying), positive bacterial cultures from the biopsy specimens, and the risk for later suspected or proven bacteremia during the follow-up. An RCT of children who needed mechanical ventilation on admission (30) showed no significant difference between groups in macroscopic bleeding, pneumonia occurrence, duration of mechanical ventilation (days) and pediatric ICU stay (days). Another RCT of children admitted to a pediatric ICU (28) did not show statistical difference in the rates of patients with important UGI hemorrhage among groups. A significant difference favoring ranitidine was found in the percentage of children with mean ph of 4 during 50% of study time compared with no treatment (one RCT 70 participants; RR, 8.67; 95% CI, 2.89 26.02). RCTs were judged as having unclear risk of bias mainly because the description of the method used to generate the sequence of randomization and to conceal the allocation was unclear (Table 2). Some markers of quality in medical research, such as performing a sample size calculation, are unlikely to have direct implications for risk of bias. However, the majority of RCTs did not calculate the sample size, which is a source of potential imprecision. Overall the quality of the reporting and design of the RCTs was poor. Effects of Interventions Treatment Versus No Treatment We found four studies that evaluated a number of medications (cimetidine, almagate, ranitidine, sucralfate, and omeprazole) vs. no treatment or placebo for different outcomes; data were not available in all RCTs for each outcome (26, 28, 30, 33). When pooling two RCTs (Fig. 2), treatment (which included almagate, ranitidine, sucralfate, and omeprazole) was significantly more effective in preventing UGI bleeding (macroscopic or important bleeding) compared with no treatment (two studies 300 participants; RR, 0.41; 95% CI, 0.19 0.91; I 2 12%) (28, 30). However, no significant difference was found when pooling both studies of treatment vs. no treatment with an additional RCT comparing treatment vs. placebo (three studies 340 participants; RR, 0.69; 95% CI, 0.41 1.17; I 2 63%) (26, 28, 30). In addition, metaanalysis of two studies (30, 33) found no significant difference in death rates among groups (two RCTs 132 participants; RR, 1.39; 95% CI, 0.70 2.79; I 2 4%). The rate of pneumonia was not significantly different when comparing treatment and no treatment in one study (30). Summary of relevant findings for primary outcomes of RCTs included in the review are detailed in Table 3. Ranitidine One RCT (33) in mechanically ventilated preterm and full-term newborns Amalgate An RCT (28) showed a significant difference favoring amalgate in the percentage of patients gastric aspirates with mean ph of 4 during 50% of study time (one RCT 70 participants; RR, 11.00; 95% CI, 3.72 32.56). No significant difference was found among groups regarding the rates of patients with important UGI hemorrhage. Sucralfate In one RCT (28), a trend favoring sucralfate was found in the percentage of patients gastric aspirates with mean ph of 4 during 50% of study time (one RCT 70 participants; RR, 3.33; 95% CI, 1.00 11.09). However, no significant difference between groups was found for other outcomes, such as rate of patients with important UGI hemorrhage. Another RCT (30) did not report significant difference among groups in rates of macroscopic bleeding, pneumonia, and deaths as well as in the duration of me- 126 Pediatr Crit Care Med 2010 Vol. 11, No. 1

Table 1. Characteristics of randomized s included in the review Study Methods Participants Interventions Outcomes Aanpreung et al (31) Behrens et al (32) Kuusela et al (33) Lacroix et al (26) patients from the pirenzepine or famotidine group were randomized. No treatment group vs. treatment group was not randomized. Double-blind randomized Lopez-Herce et al (27) Lopez-Herce et al (28) Twenty critically ill pediatric patients aged 2 mos to 12 yrs. Severity of disease was assessed using Zinner index score. Children who underwent corrective or palliative surgery for congenital heart disease. Age, weight, cardiopulmonary bypass time, aortic cross clamp period, and mean interval between endoscopy and cardiac surgery were not significantly different in the two groups. The study group was prospectively collected from mechanically ventilated preterm and full-term newborns treated in a neonatal ICU. The criterion for inclusion in the study was the start of mechanical ventilation during the first 2 hrs of life. Most of the infants were preterm; the mean gestational age was 32 wks (range 24 41 wks) and the mean birth weight was 1832 g (range 620 4550 g). Twenty-nine of the neonates were male and 24 were female. Altogether, 37 preterm infants of gestational age of 33 wks and 16 infants of gestational age of 33 wks were enrolled in the study. Forty children from birth to 18 yrs old admitted to PICU. Inclusion criteria were that illness was severe enough to preclude any oral or enteral nutrition for at least 2 days. Exclusion criteria: UGI bleeding, burns, or surgical problems; need of oral or enteral feeding; renal failure or cerebral death; treatment requiring cimeridine or antiacids. Mean SD age was 1.85 3.25 yrs. Forty patients admitted to PICU ranging from neonate to 17 yrs old were included. 165 children admitted to PICU presenting at least one of the following criteria: shock, acute renal, cardiac respiratory or liver failure, sepsis or serious focal infection, coagulopathy, acute neurologic dysfunction, multiple trauma, severe metabolic acidosis post major surgery. All patients had nasogastric tube inserted. Severity of illness was evaluated with three scores. Intravenous ranitidine 1.5 mg/kg every 6 hrs and famotidine 0.4 mg/kg every 8 hrs The first 36 patients (group 1) were not given treatment to prevent lesions of the UGI tract. Later, 43 patients (group 2) were randomized and treated either with pirenzepine, an anticholinergic (21 patients) or famotidine, a H2 antagonist (22 patients). Both drugs were given intravenously at a dosage of 1 mg/ kg/day. In older patients, the drugs were given as two doses; children who weighed 10 kg were given three doses. Fifty-three mechanically ventilated newborns were randomized into either the treatment group (prophylactic intravenous ranitidine 5 mg/kg body weight/day divided into three doses throughout 4 days) or the control group (no prophylaxis). Prophylactic treatment commenced immediately parallel to mechanical ventilation. There was no placebo treatment available. Patients were randomized to cimetidine 0.13 ml/kg/day (ampoule 150/mg/mL) or placebo every 6 hrs as long as 10 days. Patients were randomized into four groups. All of them received ranitidine at different dosages: a) 2 mg/kg by nasogastric tube every 12 hrs; b) 4 mg/kg by nasogastric tube every 12 hrs; c) 0.75 mg/kg iv every 6 hrs; d) 1.5 mg/kg iv every 6 hrs. Participants were randomized to four groups: no treatment; almagate 0.25 ml/kg every 2 hrs; ranitidine 1.5 mg/kg iv every 6 hrs; and sucralfate 0.5 gr if weighing 10 kg and 1 gr if weighing 10 kg every 6 hrs. Patients intragastric ph was measured by continuous ph monitoring digitrapper. Intensity of UGI hemorrhage was classified into three categories: nonhemorrhage; slight; and important. All patients had at least one endoscopic examination performed by the same examiner as the patients still required mechanical ventilation. To assess the severity of lesions of the UGI tract, authors developed a score based on the endoscopic findings: Normal findings 0; Mild-to-moderate inflammation, few petechiae or erosions 1; Pronounced inflammation, multiple petechiae erosions 2; Ulcer(s) 3. Additional outcomes: continuous 24-hr measurement of ph; tracheal and gastric secretions culture; daily routine chest radiographs on the first 3 postoperative days and afterwards chest radiographs were taken if indicated by the clinical findings. The primary outcome variable was mucosal lesions detected endoscopically. The procedure was planned for all patients at the age of 3 to 6 days. The findings were grouped into four categories: a) intact gastric mucosa; b) mucosal friability; c) erythema or gross blood; and d) erosions or ulcers. Gastric mucosal biopsy specimens were obtained for histological and bacteriologic evaluation if there were no contraindications. Alcian-blueperiodic-acid-Schiff and modified Giemsa stains were used to demonstrate fungi and bacteria. Biopsy specimens were also obtained for bacterial culture. UGI bleeding noted from nasogastric tube. Massive UGI bleeding was defined as brown hemorrhage from nasogastric tube and a decrease in arterial blood pressure of 20 mm Hg or with an acute decrease of hemoglobin of 2 mg/ dl ph measurements. Treatment was considered successful when gastric ph was 4 during 80% of the study time on each patient. Gastric ph evolution, UGI hemorrhage occurrence rate, microscopic upper gastrointestinal hemorrhage, mortality, adverse events. Pediatr Crit Care Med 2010 Vol. 11, No. 1 127

Table 1. Continued Study Methods Participants Interventions Outcomes Osteyee et al (29) cross over trial Sixteen critically ill children. 160 patients who needed mechanical ventilation on admission were enrolled in the study. Patients were excluded if any of the following circumstances occurred in the first 48 hrs after inclusion: extubation, death, pneumonia, or new information that the patient had received one of the study drugs in the last 48 hrs before admission. Children in group 1 received bolus dosing on day 1 and continuous infusion of ranitidine on day 2. Group 2 received the continuous infusion on day 1 and bolus dosing on day 2. Continuous infusion regimen: ranitidine bolus of 0.15 mg/kg followed by continuous infusion at 0.15 mg/kg per hour for 12 hrs. Bolus regimen: 1 mg/kg, two doses 6 hrs apart. Group S received sucralfate suspension 60 mg/kg/day in four doses via the nasogastric tube that was flushed with 10 ml of sterile water; group R received ranitidine 2 mg/kg/day intravenously in four doses; group O received omeprazole 1 mg/kg/day intravenously in two doses; and group P did not receive any medication for stress ulcer prophylaxis. Gastric ph. Ventilator-associated pneumonia was defined as the occurrence of a new or persistent radiographic infiltrate in conjunction with one of the following: positive pleural/blood culture with the same organism recovered in the tracheal aspirate or sputum, radiographic cavitation, histopathologic evidence of pneumonia; or at least two of the following: fever; leukocytosis; and purulent tracheal aspirate or sputum. Pneumonia was considered to be ventilator associated if it occurred after a minimum of 48 hrs after the initiation of mechanical ventilation. Respiratory tract culture specimens were obtained from tracheal aspirates. Hospital mortality was defined as patient death occurring in the PICU and hospital stay was defined as the days in the PICU. UGI, upper gastrointestinal; H2 antagonists, histamine H2-receptor antagonists; ICU, intensive care unit; PICU, pediatric intensive care unit; SD, standard deviation. chanical ventilation and pediatric ICU stay. Omeprazole In one RCT (28), omeprazole was not superior to no treatment in groups in rates of macroscopic bleeding, pneumonia, deaths, and in the duration of mechanical ventilation and pediatric ICU stay. Cimetidine In one RCT (26), cimetidine was not superior to placebo in rates of UGI bleeding. Drug Vs. Any Other Active Compound We found seven studies comparing diverse drugs, doses, and regimen (27 32). Most studies did not report outcomes, such as mortality, rates of UGI bleeding, and pneumonia. Pirenzepine Vs. Famotidine No significant differences were found between groups in one RCT of children who underwent corrective or palliative surgery for congenital heart disease concerning rates of pneumonia and organism cultured from the stomach and from tracheal secretion (32). Ranitidine Vs. Sucralfate One RCT (28) showed that ranitidine had a significant effect in the rate of patients gastric aspirates with mean ph of 4 during 50% compared with sucralfate (one RCT 70 participants; RR, 2.60; 95% CI, 1.49 4.55). However, no significant differences were found in the rates of patients with important UGI hemorrhage (28) and macroscopic bleeding, death and pneumonia, and in the duration of mechanical ventilation and pediatric ICU stay (29). Ranitidine Vs. Amalgate Ranitidine was significantly superior to amalgate in the rate of patients gastric aspirates with mean ph of 4 during 50% of study time (one RCT 70 participants; RR, 0.79; 95% CI, 0.64 0.97). No significant differences were found between groups in the rates of patients with important UGI hemorrhage and gastrointestinal symptoms, such as nausea, vomiting, or diarrhea (28). Amalgate Vs. Sucralfate One RCT (28) showed that amalgate had a significant effect in the rate of patients gastric aspirates with mean ph of 4 during 50% compared with sucralfate (one RCT 70 participants; RR, 3.30; 95% CI, 1.94 5.61). No significant differences were found between groups in the rates of patients with important UGI hemorrhage and gastrointestinal symptoms, such as nausea, vomiting, or diarrhea (28). Ranitidine Vs. Famotidine No significant differences were found between groups in one RCT (31) of critically ill pediatric patients in the rate of patients gastric aspirates with mean ph of 4 during 80% of study time. Other Comparisons No significant differences were found between groups in one RCT (30) of pa- 128 Pediatr Crit Care Med 2010 Vol. 11, No. 1

Table 2. Assessment of the risk of bias in randomized s included in the review a (9) Study Sequence Generation Allocation Concealment Blinding of Participants, Personnel, and Outcome Assessors Incomplete Outcome Data/ Withdrawals Free of Selective Reporting? Other Sources of Bias/Commentaries Overall Risk Aanpreung et al (31) Behrens et al (32) Kuusela et al (33) Lacroix et al (26) Lopez-Herce et al (27) Lopez-Herce et al (28) Osteyee et al (29) Yildizdas et al (30) Unclear Unclear No Yes No Yes. Description of baseline characteristics. Unclear risk of bias Unclear Unclear No Yes Unclear Yes. The clinical course of Unclear risk of bias patients in both groups was similar. Similar baseline characteristics. Unclear Unclear Unclear Yes Unclear Yes Unclear risk of bias Yes Unclear Unclear Yes Unclear Yes Unclear risk of bias Unclear Unclear No Yes No No description of baseline High risk of bias characteristics. Unclear Unclear No Unclear No Yes Unclear risk of bias Unclear Unclear No No No No description of baseline characteristics. Yes Unclear No Unclear Unclear Yes. Baseline characteristics and primary diseases in the patients were reported for both groups. High risk of bias Unclear risk of bias Yes, low risk of bias; No, high risk of bias. a According to the Cochrane Collaboration Handbook. Sequence generation: Was the allocation sequence adequately generated? Allocation concealment: Was allocation adequately concealed? Blinding of participants, personnel, and outcome assessors: Was knowledge of the allocated intervention adequately prevented during the study? Incomplete outcome data/withdrawals: Were intention to treat analyses performed? Had participants withdrawn from the study? Free of selective reporting? Other sources of bias: Was sample size calculated? Were inclusion/exclusion criteria and baseline characteristics defined? Were conflicts of interests reported? Figure 2. Meta-analysis of prophylaxis for preventing upper gastrointestinal bleeding (macroscopic or important bleeding) compared with no treatment. M-H, Mantel-Haenszel; CI, confidence interval; df, degrees of freedom. tients who needed mechanical ventilation on admission concerning rates of macroscopic bleeding, deaths, pneumonia, and the duration of mechanical ventilation and pediatric ICU stay when comparing omeprazole vs. ranitidine and omeprazol vs. sucralfate. Ranitidine Versus Ranitidine at Different Doses and Regimen One RCT (27) showed that intravenous ranitidine at 2 mg/kg and 4 mg/kg had a significant effect in the rate of patients gastric aspirates with mean ph of 4 during 80% of study time compared with ranitidine by nasogastric tube (one RCT 20 participants; RR, 0.25; 95% CI, 0.07 0.90 and RR, 0.25; 95% CI, 0.07 0.90, respectively). Another RCT did not find significant difference in patients gastric aspirates ph when comparing bolus dosing and continuous infusion dosing of 4 mg/kg per day of intravenous ranitidine (29). DISCUSSION The RCTs included in this review have assessed a broad range of treatments that resulted in limited opportunities to describe and pool useful data. Studies available for analysis are a highly heterogeneous group, with different drugs being used and different methods for assessing their efficacy (e.g., some used endoscopy on all patients, others simply monitored nasogastric output for bleeding). Furthermore, because the majority of RCTs had an unclear risk of bias, small sample size, and did not reported relevant outcomes, it was difficult to conclude whether one treatment was more beneficial than the comparator most of the time. Most RCTs focused on secondary outcomes, such as gastric ph control. In addition, methods used for diagnosis of UGI bleeding greatly varied across studies and we could not integrate data for most comparisons. Pooled data of two studies suggested that pediatric patients may benefit from receiving prophylactic treatment for preventing UGI bleeding. There was reasonable evidence that ranitidine is better than no treatment in mechanically ven- Pediatr Crit Care Med 2010 Vol. 11, No. 1 129

Table 3. Summary of relevant findings for primary outcomes of randomized s included in the review a Study Outcome Comparison (n) Relative Risk (95% CI) p Value/Heterogeneity Lopez-Herce et al (28) Lacroix et al (26) Lopez-Herce et al (28) Lopez-Herce et al (28) UGI bleeding (macroscopic or important bleeding) UGI bleeding (macroscopic or important bleeding) UGI bleeding (macroscopic or important bleeding) Treatment vs. (143) vs. no treatment (77) Treatment (162) vs. no treatment or placebo (98) Sucralfate (73) vs. no treatment (77) Kuusela et al (33) Deaths Treatment (141) vs. no treatment (67) Pneumonia Treatment (118) vs. no treatment (42) Kuusela et al (33) Kuusela et al (33) Rate of patients with normal visual endoscopic findings Rate of patients with normal histological endoscopic findings Ranitidine (23) vs. no treatment (25) Ranitidine (23) vs. no treatment (25) 0.38 (0.16 0.91) a.03; heterogeneity p.35; I 2 33% 0.66 (0.23 1.85) b ns; heterogeneity p.07; I 2 63% 0.31 (0.09 1.09) a ns; heterogeneity p.35; I 2 33% 1.24 (0.68 2.28) a ns 1.11 (0.73 1.68) ns 3.04 (1.30 7.12).011 3.53 (1.34 9.29).01 CI, confidence interval; UGI, upper gastrointestinal; ns, nonsignificant. a Relative risk and 95% CI for dichotomous primary outcomes were calculated by the Mantel-Haenszel fixed-effects model; b relative risk and 95% CI for dichotomous primary outcomes were calculated by the Mantel-Haenszel random-effects model. Table 4. Risk factors significantly associated with gastrointestinal bleeding in severely ill pediatric patients Risk Factor Participants OR or RR (95% CI) Reference Mechanically ventilated NB in PICU OR 4.06 (1.21 12.39) Kuussela et al (38) 1 15 yrs in PICU OR 5.13 (1.86 14.12) Nithiwathanapong et al (40) PRISM score 10 3 18 yrs in PICU OR 13.4 (3.7 47.9) Chaibou et al (1) Children in PICU RR 2.87 (1.55 5.32) Lacroix et al (39) Respiratory failure 3 18 yrs in PICU OR 10.2 (1.3 82.8) Chaibou et al (1) Coagulopathy 3 18 yrs in PICU OR 9.3 (2.8 30.3) Chaibou et al (1) Children in PICU RR 5.37 (2.03 14.16) Lacroix et al (39) Thrombocytopenia 1 15 yrs in PICU OR 2.26 (1.07 4.74) Nithiwathanapong et al (40) Shock NB 19 yrs in PICU OR 17.39 (3.47 87.22) Cochran et al (34) Surgery time 3 hrs NB 19 yrs in PICU OR 3.57 (1.43 8.90) Cochran et al (34) Trauma NB 19 yrs in Children OR 20.92 (1.85 24) Cochran et al (34) in PICU RR 2.46 (1.14 15.33) Lacroix et al (39) Pneumonia Children in PICU RR 3.47 (1.21 9.9) Lacroix et al (39) Enteral feeding Children in PICU RR 4.16 (2.27 7.66) Lacroix et al (39) Organ failure Mechanically ventilated RR 2.85 (1.18 6.92) Deerojanawong et al (35) Corticoid administration NB RR 1.90 (1.35 2.66) Halliday et al (36) RR 1.74 (1.02 2.98) Halliday et al (37) OR, odds ratio, RR, relative risk; CI, confidence interval; NB, newborn; PICU, pediatric intensive care unit; PRISM, Pediatric Risk of Mortality. tilated preterm and full-term newborns treated in a neonatal ICU in improving rates of normal gastric mucosal endoscopic findings by visual inspection and histologic specimens (33). In addition, ranitidine, sucralfate, and amalgate were also better than no treatment in improving rates of patients gastric aspirates with mean ph of 4 during 50% of study time (28). However, we found no evidence to support that prophylaxis medication is better than no treatment to decrease the rates of ulcers or erosion or deaths. Furthermore, no evidence was found to support that prophylaxis decreases the duration of mechanical ventilation or pediatric ICU stay. On the other hand, we did not find significant increase in the rates of pneumonia or adverse event. Finally, intravenous ranitidine was superior to ranitidine by nasogastric tube in improving rates of patients with mean ph of 4 during 80% of study time. A meta-analysis evaluating the effect of stress ulcer prophylaxis on gastrointestinal bleeding, pneumonia, and mortality in critically ill adult patients demonstrates that prophylaxis with histamine2- receptor antagonists decreases the occurrence of overt gastrointestinal bleeding (OR, 0.58; 95% CI, 0.42 0.79) and clinically important bleeding (OR, 0.44; 95% CI, 0.22 0.88). Authors also reported a trend toward an increased risk of pneumonia associated with histamine2-receptor antagonists as compared with no prophylaxis (OR, 1.25; 95% CI, 0.78 2.00). However, sucralfate was associated with a lower prevalence of nosocomial pneumonia and reduced mortality rate when compared with antacids and histamine2- receptor antagonists (5). An additional RCT found that ranitidine had a significantly lower rate of clinically important UGI bleeding than sucralfate in 1200 critically ill adult patients requiring mechanical ventilation. Authors also reported no significant differences in the rates of ventilator-associated pneumonia, the duration of the stay in the ICU, or mortality (6). A number of risk factors associated with stress ulcers, gastritis, and gastrointestinal bleeding in severelly ill pediatric patients have been described in observational studies (Table 4) (1, 34 40). A cohort study found that respiratory failure, 130 Pediatr Crit Care Med 2010 Vol. 11, No. 1

coagulopathy, and a Pediatric Risk of Mortality Score of 10 were independent risk factors for clinically significant upper GI bleeding, using multivariate analysis. Based on those findings, authors recommended that prophylaxis to prevent UGI bleeding may be limited to patients who present with at least two risk factors (1). We have produced an updated coverage of RCTs of prophylactic treatments for stress ulcers in children by summarizing the best available data. Although pooled data of two studies suggested that critically ill pediatric patients may benefit from receiving prophylactic treatment for preventing UGI bleeding, the overall quality of the evidence is low, leading to a weak recommendation (using GRADE approach) (10). Although limited evidence is available, some of the drugs studied (histamine2 receptor antagonists, sucralfate, amalgate) have been replaced in clinical use by proton pump inhibitors. However, only one study including patients treated with omeprazole was found (30). We need more evidence demonstrating the effectiveness and safety of different prophylactic drugs and improved design and reporting of RCTs. We should also investigate the use of proton pump inhibitors in children. ACKNOWLEDGMENT We thank Dr. Rodrigo Pardo of the Research Institute of the National University of Colombia for instructive commentaries to improve this manuscript. REFERENCES 1. Chaïbou M, Tucci M, Dugas MA, et al: Clinically significant upper gastrointestinal bleeding acquired in a pediatric intensive care unit: A prospective study. Pediatrics 1998; 102:933 938 2. Cook DJ, Griffith LE, Walter SD, et al: The attributable mortality and length of intensive care unit stay of clinically important gastrointestinal bleeding in critically ill patients. Crit Care 2001; 5:368 375 3. Erstad BL, Barletta JF, Jacobi J, et al: Survey of stress ulcer prophylaxis. Crit Care 1999; 3:145 149 4. Cook DJ, Fuller HD, Guyatt GH, et al: Risk factors for gastrointestinal bleeding in critically ill patients. NEnglJMed1994; 330:377 381 5. Cook DJ, Reeve BK, Guyatt GH, et al: Stress ulcer prophylaxis in critically ill patients: Resolving discordant meta-analyses. JAMA 1996; 275:308 314 6. Cook DJ, Guyatt GH, Marshall J, et al: A comparison of sucralfate and ranitidine for prevention of upper gastrointestinal bleeding in patients requiring mechanical ventilation. N Engl J Med 1998; 338:791 797 7. Messori A, Trippoli S, Vaiani M, et al: Bleeding and pneumonia in intensive care patients given ranitidine and sucralfate for prevention of stress ulcer: Meta-analysis of randomised s. BMJ 2000; 321:1103 1106 8. Spirt MJ, Stanley S: Update on stress ulcer prophylaxis in critically ill patients. Crit Care Nurse 2006; 26:18 28 9. Higgins JPT, Green S (Eds). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0 [Updated February 2008]. The Cochrane Collaboration, 2008. Available at: http://www.cochrane-handbook.org. Accessed November 23, 2009 10. Guyatt GH, Oxman AD, Vist G, et al: Rating quality of evidence and strength of recommendations GRADE: An emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008; 336:924 926 11. Golder S, McIntosh HM, Duffy S, et al: Developing efficient search strategies to identify reports of adverse effects in MEDLINE and EMBASE. Health Info Libr J 2006; 23:3 12 12. Agarwal AK, Saili A, Pandey KK, et al: Role of cimetidine in prevention and treatment of stress induced gastric bleeding in neonates. Indian Pediatr 1990; 27:465 469 13. Bonten MJ, Gaillard CA, van Tiel FH, et al: Continuous enteral feeding counteracts preventive measures for gastric colonization in intensive care unit patients. Crit Care Med 1994; 22:939 944 14. Cothran DS, Borowitz SM, Sutphen JL, et al: Alteration of normal gastric flora in neonates receiving ranitidine. J Perinatol 1997; 17: 383 388 15. Eddleston JM, Booker PD, Green JR: Use of ranitidine in children undergoing cardiopulmonary bypass. Crit Care Med 1989; 17:26 29 16. Gedeit RG, Weigle CG, Havens PL, et al: Control and variability of gastric ph in critically ill children. Crit Care Med 1993; 21:1850 1855 17. Haizlip JA, Lugo RA, Cash JJ, et al: Failure of nasogastric omeprazole suspension in pediatric intensive care patients. Pediatr Crit Care Med 2005; 6:182 187 18. Harrison AM, Lugo RA, Vernon DD: Gastric ph control in critically ill children receiving intravenous ranitidine. Crit Care Med 1998; 26:1433 1436 19. Kaufman SS, Lyden ER, Brown CR, et al: Omeprazole therapy in pediatric patients after liver and intestinal transplantation. JPediatrGastroenterol Nutr 2002; 34:194 198 20. Kentrup H, Skopnik H, Wolter L, et al: Antacids for postoperative prevention of stress ulcer in infants: A dose finding study. Klin Padiatr 1996; 208:14 16 21. Kuusela AL: Long-term gastric ph monitoring for determining optimal dose of ranitidine for critically ill preterm and term neonates. Arch Dis Child Fetal Neonatal Ed 1998; 78:151 153 22. Lloyd CW, Martin WJ, Taylor BD, et al: Pharmacokinetics and pharmacodynamics of cimetidine and metabolites in critically ill children. J Pediatr 1985; 107:295 300 23. Lugo RA, Harrison AM, Cash J, et al: Pharmacokinetics and pharmacodynamics of ranitidine in critically ill children. Crit Care Med 2001; 29:759 764 24. Wiest DB, O Neal W, Reigart JR, et al: Pharmacokinetics of ranitidine in critically ill infants. Dev Pharmacol Ther 1989; 12:7 12 25. Hencz P, Baltás B, Tekulics P: Cimetidine therapy of stress ulcer caused by lifethreatening illnesses in children. Orv Hetil 1984; 125:2005 2007 26. Lacroix J, Infante-Rivard C, Gauthier M, et al: Upper gastrointestinal tract bleeding acquired in a pediatric intensive care unit: Prophylaxis trial with cimetidine. J Pediatr 1986; 108:1015 1018 27. Lopez-Herce Cid J, Albajara Velasco L, et al: Ranitidine prophylaxis in acute gastric mucosal damage in critically ill pediatric patients. Crit Care Med 1988; 16:591 593 28. López-Herce J, Dorao P, Elola P, et al: Frequency and prophylaxis of upper gastrointestinal hemorrhage in critically ill children: A prospective study comparing the efficacy of almagate, ranitidine, and sucralfate. The Gastrointestinal Hemorrhage Study Group. Crit Care Med 1992; 20:1082 1089 29. Osteyee JL, Banner W Jr: Effects of two dosing regimens of intravenous ranitidine on gastric ph in critically ill children. Am J Crit Care 1994; 3:267 272 30. Yildizdas D, Yapicioglu H, Yilmaz HL: Occurrence of ventilator-associated pneumonia in mechanically ventilated pediatric intensive care patients during stress ulcer prophylaxis with sucralfate, ranitidine, and omeprazole. J Crit Care 2002; 17:240 245 31. Aanpreung P, Vanprapar N, Susiva C, et al: A randomized clinical trial comparing the efficacy of ranitidine and famotidine on intragastric acidity in critically ill pediatric patients. J Med Assoc Thai 1998; 81:185 189 32. Behrens R, Hofbeck M, Singer H, et al: Frequency of stress lesions of the upper gastrointestinal tract in paediatric patients after cardiac surgery: Effects of prophylaxis. Br Heart J 1994; 72:186 189 33. Kuusela AL, Ruuska T, Karikoski R, et al: A randomized, controlled study of prophylactic ranitidine in preventing stress-induced gastric mucosal lesions in neonatal intensive care unit patients. Crit Care Med 1997; 25:346 351 34. Cochran EB, Phelps SJ, Tolley EA, et al: Prevalence of, and risk factors for, upper gastrointestinal tract bleeding in critically ill pediatric patients. Crit Care Med 1992; 20:1519 1523 35. Deerojanawong J, Peongsujarit D, Vivatvakin B, et al: Incidence and risk factors of upper gastrointestinal bleeding in mechanically ventilated children. Pediatr Crit Care Med 2009; 10:91 95 36. Halliday HL, Ehrenkranz RA, Doyle LW: Early postnatal ( 96 hours) corticosteroids for preventing chronic lung disease in preterm infants. Cochrane Database Syst Rev 2003; 1:CD001146 37. Halliday HL, Ehrenkranz RA, Doyle LW: Moderately early (7 14 days) postnatal corti- Pediatr Crit Care Med 2010 Vol. 11, No. 1 131

costeroids for preventing chronic lung disease in preterm infants. Cochrane Database Syst Rev 2003; 1:CD001144 38. Kuusela AL, Mäki M, Ruuska T, et al: Stressinduced gastric findings in critically ill newborn infants: Frequency and risk factors. Intensive Care Med 2000; 26:1501 1506 39. Lacroix J, Nadeau D, Laberge S, et al: Frequency of upper gastrointestinal bleeding in a pediatric intensive care unit. Crit Care Med 1992; 20:35 42 40. Nithiwathanapong C, Reungrongrat S, Ukarapol N: Prevalence and risk factors of stress-induced gastrointestinal bleeding in critically ill children. World J Gastroenterol 2005; 11:6839 6842 APPENDIX A Search Strategy for PubMed (Randomized [pt] OR controlled clinical trial [pt] OR randomized controlled trials [mh] OR random allocation [mh] OR double-blind method [mh] OR single-blind method [mh] OR clinical trial [pt] OR clinical trials [mh] OR ( clinical trial [tw]) OR ((singl* [tw] OR doulb* [tw] OR tripl* OR trebl* [tw]) AND (mask* [tw] OR blind* [tw])) OR ( latin square [tw]) OR placebos [mh] OR placebo* [tw] OR random* [tw] OR research design [mh:noexp] OR evaluation studies [pt] OR follow-up studies [mh] OR prospective studies [mh] OR crossover studies [mh] OR control* [tw] OR prospectiv* [tw] OR volunteer* [tw]) NOT (animal [mh] NOT human [mh]) AND ( Peptic Ulcer Perforation [Mesh] OR Peptic Ulcer [Mesh] OR Duodenal Ulcer [Mesh] OR Stomach Ulcer [Mesh] OR Peptic Ulcer Hemorrhage [Mesh] OR Curling s ulcer [tw] OR ((ulcer* [tw] OR bleeding [tw] OR hemorrhage [tw] OR hemorrhage [tw] OR ulceration [tw]) AND (Peptic [tw] OR Duodenal [tw] OR Stomach [tw] OR gastric [tw] OR gastrointestinal [tw] OR gastro-intestinal [tw]))) AND (stress [tw] OR critically ill [tw] OR intensive care [tw] OR serious burns [tw] OR mechanical ventilation [tw] OR ventilator[tw] OR Critical Illness [mh] OR Ventilators, Mechanical [tw] OR severely injured [tw] OR PICU [tw] OR ICU [tw]) AND ( Anti-Ulcer Agents [Mesh] OR prophylaxis [tw] OR prevention [tw] OR prevent [tw] OR Antacid* [tw] OR Histamine H2 Antagonists [tw] OR sucralfate [tw] OR Proton-pump inhibitors [tw] OR cimetidine [tw] OR omeprazole [tw] OR Esomeprazole [tw] OR ranitidine [tw] OR Cisapride [tw] OR Famotidine [tw] OR lansoprazole [tw] OR pantoprazole [tw] OR rabeprazole [tw] OR Acid suppression [tw]) AND (Child [mh] OR children [tw] OR pediatric* [tw] OR pediatric* [tw] OR Preschool OR Infant [mh] OR infant* [tw] OR newborn [tw] OR perinatal [tw] OR neonatal [tw] OR preterm [tw] OR postnatal [tw] OR premature [tw] OR childhood [tw] OR neonate* [tw]). 132 Pediatr Crit Care Med 2010 Vol. 11, No. 1