Received: 10 April 2001Accepted: 20 March 2002Published online: 20 April 2002© Springer-Verlag 2002

Abstract Objective: To assess therisk of ventilator-associated pneumo-nia (VAP) and the incidence of upperairway colonisation related to the useof stress ulcer prophylaxis in critical-ly ill children. Design: Retrospectivestudy. Setting: Paediatric intensivecare unit (PICU) of a tertiary carecentre. Patients: All children whowere mechanically ventilated formore than 48 h. Interventions: None.Results: A total of 54 patients weregiven ranitidine, 53 patients weregiven sucralfate and 48 patients weregiven no stress ulcer prophylaxis.Thirteen (8.4%) patients developedVAP: 6 (11.1%) patients in the raniti-dine group, 4 (7.5%) in the sucral-fate group and 3 (6.2%) in the groupwithout prophylaxis. The rate of up-per airway colonisation with Gram

negative bacilli was 25.9% (14/54)in the ranitidine group, 22.6%(12/53) in the sucralfate group and37.5% (18/48) in the group withoutprophylaxis. The differences amongthe groups were not significant. Conclusions: In contrast to findingsin adults, we found that, in children,sucralfate does not decrease the inci-dence of VAP or the incidence of up-per airway colonisation with Gramnegative bacilli as compared to ra-nitidine or no stress ulcer prophyl-axis. However, the small sample sizeand study design substantially limitour conclusions.

Keywords Children · Ventilator-associated pneumonia ·Upper airway colonisation · Ranitidine · Sucralfate

Intensive Care Med (2002) 28:763–767DOI 10.1007/s00134-002-1289-3 N E O N ATA L A N D P E D I AT R I C I N T E N S I V E C A R E

Enrico LoprioreDick G. MarkhorstReinoud J. B. J. Gemke

Ventilator-associated pneumonia and upper airway colonisation with Gram negative bacilli: the role of stress ulcer prophylaxis in children

Introduction

Critically ill patients requiring mechanical ventilation areat increased risk for upper gastrointestinal bleeding fromstress ulcers. Prophylaxis against stress ulcers in intensivecare units (ICU) is, therefore, widely used. Various studieshave demonstrated that the use of agents that raise gastricpH, such as ranitidine, are associated with increased con-centrations of Gram negative bacilli in gastric aspirates [1,2, 3, 4]. Subsequent retrograde colonisation from the stom-ach to the pharynx and trachea, as well as regurgitation ofgastric contents, may then predispose patients to ventilator-associated pneumonia (VAP) caused by Gram negative ba-cilli. Three recent meta-analyses have shown that the useof ranitidine in critically ill adults is associated with a high-er incidence of VAP compared to sucralfate, which has lit-

tle or no effect on gastric pH [5, 6, 7]. So far no studiescomparing the incidence of VAP and upper airway coloni-sation in critically ill children receiving ranitidine or su-cralfate have been published. We hypothesised that the in-cidence of VAP and upper airway colonisation with Gramnegative bacilli in children is lower in patients receiving sucralfate, as compared to ranitidine.

Patients and methods

From 1998 to 2000 the hospital charts of all children who wereadmitted to a tertiary paediatric intensive care unit (PICU) andwho were mechanically ventilated, were evaluated for the pres-ence and bacterial cause of nosocomial pneumonia as well as up-per airway colonisation during mechanical ventilation. Patientswere only included in the study if they had been mechanically

E. Lopriore · D.G. Markhorst R.J.B.J. Gemke (✉ )Department of Paediatric Intensive Care,VU University Medical Centre, De Boelelaan 1117, 1007 MB Amsterdam,PO Box 7057, The Netherlandse-mail: rjbj.gemke@vumc.nlTel.: +31-20-4443000Fax: +31-20-4442419

ventilated for more than 48 h and had a nasogastric tube in place.Patients were analysed until 24 h after they were extubated or until24 h after stress ulcer prophylaxis was discontinued. Patients wereexcluded if they had pneumonia or ARDS on admission or within24 h of admission; had received histamine2 blockers, sucralfate oromeprazole within 1 week before admission; were treated withboth sucralfate and ranitidine; were treated with omeprazole orhad acute renal failure and therefore could not be treated with su-cralfate due to its potential renal side effects. Two main outcomemeasures were assessed: VAP and upper airway colonisation. Weused the following criteria for VAP, according to the Center forDisease Control and Prevention (CDC) definition of nosocomialpneumonia [8]: a chest film showing a new and persistent infil-trate that was consistent with pneumonia and at least two of thefollowing findings: (1) a temperature above 38.5°C or below35.0°C; (2) a leukocyte count of more than 10,000/mm3 or lessthan 3000/ml3; (3) isolation of pathogenic bacteria from a trachealaspirate or (4) purulent sputum.

Patients were considered to have developed a second pneumo-nia if 14 days had elapsed since the clearing of the first pneumo-nia. Early-onset pneumonia was defined according to the criteriaof Langer et al. as occurring within the first 4 days after the initia-tion of mechanical ventilation [9]. Similarly, late-onset pneumoniawas defined as occurring after the first 4 days of mechanical venti-lation. Colonisation was defined as the isolation from the trachealaspirate of more than 103 colony-forming units per millilitre of anorganism not previously found there, on two or more consecutiveoccasions. Samplings of endotracheal aspirates in mechanicallyventilated patients in our PICU are performed once weekly. Addi-tional samplings for bacterial culture are performed if clinicallyindicated. The clinical data abstracted from the records of the pa-tients included basic demographics, primary diagnosis (the diag-nosis at discharge from the PICU that was considered as the un-derlying cause of all problems presented by the patient during thePICU stay), paediatric index of mortality (PIM) score, used for se-verity of illness, and mortality. Mortality was defined as death inthe PICU or within 48 h of extubation.

We assessed two known risk factors for VAP: duration of intu-bation [10, 11] and administration of enteral feeding through naso-gastric tube [12]. Similarly, we assessed one known protective fac-tor against VAP: treatment with systemic antibiotics [13]. Further-more, we assessed the possible effect of treatment with prokinetics(cisapride or erythromycin) on the incidence of VAP. Administra-tion of enteral feeding, treatment with systemic antibiotics andtreatment with prokinetics were considered as positive whenfound during more than 75% of the duration of the study.

The attending physicians in our PICU prescribed prophylaxisstress ulcer prophylaxis according to three treatment regimens: (1) ranitidine administered i.v. in a dose of 2.5–6mg/kg bodyweight per day b.i.d.; (2) sucralfate administered orally or (3) nostress ulcer prophylaxis. Of the children included in the sucralfategroup, those children of 0–1 year of age received sucralfate at arate of 50 mg q.i.d., children of 1–2 years of age received 100 mgq.i.d., children 2–6 years of age received 250 mg q.i.d., children6–12 years received 500 mg q.i.d. and children over 12 years ofage received 1000 mg q.i.d..

Statistical analysis of the results was performed using SPSSversion 9. Chi-square test, Mann-Whitney U test and Kruskall Wal-lis test were used to evaluate the results. Logistic regression wasused where applicable. Significance was set at p less than 0.05.

Results

Patient population

During the study period, 214 consecutive patients admit-ted in our PICU were mechanically ventilated for more

than 48 h. Of these 214 eligible patients, 59 patientswere excluded according to predefined criteria, leaving atotal of 155 patients included in the study. Reasons forexclusion were: (1) pneumonia or ARDS within 24 h ofadmission (n=41), (2) treatment with both sucralfate andranitidine (n=8), (3) treatment with omeprazole (n=2) or(4) acute renal failure (n=7). Fifty-four patients weregiven ranitidine, 53 patients were given sucralfate and 48 patients received no stress ulcer prophylaxis. Thebaseline characteristics of the three groups are presentedin Table 1. Significant differences were found in age,sex, enteral feeding, duration of intubation and respirato-ry disease as primary diagnosis. These differences weredue mainly to different characteristics in the group of pa-tients receiving no stress ulcer prophylaxis. No signifi-cant differences in baseline characteristics were foundbetween the ranitidine group and the sucralfate group.No significant difference was found among the threegroups in terms of mortality, PIM score, primary diagno-sis other than respiratory disease, number of patients re-ceiving systemic antibiotics and number of patients re-ceiving cisapride or erythromycin.

Ventilator-associated pneumonia

Overall, 13 (8.4%) patients developed VAP: 6 (11.1%)patients in the ranitidine group, 4 (7.5%) in the sucralfategroup and 3 (6.2%) in the group without stress ulcer pro-phylaxis (Table 2). None of these patients developed asecond pneumonia. There was no significant differencein the development of VAP among the groups, nor be-tween the ranitidine group and the sucralfate group stud-ied separately (χ2=0.40, p=0.52, d.f.:1). The mediannumber of days until the development of VAP after initi-ation of mechanical ventilation was 5 (range: 2–44): 5.5 (range: 2–44) in the ranitidine group, 5 (range: 2–15)in the sucralfate group and 3 (range: 2–9) in the groupwithout prophylaxis. Gram negative bacilli were presentin 8 of 11 (72.7%) endotracheal aspirates of the patientswith VAP. Gram negative bacilli were isolated in 83.3%(5/6) of VAP cases in the ranitidine group, 50% (2/4) ofVAP cases in the sucralfate group and 33.3% (1/3) ofVAP cases in the group without prophylaxis. The differ-ence in rate of Gram negative bacilli isolated from pa-tients with pneumonia among the groups was not signifi-cant. When comparing only the ranitidine and the sucral-fate groups with each other, again no significant differ-ence in VAP was found (χ2=0.4, p=0.52, d.f.:1). In oneof 13 patients mortality was attributable to VAP ( 7.7%);this patient was included in the ranitidine group.

Early- and late-onset pneumonia

Overall, early-onset pneumonia occurred in 46.2% (6/13)of the cases. Three of five isolated bacteria in early-onset

764

pneumonia were Gram positive cocci (60%). Whereas inlate-onset pneumonia, only one of the eight isolated bac-teria was Gram positive cocci (12.5%). This differencetended to statistical significance (χ2=3.76, p=0.052,d.f.:1). Similarly, Gram negative bacilli were found inonly 20% (1/5) of the early-onset pneumonia. In con-trast, Gram negative bacilli were found in 87.5% (7/8) ofthe late-onset pneumonia. This difference was significant(χ2=6.64, p=0.01, d.f.:1).

Risk factors for ventilator-associated pneumonia

The duration of intubation in the group patients withVAP was significantly longer compared to the group ofpatients without VAP, and significantly more patientswith, than without, VAP, were treated with cisapride orerythromycin (Table 3). We found no difference in theuse of systemic antibiotics or enteral feeding betweenthe groups of patients with or without VAP. The number

765

Table 1 Baseline characteris-tics of the 155 study patients Ranitidine Sucralfate No prophylaxis p value

(n=54) (n=53) (n=48)

Agea (months, range) 4.5 (0–203) 22 (0–208) 2 (0–105) <0.05<1 month (n (%)) 19 (35.2) 12 (22.6) 22 (45.8) <0.051–12 months (n (%)) 12 (22.2) 13 (24.5) 12 (25) NS1–<4 years (n (%)) 7 (13) 10 (18.9) 7 (14.6) NS4–<12 years (n (%)) 11 (20.4) 13 (24.5) 5 (10.4) NS>12 years (n (%)) 5 (9.3) 6 (11.3) 1 (2.1) NS

Male sex (n (%)) 33 (61) 38 (72) 22 (46) <0.05PIM scorea (range) 6.3 (0.9–28.5) 5.4 (0.9–28.1) 6.7 (0.2–62.2) NSMortality (n (%)) 3 (5.5) 3 (5.6) 1 (2.1) NS

Primary diagnosis (n (%))Respiratory disease 11 (20.4) 13 (24.5) 26 (54.1) <0.001Abdominal surgery 9 (16.6) 4 (7.5) 4 (8.3) NSNeurological disease 5 (9.3) 5 (9.4) 4 (8.3) NSCentral nervous system surgery 5 (9.3) 7 (13.2) 2 (4.2) NSSepticaemia 9 (16.7) 10 (18.9) 2 (4.2) NSPolytrauma 3 (5.6) 4 (7.5) 0 (0) NSOther 12 (22.2) 10 (18.9) 10 (20.8) NS

Risk factorsDuration of intubationa (days, range) 6 (3–91) 5 (3–58) 11 (4–123) <0.001Enteral feeding (n (%)) 12 (22.2) 14 (26.4) 43 (89.6) <0.001Antibiotics (n (%)) 35 (64.8) 31 (58.5) 21 (43.8) NSCisapride (n (%)) 10 (18.5) 8 (15.1) 6 (12.5) NSErythromycin (n (%)) 0 (0) 2 (3.8) 4 (8.3) NS

a median

Table 2 Incidence of ventila-tor-associated pneumonia(VAP) and upper airway coloni-sation

Ranitidine Sucralfate No prophylaxis p valuegroup group group(n=54) (n=53) (n=48)

Patients with VAP (n (%)) 6 (11.1) 4 (7.5) 3 (6.2) NSIsolated Gram negative bacilli (n (%)) 5 (83.3) 2 (50) 1 (33.3) NSIsolated Gram positive cocci (n (%)) 2 (33.3) 1 (25) 1 (33.3) NS

Patients with upper airway colonisation (n (%)) 17 (31.5) 13 (24.5) 20 (41.7) NSIsolated Gram negative bacilli (n (%)) 14 (25.9) 12 (22.6) 18 (37.5) NSIsolated Gram positive cocci (n (%)) 2 (3.7) 3 (5.6) 7 (14.6) NS

Table 3 Risk factors for venti-lator-associated pneumonia(VAP)

Patients with VAP Patients without VAP p value(n=13) (n=142)

Duration of intubationa (days, range) 13 (6–91) 6 (3–123) <0.05Enteral feeding (n (%)) 7 (53.8) 62 (43.6) NSAntibiotics (n (%)) 7 (53.8) 80 (56.3) NSCisapride (n (%)) 5 (38.4) 19 (13.4) <0.05Erythromycin (n (%)) 2 (15.4) 4 (2.8) <0.05

a median

of days on mechanical ventilation was found to be theonly significant risk factor for the development of VAP(p=0.04) when using a logistic regression model to ad-just for differences in variables among the study groups.Neonatal age was almost a significant risk factor for pre-dicting VAP when using the same model (p=0.06).

Upper airway colonisation

Overall, upper airway colonisation with Gram negativebacilli was found in 44 instances: 14 (25.9%) in patientsin the ranitidine group, 12 in (22.6%) patients in the su-cralfate group and 18 (37.5%) in the group without stressulcer prophylaxis (Table 2). There was no significant dif-ference in Gram negative bacilli colonisation among thegroups nor between the ranitidine and sucralfate groupsseparately (χ2=0.15, p=0.69, d.f.:1). Of the 44 Gram neg-ative bacilli isolated from the patients with upper airwaycolonisation, 11 (25%) were Pseudomonas aeruginosa, 8 (18.2%) were Escherichia coli, 7 (15.9%) were Kleb-siella pneumoniae, 7 (15.9%) were Stenotrophomonasmaltophilia, 5 (11.3%) were Haemophilus influenzae, 3 (6.8%) were Serratia marcenscens and 3 (6.8%) wereEnterobacter aerogenes. Again, the number of days onmechanical ventilation was found to be the only signifi-cant risk factor for colonisation with Gram negative ba-cilli (p=0.04) when using a logistic regression model toadjust for differences in variables between the studygroups.

Discussion

The aim of our study was to elucidate the role of stressulcer prophylaxis in the development of VAP in criticallyill children. According to various recent studies, theoverall rate of VAP among critically ill adult patientsranges between 17.6 and 35.9% [14, 15, 16, 17]. Howev-er, only few studies have assessed the incidence of VAPin critically ill children [11, 18]. The reported incidencesof VAP in children were 9.6% and 5.2%, respectively. Inour study, we found a similar incidence of VAP in chil-dren (8.4%). The criteria used in these studies for thedefinition of VAP were similar and were all derived fromCDC definition for VAP [8]. The incidence of VAP inchildren appears, therefore, to be lower in comparison toadults.

We found a lower rate of pneumonia in patients re-ceiving sucralfate, compared to patients receiving raniti-dine (Table 2). However, our results were not significant.Moreover, we did not find a significant difference in upper airway colonisation among the three treatmentgroups (Table 2). Although the baseline characteristics ofthe ranitidine group and the sucralfate group were com-parable, some characteristics of the group of patients

without stress ulcer prophylaxis were different (Table 1).Care must therefore be taken in interpreting the results,especially when comparisons are made with the group ofpatients without prophylaxis. Heterogeneity among thethree study groups poses limitations to our study.

We found that almost half of the cases of VAP wereearly-onset pneumonia. In contrast to late-onset pneumo-nia, early-onset pneumonia was mostly caused by Grampositive cocci. Our results support the suggestion thatearly-onset pneumonia is usually caused by the introduc-tion of oropharyngeal species of bacteria usually foundthere, such as Gram positive cocci, into the trachea at thetime of intubation [14]. Modification of the gastric pHwith stress ulcer prophylaxis is therefore suggested toplay only a minor role in the development of early-onsetpneumonia.

Duration of intubation is said to be a major risk factorfor the development of VAP [10, 11]. Our results alsosuggest that the group of patients with pneumonia wereintubated longer compared to the patients without pneu-monia (Table 3). However, whether the longer durationof intubation increased the rate of pneumonia or thepneumonia itself prolonged the duration of intubation isnot clear from our study. Enteral feeding is not onlyknown to raise gastric pH, but also to increase gastricvolume. Both effects may theoretically increase the riskof nosocomial pneumonia [12]. However, when compar-ing the groups of patients with and without pneumonia,we found no evidence to support this theory (Table 3).Treatment with systemic antibiotics is known to protectagainst VAP [13]. Our results show no evidence of sucha protective effect, since similar numbers of patientswith and without pneumonia received antibiotic therapy(Table 3). We expected that prokinetics, due to their effect against gastro-oesophageal reflux, might decreasethe risk of aspiration of gastric contents and thereforeprotect against VAP. However, we found no evidence to support this hypothesis (Table 3). In contrast, wefound more pneumonia in the group of patients with pro-kinetics compared to the group of patients without.Whether treatment with prokinetics should therefore beconsidered as a risk factor for the development of VAPneeds to be evaluated in future studies.

Despite clear criteria from the CDC for VAP, reachingan unequivocal clinical and radiographic diagnosis re-mains a difficult process. The results of our study may,therefore, have been biased by different standards ofevaluation by the various radiologists in our hospital.Moreover, possible misclassification may occur when di-agnosing patients retrospectively. Nevertheless, althougherrors in diagnosing VAP may have been committed,these errors were most probably randomly distributedamong the three treatment groups. Therefore, it is un-likely that diagnostic errors grossly influenced the finalresult of our study. Since the overall rate of VAP foundin our study was low (8.4%) and comparable to previous

766

reports on the incidence of VAP in children [11, 18], it isalso unlikely that pneumonia was over-diagnosed.

In conclusion, in contrast to findings in critically illadults, the results of our preliminary study in critically illchildren show no evidence that the use of sucralfate isassociated with a lower incidence of VAP or a lower in-cidence of upper airway colonisation with Gram nega-tive bacteria. However, the sample size and the smallnumber of events per group, as well as retrospective

study design, substantially limit our conclusions. In or-der to confirm these conclusions, a large and prospectiverandomised controlled trial is mandatory. Given an inci-dence of 11% of pneumonia in the ranitidine group, asfound in our study, we estimate that a total sample sizeof 3000 patients is needed in order to give a future studya 75% power to detect a 25% reduction of pneumonia inthe sucralfate group, compared to the ranitidine group,assuming a two-sided significance test at the 0.05 level.

767

References

1. Atherton ST, White DJ (1978) Stomachas source of bacteria colonising respi-ratory tract during artificial ventilation.Lancet 2:968–969

2. Du Moulin GC, Paterson DG, Hedley-Whyte J, Lisbon A (1982) Aspirationof gastric bacteria in antacid-treatedpatients: a frequent cause of postopera-tive colonisation of the airway. Lancet1:242–245

3. Garvey BM, McCambley JA, TuxenDV (1989) Effects of gastric alkalin-ization on bacterial colonization in critically ill patients. Crit Care Med17:211–216

4. Craven DE, Kunches LM, Kilinsky V,Lichtenberg DA, Make BJ, McCabeWR (1986) Risk factors for pneumoniaand fatality in patients receiving con-tinuous mechanical ventilation. Am Rev Respir Dis 133:792–796

5. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A (2000) Bleedingand pneumonia in intensive care pa-tients given ranitidine and sucralfatefor prevention of stress ulcer: meta-analysis of randomised controlled trials. BMJ 321:1103–1106

6. Cook DJ, Reeve BK, Guyatt GH, Heyland DK, Griffith LE, BuckinghamL, Tryba M (1996) Stress ulcer prophy-laxis in critically ill patients. JAMA275:308–314

7. Cook DJ, Laine LA, Guyatt GH, RaffinTA (1991) Nosocomial pneumonia andthe role of gastric pH. Chest 100:7–13

8. Center for Disease Control (1989)CDC definitions for nosocomial infec-tions. Am Rev Respir Dis 139:1058–1059

9. Langer M, Cigada M, Mandelli M,Mosconi P, Tognoni G (1987) Early-onset pneumonia: a multicenter studyin intensive care units. Intensive CareMed 13:342–346

10. Langer M, Mosconi P, Cigada M, Mandelli M (1989) Long-term respira-tory support and risk of pneumonia incritically ill patients. Am Rev RespirDis 140:302–305

11. Fisher JE, Allen P, Fanconi S (2000)Delay of extubation in neonates andchildren after cardiac surgery: impactof ventilator-associated pneumonia. Intensive Care Med 26:942–949

12. Pingleton SK, Hinthorn DR, Liu C(1986) Enteral nutrition in patients re-ceiving mechanical ventilation: multi-ple sources of tracheal colonization in-clude the stomach. Am J Med 80:827–832

13. D’Amico R, Pifferi S, Leonetti C, TorriV, Tinazzi A, Liberati A (1998) Effec-tiveness of antibiotic prophylaxis incritically ill adult patients: systemic re-view of randomised controlled trials.BMJ 316:1275–1285

14. Prod’hom G, Leuenberger P, Koerfer J,Blum A, Chiolero R, Schaller M, Perret C, Spinnler O, Blondel J, Siegrist H, Saghafi L, Blanc D, Francioli P (1994) Nosocomial pneu-monia in mechanically ventilated pa-tients receiving antacid, ranitidine orsucral-fate as prophylaxis for stress ulcer. Ann Intern Med 120:653–662

15. Driks MR, Craven DE, Celli BR, Manning M, Burke RA, Garvin GM,Kunches LM, Farber HW, Wedel SA,McCabe WR (1987) Nosocomial pneu-monia in intubated patients given su-cralfate as compared with antacids orhistamine type 2 blockers. N Engl JMed 317:1376–1382

16. Thomason MH, Payseur ES, Hakenewerth AM, Norton J, Mehta B,Reeves TR, Moore-Swartz MW, Robbins PI (1996) Nosocomial pneu-monia in ventilated trauma patientsduring stress ulcer prophylaxis with sucralfate, antacids and ranitidine. J Trauma 41:503–508

17. Cook D, Guyatt G, Marshall J, LeasaD, Fuller H, Hall R, Peters S, RutledgeF, Griffith L, McLellan A, Wood G,Kirby A (1998) A comparison of su-cralfate and ranitidine for the preven-tion of upper gastrointestinal bleedingin patients requiring mechanical venti-lation. N Engl J Med 338:791–797

18. Singh-Naz N, Sprague BM, Patel KM,Pollack MM (1996) Risk factors fornosocomial infection in critically illchildren: a prospective cohort study.Crit Care Med 24:875–878