Nosocomial Pneumonia Risk and Stress Ulcer Prophylaxis

CHEST

Original Research CRITICAL CARE MEDICINE

Nosocomial Pneumonia Risk and Stress Ulcer Prophylaxis A Comparison of Pantoprazole vs Ranitidine in Cardiothoracic Surgery Patients Todd A. Miano, PharmD; Marc G. Reichert, PharmD; Timothy T. Houle, PhD; Drew A. MacGregor, MD, FCCP; Edward H. Kincaid, MD; and David L. Bowton, MD, FCCP

Background: Stress ulcer prophylaxis (SUP) using ranitidine, a histamine H2 receptor antagonist, has been associated with an increased risk of ventilator-associated pneumonia. The proton pump inhibitor (PPI) pantoprazole is also commonly used for SUP. PPI use has been linked to an increased risk of community-acquired pneumonia. The objective of this study was to determine whether SUP with pantoprazole increases pneumonia risk compared with ranitidine in critically ill patients. Methods: The cardiothoracic surgery database at our institution was used to identify retrospectively all patients who had received SUP with pantoprazole or ranitidine, without crossover between agents. From January 1, 2004, to March 31, 2007, 887 patients were identified, with 53 patients excluded (pantoprazole, 30 patients; ranitidine, 23 patients). Our analysis compared the incidence of nosocomial pneumonia in 377 patients who received pantoprazole with 457 patients who received ranitidine. Results: Nosocomial pneumonia developed in 35 of the 377 patients (9.3%) who received pantoprazole, compared with 7 of the 457 patients (1.5%) who received ranitidine (odds ratio [OR], 6.6; 95% confidence interval [CI], 2.9 to 14.9). Twenty-three covariates were used to estimate the probability of receiving pantoprazole as measured by propensity score (C-index, 0.77). Using this score, pantoprazole and ranitidine patients were stratified according to their probability of receiving pantoprazole. After propensity adjusted, multivariable logistic regression, pantoprazole treatment was found to be an independent risk factor for nosocomial pneumonia (OR, 2.7; 95% CI, 1.1 to 6.7; p ⴝ 0.034). Conclusion: The use of pantoprazole for SUP was associated with a higher risk of nosocomial pneumonia compared with ranitidine. This relationship warrants further study in a randomized controlled trial. (CHEST 2009; 136:440–447) Abbreviations: CI ⫽ confidence interval; HAP ⫽ hospital-acquired pneumonia; H2RA ⫽ histamine-2 receptor antagonist; IQR ⫽ interquartile range; OR ⫽ odds ratio; PPI ⫽ proton pump inhibitor; STS ⫽ Society of Thoracic Surgeons; SUP ⫽ stress ulcer prophylaxis; UGB ⫽ upper GI bleeding; VAP ⫽ ventilator-associated pneumonia; WFUBMC ⫽ Wake Forest University Baptist Medical Center

pneumonia is the leading infectious N osocomial cause of death in critically ill patients. The crude 1

mortality rate is estimated to be 20 to 50%, with an average increase in hospital stay of 7 to 9 days.2 Pneumonia treatment is also responsible for a substantial proportion of antimicrobial therapy in critically ill patients, which can impact bacterial resistance. This excess mortality, morbidity, and cost have spurred efforts directed at the identification of modifiable risk factors and prevention of nosocomial pneumonia. Acid suppressive therapy for stress ulcer prophylaxis (SUP) may increase the risk of ventilator-

associated pneumonia (VAP) in critically ill patients.3,4 The elevation of gastric pH leads to bacterial overgrowth in the stomach and potential colonization of the trachea by way of reflux into the pharynx followed by microaspiration.5,6 Consequently, the relative risks and benefits of acidsuppressive therapies for SUP are controversial. One investigation7 compared the histamine-2 receptor antagonist (H2RA) ranitidine to antacids and the acid-neutral agent sucralfate for SUP in mechanically ventilated ICU patients. The results showed no difference in clinically significant bleeding between

440

Original Research

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

the groups and a significant increase in the rate of late-onset VAP in patients receiving ranitidine or antacids. Another trial8 compared ranitidine with sucralfate in 1,200 patients receiving mechanical ventilation for ⬎ 48 h. The investigators found a significant decrease in the rate of clinically significant bleeding for patients receiving ranitidine, accompanied by a nonsignificant increase in VAP. In the past few years, proton pump inhibitors (PPIs) have been used increasingly for SUP. The rationale for their use is based mainly on their increased potency compared with H2RAs such as ranitidine.9 However, there are few data regarding their safety and efficacy compared with H2RAs in critically ill patients.10 Moreover, studies11,12 in the ambulatory population have found increased rates of community-acquired pneumonia in patients treated with PPIs compared with H2RAs. Thus, our study was designed to determine whether prophylaxis with the PPI pantoprazole increases the risk of nosocomial pneumonia compared with prophylaxis with the H2RA ranitidine in hospitalized cardiothoracic surgery patients.

Materials and Methods Study Design and Patients This is a retrospective cohort analysis of patients admitted to the Cardiothoracic Surgery Service of Wake Forest University Baptist Medical Center (WFUBMC), an 870-bed tertiary care facility. Our institutional review board approved the study protocol, and informed consent was waived. Patients were identified by query of WFUBMC patients entered into the Society of Thoracic Surgeons (STS) database.13 All patients undergoing cardiac valvular and/or revascularization surgery are entered into this database, consisting of both prospectively and retrospectively collected data elements from the clinical record. Over 600 centers nationwide contribute data to permit analysis of practice performance characteristics, major outcomes, and process-ofcare measures using rigorously defined standards. In our institution, dedicated nurses who are entirely independent of the care team and the study investigators enter these data. The STS database provides strict guidance to these data managers regarding each included data element (including pneumonia). Patients Manuscript received July 2, 2008; revision accepted February 10, 2009. Affiliations: From the Department of Pharmacy (Dr. Miano), Hospital of the University of Pennsylvania, Philadelphia, PA; the Department of Pharmacy (Dr. Reichert), North Carolina Baptist Hospital, Winston Salem, NC; and the Departments of Anesthesiology (Drs. Houle, MacGregor, and Bowton) and Cardiothoracic Surgery (Dr. Kincaid), Wake Forest University School of Medicine, Winston Salem, NC. © 2009 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal.org/site/ misc/reprints.xhtml). Correspondence to: Todd A. Miano, PharmD, Wake Forest University Baptist Medical Center, Pharmacy, 1 Medical Center Blvd, Winston Salem, NC 27157; e-mail: [email protected] DOI: 10.1378/chest.08-1634 www.chestjournal.org

were eligible for inclusion in the study if they were ⬎ 18 years of age and had been admitted to the WFUBMC Cardiothoracic Surgery Service from January 2004 through March 2007. For patients with multiple hospital admissions during the study period, only the first hospital admission was eligible for inclusion. All patients admitted to this service receive acid-suppressive therapy for SUP. Hospital admission order sets list both pantoprazole and ranitidine as options, with the choice of agent left to the discretion of the prescribing physician. For inclusion in this study, patients must have received either pantoprazole or ranitidine without crossover between the two agents during the index hospital admission. Patients were excluded if they had a diagnosis of pneumonia within the preceding 3 months, a history of clinically significant dysphagia, documented aspiration during their hospital admission, a history of immunosuppression (ie, long-term corticosteroid treatment with ⬎ 10 mg of prednisone or equivalent per day, receipt of chemotherapy within the preceding year, recent solid organ transplantation, treatment with any antirejection medication), or a diagnosis of HIV infection. Data Collection and Variables Demographic and clinical data were abstracted via query of the STS database. However, the following variables were not available in the database: history of gastroesophageal reflux disease, diabetes mellitus, asthma, cerebrovascular disease, tobacco use, use of acid suppressive therapy before hospital admission, and tracheostomy placement postoperatively. These variables were manually abstracted from the electronic and paper medical record by the primary investigator. Previous investigators have recommended including both preoperative and postoperative variables for risk assessment of cardiothoracic surgery patients. Accordingly, we chose variables that have previously been found to be predictors of outcome in this population.14 –18 Comorbid illness such as heart failure, COPD, diabetes, and stroke are elements of the validated risk adjustment model used in the STS database. Postoperative elements including transfusion requirements, duration of mechanical ventilation, and need for intraaortic balloon pump support postoperatively have been observed by others to be important predictors of outcome (including pneumonia).19,20 In contrast, physiology-based scoring systems commonly used in other critically ill populations to estimate severity of illness have proved inadequate in the postoperative cardiothoracic population,14,21 and thus were not included in our analysis. Outcome The primary outcome of interest was the incidence of nosocomial pneumonia. This end point is a metric included in the STS database. Independent data managers review the clinical record and hospital discharge summaries to identify patients treated with antibiotics for nosocomial pneumonia. They search for the following terms/diagnoses: pneumonia, ventilator pneumonia, positive respiratory cultures, respiratory infection, and hospitalacquired pneumonia (HAP). For the purposes of this study, pneumonia cases were initially identified by query of the STS database. Patients identified by the query as having nosocomial pneumonia postoperatively were then screened against the study criteria for pneumonia, as defined by the National Nosocomial Infection Surveillance system algorithm (Table 1).22 Patients with pneumonia onset after at least 48 h of mechanical ventilation or within 48 h of extubation were classified as having VAP. Quantitative cultures of tracheal aspirates are routinely used in our institution for the microbiological confirmation of the etiology of VAP. This technique has been found to perform comparably to CHEST / 136 / 2 / AUGUST, 2009

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

441

Table 1—Pneumonia Definition: National Nosocomial Infections Surveillance System Algorithm for Diagnosis of Pneumonia Radiology*

Signs/Symptoms/Laboratory

Two or more serial chest radiographs with at least one of the following: New or progressive and persistent infiltrate Consolidation

Cavitations

For any patient, at least one of the following: Fever (⬎ 38oC or ⬎ 100.4oF) with no other recognized cause Leukopenia (⬍ 4,000 WBCs/ ␮L) or leukocytosis (⬎ 12,000 WBCs/␮L) For adults ⬍ 70 years old, altered mental status with no other recognized cause And at least two of the following: New onset of purulent sputum or change in character of sputum, or increased respiratory secretions, or increased suctioning requirements New onset of worsening cough, dyspnea, or tachypnea Rales or bronchial breath sounds Worsening gas exchange (eg, O2 desaturations such as Pao2/Fio2 ratio of 240, increased oxygen requirements, or increased ventilation demand)

*In patients without underlying pulmonary or cardiac disease (eg, respiratory distress syndrome, bronchopulmonary dysplasia, pulmonary edema, or COPD), one definitive chest radiograph is acceptable.

quantitative bronchoscopic cultures in several studies. Additionally, hospital discharge summaries were reviewed for the incidence of acute upper GI bleeding (UGB). Statistical Analysis We used a propensity score adjustment to isolate the impact of medication on the development of pneumonia. Propensity score analysis is a useful means of balancing the distribution of covariates associated with the nonrandom allocation of treatment that is often found in retrospective studies. Since its introduction by Rosenbaum and Rubin23 in 1983, the propensity score has been increasingly used in observational studies to adjust for important differences between groups of interest. The propensity score for an individual is the probability of being treated conditional on the individual’s background characteristics.23,24 Propensity scores are calculated using multivariate logistic regression, with receipt of treatment (pantoprazole) as the dependent variable. We defined and controlled for 23 measured patient variables that were considered a priori to contribute to the probability that an individual would be given pantoprazole (Table 2). Using these data, the propensity for receiving pantoprazole was calculated for all patients in both treatment groups. The propensity score for each patient was then

forced into a second multivariate logistic regression model, this time with pneumonia as the dependent variable. This technique allows us to control for selection bias in our final regression model. The discriminative power of the propensity scores was quantified using the area under the receiver operating-characteristic curve (C-index). The proportion of patients in whom pneumonia developed is also presented as a function of a tertile split of the propensity score (“low, medium, and high propensity for pantoprazole”). Descriptive statistics are presented as the mean ⫾ SD, median (interquartile range [IQR]), or No. (%) as appropriate. Univariate differences between medications were assessed using the MannWhitney U test and ␹2 tests, as appropriate. All analyses were performed using a statistical software package (SPSS, version 13.0; SPSS, Inc; Chicago, IL).

Results The flow of patients through our study is depicted in Figure 1. During the study period, 1,682 patients were entered into the STS database, with 795 of these patients deemed ineligible for inclusion because of crossover therapy. Of the 887 remaining patients, 407 received pantoprazole and 480 received ranitidine. From this population, 53 patients were excluded (30 who were receiving pantoprazole, 23 who were receiving ranitidine) with the most common reasons for exclusion in both groups being a prior pneumonia diagnosis and a history of immunosuppression. As presented in Table 2, pantoprazole patients were significantly older than ranitidine patients (mean age, 65.3 ⫾ 11.8 vs 60.9 ⫾ 13.5 years, respectively), had a greater incidence of reflux disease (24.7% vs 12.0%, respectively), and New York Heart Association class IV heart failure (26.5% vs 11.4%, respectively). In addition, pantoprazole patients had a longer median duration of mechanical ventilation (13 h [IQR, 8 to 22 h] vs 10 h [7 to 17 h], respectively), greater mean use of blood products such as packed RBCs (3.8 ⫾ 4.5 vs 3.0 ⫾ 2.2 units, respectively) and need for tracheostomy (3.9% vs 0.7%, respectively). However, the utilized statistical model adequately balanced the considered covariates, adjusting their influence on the estimated risk of pneumonia associated with pantoprazole. This can be seen in Table 2, where the observed p values for all considered covariates are reduced to statistical nonsignificance after controlling for the propensity score (p value after adjustment column). The incidence of UGB in this population was low, occurring in three patients in the pantoprazole group (0.8%) vs one patient in the ranitidine group (0.2%; p ⫽ 0.333). Outcome Query of the STS database produced 101 pneumonia cases, with 34 in the crossover group. An

442

Original Research

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

Table 2—Variables Included in Propensity Score Analysis Pantoprazole (n ⫽ 377)

Variables Age, yr† Male gender GERD Diabetes mellitus Heart failure, NYHA class I II III IV COPD None Mild Moderate Severe Asthma Cerebrovascular disease Current tobacco use History of tobacco use Previous PPI Previous H2RA Mechanical ventilation, h‡ Intraoperative transfusions, units RBCs Mean (SD) Median (IQR) Transfused Fresh-frozen plasma Mean (SD) Median (IQR) Transfused Platelets Mean (SD) Median (IQR) Transfused Cryoprecipitate Mean (SD) Median (IQR) Transfused Postoperative transfusions, units† RBCs Mean (SD) Median (IQR) Transfused Fresh-frozen plasma Mean (SD) Median (IQR) Transfused Platelets Mean (SD) Median (IQR) Transfused

Ranitidine (n ⫽ 457)

65.3 (11.8) 253 (67.1) 93 (24.7) 110 (29.2)

60.9 (13.5) 132 (71.1) 55 (12.0) 157 (34.4)

112 (29.7) 83 (22.0) 86 (22.8) 96 (25.5)

159 (34.8) 125 (27.4) 121 (26.5) 52 (11.4)

270 (71.6) 83 (22.0) 13 (3.5) 11 (2.9) 10 (2.7) 35 (9.3) 87 (23.1) 128 (33.9) 104 (27.6) 16 (4.2) 13 (8–22)

363 (79.4) 73 (15.9) 17 (3.7) 4 (0.9) 14 (3.1) 34 (7.4) 123 (26.9) 143 (31.3) 21 (4.6) 35 (7.7) 10 (7–17)

2.68 (1.9) 2 (1–3) 193 (51.2)

p Value* Before After 0.001 0.21 0.001 0.11

0.99 0.99 0.95 0.99

0.001

0.99

0.14

0.99

0.72 0.34 0.20 0.41 0.001 0.04 0.001

0.99 0.99 0.99 0.99 0.63 0.98 0.15

2.10 (1.2) 2 (1–3) 171 (37.4)

0.001

0.96

2.74 (1.7) 2 (2–4) 34 (9.0)

2.52 (0.9) 2 (2–3) 32 (7.0)

0.09

0.69

2.10 (1.0) 2 (1–3) 110 (29.2)

2.12 (0.9) 2 (1–3) 122 (26.7)

0.52

0.17

1.30 (0.6) 1.36 (0.6) 1 (1–1.25) 1 (1–1.75) 0.11 25 (6.6) 19 (4.2)

0.54

3.87 (4.5) 2 (1–4) 172 (45.6)

2.74 (2.5) 2 (1–3) 94 (20.6)

0.001

0.13

3.00 (2.2) 2 (2–4) 35 (9.3)

2.58 (1.1) 2 (2–4) 17 (3.7)

0.001

0.69

1.62 (0.9) 1 (1–2) 65 (17.2)

1.82 (1.0) 1.5 (1–2) 27 (5.9)

0.001

0.06

(Continued)

www.chestjournal.org

Table 2—(Continued)

Variables Cryoprecipitate Mean (SD) Median (IQR) Transfused Tracheostomy Intraaortic balloon pump

Pantoprazole (n ⫽ 377)

Ranitidine (n ⫽ 457)

1.38 (0.6) 1 (1–2) 26 (6.9) 15 (3.9) 13 (3.5)

1.00 (0.0) 1 (1–1) 7 (1.5) 3 (0.7) 9 (1.9)

p Value* Before After

0.001

0.74

0.004 0.19

0.93 0.99

Values are given as No. (%), unless otherwise indicated. GERD ⫽ gastroesophageal reflux disease; NYHA ⫽ New York Heart Association. *The p values for all covariates are shown before and after propensity score adjustment. The analysis adequately balanced covariates between treatment groups, as reflected by nonsignificant p values after adjustment. †Values are given as the mean (SD). ‡Values are given as median (IQR).

additional 25 cases did not meet the study criteria for pneumonia (pantoprazole group, 16 cases; ranitidine group, 9 cases), leaving 42 cases for analysis. Studydefined nosocomial pneumonia occurred in 35 of 377 patients (9.3%) in the pantoprazole group vs 7 of 457 patients (1.5%) in the ranitidine group, corresponding to an unadjusted odds ratio (OR) of 6.6 (95% confidence interval [CI], 2.9 to 14.9) [Fig 2]. In the pantoprazole group, 31 of 35 cases (88.5%) were classified as VAP, with the remaining cases classified as HAP. In the ranitidine group, 5 of 7 cases (71.4%) were classified as VAP with the remainder HAP. Respiratory culture results can be found in Table 3. We analyzed the year-to-year variation in pneumonia incidence for each group using the ␹2 test. This analysis showed no significant intragroup variation in the yearly pneumonia incidence over the 37-month study period (data not shown). Pneumonia was found in 34 of 795 patients (4.3%) in the crossover group. An incidence somewhere in the middle between pantoprazole and ranitidine would seem to fit our hypothesis; however, we interpret these data with caution. The patients in the crossover group received varying durations of therapy with each agent. Moreover, pneumonia episodes often occurred during a transition phase between agents, where the acid suppressive effects of each agent could be additive. In addition, we do not have information regarding other important covariates to quantify the baseline pneumonia risk in the crossover. Propensity score analysis was conducted to account for the nonrandom treatment allocation observed during the study period. The propensity score adequately discriminated between the two groups with each predictor successfully balanced by propensity adjustment, p ⬎ 0.06, and a C-index of 0.77. CHEST / 136 / 2 / AUGUST, 2009

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

443

Wake Forest University Baptist Medical Center Society of Thoracic Surgeons Database n=1682 Cross over n=795 Non-cross over n=887

Pantoprazole n=407

Ranitidine n=480

Excluded (n=30)

Excluded (n=23)

11 Recent Pneumonia 8 Immunosuppression 4 Aspiration 3 Dysphagia 4 Repeat admission

10 Recent Pneumonia 5 Immunosuppression 3 Aspiration 4 Dysphagia 1 Repeat admission

Pantoprazole n=377

Ranitidine n=457

Figure 1. Patient flow.

Propensity scores were also used to stratify the sample, splitting the population into tertiles of probability that a patient would be given pantoprazole (Fig 3). In Figure 3, the propensity for patients to receive pantoprazole increases from left to right, as does the incidence of pneumonia. This suggests that the same factors guiding prescription of pantoprazole also conditioned the risk for pneumonia. However, in each tertile, patients in both treatment groups have a similar propensity to receive pantoprazole (and similarly balanced distributions of covariates). Thus, the difference in pneumonia incidence in each tertile reflects the isolated effect of

10 9.3

9

pneumonia percent

8 7 6 Pantoprazole Ranitidine

5 4 3 2

1.5

1 0

Figure 2. Unadjusted pneumonia incidence.

treatment with pantoprazole. In our primary analysis, a propensity adjusted, multivariable linear regression found pantoprazole treatment to be an independent risk factor for nosocomial pneumonia (OR, 2.7; 95% CI, 1.1 to 6.7; p ⫽ 0.034). Discussion This study is the largest comparison of PPIs and H2RAs in the critical care population. Our results show an increase in the risk of nosocomial pneumonia in patients receiving pantoprazole for SUP compared with those that received ranitidine. Previous studies7,8 of SUP comparing an H2RA to acidneutral therapy have found small increases in pneumonia rates. In the largest trial to date,8 the statistically insignificant increase in pneumonia was offset by a lower rate of clinically significant bleeding. Thus, many clinicians judge the uncertain increase in risk of pneumonia to be an acceptable one. However, the overall incidence of bleeding in critically ill patients has decreased substantially with improvements in modern intensive care, independent of prophylactic measures.9 Moreover, PPIs have not been shown to reduce the incidence of clinically significant bleeding compared with H2RAs.10 The overall incidence of UGB in our population was low and did not differ between treatment groups. However, it should be noted that our study was not

444

Original Research

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

Table 3—Respiratory Culture Results From Patients With VAP Organisms Gram-negative Klebsiella pneumoniae Pseudomonas aeruginosa Serratia marcescens Stenotrophomonas maltophilia Acinetobacter baumannii Escherichia coli Enterobacter cloacae Enterobacter aerogenes Enterobacter asburiae Klebsiella oxytoca Citrobacter freundii Proteus mirabilis Gram-positive Methicillin-resistant Staphylococcus aureus Methicillin-sensitive S aureus Culture-negative*

Pantoprazole (n ⫽ 31)

Ranitidine (n ⫽ 5)

8 (25.8) 6 (19.4) 6 (19.4) 5 (16.1) 3 (9.7) 3 (9.7) 1 (3.2) 1 (3.2) 0 1 (3.2) 1 (3.2) 1 (3.2)

1 (14.3) 0 1 (14.3) 0 0 2 (28.6) 0 1 (14.3) 1 (14.3) 1 (14.3) 0 0

2 (6.5)

0

1 (3.2) 2 (6.5)

0 0

Values are given as No. (%). Organisms shown are those isolated from endotracheal aspirates from patients with VAP; the total number of organisms is more than the number of VAP episodes because several patients had more than one isolate. Patients with HAP in each group did not have sputum samples obtained. *These patients received antibiotics within 48 h of sample collection.

designed to assess the risk of UGB. Data regarding important risk factors such as the presence of coagulopathy were not collected. Our findings are in agreement with the results of outpatient studies11,12 that have observed increased rates of communityacquired pneumonia in patients receiving PPIs compared with H2RAs. The underlying mechanism for the observed increase in pneumonia may be more potent acid suppression with a resultant increase in gastric colonization, although none of these studies, including ours, examined potential mechanisms.

14 12.4

pneumonia percent

12 10 8.7

8 Pantoprazole Ranitidine

6 4.4

4 1.4

2 0

0.6

0 n=

Low 64 214

Medium 103 175

High 210 68

Figure 3. Stratified pneumonia incidence. Treatment groups are stratified by their propensity to receive pantoprazole. www.chestjournal.org

Data from Madl et al25 showed PPIs to provide superior acid suppression compared with H2RAs in critically ill patients. In addition, it has been demonstrated26 that patients receiving omeprazole experience significantly greater gastric bacterial growth compared with those receiving cimetidine. To date, the largest prospective, randomized, placebo-controlled trial27 comparing the two classes of acid-suppressive agents analyzed the effectiveness of omeprazole suspension compared with cimetidine continuous infusion. The primary objective was to determine the noninferiority of omeprazole in preventing clinically significant bleeding. Analysis of the 360 patients resulted in omeprazole being judged noninferior according to the criteria of the study. The study also found a 2% absolute increase in the incidence of VAP; however, the difference was nonsignificant. Interpretations of these results are limited by the population size of the study, which was not powered to assess the risk of nosocomial pneumonia. As a consequence of its observational design, our study is limited by a lack of control over treatment assignment. Accordingly, propensity score methodology was used to balance the covariate distribution surrounding the observed nonrandom treatment allocation. However, this methodology is only able to control for observed confounding covariates and not for unobserved ones. It is possible that other important differences exist that were not included in our statistical model. As with any observational design, it is impossible to eliminate this potential for residual confounding. Only a prospective, randomized trial can adequately balance both measured and unmeasured covariates. As such, we consider our results hypothesis generating only. In addition, there was a high incidence of treatment crossover in the population, which excluded a large number of patients. The study was performed in a group of cardiothoracic surgery patients; thus, these results may not apply to other populations. The retrospective design also precluded any gastric sampling for pH analysis or culturing. We did not account for the use of nasogastric vs orogastric feeding tubes, use of enteral nutrition, or use of prior antibiotic therapy. In addition, we did not document rates of using chlorhexidine mouthwash or semirecumbent positioning. However, during the period of data collection these prophylactic measures were mandated in all patients by an ICU treatment protocol. Additionally, treatment protocols call for the early advancement of enteral nutrition for all patients and the preferential placement of orogastric feeding tubes. We were unable to account for any differences in duration of acid suppression therapy between the two treatment groups. Differences in CHEST / 136 / 2 / AUGUST, 2009

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

445

exposure to acid suppression could have affected pneumonia risk. However, the general practice in our ICU is to continue SUP until patients are transferred to a lower level of care, regardless of which agent is being used. The incidence of pneumonia in our population was relatively low and may not reflect the incidence in the general population. We do not report secondary end points such as length of stay, mortality, or overall use of antimicrobial therapy. Although these end points would further put into context our findings, their interpretation would be difficult considering the many factors affecting these outcomes in critically ill patients. Finally, there were significant differences between the treatment groups, resulting in large CIs for the primary end point. VAP is associated with prolonged duration of mechanical ventilation,28 which may also increase the rate of tracheostomy. Hence, the OR of the development of pneumonia in patients receiving pantoprazole was potentially underestimated due to the inclusion of these variables in our propensity analysis. We cannot determine how much of the observed difference in these variables is attributable to a higher initial severity of illness in the pantoprazole group or directly to the use of pantoprazole. Conclusions In a population of cardiothoracic surgery patients, the use of pantoprazole for SUP was associated with a higher risk of nosocomial pneumonia compared with ranitidine. This relationship warrants further study in a randomized controlled trial. Acknowledgments Author contributions: Dr. Miano had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Specifically, he contributed to study design, served as primary data collector, and maintained the study data spreadsheet. In addition, he contributed to data analysis, interpretation, and drafted the initial manuscript and all subsequent revisions. Dr. Reichert was responsible for study conception, design, and interpretation of the data. In addition, he contributed to data collection, provided critical review of the manuscript, and gave final approval of the revised draft for publication. Dr. Houle contributed to study design, statistical analysis, interpretation of data, provided critical review of the manuscript, and gave final approval of the revised draft for publication. Dr. MacGregor participated in study design, provided critical review of the manuscript, and gave final approval of the revised draft for publication. Dr. Kincaid participated in study design, provided critical review of the manuscript, and gave final approval of the revised draft for publication. Dr. Bowton was responsible for study design, interpretation of the data, provided critical review of the manuscript, and gave final approval of the revised draft for publication.

Financial/nonfinancial disclosures: Dr. Houle has received grant support from ENDO pharmaceuticals and the National Institutes of Health, and royalties from Monster Worldwide, Inc, and has served as a speaker for Bayer Pharmaceuticals. Dr. Bowton has received grant support from Theravance and PPD, and has been a consultant to Wyeth and Spacelabs. Drs. Miano, Reichert, MacGregor, and Kincaid have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

References 1 Porzecanski I, Bowton DL. Diagnosis and treatment of ventilator-associated pneumonia. Chest 2006; 130:597– 604 2 Tablan OC, Anderson LJ, Besser R, et al. Guidelines for preventing health-care–associated pneumonia, 2003: recommendations of the CDC and the Healthcare Infection Control Practices Advisory Committee. MMWR Recomm Rep 2004; 53:1–36 3 Cook DJ, Reeve BK, Guyatt GH, et al. Stress ulcer prophylaxis in critically ill patients: resolving discordant metaanalyses. JAMA 1996; 275:308 –314 4 Messori A, Trippoli S, Vaiani M, et al. Bleeding and pneumonia in intensive care patients given ranitidine or sucralfate for prevention of stress-ulcer: meta-analysis of randomized controlled trials. BMJ 2000; 321:1103–1106 5 du Moulin GC, Paterson DG, Hedley-Whyte J, et al. Aspiration of gastric bacteria in antacid-treated patients: a frequent cause of postoperative colonization of the airway. Lancet 1982; 319:242–245 6 Donowitz LG, Page MC, Mileur BL, et al. Alteration of normal gastric flora in critical care patients receiving antacid and cimetidine therapy. Infect Control 1986; 7:23–26 7 Prod’hom G, Leuenberger P, Koerfer J, et al. Nosocomial pneumonia in mechanically ventilated patients receiving antacid, ranitidine, or sucralfate as prophylaxis for stress ulcer: a randomized controlled trial. Ann Intern Med 1994; 120:653– 662 8 Cook D, Guyatt G, Marshall J, et al. A comparison of sucralfate and ranitidine for the prevention of upper gastrointestinal bleeding in patients requiring mechanical ventilation. N Engl J Med 1998; 338:791–797 9 Hsu TC, Su CF, Leu SC, et al. Omeprazole is more effective than a histamine H2-receptor blocker for maintaining a persistent elevation of gastric pH after colon resection for cancer. Am J Surg 2004; 187:20 –23 10 Kantorova I, Svoboda P, Scheer P, et al. Stress ulcer prophylaxis in critically ill patients: a randomized controlled trial. Hepatogastroenterology 2004; 51:757–761 11 Laheij RJ, Sturkenboom MC, Hassing RJ, et al. Risk of community-acquired pneumonia and use of gastric acidsuppressive drugs. JAMA 2004; 292:1955–1960 12 Gulmez SE, Holm A, Frederiksen H, et al. Use of proton pump inhibitors and the risk of community-acquired pneumonia: a population-based case-control study. Arch Intern Med 2007; 167:950 –955 13 Society of Thoracic Surgeons. Database. Available at: http:// www.sts.org/sections/stsnationaldatabase/. Accessed June 1, 2009 14 Michalopoulos A, Tzelepis G, Dafni U, et al. Determinants of hospital mortality after coronary artery bypass grafting. Chest 1999; 115:1598 –1603 15 Rady MY, Ryan T, Starr NJ. Perioperative determinants of morbidity and mortality in elderly patients undergoing cardiac surgery. Crit Care Med 1998; 26:225–235

446

Original Research

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

16 Halkos ME, Puskas JD, Lattouf OM, et al. Impact of preoperative neurologic events on outcomes after coronary artery bypass grafting. Ann Thorac Surg 2008; 86:504 –510 17 Magovern JA, Sakert T, Magovern GJ, et al. A model that predicts morbidity and mortality after coronary artery bypass graft surgery. J Am Coll Cardiol 1996; 28:1147–1153 18 Higgins TL, Estafanous FG, Loop FD, et al. Stratification of morbidity and mortality outcome by preoperative risk factors in coronary artery bypass patients: a clinical severity score. JAMA 1992; 267:2344 –2348 19 Bonten MJ, Kollef MH, Hall JB. Risk factors for ventilatorassociated pneumonia: from epidemiology to patient management. Clin Infect Dis 2004; 38:1141–1149 20 Shorr AF, Duh MS, Kelly KM, et al. Red blood cell transfusion and ventilator-associated pneumonia: a potential link? Crit Care Med 2004; 32:666 – 674 21 Turner JS, Morgan CJ, Thakrar B, et al. Difficulties in predicting outcome in cardiac surgery patients. Crit Care Med 1995; 23:1843–1850 22 Centers for Disease Control and Prevention. National Nosocomial Infections Surveillance System (NNIS). Available at: http://www.cdc.gov/ncidod/dhqp/nnis.html. Accessed June 1, 2009

www.chestjournal.org

23 Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983; 70:41–55 24 D’ Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998; 17:2265–2281 25 Madl C, Kitzberger R, Warzawska J, et al. Effect of intravenous esomeprazole verus ranitidine on gastric pH in critically ill patients: a prospective, randomized, double blind, multicenter study [abstract]. Crit Care Med 2007; 35:A249 26 Thorens J, Froehlich F, Schwizer W, et al. Bacterial overgrowth during treatment with omeprazole compared with cimetidine: a prospective randomised double blind study. Gut 1996; 39:54 –59 27 Conrad SA, Gabrielli A, Margolis B, et al. Randomized, double-blind comparison of immediate-release omeprazole oral suspension versus intravenous cimetidine for the prevention of upper gastrointestinal bleeding in critically ill patients. Crit Care Med 2005; 33:760 –765 28 Safdar N, Dezfulian C, Callard HR, et al. Clinical and economic consequences of ventilator-associated pneumonia: a systematic review. Crit Care Med 2005; 33:2184 –2193

CHEST / 136 / 2 / AUGUST, 2009

Downloaded from www.chestjournal.org by Kimberly Henricks on August 10, 2009 Copyright © 2009 American College of Chest Physicians

447