Research Original Investigation
H2 Receptor Antagonists vs Proton Pump Inhibitors
10. Barkun AN, Bardou M, Pham CQ, Martel M. Proton pump inhibitors vs. histamine 2 receptor antagonists for stress-related mucosal bleeding prophylaxis in critically ill patients: a meta-analysis. Am J Gastroenterol. 2012;107(4):507-520. 11. Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA. 2004;292(16): 1955-1960. 12. Gulmez SE, Holm A, Frederiksen H, Jensen TG, Pedersen C, Hallas J. Use of proton pump inhibitors and the risk of community-acquired pneumonia: a population-based case-control study. Arch Intern Med. 2007;167(9):950-955. 13. Sarkar M, Hennessy S, Yang YX. Proton-pump inhibitor use and the risk for community-acquired pneumonia. Ann Intern Med. 2008;149(6): 391-398. 14. Eom CS, Jeon CY, Lim JW, Cho EG, Park SM, Lee KS. Use of acid-suppressive drugs and risk of pneumonia: a systematic review and meta-analysis. CMAJ. 2011;183(3):310-319. 15. Herzig SJ, Howell MD, Ngo LH, Marcantonio ER. Acid-suppressive medication use and the risk for hospital-acquired pneumonia. JAMA. 2009;301(20):2120-2128. 16. Kwok CS, Arthur AK, Anibueze CI, Singh S, Cavallazzi R, Loke YK. Risk of Clostridium difficile infection with acid suppressing drugs and antibiotics: meta-analysis. Am J Gastroenterol. 2012;107(7):1011-1019. 17. Howell MD, Novack V, Grgurich P, et al. Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection. Arch Intern Med. 2010;170(9):784-790. 18. Safdar N, Dezfulian C, Collard HR, Saint S. Clinical and economic consequences of ventilator-associated pneumonia: a systematic review. Crit Care Med. 2005;33(10):2184-2193. 19. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ. Estimating the proportion of healthcare-associated infections that are reasonably preventable and the related
mortality and costs. Infect Control Hosp Epidemiol. 2011;32(2):101-114. 20. Burgmann H, Hiesmayr JM, Savey A, Bauer P, Metnitz B, Metnitz PG. Impact of nosocomial infections on clinical outcome and resource consumption in critically ill patients. Intensive Care Med. 2010;36(9):1597-1601. 21. Kollef MH, Hamilton CW, Ernst FR. Economic impact of ventilator-associated pneumonia in a large matched cohort. Infect Control Hosp Epidemiol. 2012;33(3):250-256. 22. Lipp MJ, Nero DC, Callahan MA. Impact of hospital-acquired Clostridium difficile. J Gastroenterol Hepatol. 2012;27(11):1733-1737. 23. Ghantoji SS, Sail K, Lairson DR, DuPont HL, Garey KW. Economic healthcare costs of Clostridium difficile infection: a systematic review. J Hosp Infect. 2010;74(4):309-318. 24. McGlone SM, Bailey RR, Zimmer SM, et al. The economic burden of Clostridium difficile. Clin Microbiol Infect. 2012;18(3):282-289. 25. Streiner DL, Norman GR. The pros and cons of propensity scores. Chest. 2012;142(6):1380-1382. 26. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424.
Campaign: international guidelines for management of severe sepsis and septic shock, 2012. Intensive Care Med. 2013;39(2):165-228. 31. Cook D, Heyland D, Griffith L, Cook R, Marshall J, Pagliarello J; Canadian Critical Care Trials Group. Risk factors for clinically important upper gastrointestinal bleeding in patients requiring mechanical ventilation. Crit Care Med. 1999;27(12):2812-2817. 32. Cook DJ, Fuller HD, Guyatt GH, et al; Canadian Critical Care Trials Group. Risk factors for gastrointestinal bleeding in critically ill patients. N Engl J Med. 1994;330(6):377-381. 33. Brown RB, Klar J, Teres D, Lemeshow S, Sands M. Prospective study of clinical bleeding in intensive care unit patients. Crit Care Med. 1988;16(12):1171-1176. 34. Cochard JF, Leger A, Pinaquy C, et al. Gastrointestinal bleeding in trauma patients: incidence and risk factors [abstract]. Intensive Care Med. 1997;23:S140. 35. Schuster DP, Rowley H, Feinstein S, McGue MK, Zuckerman GR. Prospective evaluation of the risk of upper gastrointestinal bleeding after admission to a medical intensive care unit. Am J Med. 1984;76(4):623-630.
27. Herzig SJ, Vaughn BP, Howell MD, Ngo LH, Marcantonio ER. Acid-suppressive medication use and the risk for nosocomial gastrointestinal tract bleeding. Arch Intern Med. 2011;171(11):991-997.
36. Huang J, Cao Y, Liao C, Wu L, Gao F. Effect of histamine-2-receptor antagonists versus sucralfate on stress ulcer prophylaxis in mechanically ventilated patients: a meta-analysis of 10 randomized controlled trials. Crit Care. 2010;14(5):R194.
28. 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(4):760-765.
37. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A. Bleeding and pneumonia in intensive care patients given ranitidine and sucralfate for prevention of stress ulcer: meta-analysis of randomised controlled trials. BMJ. 2000;321(7269):1103-1106.
29. Levy MJ, Seelig CB, Robinson NJ, Ranney JE. Comparison of omeprazole and ranitidine for stress ulcer prophylaxis. Dig Dis Sci. 1997;42(6):1255-1259.
38. Jhung MA, Banerjee SN. Administrative coding data and health care-associated infections. Clin Infect Dis. 2009;49(6):949-955.
30. Dellinger RP, Levy MM, Rhodes A, et al; Surviving Sepsis Campaign Guidelines Committee including The Pediatric Subgroup. Surviving Sepsis
Invited Commentary
Unintended Consequences of Therapy in the Intensive Care Unit Richard Kim, MD; Christopher H. Goss, MD, MS
Gastric acidity suppression is recommended for critically ill patients,1 with strong clinical evidence for efficacy, particularly in mechanically ventilated patients. Acid-suppressive agents such as proton pump inhibitors (PPIs) and histamine-2 receptor antagonists (H2RA) are thought to decrease the Related article page 564 risk of acute gastrointestinal tract (GI) hemorrhage from stress-induced ulcers. However, the suppression of gastric acidity has been shown in several observational studies to lead to higher rates of pneumonia and Clostridium difficile infection (CDI) in settings other than the 574
intensive care unit (ICU).2,3 It is unclear whether these findings extend to the ICU, where nosocomial infections entail considerable mortality and costs. In this issue of the journal, MacLaren and colleagues4 performed a large retrospective cohort study using billing data on mechanically ventilated patients who were administered a PPI or H2RA for at least 48 hours while intubated. For their main outcomes, they measured not only the number of pneumonia and CDI events but also the intended primary effect of acidsuppressive therapy: the number of GI hemorrhage events occurring more than 48 hours after initiation of mechanical
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
jamainternalmedicine.com
H2 Receptor Antagonists vs Proton Pump Inhibitors
Original Investigation Research
ventilation. Because there were not enough patients who received no acid-suppressive therapy, they compared those who received PPI with those who received H2RA. They hypothesized that the more potent acid suppression effects of PPI would translate into fewer GI bleeding events compared with patients receiving H2RA prophylaxis but a higher number of pneumonia and CDI cases. Adjustments were made using a number of covariates derived from the billing data, as well as a propensity score for treatment assignment. Consistent with their hypothesis, they found that pneumonia and CDI were more frequent in those receiving a PPI compared with those receiving an H2RA. However, they unexpectedly found that the risk of GI hemorrhage was higher in those receiving a PPI. An analysis of other risk factors associated with GI hemorrhage events found that acute or chronic hepatic injury and coagulopathy were associated with increased risk, while thrombocytopenia and use of platelet inhibitors was protective. Increased duration of mechanical ventilation was associated with an increased risk of both pneumonia and CDI, and use of carbapenems and piperacillin was associated with increased risk of CDI, as one might expect. One of the challenges with this analysis is that it uses a large administrative claims data set, where outcomes are inferred using International Classification of Diseases, Ninth Revision, Clinical Modification codes. The possibility of misclassification is suggested by the fact that the frequency of their primary outcomes (pneumonia, CDI, and GI hemorrhage) are strikingly high—in the case of pneumonia, an overall rate of 34.2%. A previous study by Lawson and colleagues5 comparing administrative claims data with medical chart review have found significant limitations in the accuracy of identifying surgical complications. To address this, the authors used probabilistic modeling of potential misclassification of the outcome measures and covariates as a sensitivity analysis. Their Monte Carlo simulation addressed the effect of random variance by allowing parameters to vary by ±25%. However, Lawson and colleagues5 found that the percentage of recorded pneumonia complications as determined by claims data that were later determined to be false positive by medical chart review was 63%, suggesting that even larger bounds for the parameter misclassification in the sensitivity analysis may be warranted to support the conclusions in this study. Another threat to the study’s conclusions is the possibility of indication bias (also termed confounding by indication)—a risk that has been clearly acknowledged by MacLaren and colleagues.4 The concern is that the outcomes of one group of patients (those who received PPI) are confounded by the fact they may have been sicker or otherwise predisposed to the study outcomes compared with those administered an alternative therapy (H2RA). In fact, indication bias provides a unifying explanation for why patients administered a PPI were at higher risk for all the end points noted. The authors attempt to address this and other potential confounders by adjusting for covariates in
ARTICLE INFORMATION Author Affiliations: Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of Washington, Seattle (Kim, Goss);
their regression model, as well as generating a propensity score. Propensity score matching, in principle, is superior to traditional adjustment by creating pseudorandomization of measured confounders. MacLaren and colleagues4 performed covariate adjustment using the propensity score (based on the probability of receiving H2RA vs PPI) in their multivariate analysis, as well as a separate propensity-matched analysis. After matching, the authors found that the propensity score was successful in generating 2 populations with very similar clinical and hospital characteristics, and their results remained robust. Other methods have been developed to address the problem of confounding by indication in observational studies. These include the use of an instrumental variable and restriction. Instrumental variables, widely used in econometrics, require the presence of a suitable variable (or combination of variables) that is strongly associated with the treatment but not the outcome, which may not have been available in this data set. Furthermore, the statistical literature suggests that none of the aforementioned methods are fully successful in removing indication bias.6 It should be noted that the potential for misclassification and indication bias exists in any retrospective study of administrative claims data. One should not discount the findings of MacLaren and colleagues4 based solely on the challenges of addressing these limitations. After selecting appropriate methods for addressing indication bias and misclassification, one should ask the following additional questions to infer causality of the association noted: (1) Is the effect biologically plausible; (2) Is the effect large; (3) Is the anticipated bias in the direction of the null hypothesis; (4) Does the risk factor precede the outcomes? The ability to answer yes to these questions adds greater confidence to the study conclusions. In fact, these questions and the aforementioned methods to address indication bias are part of the toolkit used by those who conduct comparative effectiveness research. Ultimately, the findings from this study are provocative and represent the best efforts of the authors with the data and statistical techniques at hand to evaluate the intended and unintended consequences of stress ulcer prophylaxis in the intensive care unit. As the authors note in their discussion, a recent systematic review and meta-analysis7 came to the exact opposite conclusions of their study. While well-designed randomized clinical trials remain the gold standard for determining efficacy, observational cohort studies provide key insights into clinical effectiveness in real-world settings. The primary challenges to interpreting the results of MacLaren and colleagues4 is the absence of a clear biologic model to explain the results and the potential for persistent indication bias away from the null hypothesis. Additional studies will be needed to determine if the increased risk of pneumonia, CDI, and acute GI hemorrhage in mechanically ventilated patients in this study is due to the administration of a PPI instead of an H2RA or a result of residual indication bias.
Division of Pulmonary Medicine, Department of Pediatrics, Seattle Children’s Hospital, Seattle, Washington (Goss).
jamainternalmedicine.com
Corresponding Author: Christopher H. Goss, MD, MSc, Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
575
Research Original Investigation
H2 Receptor Antagonists vs Proton Pump Inhibitors
Washington Medical Center, 1959 NE Pacific, Campus Box 356522, Seattle, WA 98195 ([email protected]). Published Online: February 17, 2014. doi:10.1001/jamainternmed.2013.13509. Conflict of Interest Disclosures: None reported. Funding/Support: Dr Goss is supported by grants 1R01HL103965, 1R41HL098985, P30DK089507, and P30ES0070333 from the National Institutes of Health; grant 1R01FD003704 from the Food and Drug Administration; and the CF Foundation. Dr Kim is supported by grant HL07287 from the National Institutes of Health. Role of the Sponsor: The sponsors had no role in the preparation, review, or approval of the manuscript or the decision to submit the manuscript for publication.
576
REFERENCES 1. ASHP Therapeutic Guidelines on Stress Ulcer Prophylaxis. ASHP Commission on Therapeutics and approved by the ASHP Board of Directors on November 14, 1998. Am J Health Syst Pharm. 1999;56(4):347-379. 2. Dial S, Delaney JA, Barkun AN, Suissa S. Use of gastric acid-suppressive agents and the risk of community-acquired Clostridium difficile-associated disease. JAMA. 2005;294(23):2989-2995. 3. Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA. 2004;292(16): 1955-1960. 4. MacLaren R, Reynolds PM, Allen RR. Histamine-2 receptor antagonists vs proton pump inhibitors on gastrointestinal tract hemorrhage and infectious complications in the intensive care
unit [published online February 17, 2014]. JAMA Intern Med. doi:10.1001 /jamainternmed.2013.14673. 5. Lawson EH, Louie R, Zingmond DS, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg. 2012;256(6):973-981. 6. Bosco JL, Silliman RA, Thwin SS, et al. A most stubborn bias: no adjustment method fully resolves confounding by indication in observational studies. J Clin Epidemiol. 2010;63(1):64-74. 7. Alhazzani W, Alenezi F, Jaeschke RZ, Moayyedi P, Cook DJ. Proton pump inhibitors versus histamine 2 receptor antagonists for stress ulcer prophylaxis in critically ill patients: a systematic review and meta-analysis. Crit Care Med. 2013;41 (3):693-705.
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
jamainternalmedicine.com
H2 Receptor Antagonists vs Proton Pump Inhibitors
10. Barkun AN, Bardou M, Pham CQ, Martel M. Proton pump inhibitors vs. histamine 2 receptor antagonists for stress-related mucosal bleeding prophylaxis in critically ill patients: a meta-analysis. Am J Gastroenterol. 2012;107(4):507-520. 11. Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA. 2004;292(16): 1955-1960. 12. Gulmez SE, Holm A, Frederiksen H, Jensen TG, Pedersen C, Hallas J. Use of proton pump inhibitors and the risk of community-acquired pneumonia: a population-based case-control study. Arch Intern Med. 2007;167(9):950-955. 13. Sarkar M, Hennessy S, Yang YX. Proton-pump inhibitor use and the risk for community-acquired pneumonia. Ann Intern Med. 2008;149(6): 391-398. 14. Eom CS, Jeon CY, Lim JW, Cho EG, Park SM, Lee KS. Use of acid-suppressive drugs and risk of pneumonia: a systematic review and meta-analysis. CMAJ. 2011;183(3):310-319. 15. Herzig SJ, Howell MD, Ngo LH, Marcantonio ER. Acid-suppressive medication use and the risk for hospital-acquired pneumonia. JAMA. 2009;301(20):2120-2128. 16. Kwok CS, Arthur AK, Anibueze CI, Singh S, Cavallazzi R, Loke YK. Risk of Clostridium difficile infection with acid suppressing drugs and antibiotics: meta-analysis. Am J Gastroenterol. 2012;107(7):1011-1019. 17. Howell MD, Novack V, Grgurich P, et al. Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection. Arch Intern Med. 2010;170(9):784-790. 18. Safdar N, Dezfulian C, Collard HR, Saint S. Clinical and economic consequences of ventilator-associated pneumonia: a systematic review. Crit Care Med. 2005;33(10):2184-2193. 19. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ. Estimating the proportion of healthcare-associated infections that are reasonably preventable and the related
mortality and costs. Infect Control Hosp Epidemiol. 2011;32(2):101-114. 20. Burgmann H, Hiesmayr JM, Savey A, Bauer P, Metnitz B, Metnitz PG. Impact of nosocomial infections on clinical outcome and resource consumption in critically ill patients. Intensive Care Med. 2010;36(9):1597-1601. 21. Kollef MH, Hamilton CW, Ernst FR. Economic impact of ventilator-associated pneumonia in a large matched cohort. Infect Control Hosp Epidemiol. 2012;33(3):250-256. 22. Lipp MJ, Nero DC, Callahan MA. Impact of hospital-acquired Clostridium difficile. J Gastroenterol Hepatol. 2012;27(11):1733-1737. 23. Ghantoji SS, Sail K, Lairson DR, DuPont HL, Garey KW. Economic healthcare costs of Clostridium difficile infection: a systematic review. J Hosp Infect. 2010;74(4):309-318. 24. McGlone SM, Bailey RR, Zimmer SM, et al. The economic burden of Clostridium difficile. Clin Microbiol Infect. 2012;18(3):282-289. 25. Streiner DL, Norman GR. The pros and cons of propensity scores. Chest. 2012;142(6):1380-1382. 26. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424.
Campaign: international guidelines for management of severe sepsis and septic shock, 2012. Intensive Care Med. 2013;39(2):165-228. 31. Cook D, Heyland D, Griffith L, Cook R, Marshall J, Pagliarello J; Canadian Critical Care Trials Group. Risk factors for clinically important upper gastrointestinal bleeding in patients requiring mechanical ventilation. Crit Care Med. 1999;27(12):2812-2817. 32. Cook DJ, Fuller HD, Guyatt GH, et al; Canadian Critical Care Trials Group. Risk factors for gastrointestinal bleeding in critically ill patients. N Engl J Med. 1994;330(6):377-381. 33. Brown RB, Klar J, Teres D, Lemeshow S, Sands M. Prospective study of clinical bleeding in intensive care unit patients. Crit Care Med. 1988;16(12):1171-1176. 34. Cochard JF, Leger A, Pinaquy C, et al. Gastrointestinal bleeding in trauma patients: incidence and risk factors [abstract]. Intensive Care Med. 1997;23:S140. 35. Schuster DP, Rowley H, Feinstein S, McGue MK, Zuckerman GR. Prospective evaluation of the risk of upper gastrointestinal bleeding after admission to a medical intensive care unit. Am J Med. 1984;76(4):623-630.
27. Herzig SJ, Vaughn BP, Howell MD, Ngo LH, Marcantonio ER. Acid-suppressive medication use and the risk for nosocomial gastrointestinal tract bleeding. Arch Intern Med. 2011;171(11):991-997.
36. Huang J, Cao Y, Liao C, Wu L, Gao F. Effect of histamine-2-receptor antagonists versus sucralfate on stress ulcer prophylaxis in mechanically ventilated patients: a meta-analysis of 10 randomized controlled trials. Crit Care. 2010;14(5):R194.
28. 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(4):760-765.
37. Messori A, Trippoli S, Vaiani M, Gorini M, Corrado A. Bleeding and pneumonia in intensive care patients given ranitidine and sucralfate for prevention of stress ulcer: meta-analysis of randomised controlled trials. BMJ. 2000;321(7269):1103-1106.
29. Levy MJ, Seelig CB, Robinson NJ, Ranney JE. Comparison of omeprazole and ranitidine for stress ulcer prophylaxis. Dig Dis Sci. 1997;42(6):1255-1259.
38. Jhung MA, Banerjee SN. Administrative coding data and health care-associated infections. Clin Infect Dis. 2009;49(6):949-955.
30. Dellinger RP, Levy MM, Rhodes A, et al; Surviving Sepsis Campaign Guidelines Committee including The Pediatric Subgroup. Surviving Sepsis
Invited Commentary
Unintended Consequences of Therapy in the Intensive Care Unit Richard Kim, MD; Christopher H. Goss, MD, MS
Gastric acidity suppression is recommended for critically ill patients,1 with strong clinical evidence for efficacy, particularly in mechanically ventilated patients. Acid-suppressive agents such as proton pump inhibitors (PPIs) and histamine-2 receptor antagonists (H2RA) are thought to decrease the Related article page 564 risk of acute gastrointestinal tract (GI) hemorrhage from stress-induced ulcers. However, the suppression of gastric acidity has been shown in several observational studies to lead to higher rates of pneumonia and Clostridium difficile infection (CDI) in settings other than the 574
intensive care unit (ICU).2,3 It is unclear whether these findings extend to the ICU, where nosocomial infections entail considerable mortality and costs. In this issue of the journal, MacLaren and colleagues4 performed a large retrospective cohort study using billing data on mechanically ventilated patients who were administered a PPI or H2RA for at least 48 hours while intubated. For their main outcomes, they measured not only the number of pneumonia and CDI events but also the intended primary effect of acidsuppressive therapy: the number of GI hemorrhage events occurring more than 48 hours after initiation of mechanical
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
jamainternalmedicine.com
H2 Receptor Antagonists vs Proton Pump Inhibitors
Original Investigation Research
ventilation. Because there were not enough patients who received no acid-suppressive therapy, they compared those who received PPI with those who received H2RA. They hypothesized that the more potent acid suppression effects of PPI would translate into fewer GI bleeding events compared with patients receiving H2RA prophylaxis but a higher number of pneumonia and CDI cases. Adjustments were made using a number of covariates derived from the billing data, as well as a propensity score for treatment assignment. Consistent with their hypothesis, they found that pneumonia and CDI were more frequent in those receiving a PPI compared with those receiving an H2RA. However, they unexpectedly found that the risk of GI hemorrhage was higher in those receiving a PPI. An analysis of other risk factors associated with GI hemorrhage events found that acute or chronic hepatic injury and coagulopathy were associated with increased risk, while thrombocytopenia and use of platelet inhibitors was protective. Increased duration of mechanical ventilation was associated with an increased risk of both pneumonia and CDI, and use of carbapenems and piperacillin was associated with increased risk of CDI, as one might expect. One of the challenges with this analysis is that it uses a large administrative claims data set, where outcomes are inferred using International Classification of Diseases, Ninth Revision, Clinical Modification codes. The possibility of misclassification is suggested by the fact that the frequency of their primary outcomes (pneumonia, CDI, and GI hemorrhage) are strikingly high—in the case of pneumonia, an overall rate of 34.2%. A previous study by Lawson and colleagues5 comparing administrative claims data with medical chart review have found significant limitations in the accuracy of identifying surgical complications. To address this, the authors used probabilistic modeling of potential misclassification of the outcome measures and covariates as a sensitivity analysis. Their Monte Carlo simulation addressed the effect of random variance by allowing parameters to vary by ±25%. However, Lawson and colleagues5 found that the percentage of recorded pneumonia complications as determined by claims data that were later determined to be false positive by medical chart review was 63%, suggesting that even larger bounds for the parameter misclassification in the sensitivity analysis may be warranted to support the conclusions in this study. Another threat to the study’s conclusions is the possibility of indication bias (also termed confounding by indication)—a risk that has been clearly acknowledged by MacLaren and colleagues.4 The concern is that the outcomes of one group of patients (those who received PPI) are confounded by the fact they may have been sicker or otherwise predisposed to the study outcomes compared with those administered an alternative therapy (H2RA). In fact, indication bias provides a unifying explanation for why patients administered a PPI were at higher risk for all the end points noted. The authors attempt to address this and other potential confounders by adjusting for covariates in
ARTICLE INFORMATION Author Affiliations: Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of Washington, Seattle (Kim, Goss);
their regression model, as well as generating a propensity score. Propensity score matching, in principle, is superior to traditional adjustment by creating pseudorandomization of measured confounders. MacLaren and colleagues4 performed covariate adjustment using the propensity score (based on the probability of receiving H2RA vs PPI) in their multivariate analysis, as well as a separate propensity-matched analysis. After matching, the authors found that the propensity score was successful in generating 2 populations with very similar clinical and hospital characteristics, and their results remained robust. Other methods have been developed to address the problem of confounding by indication in observational studies. These include the use of an instrumental variable and restriction. Instrumental variables, widely used in econometrics, require the presence of a suitable variable (or combination of variables) that is strongly associated with the treatment but not the outcome, which may not have been available in this data set. Furthermore, the statistical literature suggests that none of the aforementioned methods are fully successful in removing indication bias.6 It should be noted that the potential for misclassification and indication bias exists in any retrospective study of administrative claims data. One should not discount the findings of MacLaren and colleagues4 based solely on the challenges of addressing these limitations. After selecting appropriate methods for addressing indication bias and misclassification, one should ask the following additional questions to infer causality of the association noted: (1) Is the effect biologically plausible; (2) Is the effect large; (3) Is the anticipated bias in the direction of the null hypothesis; (4) Does the risk factor precede the outcomes? The ability to answer yes to these questions adds greater confidence to the study conclusions. In fact, these questions and the aforementioned methods to address indication bias are part of the toolkit used by those who conduct comparative effectiveness research. Ultimately, the findings from this study are provocative and represent the best efforts of the authors with the data and statistical techniques at hand to evaluate the intended and unintended consequences of stress ulcer prophylaxis in the intensive care unit. As the authors note in their discussion, a recent systematic review and meta-analysis7 came to the exact opposite conclusions of their study. While well-designed randomized clinical trials remain the gold standard for determining efficacy, observational cohort studies provide key insights into clinical effectiveness in real-world settings. The primary challenges to interpreting the results of MacLaren and colleagues4 is the absence of a clear biologic model to explain the results and the potential for persistent indication bias away from the null hypothesis. Additional studies will be needed to determine if the increased risk of pneumonia, CDI, and acute GI hemorrhage in mechanically ventilated patients in this study is due to the administration of a PPI instead of an H2RA or a result of residual indication bias.
Division of Pulmonary Medicine, Department of Pediatrics, Seattle Children’s Hospital, Seattle, Washington (Goss).
jamainternalmedicine.com
Corresponding Author: Christopher H. Goss, MD, MSc, Division of Pulmonary and Critical Care Medicine, Department of Medicine, University of
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
575
Research Original Investigation
H2 Receptor Antagonists vs Proton Pump Inhibitors
Washington Medical Center, 1959 NE Pacific, Campus Box 356522, Seattle, WA 98195 ([email protected]). Published Online: February 17, 2014. doi:10.1001/jamainternmed.2013.13509. Conflict of Interest Disclosures: None reported. Funding/Support: Dr Goss is supported by grants 1R01HL103965, 1R41HL098985, P30DK089507, and P30ES0070333 from the National Institutes of Health; grant 1R01FD003704 from the Food and Drug Administration; and the CF Foundation. Dr Kim is supported by grant HL07287 from the National Institutes of Health. Role of the Sponsor: The sponsors had no role in the preparation, review, or approval of the manuscript or the decision to submit the manuscript for publication.
576
REFERENCES 1. ASHP Therapeutic Guidelines on Stress Ulcer Prophylaxis. ASHP Commission on Therapeutics and approved by the ASHP Board of Directors on November 14, 1998. Am J Health Syst Pharm. 1999;56(4):347-379. 2. Dial S, Delaney JA, Barkun AN, Suissa S. Use of gastric acid-suppressive agents and the risk of community-acquired Clostridium difficile-associated disease. JAMA. 2005;294(23):2989-2995. 3. Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA. 2004;292(16): 1955-1960. 4. MacLaren R, Reynolds PM, Allen RR. Histamine-2 receptor antagonists vs proton pump inhibitors on gastrointestinal tract hemorrhage and infectious complications in the intensive care
unit [published online February 17, 2014]. JAMA Intern Med. doi:10.1001 /jamainternmed.2013.14673. 5. Lawson EH, Louie R, Zingmond DS, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg. 2012;256(6):973-981. 6. Bosco JL, Silliman RA, Thwin SS, et al. A most stubborn bias: no adjustment method fully resolves confounding by indication in observational studies. J Clin Epidemiol. 2010;63(1):64-74. 7. Alhazzani W, Alenezi F, Jaeschke RZ, Moayyedi P, Cook DJ. Proton pump inhibitors versus histamine 2 receptor antagonists for stress ulcer prophylaxis in critically ill patients: a systematic review and meta-analysis. Crit Care Med. 2013;41 (3):693-705.
JAMA Internal Medicine April 2014 Volume 174, Number 4
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: http://archinte.jamanetwork.com/ by a Fudan University User on 05/17/2015
jamainternalmedicine.com
