Erythromycin prior to endoscopy for acute upper gastrointestinal haemorrhage: a cost-effectiveness analysis

Transcription

1 Alimentary Pharmacology & Therapeutics Erythromycin prior to endoscopy for acute upper gastrointestinal haemorrhage: a cost-effectiveness analysis N. S. WINSTEAD & C. M. WILCOX Gastroenterology and Hepatology, University of Alabama-Birmingham, Birmingham, AL, USA Correspondence to: Dr N. Winstead, 703 S. 19th Street, Birmingham, Alabama, USA twinstead@uab.edu Publication data Submitted 26 July 2007 First decision 16 August 2007 Resubmitted 5 September 2007 Accepted 6 September 2007 SUMMARY Background Erythromycin is a potent stimulator of gastrointestinal motility. Recent studies have examined the use of intravenous erythromycin to clear the stomach of blood before oesophago-gastroduodenoscopy (EGD) for acute upper gastrointestinal haemorrhage (UGIH). These studies have shown clinical effectiveness. Aim To evaluate the cost-effectiveness of this intervention. Methods We sought to determine the cost-effectiveness of erythromycin before EGD from the payer s perspective. We found three relevant studies of erythromycin and used these data for the analysis. We obtained costs for intravenous erythromycin and charges for peptic ulcer hospitalization, EGD, surgery, and angiographic embolization. Complication rates were also incorporated from the literature. We implemented a model of health-related quality of life to measure the impact of the intervention. We created a decision-analysis tree and performed a probabilistic sensitivity analysis. Results A strategy of erythromycin prior to EGD resulted in a cost-effective outcome in a majority of trials using willingness-to-pay figures of $0, $ and $ (US) per quality-adjusted life-year (QALY). Conclusion Because of the implications for cost saving and increase in QALY, we would recommend giving erythromycin prior to EGD for UGIH. Aliment Pharmacol Ther 26, ª 2007 The Authors 1371 doi: /j x

2 1372 N. S. WINSTEAD AND C. M. WILCOX INTRODUCTION Upper gastrointestinal bleeding is a common problem with an estimated incidence of episodes per people. 1 3 Despite recent advances in endoscopic therapy and acid-suppressing medications, the estimated direct costs in the USA are over 1 billion dollars (US) annually. 2, 4 In 2004, the most recent year for which data are available, the Healthcare Utilization Project National Inpatient Sample (HCUP NIS) database showed a mortality of 2.7% for all stomach and duodenal ulcer diseases classified under ICD-9 codes. 5 Urgent endoscopy has reduced mortality and cost, particularly in variceal haemorrhage. 6, 7 However, one challenge of urgent endoscopy is the presence of retained blood in the stomach, which may interfere with endoscopic evaluation. In patients with retained clot in the fundus, lesions may be missed up to 41% of the time. 8 Missed lesions at endoscopy are associated with increased morbidity and mortality, increased need for surgery, increased length-of-stay and associated costs While gastric lavage may reduce the interference of retained clot with endoscopy, performance of gastric lavage is onerous and time-consuming and had no impact on the need for repeat endoscopy in one study. 13 The macrolide antibiotic, erythromycin, is motilin receptor agonist, and thus a potent stimulator of gastrointestinal motility Given this effect, some case studies and three recent clinical trials have evaluated the use of erythromycin prior to endoscopy to clear the stomach of retained blood and clots These studies indicated that erythromycin might be useful in the setting of urgent endoscopy for acute upper gastrointestinal haemorrhage. Because erythromycin is generic and inexpensive, it is possible that this intervention would be cost-effective or cost-saving because of the decreased need for repeat endoscopy and possibly decreased rates of surgical interventions; however, the cost-effectiveness of this intervention has not been rigorously evaluated. METHODS Using TreeAge Pro software, we developed a model of upper GI bleeding into which patients entered one of two competing strategies erythromycin infusion before endoscopy for acute upper gastrointestinal haemorrhage or endoscopy without erythromycin infusion. The model accounts for patients during their hospitalization and one year of follow-up because most outcomes are observed in a short time period (Figure 1). We performed a PubMed search in 2006 for relevant randomized, controlled trials of erythromycin prior to No further endoscopy Surgery for bleeding or perforation psurgery UGI haemorrhage EGD with erythromycin premedication Angiographic embolization pangio Second endoscopy pery2nd p No further endoscopy Recovery p surg EGD without erythromycin premedication Surgery for bleeding or perforation psurgery Angiographic embolization pangio Second endoscopy pnoery2nd p Recovery p surg Figure 1. The model specification.

3 COST-EFFECTIVENESS OF ERYTHROMYCIN BEFORE ENDOSCOPY 1373 endoscopy for acute upper gastrointestinal haemorrhage in the English language medical literature. We used the search terms erythromycin combined with either endoscopy or haemorrhage. References from identified articles were also searched. Our original search identified 85 articles, in which we found three relevant randomized, controlled trials with a common outcome of need for second- endoscopy One randomized, controlled trial was excluded because it did not include this outcome. 25 One randomized, controlled trial showing benefit was excluded because it was published in Polish-language and could not be critically evaluated. 26 All articles included need for second- endoscopy as either a primary or secondary endpoint, and these data were pooled and analysed (Table 1). This pooled analysis showed a significant advantage to erythromycin infusion prior to endoscopy over placebo. We also performed a systematic review of the literature to obtain base-case estimates for our model for the incidence of the various outcomes (Table 2). In cases where limited peer-reviewed data were available, we obtained estimates from the HCUP NIS database. 5 The HCUP NIS database is a collection of discharge data from 1004 hospitals in 37 states, including academic, public, private, and notfor-profit hospitals and comprises 90% of hospital discharges in the United States for Quality-adjusted life-years (QALYs) were adopted from Spiegel s model of health-related quality of life 27, 28 (HRQOL) in upper GI bleeding. Length-of-stay data were validated against the NIS database. 5 The QALY estimates in Spiegel s model were derived from previously published research examining the healthrelated quality of life in uninvestigated mild, moderate, and severe dyspepsia and ulcers requiring hospitalization using time-trade off interviews 29 of patients with dyspepsia See Table 3 for disutilities used in the model. Cost estimates for our model were derived from the 2006 Medicare fee schedule for hospitalizations and procedures. 33 In the case of endoscopy charges, we used the charges for therapeutic endoscopy. Costs for Table 1. Pooled analysis of the need for second- endoscopy in the three randomized, controlled trials of intravenous erythromycin prior to upper gastrointestinal endoscopy for acute upper gastrointestinal haemorrhage Erythromycin Second No second Total Placebo Second No second Total Frossard Coffin Carbonell Totals Proportions pooled p pooled Table 2. Base-case estimates and ranges for variables used in the model Variable Base-case estimate (%) Range in literature (%) Range tested in analysis (%) References Probability of second- endoscopy with erythromycin Probability of second- endoscopy without erythromycin Probability of surgery for perforation , 5, 38 or uncontrollable bleeding Probability of angiography and embolization 0.39 N A for uncontrollable bleeding Probability of death , 5, 39, 40 Probability of death after surgery for perforation or uncontrollable bleeding , 5, 39, 40

4 1374 N. S. WINSTEAD AND C. M. WILCOX Table 3. Disutilities of health states and ranges of disutilities used in the model Health state Disutility of health state (QALY = 1 - Disutility) Range tested in analysis References Uncomplicated peptic ulcer with 4-day hospital , 28, stay and outpatient recovery Complicated peptic ulcer with 8-day hospital , 28, stay and outpatient recovery Peptic ulcer with surgery for perforation or bleeding , 28, (bleeding or postsurgery) 1 N A N A IV erythromycin were obtained from our hospital pharmacy ($5.85 (US) per 500 mg vial, wholesale) and from a time and motion study in the medical literature ($30.55 (US) per 500 mg vial infusion). 34 When necessary, costs were inflated to the levels of 2006 US dollars using year-over-year gross domestic product (GDP) growth indices and using exchange rate conversions at December 2006 levels. Because of the wide variation in cost for IV erythromycin we used the entire range in our model. This creates a model which is intentionally biased away from erythromycin use because from the payer s standpoint, there would not be additional charges for nursing labor, IV tubing, the IV pump, etc. since these costs would be covered under the diagnosis-related group (DRG) charge for the hospitalization (Table 4). RESULTS Using our model and the base-case estimates, we performed a rollback analysis that showed the strategy of infusion of IV erythromycin before endoscopy to be both less expensive and resulting in higher QALYs (cost savings of $486 (US), QALYs gained ). We also performed a sensitivity analysis on all variables (tornado analysis) and found our model to be sensitive only to rates of repeat endoscopy and charges for uncomplicated and complicated peptic ulcer. In one-way sensitivity analysis, a strategy of erythromycin before endoscopy for upper gastrointestinal haemorrhage was cost-effective when the rate of second- endoscopy in the erythromycin strategy was less than 0.29 and the rate of second- endoscopy in the no-pretreatment strategy was greater than It was also cost-effective when the charges for uncomplicated peptic ulcer were less than $8000 (US) and the charges for complicated peptic ulcer were greater than $6000 (US). Because of these results, we conducted a Monte Carlo simulation using 1000 simulated patients. With each pass through the model, the incidence and costs of various outcomes were varied randomly throughout their specified range. This enables modeling of the variation in reported values of any of these variables. Cost Base-case estimate Range tested in analysis DRG for UGI haemorrhage $ $ Erythromycin 500 mg IV $30.55 $ Cost of EGD Consultation fee $ Endoscopy fee (including haemostasis) $ Total $ $100-$ Surgery for bowel perforation or bleeding Consultation fee $ Surgery fee $ DRG for perforation $ Total $ $ DRG for complicated peptic ulcer $ $ Angiography and embolization $ $ References Table 4. Base-case estimates of charges and ranges of charges (in US dollars) used in the model

5 COST-EFFECTIVENESS OF ERYTHROMYCIN BEFORE ENDOSCOPY Incremental cost (US $) Incremental QALYs Incremental cost effectiveness ratios WTP = $50,000 (US)/QALY WTP = $100,000 (US)/QALY Figure 2. Incremental cost-effectiveness graph and willingness-to-pay from the probabilistic sensitivity analysis. Interventions which lay below the willingness-to-pay (WTP) line and to the right of the y-axis are considered cost-effective as they either result in improved QALYs at an acceptable WTP (in cost QALY) or cost-saving (if they lay below the x-axis). Interventions which lay below the WTP line and to the left of the y-axis result in cost-savings and decreased QALYs, but at a rate which is acceptable based upon the WTP (thought this may or may not be desirable). Interventions which lay above the WTP line and to the right of the y-axis result in improved QALYs but cost more than the WTP. Interventions which lay above the WTP line and to the left of the y-axis result in lower QALYs and either increased cost (if above the x-axis) or cost-savings which are unacceptable based upon the WTP (if below the x-axis). The Monte Carlo simulation results are presented in Figure 2 and Table 5 for a variety of willingnessto-pay figures. Under all scenarios, the strategy of infusing IV erythromycin prior to oesophago-gastroduodenoscopy results in cost savings or increases QA- LYs at a cost less than the willingness-to-pay. Further, under any willingness-to-pay scenario, the erythromycin strategy results in a cost-savings 63.7% of the time and reduces the need for repeat endoscopy from an Table 5. Percentage of trials resulting in cost-saving, cost-effective and cost ineffective outcomes for three WTP figures WTP (US $ QALY) Costsaving (%) Cost effective (cost WTP) (%) Cost ineffective (cost > WTP) (%) average of 1.33 endoscopies per patient to an average of 1.18 endoscopies per patient. DISCUSSION The concept of QALYs is important in cost-effectiveness research, as it allows researchers to agree on a common outcome or common denominator when comparing competing strategies. The QALY model has been used in a number of acute and chronic diseases and is widely accepted by health systems researchers. In our model, the strategy of infusion of IV erythromycin prior to endoscopy for acute upper gastrointestinal haemorrhage appears to result in modest cost savings and increase in QALYs at a willingness to pay of $ (US) per QALY. These results were in spite of the model being intentionally biased away from the erythromycin strategy by using an inflated estimate of the cost of erythromycin. We believe that these results are particularly important because of the potential for reducing repeat endoscopies and morbidity and mortality. An additional reason to consider infusion of erythromycin

6 1376 N. S. WINSTEAD AND C. M. WILCOX prior to endoscopy is the low rate of overall side effects or adverse outcomes from a single intravenous dose. There is a theoretical risk of causing vomiting and aspiration, but no case reports of this adverse reaction could be found so it was not included in our model. There are also reports of cardiac arrhythmias (torsade des pointes and ventricular tachycardia in particular) when erythromycin is given in conjunction with other drugs; however, these risks can be mitigated by taking a careful history of medication from the patients prior to administration of erythromycin. 35, 36 We therefore also did not include this adverse event in our model. Additional possible side effects of erythromycin include altered drug metabolism of other drugs, abdominal pain, diarrhoea, and seizure. All of these adverse events are uncommon, and accordingly they were also not included in our model. Our study does have several limitations. First, our estimates of need for repeat endoscopy in the two strategies are based on studies that included significant proportions of patients with portal hypertensive gastrointestinal haemorrhage. Because many times one cannot predict the etiology of gastrointestinal bleeding prior to endoscopy, we feel that these estimates are useful and appropriate for our model. Second, our study is being performed from the payer s perspective. While Medicare fee schedules were used for cost estimation, it may be appropriate to apply a cost-tocharge ratio to the cost estimates before using them in the model. However, cost-to-charge ratios may vary between hospitals and between payers, therefore no such ratio was applied to our model. Further, for an individual patient, it is likely that the cost-to-charge ratio would remain uniform for all charges incurred during a single hospital stay. Third, utilities (QALYs) were derived from time trade off interviews of patients with chronic dyspepsia rather than patients with acute upper gastrointestinal haemorrhage. It has been shown that patients with chronic illness may rate their health state higher than a person without that illness. 37 Further, patients who experience acute peptic ulcer may experience higher disutility from that health state than a patient who lives with dyspepsia chronically. These factors may however, have led to an under-estimation of the cost-effectiveness of this intervention. Finally, this complex model may fail to capture all the slight variations in patterns of care among different hospitals, different patients, and counties outside of the United States. We have attempted to mitigate this limitation by including broad, conservative ranges of values in our sensitivity analyses and Monte-Carlo simulation. We have also used data from the NIS database, which includes a range of academic, public, private, and not-for-profit hospitals across the United States of America. Unfortunately, this is a limitation of cost-effectiveness research in general. CONCLUSION Because of the modest cost savings, favourable incremental cost-effectiveness ratio, and improved QALYs among patients receiving erythromycin infusion prior to endoscopy for acute upper gastrointestinal haemorrhage, we would recommend this intervention to be made a standard practice. Because of the occasional difficulty in obtaining medications from the pharmacy promptly (particularly at odd hours), it may be useful to consider storing erythromycin on the emergency endoscopy cart or in the endoscopy lab. We would also propose developing a similar model for portal hypertensive upper gastrointestinal haemorrhage. ACKNOWLEDGEMENTS Declaration of personal interests: None. Declaration of funding interests: This study was funded in part by US Agency for Healthcare Research and Quality, grant number 5 T32 HS REFERENCES 1 Gilbert DA. Epidemiology of upper gastrointestinal bleeding. Gastrointest Endosc 1990; 5(Suppl): S Longstreth GF. Epidemiology of hospitalization for acute upper gastrointestinal hemorrhage: a population-based study. Am J Gastroenterol 1995; 90: Rockall TA, Logan RF, Devlin HB, Northfield TC. Incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. Steering Committee and members of the National Audit of Acute Upper Gastrointestinal Haemorrhage. Bmj 1995; 311: Gralnek IM, Jensen DM, Kovacs TO, et al. An economic analysis of patients with active arterial peptic ulcer hemorrhage treated with endoscopic heater probe, injection sclerosis, or surgery in a prospective, randomized trial. Gastrointest Endosc 1997; 46: AHRQ. HCUPnet. Available from: http: hcupnet.ahrq.gov [accessed November 2006].

7 COST-EFFECTIVENESS OF ERYTHROMYCIN BEFORE ENDOSCOPY Chalasani N, Kahi C, Francois F, et al. Improved patient survival after acute variceal bleeding: a multicenter, cohort study. Am J Gastroenterol 2003; 98: El-Serag HB, Everhart JE. Improved survival after variceal hemorrhage over an 11-year period in the Department of Veterans Affairs. Am J Gastroenterol 2000; 95: Stollman NH, Putcha RV, Neustater BR, Tagle M, Raskin JB, Rogers AI. The uncleared fundal pool in acute upper gastrointestinal bleeding: implications and outcomes. Gastrointest Endosc 1997; 46: Cheng CL, Lee CS, Liu NJ, Chen PC, Chiu CT, Wu CS. Overed lesions at emergency endoscopy for acute nonvariceal upper gastrointestinal bleeding. Endoscopy 2002; 34: Chung YF, Wong WK, Soo KC. Diagnostic failures in endoscopy for acute upper gastrointestinal haemorrhage. Br J Surg 2000; 87: Pardo A, Durandez R, Hernandez M, et al. Impact of physician specialty on the cost of nonvariceal upper GI bleeding care. Am J Gastroenterol 2002; 97: Quirk DM, Barry MJ, Aserkoff B, Podolsky DK. Physician specialty and variations in the cost of treating patients with acute upper gastrointestinal bleeding. Gastroenterology, 1997; 113: Lee SD, Kearney DJ. A randomized controlled trial of gastric lavage prior to endoscopy for acute upper gastrointestinal bleeding. J Clin Gastroenterol 2004; 38: Peeters T, Matthijs G, Depoortere I, Cachet T, Hoogmartens J, Vantrappen G. Erythromycin is a motilin receptor agonist. Am J Physiol 1989; 1: G Xu L, Depoortere I, Vertongen P, Waelbroeck M, Robberecht P, Peeters TL. Motilin and erythromycin-a share a common binding site in the third transmembrane segment of the motilin receptor. Biochem Pharmacol, 2005; 70: Peeters TL. Erythromycin and other macrolides as prokinetic agents. Gastroenterology 1993; 105: Tack J, Janssens J, Vantrappen G, et al. Effect of erythromycin on gastric motility in controls and in diabetic gastroparesis. Gastroenterology 1992; 103: Weber FH, Jr. Richards RD, McCallum RW. Erythromycin: a motilin agonist and gastrointestinal prokinetic agent. Am J Gastroenterol 1993; 88: Lawrence SP, McNally PR. Intravenous erythromycin to clear the stomach of obstructing clot. Gastrointest Endosc 1994; 40: Aubertin JM, Levoir D, Becheur H, Bloch F, Petite JP. Intérêt de l érythromycine intra-veineuse avant gastroscopie dans les hémorragies digestives hautes. Gastroenterol Clin Biol 1995; 19: Nion I, Andant C, Jouet P, Leport J, Soule JC. Intèrít de l èrythromycine intraveineuse dans la prèparation l endoscopie en cas d hêmorragie digestive haute. Gastroenterologie Clinique et Biolique 1998; 22: Carbonell N, Pauwels A, Serfaty L, Boelle PY, Becquemont L, Poupon R. Erythromycin infusion prior to endoscopy for acute upper gastrointestinal bleeding: a randomized, controlled, double-blind trial. Am J Gastroenterol 2006; 101: Coffin B, Pocard M, Panis Y, et al. Erythromycin improves the quality of EGD in patients with acute upper GI bleeding: a randomized controlled study. Gastrointest Endosc 2002; 56: Frossard JL, Spahr L, Queneau PE, et al. Erythromycin intravenous bolus infusion in acute upper gastrointestinal bleeding: a randomized, controlled, double-blind trial. Gastroenterology, 2002; 123: Sears R, Duckworth C, Balaban D, et al. Erythromycin versus gastric lavage for pre-endoscopic preparation of patients with upper gastrointestinal bleeding. Gastrointest Endosc, 1996; 43: Rudzki S, Czekalowski S, Michalak K, Kusz W, Fularz A. Erytromycyna jako czynnik poprawiający jakość badania endoskopowego w ostrych krwawieniach z górnego odcinka przeqodu pokarmowego. Wiadomości Lekarskie 2006; 59: Spiegel BM, Dulai GS, Lim BS, Mann N, Kanwal F, Gralnek IM. The cost-effectiveness and budget impact of intravenous versus oral proton pump inhibitors in peptic ulcer hemorrhage. Clin Gastroenterol Hepatol 2006; 4: Spiegel BM, Ofman JJ, Woods K, Vakil NB. Minimizing recurrent peptic ulcer hemorrhage after endoscopic hemostasis: the cost-effectiveness of competing strategies. Am J Gastroenterol 2003; 98: Torrance GW, Thomas WH, Sackett DL. A utility maximization model for evaluation of health care programs. Health Serv Res 1972; 7: Ebell MH, Warbasse L, Brenner C. Evaluation of the dyspeptic patient: a costutility study. J Fam Pract 1997; 44: Groeneveld PW, Lieu TA, Fendrick AM, et al. Quality of life measurement clarifies the cost-effectiveness of Helicobacter pylori eradication in peptic ulcer disease and uninvestigated dyspepsia. Am J Gastroenterol 2001; 96: Barkun AN, Herba K, Adam V, Kennedy W, Fallone CA, Bardou M. High-dose intravenous proton pump inhibition following endoscopic therapy in the acute management of patients with bleeding peptic ulcers in the USA and Canada: a cost-effectiveness analysis. Aliment Pharmacol Ther 2004; 19: Center for Medicare and Medicaid Services Fee Schedule. Available from: [accessed November 2006]. 34 van Zanten AR, Engelfriet PM, van Dillen K, van Veen M, Nuijten MJ, Polderman KH. Importance of nondrug costs of intravenous antibiotic therapy. Crit Care 2003; 7: R McComb JM, Campbell NP, Cleland J. Recurrent ventricular tachycardia associated with QT prolongation after mitral valve replacement and its association with intravenous administration of erythromycin. Am J Cardiol 1984; 54: Guelon D, Bedock B, Chartier C, Haberer JP. QT prolongation and recurrent torsades de pointes during erythromycin lactobionate infusion. Am J Cardiol 1986; 58: Gold MR. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996; Xxiii: 425p 38 Marmo R, Rotondano G, Piscopo R, Bianco MA, D Angella R, Cipolletta L. Dual therapy versus monotherapy in the endoscopic treatment of high-risk bleeding ulcers: a meta-analysis of controlled trials. Am J Gastroenterol 2007; 102: quiz Katschinski B, Logan R, Davies J, Faulkner G, Pearson J, Langman M. Prognostic factors in upper gastrointestinal bleeding. Dig Dis Sci 1994; 39: Targownik LE, Nabalamba A. Trends in management and outcomes of acute nonvariceal upper gastrointestinal bleeding: Clin Gastroenterol Hepatol 2006; 4: