Inappropriate Use of Intravenous Pantoprazole: Extent of the Problem and Successful Solutions

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3: Inappropriate Use of Intravenous Pantoprazole: Extent of the Problem and Successful Solutions GILAAD G. KAPLAN,* DUANE BATES, DAWN MCDONALD, REMO PANACCIONE,* and JOSEPH ROMAGNUOLO*,, *Division of Gastroenterology, Departments of Medicine, Department of Pharmacy (Calgary Health Region), and Department of Community Health Sciences, University of Calgary, Calgary, Alberta, Canada; and Medical University of South Carolina, Charleston, South Carolina Background & Aims: Indications for intravenous proton pump inhibitors (IV PPI) include upper gastrointestinal bleeding (UGIB) from peptic ulcer disease with high-risk stigmata and patients receiving nothing by mouth (NPO). The objectives were to assess the extent of errors in indications for IV PPI use and to determine whether multidisciplinary interventions could improve IV PPI use and costs. Methods: Part 1: Patients prescribed IV PPI during a period of 4 months were divided into 2 settings, UGIB or non-ugib. The setting-specific appropriateness of the IV PPI indication and dosing regimen was determined. Part 2: Patients prescribed IV PPI before and after multidisciplinary interventions (educating physicians, a computerized dose template, pharmacists altering IV PPI orders in non-ugib patients who were not NPO, and recommending a GI consult when a continuous infusion was ordered) were studied. Incidence of prescribing errors, IV PPI costs, and potential confounders were compared. Results: Part 1: Only 50% of UGIB (n 145) patients received IV PPI for an appropriate indication. Both indication and dosing regimen were appropriate in 21%. In the non-ugib group (n 95), 33% were truly NPO; 51% had a correct dosing frequency. Part 2: The postintervention (n 105) group (vs the preintervention group, n 113) showed a significant absolute reduction in the degree of inappropriate indication in the UGIB (26%; 95% confidence interval [CI], 10% 42%; P <.0001) and in the non-ugib (41%; 95% CI, 24% 58%; P <.0001) subgroups. However, a greater improvement in underspending than overspending meant that overall costs were unchanged. Conclusions: IV PPI was frequently prescribed inappropriately and incorrectly; simple maneuvers resulted in reductions in errors. Upper gastrointestinal bleeding (UGIB) is an important clinical problem accounting for 300,000 hospital admissions annually and an expenditure of 2.5 billion health care dollars per year in the United States. 1 There is Level I evidence from randomized trials showing that patients with peptic ulcer disease (PUD) associated with high-risk stigmata (HRS) treated with high-dose bolus followed by continuous infusion of an IV proton pump inhibitor (PPI) have a reduced frequency of rebleeding. 2 6 A recent multivariate analysis of a large registry of UGIB patients suggested that IV PPI might also reduce mortality in UGIB. 7 IV PPI has also been shown to be an effective substitute in patients who cannot tolerate oral medications but require a PPI for various reasons IV pantoprazole was released on the Canadian market in 1999 and is currently the only approved IV PPI formulation in Canada. Since its addition to the Calgary Health Region adult acute care formulary in April 1999, annual IV PPI expenditures rose 4-fold between the first and third years to nearly half a million dollars (Canadian). 11 In part this increase in spending is believed to be a result of inappropriate use of IV pantaprazole, because 3 North American institutions have shown misuse of this medication A retrospective study was conducted to assess adherence to published indications for the use of IV PPI, as well as appropriate dosing of the drug. Subsequently, a prospective study was carried out to determine whether simple institutional interventions could improve physician prescribing behavior and decrease drug costs. Methods Part 1: Extent of the Problem Consecutive inpatients receiving IV pantoprazole (bolus or infusion) during 4 selected months between 2001 and 2002 at the Foothills Medical Centre (University of Calgary, Alberta, Canada) were identified through the inpatient pharmacy computer database. Two summer months and 2 winter months were chosen without prior knowledge of their respective expenditures. The months were purposefully spaced to Abbreviations used in this paper: CI, confidence interval; HRS, highrisk stigmata; NPO, nothing by mouth; PUD, peptic ulcer disease; PPI, proton pump inhibitor; UGIB, upper gastrointestinal bleeding by the American Gastroenterological Association /05/$30.00 PII: /S (05)

2 1208 KAPLAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 12 allow a determination of the presence of possible time trends in regional usage. Each chart was manually reviewed by one of the investigators by using an a priori developed data collection sheet. The following information was abstracted: indication for IV pantoprazole, dosing regimen and duration of IV pantoprazole, prescribing service, recent myocardial infarction or other significant comorbidity making endoscopy potentially unsafe, previous use of oral PPI, oral or nasogastric diet, medication orders during IV pantoprazole use, gastroenterology consultation, endoscopy results, and length of stay. Because the appropriateness criteria and dosing of IV PPI differ between UGIB patients and other patients given IV PPI, the patients were stratified into 2 subgroups, UGIB and non-ugib (generally a patient who required a PPI but could not take it enterally). Institutional review board approval was granted through the Calgary Health Region Quality Improvement Health Information Group. Upper gastrointestinal bleeding subgroup. Definition of appropriate indication. In the study by Lau et al 2 and in others, patients with an UGIB caused by PUD associated with HRS benefited from high-dose IV PPI infusion. 3,5,6 The presence of HRS was defined as arterial or venous bleeding, nonbleeding visible vessel, or adherent clot. 2,15,16 Starting IV pantoprazole before endoscopy was considered acceptable, as long as it was discontinued within 12 hours of endoscopy when HRS were not confirmed. Empiric IV PPI (up to 72 hours) for patients with a suspected UGIB but determined by the consultant to be at high risk for endoscopy (eg, recent myocardial infarction) was considered acceptable. Furthermore, it was believed that when IV PPI was discontinued within 24 hours, it would not be labeled as inappropriate in this study, because brief diagnostic uncertainty regarding whether significant UGIB is indeed occurring is a real-life phenomenon that clinicians contend with. Definition of appropriate dosing regimen. The appropriate IV pantoprazole dosing regimen included an initial 80-mg bolus followed by an 8 mg/h infusion for 72 hours used by Lau et al 2 ; nearly identical regimens were used by other authors, 2,3,5,6 and this is the regimen supported by studies in healthy volunteers 17 and a recent consensus document. 18 If rebleeding occurred, diagnosed on clinical and/or endoscopic grounds, the patient was allowed to receive IV PPI for an additional 72 hours. Non bleeding subgroup. Definition of appropriate indication and dosing. In the non-ugib subgroup, IV pantoprazole was considered to be indicated when a PPI was judged by the attending physician to be necessary, but the patient was unable to take an oral PPI preparation; the patient could not be concurrently receiving enteral (including oral or enteral tube routes) food or medications. 8,19,20 The non-ugib appropriate dose was defined as a 40-mg IV bolus once daily, 8 unless the patient had an accepted indication for twice daily therapy (eg, failure on once daily therapy). Statistical analysis. Multivariate analysis was used in each of the 2 groups (UGIB and non-ugib) to determine Table 1. Clinical Scenarios and Descriptions of How Corresponding Costs Were Calculated for the Ideal Model (Part 1 of the Study) Clinical scenario Patients with HRS or considered high risk for endoscopy No HRS were identified and IV PPI discontinued within 12 h of endoscopy IV PPI continued for 12 h after endoscopy when no HRS are identified IV PPI started for suspected UGIB, without endoscopy attempted, but discontinued within 24 h IV PPI started for suspected UGIB, without endoscopy, but continued for 24 h Cost calculation for ideal model a Cost of 80-mg bolus and 8 mg/h 72 h PO PPI thereafter Cost of 80-mg bolus and 8 mg/h for time prescribed as ordered PO PPI thereafter Cost of 80-mg bolus and 8 mg/h infusion before endoscopy PO PPI thereafter Cost of IV PPI as ordered Cost of PO PPI a For each clinical scenario, the ideal cost model estimates the cost of IV PPI plus that of the oral PPI that would have been incurred if the physician had ordered the PPI appropriately with the correct dosing regimen. predictors of inappropriate prescribing (logistic regression) within the inappropriate group, correcting for confounding by time period if needed. The variables included in this analysis were age, gender, month observed, prescribing service (modeled as several dummy variables; the reference category was gastroenterology), performance of an endoscopy, time to endoscopy, inpatient or outpatient status, nonsteroidal anti-inflammatory drug use, and time of day of first dose (daytime vs night-time). Transformations were used if indicated by residual plots. Likelihood ratio tests of nested models were used in model building. The trend in monthly costs was evaluated qualitatively and with linear regression to explore for any time trends. Costs. The pharmaceutical cost was based on the Calgary regional formulary cost of a 40-mg IV vial of pantoprazole being $13.70 (Canadian) compared with a 40-mg tablet being $0.45 (Canadian) in hospital patients. Costs were expressed in Canadian dollars. Costs were tabulated for the entire background period. Theoretical or ideal costs were calculated by using the rules in Table 1. Part 2: Effect of Multidisciplinary Intervention Consecutive inpatients prescribed IV pantoprazole (bolus or infusion) at any time during a 60-day period, before and after instituting interventions directed at improving prescribing behavior, were identified through the inpatient pharmacy

3 December 2005 IV PROTON PUMP INHIBITOR USE 1209 computer database. The 2 periods were chosen without prior knowledge of their respective expenditures. The study period was separated by 3 months from the control period to allow time for interventions to work. Each chart was manually reviewed as outlined previously. Stratification of patients into UGIB and non-ugib subgroups and the definitions for appropriateness criteria and dosing of IV PPI were identical to those described in Part 1 of the study. Baseline characteristics of the patients and balance measures, such as the total number of admissions to hospital, the number of UGIB cases presenting to the emergency department, and the total number of GI consults were compared between the 2 study periods to assess for potential confounders in the comparison of the 2 periods. Interventions. The interventions conducted between the control and study periods were 4-fold: (1) a newsletter was sent out educating physicians on the appropriate indications and dosing of IV pantoprazole; (2) a dose template for ordering IV PPI that directed physicians to the correct indication and dosing of the drug was instituted in our computerized inpatient order system including an auto-stop requiring reordering after 72 hours; (3) autosubstitution of IV pantoprazole for the oral dose equivalent was performed by pharmacists if the patient was receiving IV pantoprazole for a Non-UGIB indication but was concurrently receiving either oral/nasogastric feeds or medications (ie, not truly nothing by mouth [NPO]); (4) a gastroenterology consult was recommended when physicians ordered continuous IV pantoprazole infusions, so that an endoscopy would be considered to confirm HRS and perform endoscopic therapy if present. With a phone call to pharmacy, the pharmacy suggestions could be readily overridden by the prescribing physician and the consult request easily cancelled if the referring service disagreed. Costs. The total pharmaceutical costs of administered IV pantoprazole in Canadian dollars were tabulated weekly by the pharmacy department for the preintervention and postintervention periods. Statistical analysis. The 2 test was used to compare proportions with appropriate prescribing during the preintervention and postintervention periods. Differences in average weekly costs were compared with a Wilcoxon rank-sum test. Results Part 1: Extent of the Problem A total of 240 patients were prescribed either a bolus or infusion of IV pantoprazole during the 4 months studied (January 2001, January 2002, July 2002, August 2002). One hundred forty-five (60.4%) were prescribed IV PPI for a suspected or confirmed UGIB, whereas 95 were prescribed PPI for other reasons (non-ugib). The distribution of services (admitting service or consulting service) that initiated the order to prescribe IV pantoprazole in both the UGIB and non-ugib subgroups was the following: (1) UGIB: medicine 28%, hospitalist 23%, critical care 13%, surgery 11%, gastroenterology Table 2. Baseline Characteristics of the Patients Receiving IV Pantoprazole for the Indication of UGIB and non- UGIB subgroups Characteristic Mean (standard deviation), median (range) or % UGIB Age (y) 65 (17) Gender 62% male History of recent 34% NSAID use Gastroenterology 75% consult Endoscopy performed 67% Inpatient when bleed 52% commenced Length of stay (days) 9 (1 93) Time to endoscopy (h) 12.6 (2.5 80) Non-UGIB Age (y) 57 (16) Gender 59% male Gastroenterology 14% consult History of oral PPI use 33% Length of stay (days) 14 (3 133) 9%, cardiology 6%, and other services 10%; (2) non- UGIB: surgery 40%, hematology-oncology 26%, hospitalist 13%, cardiology 10%, critical care 6%, and other services 5%. Table 2 demonstrates the baseline characteristics of the patients assessed during the study period. Upper gastrointestinal bleeding subgroup. IV pantoprazole was withheld until endoscopy confirmed PUD with HRS in only 5.5% of cases. Of the 145 people who received the drug, 67% (97/145) underwent endoscopy. Of the 48 patients who did not undergo endoscopy, 54% (26/48) continued to receive the medication for more than 24 hours. Of note, in 78% of these 48 cases, a gastroenterologist was not consulted. Forty-seven percent (46/97) of those who underwent endoscopy continued to receive IV pantoprazole after the endoscopy, despite the absence of HRS. The remainder either had HRS (26%) or had IV pantoprazole discontinued appropriately after the endoscopy did not confirm HRS (27%). Overall, in 72 (50%) of the 145 cases, the IV PPI was determined to have been ordered for an inappropriate indication (Figure 1). When IV pantoprazole was used in the UGIB group, physicians correctly ordered the initial 80-mg bolus in only 44% of cases. The correct infusion (8 mg/h) was used in 67% of the cases. The remainder ordered pantoprazole 40 mg IV twice daily instead of a continuous infusion protocol, despite the fact that the indication for ordering the PPI was an UGIB. In only 30% of the cases was the duration of IV pantoprazole literaturebased. 2,3,5,6 All 3 components (bolus, infusion, and duration) of the dosing regimen were correct in only 25%

4 1210 KAPLAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 12 Figure 1. Distribution of patients with suspected or confirmed UGIB who were prescribed IV pantoprazole appropriately (white boxes) or inappropriately (shaded boxes) in part 1 of the study. Percentages in parentheses are with respect to the total of the box immediately preceding. PZ, pantoprazole. HRS, high-risk stigmata. of the cases. The drug was ordered appropriately, in terms of indication and all 3 components of the correct dosing regimen, in only 21% (95% CI, 14% 27%) of the cases. During the time period studied, there was no significant time trend in the proportion of patients with an appropriate indication and/or dose. Non bleeding subgroup. Of the 95 patients who were prescribed IV pantoprazole because they required a PPI that they could not take orally, only 33% were truly NPO, and only half (51%) were prescribed the correct dosing frequency. Physicians prescribed IV pantoprazole for an appropriate indication and at the correct dosing frequency in only 14% (95% CI, 7% 22%) of these patients. The proportion of patients with an appropriate indication and/or dose did not significantly vary with time. Multivariate analysis. None of the variables significantly predicted inappropriate prescribing in either the UGIB or non-ugib subgroups. In the UGIB group, within the subgroup that underwent an endoscopy, a model significantly explained variation in inappropriate prescribing (P.0004). This model contained 2 significant variables, time delay to endoscopy (odds ratio per extra hour, 1.06 [95% CI, ]; P.006) and age (odds ratio per increased decade, 0.72 [95% CI, ]; P.02). Thus a longer time delay to endoscopy and younger patient age were independently associated with a higher chance of prescribing IV PPI inappropriately. Cost analysis. The costs of PPI therapy in all patients prescribed IV PPI during the background period were $33,164 compared with $14,566 in the ideal hypothetical model, with a potential cost savings of $18,598. For the UGIB subgroup these cost figures were $21,552 (actual) and $12,126 (ideal), respectively. The cost figures for non-ugib subgroup were $11,612 and $2,440, respectively. Part 2: Effect of Multidisciplinary Intervention A total of 113 (68 for UGIB, 45 for non-ugib) patients were prescribed and given a bolus and/or infusion of IV pantoprazole during the preintervention period and 105 (67 for UGIB, 38 for non-ugib) patients during the postintervention period. An additional 30 patients were prescribed IV pantoprazole in the postintervention non-ugib subgroup but had an oral PPI substituted by a pharmacist because they were not truly NPO. Upper gastrointestinal bleeding subgroup. In the UGIB subgroup, there was a significant absolute reduction in the proportion without an appropriate indication of 26% (95% CI, 10% 42%; P.0001) during the postintervention period (Figure 2). Subgroup analysis explored which errors of inappropriate use of IV pantoprazole were most often corrected (Tables 3 and 4). Table 3 represents errors by physicians that resulted in overspending of IV pantoprazole, whereas Table 4 corresponds to underspending errors. Non bleeding subgroup. In the non-ugib subgroup, there was a significant absolute reduction in the Figure 2. The proportion of physicians who prescribed IV pantoprazole appropriately for the UGIB and the non-ugib subgroups in the preintervention (gray) and postintervention periods (black).

5 December 2005 IV PROTON PUMP INHIBITOR USE 1211 Table 3. Influence of the Intervention on Selected Physician Prescribing Errors That, When Corrected, Contributed to a Decrease in Drug Costs Between the Preintervention and Postintervention Periods (Part 2 of the Study) Control (n 68), (%) Study (n 67), (%) Difference (95% CI) P value Proportion of patients who continued to receive IV PZ despite no HRS seen on endoscopy Proportion of patients who received IV PZ for more than 24 h without confirming HRS 21 (31) 13 (19) 12% ( 3% to 26%).1 15 (22) 5 (8) 14% (3% to 26%).02 PZ, pantoprazole. proportion without an appropriate indication by 41% (95% CI, 24% 58%; P.0001) during the postintervention period (Figure 2). As a secondary analysis, when the 30 patients with inappropriate IV pantoprazole that was corrected by the hospital pharmacist were considered errors, there was still an absolute improvement of 22%. Cost analysis. There was no statistically significant pharmaceutical cost difference between the preintervention period and the postintervention period, with median weekly costs of $4188 and $4485, respectively. Baseline and balance measures. The mean age of the patients in the 2 study periods was identical (62 years); there was a nonsignificant difference in the gender proportion, with 58% male during the preintervention period compared with 64% during the postintervention period. The preintervention period was comparable to the postintervention period in balance measures, including the total number of admissions to hospital (5739 vs 5884), the total number of UGIB cases presenting to the emergency department (106 vs 117), and the total number of gastroenterology consults (286 vs 331). Discussion An IV PPI is indicated in the treatment of UGIB caused by PUD associated with HRS and in patients requiring a PPI but who cannot take medications enterally. The results of this study highlight poor physician prescribing behavior in a university health care center. This was evident both for the ordering of IV PPI for the appropriate indication and for the selection of the correct dosing regimen. Similar phenomena have been noted by other groups, suggesting that inappropriate use of IV PPI might be a widespread problem in North America. A tertiary care center in the US reported that 56% of patients receiving IV PPIs had an inappropriate indication and estimated $100,000 in excess spending in a simple cost model. 14 The study s definition of appropriate use was liberal, accepting any source of UGIB (including variceal) instead of limiting the definition of appropriateness of the high-dose infusions to patients with UGIB caused by PUD with HRS. In addition, detailed analysis of the appropriateness of dosing was not done. 14 A study conducted in Toronto reported that only 35% of patients were prescribed IV pantoprazole appropriately for an UGIB (vs 50% in our study), and only 55% of patients prescribed IV pantoprazole for a non-ugib were actually NPO (vs 33% in our study). 12 A third study from Vancouver showed 57% of patients presenting with an UGIB received IV PPI before endoscopy. 13 Of the patients who had endoscopy, 57% continued to receive IV PPI despite the absence of HRS on endoscopy (vs 47% in our study). 13 The inclusion criteria, dosing regimens, and infusion durations used in 4 randomized trials formed the basis of our definition of appropriate dosing for the UGIB subgroup. 2,3,5,6 Although a clear benefit has not been proven for prescribing IV PPI pending en- Table 4. Influence of the Intervention on Selected Physician Prescribing Errors That, When Corrected, Contributed to an Increase in Drug Costs Between the Preintervention and Postintervention Periods (Part 2 of the Study) Control (n 68), (%) Study (n 67), (%) Difference (95% CI) P value UGIB treated appropriately with continuous infusion instead of intermittent bolus IV PZ Appropriate 80-mg bolus used instead of 40-mg or no bolus Continuous infusion ordered for 48 h when a 72- h infusion was indicated 44 (65) 63 (94) 29% (4% to 17%) (46) 45 (67) 21% (5% to 37%) (31) 13 (19) 12% ( 3% to 26%).1 PZ, pantoprazole.

6 1212 KAPLAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 12 doscopy in comparison to withholding it until after endoscopy, an adequately powered randomized trial answering this question would be difficult; the absolute risk reduction attributable to a few hours of extra early IV PPI would be anticipated to be very small. It was therefore decided that, pending further evidence, starting IV pantoprazole before endoscopy would be conservatively considered to be acceptable (although not necessary) 18,21 for the purposes of this study, as long as it was discontinued within 12 hours of endoscopy if HRS were not confirmed. Excluding all IV PPI use before confirmation of HRS on endoscopy from the appropriate use definition would have resulted in an even larger point estimate of the extent of the errors (ie, more use deemed inappropriate), but it would have arguably been less generalizable. The use of IV PPI pending endoscopy is already widespread and has a theoretical basis for being marginally effective. Despite dramatic evidence of inappropriate IV pantoprazole use, the second portion of the study showed how physicians and pharmacists can work together to create solutions to inappropriate drug use. Through simple interventions that involve improving physician education, simplifying the ordering process of IV pantoprazole, empowering pharmacists to look for and correct physician mistakes (while allowing physicians to override these corrections as needed for individualized care), and optimizing the use of endoscopy to confirm HRS (and perform endoscopic therapy as needed) in those patients prescribed high-dose continuous infusion of IV pantoprazole, we have been able to dramatically improve physician prescribing behavior. Physician prescribing errors are a common international problem with the potential for serious patient complications. 22,23 Dean et al 22 studied the causes of prescribing errors in their hospital and determined that lack of knowledge was an important etiology of physician error. Likewise, the errors associated with using IV pantoprazole inappropriately may stem from a misunderstanding of its role. Improving physician knowledge is paramount to improving their prescribing behaviors. However, at times education alone is not sufficient; systems to catch and correct physician errors in real time are also very important. Several studies have shown that pharmacists play a critical role in catching physicians medication mistakes and in turn lead to better patient care. 24,25 In addition, computerized prescription ordering systems have been shown to decrease prescribing errors. 26,27 The computerized nature of our pharmacy ordering system and consultant request systems and the involved role of our pharmacists give our center several advantages. Also, the ability for physicians to override the system when they believed the case justified nonstandard therapy is of absolute importance for physician buy-in. The goal was to educate, guide, and attempt to correct unrecognized errors, not to overturn physiciandirected individualized care. Despite the significant improvement in physician prescribing behavior, the interventions instituted did not translate into significant drug cost savings. The reason for this appeared to be that overspending prescribing behaviors might be more resistant to change than underspending ones, resulting in improvements in appropriate prescribing but a lesser than anticipated cost savings. Table 4 provides several examples of our successes in correcting underspending errors. For example, during the preintervention period, 35% of hospital physicians treated an UGIB with IV pantoprazole bolus doses instead of an infusion as outlined by Lau et al 2 (an error that actually saves money), as opposed to only 6% during the postintervention period. In addition, patients in the postintervention period were more likely to get the correct bolus for an UGIB (80 mg, instead of 0 40 mg) and to receive the infusion for a complete 72 hours, both of which cost more. Cost-saving prescribing errors were also improved on, but to a slightly lesser extent. For example, during the preintervention period 23% of patients were prescribed IV PPI for more than 24 hours without ever having had endoscopy to confirm HRS. The rate of this error was reduced to 8% during the postintervention period. Although the interventions taken have improved physician prescribing behavior, potential supplementary costs might have occurred. The recommended gastroenterology consult was designed to minimize non-gi services from ordering infusions for UGIB without seeking an endoscopist consultation, because evidence for the infusion is currently limited to those patients whose UGIB is caused by PUD associated with HRS. Sung et al 28 have shown that IV PPI high-dose infusion combined with endoscopic therapy is far more effective for nonbleeding visible vessels than IV PPI infusion alone. Overall, GI consults increased by 16% from the control period. Although a portion of this increase was likely due to the GI consult attached to infusion orders, there was a 10% higher volume of UGIB seen during the postintervention period, which could have accounted for some of this increase. Corrections in dosing and duration of IV PPI for UGIB patients and increasing the appropriate usage of high-dose infusions instead of bolus regimens in UGIB increased the total drug

7 December 2005 IV PROTON PUMP INHIBITOR USE 1213 budget, but in theory this might have resulted in better outcomes for patients and reduced rebleedingassociated costs. 2 By using a number needed to treat of for high-dose PPI infusions in preventing rebleeding and just under 20 extra patients treated with infusions in our intervention group, 3 rebleeding episodes might have been avoided in our 67 intervention patients (number needed to treat of 22 for our intervention to prevent rebleeding). Lastly, nursing costs and other miscellaneous costs related to infusion contribute. It takes a nurse approximately 10 minutes to reconstitute IV pantoprazole and then administer it versus approximately 2 minutes for the oral tablet. The cost of the tubing and mini-bag required to administer the IV drug in our center is $2.50 and $1.00, respectively, for each bag. When nursing and equipment costs were considered for part 1 of the study, the total cost of IV pantoprazole increased from $33,164 to $48,112 during the 4 month period studied. This study is limited in its pre-post design. A cluster randomized trial would be needed, but this is currently limited by the different physician-pharmacy relationships and computer systems at different institutions. Our baseline and balance measures were comparable between the 2 study periods. There are also potentially subjective errors in review of medical charts. However, because the data abstracted were complete in all cases, and the outcomes were reasonably objective (recording of HRS, coprescription of oral medications, oral diet), this bias is likely minimal. Inappropriate use of IV pantoprazole is a common problem in our institution. This study demonstrated how a multidisciplinary approach involving simple interventions resulted in improved physician prescribing behavior with respect to IV pantoprazole use. In some cases, improving physician prescribing behavior saved costs, whereas in others, improved behavior resulted in excess costs. Because underspending errors appeared to be more readily corrected than overspending errors, an overall cost savings was not achieved, but an assortment of prescribing errors were reduced by both a statistically and a clinically significant degree. References 1. Gilbert DA. Epidemiology of upper gastrointestinal bleeding. Gastrointest Endosc 1990;36(Suppl): S8 S Lau JY, Sung JJ, Lee KK, et al. Effect of intravenous omeprazole on recurrent bleeding after endoscopic treatment of bleeding peptic ulcers. N Engl J Med 2000;343: Lin HJ, Lo WC, Lee FY, et al. A prospective randomized comparative trial showing that omeprazole prevents rebleeding in patients with bleeding peptic ulcer after successful endoscopic therapy. Arch Intern Med 1998;158: Zed PJ, Lowewen PS, Slavik RS, et al. Meta-analysis of proton pump inhibitors in treatment of bleeding peptic ulcers. Ann Pharmacother 2001; 35: Hasselgren G, Lind T, Lundell L, et al. Continuous intravenous infusion of omeprazole in elderly patients with peptic ulcer bleeding. Scand J Gastroenterol 1997;32: Schaffalitzky de Muckadell OB, Havelund T, Harling H, et al. Effect of omeprazole on the outcome of endoscopically treated bleeding peptic ulcers: randomized double-blind placebo-controlled multicentre study. Scand J Gastroenterol 1997;32: Barkun A, Sabbah S, Enns R, et al. The Canadian Registry on Nonvariceal Upper Gastrointestinal Bleeding and Endoscopy (RUGBE): endoscopic hemostasis and proton pump inhibition are associated with improved outcomes in a real-life setting. Am J Gastroenterol 2004;99: Paul J, Metz D, Maton PA. Pantoprazole IV treatment decreases antacid usage in patients with gastroesophageal reflux disease (abstract). Gastroenterology 1999;116:G Metz DC, Forsmark C, Lew EA, et al. Replacement of oral proton pump inhibitors with intravenous pantoprazole to effectively control gastric acid hypersecretion in patients with Zollinger-Ellison syndrome. Am J Gastroenterol 2001;96: Hartmann M, Ehrlich A, Fuder H, et al. Equipotent inhibition of gastric acid secretion by equal doses of oral or intravenous pantoprazole. Aliment Pharmacol Ther 1998;12: Kaplan G, Bates D, McDonald D, et al. Inappropriate prescribing leading to rapidly escalating costs of IV proton pump inhibitors in the Calgary Health Region. Pharma News 2002;26: Cornish P, Papastergiou J, Saibil F. Audit of IV pantoprazole: patterns of use and compliance with guidelines. Can J Hosp Pharm 2002;55: Enns R, Andrews CN, Fishman M, Hahn M, Atkinson K, Kwan P, Levy A. Description of prescribing practices in patients with upper gastrointestinal bleeding receiving intravenous proton pump inhibitors: a multicentre evaluation. Can J Gastroenterol 2004;18: Guda NM, Noonan M, Kreiner MJ, et al. Use of intravenous proton pump inhibitors in community practice: an explanation for the shortage? Am J Gastroenterol 2004;99: Johnston JH. Endoscopic risk factors for bleeding peptic ulcer. Gastrointest Endosc 1990;36(Suppl): S16 S Laine L, Peterson WL. Bleeding peptic ulcer. N Engl J Med 1994; 331: Brunner G, Luna P, Hartmann M, et al. Optimizing the intragastric ph as a supportive therapy in upper GI bleeding. Yale J Biol Med 1996;69: Barkun A, Bardou M, Marshall JK, et. al. Nonvariceal upper GI bleeding consensus conference group: consensus recommendations for managing patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med 2003;139: Freston J, Chiu YL, Pan WJ, et al. Effects on 24-hour intragastric ph: a comparison of lansoprazole administered nasogastrically in apple juice and pantoprazole administered intravenously. Am J Gastroenterol 2001;96: Taubel JJ, Sharma VK, Chiu YL, et al. A comparison of simplified lansoprazole suspension administered nasogastrically and pantoprazole administered intravenously: effects on 24 intragastric ph. Aliment Pharmacol Ther 2001;15: Enns RA, Gagnon YM, Rioux KP, et al. Cost-effectiveness in Canada of intravenous proton pump inhibitors for all patients presenting with acute upper gastrointestinal bleeding. Aliment Pharmacol Ther 2003;17: Dean B, Schachter M, Vincent C, et al. Causes of prescribing errors in hospital inpatients: a prospective study. Lancet 2002; 359:

8 1214 KAPLAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No Lesar TS, Briceland L, Stein DS. Factors related to errors in medication prescribing. JAMA 1997;277: Leape LL, Cullen DJ, Clapp MD, et al. Pharmacist participation on physician rounds and adverse drug events in the intensive care unit. JAMA 1999;282; Kucukarslan SN, Peters M, Mlynarek M, et al. Pharmacists on rounding teams reduce preventable adverse drug events in hospital general medicine units. Arch Intern Med 2003;163: Fontan JE, Maneglier V, Nguyen VX, et al. Medication errors in hospitals: computerized unit dose drug dispensing system versus ward stock distribution system. Pharm World Sci 2003;5: Evans KD, Benham SW, Garrard CS. A comparison of handwritten and computer-assisted prescription in an intensive care unit. Crit Care (Lond) 1998;2: Sung JJ, Chan FK, Lau JY, et al. The effect of endoscopic therapy in patients receiving omeprazole for bleeding ulcers with nonbleeding visible vessels or adherent clots: a randomized comparison. Ann Intern Med 2004;139: Address requests for reprints to: J. Romagnuolo, MD, Medical University of South Carolina, 96 Jonathon Lucas Street, CSB #210, PO Box , Charleston, South Carolina romagnuo@ musc.edu; fax: Supported as an independent investigator (J.R.) by the Alberta Heritage Foundation for Medical Research. Dr Romagnuolo has served on the AstraZeneca advisory panel, including IV Nexium/Esomeprazole resource utilization study planning, and has received minor honoraria from all proton pump inhibitor companies, including pantoprazole for the CME lectures. Abstract presented (oral presentation) at the American College of Gastroenterology annual meeting, October 2002, Seattle, Washington, and published in the American Journal of Gastroenterology supplement (2002;97:S236). A related poster was also presented in Banff at the Canadian Association of Gastroenterology annual meeting in February 2004 and published in the Canadian Journal of Gastroenterology (2004;18:133A 134A). The second portion of the study was presented (poster presentation) at the Digestive Disease Week conference, May 2004, New Orleans, Louisiana, and published in Gastroenterology (2004;126:A433).