Pre-Endoscopy Use of Proton Pump Inhibitor Intravenous Bolus Dosing in Hemodynamically Stable Patients With Suspected Upper Gastrointestinal Bleeding: Results of a Pharmacist-Managed Hospital Protocol to Reduce Continuous Infusion Pantoprazole Use

Abstract

Background: Guidelines for acute upper gastrointestinal bleeding (UGIB) recommend use of proton pump inhibitors (PPI) administered by continuous IV infusion (CI). Although data suggest comparable outcomes with CI and IV push (IVP) dosing post-endoscopy, there are limited data to support IVP PPI as the pre-endoscopy regimen. Objective: To evaluate the impact of a pharmacist-managed protocol for reducing PPI CIs and substitution of PPI IVP dosing in hemodynamically stable patients with suspected acute upper gastrointestinal bleeding (UGIB) prior to endoscopic intervention. Design, Setting, and Participants: Retrospective study; Tertiary-care community teaching hospital; Hemodynamically stable adults with confirmed or suspected UGIB. Hemodynamic stability was defined as a systolic blood pressure >90 mmHg, heart rate <100 beats, mean arterial pressure >65 mmHg, and no requirement for vasopressors. Intervention: All iterations of treatment recommendations encouraged an initial pantoprazole 80 mg IVP dose. In the pre-intervention group, patients were then treated at the at the provider’s discretion with the majority receiving CI pantoprazole. After implementation of the original protocol (Phase I), all hemodynamically stable patients were allowed 1 bag of CI pantoprazole (80 mg infused over 10 hours) before being transitioned by the pharmacist to pantoprazole 40 mg IVP every 12 hours. After internal analysis, the protocol was revised to allow patients to be immediately transitioned to IVP dosing without an initial CI (Phase II). Main Outcome: Incidence of continued bleeding or re-bleeding within 7 days of initial PPI dose. Results: A total of 325 patients were included across all 3 study phases. The median number of CI bags per patient was reduced from 4 pre-intervention, to 1.5 in phase I, and to 0 in phase II (P < .001). The primary endpoint of continued bleeding or re-bleeding within 7 days was similar across all 3 groups (5.0% vs 6.5% vs 5.2%, P = .92). Mean intravenous pantoprazole costs were reduced by $21.73/patient. Conclusions: Movement toward preferential use of IVP PPI prior to endoscopy for hemodynamically stable patients with confirmed or suspected UGIBs resulted in similar rates of continued bleeding or re-bleeding and generated modest cost savings. These findings warrant further investigation.

Introduction

Acute upper gastrointestinal bleeding (UGIB) is a medical emergency requiring immediate evaluation of patient hemodynamic stability, coagulation status, and bleeding source. For over 40 years, it has been recognized that elevating gastric pH promotes local platelet aggregation. ¹ This has led to numerous studies and guidelines demonstrating that acid suppressive therapy, specifically with PPIs, results in reduced rates of continued bleeding or re-bleeding, need for surgical intervention, and possibly mortality.^2,3 Both national and international guidelines from the past 10 years recommend an initial PPI bolus followed by a PPI continuous infusion (eg, pantoprazole 80 mg IVP followed by 8 mg/hour CI) for patients with high-risk stigmata of hemorrhage.^2,4,5 A 2021 update to the American College of Gastroenterology (ACG) guidelines for UGIB management could not recommend for or against use of pre-endoscopy PPI therapy given limited data, although the cited data utilized high-dose, continuous infusion PPI. ⁶ The 2021 European Society of Gastrointestinal Endoscopy (ESGE) guidelines give a weak recommendation to consider pre-endoscopic, high-dose PPI therapy. ⁷ Following endoscopic intervention, both ACG and ESGE recommend high-dose PPI therapy, with some preference given to CI PPI. In a cascade guideline for resource-limited areas, ESGE recommends at least provision of oral or IV push (IVP) PPI, but when resources are not a factor, ESGE still recommend PPI CI as the preferential treatment modality. ⁸

Alternatively, data have brought the necessity of a PPI CI into question, especially after an endoscopic intervention. ⁹ A Cochrane review comparing high dose PPI regimens to lower doses found insufficient evidence to support high dose regimens. ¹⁰ A systematic review and meta-analysis the following year evaluated studies where patients with high-risk stigmata were to be initially treated with endoscopy and then managed with either PPI IVP or CI found IVP dosing to be non-inferior, prompting the authors to recommend a guideline revision. ¹¹ Most recently, a meta-analysis and meta-regression analysis concluded low-dose PPI regimens are as efficacious as high-dose PPI regimens in regards to hemostasis post-endoscopy. ¹²

Based on literature comparing CI and IVP strategies as well as a nationwide intravenous pantoprazole shortage in 2017, this study describes our center’s progression toward preferential use of pre-endoscopy pantoprazole IVP in hemodynamically stable patients with a confirmed or suspected UGIB.

Methods

Study Design

Due to an impending shortage of intravenous pantoprazole in the summer of 2017, the Texas Health Dallas Pharmacy and Therapeutics committee approved a conservation strategy to reduce use of pantoprazole CIs. For all patients treated for a suspected UGIB, a single initial 80 mg IVP dose was recommended, regardless of the phase of this study. During Phase I of the protocol, when pantoprazole CI was ordered for a hemodynamically stable patient (defined as having a systolic blood pressure >90 mmHg, a heart rate <100 beats per minute, a mean arterial pressure >65 mmHg, and no requirement for vasopressors), the pharmacist would limit the pantoprazole CI to 1 bag (80 mg infused over 10 hours) and then change the pantoprazole to IVP (40 mg every 12 hours) starting within 2 hours of the end of the CI. The impact of this protocol was analyzed by 2 of the authors and reported internally. In response to these findings, the protocol was updated in July 2018 to allow pharmacists to immediately transition hemodynamically stable patients to PPI IVP (40 mg every 12 hours) following a single initial 80 mg bolus (Phase II). In both iterations of this protocol, providers were able to override the automatic transition to IVP by explicitly documenting the need for a CI or an intended duration of therapy.

This study is a retrospective chart review of all patients with a confirmed or suspected acute UGIB who received 2 or more doses of intravenous pantoprazole. Data reports used to identify all patients who received intravenous pantoprazole were generated by the hospital billing department. The pre-intervention period enrolled patients from January 1, 2017 to May 31, 2017. Phase I patients were enrolled from July 1, 2017 to November 30, 2017 and May 1, 2018 to July 31, 2018. Phase II patients were enrolled from August 7, 2018 to October 31, 2018. All phases of this project were approved by the Texas Health Resources Investigational Review Board.

Inclusion/Exclusion Criteria

Patients were included if they were 18 years of age or older and received 2 or more doses of intravenous pantoprazole for a confirmed or suspected acute UGIB. This study was restricted to patients who met criteria for hemodynamic stability, as defined above. Patients were excluded if they only received once daily pantoprazole, were on pantoprazole for reasons other than an upper gastrointestinal bleed (eg, stress ulcer prophylaxis, another gastrointestinal diagnosis, such as acid reflux or a non-gastrointestinal diagnosis), or if they had transferred from an outside hospital where they had been admitted for longer than 24 hours.

Outcomes

The primary endpoint evaluated the incidence of upper gastrointestinal continued bleeding or re-bleeding within 7 days of treatment with pantoprazole. Continued bleeding or re-bleeding was defined as bleeding visualized on endoscopy or a drop in hemoglobin of at least 2 g/dL in addition to documentation of bleeding in the patients’ chart. A comprehensive search for the terms “hematemesis,” “hematochezia,” “melena,” “blood,” or “bleed” were reviewed in each chart. Secondary endpoints included the number of pantoprazole continuous infusion bags and IVP doses, continued bleeding or re-bleeding within 3 days of pantoprazole treatment, requirement of initial and repeat blood transfusions, number of endoscopies performed and endoscopic interventions made, length of hospital stay, and in-hospital mortality. Pharmacy costs of intravenous pantoprazole, including drug, diluent, and IV tubing sets, were compared between all groups.

Statistical Analysis

This was a retrospective analysis where all patients in the predefined study periods who met inclusion criteria were included in the study. Thus, an a priori sample size and power analysis was not performed. A 3-way Chi-squared test was performed for categorical data and a Kruskal-Wallis 1-way analysis of variance test was used for continuous data assuming non-normal distributions. Two post hoc analyses were conducted, one between patients initially treated with CI pantoprazole compared to those treated solely with IVP pantoprazole and another between patients who underwent endoscopy versus those who did not. For these analyses, a Fisher’s exact test based on a 2 × 2 contingency table was utilized. A P-value of <.05 was considered statistically significant for the interaction between the groups. Statistics were calculated using web-based, free to access calculators through Social Science Statistics and GraphPad (www.socscistatistics.com; GraphPad Software, San Diego, CA; Last accessed August 2020).

Results

Patient and Treatment Characteristics

Baseline demographics are summarized in Table 1. In total, 1679 patients were screened for inclusion in either Phase I or Phase II. An initial 1275 patients were excluded with the most common reasons being receipt of once daily pantoprazole or use of pantoprazole for a non-UGIB indication. Of the remaining 404 patients, an additional 79 patients were excluded for not meeting the criteria for hemodynamic stability. This left a total of 325 hemodynamically stable patients in the final analysis. There were 80 patients in the pre-intervention group, 168 patients in the Phase I original protocol group, and 77 patients in the Phase II revised protocol group. There were statistically significant differences between the groups with regards to gender, admission for a chief complaint of UGIB, antiplatelet use, history of alcohol use, and ICU admission for a GIB. The median pre-endoscopy Rockall et al^13,14 score was 3 for all groups indicating an 11% risk of mortality, with 60% to 70% of all patients having a score ≤3.

Table 1.

Patient Characteristics.

	Pre-intervention (N = 80)	Phase I (N = 168)	Phase II (N = 77)	P-value
Age (median IQR)	66 (52.3-81.3)	68 (52-78.3)	70 (56-81)	.48
Male (%)	49 (61.3)	100 (59.5)	33 (42.8)	.03
Admission diagnosis of suspected UGIB (%)	62 (77.5)	139 (82.7)	51 (66.2)	.02
Pre-endoscopy Rockall scores
Median (IQR)	3 (2-4)	3 (2-4)	3 (2-4)	.29
0	7 (8.8%)	22 (13.1%)	6 (7.8%)
1	5 (6.3%)	12 (7.2%)	7 (9.1%)
2	15 (18.8%)	27 (16.1%)	10 (12.9%)
3	32 (40%)	60 (35.8%)	25 (32.5%)
4	19 (23.8%)	44 (26.2%)	25 (32.5%)
5	2 (2.5%)	3 (1.8%)	4 (5.2%)
HgB ≤7 g/dL at PPI start (%)	30 (37.5)	48 (28.6)	23 (29.8)	.35
ICU admit for GIB (%)	28 (35)	32 (19.1)	17 (22.1)	.02
Prior to admission history (%)
Hospitalized in last 30 d	6 (7.5)	14 (8.3)	1 (1.3)	.11
Alcohol use	31 (38.8)	54 (32.1)	14 (18.2)	.02
NSAID use	17 (21.3)	44 (26.2)	14 (18.2)	.35
Antiplatelet use	34 (42.5)	45 (26.8)	29 (37.7)	.03
Anticoagulant use	13 (16.3)	27 (16.1)	13 (16.9)	.99
Acid suppression use	34 (42.5)	69 (41.1)	29 (37.7)	.81
History of UGIB	16 (20)	37 (22)	14 (18.2)	.78

Note. IQR = interquartile range; UGIB = upper gastrointestinal bleeding; HgB = hemoglobin; ICU = intensive care unit; NSAID = non-steroidal anti-inflammatory drugs (excluding aspirin).

Treatment for UGIB is summarized in Table 2. A similar number of patients between groups were initially treated with octreotide in addition to pantoprazole. Endoscopy was performed in somewhat fewer patients in the Phase II group. Of those patients requiring upper endoscopy, fewer incidence of patients in the Phase II group received intervention during the endoscopy.

Table 2.

Treatment of UGIB.

	Pre-intervention (N = 80)	Phase I (N = 168)	Phase II (N = 77)	P-value
Octreotide use (%)	20 (25)	38 (22.6)	16 (20.8)	.82
Endoscopy (%)	62 (77.5)	133 (79.2)	49 (63.6)	.03
Intervention during endoscopy (%)	40/62 (64.5)	90/133 (67.7)	22/49 (44.9)	.02
RBC transfusion (%)	49 (61.2)	99 (58.9)	36 (46.7)	.13
Need for repeat transfusion (%)	24/49 (49)	55/99 (55.5)	21/36 (58.3)	.65
Proton pump inhibitor treatment
Initial every 12 h frequency (%)	12 (15)	36 (21.4)	45 (58.5)	<.001
Number of CI bags (median IQR)	4 (2-7)	1.5 (1-3)	0 (0-2)	<.001
Transitioned from initial CI to IVP (%)	17/68 (25)	101/132 (76.5)	29/32 (90.6)	<.001
Number of IVP doses (median IQR)	2 (2-4)	3 (1-6)	4 (2-6)	.003
Time to transition to oral regimen (days; median IQR)	3 (2-3.8)	2 (1-3.8)	3 (2-3.3)	.69
Medications received after hemostasis achieved
Antiplatelet (%)	15 (18.8)	13 (7.7)	7 (9.1)	.03
Anticoagulation a (%)	17 (21.3)	22 (13.1)	9 (11.7)	.16

Note. UGIB = upper gastrointestinal bleeding; RBC = red blood cell; CI = continuous infusion; IQR = interquartile range.

^aIncludes both therapeutic and prophylactic anticoagulation.

Outcomes

As shown in Table 3, there were no differences across all 3 groups regarding the primary outcome of continued bleeding or re-bleeding rate at 7 days (5.0% vs 6.5% vs 5.2%; P = .92). Similar to the primary outcome, continued bleeding or re-bleeding at 3 days and mortality were no different between the 3 groups. Median hospital length of stay was 4 days in all 3 groups.

Table 3.

Outcomes.

	Pre-intervention (N = 80)	Phase I (N = 168)	Phase II (N = 77)	P-value
Rebleed within 7 d (%)	4 (5)	11 (6.5)	4 (5.2)	.92
Rebleed within 3 d (%)	4 (5)	8 (4.8)	4 (5.2)	.99
Mortality (%)	2 (2.5)	5 (3)	3 (3.8)	.87
Hospital LOS (d; median IQR)	4 (3-8)	4 (2-7)	4 (3-5)	.61

Note. LOS = length of stay.

Regarding the pantoprazole management, the use of initial IVP pantoprazole increased over the different iterations of the protocol (15% vs 21.4% vs 58.5%; P < .001) with a subsequent decrease in the use of CI pantoprazole. The median number of CI pantoprazole bags (80 mg/100 mL) per patient was reduced from 4 (IQR 2-7) to 1.5 (IQR 1-3) to 0 (IQR 0-2) over the course of the different phases of the protocol (P < .001). Mean per patient intravenous pantoprazole costs in 2019 USD (drug, diluent, and IV tubing sets) were reduced by 42% over the progression of the protocol, from $51.30 ± 28.20 to $29.57 ± 19.89 (difference $21.73; 95% CI, $14.04-29.41).

Subgroup Analyses

Patients were also compared based on initial pantoprazole regimen. A total of 231 patients were initially treated with CI pantoprazole compared to 94 treated solely with IVP. The outcomes in this analysis were similar to the overall analysis for the study. Continued bleeding or re-bleeding at 7 days was not statistically significant between groups (6.5% in CI arm vs 4.3% in IVP arm; P = .60), nor was hospital mortality different (3.5% in CI arm vs 2.1% in IVP arm; P = .73).

Among the 244 patients who underwent upper endoscopy for management of their UGIB, 186 were initially treated with PPI CI prior to endoscopy compared to 58 initially treated with an IVP regimen. Continued bleeding or re-bleeding at 7 days was not significantly different between those treated with CI versus IVP (6.4% vs 3.4%, P = .53). Similarly, continued bleeding or re-bleeding at 3 days (4.8% vs 3.4%, P = 1.0) and mortality (1.6% vs 0%, P = 1.0) were not different between the groups.

Discussion

The findings of this retrospective analysis demonstrated that patients with a confirmed or suspected UGIB who were managed with a pre-endoscopy protocol which transitioned from primarily CI PPI to primarily IVP PPI had no significant differences in the rate of continued bleeding or re-bleeding. These results align with those of previous studies in post-endoscopic management that have shown no significant difference between intermittent and continuous dosing regimens.^9-12,15,16 While current treatment guidelines give consideration to high-dose CI PPI pre-endoscopy, our data suggest that for lower risk patients, a lower dose, IVP PPI may be an option.^6,7

The physiologic basis for use of a potent acid suppressive regimen for management of acute UGIBs dates to the 1970s. Maintaining a gastric pH above 6 has been shown to enhance the formation and stabilization of gastric blood clots.^1,17 At a gastric pH less than 5, platelet aggregation is reduced and the likelihood of clot stabilization and bleeding cessation is diminished. It is for this reason that other, less potent acid suppressive agents, such as histamine-2 receptor antagonists, have not shown significant benefit in patients with UGIBs.^3,18 Use of a CI theoretically maintains a more constant inhibition of the parietal cell and thus, should result in more consistent elevation to gastric pH. However, a 2007 study demonstrated similar pH values when pantoprazole 80 mg IVP given once followed by a CI of 8 mg/hour for 72 hours or pantoprazole 80 mg IVP given once followed by 40 mg IVP every 12 hours. ⁹ The mean gastric pH (5.8 vs 5.3) and the duration of time with a gastric pH above 6 (59% vs 49%) were not significantly different between the CI and IVP groups. Therefore, there is some biologic rationale to support intermittent dosing.

Prior to endoscopy, most current guidelines suggest high-dose, CI PPI therapy to reduce high-risk stigmata.^2,4,5 The majority of data comparing high-dose, or CI, PPI to lower, IVP doses exists in UGIB patients post-endoscopy. Data for patients with endoscopically-confirmed bleeding peptic ulcers were compiled for a 2013 Cochrane Database Review. This meta-analysis concluded that following endoscopy, there is insufficient evidence to support the superiority of a high-dose (ie, continuous infusion) regimen over lower doses. ¹⁰ Similarly, a 2018 meta-analysis comparing high versus low-dose PPI following endoscopy corroborated the aforementioned review. ¹² High-dose PPI regimens largely consisted of pantoprazole 80 mg IVP followed by an infusion whereas the low-dose regimens ranged from omeprazole 20 mg daily to pantoprazole 80 mg twice daily. This analysis concluded low-dose regimens following endoscopy were equally efficacious as higher doses with regards to re-bleeding, surgical intervention, and mortality. While there appears to be ample data to support lower dose PPI in patients with hemostasis achieved via endoscopy, there are limited data to support a pre-endoscopic approach of limiting CI PPIs, making our study novel. Our study included all patients with suspected or confirmed UGIB, unlike many other studies which only included patients who underwent endoscopy promptly on admission. ¹⁹ In a recent study by Khan and colleagues, high-risk UGIB patients who underwent endoscopy had higher rates of re-bleeding, need for embolization or surgery, or mortality if they were initially managed with an IVP regimen. While Rockall scores were not included for the Khan cohort, all patients included in the study had endoscopy findings of ulcers consistent with high-risk stigmata (eg, Forrest classification I or II). It may be postulated that in higher-risk, less stable patients, a PPI infusion regimen remains the preferred approach whereas our data would suggest in lower risk patients, an IVP regimen may be sufficient.

While there were no observed differences in clinical outcomes with the migration toward initial IVP PPI, we did find a modest drug cost savings. A 2016 cost-analysis by Lu et al ²⁰ utilized decision tree modeling and found an IVP regimen was associated with lower overall costs of care. When compared to the standard CI PPI, use of an intermittent IVP regimen both pre- and post-endoscopy was projected to reduce overall costs by $233/patient. Another study estimated potential cost savings of $121 000/year in the authors’ institution by preferential use of IVP PPI. ²¹ While IV pantoprazole has a relatively innocuous compatibility and tolerability profile, a prolonged CI could create the need for additional IV access, leading to the potential for midline or central venous access. Thus, there may be other cost savings related to venous access devices that we did not capture in this study. Other benefits may include reduced pharmacy labor and a lower chance for compounding errors. Since neither PPI CI nor IVP regimens demonstrated clear superiority for clinical outcomes in hemodynamically stable patients, the operational advantages merit consideration of IVP regimens as the preferential initial treatment.

The purpose of our study was to evaluate the safety of reducing pantoprazole requirements for patients with UGIBs. Though this evidence supports an alternative treatment regimen for these patients, it is imperative to note the importance of endoscopic interventions. Endoscopy remains the standard of care for both identification and intervention for UGIBs. Pre-endoscopic PPI therapy has been shown to reduce the stigmata of recent hemorrhage (active bleeding, non-bleeding visible vessels, or adherent clot) at endoscopy, but the primary method for achieving hemostasis and reducing the risk of re-bleeding is endoscopic intervention. ²² Thus, this study only serves to provide an alternative, cost-saving strategy to medically manage hemodynamically stable patients with UGIBs.

Due to the retrospective nature of this study, a limitation of our analysis is that the primary endpoint of continued bleeding or re-bleeding relied on appropriate documentation by providers and accurate chart review by the investigators. This was mitigated by using the predetermined definition of continued bleeding or re-bleeding for all phases of the chart review and by limiting the number of chart reviewers, reducing chance of variability. Additionally, data regarding endoscopic findings were not collected as there was inconsistent documentation of hemorrhagic stigmata and Forrest classifications. In lieu of Forrest classification, we were able to calculate pre-endoscopy Rockall scores to estimate mortality rates based on patient age, presence of shock as evidenced by tachycardia and/or hypotension and comorbidities including renal failure, liver failure and/or malignancy.^13,14 In the future, a prospective, randomized trial of initial IVP compared to CI PPI should include risk stratification and details on endoscopic findings, as efficacy data in higher-risk patients may not be as favorable. ¹⁹ There were some significant differences in baseline demographics which likely is the consequence of an uncontrolled, retrospective study. Lastly, this study was not a direct comparison between CI and IVP regimens but a comparison of different iterations of a progressively restrictive CI protocol. Even in phase II of the study, 41% of patients still received at least 1 CI bag. While our post-hoc analysis of initial management with IVP versus CI indicated similar outcomes, such an analysis likely introduces bias. Ideally, the question of equivalence between IVP and CI pre-endoscopy would need to be answered with a randomized, controlled trial comparing the 2 drug regimens.

Conclusion

Transition from pantoprazole CI to pantoprazole IVP as the initial medical management strategy in hemodynamically stable patients with a suspected UGIB resulted no differences in continued bleeding or rebleeding with a modest reduction in drug therapy costs. Pre-endoscopy IVP PPI should be considered as an alternative dosing regimen in hemodynamically stable patients with an acute UGIB, especially in resource-limited areas or during drug shortage. Further investigation between IVP and CI PPI as the initial pharmacotherapeutic approach in this patient population is warranted.