Prescribing Portraits to Optimize Prescribing of Proton Pump Inhibitors in Long-Term Care: PPI-T STOP Study

Abstract

Background

Proton pump inhibitors (PPIs) are among the most commonly prescribed medications in Canada, particularly for older adults (at least 65 years of age). Overprescribing of long-term PPIs leads to health care system waste and is associated with adverse effects, including infections and fractures. The high prevalence of PPI prescribing in long-term care (LTC) facilities prompted an evaluation of systematic approaches to PPI deprescribing.

Objective

To assess the impact of individualized prescribing portraits, a type of audit-and-feedback quality improvement intervention, on PPI deprescribing in the LTC setting.

Methods

This prospective, nonblinded, uncontrolled, pre–post quality improvement study was conducted from December 2021 to April 2022 at a 126-bed LTC facility in Vancouver, British Columbia. A PPI prescribing portrait was developed for each prescriber (n = 5) at the LTC facility, containing the prescriber’s personal PPI prescribing metrics as compared with those of their peers across all LTC facilities within the same health authority; an evidence summary for PPI deprescribing; and a personalized list of the prescriber’s PPI-treated residents, along with their respective PPI indications and strategies for PPI deprescribing. Three months after the prescribers received their PPI prescribing portraits, the number and types of PPI deprescribing orders were recorded.

Results

The implementation of prescribing portraits resulted in 17 (61%) of 28 PPI-treated residents receiving a deprescribing order by the end of the study period. Of the 28 PPI-treated residents, 20 were determined to be eligible for PPI deprescribing according to the evidence summary presented in the prescribing portrait; of these 20 residents, 16 (80%) appropriately received PPI deprescribing.

Conclusions

Individualized prescribing portraits had the potential to increase evidence-based PPI deprescribing among LTC residents, beyond the extent of deprescribing previously achieved through standard of care.

INTRODUCTION

Proton pump inhibitors (PPIs) are indicated for a variety of gastrointestinal conditions, including gastroesophageal reflux disease, peptic ulcer disease, and prevention of ulcers induced by nonsteroidal anti-inflammatory drugs (NSAIDs).¹ PPIs are among the most widely prescribed drugs in Canada, accounting for $253.8 million in public program spending in 2020.² In 2016, 29% of older adults in Canada (age 65 years or older) were taking PPIs, making them the second most frequently prescribed medication in this age group. PPI use was even higher among older adults living in long-term care (LTC) facilities, with an estimated 38% of such residents receiving a PPI.³

The high prevalence of PPI prescribing calls into question the appropriateness and duration of their use, especially in the LTC setting. A cross-sectional study based on data collected in 2004 from 1174 nursing home facilities in the United States⁴ showed that among older adult residents receiving PPIs, 48.95% of PPI use was not evidence-based, as defined by the US Food and Drug Administration and the UK National Institute for Health and Care Excellence guidelines. These findings were echoed by a retrospective cross-sectional study conducted in 2015/16 across 6 LTC facilities in British Columbia, which found that 43.7% of PPI orders did not have a documented common evidence-based indication, defined by the authors as gastroesophageal reflux disease or peptic ulcer disease, and 33.9% of PPI orders had no documented indication at all.⁵

Overprescribing of PPIs causes health care system waste. Furthermore, prolonged use of PPIs has been associated with adverse effects, including hypomagnesemia, fractures, and enteric infections, although the evidence is very uncertain and causality has not been established.^6–8 The American Geriatrics Society 2019 Beers Criteria recommend, with a high level of evidence and a strong strength of recommendation, to “avoid scheduled use for >8 weeks” in most older adults.⁹

Given the concern about inappropriate long-term PPI use, there is growing interest in planned, supervised discontinuation or reduction in PPI therapy (termed “deprescribing”). Various deprescribing interventions have been attempted in LTC facilities to date.¹⁰ However, quality improvement (QI) approaches to deprescribing PPIs in LTC have not been examined.

One systematic approach to QI is the audit-and-feedback method, whereby clinicians are given performance feedback and prompted to modify their practice when it is not aligned with a desirable target.¹¹ One example of audit and feedback is a prescribing portrait, which outlines a clinician’s prescribing habits compared with their peers.¹² The aim of this study was to assess the impact of individualized prescribing portraits on PPI deprescribing in LTC.

METHODS

Study Design and Setting

This single-site, prospective, nonblinded, uncontrolled pre–post QI study was conducted from December 19, 2021, to April 17, 2022, at Holy Family Hospital, an LTC facility in Vancouver, British Columbia. The site was recruited from Providence Health Care (a regional health authority in the province) according to the following criteria: an LTC facility staffed by physicians and a clinical pharmacist who were capable and willing to engage in a QI initiative aimed at PPI deprescribing.

The rationale for selecting a single site, rather than multiple sites, for this study was based on principles of the plan–do–study–act (PDSA) cycle method, particularly the principles of iterative cycles and small-scale testing.¹³ Starting with a single site, feedback will be collected to inform the next iteration of this QI initiative and, in doing so, increase the likelihood of success as the QI initiative expands in scope.

The Providence Health Care Research Ethics Board reviewed the protocol of this study and determined that the study was exempt from the requirements for ethics approval because it met the criteria of a QI study.

Evidence-Based and Long-Term PPI Indications

A review of the literature was conducted to identify evidence-based indications for PPI therapy and to identify those indications for which long-term PPI therapy may be warranted (search details available by request to the corresponding author). A summary of evidence-based indications and a PPI deprescribing algorithm were presented as part of the individualized prescribing portrait (see “Intervention”, below). The following indications were identified as warranting long-term PPI therapy: Barrett esophagitis; hypersecretory conditions (e.g., Zollinger–Ellison syndrome); severe esophagitis, defined as Los Angeles grade C or D esophagitis, as determined by endoscopic assessment; dual antiplatelet therapy or anticoagulant + antiplatelet therapy if patient had history of upper gastrointestinal bleeding or at least 1 additional risk factor (age > 65 years, high-dose NSAID therapy, history of uncomplicated ulcer, concurrent use of acetylsalicylic acid [ASA], corticosteroid, or anticoagulant); and long-term use of NSAID plus at least 3 risk factors (age > 65 years, high-dose NSAID therapy, history of uncomplicated ulcer, concurrent use of ASA, corticosteroid, or anticoagulant).^14,15 See Appendix 1 for the PPI deprescribing algorithm.

Criteria for PPI Deprescribing Eligibility

The following criteria, based on the most recent clinical practice guidelines, were used to identify residents eligible for PPI deprescribing: no documented PPI indication, mild or moderate gastroesophageal reflux disease treated for longer than 8 weeks, peptic ulcer disease treated for longer than 12 weeks, Helicobacter pylori–induced ulcer treated for longer than 2 weeks, dyspepsia secondary to prescribing cascade (e.g., PPI prescribed to treat anticholinergic drug-induced dyspepsia), and stress ulcer prophylaxis continued beyond critical care admission.^14,15 The deprescribing eligibility status of the residents was not explicitly disclosed to the physicians, but these criteria were presented within the algorithm provided as part of the prescribing portrait. See Appendix 1 for the PPI deprescribing algorithm.

PPI Deprescribing Methods

Several methods of PPI deprescribing, based on the most recent clinical practice guidelines, systematic reviews, and randomized controlled trials, were identified and ranked by preference, according to efficacy and tolerability.^14,16,17 The preferred methods of PPI deprescribing were reducing the PPI dose by decreasing drug strength or frequency of administration; and switching to on-demand dosing, whereby PPI was given once daily for several consecutive days while the patient was symptomatic, then discontinued once symptoms resolved, with the option to repeat this treatment if symptoms recurred.^18,19 Other methods of PPI deprescribing included discontinuing the PPI and switching to an H2 receptor antagonist; PRN dosing, whereby a single PPI dose is administered in response to symptoms; and discontinuing the PPI abruptly.^20,21 See Appendix 1 for the PPI deprescribing algorithm.

Study Population

All residents at Providence Health Care LTC sites were used to calculate the proportion of residents with an active PPI order across all sites, as well as the median duration of active PPI orders across all sites.

Similarly, all residents of Holy Family Hospital were included in calculating individual Holy Family Hospital physicians’ proportion of residents with an active PPI order, as well as individual physicians’ median duration of active PPI orders.

Residents within Holy Family Hospital who had an active PPI order on December 13, 2021, were included in the portion of the prescribing portrait that reported the PPI usage and indication of individual residents.

For further details about the study population, see the section entitled “Intervention,” below.

Intervention

The intervention for this QI study was a prescribing portrait. To inform the portrait, one of the authors (Y.W.) conducted a literature review (search strategy available by request to the corresponding author). An individualized prescribing portrait was generated for each physician (n = 5). Each portrait was made up of 3 components: the personalized prescribing data of the physician compared with those of their peers; an evidence summary of PPI usage alongside a PPI deprescribing algorithm; and a list of the physician’s PPI-treated patients, their PPI indication (if any), and strategies for PPI deprescribing.

Two metrics constituted the physician’s personalized prescribing data. First, each physician’s proportion of PPI-treated patients relative to the total number of patients under their care was compared with the proportion of PPI-treated patients across all 5 Providence Health Care LTC facilities. Second, the average duration of the physician’s active PPI orders was compared with the average duration of all active PPI orders within the Providence Health Care LTC facilities.

We also presented a succinct summary of evidence-based PPI indications, the potential harms of long-term PPI use, and an evidence-based PPI deprescribing algorithm. The methods of literature review used to inform this evidence summary and deprescribing algorithm are detailed above, in the section entitled “Evidence-Based and Long-Term PPI Indications”.

Finally, an individualized list of the prescriber’s PPI-treated patients was generated, comprising each patient’s PPI order (drug, strength, route, frequency); the duration of their PPI order; the PPI indication; and a list of preferred deprescribing methods, based on the aforementioned PPI deprescribing algorithm. Preliminary drafts of the individualized prescribing portraits were reviewed by the authors, the coordinator of pharmacy services for Providence Health Care Long Term Care, the clinical pharmacist at Holy Family Hospital, and an expert consultant in deprescribing (also a clinical pharmacist). Subsequent revisions to the draft were made before finalization, based on each reviewer’s feedback. See Appendix 1 for an example of a PPI prescribing portrait.

One month before distribution of the individualized prescribing portraits, the LTC physicians were notified of the project by means of an infographic (Appendix 2).

The clinical pharmacist met with 4 of the 5 physicians in person to review their respective prescribing portraits individually and to provide recommendations for each patient listed in the portraits. The fifth prescriber wished to review their portrait with their medical resident in lieu of discussing with the clinical pharmacist. The clinical pharmacist followed up on all patients identified as having an active PPI order by monitoring the electronic medical record to confirm that the previously discussed actions were taken.

Outcomes

The primary outcome of this study was the proportion of patients who received a PPI deprescribing order, from among those initially identified as having an active PPI order, within the 3 months following distribution of the individualized PPI prescribing portraits.

Additional (secondary) outcomes of this study were the proportion of deprescribing-eligible residents with a deprescribing order within 3 months (with deprescribing eligibility as defined by the criteria presented in the section entitled “PPI Deprescribing Methods”); the method of deprescribing; and the sustainability of deprescribing, defined as the proportion of PPI deprescribing orders that remained in place at 3 months following the distribution of the individualized PPI prescribing portraits. If any PPI deprescribing orders were reversed, the authors investigated the reason for reversal.

Data Analysis

Descriptive statistics were used to summarize and describe the data. Residents who died between the time that the prescribing portraits were generated and the time that deprescribing interventions were made were excluded from certain aspects of the data analysis.

RESULTS

A total of 28 patients met the inclusion criteria. Their baseline characteristics are reported in Table 1.

TABLE 1.

Baseline Characteristics of Residents at Holy Family Hospital with Active PPI Order

Characteristic	No. (%) of Patientsa (n = 28)
Sex
Male	9 (32)
Female	19 (68)
Age (years) (mean ± SD)	84.3 ± 9.9
Duration of admission (days) (mean ± SD)	998 ± 1242
Duration of current PPI order (days) (mean ± SD)	243 ± 239
PPI regimen
Pantoprazole
20 mg once daily	4 (14)
40 mg once daily	10 (36)
40 mg twice daily	5 (18)
Esomeprazole
20 mg once daily	2 (7)
40 mg once daily	7 (25)
PPI indication
Gastroesophageal reflux disease	10 (36)
History of GI bleeding	8 (29)
GI ulcer	1 (4)
Gastroprotection for concomitant NSAID or antithrombotic agent	1 (4)
Stress ulcer prophylaxis from previous critical care admission	1 (4)
None documented	7 (25)
Eligibility for deprescribingb
Eligible	20 (71)
Not eligible	8 (29)

GI = gastrointestinal, NSAID = nonsteroidal anti-inflammatory drug, PPI = proton pump inhibitor, SD = standard deviation.

^aExcept where indicated otherwise.

^bEligibility for deprescribing defined in the text (see section entitled “Criteria for PPI Deprescribing Eligibility” in Methods).

Primary Outcome

Of the 28 PPI-treated residents, 17 (61%) had a deprescribing order within the 3 months following distribution of the PPI prescribing portraits (Figure 1). Seven residents did not receive deprescribing orders, and 4 residents died. Of the 4 residents who died, 2 received deprescribing orders before their death. Deceased residents were excluded from the reported proportion of residents who received deprescribing.

FIGURE 1.

Outcomes for all residents treated with proton pump inhibitors (PPIs; n = 28) at 3 months after distribution of the PPI prescribing portraits.

Secondary Outcomes

Of the 28 PPI-treated residents, 20 were identified as being eligible for PPI deprescribing. The proportion of deprescribing-eligible residents who received a PPI deprescribing order within the 3 months following distribution of the PPI prescribing portraits was 80% (16/20) (Figure 2). One resident who died before the end of the 3-month follow-up had received a deprescribing order, but deceased residents were excluded from the reported proportion of residents who received deprescribing.

FIGURE 2.

Outcomes for deprescribing-eligible residents (n = 20) at 3 months after distribution of the proton pump inhibitor prescribing portraits.

The most common method of deprescribing was dose reduction, observed in 94% (16/17) of the residents who received deprescribing orders. Examples of dose reduction included decreasing the unit dose (e.g., from 40 to 20 mg) or the dosing frequency (e.g., from twice daily to once daily). Tapering until discontinuation of PPI was applied for 1 resident. There were no orders to discontinue the PPI abruptly, to switch to on-demand PPI dosing, or to switch to a histamine H2-receptor antagonist or antacid.

Of the 17 deprescribing orders, 14 (82%) remained in place and 3 (18%) had been reversed by the end of 3 months after distribution of the PPI prescribing portraits. Reasons for reversal included new-onset symptoms of epigastric pain, esophageal reflux, and nausea. The 3 reversals of PPI deprescribing were ordered by a single physician.

DISCUSSION

Our study has demonstrated the feasibility and efficacy of prescribing portraits as a form of audit and feedback to achieve appropriate PPI deprescribing in LTC. Within 3 months of distributing these prescribing portraits, the majority of PPI-treated residents had received a deprescribing order. An even higher proportion of deprescribing was seen among patients identified, a priori, as being eligible for PPI deprescribing.

Notably, throughout their admission, these residents had already been assessed for appropriate pharmacotherapy, including PPI use, during biannual medication reviews, according to the facility’s standard of care. The mean duration of admission for these residents was 998 days, which implies that the vast majority would have undergone at least 1 medication review. Yet despite regular medication reviews, many residents did not receive PPI deprescribing until implementation of our PPI prescribing portrait intervention. This finding gives us confidence that our intervention had a true effect on PPI deprescribing.

The methods of deprescribing used by the prescribers in this study were tapering before discontinuation and dose reduction, with the latter being used for 16 of the 17 deprescribing orders. The extensive use of dose reduction as the method of deprescribing can be attributed to a few factors. First, the deprescribing algorithm supplied within the prescribing portraits favoured dose reduction. This recommendation was based on literature stating that there is no increase in the risk of symptom relapse with lower PPI doses relative to higher PPI doses.¹⁹ Second, the standard of care in Providence Health Care LTC facilities is to conduct a medication review every 6 months for reassessment of long-term medications; as such, this study’s 3-month follow up period may not have been long enough to capture further PPI dose tapering with eventual discontinuation.

The findings of our study are consistent with those of a previous evaluation by Tandun and others,²² conducted at 2 LTC facilities in British Columbia. In that study, a clinical pharmacist presented prescribers with a list of their PPI-treated residents, and in one of these facilities, the pharmacist also conducted weekly reviews of residents’ medical records, offering deprescribing recommendations. Overall, 62% of the residents in the study received PPI deprescribing by the end of the 4-month study period, with a higher rate of PPI deprescribing in the facility where the pharmacist offered more direct patient care.

In contrast to the study by Tandun and others,²² we developed and distributed prescribing portraits as the intervention; the role of the clinical pharmacist was to provide patient care as per existing standards of practice. In our study, the need for additional pharmacist workload was minimal, yet we achieved a similarly high rate of PPI deprescribing relative to the study of Tandun and others,²² which required a higher level of active pharmacist intervention.

To our knowledge, our study is the first to use prescribing portraits as a form of audit and feedback for PPI deprescribing in LTC. Given that our study standardized the protocol for generating prescribing portraits, these portraits can be replicated for use across multiple sites and scaled to larger populations. Indeed, other drug classes have been identified as warranting QI for appropriate prescribing in the LTC setting, particularly antipsychotic medications.

Limitations

The study reported here was the first iteration of this prescribing portrait–based QI project. As such, compared with other prospective studies, our study had a smaller sample, recruited from a single site, which limits its generalizability. The short follow-up period of 3 months limited our ability to capture the full scope of continued PPI deprescribing versus reversal of deprescribing orders over time. Conversely, this short follow-up period may have led us to overestimate the benefit of prescribing portraits, given that previous literature has shown that PPI deprescribing interventions led to an initial reduction in PPI use that was not sustained over a longer period.²³ Additionally, outpatient data from before residents’ admission to LTC was unavailable, and our electronic medical record system contained information only from November 2019 onward; thus, the average duration of PPI orders was likely an underestimation, and some documented PPI indications could have been missed. Finally, having the clinical pharmacists discuss the portrait with each prescriber was an important part of the intervention but may be impractical; if such individualized discussions are not feasible, the effectiveness of the intervention may be limited.

In terms of the PDSA cycle method, future iterations of this QI project would benefit from an a priori protocol for handling data from patients who die within the study period, as well as an expansion to other drug classes such as antipsychotics.

CONCLUSION

Prescribing portraits aimed at fostering appropriate PPI use in LTC led to PPI deprescribing in 61% of all PPI-treated residents in a single LTC facility. Deprescribing was initiated for 80% of the residents who were identified as being eligible for PPI deprescribing. Our results showed that prescribing portraits are a feasible approach to improving the quality of PPI use in LTC. In our study, the use of prescribing portraits increased appropriate PPI deprescribing beyond what had been previously achieved by the existing clinical practice of biannual medication reviews. Future studies should explore scale-up of this approach and evaluation of other medication classes. Studies employing this QI approach would benefit from longer follow-up to assess the sustainability of appropriate prescribing over time.

APPENDIX 1: Example of a PPI prescribing portrait. © 2021 Lower Mainland Pharmacy Services. Reproduced with permission

Definitions of abbreviations for all parts of Appendix 1: ASA = acetylsalicylic acid, C. difficile = Clostridioides difficile, CAP = community-acquired pneumonia, DAPT = dual antiplatelet therapy, Dx = diagnosis, GERD = gastroesophageal reflux disease, GI = gastrointestinal, H. pylori = Helicobacter pylori, H₂RA = histamine H2-receptor antagonist, HAP = hospital-acquired pneumonia, ICU = intensive care unit, MRN = medical record number, NSAID = nonsteroidal anti-inflammatory drug, PHC = Providence Health Care, PPI = proton pump inhibitor, PUD = peptic ulcer disease, UGIB = upper gastrointestinal bleeding.

APPENDIX 2: Infographic for physician notification. © 2021 Lower Mainland Pharmacy Services. Reproduced with permission

C. difficile = Clostridioides difficile, GERD = gastroesophageal reflux disease, H. pylori = Helicobacter pylori, LTC = long-term care, PHC = Providence Health Care, PPI = proton pump inhibitor.