Deprescription of proton pump inhibitors in the primary care setting in the Canary Islands. A randomized massive educational intervention

Abstract

Background

It is estimated that between 20% and 80% of Proton Pump Inhibitor (PPI) prescriptions worldwide lack an appropriate clinical indication. Since their introduction in 1989, PPIs have proven highly effective for treating acid-related digestive disorders. However, over three decades of global use have revealed notable side effects, notwithstanding their overall safety profile. The chronic and indiscriminate use of PPIs by millions of people without proper indications represents a substantial public health issue and threatens the sustainability of health services worldwide.

Objective

To evaluate the clinical and economic outcomes of deprescribing PPIs that are indefinitely and incorrectly prescribed in primary care centres across the Canary Islands after an educational intervention.

Method

An online training course was developed to educate family physicians on deprescribing PPIs in accordance with internationally recognized guidelines. Indications for PPI prescriptions were assessed and compared among an intervention group and two control cohorts: physicians interested in but not participating in the training, and those who did not express any interest on it. Following an initial assessment at twelve weeks, the second interested group received the same training. Chronic prescribing practices were then evaluated for all three groups at twelve and twenty-four weeks post-study initiation.

Results

Before the intervention the 8% of indefinite PPI prescriptions in the primary care setting met the appropriate criteria according to the guideline used. This percentage did not increase in the groups of trained physicians. However, PPI-related expenditures were significantly lower in the intervention group compared to both control groups at 24 weeks, with the intervention group showing a median reduction of €82 per physician, while the control and anonymous groups showed increases of €15 and €160 respectively (p = 0.020 for the initial to 12-week comparison).

Conclusions

Given the high volume of use of these medicines, any strategy aimed at improving their management by physicians can have a significant beneficial effect on the health of the population, and an exponential impact on the consumption of public resources.

Longer-term follow-up, improved documentation of indications, and increased longitudinal patient care may play a critical role in the impact of educational interventions.

Introduction

Proton Pump Inhibitors (PPIs) are medications that irreversibly inhibit the H+/K+-ATPase enzyme in the parietal cells of the gastric mucosa, blocking gastric acid secretion through hydrogen ion exchange for potassium. PPIs have a plasma half-life of 1–2 h, usually achieving maximal acid suppression after three days of therapy [1]. Unlike H2-receptor antagonists, PPIs inhibit both basal and meal-stimulated acid secretion, providing sustained pharmacodynamic effects for 16–18 h, which allows for once-daily dosing. Notably, tolerance does not develop because receptor activation does not produce downregulation [2].

Since omeprazole’s introduction in 1989, PPIs have become central in managing acid-related disorders [1, 3]. Compared to older agents (synthetic prostaglandin analogs, anticholinergics, H2 antagonists), PPIs demonstrate superior tolerability, safety, and efficacy in acid suppression [1].

In Spain, PPIs are indicated for gastroesophageal reflux disease (GERD), gastroduodenal ulcer, Helicobacter Pylori eradication treatment, Zollinger-Ellison syndrome and prevention of gastropathy due to chronic non-steroidal anti-inflammatory drugs (NSAID) use in high-risk patients (age over 65, history of uncomplicated gastroduodenal ulcer, concomitant use of acetylsalicylic acid, corticosteroids, and/or oral anticoagulants) [4].

Worldwide use

It is estimated that 20–80% of global PPI users lack a valid medical indication [2, 3, 5–8]. The economic impact of unnecessary PPI use is substantial. For example, in France, PPI sales increased by ~ 20% from 2010 to 2013, with 80 million packs sold in 2013, ranking esomeprazole, omeprazole, and pantoprazole among the top 30 dispensed drugs; in 2015, 29.8% of the French adult population (more than 15,000,000 inhabitants aged 18 and over) consumed at least one pack of PPIs. Almost half of them were new users of these drugs, and their indication was not documented in almost one-third of the cases [5].

In England (2006), PPIs accounted for £425 million out of £7 billion in drug spending (6%) [7]. It is the third most prescribed group of drugs in the United States [9]. Its cost is increased by the use of brand-name drugs instead of generics. It has been estimated that the 5-year excess cost of using brand-name PPIs in the United States exceeded $47 billion [10]. According to British data, it is estimated that £2 billion is potentially spent annually unnecessarily on these drugs throughout the world [11]. In 2009, the use of these drugs accounted for an expenditure of $13.6 billion worldwide [2].

Inappropiate indications

Many studies highlight the frequent overuse of PPIs in both hospital and primary care settings [2]. In France, misuse of PPIs has been documented to range from 40% to more than 80%, depending on the definition used [5].

A US study found that 73% of nearly one million hospitalized patients received PPIs without proper indication. The risk of Clostridium difficile infection was higher among those prescribed PPIs; 56% lacked a valid indication, primarily for gastrointestinal bleeding prophylaxis [9].

Inappropriate continuation of PPI therapy post-discharge is pervasive, as shown in longitudinal studies assessing both hospital and primary care practices [12–14]. Functional dyspepsia and chronic prescriptions—often initiated by family physicians—are significant sources of unnecessary PPI use. PPIs provide low success rates (10–30%), so shorter, low-dose regimens with regular re-evaluation are advisable [2].

Additional inappropriate indications might include the prescription of corticosteroids without concurrent NSAIDs [12, 13]; NSAID gastroprotection below age 65 without bleeding risk [2], anticoagulants/antiplatelets without gastric risk factors [14, 15] or SSRIs, portal hypertension in cirrhosis, and other off-label uses [2].

The problem of PPI overuse encompasses several distinct but overlapping phenomena: inappropriate prescribing (prescribing without a valid clinical indication), misuse (prescribing outside validated indications or at incorrect doses), and overuse (prescribing at excessive duration or with automatic renewal without clinical reassessment). Deprescribing, in turn, is not limited to complete discontinuation, but encompasses dose tapering, transition to on-demand therapy, reduction in administration frequency, and step-down to less intensive therapeutic strategies. This conceptual distinction is relevant because a patient may experience overuse without their prescription being formally classified as ‘inappropriate’ under a binary classification framework. In this study, the primary outcome captures only the binary dimension; the economic outcomes may therefore be more sensitive to the full spectrum of prescribing behaviour changes.

PPI deprescribing guide

Deprescribing is a planned, supervised process to reduce or cease medications causing harm or no longer offering benefit [6, 16]. The aim is to reduce drug burden and adverse effects, while maintaining or improving quality of life [6].

The practice of PPI deprescribing has evolved very favourably in the last few years, but in the recent past clinical guidelines lacked robust recommendations for tapering or discontinuing PPIs, with few evidence-based deprescribing protocols available. The Canadian practice guideline (2017) supports clinicians in safely deprescribing PPIs [6]. This resource targets adults on continuous PPI therapy (more than 28 days), stratifying recommendations by evidence strength.

According to the aforementioned guideline, indefinite prescription is considered to be correct:

• Barrett’s esophagus.

• Severe esophagitis (Los Angeles grades C, D).

• Documented history, with endoscopy, of bleeding gastroduodenal ulcer.

• Chronic use of NSAIDs in patients at risk of bleeding:

1. Over 65 years of age.

2. Previous uncomplicated gastroduodenal ulcer.

3. Concomitant use of acetylsalicylic acid, corticosteroids, and oral anticoagulants.

Prescriptions not meeting these criteria should be targeted for deprescribing (see algorithm Fig. 1). The Canadian guide [6] has been used in the training project described in this article.

Fig. 1.

Proton Pump Inhibitor (PPI) Deprescribing Algorithm

Objectives

Primary

Assess PPI prescribing practices in the Canary Islands primary care after an educational intervention.

Secondary

1. Analyze inappropriate indefinite PPI prescriptions.

2. Develop an open online deprescribing course.

3. Evaluate course impact.

4. Assess the economic consequences of PPI prescribing.

Methods

PPI prescription analysis in the Canary Islands

Prevalence data for PPI prescriptions from primary care physicians (February 2020) were obtained and coded for evaluation. Inappropriate chronic prescriptions were defined as those lacking an indication in accordance with established guidelines. In the Canary Islands, medical records and prescriptions—whether for acute, chronic conditions—are computerised. This enables the present study to carry out a two-pronged assessment: qualitative (correct/incorrect indefinite prescriptions according to the guidelines used) and quantitative (costs), before, during and after the intervention.

Intervention design and implementation

A prospective, randomized, Massive, Open, Online, educational intervention (MOOC format) was developed, focusing on PPI deprescribing guidelines and strategies, including their contents and structure (Fig. 2). The nature of this MOOC format prevents students from progressing to the next task unless they have successfully completed the previous one (rating 8/10). Consequently, only those who successfully completed all the tasks could be assessed.

Fig. 2.

PPI deprescribing course content

Study design and sample selection

The course was offered on the Health Service intranet. Registrants were assigned to either the intervention or control cohorts (with an anonymous comparison group), matched for baseline characteristics to avoid selection bias. This approach aims to compare the behaviour of a group of doctors who are highly motivated to engage in deprescribing—so much so that they are willing to undertake a voluntary, academically demanding course— (intervention and control cohorts) with a group showing no such motivation (anonymous cohort).

A total of 70 family physicians registered for the course, covering three different islands. They were randomly divided into two groups; the first (“intervention group”, 35 physicians) received the course for a period of three weeks and had follow-up assessments at 12 and 24 weeks, whereas the remaining group (“control group”, 35 physicians) did not receive the course until 12 weeks after the “intervention group”. The randomisation was simple and allocation was not concealed.

Furthermore, out of the primary care physicians who showed no interest in the course, a group of 40 was selected (“anonymous group”) from those working in the same health areas, matched by work experience, age, and educational background. The data of PPI prescriptions of those physicians were analyzed after masking their personal identities. The evaluation of the outcomes of prescriptions was performed initially at 12 and 24 weeks for each group. Finally, due to incomplete data, only complete data from 18 physicians in the “intervention group”, 16 in the “control group”, and 37 in the “anonymous group” could be included in the study. Exclusions included physicians without fixed medical quotas, those absent from practice during follow-up, and those with medical quotas below 900 patients (Fig. 3).

Fig. 3.

Sample and groups under study

The rationale for the three-group design was to separate two distinct sources of variation: the effect of the training itself, and the effect of physician motivation. The intervention and control cohorts both comprised physicians sufficiently motivated to voluntarily register for a demanding online course; their comparison isolates the effect of the training (received at different time points). The anonymous cohort — physicians who expressed no interest in the course — provides a comparator for the combined effect of motivation and training on prescribing behaviour. Because the control group received the training 12 weeks after the intervention group, the design effectively becomes a partial crossover after week 12. Comparisons between the intervention and control groups at 24 weeks should therefore be interpreted with caution, as both groups had received the training by that point. This limitation is discussed further in the Limitations section.

Variables/algorithm

Prescriptions were classified (appropriate/inappropriate), quantifying omeprazole and other PPI consumption, and associated costs. Omeprazole is the best-selling PPI in Spain, and it is also the cheapest and most cost-effective. For this reason, its use is promoted by the National Health Service throughout the country. The course offered by this study strongly promotes its use over other PPIs and devotes particular attention to it in two of its modules: Module 1 (Pharmacological aspects of PPIs) and Module 3 (Pharmacoeconomic aspects of deprescribing) (Fig. 2).

Analysis of the complete dataset confirmed that omeprazole accounted for approximately 65% of all PPI prescriptions at baseline across the three groups, making it the dominant molecule and justifying its separate reporting.

The measurement of the variables was performed in the three groups at the beginning of the course and twelve weeks after the course ended (1st edition) for the “intervention group”. In addition, an evaluation of the three groups was conducted twelve weeks after the course was given to the “control group” (2nd edition), i.e., 24 weeks after the completion of the 1st edition of the course. The number of PPIs prescribed correctly and incorrectly, and their percentage, and the difference between the start of the course and the two subsequent cuts were assessed, specifying how many of these were omeprazole. The cost in euros was also evaluated, estimating the difference in the three groups with respect to the results obtained before the intervention (Table 1).

Table 1.

Chronological scheme of the variable measurements

	Start	Course Intervention group	12 weeks	Course Control group	12 weeks
Groups	1/18/21	1/18/21 to 02/14/21	5/10/21	4/19/21 to 5/17/21	8/02/21
Intervention	Evaluation	Course	Evaluation	Regular work	Evaluation
Control	Evaluation	Regular work	Evaluation	Course	Evaluation
Anonymous	Evaluation	Regular work	Evaluation	Regular work	Evaluation

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics (version 26.0). The primary outcome — the percentage of appropriate chronic PPI prescriptions — and the secondary cost outcomes were analysed using repeated-measures analysis of variance (RM-ANOVA) with Group (intervention, control, anonymous) as the between-subjects factor and Time (baseline, 12 weeks, 24 weeks) as the within-subjects factor. The significance threshold was set at α = 0.05. Prior to the main analysis, the assumption of normality was assessed using the Shapiro-Wilk test and the assumption of sphericity was evaluated using Mauchly’s test; where sphericity was violated, the Greenhouse-Geisser correction was applied. Effect sizes are reported as partial eta-squared (η²p), with values of 0.01, 0.06, and 0.14 considered small, medium, and large, respectively. For between-group pairwise comparisons at each time point, 95% confidence intervals are reported using parameter estimates from the RM-ANOVA model, with the anonymous group as the reference category.

Ethical aspects

Participants were fully informed, both verbally and in writing, and provided informed consent. Approval was granted by the Ethics Committee for Research with Medicines (CEIm) of Santa Cruz de Tenerife province / Hospital Universitario de Canarias (CHUC). (CHUC_2019_102). Anonymity and data protection principles were maintained; only aggregate data were accessible. Although this study does not imply clinical research, but only educational interventions, it was conducted in accordance with the Declaration of Helsinki (Revision Fortaleza, 2013) in the aspects required.

Results

To assess the appropriate chronic prescribing of PPIs in PC physicians of the Canary Islands Health System (SCS)

The entire database of clinical records from the SCS (over 2.2 million records) was anonymously assessed: only 8% of indefinite PPI prescriptions met the appropriate criteria according to the guideline used [6].

Across all 71 physicians included in the analysis (18 in the intervention group, 16 in the control group, and 37 in the anonymous group), a total of approximately 54,500 PPI packages were assessed across the three measurement timepoints (baseline, 12 weeks, 24 weeks), of which approximately 65% corresponded to omeprazole.

Develop an open online course for training in the correct use of PPIs

The course was developed in accordance with the characteristics outlined in the Methodology section, as a Massive Open Online Course (MOOC). It was available through an official MLS of the SCS (Fig. 3). It was offered to all Primary care physicians through the intranet of the Health System (1,880 physicians covering a population of 2.2 million inhabitants distributed across seven islands).

The course was very highly rated by the doctors who completed it in both sessions.

In the satisfaction survey of the 26 participants (only 1/27 did not complete it) in the first session of the course, we received the following responses:

1. The course was focused on topics they found interesting ‘often’ or ‘almost always’ : 96%.

2. What they learnt on the course was important for their professional practice ‘often’ or ‘almost always’: 96%.

3. The course helped them learn how to improve their professional practice ‘often’ or ‘almost always’: 100%.

Regarding knowledge assessment: the MOOC design required participants to achieve a minimum score of 8/10 on each module questionnaire before advancing to the next module. Completion of all module questionnaires was therefore a prerequisite for inclusion in the study analysis, serving as a minimum competency threshold. Formal pre/post knowledge scores by module were not collected as a separate outcome variable.

Evaluate the outcomes of the course

The difference in appropriate chronic prescribing of PPIs did not change in the “intervention group” at 12 weeks and 24 weeks after completing the training, nor in the “control” and “anonymous groups”, maintaining very similar values between the three groups (Table 2). Using repeated measures ANOVA (MR-ANOVA), no significant differences were found between groups (p = 0.186; η²p = 0.043), nor between times (p = 0.155; η²p = 0.013 for baseline to 12 weeks; η²p = 0.034 for baseline to 24 weeks), nor between the evolution of the groups by time (p = 0.676). The behaviour of appropriate chronic prescribing of omeprazole followed the same pattern as for PPIs as a whole.

Table 2.

Correct chronic prescription PPIs

	Mean % PPI correct (initial) SD (SE)	Mean % PPI correct (12w) SD (SE)	Mean % PPI correct (24w) SD (SE)
Intervention	6,4 3,18 (0,71)	6,2 3,33 (0,74)	6 2,96 (0,66)
Control	7,4 3,18 (0,73)	7,4 3,74 (0,86)	7,4 3,85 (0,88)
Anonymous	8,3 3,78 (0,60)	7,9 3,81 (0,60)	7,8 3,73 (0,59)

Percentage of correct PPIs at baseline at 12 and 24 weeks

SD Standard deviation, SE Standard error of the mean

For the between-group comparison at each timepoint, with the anonymous group as reference, the intervention group showed consistently lower rates of appropriate prescribing (difference of approximately − 1.9% points at baseline, 95% CI: −3.81 to 0.01, p = 0.051; −1.7 pp at 12 weeks, 95% CI: −3.70 to 0.31, p = 0.097; −1.8 pp at 24 weeks, 95% CI: −3.71 to 0.20, p = 0.077). These differences approached but did not reach statistical significance, and the effect sizes were small (η²p < 0.05) throughout.

However, the cost of prescribing PPIs showed a modest reduction in the “intervention group” compared to the other two groups at 24 weeks after the training (Table 3). The SCS calculated the total expenditures on PPI for the 12 weeks preceding the intervention for each participant’s medical quota. In the “intervention group”, each physician prescribed PPI for a median value of €82 less during the 24 weeks of the study. In contrast, the two control groups increased the PPI prescribing expenditures by €15 (“control group”) and €160 (“anonymous group”) (Table 3). Using repeated measures ANOVA (MR-ANOVA), these differences are statistically significant (initial period and 12 weeks) (p = 0.020). In contrast, the evolution between week 12 and week 24 in the three groups did not differ significantly (p = 0.366). The difference in the evolution between the initial period and 12 weeks is due to the differentiated behaviour of the “intervention group” (p = 0.006) compared to the other two groups (“control” and “anonymous”), which do not differ from each other (p = 0.409).

Table 3.

PP cost by group and total measures in euros

	Total PPI cost initial incorrect(correct)	Total PPI cost 12 weeks incorrect(correct)	Total PPI cost 12–24 weeks incorrect(correct)	Total cost difference at 24 weeks. Group (per participant)
Intervention(18 participants)	78.976 73.684 (5.292)	78.353 73.260 (5.093)	77.497 72.614 (4.883)	-1.479 (-82)
Control (16 participants)	68.915 63.677 (5.238)	70.414 65.203 (5.211)	69.150 63.964 (5.186)	+ 235 (+ 15)
Anonymous (37 participants)	149.772 138.389 (11.383)	156.944 145.644 (11.300)	155.704 144.649 (11.055)	+ 5.932 (+ 160)
Total (71 participants)	297.663 275.750 (21.913)	305.711 284.107 (21.604)	302.351 281.227 (21.124)	+ 4.688 (+ 66)

PPI cost in euros in each sample group at baseline, 12 and 24 weeks. Cost corresponding to the percentage incorrect/correct consumption in each group. Total cost of all measurements

Assessing the economic consequences in the Canary Islands

If the differences in costs of the intervention group between the pre-intervention costs and the costs at 24 weeks after intervention were extrapolated to the total number of medical quotas in the Canary Islands Health Service (1,188), the savings in a year could have represented a total of €211,504 for all medical quotas in the Canary Islands (Table 4). Were these results generalisable to all medical quotas in the Canary Islands Health Service (1,188), and assuming that the cost reduction observed in the intervention group were sustained over a full year, the potential savings could represent approximately €211,504 annually. This projection rests on several assumptions — a uniform effect across all physicians, stability of the effect beyond 24 weeks, and the absence of pandemic-related distortions — none of which can be validated from the present data. This is a simple mathematical projection and should therefore be treated with the utmost caution, particularly given the exceptional circumstances created by the SARS-CoV-2/COVID-19 pandemic.

Table 4.

Projected decrease in expenditure for the whole Canary Islands (1,188 medical quotas) measured in euros at 24 weeks and one year, according to the intervention group

	Decrease in expenditure in the intervention group at 24 weeks (18 participants)	Projected decrease in expenditure on total medical quotas at 24 weeks (1,188)	Projected decrease in expenditure on total medical quotas at 1 year (1,188)
Total	-1.479	-97.617	-211.504
*The savings in the intervention group 24 weeks after the training could have translated into a saving of €211.504 per year (in the total of 1,188 medical quotas)

Discussion

The most important difference between this study and those published to date [17] that evaluated PPI deprescribing strategies is that all these studies selected a sample of patients with inappropriate PPI prescribing and proposed different protocols to replace, reduce, or eliminate them, and then measured the results obtained. In contrast, this study draws its sample from a group of primary care physicians who are trained in a specific PPI deprescribing strategy and then measures the difference in terms of their undefined correct/incorrect prescribing of these drugs. In this study, we have considered targeting prescribers rather than the patients themselves, in an attempt to multiply the effectiveness of our training strategy.

The 71 physicians included in the analysis collectively managed over 54,000 PPI packages across the three measurement timepoints, underscoring the potential public health impact of even modest improvements in prescribing behaviour at the individual physician level.

Analysing published studies on deprescribing strategies [17], we consider that the strengths of this study are predominantly two: using a “control group” and an “anonymous group” to measure the results of the intervention, thus avoiding selection bias, and adding a cost assessment to measure the outcome of the proposed intervention. While no qualitative change in the appropriate chronic prescribing of PPIs was found after the intervention, a reduction in the number of incorrectly prescribed packs was detected in the groups that received the training, compared to a sustained increase in the group that did not receive the training. These changes in the prescribing behaviour, in turn, lead to a reduction in the cost associated with chronic PPI prescribing in the groups that received training and a marked increase in the group that did not.

One possible explanation for this phenomenon is that in order to adjust the prescription to the criteria set out in the guidelines used, the doctors had to carry out a significant review of the medical records, something complex in the absence of their patients, given that the consultations during the development of this study were mostly by telephone (SARS-CoV-2/COVID-19). The unpredictable, wave-like progression of the pandemic over the three years following its emergence in Spain in March 2020, with its constant shifts in public health conditions and healthcare provision—ranging from periods of greater openness to periods of stricter restrictions—enabled us to remain committed to completing the study. Furthermore, the fact that the core of the project involved delivering an online course compatible with the pandemic situation was decisive in our decision to continue.

The option of postponing the study until conditions normalised was considered but ultimately not pursued, principally because the end of pandemic restrictions could not be predicted at any point during 2020 or 2021, and because the online format of the intervention was fully compatible with remote working conditions. We acknowledge, however, that conducting a deprescribing intervention during a period when clinical consultations were predominantly remote, patient records were incompletely accessible, and automatic renewal protocols were in place constituted a structural impediment to the application of the training. It is plausible that the anonymous group was differentially less affected by these constraints, since their baseline behaviour was already unlikely to involve proactive medication review. The differential impact of these conditions on the three groups cannot be quantified from the available data, and this uncertainty should be considered when interpreting the primary outcome.

In this regard, it is worth noting that the most frequently reported barrier by researchers in studies on deprescribing strategies was access to patient information [17]. Mostly, the researchers used electronic records, but they systematically met the obstacle that these records are not accessible to all levels of care. Thus, the records, used by the most diverse specialists attending the same patient, known as polydoctoring [18], have different formats and contain information in which not all the responsible doctors participate. This is also the case in our setting, where access to medical records and prescriptions is not available to all responsible physicians, and most of them are not kept up to date.

It may also be relevant to note that the qualitative assessment in this study is binary (correct/incorrect chronic prescription) and does not allow for the identification of the method by which the clinician arrived at deprescribing (complete withdrawal, gradual reduction or on-demand prescription), nor does it account for overprescription in the case of correct prescription or a change of drug. It seems reasonable to assume that doctors trained using the proposed course may have implemented measures to improve the use of PPIs that the study does not detect, the results of which could be objectively assessed through longer-term measurements.

When considering a deprescribing strategy, it is crucial to consider the feasibility of the procedure. As stated in the study by Reeve et al. [19], it is critical to assess how much time and effort is devoted to the deprescribing task since, although it is relevant that the protocols are clear and straightforward, it is also decisive that they can be carried out in ordinary consultations, avoiding duplicated parallel practices that drastically reduce the efficiency of these procedures. A recently published study carried out in France examines a deprescribing strategy aimed at educating patients and doctors, with very promising results [20]. However, one might question the feasibility of sustaining public awareness campaigns over time, particularly given their cost.

Given the high volume of use of these drugs, a slight change in prescribing habits can make a significant difference in terms of resource consumption. We believe that training doctors appropriately in how to use these drugs and how to deprescribe them may be a practical way to deal with their massive and increasing use. To make prescribing not only more efficient but also qualitatively more appropriate, it would be necessary to monitor the samples studied more closely and for a more extended period after training.

Limitations

The primary and most significant limitation of this study was the SARS-CoV-2/COVID-19 pandemic, which was at its peak in Spain during the study’s development and throughout the outcome measurement period.

The pandemic brought about a significant structural change, as primary care consultations shifted to telemedicine, impeding direct deprescribing efforts and routine reviews of chronic medications. Automatic refill protocols further hindered the evaluation of medications. Furthermore, physicians have had to severely limit the time they spent on any other pathology than COVID-19 due to its importance and lethality.

Prescribing measurements were practice-based rather than individual provider-based, and prescribing appropriateness depended on diagnoses entered into electronic medical records, which were variably updated, potentially leading to the misclassification of prescriptions as inappropriate.

It also seems worth noting that, of the 70 doctors enrolled in the course, due to incomplete data, results could only be assessed for 34 (18 from the “intervention group” and 16 from the “control group”). It seems reasonable to assume that this reduction in the sample size may increase the likelihood of committing a Type II error, which means that the study lacks the necessary power to demonstrate a significant difference in the appropriate chronic prescription of PPIs between the cohorts.

A post-hoc power analysis confirmed that, with the achieved sample sizes of n = 18 (intervention) and n = 16 (control), the study had approximately 15% power to detect a 1%-point difference in appropriate prescribing rates (SD ≈ 3.2%; α = 0.05, two-tailed). To achieve 80% power with these sample sizes, a true between-group difference of approximately 3.0–3.5% points would have been required. The absence of a statistically significant finding on the primary outcome should therefore be interpreted as reflecting inadequate statistical power rather than as evidence of no effect.

Because the control group received the training at 12 weeks, the design becomes a partial crossover after that point. The 24-week comparison between intervention and control groups should be interpreted with caution, as both groups had received the same training by week 24, and any remaining difference may reflect only the timing of exposure rather than a sustained training effect.

The formal evaluation of training efficacy was limited. No pre/post knowledge assessments by module were collected as study outcome variables, beyond the mandatory minimum competency thresholds built into the MOOC format. This limits the ability to attribute the observed cost reductions directly to knowledge acquisition. Additionally, knowledge and satisfaction data from the second course edition (control group) were not systematically collected, preventing a complete per-protocol analysis.

Finally, to more accurately assess the changes in prescribing after training, it would be helpful to monitor the results over a more extended period, specifically 12 and 24 months later.

Conclusions

Inappropriate chronic prescribing of PPIs accounts for approximately 92% of prescriptions in the Canary Islands, in line with the current trend in all Western countries. The cost of inappropriate prescribing is an unjustifiably high burden on public health services.

Training on PPI deprescribing was designed for primary care doctors in the Canary Islands; at that point the study was unable to demonstrate an statistically significant improvement by increasing the percentage of appropriate chronic prescribing of these drugs. However, it did show a reduction in the number of incorrectly prescribed packages of PPIs among the physicians who received it. It also showed a statistically significant reduction in total PPI expenditure in the group that received it first. It is reasonable to conclude that these results may have been influenced by the logistical difficulties arising from the pandemic.

Given the high volume of use of these medicines, which are not free from side effects, any strategy aimed at improving their management by physicians could have a significant beneficial effect on the health of the population, and, if the findings of this study are confirmed, an exponential impact on the consumption of public resources. Any such training efforts should be closely monitored and evaluated over the medium and long term, at least 12 and 24 months. Among its main objectives should be to reduce the percentage of unjustified chronic prescriptions and to increase the percentage of prescribing the cheapest, equally effective, and safe drug.

Authors’ contributions

Both authors developed the study design, provided training to the doctors, and analysed the datasets. MD: writing-original draft. ES: project administration, supervision, and writing-review and editing.

Funding

No funding was available for this study. It was done within the research time allotted to the authors by their institutions.

Data availability

The datasets analysed during the current study are available from the corresponding author without undue reservation.

Declarations

Ethics approval and consent to participate

Participants were fully informed, both verbally and in writing, and provided informed consent. Approval was granted by the Ethics Committee for Research with Medicines (CEIm) of Santa Cruz de Tenerife province / Hospital Universitario de Canarias (CHUC). (CHUC_2019_102). Anonymity and data protection principles were maintained; only aggregate data were accessible. Although this study does not imply clinical research, but only educational interventions, it was conducted in accordance with the Declaration of Helsinki (Revision Fortaleza, 2013) in the aspects required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.