1 Introduction
Non-steroidal anti-inflammatory drugs (NSAIDs) and low-dose acetylsalicylic acid (LDASA) may cause serious gastrointestinal (GI) adverse effects such as dyspepsia, peptic ulcers, peptic bleeding and ulcer perforations [
1,
2]. The use of these drugs substantially contributes to the number of potentially preventable, drug-related hospital admissions [
3,
4]. Proton pump inhibitors (PPIs) effectively prevent GI complications in NSAID and LDASA users [
5,
6]. Consequently, PPIs are recommended as a protective co-medication in patients at risk of GI damage during NSAID or LDASA treatment [
5‐
9]. In the Netherlands, since 2012, PPIs have been recommended for all patients using NSAIDs aged 70 years and older and for those using LDASA aged 80 years and older. For patients using NSAIDs aged 60 years and older, and LDASA users aged 70 years and older, PPIs are recommended in combination with GI risk-increasing co-medication or co-morbidity [
9]. This policy was supported by the development of quality indicators for general practitioners and community pharmacists [
10,
11].
Incorporated in the prescribing and dispensing process, these quality indicators measure the proportion of patients with a protective co-medication in NSAID or LDASA users at increased GI risk. Over the years in the Netherlands, the concomitant use of PPIs in (older) patients prescribed NSAIDs and LDASA has considerably increased. In NSAID users at risk of GI adverse effects, PPI use increased from 55% in 1996 [
12] to 85% in 2012; the proportion of susceptible LDASA users with a PPI co-medication grew from 8% in 2000 to 31% in 2012 [
10]. In 2013, 2.75 million people in the Netherlands (17% of the total population) used PPIs with on average 235 daily defined doses [
13]. In 2015, the PPI omeprazole was the most frequently dispensed drug in the Netherlands with 5.2% of all drugs dispensed [
14].
Although PPIs effectively reduce upper GI damage in patients at risk [
7,
15], a number of concerns have been raised concerning their safety [
16]. In several studies, PPIs were shown to modestly increase the risk of bone fractures (hip, wrist and vertebrae) in long-term users [
17‐
21]. Furthermore, there is some evidence that PPIs increase the risk of community-acquired pneumonia (CAP), especially shortly after initiation of treatment [
22‐
25].
Finally, there is a consensus that PPI use without a disease or medication-based indication should be avoided [
26]. This inappropriate PPI use has been described before [
27‐
31]. The cost effectiveness to add a protective PPI co-medication to NSAID or LDASA treatment in patients with an increased risk of GI damage has been well demonstrated [
32‐
34]. However, insight into the potential cost savings and benefits from the discontinuation of inappropriate PPI use following cessation of NSAID or LDASA treatment is lacking.
Consequently, the present study aimed to, first, document the extent of inappropriate PPI use initially related to NSAID and LDASA treatment in the Netherlands. Second, to assess the benefits of the strategy to stop inappropriate PPI use in former NSAID and LDASA users at risk of GI adverse effects as compared to no intervention using a Markov model. A budget impact analysis (BIA) was performed for the potential cost savings resulting from the successful use of interventions to prevent inappropriate PPI use.
4 Discussion
In 2014, a substantial number of patients in the Netherlands who had been prescribed an NSAID or LDASA in combination with a PPI to prevent GI adverse effects were found to have inappropriately continued their PPI use after cessation of NSAID or LDASA treatment. The highest percentage (16.4%) of patients inappropriately using a PPI was found in the oldest age category (> 80 years) after NSAID discontinuation. Our analysis showed that discontinuation of inappropriate PPI use resulted in substantial cost savings and improvement of quality of life. From the costs’ perspective, this was to be expected because of the savings of the direct medication costs for the PPIs. Additional gains in QALYs and cost savings were achieved by preventing hip fractures and CAP as PPI adverse effects.
Earlier studies reported substantially higher percentages of inappropriate PPI use. In a recent US study, about 80% of PPI continuation was found to be inappropriate because of the absence of a valid indication after hospital discharge [
29]. An Australian study showed that 63% of all PPI prescriptions were not based on actual guideline recommendations [
30]. An Irish study considered 33% of PPI use as inappropriate [
31]. The considerable lower percentages found in the present study might be explained by the fact that our study focused on PPI use as a co-medication to NSAID or LDASA treatment. Thus, inappropriate PPI use due to other reasons such as PPI continuation after hospital discharge was not taken into account. A more reticent prescribing attitude in the Netherlands may also have contributed to the lower percentages of inappropriate PPI use observed in our study. Finally, the quality of prescribing, and dispensing, may be higher in the Netherlands as has been described previously for other aspects of inappropriate prescribing [
47].
Except as a co-medication in patients using an NSAID or LDASA, PPIs could also have been prescribed for other reasons such as Zollinger–Ellison syndrome, Barret’s esophagus or esophagitis. A limitation of dispensing data used in the present study is that they did not provide information on these disorders. However, as their prevalence is low, patients with these disorders were unlikely to have substantially influenced our results. Furthermore, we have only included patients who started PPI treatment and who had not been dispensed a PPI in the prior 12 months. These patients also had to have a period of NSAID or LDASA use within a narrow time window of 10 days around the PPI treatment initiation. Thus, patients included were most likely to use PPIs as a protective co-medication.
According to the guidelines, PPI use should be periodically evaluated to prevent long-term use and the development of rebound effects on gastric acid production. As the result of strict inclusion criteria, inappropriate PPI continuation might have been missed and our findings therefore may be too conservative. In addition, we only included costs and QALYs for hip fractures but not for other fractures that might occur as PPI side effects. This could also make our estimates for the consequences of PPI adverse effects too conservative. Finally, drug prices used in the BIA are only an indication for the real prices paid by the health insurers.
Nevertheless, unlike list prices, they are probably the best indication of the actual prices in the Netherlands, which as such are not transparent because of direct price negotiations between manufacturers on the one hand, and health insurers or hospitals on the other hand. A strength of this study is that we used the data of nearly 90% of all Dutch community pharmacies. Our results therefore are representative of the Netherlands. Because our information was extracted from routinely collected dispensing data, reporting bias is also unlikely. In our model, we assumed that event rates were constant, whereas this is likely to change over time. However, because to our knowledge there are no data available on the rates over time, under the current conditions the assumptions are the best available.
The BIA showed that an intervention to discontinue inappropriate PPI use related to previous NSAID or LDASA treatment would cost up to €29 extra per patient. These costs are determined by the costs already invested for an intervention to stop inappropriate PPI use, minus the savings resulting from the prevention of PPI-induced adverse effects and their detrimental consequences. The cost–utility analysis shows that this intervention will easily pay for itself in the following years (as indicated by the negative incremental costs). Pharmacists in cooperation with general practitioners could effectively intervene by means of adequate patient counselling or increasing awareness, for instance by a letter, to stop the inappropriate use of PPI [
48].
In a study in outpatient clinics of Australian hospitals, a strategy to reduce the inappropriate use of PPI by more intensive patient counselling proved effective [
37,
49]. Based on our results, in Dutch community pharmacies, by monthly searches of dispensing data, 18 potentially inappropriate PPI users per pharmacy on average can be easily identified. Subsequently, in consultation with the general practitioners and in the absence of other reasons for PPI discontinuation, these patients could be labelled for not dispensing further PPIs. With regard to the BIA, it is important to note that the prices used to calculate the BIA more or less reflected the actual prices used by the healthcare insurers.
In the Netherlands, the prices of generic medicines are very low as the result of the legal power of health insurers to impose price-restricting measures. For a given drug, these prices may differ between insurers while the basis for pricing is not transparent. We believe that under these conditions the prices used in the present study are the best available. As the risks of adverse effects increases with older age, discontinuation of inappropriate PPI use proved more cost effective in older patients.
Compliance with Ethical Standards