Proton pump inhibitor prescribing patterns in the UK: a primary care database study

To determine the prevalence and pattern of proton pump inhibitor (PPI) prescription and the practices employed to reduce PPI use in the UK general population.


INTRODUCTION
The introduction of proton pump inhibitors (PPIs) has revolutionised the management of acid-related gastrointestinal disorders. 1 In the UK, 11 126 000 prescriptions for PPIs were dispensed in 2000, 2 and this increased to 43 127 000 in 2011. 3 Although the expenditure on PPIs has decreased in the UK since 2006 as a result of government efforts to encourage the use of low-cost generic PPIs, 4,5 there has still been an overall increase in the total number of PPI prescriptions dispensed. 4 For instance, in 2010, PPIs became one of the top 20 drugs with the greatest net ingredient cost in the UK. 4 The importance of reducing any overuse of PPIs is not only limited to the associated costs but also to the risks of taking the drugs on a long-term basis. 6 As PPIs have become commonly used for long-term maintenance, concerns have been raised about the safety of such use. 6 Studies have showed that PPI use is associated with malabsorption of vitamins and minerals, 7,8 an increased risk of infections, such as pneumonia 9 and enteric infection, 10 and an increased fracture risk. 11,12 These potential side effects can be minimised through appropriate prescription practices in terms of stepping down the dose or stopping longterm treatment altogether.
As a consequence of this dramatic increase in PPI use and the associated potential risks, clinical guidelines in the UK have recommended rationing the use of the PPI in the primary care setting, either by stepping down the dose or stopping treatment all together. 13 However, very few research studies have examined the extent to which the clinical guidelines are being followed in the UK. [14][15][16][17][18][19] The aim of this study was to determine the prevalence of PPI use and assess the practices employed to reduce PPI use in the general UK population. It is anticipated that research of this nature will help to inform future attempts to moderate the use of PPIs.

Study type and data source
We conducted an observational study with repeated cross-sectional analyses to estimate the prevalence of PPI use annually and a cohort design to describe the patterns of PPI utilisation by using data from the UK Clinical Practice Research Datalink (CPRD). 20 CPRD is a large database drawn from the computerised records of primary care practices throughout the UK and encompassing a representative sample of around 6% of UK population. [21][22][23] The CPRD comprises data about patients' medical diagnoses, GPs' prescriptions, investigations, hospital referrals and discharges, together with basic demographic information. The information on prescriptions includes their issue dates, the drug prescribed, numeric daily dose, daily quantity and the number of packs/pack size prescribed. Many studies have validated CPRD for use in pharmacoepidemiological research. 21,24

Study population
We studied adult patients with at least 1 month of prospective records after either the date of their current registration or the date after the practice became 'up to standard' on CPRD 21 whichever was the latest and an 'acceptable' registration status as defined by CPRD 21 between 1 January 1990 and 31 December 2014. This population formed our denominator for studies of prevalence. Patients who received ≥1 PPI prescription (BNF 1.3.5 25 ) were classified as exposed subjects in the study (i.e. the numerator).

Prescription duration
The earliest PPI prescription for each patient was considered their index date. Prescription duration was taken as the number of treatment days recorded by the GP or calculated from the prescribed quantity and numeric daily dose prescribed. If information on both was missing, the individual median duration was imputed. The duration was recalculated if the calculated prescription duration was less than or equal to 7 days assuming that the prescription quantity was referring to the number of individual product packs prescribed.

Prescribing patterns
To describe the prescribing practices of long-term PPI use in general practice in terms of discontinuation, stepping down or switching to histamine 2 receptor antagonists (H2RA), we identified new PPI therapy users, that is, patients with at least 12 months of registration on CPRD prior to their index date who had ≥5 years of prospective follow-up data.
The NICE guidelines 13 were used to determine what constitutes expected long-term PPI use within this study. PPIs are used for the short-term management in conditions such as dyspepsia, gastro-oesophageal reflux disease (GORD) and gastric and duodenal ulcers. Long-term PPI therapy is often prescribed to prevent recurrence of GORD complications and as prophylactic therapy to prevent peptic ulcers in patients who are coprescribed non-steroidal anti-inflammatory (NSAID) therapy. 13 Exposure to PPIs was considered to begin on the date of a prescription for them and end after its calculated duration unless another prescription was issued ≤30 days after this date in which case we considered exposure continuous. We refer to one set of continuous prescriptions as one course. Courses were classified as short (<12 months) or long (≥12 months); this time period being chosen as 12 months is the minimum frequency with which NICE recommends that these prescriptions should be reviewed and stopped or stepped down if possible. Individuals receiving exclusively short courses were classified as short-term users, while individuals who received at least one long course were classified as long-term users even if their records contained other short courses.
Discontinuation (no subsequent PPI prescription issued within 30 days after the end of the previous one) was categorised as temporary (patients subsequently re-prescribed PPI) or permanent (no further prescriptions received up to the end of the patient's follow-up). A step down of PPI therapy was defined as a reduction in daily dose of the subsequent PPI prescription. If a following prescription was for a different PPI, the dose was converted to an equivalent dose based on the recommended dosing in the BNF. 25 A successful step down was defined as maintaining the stepped down dose for 12 months from the step-down date. Lastly, a switch to H2RA medication was defined as receiving H2RA prescription within 1 month before or after discontinuation or stepping down attempt.

Covariates
We abstracted data on patients' age at the index date (in 10-year age bands), gender and socioeconomic status (derived through linking CPRD to the Index of Multiple Deprivation (IMD) 2007). For each course, the potential indications as specified in the BNF 25 were identified by the presence of relevant read codes on the first prescription date of a course or within 30 days before and 12 months after that date. We considered prevention and treatment of NSAIDassociated ulcer the indication if NSAID prescription date fell on the same date as the PPI prescription. Potential indications were then classified into eight categories (Supplementary Table 1), and missing initial indication was recorded in a separate category.

Statistical analysis
Prevalence of PPI use. For each year, we calculated the period prevalence by dividing the number of patients who received at least one PPI prescription during that year by the corresponding mid-year adult population of the CPRD. We also calculated annual point prevalence as the number of patients with an ongoing PPI prescription on 30 June divided by the corresponding mid-year population. We stratified these prevalence estimates by gender and age (calculated on 30 June and grouped into 10-year age bands).
Patterns of PPI use. The baseline patient characteristics and the use of PPIs among new users were described as proportions of age bands, genders and quintiles of IMD (to represent socioeconomic status). We calculated the percentage of patients who continued their first PPI course, from the index date to the end of 5 years of follow-up during the study period.
Kaplan-Meier survival curves were constructed among all new PPI patients to graphically describe (1) time to discontinuation (permanent or temporary) of the first PPI course during the 5-year follow-up and (2) time to permanent discontinuation of all PPI therapy during the 5-year follow-up. Time to discontinuation of the first PPI course was calculated from the index date to the first PPI course's end date. Time to permanent discontinuation was calculated from the index date to the end date of the last PPI course that each patient received during the follow-up period.
The proportions of patients, who stepped down, or substituted PPIs, were calculated for long-term users as NICE guidelines 13 recommends reviewing long-term PPI user on an annual basis at a minimum. To determine successful step-down attempts accurately, patients were required to have a 12-month window after the step-down date. The analysis of successful step-down attempts was therefore limited to patients who had step-down attempts within the first 4 years of the follow-up to allow adequate follow-up within the final year of the cohort. We repeated this analysis restricted to patients who started PPI therapy Table 1. Descriptive characteristics of new users of proton pump inhibitor therapy with ≥5 years of follow-up data (patients with exclusively short-term courses and patients with at least one long-term course) and the duration (in days) for the first short and long courses as long term and whose indication might not suggest an ongoing need for long-term PPI use; therefore, patients with recorded indication of complicated GORD, NSAID-associated ulcers, prophylaxis or reducing the degradation of pancreatic enzyme supplements were excluded. Analyses were performed using STATA 12 (Stata Corp., College Station, Texas).

Prevalence of prescribing
We identified 31 956 396 PPI prescriptions in 1 828 141 adult patients during the study period. The point and period prevalence of PPI increased between 1990 and 2014 ( Figure 1A), and it varied substantially by age group ( Figure 1B). The point prevalence of PPI use was similar between males and females, increasing during the study period from 0.04% in 1990 to 7.05% in 2014 in males and from 0.03% in 1990 to 8.35% in 2014 in females. The female to male prevalence ratio of PPI use was 1.14 (95% confidence intervals (CI) 1.12-1.17) from 1990 to 2014.

Prescribing patterns
During the study period, 596 334 new users of PPI therapy with at least 5 years of follow-up data were identified. Their mean age was 54.2 years (standard deviation (SD) 16.3), and 55% were females. They received a total of 8 784 272 prescriptions, and 26.5% had one PPI prescription recorded. The median duration for all PPI prescriptions was 28 days (interquartile range (IQR) 18-56 days).
Individual prescriptions were combined to create 1 708 513 PPI courses. The median duration of all courses was 55 days (IQR 28-125 days), and there were a median of two courses per patient (IQR 1-4 courses). Patients received prescriptions for enough PPI to cover 96.69% (95% CI 96.68-96.71) of days in these courses.
Of the courses, 1 505 758 (88.1%) were categorised as short courses and 202 755 (11.8%) were categorised as long courses with median durations of 28 days (IQR 28-79 days) and 805 days (IQR 526-1345 days) for short and long courses respectively. Of the cohort, 73.2% received exclusively short courses with a mean age of 51.6 years (SD 16.3 years), and 26.7% received at least one long course with a mean age of 61.2 years (SD 14.3 years) (

Prescription indications
Initially, 365 481 PPI courses (21.3%) had no coded indication for PPI prescription. This fell to 14.0% after assuming that prescriptions concurrent with NSAID prescriptions were intended for gastroprotection. Dyspepsia was the most frequent recorded indication (Table 2). Figure 2 shows the proportion of patients who discontinued the first PPI course ( Figure 2A) and patients who permanently discontinued all PPI courses ( Figure 2B). When considering only long-term PPI patients, 25% had temporarily discontinued their therapy at 1 year and 3 months after starting their long-term PPI course, 50% at 1 year and 7 months, and by 2 years and 3 months, 75% had temporarily discontinued their long-term PPI course. Of those discontinuing, 9557 (9%) received a prescription for H2RA within 1 month before or after this occurred.

Discontinuation, step down and substitution
Of the 159 259 patients who received long-term PPIs, 63 640 (39.9%) had an attempt to step down their PPI dose (Table 3). Of these, 6388 (10%) had received an H2RA prescription within 1 month before or after stepping down PPI dose.
Of 59 734 patients in whom the initial indication for PPI prescription did not suggest a recognised need for PPI use to be prolonged, uncomplicated GORD was the most frequent recorded indication and 39 164 (65.5%) discontinued PPI therapy (temporarily or permanently). For those patients who temporarily discontinued their PPI therapy, the median time to this was 3 years and 3 months after starting their PPI course. In those using PPI long term without recognised indication for such use, a step-down attempt was identified in 47% (Table 4).

Summary
This study describes the pattern of PPI prescription in UK general practice in terms of its prevalence and the practices employed to reduce long-term use. The proportion of the population using PPIs within each year increased from 0.2% in 1990 to 15.0% in 2014. Of those new PPI users who had 5 years of follow-up available, 26.7% used PPI therapy for more than 1 year, and 3.9% remained on PPI therapy for five 5 years. Clear attempts to step down long-term use  were identified in about 39%, and 8.7% of long-term users received a H2RA prescription around the time they attempted to step down and/or discontinue their use of PPI. Amongst patients whose initial PPI prescription indication did not necessarily warrant long-term PPI use, 47% had attempts to step down their PPI dose.

Comparison with previous work
Our findings pertaining to the prevalence of PPI use in the early years of our study were consistent with the findings of earlier studies involving general practice in the UK. 14,16,18,26 In addition, our result revealed that the use of PPI has continued to rise. These trends are not limited to the UK: Similar increases in prescription rates have been observed in the USA, 27 Australia 28 and many European countries. This widespread increase supports the evidence that PPI prescriptions remain highly prevalent in many health care systems despite the extensive literature that indicates overprescribing PPI in both the primary and secondary care setting. 29,30 In this study, the proportion of patients who were on long-term PPI (26%) was higher than that reported in previous studies, [16][17][18] which have reported rates of long-term PPI usage between 0.05 and 4.4%, according to varying definitions of long-term use. Studies have shown that repeat prescription practices account for approximately 32 to 81% of the total cost of prescribed drugs. 31 The continuous increase in PPI use, specifically the increase in the proportion of long-term users, may therefore have important cost implications despite the availability of low-cost PPI.
PPIs provide effective symptomatic relief for patients who suffer from dyspepsia symptoms. However, while clinical guidelines suggest the use of PPI therapy over short durations to treat dyspepsia symptoms, 13 it seems that PPIs had been prescribed as a form of maintenance therapy without specific PPI, proton pump inhibitor; GORD, gastro-oesophageal reflux disease; NSAID, non-steroidal anti-inflammatory drugs; H2RA, histamine 2 receptor antagonist. *Percentages were calculated from the total number of long-term PPI patients. † Percentages were calculated from the number of step-down patients who stepped down within the first 4 years of follow-up and successfully stepped down for 12 months (number of patients (59 458)).
underlying cause. Our study revealed that dyspepsia symptoms were the initial indication in 23% of longterm PPI courses. However, as most patients on first presentation in primary care will not have a final endoscopic diagnosis, it is inevitable that the GPs will have recorded less-specific indications in subjects who had other underlying diagnoses. Our results concur with those of several studies that have reported that the majority of patients on PPI therapy are prescribed PPI for the purpose of relieving symptoms without any other clear indications. 32,33 In addition, although its clinical relevance is unproven, it has been proposed that rebound acid hypersecretion following PPI therapy withdrawal may help perpetuate the use of PPIs in patients with uncertain indications or who have received them for symptomatic relief of relatively mild symptoms for more than 6 weeks. 34 The issue of appropriateness in terms of prescription practices has been discussed in existing literature. 29,35,36 Despite this, PPIs are still being administered to patients for a variety of complaints that are not known to be acid-induced and over a long-term basis.
In the view of the emerging concerns regarding adverse events from long-term PPI use, clinical guidelines 13 have encouraged GPs to use PPIs carefully and to continually review long-term patients to try to step down or stop treatment. Our results suggest that GPs are actively attempting to reduce PPI use by stepping down and substituting alternative medication. Previous studies [36][37][38][39] reported discontinuation rates that differed from those identified in our study; however, these can be explained by variations in the study population and the discontinuation strategies employed. 38 Reports regarding the outcomes of step-down therapy have been conflicting. 40,41 For example, one study reported that more than half of the patients involved in the study remained asymptomatic after the step down, 41 while another reported that 19% of patients whose PPI therapy was stepped down experienced relapsed symptoms and resumed PPI use. 40 In our study, 60% of the long-term PPI users maintained lower doses for more than 1 year. However, while we identified an appreciable proportion of long-term PPI users who could potentially reduce the use of the drug, we were unable to find evidence of such attempts in a large proportion of those individuals. Non-adherence to the step-down therapy, therefore, allows the maintenance of inappropriate PPI prescription that may sustain overuse of PPIs.  (27 473)).

Strengths and limitations
Our study used data from a large database of UK primary care records that has been extensively used and validated for pharmacoepidemiological research. 21,24 The population in our study is therefore representative of the general practice population of the UK to whom our results should be generalisable. 21 The large sample size has allowed us to stratify our analyses by age groups and gender and to show trends in PPI use over time. It has also provided us with adequate power to identify the relatively small proportion of patients who took PPIs on a long-term basis and describe the management of their prescriptions.
Weaknesses in our study include that we may have underestimated PPI use as neither hospital prescriptions nor over the counter (OTC) use are captured in the data. However, because secondary care-initiated PPI treatment will often be continued by GPs afterwards 35 and prescribed PPI use continued to rise after they became available OTC, 42 we think it unlikely that this has led to massive underestimation. Additionally, we focused on long-term users who would be the most likely to obtain their prescriptions from their GPs. Furthermore, the period of PPI exposure for those who took PPI intermittently may have been underestimated, because the calculation of the prescription duration was based on the assumption that the dispensed prescription was consumed as directed. Indeed, CPRD only contains information about the prescriptions of medications; as such, it is not possible to assess whether patients actually collected or consumed the prescribed medication. In addition, our definition of a successful step down may underestimate the proportion of patients whose long-term PPI therapy was stepped down but then required a smaller increase in dose lower than the initial dose. However, including this in our definition only identified an additional 997 patients (an additional 1.5% of attempted step downs), so for clarity, we retained our initial stricter definition. Furthermore, our method of estimating successful step-down attempts within the initial 4 years of follow-up would not have led to a substantial underestimation, as it is expected that long-term patients should have been offered a step-down attempt at least within the first year of their continuous use of PPI therapy.

CONCLUSION
During the study period, a considerable increase in the administration of PPI prescriptions was observed in UK general practice. The majority of patients use PPIs on a short-term basis with 26% of the identified use long term. Our results suggest that GPs are actively attempting to decrease the use of PPI by stepping down and discontinuing prescriptions; however, this is not universally practised nor is it always successful when attempted. If the cost and potential risks of the continuing increase of PPI are to be minimised, a proactive clinical review and adherence to the guidelines is likely to be required.

CONFLICT OF INTEREST
King Saud bin Abdulaziz University for Health Sciences-Saudi Arabia has sponsored Fatmah Othman studies at University of Nottingham. There is no other support from any other organisation for the submitted work. There are no financial relationships with any other organisations that might have an interest in the submitted work in the previous 3 years, and there are no other relationships or activities that could appear to have influenced the submitted work.

KEY POINTS
• The prevalence of PPI use in the UK general population is high and still increasing.
• The majority of patients only use PPIs short term, with only 26% using them long term.
• Clear attempts to step down long-term use were identified in two fifths of the patients, so there remain further opportunities for reducing the cost and side effects of PPI use through improving adherence to recommended withdrawal strategies.

ETHICS STATEMENT
This study was approved by the Independent Scientific Advisory Committee (ISAC) with CPRD number 13_214 and 13-214Mn.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web site.
Supplementary Tables: Categories of potential indications for proton pump inhibitor prescription with the proposed management for each indication listed.