Impact of EMA regulatory label changes on hydroxyzine initiation, discontinuation and switching to other medicines in Denmark, Scotland, England and the Netherlands: An interrupted time series regression analysis

Hydroxyzine is indicated for the management of anxiety, skin and sleep disorders. In 2015, the European Medicines Agency (EMA) concluded that hydroxyzine was pro‐arrhythmogenic and changes to the product information were implemented in Europe. This study aimed to evaluate their impact in Denmark, Scotland, England and the Netherlands.

impact on discontinuation with no strong evidence suggesting unintended consequences of major switching to other antihistamines, benzodiazepines or antidepressants. clinical and post-marketing safety data. 1 The EMA safety review concluded that although hydroxyzine containing medicinal products are effective treatments for their approved indications, in order for the benefit-risk balance of hydroxyzine to remain favourable, contraindications, warnings, and changes to the product information, including a direct healthcare professional communication (DHPC) were required to be implemented across the European Union (EU).
Hydroxyzine was contraindicated in people with cardiovascular disease (CVD) and in those prescribed medicines known to prolong the QT interval. 1 When new safety information emerges about a medicine, effective communication is essential to minimise harm. The benefit-risk profile of medicines is constantly evaluated by the EMA, which is responsible for cascading new safety information through the European pharmacovigilance network to alert prescribers and patients. However, the impact of such regulatory actions on healthcare professional behaviour is often poorly understood despite their enormous potential to affect public health. 2,3 One of the EMAs responsibilities now includes measuring the effectiveness of their regulatory actions, which led the EMA to commission a study on hydroxyzine to support the EMA Pharmacovigilance Risk Assessment Committee (PRAC) decision making. The aim of this study was therefore to evaluate the impact of the EMA risk minimisation measures implemented in 2015 following the EMA PRAC referral procedure for hydroxyzine using data from Denmark, England, Scotland, and the Netherlands.

| Data sources
Four data sources were analysed (please see Supporting Information Methods for further details). In brief these were: • The Clinical Practice Research Datalink (CPRD), which contains primary care data. For this analysis only up-to-standard data from English practices within the UK was used. 4 • The Scottish Prescribing Information System (PIS), which records all dispensed medicines from pharmacies in Scotland (UK) that can be record-linked to demographic data, hospital admissions and attendances, and death registrations nationally within Scotland. 5 • The Danish Register of Medicinal Products, which records all outof-hospital prescriptions and allows linkage of drug exposures to inpatient and outpatient hospital contacts in the Danish National Patient Registry, including death data from the Civil Registration System. [6][7][8] • The Dutch PHARMO Database Network, which combines data from primary and secondary care in the Netherlands. These different data sources, including data from general practices, in-and out-patient pharmacies, clinical laboratories, hospitals, the cancer registry, pathology registry and perinatal registry, and which are linked on a patient level through validated algorithms. 9

| Study population
Cohorts were generated to provide time series data for analysis using a common protocol (EU PAS Register number EUPAS24089). 10 The start period varied by the availability of data from each database.
Cohort entry was defined as the latest of the following: study start date; date of registration with the general practice (in CPRD and PHARMO); availability of data collection; having at least 1 year of observation (lookback period). For cohorts associated with the target condition (a new contraindication), cohort entry was additionally defined by the date when the condition was first recorded. Contraindicated patients included those patients with: (a) a known risk factor to QT interval prolongation consisting of CVD disease and;

KEY POINTS
• Hydroxyzine initiation fell following the 2015 EMA regulatory action • EMA regulatory action had a variable effect on hydroxyzine initiation • EMA regulatory action was associated with no major switching to other antihistamines, benzodiazepines or antidepressants (b) concomitant use of medicines known to prolong the QT interval and/or induce Torsades de Pointes (please see Supporting Information Methods for further details).
A patient's index date was the latest of the study period start date, the date of birth, or their first database follow up date plus 1 year (to allow sufficient time to determine prevalent versus incident use of medicines). A patient's last follow up date was the first occurrence of the following: death (all databases); end of study period (varies between countries); end of registration within the data source (end of registration would not significantly affect data from Denmark and Scotland because they use national data that captures patients moving within the health system). A patient was included in the time period aggregate if the first and last day both lay between the patient's index date and their last follow up date, so patients were observable for the entire quarter.

| Exposure
The exposure was the conclusion of the 2015 EMA PRAC referral procedure when recommendations aimed at reducing the risk of heart rhythm disturbances associated with use of hydroxyzine first became public (Box 1). 1

| Outcomes
The outcomes of interest evaluated whether there was any immediate change in prescribing at the time of the regulatory action (prespecified as February 2015) and/or change in prescribing trend postintervention compared to the baseline trend. These were analysed as a series of proportions from aggregated patient counts evaluated in each quarter over the study period.
• Hydroxyzine prescribing initiation overall • Hydroxyzine prescribing initiation by recorded history of an indication, age, gender, and exposure type • Hydroxyzine prescribing discontinuation overall • Switching patterns to alternative medicines following hydroxyzine discontinuation Hydroxyzine initiation was defined as a prescription for hydroxyzine with no exposure to hydroxyzine in the preceding 92 days. The denominator was the number of non-users on the first day of the time period defined as no exposure to hydroxyzine in the previous 92 days.
Exposure type was defined as one-off use, sporadic use or chronic use.
One-off users were defined as patients prescribed a single hydroxyzine prescription only. To define sporadic and chronic users we calculated a possession ratio for each patient defined by using the number of days prescribed (or supplied) assuming a standard daily dose divided by the number of days between hydroxyzine prescriptions. We defined sporadic users as patients with a hydroxyzine possession ratio of less than 1 standard day of therapy per 3 days. Patients with a hydroxyzine possession ratio of more than 1 standard day of therapy per 3 days were defined as chronic users. Hydroxyzine discontinuation was defined as the number of patients with a prescription for hydroxyzine with no further exposure to hydroxyzine in the 92 days following the date of prescription. The denominator was the number of patients prescribed hydroxyzine in the time period. The numerator was the number of patients discontinuing. For patients with contraindications to hydroxyzine, only hydroxyzine initiation was examined. A switch to an alternative medicine class was defined as those patients who discontinued hydroxyzine and who then initiated a drug in the class listed in Table S6. Initiation of an alternative medicine was defined as the first prescription of a drug in that class prescribed within 92 days following the date of the last hydroxyzine prescription. Please see online Supporting information methods section for further details.

| Statistical analysis
The study design was an interrupted time series regression analysis of prescribing trends. The primary analysis used interrupted time series regression to fit quarterly time trends for each country. The effect of the intervention for each country was represented either by a step function, or by a continuous linear function modelling the baseline slope before the intervention time point, the change in slope from the baseline time periods to the post-intervention time periods and the immediate change associated with the intervention time point as described by Wagner et al. 11 Interrupted time series regression was done for hydroxyzine initiation rates, hydroxyzine discontinuation rates, and for patients switching to the alternative classes of medicines stratified by country. Time points with cell counts fewer than five were suppressed. Before fitting all regression models, the data was visualised graphically. Trends were modelled using weighted linear regression, the weights being the denominators in each proportion. All models were checked for autocorrelation using the Durbin-Box 1 Summary of EMA recommended measures to minimise the risk of heart rhythm problems with hydroxyzine use • Use must be avoided in patients who already have risk factors for heart rhythm disturbances orare taking other medicines that increase the risk of QT prolongation.
• Use is not recommended in the elderly.
• Use hydroxyzine at the lowest effective dose for as short a time a s possible.
• The maximum daily dose of hydroxyzine should be no more than 100 mg in adults (50 mg in theelderly if use cannot be avoided), and 2 mg per kg body weight where used in children up to 40kg in weight.
• Care is needed in patients taking medicines that slow the heart rate or decrease the level of potassium in the blood, as these also increase the risk of problems with heart rhythm.
Watson statistic. All analyses were carried out using SAS Version 9.4 Copyright 2002-2012 SAS Institute Inc., Cary, North Carolina.

| Ethical permissions
Permission to conduct the study in each database was obtained from the relevant source from each country, according to each databases' standard terms and conditions.

| RESULTS
Over the study period the population consisted of approximately: 5

| Impact of the regulatory action on hydroxyzine initiation
Trends in overall hydroxyzine initiation in each country are shown in Figure 1 and in

| Trends in hydroxyzine initiation stratified by age, gender and exposure type
Trends in hydroxyzine initiation by age, gender and exposure type are shown in Figures S1 to S4 and Tables S1 to S4. Hydroxyzine initiation was greater in women than men and consisted mainly of one-off use.
The regulatory intervention was associated with immediate reductions: in all exposure types in England; one-off and sporadic exposure in Scotland; sporadic exposure in the Netherlands. Post-intervention, there was a change to a negative trend in all exposure types in England and one-off and chronic exposure in Scotland.

| Impact of the regulatory action on hydroxyzine discontinuation
The results for hydroxyzine discontinuation are shown in Figure 2 and in Table 2  3.4 | Impact of the regulatory action on switching to other antihistamines, benzodiazepines or antidepressants Trends in switching to alternative medicines are shown in Figure 3, Table 3. The regulatory action was associated with no immediate change in switching to other antihistamines, benzodiazepines or antidepressant and no significant change in trend post-intervention in any country.

| Impact of the regulatory action on hydroxyzine initiation in patients with contraindications
Trends in hydroxyzine initiation by contraindication are shown in Table 4 and Figures

| DISCUSSION
Over the study period, the prevalence of hydroxyzine initiation fell by 36, 26  For regulatory actions to be effective, knowledge should be effectively disseminated, communicated and understood, and how this occurs may affect the size of any impact. 13 If prescribing levels are already very low, it is possible that regulatory actions may struggle to demonstrate significant impact unless particularly focused on a narrow target population. This is because healthcare professionals may perceive them as being less important and may not be widely discussed among educational resources or incorporated into guidelines.
In this regard, it is possible that use may already be limited to those patients who have a real need for the medicine (i.e., that it has established effectiveness so that they are less likely to discontinue treatment Regulatory interventions may be associated with unintended consequences such as substitute prescribing or switching that may themselves lead to adverse effects. 20,26 Evaluating such effects, is often subject to methodological limitations. 2,3 Specific to hydroxyzine, it would be of concern if patients switched to benzodiazepines that are associated with an increased risk of addiction and mortality to manage anxiety. 27 Similarly, switching to an antihistamine with more sedative properties may also have unintended effects such as increasing road traffic accidents. 28 Reassuringly, we observed there is no strong evidence suggesting the regulatory action for hydroxyzine caused widespread substitute prescribing for those medicines that we evaluated.
We noted that the regulatory action seemed to have less impact on hydroxyzine discontinuation compared to initiation. This would be in keeping with the findings from a systematic review examining the impact of United States risk advisories, which suggested such advisories were more effective at decreasing initiation of targeted medicines and less effective at bringing about their discontinuation. 2 This has also been seen with another recent European study. 20 While the immediate rise in hydroxyzine discontinuation in England is considered an intended effect, it is uncertain why hydroxyzine discontinuation should have fallen in Denmark particularly when the rate of hydroxyzine initiation did not change. However, the post-intervention negative trend in discontinuation in England may be because the pool of hydroxyzine initiators was falling, an effect seen elsewhere. 20

| Strengths and limitations
To our knowledge, this study is the largest study examining hydroxyzine prescribing in Europe, the findings of which may support the EMA strategy for examining the impact of pharmacovigilance decisions. It uses high quality data sources and a common protocol and data extraction method to standardise data from each country. However, the study has several potential limitations. Firstly, our data sources do not capture over the counter antihistamine use that may underestimate the prevalence of switching to other antihistamines. Furthermore, indications were determined by coded diagnoses that are more commonly recorded in primary care than hospital data sources. For this reason the proportion of patients with an indication was greater for England and the Netherland compared to Denmark and Scotland, which report trends for patients with indications attending secondary care. Secondly, due to the lower prevalence of hydroxyzine exposure, many subgroup analyses meant that cell counts less than five were detected that affected precision and their ability to be reported.
Thirdly, although ITS analysis is a robust quasi-experimental design commonly used to evaluate policy interventions, it only examines associations that may be confounded by other interventions occurring simultaneously. 29 Data presented in this manuscript are population-average estimates. Therefore variation may still exist among other groups that have not been studies such as general practices. Recommendations from this study may not be generalizable to areas within Europe that were not studied.
In conclusion, the EMA 2015 regulatory action targeting the safety of hydroxyzine products was associated with consistent reductions in hydroxyzine initiation in two of the four countries, whilst having limited impact on discontinuation and no switching of switching to the classes of medication that we examined.