Volume 82, Issue 2 p. 487-497
Pharmacoepidemiology
Free Access

Prescribing pattern of antipsychotic drugs during the years 1996–2010: a population-based database study in Europe with a focus on torsadogenic drugs

Alessandro Oteri

Alessandro Oteri

Department of Medical Informatics, Erasmus University Medical Center, Rotterdam, the Netherlands

Department of Clinical and Experimental Medicine, University of Messina, Messina, Italy

Search for more papers by this author
Giampiero Mazzaglia

Giampiero Mazzaglia

Health Search, Italian College of General Practitioners, Florence, Italy

Search for more papers by this author
Serena Pecchioli

Serena Pecchioli

Health Search, Italian College of General Practitioners, Florence, Italy

Regional Agency for Healthcare Services of Tuscany, Florence, Italy

Search for more papers by this author
Mariam Molokhia

Mariam Molokhia

Department of Primary Care and Public Health Sciences, King′s College London, London, UK

Search for more papers by this author
Sinna Pilgaard Ulrichsen

Sinna Pilgaard Ulrichsen

Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark

Search for more papers by this author
Lars Pedersen

Lars Pedersen

Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark

Search for more papers by this author
Elisabetta Poluzzi

Elisabetta Poluzzi

Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy

Search for more papers by this author
Fabrizio De Ponti

Fabrizio De Ponti

Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy

Search for more papers by this author
Edeltraut Garbe

Edeltraut Garbe

Leibniz Institute for Epidemiology and Prevention Research – BIPS, Bremen, Germany

Search for more papers by this author
Tania Schink

Tania Schink

Leibniz Institute for Epidemiology and Prevention Research – BIPS, Bremen, Germany

Search for more papers by this author
Ron Herings

Ron Herings

PHARMO Institute for Drug Outcomes Research, Utrecht, the Netherlands

Search for more papers by this author
Irene D. Bezemer

Irene D. Bezemer

PHARMO Institute for Drug Outcomes Research, Utrecht, the Netherlands

Search for more papers by this author
Miriam C. J. M. Sturkenboom

Miriam C. J. M. Sturkenboom

Department of Medical Informatics, Erasmus University Medical Center, Rotterdam, the Netherlands

Search for more papers by this author
Gianluca Trifirò

Corresponding Author

Gianluca Trifirò

Department of Medical Informatics, Erasmus University Medical Center, Rotterdam, the Netherlands

Department of Clinical and Experimental Medicine, University of Messina, Messina, Italy

Correspondence Dr Gianluca Trifirò MD, PhD, Department of Clinical and Experimental Medicine, University of Messina, Policlinico Universitario G. Martino, Via Consolare Valeria Gazzi, 98125 Messina, Italy. Tel.: +39 09 0221 3300 Fax: +39 09 0221 3264; E-mail: [email protected]Search for more papers by this author
First published: 06 April 2016
Citations: 22

Abstract

Introduction

Antipsychotic drugs (APDs) are used to treat several mental illnesses. Some APDs have long been known to be associated with QT prolongation, potentially leading to torsades de pointes (TdP) and sudden cardiac death (SCD). In 2005, thioridazine was withdrawn because of the risk of SCD, bringing further attention to the arrhythmogenic potential of APDs.

Aim

The aim of the current study was to evaluate the use of APDs in five European countries during the years 1996–2010.

Methods

A cohort study was conducted using prescription/dispensing data from seven healthcare databases [the AARHUS University Hospital Database (Denmark), the German Pharmacoepidemiological Research Database (GePaRD) (Germany), Health Search Database/Thales (HSD) and Emilia Romagna Regional Database (ERD) (Italy), PHARMO Database Network and Integrated Primary Care Information (IPCI) (the Netherlands) and The Health Improvement Network (THIN) (the UK), covering a population of 27 million individuals.

The annual prescription rate of APDs was measured overall and for individual medications. APDs were classified as torsadogenic according to the Arizona-CERT list. All analyses were stratified by age, gender and calendar year.

Results

A total of 559 276 person-years (PYs) of exposure to APDs was captured. The crude annual prescription rate of APD use ranged from 3.0/1000 PYs in ERD to 7.7/1000 PYs in AARHUS. Among APDs with established torsadogenic potential, thioridazine was the most frequently used medication in the UK. Haloperidol was commonly prescribed in Italy and the Netherlands. The use of APDs with torsadogenic potential was much higher in elderly patients.

Conclusions

Substantial use of APDs with torsadogenic potential has been reported in Europe in recent years, in spite of increasing concerns about their arrhythmogenic potential. This use was even greater in elderly patients, who are at higher risk of SCD.

What is Already Known about this Subject

  • Several antipsychotic drugs (APDs), belonging to both typical and atypical classes, (e.g. sertindole, thioridazine or haloperidol) are known to be associated with an increased risk of QTc prolongation, potentially leading to serious ventricular arrhythmias (VAs) such as torsades de pointes (TdP) and sudden cardiac death (SCD). In recent years, regulatory authorities from several countries have expressed concerns about a relationship between APD use and the risk of QTc prolongation, serious VAs and SCD. Consequently, some APDs have been withdrawn from the market and the use of several others has been restricted.

What this Study Adds

  • Widespread use of APDs with torsadogenic potential was reported in a large population from five European countries, especially in Italy and the Netherlands. The safety warnings issued by regulatory agencies contributed to reduce the use of thioridazine in the UK and haloperidol in Italy. The widespread use of medications with torsadogenic potential in elderly patients, as observed in our study, should be taken into account, given the higher risk of VAs potentially leading to SCD in this population.

Introduction

Antipsychotic drugs (APDs) are the mainstay of therapy for several mental illnesses, such as schizophrenia and other psychoses. A high level of use is also reported for the treatment of behavioural symptoms in patients with dementia. In recent decades, an increasing trend in APD use has been observed worldwide, with atypical drugs progressively replacing typical APDs owing to a better extrapyramidal safety profile 1. Several APDs, belonging both to typical and atypical classes (e.g. sertindole, thioridazine and haloperidol), are known to be associated with an increased risk of QTc prolongation, potentially leading to serious ventricular arrhythmias (VAs) such as torsades de pointes (TdP) and sudden cardiac death (SCD) 2-4. In recent years, regulatory authorities from several countries have expressed concerns about a relationship between APD use and the risk of QTc prolongation, serious VAs and SCD 5-8. Consequently, some APDs have been withdrawn from the market and the use of several others has been restricted. In 1998, the Committee on Safety of Medicines (CSM) in the UK reported 13 cases of VA following the use of sertindole. This led to the voluntary withdrawal of the drug from the market, in order to collect additional evidence 9. In 2001, the drug was reintroduced in Europe under the restriction that treated patients would be enrolled in a study focusing on mortality, to ensure that all patients treated with sertindole were selected and monitored carefully 10. In 2010, the results of the Sertindole Cohort Prospective (SCoP) study, on the safety of sertindole in relation to all-cause mortality, serious cardiac events, including arrhythmias, and cardiac mortality, showed a statistically significant increased risk of cardiac mortality in comparison with risperidone, further confirming the previous evidence showing the increased cardiac toxicity associated with this medication 11.

In 1998, thioridazine, mesoridazine and droperidol received a black-box warning from the US Food and Drug Administration (FDA) regarding the risk of QTc prolongation. By the end of 2000, 21 suspected cases of SCD had been reported in patients prescribed thioridazine. This led the CSM to restrict the drug to a second-line option for the treatment of schizophrenia under specialist supervision. Baseline and periodic electrocardiogram (ECG) measurements and the monitoring of serum potassium, calcium and magnesium were also recommended 12. A few months later, droperidol was withdrawn from the market because of its torsadogenic potential and, in 2005, thioridazine was withdrawn from the global market because of the risk of SCD. Finally, in 2007, the FDA released an alert on the torsadogenic risk of haloperidol, especially for high-dose and intravenous administration. It recommended periodic ECG monitoring upon prescription of haloperidol in patients with other QT-prolonging conditions, including electrolyte imbalance, underlying cardiac abnormalities and hypothyroidism, and in those already treated with other drugs known to prolong the QT interval 13. More recently, a meta-analysis of observational studies, conducted in the context of the Arrhythmogenic Potential of Drugs (ARITMO) project, showed a variable risk of VA and SCD between individual APDs in relation to their differing ability to block the human ether-a-go-go-related gene (hERG) channel. Medications with a higher potency for blocking the hERG channel (thioridazine, clozapine, risperidone, haloperidol, olanzapine and quetiapine) were associated with a higher risk of VA and SCD. 14. Conversely, a large study on 18 154 patients with schizophrenia failed to show an association between ziprasidone and SCD in comparison with olanzapine, in spite of its known risk of QTc prolongation 15.

Although several studies have analysed the prescribing patterns of APDs in the general population 16-18, to our knowledge no previous investigation has evaluated the prescription rate of such medications in relation to their torsadogenic potential in a large European population. The aims of the present study were: (i) to investigate the possible changes in the prescribing patterns of APDs in relation to their torsadogenic potential in five European countries (Denmark, Germany, Italy, the Netherlands and the UK) between 1996 and 2010 and (ii) to evaluate the possible impact of safety warnings and regulatory measures on the use of these medications.

Methods

Data sources

Seven healthcare databases were involved in the present study: Health Search Database/Thales (HSD, Italy), Integrated Primary Care Information (IPCI, the Netherlands), the PHARMO Database Network (the Netherlands), the AARHUS University Hospital Database (Denmark), The Health Improvement Network (THIN, UK), the German Pharmacoepidemiological Research Database (GePaRD, Germany) and the Emilia Romagna Regional Database (ERD, Italy). All these databases contain information from the healthcare records of almost 27 million European citizens. HSD, IPCI and THIN contain records from general practice (GP), while PHARMO, AARHUS, GePARD and ERD are comprehensive administrative/record-linkage systems in which drug dispensing data for a well-defined population are linked to a registry of hospital discharge diagnoses and various other registries. Information collected by these databases includes a registry of inhabitants, and drug dispensing/prescription records. Drug prescriptions/dispensing are coded using the Anatomical Therapeutic Chemical (ATC) classification system in HSD, IPCI, PHARMO, AARHUS, GePARD and ERD, or the British National Formulary (BNF)/Multilex classification system in THIN. Table 1 reports the characteristics of each database as well as the number and percentage of exposed individuals by age groups. For this analysis, databases contributed data for the period between 1 January 1996 and 31 December 2010.

Table 1. Characteristics of healthcare databases involved in the Arrhythmogenic Potential of Drugs (ARITMO) project
Characteristics AARHUS (Denmark) GePaRD (Germany) Emilia Romagna Regional Database (Italy) Health Search Database/Thales (Italy) PHARMO (the Netherlands) IPCI (the Netherlands) THIN (UK)
Current source population 1 559 718 7 108 746 6 079 798 1 240 561 3 989 797 1 016 632 5 894 468
Observational years 2001–2008 2005–2008 2006–2010 2000–2010 1999–2009 1997–2010 1996–2009
Type of database Data warehouse record linkage system with:

1) Registry inhabitants

2) Regional drug dispensing records

3) Hospital claims database

Data warehouse record linkage system with:

1) Registry inhabitants

2) Outpatient prescription drug data

3) Hospital claims database

Data warehouse record linkage system with:

1) Registry inhabitants

2) Regional drug dispensation records

3) Hospital claims database

General practice database Data warehouse record linkage system with:

1) Registry inhabitants

2) Regional drug dispensing records

3) Hospital discharge diagnosis records

General practice database General practice database
Age range Exposed individuals N (%) Exposed individuals N (%) Exposed individuals N (%) Exposed individuals N (%) Exposed individuals N (%) Exposed individuals N (%) Exposed individuals N (%)
0–19 4184 (3.4) 6342 (3.1) 1366 (1.6) 608 (0.9) 13 246 (7.9) 1260 (6.1) 4634 (2.6)
20–34 16 143 (13.2) 22 484 (11.1) 6333 (7.4) 5939 (8.3) 26 498 (15.7) 2520 (12.1) 25 804 (14.4)
35–49 27 282 (22.3) 50 916 (25.0) 14 376 (16.7) 11 867 (16.6) 39 553 (23.5) 4916 (23.6) 40 139 (22.4)
50–64 30 367 (24.9) 49 504 (24.4) 15 104 (17.6) 14 885 (20.8) 30 400 (18.1) 4744 (22.8) 36 521 (20.3)
65–79 26 252 (21.5) 48 415 (23.8) 21 584 (25.1) 19 915 (27.8) 28 761 (17.1) 3865 (18.6) 37 789 (21.0)
≥ 80 17 907 (14.7) 25 602 (12.6) 27 278 (31.7) 18 306 (25.6) 29 943 (17.8) 3492 (16.8) 34 673 (19.3)
  • AARHUS, the AARHUS University Hospital Database; GePARD, German Pharmacoepidemiological Research Database; IPCI, Integrated Primary Care Information; PHARMO, the PHARMO Database Network; THIN, The Health Improvement Network.

Study population

All subjects registered in the databases during the study period, with at least one year of valid records, were included in the analysis. The total source population included around 27 million individuals from the five European countries.

Study drugs

Within the source population, all persons with at least one prescription/dispensing of APDs (ATC: N05 A or corresponding BNF chapter) during the study period were identified.

On the basis of the Arizona-Cert List website (https://www.crediblemeds.org/new-drug-list/; accessed 11 August 2015), APDs were divided into three mutually exclusive groups: (i) APDs with established torsadogenic potential (chlorpromazine, droperidol, haloperidol, mesoridazine, pimozide, sulpiride and thioridazine): when ‘substantial evidence supports the conclusion that these drugs prolong the QT interval and are clearly associated with a risk of TdP, even when taken as directed in official labelling’; (ii) APDs with possible torsadogenic potential (aripiprazol, clozapine, iloperidone, olanzapine, paliperidone, quetiapine, risperidone, sertindole and ziprasidone): when ‘substantial evidence supports the conclusion that these drugs can cause QT prolongation but there is insufficient evidence at this time that these drugs, when used as directed in official labelling, are associated with a risk of causing TdP’ and (iii) all other APDs 19.

Data analysis

For each observation year, the annual and monthly prescription rate of APD treatment was evaluated by dividing the number of patients who received at least one day of exposure to an APD in that calendar year or month by the total person-time in that person-year/month for all subjects alive and registered in the seven databases on 1 July of the same year. The prescription rate of APDs was expressed as the rate per 1000 person-years (PYs), with 95% confidence intervals (95% CIs). All of the analyses were also stratified by age, database and calendar year. Data transformation was conducted locally, based on standardized common input files, and aggregated data on annual and monthly prescription rates were produced by using Jerboa© v2.9.21, the software has been built at the Erasmus Medical Center, Rotterdam (NL) 20. All subsequent analyses on the aggregated data were carried out using SAS v9.2 (SAS Institute Inc., Cary, NC, USA) on the central remote research environment Octopus. A detailed description of these procedures has been published elsewhere 21.

Results

The source population included in the ARITMO study comprised 26 889 720 individuals from five European countries (Table 1). Overall, 851 717 patients received at least one prescription/dispensing of APDs during the study period, generating 559 276 PYs of exposure to APDs.

The crude average prescription rate of APDs ranged between 3.0/1000 PYs in ERD to 4.1/1000 PYs in IPCI, 4.3/1000 PYs in PHARMO, 4.8/1000 PYs in THIN, 5.8/1000 PYs in HSD, 7.0/1000 PYs in GePARD and 7.7/1000 PYs in AARHUS. Figure 1 shows the monthly prescription rates, demonstrating constant higher use in Denmark. Stratifying the analysis by age group, an increasing trend in the overall prescription rate of APDs was reported in patients over 65 years of age (Figure 2). Drugs with torsadogenic potential accounted for 54.2% of the total exposure in elderly patients (age ≥ 65 years) (data not shown).

Details are in the caption following the image
Monthly prescription rate [× 1000 person-years (PYs)] of antipsychotic drug use for each calendar month and database during the observed years. (image) AARHUS, the AARHUS University Hospital Database; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network
Details are in the caption following the image
Overall prescription rate [× 1000 person-years (PYs)] of antipsychotic drug (APD) use per age and database during the observed years. (image) AARHUS, the AARHUS University Hospital Database; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network

Use by torsadogenic potential

Looking at the distribution of APD use across countries, different percentages of the total exposure to drugs with established and possible torsadogenic potential, according to the Arizona-CERT list, were reported in the seven databases involved in the present study: 60.5% in AARHUS (Denmark), 54.0% in GePARD (Germany), 47.9% in ERD (Italy), 72.2% in HSD (Italy), 64.9% in PHARMO (the Netherlands), 72.1% in IPCI (the Netherlands) and 52.2% in THIN (the UK) (Figure 3). Among APDs that had undergone regulatory action, only thioridazine and haloperidol were prescribed in the databases included in the present study (Figures 4, 5). Annual prescription rates for thioridazine were particularly high in the UK, showing an increasing trend in use from 1996 to 2000, followed by a sharp fall after the first warning released by the CSM. A similar declining trend in the use of thioridazine until 2005 (the year of withdrawal) was reported in Italy and the Netherlands. In addition, haloperidol was mostly used in Italy and the Netherlands. However, different trends of use were observed in these two countries. In Italy, an increasing trend in use was observed until 2007, followed by a decline after an alert released by the Italian Medicines Agency (AIFA) following the warning about the torsadogenic risk of haloperidol issued by the FDA. Conversely, a constant increasing trend was seen in PHARMO and IPCI (the Netherlands). Among drugs with possible torsadogenic potential, quetiapine and risperidone showed the highest prescription rate, with an increasing trend in use constantly observed for quetiapine, from its marketing in 2000 until the end of the follow-up period (Figure 6). Finally, the highest use of risperidone was seen in Denmark and the Netherlands (Figure 7).

Details are in the caption following the image
Distribution of antipsychotic drug use across different countries (% of total exposure). AARHUS, the AARHUS University Hospital Database; ERD, Emilia Romagna Regional Database; GePARD, German Pharmacoepidemiological Research Database; HSD, Health Search Database; IPCI, Integrated Primary Care Information; PHARMO, the PHARMO Database Network; THIN, The Health Improvement Network. (image) Other antipsychotic agents; (image) Possible risk; (image) Establish risk
Details are in the caption following the image
Annual prescription rate [× 1000 person-years (PYs)] of antipsychotic drug subjected to regulatory action: thioridazine. (image) AARHUS, the AARHUS University Hospital Database; CSM, Committee on Safety of Medicines; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network
Details are in the caption following the image
Annual prescription rate [× 1000 person-years (PYs)] of antipsychotic drug subjected to regulatory action: haloperidol. (image) AARHUS, the AARHUS University Hospital Database; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network
Details are in the caption following the image
Annual prescription rate [× 1000 person-years (PYs)] of antipsychotic drug use with possible torsadogenic potential: quetiapine. (image) AARHUS, the AARHUS University Hospital Database; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network
Details are in the caption following the image
Annual prescription rate [× 1000 person-years (PYs)] of antipsychotic drug use with possible torsadogenic potential: risperidone. (image) AARHUS, the AARHUS University Hospital Database; (image) ERD, Emilia Romagna Regional Database; (image) GePARD, German Pharmacoepidemiological Research Database; (image) HSD, Health Search Database/Thales; (image) IPCI, Integrated Primary Care Information; (image) PHARMO, the PHARMO Database Network; (image) THIN, The Health Improvement Network

Discussion

The present population-based study evaluated the prescribing patterns of APDs in five European countries in relation to their perceived torsadogenic potential, as per the Arizona-CERT list, and before and after safety warnings on the risk of QTc prolongation for thioridazine and haloperidol. The use was highest in Denmark and lowest in the Netherlands and Italy, with a fairly constant prescription rate observed during the whole study period. The prescribing patterns of APDs in the general population have been evaluated in several previous investigations. In their analysis, Kaye et al. observed an increasing use of APDs in the UK between 1991 and 2000 22. This finding is consistent with our data, showing a similar trend during the overlapping periods in the THIN database (UK). Another study, carried out in Italy, found a fairly stable prevalence of use between 1999 and 2002 18, which was also comparable with our results, showing a steady use of APDs in Italy during the same time window. Furthermore, the higher prescription rate of APDs observed in PHARMO, compared with IPCI, might be explained by the fact that the additional dispensings in PHARMO are from specialist prescriptions. Analysing the distribution of APD use based on their torsadogenic potential, higher percentages of the total exposure to drugs with established and possible torsadogenic potential were reported in Italy and the Netherlands, with more than 70% of the total exposure observed in HSD and PHARMO. Such evidence is in line with a recent investigation, also produced in the context of the ARITMO project, showing a higher exposure to APDs with strong torsadogenic signals, based on pharmacovigilance data from the FDA Adverse Event Reporting System across Europe 23, 24. Among drugs with established torsadogenic potential, only thioridazine and haloperidol were prescribed in the databases included in the present study. Thioridazine was the most frequently prescribed medication in the UK between 1996 and 2000, followed by a sharp decline after the first alert released by the CSM in 2000 11. The increasing trend in thioridazine use observed in THIN is consistent with the above-mentioned UK study, reporting this drug to be the most commonly prescribed APD in the UK between 1991 and 2000 22. Although less representative, a similar reduction in the use of thioridazine was reported in databases with a longer follow-up, such as HSD, PHARMO and IPCI. This is also in line with the results of a study carried out in Australia, showing an 80% fall in thioridazine use from 1995 to 2001 25.

Concerning haloperidol, higher prescription rate was reported in Italy and the Netherlands. However, different trends in use were observed in databases from these two countries. Increasing use of haloperidol was reported in HSD (Italy) until 2007, followed by a reduction, probably due to the introduction by the AIFA of a warning about the risk of serious VAs and SCD, especially when the drug is administered intravenously or in high doses 25. A similar declining trend was seen in ERD during the same period (see below for an interpretation of ERD values). Based on this warning, the AIFA changed the Summary of Product Characteristics, recommending baseline and periodic ECG monitoring in patients at high cardiovascular risk 26. Conversely, an increasing trend in haloperidol use was observed in PHARMO and IPCI over the entire study period. The effect of safety warnings on APD use was further investigated by Dorsey et al. 27 and Trifirò et al. 28 in 2010. A recent US study showed a reduction in the use of atypical APDs as consequence of an FDA warning linking atypical APDs to an increased risk of all-cause mortality in elderly patients affected by dementia 27. A similar falling trend was seen in another study, evaluating the prescribing pattern of atypical APDs in an Italian setting of elderly demented patients 28. A recent Spanish investigation analysed the changing pattern of use of atypical APDs (olanzapine and risperidone) by means of interrupted time-series analysis, after the introduction of specific regulatory measures concerning such medications. The study showed a reduction in the prescribing pattern of risperidone after the introduction of a safety warning about the drug 29. Sumilarly, in the present study, the use of APDs in Europe seemed to be highly influenced by safety warnings and health regulatory measures on QTc prolongation and severe VAs potentially leading to SCD. This finding confirms the importance of health regulatory measures in influencing the use of APDs in the general population.

Among drugs with possible torsadogenic potential, quetiapine and risperidone showed the highest use. An increasing trend in the prescription rate was constantly reported for quetiapine, from its marketing in 2000 until the end of the follow-up period. On one hand, this result might be due to the tendency of new drugs to be widely prescribed soon after their marketing 30, 31 but, on the other hand, it may also reflect the increasing trend in the use of atypical APDs in the general population because of their better extrapyramidal safety profile 1. This observation could also explain the increasing trend in the use of risperidone observed mainly in Denmark and the Netherlands. The increasing prescription rate of APDs with advancing age reported in the present study is also consistent with several previous investigations showing higher use of APDs in elderly patients 18. This trend might be explained by the high use of APDs for the treatment of behavioural and psychological symptoms in elderly demented patients, which has been documented in previous investigations 27, 28.

Of note, drugs with torsadogenic potential accounted for 54.2% of the total PYs of exposure in the elderly population. As previous evidence has shown an increasing risk of VAs and SCD in elderly patients treated with APDs, especially those with dementia 2, 32, a more cautious use of drugs with torsadogenic potential would have been expected in this higher-risk group, compared with younger patients, in the present study.

Limitations of the study

The results of the present study should be considered in view of some limitations. Firstly, we used outpatient prescription/dispensing data and had no information about the actual use of the medications. Secondly, the study was performed using computerized medical records from seven healthcare databases, thus limiting the generalizability of the results, as some categories of patients, such as inpatients or elderly persons with mental illnesses residing in nursing homes, were missing. Furthermore, the very low prescription rate of atypical APDs observed in ERD might be explained by the fact that, in Italy, these drugs are partially dispensed through direct distribution from local psychiatric services and cannot be captured in the outpatient pharmaceutical dispensing flow, which was considered in the present study. As a consequence, an underestimation of the use of atypical APDs may have occurred in this database. Nevertheless, because GPs in Italy initiate APD treatment for some patients and continue treatments begun by specialists for other patients, data from HSD are likely to be more representative of national trends than those from ERD, especially for atypical APDs.

Conclusions

In conclusion, high use of APDs with torsadogenic potential was reported in recent years in the five European countries studied, despite increasing concerns about their arrhythmogenic potential. The safety warnings issued by regulatory agencies contributed to reducing the use of thioridazine in the UK and haloperidol in Italy. The wide use of these medications in elderly patients should be taken into account, given the higher risk of VAs potentially leading to SCD in this population.

Competing Interests

All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work. EG runs a department that occasionally performs studies for pharmaceutical companies. These include Mundipharma, Bayer-Schering, Stada, Sanofi-Aventis, Sanofi-Pasteur, Novartis, Celgene and GSK. EG has been a consultant to Bayer-Schering, Nycomed, Teva and Novartis in the past. MS runs a group that occasionally performs studies for pharmaceutical companies with the full freedom to publish. These include Pfizer, EliLilly and AstraZeneca.

This research was supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at Guy′s and St Thomas′ NHS Foundation Trust and King′s College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

Contributors

AO, GM, MCJMS and GT conceived and designed the experiments. AO, SP, SPU, EP, TS and IDB collected and analysed the data. AO and GT wrote the manuscript. MM, LP, FDP, EG and RH made substantive suggestions for data interpretation and contributed to the discussion. All the authors revised and approved the final submitted manuscript.

The research leading to these results received funding from the European Community's Seventh Framework Programme (FP7/2007–2013) under grant agreement n° 241 679 – the ARITMO project. The funders had no role in study design, data collection and analysis, the decision to publish or preparation of the manuscript.