Changes in antipsychotic drug prescribing by general practitioners in the United Kingdom from 1991 to 2000: a population-based observational study

Abstract

Aims

To estimate changes in the frequency of use of various antipsychotic drugs in the UK from 1991 to 2000 and to relate these changes to patients' characteristics.

Methods

We conducted a population-based observational study using data from general practices that contribute information to the General Practice Research Database (GPRD). The study population comprised men and women 10–99 years old. We estimated annual use, first-time use, duration of use and, in a sample of 200 patients, indications for use of various antipsychotic drugs, and we observed how these measures had changed over the past decade.

Results

The annual use of antipsychotic drugs increased from 10.5 per thousand in 1991 to 12.2 per thousand in 2000, an overall increase of 16%. The increase was greater in men (25.2%) than women (2.7%). At the same time, the rate of new use of antipsychotic drugs was stable in men and decreased by 21% in women. The difference between patterns of annual use and rates of new use is attributable to the increasing average annual duration of treatment in both men and women during the past decade. Thioridazine, which the UK Committee on the Safety of Medicines (CSM) recently recommended should be used only for second-line treatment of schizophrenia in adults, was the most commonly prescribed antipsychotic throughout the study period. Its use increased as a proportion of all antipsychotic drug use from 1991 to 2000 in men and women aged 10–69 years, but decreased in older users. More than half of all first-time use of antipsychotic drugs in the sample of patients we evaluated was for treatment of depression, anxiety states, and panic disorders, while less than 10% was for treatment of schizophrenia and other psychoses.

Conclusion

The use of antipsychotic drugs has increased in the UK during the past decade, primarily due to increased average annual duration of use rather than higher rates of new use. Most antipsychotic drug use appears to have been prescribed to treat nonpsychotic disorders. It will be of interest to see whether the use of thioridazine, which was the most widely prescribed antipsychotic during 1991–2000, decreases during the next decade in response to the recent CSM recommendation.

Introduction

Antipsychotic drugs are used to treat schizophrenia and other psychoses, as well as less severe psychiatric disorders and a variety of physical symptoms. Several antipsychotic drugs, such as thioridazine and haloperidol, have been used widely for several decades, while newer ‘atypical’ antipsychotics such as risperidone and olanzapine, which have been reported to cause fewer adverse effects, were introduced during the 1990s. The UK Committee on the Safety of Medicines only recently recommended (November 2001) that the use of thioridazine should be restricted to the second-line treatment of schizophrenia in adults. This recommendation was made because the balance of benefits and risks (especially cardiotoxicity [QTc interval prolongation and life-threatening ventricular arrhymias]) was judged to be unfavourable for previous thioridazine indications [1], which include anxiety, agitation and restlessness in the elderly, moderate to severe psychomotor agitation, violent and dangerously impulsive behaviour, mania/hypomania, and behavioural disorders and epilepsy in children. Because antipsychotic drugs have been used to treat many psychiatric and nonpsychiatric conditions and are often prescribed chronically, general practitioners will share responsibility for responding to this recommendation. Since no recent population-based studies of antipsychotic drug use in the UK have been published, we investigated patterns of antipsychotic drug prescribing by general practitioners in the UK during the past decade.

Methods

We studied the prescribing of antipsychotic drugs to patients aged 10–99 years who were registered in the General Practice Research Database at any time from 1 January 1991 to 31 December 2000.

General Practice Research Database

We used data from 270 general practices that have contributed information continuously to the General Practice Research Database (GPRD) maintained by the Boston Collaborative Drug Surveillance Program. Our group has confirmed in many previous studies that these practices provide clinical and prescription data that are of high quality and completeness.

Antipsychotic drugs

We studied drugs listed as antipsychotic agents in the British National Formulary (September 2002). We did not include fixed combinations of an antipsychotic and a tricyclic antidepressant (fluphenazine and nortriptylene, for example) because these are classified as antidepressant drugs. We excluded five antipsychotic drugs (clozapine, benperidol, sertindole, amisulpride, and quetiapine) that were each prescribed to fewer than 1000 patients during the study period.

Annual use and first-time use of antipsychotic drugs

The base population for estimating annual antipsychotic drug use comprised all subjects registered in the database at any time during a given year. We defined users of antipsychotic drugs (or of a particular antipsychotic drug) as patients who had at least one such prescription recorded in their computerized medical record during the year. We estimated annual use as the number of users divided by the number of subjects in the base population each year. The number of users of any antispychotic drug ranged from 15 800 to 21 500 annually during the study period. The number of subjects aged 10–99 years registered annually in the database ranged from 1.4 million to 1.7 million. The base population for estimating annual rates of first-time antipsychotic drug use included all subjects registered in the database during a given year who had at least 2 years of previous information recorded in the GPRD and during that time, had no antipsychotic drug prescription recorded. We defined first-time users of antipsychotic drugs as patients who had a first recorded prescription for such a drug after at least 2 years registration with no recorded antipsychotic drug prescription. We estimated the annual rate of first-time use as the number of first-time users in a given year divided by the person-time free from antipsychotic drug use contributed by individuals in the base population.

Number of prescriptions and annual duration of treatment

To further assess changes in patterns of antipsychotic drug prescribing, we analysed the annual number of antipsychotic drug prescriptions recorded for each user. We also investigated the annual duration of treatment for oral antipsychotic drugs, which we estimated from the number of pills or capsules prescribed and the recorded instructions for use. For these analyses we excluded information from each subject's first and last year of recorded antipsychotic drug use. We examined data from 1990 and 2001, when available, to allow inclusion of appropriate information from 1991 and 2000. We also excluded individual antipsychotic drug prescriptions with instructions for use ‘as directed’, those with uninterpretable directions for use, and those apparently covering less than 1 week or more than 3 months of treatment.

Indications for antipsychotic drug use

To explore the indications for antipsychotic drug use in the study population, we reviewed computerized medical records for a random sample of 200 first-time users of antipsychotic drugs and noted the diagnosis of psychiatric illnesses and other medical conditions. In many instances, more than one potential indication was listed before the first prescription, or the most medically plausible explanation for the prescription did not immediately precede the actual prescription recording, so we used the following hierarchy to categorize indications for treatment. If at any time in the patient profile before the first recording of an antipsychotic medication there was an indication of schizophrenia or other unspecified psychosis, bipolar disorder, or depression, anxiety or panic disorder, we chose that diagnosis (in order listed) as the indication for treatment. If none of the above indications was listed in the patient profile before the first prescription, we then grouped indications for treatment into one of the following categories: pain; agitation or dementia; miscellaneous symptoms and signs, which included stress, insomnia, fatigue, nausea, hallucinations, substance abuse/withdrawal or other personality disorders; multiple symptoms and signs (any combination of two or more symptoms/signs); or unknown.

Statistical analysis

To study changes in antipsychotic drug use over time, we standardized estimates of annual use to the sex-specific age distributions of the population in the year 2000. We also modelled the sex-specific changes over time by adjusting for age in years as a continuous variable using fractional polynomial Poisson regression (STATA, version 7.0; Stata Corporation, College Station, TX, USA). We modelled sex-specific changes in the rates of first-time antipsychotic drug use controlling for age by Poisson regression (PROC GENMOD, SAS, version 8.02; SAS Institute, Inc., Cary, NC, USA). In modelling changes in the annual number of antipsychotic drug prescriptions over time, we used PROC MIXED (SAS, version 8.02) to adjust for correlation between years within patients. Figures presenting patterns of use by age were generated using STATA (version 8.0).

Results

Annual antipsychotic drug use

The annual use of antipsychotic drugs increased from 10.5 per thousand in 1991 to 12.2 per thousand in 2000, an overall increase of 1.7 per thousand (16%). Average annual use during the study period was higher in women (14.7 per thousand) than men (9.4 per thousand). Annual use increased with older age in both men and women. Although annual use increased with age through most of the age range we studied, the change was greater in the 10–29-year-old and 70–89-year-old groups than in 30–69-year-olds (Figure 1) Adjusted for age, the annual use of antipsychotic drugs increased by an average of 2.8% (95% confidence interval [CI] 2.5, 3.0) per year in men and 0.3% (0.1, 0.5) per year in women; a total of 25.2% in men and 2.7% in women over the decade we studied.

Figure 1.

Observed (°) and model-adjusted (‘predicted’ (—)) annual use of antipsychotic drugs. Multiple data points at each age represent different years of the study. Note that the ordinate is plotted on a logarithmic scale.

Thioridazine was the most commonly prescribed antipsychotic agent, accounting on average for 36% of annual antipsychotic drug use. Flupentixol, chlorpromazine, trifluoperazine, and haloperidol each accounted for 10–15% of antipsychotic drug use, while sulpiride, fluphenazine, and droperidol each accounted for only 1–4% of use. Use of the ‘atypical’ antipsychotic drugs risperidone and olanzapine increased continuously from when they were introduced (1993 and 1996, respectively), and by the year 2000, these drugs accounted for 10% and 6% of annual antipsychotic drug use, respectively. The proportional use of three older drugs (thioridazine, sulpiride, and droperidol) increased during the study period, while the use of flupentixol, chlorpromazine, trifluoperazine, and fluphenazine decreased.

For the years 1991 and 2000, (Figure 2 (A,B) presents age-specific patterns of the use of selected antipsychotic drugs as proportions of all antipsychotic drug use during those years. In both men and women 10–69 years old, the use of thioridazine increased as a percentage of all antipsychotic drug use from 1991 to 2000, but its use decreased during this period in older age groups. The proportional use of chlorpromazine, flupentixol, and trifluoperizine decreased in nearly all age groups in both men and women. By 2000, more than 20% of male antipsychotic drug users aged 10–19 years were treated with risperidone, and a second peak in the pattern of its proportional use was evident in men aged 80–89 years old. The use of olanzapine was proportionately highest in 20–29-year-old men (nearly 20%), but it was used less frequently than risperidone in older men. We observed similar patterns of risperidone and olanzapine use by age in women.

Proportion of Users of Selected Antipsychotic Drugs Among Male Users, Standardized to Age Distribution of Male Users in 2000. 1991 () and 2000 ().

Proportion of Users of Selected Antipsychotic Drugs Among Female Users, Standardized to Age Distribution of Male Users in 2000. 1991 () and 2000 ().

Annual rates of first-time antipsychotic drug use

Women had a higher rate of first-time antipsychotic drug use (5.0 per 1,000 person-years) than men (3.3 per 1,000 person-years). Over the 10-year study period, the age-adjusted annual rate of first-time use decreased by 21% in women (P < 0.0001) but there was no temporal change in the age-adjusted rate of first-time use in men (P = 0.24).

Figure 3 presents the rates of first-time antipsychotic drug use by age in men and women. Although the rate was slightly higher in women than in men throughout the age range we studied, the age dependence of first-time use was nearly identical for men and women, rising gradually in the 10–59-year-old age groups and more sharply in the 60–99-year-old age groups.

Figure 3.

Sex-specific rates of first-time antipsychotic use by age.

Number of antipsychotic drug prescriptions and annual duration of use

The average annual number of antipsychotic prescriptions increased from 9.9 in 1991 to 12.2 in 2000, an increase of 2.3 prescriptions (23%) per year. Adjusting for correlation in the number of prescriptions from year to year within individual users, this increase was statistically significant (P < 0.0001). The increase occurred both in men (from 10.6 to 12.4 per year) and in women (from 9.6 to 12.0 per year). The increase in the annual number of antipsychotic drug prescriptions during the study period was not attributable to a change in the number of days covered by each prescription, which decreased only slightly from a median of 30 days in 1991 to 28 days in 2000. Taken together, these findings indicate that patients treated with antipsychotic drugs were treated on average for progressively longer times each year during the study period. There was no material variation in the number of prescriptions or the duration of individual prescriptions between the various antipsychotic agents we studied (data not shown).

Indications for antipsychotic drug treatment

Less than 10% of first-time users were treated for schizophrenia and other unspecified psychoses, and a much smaller proportion, 1.0%, were treated for bipolar disorder. Much more frequent indications for use of antipsychotic drugs were anxiety states, depression (frequently occurring with agitation, anxiety or end-stage disease) and panic disorders, which combined accounted for over 50% of prescriptions. Additionally, nearly 15% of prescriptions were given to treat agitation and dementia.

Further examination of the indications for treatment revealed several age-related differences. Prescriptions to treat depression, anxiety and panic attacks accounted for nearly 65% of first-time use in younger users (10–39 years), 57% in middle-aged users (40–69 years), but only 39% in the elderly (70+ years). In contrast, prescriptions to treat agitation or dementia accounted for 29% of new use in the elderly but only 4% and 7% in younger and middle-aged users, respectively.

Discussion

Although the rate of first-time antipsychotic drug use was constant in men and decreased in women from 1991 to 2000, we found that the annual use of antipsychotic drugs increased in both sexes during the study period, more so in men than in women. Our data suggest that this increase was due to an increase in the average annual duration of treatment with antipsychotic drugs. Because treatment with antipsychotic drugs was, on average, more prolonged in both men and women during the later years of the study period compared with the earlier years, the proportion of the study population using these drugs each year increased even though first-time use was stable (in men) or decreasing (in women).

Thioridazine was the most commonly used antipsychotic drug in both men and women from 1991 to 2000. Its use increased proportionally in most age groups during the study period while the use of other drugs (chlorpromazine, flupentixol, and trifluoperizine) decreased proportionally with the availability of newer ‘atypical’ antipsychotic drugs. The use of these newer ‘atypical’ antipsychotic drugs appeared to be still increasing at the end of the decade we studied.

The increases we observed in the average annual number of antipsychotic drug prescriptions given to each user remains unexplained, but could be related to several factors, including better patient tolerance of treatment (allowing patients to continue treatment longer) or changes over time in the duration of treatment required for the types of indications being treated (i,e. a shift toward treatment of more chronic conditions). Further analysis of individual patients’ characteristics and clinical outcome data would be needed to distinguish between these and other possible explanations. The sample of patients in whom we evaluated indications for antipsychotic drug use was too small to yield useful information on sex differences, ‘off-label’ use of specific antipsychotic drugs, or the use of antipsychotic agents in relation to other psychoactive drugs (such as benzodiazepines for anxiety or panic disorders). Evaluation of possible changes in the incidence of various psychiatric disorders over time was also beyond the scope of our study.

Our results were obtained using computerized medical records from general practices in the UK. The changes we observed may not pertain directly to patients treated in other settings (for example, psychiatric inpatients or individuals in nursing homes). Indeed, the use of ‘atypical’ antipsychotics by London psychiatrists in a small study was reported to be proportionately higher than we observed in the present study [2]. Nevertheless, because general practitioners in the UK initiate antipsychotic drug therapy for some patients and continue treatments begun by specialists for other patients, we believe that the data we have presented are likely to be more representative of national trends than data from psychiatric hospitals or clinics.

Other studies [3–5] have also found increases in the use of atypical antipsychotic drugs during the decade we studied but major differences in study design, such as the populations and providers studied, preclude direct comparison with our findings. In the United States, Hermann et al.[6] studied the prescription of antipsychotic drugs by office-based physicians using data from the National Ambulatory Medical Care Surveys from 1989 to 1997. They found that the rate of prescriptions for antipsychotic drugs increased approximately two-fold during this time period, and by 1997 the atypical antipsychotics risperidone and olanzapine were the most widely prescribed antipsychotics. This clearly differs from what we observed in the UK. In Australia, Mond et al. [7], studied prescriptions for antipsychotic drugs dispensed through community pharmacies during 1995–2001 using data from the national Health Insurance Commission, and estimated that more than 80% of antipsychotic drug use was prescribed by general practitioners. Although the dispensing of antipsychotic drugs increased by approximately two-thirds during the study period, the use of thioridazine decreased by 80%. No information was presented regarding age- or sex-specific use of various antipsychotics, nor were rates of new use or annual duration of use studied.

Our study provides the first large population-based estimates of changes in the patterns of use of antipsychotic drugs in the UK during the past decade. Since the CSM has recommended that thioridazine use now be restricted to second-line treatment of schizophrenia in adults, it will be of interest to see whether the use of thioridazine decreases over time, and if so, in what age groups and what types of patients. It will also be of interest to observe whether the use of newer ‘atypical’ antipsychotic drugs continues to supplant the use of older agents such as chlorpromazine, flupentixol, and trifluoperizine.