Journal of Affective Disorders, 21 (1991) 45-55 0 1991 Elsevier Science Publishers B.V. (Biomedical ADONIS 016503279100057V

45 Division)

0165-0327/91/$03.50

JAD 00770

Antidepressant

drug prescribing

in general practice:

a 6-year study

Matte0 Balestrieri I, Nicoletta Bragagnoli ’ and Cesario Bellantuono ’ I Servizio di Psicologia Medica, Istiruro di Psichmrria, Universirci di Verona and ’ Settore Farmaceutico,

ULSS 25, Verona, Izaly

(Received 26 June 1990) (Revision received 27 August 1990) (Accepted 25 September 1990)

We analysed antidepressant drugs (AD) prescription ratios of the GPs working in Verona, Italy, over a system (SIF-USL), were 6-year period (1983-1988). The data, provided by a local dru, (J information calculated as Defined Daily Dose (DDD), which is the unit of drug consumption recommended by WHO. We found that DDD/lOOO patients/day increased over the period, mainly because of an increase in the use of ‘second-generation’ antidepressants and other non-tricyclic antidepressants. An increase in the levels of prescription of AD was observed over the 6 years. This increase was statistically significant when comparing the first (1983) with the other years. Low correlations were found between DDD/patient/year ratios and GPs’ age, sex and list size. Harmonic analysis of the seasonal variations in prescriptions of AD revealed a substantial pattern of seasonality, in which the first four harmonics accounted for the greater part (95.5%) of the seasonality. AD prescribing may be linked more closely to seasonal holiday patterns than to seasonality in the onset of depressive disorders.

Key words: Antidepressant

prescribing;

Seasonality;

Introduction Many studies conducted in the last 20 years have documented that the vast majority of psychiatric disorders in the community are treated by GPs rather than by psychiatrists (Shepherd, 1966; Goldberg and Huxley, 1980). It has also been

Address for correspondence: Dr. Matte0 Balestrieri, Servizio di Psicologia Medica. Istituto di Psichiatria, Ospedale Policlinico, I-37134 Verona, Italy.

Trends;

Variation

demonstrated that the most common treatment strategy for emotional and psychiatric problems at the primary care level is the prescription of a psychotropic drug (Williams, 1979; Gabe and Williams, 1986; Williams and Bellantuono, 1990). A considerable amount of this research has been conducted in English-speaking countries (U.K., U.S.A., Australia) while less information is available from mainland Europe (Tognoni et al., 1981; Cooperstock and Pamell, 1982). In Italy a number of studies have been published on psychotropic drug use. These are (a) a

46

community survey of self-reported psychiatric drug consumption (Siciliani et al., 1985); (b) a study in which national data on psychotropic drug sales were examined for the years 1975-1984 (Williams et al., 1986); (c) two small-scale investigations carried out in two different geographical areas in Northern Italy, based on drug prescriptions collected by GPs (Bellantuono et al., 1988; Fiorio et al., 1988); (d) a large-scale cross-sectional survey in which nearly 100 GPs were involved (Bellantuono et al., 1989a); and (e) a psychotropic drug monitoting study in general practice over a 2-year period (1983-1984) using a computerised drug information system (Bellantuono et al., 1987). This latter study was considered a preliminary investigation carried out not only to describe the patterns of psychotropic drug prescription in general practice, but also to assess the utility of a drug information system as a technique for monitoring psychotropic drug use in a well-defined geographical area. However? it should be noted that the system, the Sistema Informativo Farmaci of the Unit2 Sanitaria Locale (Local Health Unit) (SIF-USL), is based on computerised monitoring of drugs included in the Italian National Formulary (INF). For this reason benzodiazepine prescriptions were only partially recorded, as these drugs were excluded from the INF from September 1984. In the present study the SIF-USL system has been used to monitor antidepressant drug (AD) prescriptions. This drug category is included in the INF and is therefore recorded by the system. The main aims of this study can be set out as follows: (a) to describe the trends of AD prescription in general practice from 1983 to 1988; (b) to investigate the monthly variations in these prescriptions over the same period; (c) to assess between-doctor variations in AD prescriptions. Methods The area The geographical area of investigation is that covered by Local Health Unit No. 25 of the Veneto region and corresponds to Verona and surrounding villages. Verona is a town sited in Northern Italy half-way between Milan and Venice. On 31 December 1988 its population was

258.196 (density 1297 people/km2), with a male/female ratio of 0.89. About 12% of the population is under the age of 14 years, 71.5% between 14 and 64 years and 16.5% over 64 years. In the last 10 years the population has been declining slightly because of a falling birth rate. The Local Health Unit (USL) is responsible for providing health services for a population of about 300,000 (7% of all population of the Veneto region). In the area there are three general hospitals, 78 municipal pharmacies and the practices of about 300 general practitioners (GPs). All GPs are in practice by themselves, like the great majority of Italian GPs.

The drug information system SIF-USL In 1983 the Department of Pharmacy of USL 25 set up a drug information system, the SIF-USL, based on intensive computerised monitoring of all drugs included in the INF that were prescribed by GPs operating in USL 25 and collected by the pharmacies located in the same area (Bozzini and Tognoni, 1987). For each prescription the SIFUSL system records the standard input of pharmacy code, prescribing doctor and drug name. Prescription monitoring can be done monthly on the total drug prescriptions or on a sample for a selected period of time. Examples of output programmes run by SIF-USL include: identification of general prescription indices and administrative data (for example, total drug expenditure, cost per patient, cost per doctor); analysis of drug prescriptions by therapeutic class and by active ingredient; and identification of prescription indices and drugs prescribed by each doctor. The SIF-USL system classifies drugs into 27 categories. Due to the exclusion of the benzodiazepines (see above) only two groups of psychotropic drugs remain in the National Formulary: antidepressants (AD) and neuroleptics. The category AD includes: tricyclic antidepressants (TCA), second-generation antidepressants (2GAD) and other antidepressants (OAD). Table 1 shows all AD marketed in Italy between 1983 and 1988. From the end of January 1984 each doctor working as a GP in USL 25 received information, every 6 months, about drug prescriptions written by himself in the previous 6 months. This feed-

47 TABLE 1 PRESCRIPTION

OF ANTIDEPRESSANTS,

CALCULATED

AS DDD/lOOO REGISTERED

PATIENTS/DAY

1984

1985

1986

1987

1988

Mean

277,016 310

274,125 309

272,400 288

272,194 282

272,422 267

274.779.0 295.2

1.42 0.03 0.71 0.01 0.02

1.41 0.03 0.66 0.01 0.02

1.38 0.01 0.61 0.01 0.01 0.12

0.00 0.73

1.25 0.02 0.62 0.01 0.01 0.15. 0.00 0.59 _

1.30 0.02 0.60 0.01 0.01 0.15

0.00 0.80

1.32 0.03 0.62 0.01 0.01 0.09 0.00 0.65

0.57

0.51 0.01 0.38 0.07

0.58 0.01 0.35 0.07

0.59 0.01 0.38 0.07

0.67 0.01 0.41 0.06

0.75 0.00 0.45 0.06

0.53 0.00 0.70 0.00 0.43 0.05

1.35 0.02 0.64 0.01 0.01 0.13 0.00 0.64 0.00 0.63 0.01 0.40 0.06

3.96

3.86

3.78

3.79

3.91

3.88

3.86

antidepressants 0.18 0.51

1.15 0.63

0.18 0.10

0.19 0.10

1.43 0.63 0.11 0.20 0.06

1.73 0.73 0.30 0.21 0.05

1.70 0.67 0.41 0.21 0.04

1.92 0.63 0.46 0.19 0.03

1.35 0.63 0.32 0.20 0.06

Total

0.96

2.07

2.42

3.01

3.04

3.24

2.46

Other antidepressants Ademetionine oxitriptan

0.18

0.11

0.07 0.11

0.38 0.07

0.74 0.07

0.74 0.07

0.48 0.10

Total

0.18

0.11

0.19

0.45

0.80

0.81

0.42

All antidepressants Total

5.10

6.04

6.38

7.25

7.75

7.93

6.74

1983 Registered patients (n) 280,517 GPs (n) 315 Tricyclic antidepressants Amitriptyline Butriptyline Clomipramine Desipramine Dibenzepin Dothiepin Doxepin Imipramine Lofepramine Maprotiline Melitracen NortriptyIine Trimipramine Total Second-generation Amineptine Mianserin Minaprine Trazodone Viloxazine

back procedure was adopted in order to provide direct information on drug prescriptions to GPs. Following the advice of WHO, which recommends the Defined Daily Dose (DDD, see below) as a unit of measurement for comparative drug consumption statistics, the data provided by the SIF-USL system are calculated both as number of items sold and as DDD. Indices of prescriptions Drug consumption in a population can be measured using several indices. Some indices are expressed in terms of items sold (boxes, tablets,

capsules, suppositories, vials and so on), others in terms of physical units (grams, millimoles, liters, international units with their respective multiples and fractions), yet more are based on cost. Indices based on the number of items sold are only adequate when the amount of drugs prescribed is compared with that existing in a previous adjacent period in the same or a homogeneous setting. If this is not the case, difficulties can arise: for example, the product marketed by a pharmaceutical company in an area may differ in terms of quantity and concentration of active ingredient from the corresponding product marketed

48

by another company in a nearby area, and these may also differ between regions and countries. The comparison of quantities expressed in common physical units is a more reliable way of studying drug consumption, but there may still be problems. For example, a number of preparations are sold in combination at a fixed dose; moreover, with this method it is difficult to compare drugs of the same therapeutic group, but with different pharmacological power. Comparisons based on cost can also be misleading, because there are price differences between alternative preparations marketed by different companies, cost levels are dissimilar between countries and prices and currencies are constantly changing. For all these reasons WHO recommends the use of a different unit of measurement: the Defined Daily Dose (DDD). The DDD is a technical unit of measurement, defined as the assumed auerage dose of the preparation per 24 h, used for the main indication in adults. For drugs which have an initial dose and a maintenance dose, the latter is chosen. The statistics are generally given in DDD per 1000 inhabitants per day. Alternatively, one can calculate the DDD per 1000 patients per day ratio. When a study is carried out in hospitals another statistic can be used: DDD per 100 bed days. Drug consumption, expressed in DDD/lOOO patients/day, may also give a rough indication of how many patients receive ‘standard’ treatment with a drug. However, it does not indicate how many patients are treated over a given period. Only for a drug used continuously can consumption data be used to indicate the number of people on treatment. The DDD was first defined in 1970 by a group of experts from Northern Ireland, Sweden and Norway. A few years later this group was enlarged to include experts from other Nordic countries (Denmark, Finland, Iceland, Norway and Sweden) and became the WHO-Drug Utilization Research Group (WHO-DURG). The first complete DDD list was published in Norway, for most of the drugs for systemic use registered in that country (Baksaas et al., 1975). The Nordic Council of Medicines (1979) published a booklet (Drug Statistics on Medicines) with a DDD list for all drugs

registered in the Nordic countries. Since then, every 3 years a new edition of that booklet has been published; the last one was published in 1988 (Nordic Council on Medicines, 1988). In Italy a partial DDD list was published in a pilot study in 1988 (Bozzini et al., 1988), and the first complete DDD list appeared the following year (Castellani et al., 1989). Data-set We analysed AD prescription data from GPs working in Verona over a 6-year period (19831988). Prescriptions were collected by the municipal pharmacies and sent to the Department of Pharmacy of USL 25, for processing by SIF-USL 25. This data-base provided data on sex, age, list size and number of AD prescribed (expressed in number of items as well as in DDD) per GP, and per year. Moreover we received from SIF-USL a printout containing the list of DDD sold in every year for each AD. Eventually we were provided with a DDD list of AD sold in every month during the 1983-1988 period. A preliminary analysis was performed on the amount of DDD/lOOO patients/day sold throughout the 6 years, comparing the three classes of AD: TCA, 2GAD and OAD. The GPs were then grouped into five classes, according to their rate of prescription (expressed as DDD/patient/year), as follows: - very low prescribers: up to 1; - low prescribers: more than 1 and up to 2; _ medium prescribers: more than 2 and up to 3; - high prescribers: more than 3 and up to 4; _ very high prescribers: more than 4. Statistics We tested the 6 years in pairs in order to check for differences among the years in the distribution of the five classes of GPs. For this purpose we used the two-tailed Kolmogorov-Smimov test for two samples with the Bonferroni correction. A significance level of a = 0.05 was adopted. Then for each year we used Pearson’s r for correlating the distribution of the rates of prescription of the GPs with the distribution of their age, sex and list size. Finally, we tested the data for seasonality of DDD prescriptions of AD. A number of tests to

49

estimate the significance of cyclical variation are available. However, the lack of independence between observations in successive months and the possibility of substantial linear trends over time preclude the use of many, including Edwards’ test (1961) or Walter and Elwood’s (1975) modification of it, or of Freedman’s (1979) non-parametric test. On the other hand, Hewitt et al.‘s (1971) rank-sum test, which does not assume independence, is insufficiently powerful (Walter, 1980). Furthermore the parametric tests assume a model for the seasonal variation of one cycle per year, which may be not accurate. Thus, we have applied harmonic analysis (Pocock, 1974) which makes it possible to distinguish between seasonal and nonseasonal components. To remove all trends, i.e., the mean differences between the years, an analysis of variance approach was used. The GLIM package was utilized (Payne, 1986), fitting to monthly data a generalised linear model, specifying a Poisson error with link identity. The main effects of years were then subtracted from the original data and the residuals so obtained were subjected to harmonic analysis, using the technique described by Pocock (1974). In this method all the harmonics are estimated and then categorised as seasonal or nonseasonal, depending on whether . or not the frequency per annum is an integer. Variance can be partitioned into seasonal, non-seasonal and random components and the level of significance of each component assessed. The contribution of individual harmonics can also be estimated. Since harmonic analysis estimates frequencies of discrete numbers, we used the absolute numbers of DDD sold each month instead of population ratios. Results Table 1 shows the number of GPs whose prescriptions were collected in the 1983-1988 period by the SIF-USL system, the number of patients registered with them and the ratio DDD/lOOO registered patients/day for each drug, for the three classes of AD and for the total. A mean of 295 GPs operated in USL 25, with a decreasing trend over time. The number of patients also declined over the 6 years, from 280,517 to 272,422 (mean 274,779). This decline was not due to changes in the method of data collection.

B

_

fl TCA B End-genero:ion

AD

27 D \ ; 6

i ;e 5 c 0 48 : 3: 0 El1983

I984

I985

1986

’ 987

1988

-+calrs

Fig. 1. Consumption of the three classes of AD during 6 years (1983-1988).

The mean list size over the 6 years was 931 patients per doctor (range of the years: 887-1020). The rise of AD prescriptions

If we assume that all AD sold were also consumed, a mean of 6.7 people per 1000 patients TABLE 2 PERCENTAGES OF ALL PRESCRIPTIONS (DDD/lOOO REGISTERED PATIENTS/DAY) OF ANTIDEPRESSANTS: THREE CLASSES OF ANTIDEPRESSANTS AND SIX MOST PRESCRIBED DRUGS 1983

1988

(S)

(%)

Mean 1983-1988 (S of total)

Difference between 1983 and 1988 percentages

Tricylic antidepressants Amitriptyline 21.8 Clomipramine 13.9 Imipramine 15.7 Maprotihne 10.0

17.4 7.7 6.7 8.8

20.0 9.5 9.5 9.3

- 10.4 -6.2 - 9.0 -1.2

Total

77.6

48.9

57.3

- 28.7

Second-generation Amineptine Mianserin

antidepressants 3.5 24.2 10.0 7.9

20.0 9.3

+ 20.7 - 2.1

Total

18.8

40.8

36.5

+ 22.0

Other antidepressants Total 3.5

10.2

6.2

+ 6.7

AN antidepressants Total 100.0

100.0

100.0

0.0

50

took an AD every day. The DDD/lOOO patients/day ratio increased over the period (+55.5%), from 5.1 to 7.9. Thirteen TCA, five 2GAD and two OAD were prescribed. Fig. 1 shows that the consumption of all TCA remained quite stable (range: 3.78-3.96), while that of 2GAD increased more than three times (from 0.96 to 3.24) and that of OAD more than four times (from 0.18 to 0.81). Table 2 reports the DDD/lOOO patients/day ratios in 1983 and 1988 as percentages compared to the total. Only the six most prescribed AD are shown. It can be seen that the 1983-1988 proportion of TCA within the group of AD decreased ( - 28.7%), while the proportions of 2GAD and of OAD increased (+ 22% and + 6.7%, respectively). In 1988 the four most prescribed TCA (amitriptyline, clomipramine, imipramine, maprotiline) represented 83.0% of all TCA (in 1983: 86.8%) and the two most prescribed 2GAD (amineptine, mianserin) represented 78.7% of all 2GAD (in 1983: 71.8%). As for the OAD, it is noteworthy that 91.3% of the prescriptions were for ademetionine. Between-doctor Fig. 2 shows to their rate of bars represent year. It can be than one third

variation the distribution of GPs according prescription over the 6 years. The percentages of the total for each seen that, apart from 1983, more of the GPs (range 33.2-36.5s)

1 ;40

1984

I 1983

Ill

0 1986

K31987

B

1985

a

> >

ii3 t i 20 i ; : IO ; II

0.0-I

.o

Fig. 2. Distribution

I .o-2.0

2.0-3.0

3.0-4.0

of GPs over five prescription 6 years (1983-1988).

4.ot categories

in

TABLE

3

HARMONIC SOLD EACH

ANALYSIS MONTH

OF

THE

NUMBER

Total variance First harmonic: Second harmonic: Third harmonic: Fourth harmonic: Fifth harmonic: Sixth harmonic: Seasonal

variation

Non-seasonal variation Random

Seasonal variance

1 cycle/year 2 cycles/year 3 cycles/year 4 cycles/year 5 cycles/year 6 cycles/year

9.8 30.7 12.0 8.9 2.6 0.3

15.2 41.8 18.6 13.9 4.0 0.5

(all harmonics)

64.3

100.0

DDD

(%)

and non-random

variation

Total sample

(8)

OF

35.5 0.2 100.0

were medium prescribers. In 1983 there was a relatively large proportion of GPs (about 40%) that were low prescribers. A comparison of the outlines of the five groups of bars (Fig. 2) shows that there was an increase in the numbers of GPs classified as high and very high prescribers over the 6 years. However, only 1983 was significantly different from the other years using the Kolmogorov-Smimov test with the Bonferroni correction. No significant differences emerged between each pair of years. Low correlations were found between DDD/ patient/year ratios and list size in the various years (range of r coefficients: -0.04-0.13). Similar coefficients resulted also from correlations between DDD/patient/year ratios and other indices, such as GP’s age and sex. Seasonal variation As far as the harmonic analysis of the prescriptions of AD is concerned, Table 3 shows that the seasonal harmonics accounted for about 64% of the total sample variance (P -C 0.001). x2 tests for single harmonics were also highly significant (P < O.OOl), but the first four harmonics (1, 2, 3 and 4 cycles per year) accounted for the greater part (95.5%) of the seasonality. Fig. 3 shows the similarity between the observed and the expected DDD/month values. The

51

1 \/

60000 L ; 55000 1 ; 50000 ; 45000 40000

30000

L

JFMRHJJRSONO

--

1

JFHRMJJASONOJFHAMJJRSONOJFnRHJJASONOJFHRMJJRSONO

Fig. 3. Seasonal variation of the prescriptions of AD during 6 years numbers of DDD sold in each month, while the dotted tine represents

expected values derived from a harmonic analysis model in which only the above four harmonics were included. There is a trend of increasing AD consumption during the spring, culminating in a May peak. Afterwards there is a marked drop in consumption during June and July, followed by a rise in the autumn, with a peak in October. In terms of number of people consuming AD, the difference between the May peak and the subsequent trough corresponds to about 10,481 people in a month. Discussion This is the first AD monitoring study conducted in Italian general practice in which drug prescription data have been expressed in DDD/lOOO patients/day. It should be noted that DDD data are only available in about 20% of Italian Local Health Units so that no comparison can be made between our data and DDD data at national and regional levels. Psychotropic drug sale data for Italy were examined for the years 1975-1984 by Williams et al. (1986). They found a consistent annual increase in sales of AD as well as of benzodiazepines and, to a lesser extent, of neuroleptic drugs. The largest

JFHAMJJRSONO

(1983-1988). The continuous line represents the observed the corresponding values predicted by harmonic analysis.

percentage increase was for the category of AD, which by 1984 had increased by 119% over the 1975 level; the corresponding figure for benzodiazepines and neuroleptics were, respectively, 95% and 20%. Regional differences in psychotropic drug sales for 1983-1984 were also examined. Marked differences were found and, in general, levels of use were higher in North-Central than in Southern regions. The variation was greatest for AD, the prescribing of which ranged from 54% of the regional average (in Puglia) to 234% (in Liguria). Looking at these regional differences, our data are more likely to be representative of the practice of GPs in Northern Italy than that of those in the Southern regions. The rise of AD prescriptions The results of this 6-year study clearly show a consistent and significant annual increase in AD prescription from 1983 to 1988 (5.10 to 7.93 DDD/lOOO patients/day) with an increase of about 60%. Even though these data cannot be directly correlated with AD sale data analysed by Williams et al. (1986) it seems that in Italy in the general practice setting, at least for the AD, the ‘psychotropic drug juggernaut’ (see Trethowan,

52

1975) has continued its relentless progress over the years. More than one reason can account for this rise in AD prescription in Italy. The first relevant factor could be an improvement of GPs’ ability to manage patients with depressive disorders. This might be the consequence of their growing interest in and concern for psychiatric problems over the last decade. The 1978 Italian Psychiatric Reform has, in fact, had a significant impact not only among psychiatrists but also among GPs, especially those working in areas with a good-practice community psychiatric service (Tansella, 1991). A number of studies conducted in general practice in Italy have documented that on average about one third of patients consulting the GP in one day are affected by emotional problems, mostly anxiety states and/or depressive disorders (Bellantuono et al., 1991). A minority of these patients (about 27%) are referred by GPs to specialist psychiatric services (Arreghini et al., 1990). This confirms that the majority of patients affected by ‘minor’ psychiatric morbidity are managed by GPs (Shepherd et al., 1966; Goldberg and Huxley, 1980; Tansella et al., 1986). It must be remembered, however, that not all depressive disorders respond to AD treatment. A statistical review of placebo-controlled trials of imipramine in the treatment of depression supports the effectiveness of this TCA in acute ‘endogenous’ depression, but indicates that the value of AD drugs in ‘chronic, atypical, neurotic, reactive’ depression is much less clear (Roger and Clay, 1975). The implication for clinical practice is that GPs should be trained to recognise and treat patients who are more likely to respond to AD (i.e., patients with ‘major, bipolar, recurrent’ depression). A second possible factor which might have contributed to the increased AD prescription is the availability in the Italian pharmaceutical market of several ‘new’ antidepressants (2GAD plus OAD, see Table 1), which, with TCA and neuroleptics, are provided by the National Health Service even though a fee (20% of the price) must be paid by the patients. On the other hand, they must pay the fulI price for benzodiazepines. These ‘new’ AD drugs are promoted in the general practice setting as having a better side-effect profile

and less toxicity than the ‘old’ TCA (no significant anticholinergic and cardiovascular effects). The possible advantages of these AD could have encouraged GPs to treat more depressed patients themselves whom they previously would have referred to psychiatrists. Analysis of our DDD data seems to support this hypothesis, as the increase of AD prescriptions is only confined to these compounds, while the level of TCA prescriptions has remained stable over the 6 years (Fig. 1). However, it is worth noting that despite claims of their therapeutic efficacy and safety, at the specialist level of care these drugs are usually considered as ‘AD of second choice’. As Blackwell (1987) and Gelemberg (1989) pointed out, those AD are more likely to be used after a TCA has already failed or is contraindicated. Furthermore for drugs like amineptine, minaprine and ademetionine there is little convincing evidence of antidepressant activity since the studies conducted so far contain significant methodological failings (Bellantuono et al., 1989b). A third explanation for the rise of AD prescription in general practice could be the campaign against benzodiazepines (Taylor, 1989) which for the last 10 years has been carried out not only in scientific journals but also by the mass media. General practice patients are now aware of the possibility that benzodiazepines can induce dependence and become increasingly concerned about this risk. It may be hypothesised that this concern could have induced the GPs to prescribe an AD instead of a benzodiazepine (for example for patients with mixed anxiety and depression). Unfortunately we do not know if during the period in question the prescription of benzodiazepines increased, remained stable, or decreased. A study on benzodiazepine prescription over the last 10 years is crucial to test such an hypothesis and will be a priority in our future research. Between-doctor variation Between-doctor variation in psychotropic drug prescribing has been frequently observed. For example, the Birmingham Research Group of the Royal College of General Practitioners (1978) observed a lo-fold variation in psychotropic prescribing by 100 doctors. Fleming and Cross (1984)

found a two-fold difference in the prescribing of new psychotropic drug prescriptions, a four-fold difference for continuing prescriptions and a lofold difference for repeat prescriptions. In a previous study conducted in South-Verona the variation in the overall rate of psychotropic prescribing was almost 20-fold (Bellantuono et al., 1987). A number of studies have tried to determine the causes of such differences. Among them, the doctor’s personal characteristics (i.e., age, sex, medical education), attitudes, workload and the urban/rural location of the clinic have been invoked. List size is usually taken as a measure of the GP’s workload and is often considered to be an index of the amount of time available to patients. In the present study no correlations between prescribing rate and variables such as GP’s sex, age and list size have been found. In the literature two studies have been published reporting a correlation between list size and number of psychotropic prescriptions (Stolley et al., 1972; Hartzema and Christiansen, 1982). But a greater number of studies have not found such a correlation (Fleming and Cross, 1984; King et al., 1982; Gabe and Williams, 1986; Bellantuono et al., 1987; Holm and Olesen, 1988). Moreover, Holm and Olesen (1988) found no correlation between psychotropic prescribing and GP’s sex. and age. When considering our findings it should be noted that all GPs entering the study were in practice by themselves. The mean list size (930.8) was slightly less than that observed by Holm and Olesen (1988) in Denmark (1135.4) and less than half of that observed by Fleming and Cross (1984) and by Gabe and Williams (1986) in the United Kingdom (2448 and 2369, respectively). Since the present investigation found no important single explanation for the great differences in prescription rates, we are now exploring other hypotheses. A study in progress is investigating the influence of practice location on the pattern of AD prescribing. Seasonal variation There is a growing body of literature on seasonal variations in the incidence and prevalence of various psychiatric disorders. Several analyses have been conducted at various levels; however, to our

knowledge, so far only two studies have examined monthly variation of psychotropic prescribing. In one, Williams and Dunn (1981) analysed the trend in prescriptions for five categories of psychotropic drug (stimulants/appetite suppressants, tranquillisers, antidepressants, barbiturate hypnotics and non-barbiturate hypnotics) dispensed at retail pharmacies in England during the years 19691975. These authors observed that patterns of cyclical variation for the different types of drug varied widely and that antidepressants, in particular, exhibited a thrice-yearly cycle. All psychotropic drugs were consistent in showing a relative deficit of prescribing in August. The second study was conducted by our research group on South-Verona psychotropic drug prescribing data provided by the SIF-USL and concerned a 2-year period (1983-1984) (Bellantuono et al., 1987). The results indicated that monthly variation for the different drug classes (benzodiazepine hypnotics, antidepressants, neuroleptics) followed a similar pattern. However, a thrice-yearly cycle was evident only in 1984, while 1983 did not show any clear cyclical pattern, apart from a summer trough. The present study differs from the previous one in three ways: (a) the area of investigation (the whole municipality of Verona and surroundings) has a population three times bigger than that of the South-Verona area; (b) the period of observation is three times longer; and (c) the research focussed only on AD prescriptions. For these reasons our present findings can be considered more robust than the previous ones in describing the pattern of AD prescription. Compared with the study by Williams and DUM (1981) our study has confirmed the presence of a seasonal variation. The model which best describes our data includes four kinds of harmonics (1, 2, 3 and 4 cycles/year), one of which was also present in the English study (3 cycles/year). A summer trough is present in both studies, and is also confirmed by Harris (1984), in his study on consultations for depression in five London general practices. The seasonal variation identified by our study (a rise in spring, a drop in summer, a further rise in autumn, followed by a small decline in winter) suggests that AD prescribing may be linked more

54

closely to the seasonal holiday patterns than to seasonality in the onset of depressive disorders. Acknowledgements We are grateful to Dr. L. Bozzini, Director of the Pharmaceutical Department of USL 25, Verona, for permission to use SIF-USL data and to Mr. G. Nota for the provision of those data. We are also grateful to Dr. R. Micciolo for his help with the statistical analysis and to Prof. M. Tansella and Dr. G. Wilkinson for their comments on the first draft of this paper. The present study was supported by the Consiglio Nazionale delle Ricerche (CNR, Rome), Progetto Finalizzato Fattori di Malattia (FATMA), Sottoprogetto ‘Stress’, Contract to Professor M. Tansella. References Arreghini, E., Agostini, C. and Wilkinson, G. (1990) GP referral to specialist psychiatric services. A study comparing GPs’ practices in North- and South-Verona. Psychol. Med. (in press). Baksaas, AI. et al. (1975) Drug Dose Statistics. List of Defined Doses for Drugs Registered in Norway. Norsk Medisinaldepot, Oslo. Bellantuono, C., Fiorio, R., Williams, P., Martini, N. and Bozzini, L. (1987) Psychotropic drug monitoring in general practice in Italy: a two-year study. Family Pratt. 4, 41-49. Bellantuono, C., Fiorio, R., Williams, P., Arreghini, E. and Cason, G. (1988) Urban-rural differences in psychotropic prescribing in northern Italy. Eur. Arch. Psychiatry Neural. Sci. 237, 347-350. Bellantuono, C., Arreghini, E., Adarni, M., Bodini, F., Gastaldo, M. and Micciolo, R. (1989a) Psychotropic drug prescription in Italy. A survey in general practice. Sot. Psychiatry Psychiatr. Epidemiol. 24, 212-218. Bellantuono, C., Balestrieri, M. and Adami, M. (1989b) I farmaci antidepressivi. In: C. Bellantuono and M. Tansella (Eds.), Gli Psicofarmaci nella Pratica Terapeutica, 2nd edn. 11 Pensiero Scientifico, Rome, pp. 149-200. Bellantuono, C., Williams, P. and Tansella, M. (1991) Psychiatric morbidity in general practice. In: M. Tansella (Ed.), Community-based Psychiatry: Long-term Patterns of Care in South-Verona, Psychological Medicine Monograph Supplement. Cambridge University Press, Cambridge. BlackwelL B. (1987) Newer antidepressant drugs. In: H.Y. Meltzer (Ed.), Psychopharmacology, The Third Generation of Progress. Raven Press, New York, NY, pp. 1041-1049. Bozzini, L. and Tognoni, G. (1987) SIF-USL: 11 Sistema Informativo Farmaci per I’UnitB Sanitaria Locale. G. Ital. Farm. Clin. 1, 99-108. Bozzini, L., Castellani, L. and Tognoni, G. (1988) Analisi e

valutazione delle prescrizioni farmaceutiche in una USL de1 Veneto negli anni 1984-1985-1986. G. Ital. Farm. Clin. 2, 81-99. Castellani, L., Bozzini, L., Pedrini, A., Carrara. F., Nasi, G.F., Ferrarese, A. and Tognoni, G. (1989) Prima lista italiana delle DDD di alcuni gruppi di farmaci. G. Ital. Farm. Clin. 3, 81-135. Cooperstock, R. and Pamell, P. (1982) Research on psychotropic drug use. A review of findings and methods. Sot. Sci. Med. 16, 1179-1196. Edwards, J.H. (1961) The recognition and estimation of cyclic trends. Ann. Hum. Genet. 25, 83-87. Fleming, D.M. and Cross, K.W. (1984) Psychotropic drug prescribing. J. R. Coll. Gen. Practit. 34, 216-220. Freedman, L.S. (1979) The use of a Kolmogorov-Smimov type statistic in testing hypotheses about seasonal variation. J. Epidemiol. Comm. Health, 33, 223-228. Gabe, J. and Williams, P. (1986) Rural tranquillity? Urban-rural differences in tranquillizer prescribing. Sot. Sci. Med. 22, 1059-1066. Gelemberg, A.J. (1989) New perspectives on the use of tricyclic antidepressants. J. Clin. Psychiatry 50 (Suppl.). 7. Goldberg, D.P. and Huxley, P. (1980) Mental Illness in the Community: The Pathway to Community Care. Tavistock. London. Harris, CM. (1984) Seasonal variations in depression and osteoarthritis. J. R. Coll. Gen. Practit. 34, 436-439. Hartzema, A.G. and Christiansen, D.B. (1982) Nonmedical factors associated with the prescribing volume among family practitioners in an HMO. Med. Care 10, 7-28. Hewitt, D., Milner, J., Csima, A. and Pakula, A. (1971) On Edwards’ criterion of seasonality and a non-parametric alternative. Br. J. Prev. Sot. Med. 25, 174-176. Holm, M. and Olesen, F. (1988) Factors affecting prescription of psychotropic drugs in general practice. Scand.:$. Prim. Health Care 6. 169-173. King, D., Griffiths, K., Reilly, P. and Merret, J.D. (1982) Psychotropic drug use in Northern Ireland 1966-1980: prescribing inter- and intra-regional comparisons and relationship to demographic and socioeconomic variables. Psychol. Med. 12, 819-833. Martini, N., Castellani, L. and Scroccaro, G. (1985) Monitoring of prescription patterns in district medicine by a drug information system (SIF-USL). In: J. Bona1 and J. Poston @Is.), Progress in Clinical Pharmacy. Cambridge University Press, Cambridge. Nordic Council on Medicines (1979) Nordic Statistics on Medicines, 1975-1977. Helsingfors. Nordic Council on Medicines (1988) Nordic Statistics on Medicines. 1984-1986. Uppsala. Payne. C.D. (1986) The GLIM System Release 3.77 Manual. Generahsed Interactive Modelling. Royal Statistical Society, London. Pocock, S.J. (1974) Harmonic analysis applied to seasonal variations in sickness absence. Appl. Stat. 23. 103-120. Roger, S.C. and Clay. P.M. (1975) A statistical review of controlled trials of imipramine and placebo in the treatment of depressive illness. Br. J. Psychiatry 127. 590-603.

55 Shepherd, M., Cooper, B., Brown, A.C. and Kalton, G. (1966) Psychiatric Illness in General Practice. Oxford University Press, London. Siciliani, O., Bellantuono, C., Williams, P. and Tansella, M. (1985) Self-reported use of psychotropic drugs and alcohol abuse in South-Verona. Psychol. Med. 15, 821-826. Stolley, P.D., Becker, M.H., Lasagna, L., McEvilla, J.D. and Sloane, L.M. (1972) The relationship between physician characteristics and prescribing appropriateness. Med. Care 10, 17-28. Tansella, M. (Ed.) (1991) Community-Based Psychiatry. Long-Term Patterns of Care in South-Verona. Psychol. Med. Monograph Supplement, Cambridge University Press, Cambridge. Tansella, M., Williams, P., Balestrieri, M., Bellautuono, C. and Martini, N. (1986) The management of affective disorders in the community. J. Affect. Disord. 11, 73-79. Taylor, F.K. (1989) The damnation of benzodiazepines. Br. J. Psychiatry 154, 697-704. Tognoni, G., Bellantuono, C. and Lader, M. (Eds.) (1981) The Epidemiological Impact of Psychotropic Drugs. Elsevier/ North Holland, Amsterdam.

Trethowan, W.H. (1975) Pills for personal problems. Br. Med. J. iii, 749-751. Walter, S.D. (1980) Exact significance levels for Hewitt’s test for seasonality. J. Epidemiol. Comm. Health 34, 147-149. Walter, S.D. and Elwood, J.M. (1975) A test for seasonality of events with a variable population at risk. Br. J. Prev. Sot. Med., 29, 18-21. Williams, P. (1979) The extent of psychotropic drug prescription. In: P. Williams and A. Glare (Eds.), Psychological Disorders in General Practice, Academic Press, London, pp. 151-160. Williams, P. and Dunn, G. (1981) Cyclical variation in psychotropic drug prescription. J. Epidemiol. Comm. Health 35, 136-138. Williams, P. and Bellantuono, C. (1990) Long-term tranquillizer use. The contribution of epidemiology. In: J. Gabe (Ed.), Understanding Long-term Tranquillizer Use. Routledge, London. Williams, P., Bellantuono, C., Fiorio, R. and Tansella. M. (1986) Psychotropic drug use in Italy: national trends and regional differences. Psychol. Med., 16, 841-850.

Antidepressant drug prescribing in general practice: a 6-year study.

We analysed antidepressant drugs (AD) prescription ratios of the GPs working in Verona, Italy, over a 6-year period (1983-1988). The data, provided by...
1MB Sizes 0 Downloads 0 Views