European Journal of Clinical Pharmacology

, Volume 60, Issue 1, pp 51–55

Socio-economic inequalities in first-time use of antidepressants: a population-based study

Authors

    • Research Unit of General PracticeUniversity of Southern Denmark
  • J. Søndergaard
    • Research Unit of General PracticeUniversity of Southern Denmark
  • W. Vach
    • Department of StatisticsUniversity of Southern Denmark
  • L. F. Gram
    • Clinical PharmacologyUniversity of Southern Denmark
  • J. U. Rosholm
    • Department of GeriatricsOdense University Hospital
  • P. B. Mortensen
    • National Center of Register-based ResearchAarhus University
  • J. Kragstrup
    • Research Unit of General PracticeUniversity of Southern Denmark
Pharmacoepidemiology and Prescription

DOI: 10.1007/s00228-003-0723-y

Cite this article as:
Hansen, D.G., Søndergaard, J., Vach, W. et al. Eur J Clin Pharmacol (2004) 60: 51. doi:10.1007/s00228-003-0723-y

Abstract

Objective

To analyse whether first-time use of antidepressants (incidence) and selection of TCAs (tricyclic antidepressants) versus new-generation drugs are associated with socio-economic status and psychiatric history.

Method

We conducted a population-based cohort study using registry data covering Funen County, Denmark. A total of 305,953 adult residents without antidepressant prescriptions 5 years prior to the study period (1998) were included.

Results

The 1-year incidence rate of antidepressant prescription (1.7%) increased with age. It was higher in people who were female, less educated, unemployed, those receiving old-age or disability pension, low-income groups, and singles. The proportion prescribed new-generation antidepressants (82%) showed no difference according to socio-economic variables (education, annual income and socio-economic group), but was higher among the young and single. Admission to psychiatric hospital within 4 years prior to the study period was associated with high-incidence rate of antidepressant prescription and overall a preference for the new-generation antidepressants.

Conclusion

Socio-economic status did not seem to influence the selection of TCAs versus new-generation antidepressants. Compatible with the general epidemiology of depression, low socio-economic status was associated with a high number of first-time users of antidepressants in the population, and the incidence rate increased with age.

Keywords

Antidepressive agentsIncidenceSocio-economic status

Introduction

The consumption of antidepressants has been rapidly increasing since the introduction of new-generation drugs with milder adverse effects and a broader range of indications [1, 2]. Selection of a specific drug is an interplay between physician and patient, where many factors are taken into consideration. The drug price and the patient’s knowledge about drugs are likely to influence the final decision to prescribe. Patients’ requests for the new and more expensive preparations may be higher among the well-paid and higher educated patients [3]. Consequently, we set forth the hypothesis that the selection between a new- or an old-generation drug for antidepressant treatment is influenced by socio-economic status.

Low socio-economic status is generally associated with high psychiatric morbidity and disability [4]. Poorer coping styles, ongoing life events, stress exposure, and weaker social support are examples of psychiatric risk factors that are more frequent in lower socio-economic groups [5]. Use of antidepressants is therefore expected to be higher among these groups. To the best of our knowledge, an analysis of socio-economic status as predictor of antidepressant drug use has never previously been published.

Aims of the study

Based on registry data on a population covering 470,000 individuals, the aim of this study was to analyse whether first-time use of antidepressants (incidence) and selection of tricyclic antidepressants (TCAs) versus new-generation drugs are associated with socio-economic status and psychiatric history.

Materials and methods

We conducted a cohort study in the Danish County of Funen (~470,000 inhabitants) targeting first-time use of antidepressants. The cohort was defined among adults residing in the County on 1 January 1998. The study period was 1 year (1998). Eligibility criteria were residence in the County without prescriptions for antidepressants within 5 years prior to 1 January 1998. Data were obtained from established registers: the prescription database OPED (Odense University Pharmacoepidemiologic Database), Statistics Denmark and the Danish Psychiatric Central Register and linked by means of the Central Person Register number.

Data sources and variables

OPED comprises individual data on all prescriptions redeemed at all pharmacies in the County of Funen since mid 1992 [6]. For the period 1993–1998, we retrieved individual information on antidepressant prescriptions (Anatomical Therapeutical Chemical classification, ATC: N06A) including Central Person Register number, age, sex, date of dispense and drug type (ATC). We classified antidepressants into: (i) TCAs (ii) new-generation antidepressants: selective serotonin reuptake inhibitors (SSRIs—fluoxetine, citalopram, paroxetine, sertraline and fluvoxamine) and other new antidepressants (venlafaxine, reboxetine, nefazodone and mirtazapine) and (iii) tetracyclic antidepressants and monoamine oxidase (MAO) inhibitors. We retrieved a second file from OPED, including dates of deaths and migration to or from the county. No information on indication is recorded in OPED.

The integrated database for longitudinal labour market research (IDA by Statistics Denmark) contains yearly information from national administrative registers. We collected demographic and socio-economic data from IDA for 1997. We used three explanatory variables to indicate socio-economic status: education, socio-economic group, and annual income (Table 1). Educational level was described by different types of education, ranging from primary school level to university degree. These were re-coded into five categories. We defined socio-economic group in the form of nine levels describing main employment status; among employed, the ordinal ranking of occupations was based on the level of expertise it requires. Finally, we defined annual income in terms of quartile income groups. Subdivision of citizenship and family structure appears from Table 1. Citizenship was not considered a socioeconomic variable since individuals with foreign citizenship constitute a very heterogeneous group.
Table 1

Association between population characteristics and first-time prescribing of antidepressants (ADs) in Denmark, 1998 (Funen County). Data on incidence include 1-year incidence rates (number of first-time users per 100 person years) and incidence rate ratios (IRR). The proportion prescribed the new-generation antidepressants are shown together with odds ratios (ORs). IRRs and ORs are adjusted for age and sex. In brackets are 95% confidence intervals (CI). TCAs tricyclic antidepressants; new-generation antidepressants selective serotonin reuptake inhibitors (SSRIs) among others

 

Data on incidence

Data on selection of antidepressant

 

n (%)

1-Year incidence rate

Adj. IRR

P value

n

New-generation ADs (%)

Adj. OR

P value

  

New-generation ADs

TCAs

TCAs and new-generation ADs

     

Total

305,953 (100)

1.43 (1.38–1.47)

0.29 (0.27–0.31)

5131

83.1

Citizenship

  Danish

300,457 (98.2)

1.43 (1.39–1.48)

0.29 (0.27–0.31)

1.00

0.5

5049

83.3

1.0

0.0048

  Other

5496 (1.8)

1.10 (0.85–1.42)

0.43 (0.28–0.64)

1.08 (0.87–1.34)

 

82

72.0

0.49 (0.30–0.80)

 

Age (years)

  18–25

36,342 (11.9)

0.65 (0.57–0.74)

0.08 (0.06–0.12)

1.00*

<0.0001

255

89.0

1.0*

<0.0001

  26–40

81,180 (26.5)

1.11 (1.04–1.19)

0.15 (0.13–0.18)

1.72 (1.50–1.98)

 

1010

88.1

0.91 (0.59–1.41)

 

  41–60

109,851 (35.9)

1.29 (1.23–1.36)

0.32 (0.29–0.36)

2.20 (1.93–2.51)

 

1752

80.2

0.50 (0.33–0.75)

 

  61–70

36,455 (11.9)

1.49 (1.37–1.62)

0.43 (0.37–0.50)

2.59 (2.25–3.00)

 

685

77.7

0.43 (0.28–0.66)

 

  >70

42,125 (13.8)

3.02 (2.86–3.20)

0.54 (0.48–0.62)

4.70 (4.11–5.37)

 

1429

84.7

0.68 (0.45–1.03)

 

Sex

  Male

153,612 (50.2)

1.12 (1.07–1.18)

0.25 (0.22–0.27)

1.00†

<0.0001

2057

82.0

1.0†

0.2

  Female

152,341 (49.8)

1.73 (1.67–1.80)

0.33 (0.31–0.36)

1.41 (1.33–1.49)

 

3074

83.9

1.10 (0.95–1.28)

 

Education (years)

  <11

114,286 (37.4)

1.51 (1.44–1.58)

0.32 (0.28–0.35)

1.00

<0.0001

2048

82.7

1.0

0.3

  11–12

16,726 (5.5)

0.82 (0.69–0.97)

0.16 (0.11–0.24)

0.78 (0.66–0.92)

 

157

83.4

0.69 (0.44–1.09)

 

  13–14

111,675 (36.5)

1.10 (1.04–1.16)

0.25 (0.22–0.28)

0.82 (0.77–0.88)

 

1482

81.4

0.93 (0.78–1.11)

 

  >14

35,905 (11.7)

1.08 (0.97–1.19)

0.23 (0.19–0.29)

0.76 (0.68–0.84)

 

463

82.1

1.02 (0.78–1.33)

 

  Unknown or none

27,361 (8.9)

3.32 (3.10–3.55)

0.49 (0.41–0.58)

1.34 (1.20–1.50)

 

981

87.2

1.18 (0.86–1.63)

 

Socio-economic group

  Low-level employee

117,290 (38.3)

0.94 (0.89–1.00)

0.17 (0.15–0.19)

1.00

<0.0001

1286

84.8

1.0

0.4

  High-level employee

43,576 (14.2)

0.87 (0.78–0.96)

0.17 (0.14–0.22)

0.87 (0.78–0.97)

 

448

83.3

1.02 (0.76–1.37)

 

  Self-employed/senior manager

21,021 (6.9)

0.89 (0.77–1.02)

0.24 (0.18–0.31)

0.97 (0.84–1.11)

 

233

79.0

0.83 (0.58–1.19)

 

  Early retirement pensioner

14,458 (4.7)

1.35 (1.17–1.55)

0.39 (0.30–0.51)

1.14 (0.98–1.32)

 

248

77.4

0.89 (0.62–1.29)

 

  Disability pensioner

19,848 (6.5)

2.58 (2.37–2.82)

0.67 (0.57–0.80)

2.34 (2.11–2.59)

 

630

79.4

0.88 (0.68–1.14)

 

  Old-age pensioner

53,161 (17.4)

2.75 (2.61–2.90)

0.51 (0.45–0.58)

1.67 (1.46–1.91)

 

1660

84.3

0.99 (0.68–1.44)

 

  Student

11,673 (3.8)

0.82 (0.67–1.01)

0.13 (0.07–0.21)

1.03 (0.84–1.27)

 

106

86.8

0.59 (0.32–1.09)

 

  Unemployed

10,089 (3.3)

1.57 (1.34–1.83)

0.27 (0.19–0.40)

1.52 (1.30–1.78)

 

182

85.2

1.08 (0.70–1.68)

 

  Others out of labour force

14,835 (4.9)

1.93 (1.72–2.17)

0.42 (0.32–0.54)

1.98 (1.76–2.23)

 

338

82.3

0.74 (0.54–1.02)

 

Annual income

  Lower quartile

76,489 (25.0)

1.98 (1.88–2.09)

0.41 (0.36–0.46)

1.00

<0.0001

1758

82.9

1.0

0.6

  Second quartile

76,488 (25.0)

1.80 (1.70–1.90)

0.35 (0.31–0.39)

1.03 (0.96–1.11)

 

1604

83.8

1.13 (0.93–1.36)

 

  Third quartile

76,488 (25.0)

1.08 (1.00–1.15)

0.21 (0.18–0.25)

0.72 (0.66–0.78)

 

972

83.5

1.10 (0.87–1.38)

 

  Upper quartile

76,488 (25.0)

0.86 (0.80–0.93)

0.19 (0.16–0.23)

0.61 (0.56–0.67)

 

797

81.7

1.12 (0.88–1.43)

 

Family structure

  Co-habiting

201,260 (65.8)

1.22 (1.17–1.27)

0.29 (0.27–0.31)

1.00

<0.0001

2987

80.9

1.0

0.0032

  Single

104,693 (34.2)

1.83 (1.75–1.92)

0.29 (0.26–0.33)

1.33 (1.25–1.41)

 

2144

86.3

1.28 (1.08–1.50)

 

  Children at home

88,012 (28.8)

1.12 (1.05–1.19)

0.18 (0.16–0.22)

1.00

0.4

1133

85.8

1.0

0.1

  No children at home

217,941 (71.2)

1.55 (1.50–1.61)

0.33 (0.31–0.36)

0.97 (0.90–1.04)

 

3998

82.4

0.84 (0.68–1.04)

 

Admitted to psychiatric hospital 1994–1997

  No

301,268 (98.5)

1.34 (1.30–1.38)

0.28 (0.26–0.30)

1.00

<0.0001

4787

82.6

1.0

0.0009

  Yes

4685 (1.5)

7.07 (6.33–7.91)

0.75 (0.53–1.06)

5.66 (5.07–6.31)

 

344

90.4

1.88 (1.30–2.73)

 

Diagnosis at discharge if admitted

  No affective disorder

4406 (94.0)

6.90 (6.14–7.75)

0.53 (0.35–0.81)

1.00

0.0015

308

92.9

1.0

<0.0001

  Including affective disorder

279 (6.0)

9.97 (6.74–14.8)

4.39 (2.43–7.92)

1.76 (1.24–2.50)

 

36

69.4

0.15 (0.06–0.36)

 

  No anxiety disorder

4480 (95.6)

6.85 (6.10–7.69)

0.74 (0.52–1.05)

1.00

0.036

319

90.3

1.0

0.7

  Including anxiety disorder

205 (4.4)

12.0 (7.97–18.04)

1.04 (0.26–4.17)

1.56 (1.03–2.35)

 

25

92.0

1.33 (0.29–6.07)

 

  No substance abuse

2590 (55.3)

7.51 (6.50–8.69)

0.99 (0.66–1.48)

1.00

0.6

206

88.4

1.0

0.014

  Including substance abuse

2095 (44.7)

6.53 (5.50–7.77)

0.46 (0.24–0.88)

0.94 (0.74–1.20)

 

138

93.5

2.91 (1.24–6.81)

 

*Adjusted for sex

†Adjusted for age

In Denmark, all psychiatric hospital treatment is free of charge. From the Danish Psychiatric Central Register [7], we had information on admissions 1994–1997 and outpatient services 1995–1997 (the latter only available from 1995). Our data included day of admission and discharge, patient status (in- or outpatient) and diagnoses at discharge (according to ICD-10) [8]. We classified the diagnoses as affective disorders (F30–39), anxiety disorders (F40–42), substance abuse (F10–19) and others. We considered main and auxiliary diagnoses to be of equal importance. For each individual with a history of mental illness during 1994–1997, we dichotomised the diagnostic categories into present or not. Psychiatric history was limited to 4 years to avoid inclusion of diagnosis classified according to ICD-8, which differs fundamentally from ICD-10 introduced in Denmark 1 January 1994 [9].

Definition of sample and outcome

Adults without antidepressant prescriptions within 5 years prior to the study period were considered at risk of being prescribed an antidepressant for the first time during the study period, 1998. As drug information was restricted to the County, we excluded people immigrating during 1993–1997 from the cohort. Among 371,748 adults (>17 years) living in the County on 1 January 1998, we excluded 38,759 because of immigration and 27,036 because of antidepressant prescriptions prior to the study period. A study sample of 305,953 was thus identified. We identified first-time users of antidepressants by their first antidepressant prescription in the study period.

Analysis

Individuals were considered at risk of becoming first-time users of antidepressants until first antidepressant prescription, emigration, death or end of study period, whichever came first. We estimated incidence rates as the number of first-time users per 100 person years. We approached the association between social position, mental disorder and antidepressant prescription using Poisson regression analysis. First, we modelled incidence rate ratios using univariate analyses, and, second, we used multivariate analyses to control for confounding by age and sex. We repeated the multivariate analyses for individuals without a history of psychiatric hospital admission. We analysed the association between the selection of new-generation antidepressants for first-time prescriptions and patient characteristics using linear logistic regression adjusting for age and sex. We regarded P<0.05 as significant.

Results

We identified 5204 first-time users of antidepressants (1.7%). The initial drug was new-generation antidepressants for 4265 patients (82.0%), TCA for 866 (16.6%), tetracyclic and MAO inhibitors for 63 (1.2%), and two or more types of antidepressants for 10 (0.2%). Due to the low frequency, we did not analyse the two latter categories further. The incidence rate increased with age and was higher among females (Table 1). Prior psychiatric hospital admission was observed for 1.5% of the study population.

The results of the Poisson regression analysis concerning incident use of TCA and new-generation antidepressants are shown in Table 1. The incidence rate of antidepressants increased with age and was higher in people who were less educated, unemployed, receiving old-age or disability pension, in the lower income categories, who were single, or female. We found higher incidence rates in psychiatric patients. Compared with the general effect of mental disorder requiring hospital admission, the specific diagnostic category had a more limited effect. Separate analyses of each group of antidepressants and of people who had never been admitted to psychiatric hospitals showed associations similar to the above (data not shown).

The results from the logistic regression analysis concerning type of antidepressant prescribed among first-time users are presented in Table 1. The proportion prescribed new-generation antidepressants showed no difference according to socio-economic status, but young and single people were more likely to use the new-generation antidepressants. Furthermore, people without Danish citizenship were more likely to be prescribed tricyclics. Overall, patients with former psychiatric admission were more often prescribed new-generation antidepressants. Affective disorder as a specific diagnostic category lowered the preference for new-generation antidepressants, whereas substance abuse was associated with an increase.

Discussion

We undertook this study with the hypothesis that the prescribing of antidepressants is associated with socio-economic status regarding incidence and selection between new- and old-generation antidepressants. We found, however, no influence of socio-economic status on selection of antidepressants, but confirmed the association with incidence. Prior psychiatric hospital admission was a major risk factor for use of antidepressants.

All analyses concerned first-time prescribing in a large-scale community setting and were based on individual data from validated registers linked by the unique Danish personal registration number. Data were collected independently of the study, thus excluding information bias. All types of antidepressants are included in OPED, and less than 0.5% of the total quantity of drugs was purchased at pharmacies outside the County [6]. By definition, first-time users may have received antidepressants more than 5 years prior to the study, but misclassification due to missing prescription data was likely to be minimal. In contrast, most previous studies of incidence of antidepressants or depression have used a shorter period, which implies classification of prevalent cases as incident. An eligibility criterion of 5 years without symptoms eliciting antidepressant prescription strengthens this study, because it minimises the risk of the observed association between socio-economic status and use of antidepressants reflecting an effect of mental disorder on the socio-economic status, and not vice versa. A limitation of our study is the fact that no data were available about the symptoms and indications eliciting prescriptions; hence the study does not target social differences in the appropriateness of the prescribing of antidepressants.

The three socio-economic status variables included in the study do not only reflect economic, educational and occupational realities for each individual, but also various personality traits and individual circumstances at different stages of life [10]. The three variables are associated with each other and may be linked to mental health because they tend to define important differences in stress exposure and in the availability of coping resources [5]. No previous studies have analysed the association between socio-economic status and use of antidepressants. Well-informed patients may better formulate opinions and demands, and differences in costs and patients’ demands may imply that new-generation antidepressants are selected more often among higher social groups. However, we did not observe social inequality in the selection between new- and old-generation antidepressants. Therefore, the use of new-generation antidepressants does not seem to have proliferated extraordinarily among higher socio-economic groups, and the results did not suggest that non-medical factors have impact on the selection of antidepressants when a treatment is initiated. The latter point is consistent with a previous study of antidepressant prescribing among general practitioners, where we observed surprisingly little inter-practice variation beyond randomness in the relative use of new-generation antidepressants [11]. Furthermore, selection of antidepressants for specific patients has mostly been studied in relation to depression and in questionnaires based on clinical case vignettes [12, 13]. A relatively more frequent use of TCAs was seen in the age group 41–70 years and was particularly pronounced in patients with a history of psychiatric hospitalisation and a diagnosis of affective disorder. The latter finding may reflect the use of TCAs as first choice for hospitalised patients with severe or melancholic depression [14]. The distribution of diagnoses among first-time users of antidepressants in the community is poorly described. We did not have the opportunity either and, therefore, social differences in the prescribing of antidepressants for selected diagnosis cannot be excluded.

We confirmed the hypothesis that first-time use of antidepressants is higher among socio-economically underprivileged individuals. This is in line with previous knowledge of the distribution of psychiatric risk factors and depression in the population [15]. Among the various studies targeting the association between depression and socio-economic status, only few were actually designed to analyse incidence of depression [16, 17, 18, 19, 20]. Most results suggest that prevalence and incidence of depression are higher among populations with low socio-economic status [15]. Affective disorders including major depression constitute only part of the clinical scenario of antidepressant treatment [2]. Consequently, treatment rates do not correspond to rates of major depression, but represent to a large extent patients with depressive symptoms.

A further understanding of the association between social position and initiation of antidepressant treatment would require studies with data identifying those symptoms or diagnostic categories that in particular account for the association with socio-economic status.

Acknowledgements

The study was funded by The Danish Research Foundation for general practice, grant no. 585–148509. We thank Secretary Lise Stark for proofreading the manuscript.

Copyright information

© Springer-Verlag 2004