Family Practice Advance Access originally published online on November 1, 2004
Family Practice 2004 21(6):623-629; doi:10.1093/fampra/cmh608
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Family Practice Vol. 21, No. 6 © Oxford University Press 2004, all rights reserved.
Early discontinuation of antidepressants in general practice: association with patient and prescriber characteristics
a Research Unit of General Practice, b Department of Statistics and d Clinical Pharmacology, University of Southern Denmark and c Department of Geriatrics, Odense University Hospital, Odense, Denmark
E-mail: DGilsaa{at}health.sdu.dk
Received 8 July 2004; Accepted 20 July 2004.
Hansen DG, Vach W, Rosholm J-U, Søndergaard J, Gram LF and Kragstrup J. Early discontinuation of antidepressants in general practice: association with patient and prescriber characteristics. Family Practice 2004; 21: 625–631.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
Background. Most antidepressant treatment is initiated and continued in general practice but, despite current guidelines, treatment duration is often short among patients with depression. Discontinuation may, however, be caused by a complexity of factors, but so far research has focused on drug effects, adverse effects and drug regimens.
Objective. Our aim was to analyse whether early discontinuation of first-time antidepressant treatment in general practice may be predicted by (i) social position and psychiatric history of the patient; and (ii) demography, practice activity and the general prescribing behaviour of the GP.
Methods. Early discontinuation, i.e. that patients do not purchase antidepressants in the 6 months following first prescription, was analysed using established databases. Among patients presenting in 174 general practices in Funen County, Denmark, 4860 adult first-time users of antidepressants were identified (regardless of diagnosis). The inclusion period was January 1998–June 1999.
Results. One in three patients did not purchase antidepressants in the 6 months following first prescription, but rates were higher among those prescribed tricyclic compared with new generation antidepressants. Patients' age and sex did not have an influence, but early discontinuation was more frequent among patients of low socio-economic status and patients prescribed in practices characterized by high prescribing rates. No association with psychiatric history was observed.
Conclusion. Early discontinuation is frequent in general practice, and patients of low social status are at greater risk. Adherence-promoting strategies should pay attention to the high prescribing doctors. Further studies may answer the question of whether the association between doctors' prescribing behaviour and early discontinuation is a feature specific to antidepressants or a more general phenomenon.
Keywords. Adherence, antidepressive agents, family practice, physician's practice patterns, socio-economic status.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
Short duration of antidepressant treatment is very common among patients treated for depression.1–3 It is a major problem because long-term treatment may relieve symptoms and prevent relapse and recurrence of new episodes of depression.4,5 So far, research on adherence to antidepressants has focused on drug effects, adverse effects and dose regimens. Discontinuation may, however, be caused by a complexity of factors such as adverse effects, severity of illness, co-morbidity, personality traits, support from health providers, etc.6–8 Socio-economic status and level of education may influence the ways patients perceive their treatments and thereby affect adherence, and unemployment may, as another example, seriously affect the ability and motivation of an individual or family with regard to managing chronic illness and appropriate medical treatment.9 Consequently, we hypothesized that low socio-economic status is associated with higher rates of early discontinuation of antidepressant treatment. Adherence to treatment seems, furthermore, to be related to the quality of the clinical consultation,10 and the fact that some patients only purchase one prescription for a course of treatment may be associated with factors such as diagnostic skills and general prescribing attitude of the prescriber.
If interventions to increase adherence to treatment could be carefully targeted to specific patient and prescriber groups, early discontinuation from treatment may decrease11 and response rates increase.12 Most antidepressant treatments are initiated and continued in general practice. Therefore, in a naturalistic setting, the aim of this study was to analyse whether early discontinuation of first-time antidepressant treatment initiated in general practice may be predicted by (i) social position and psychiatric history of the patient; and (ii) demography, practice activity and the general prescribing behaviour of the GP.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
We used a Danish prescription database covering the County of Funen (470 000 inhabitants/186 general practices) to identify first-time users of antidepressants from 1 January 1998 to 30 June 1999. The source population included adults (at least 18 years), who resided in the county on 1 January 1998 and 5 years previously without prescriptions of antidepressants (n = 305 953). A period of 5 years without antidepressants was used to define an individual as a 'first-time user'. We identified 7047 first-time users of tricyclic or new generation antidepressants; among them 5694 (80.5%) received the prescription in general practice. Individuals who received the first prescription from the 174 practices under study (see below) constituted 5149, but 251 were excluded due to death and 25 because of emigration during follow-up (6 months following first prescription). A total of 4860 individuals who had been prescribed one type of antidepressants were included (13 were excluded because they had both types or prescriptions from two practices on the day of inclusion).
Data and variables included
Patient data were obtained from Odense University Pharmacoepidemiologic Database (OPED), the Danish Psychiatric Central Register and Statistics Denmark; and data on GPs were obtained from OPED and the National Health Insurance Register. Linkage was maintained via a unique person identifier (civil registration number) and a unique prescriber code (giving the prescribing practice but not the individual doctor).
Patient information
OPED contains individual data on all prescriptions redeemed at all pharmacies in the County of Funen since mid 1992.13 All antidepressant prescriptions were retrieved for 1993–1999. Data included person identifier, age, sex, date of dispensing, prescriber code, drug code (ATC: Anatomical Therapeutic Chemical classification), strengths and number of tablets. Two types of antidepressants were under study: new generation antidepressants: (i) selective serotonin re-uptake inhibitors (SSRIs; fluoxetine, citalopram, paroxetine, sertraline and fluvoxamine), and other new antidepressants (venlafaxine, reboxetine, nefazodone and mirtazapine); and (ii) tricyclic antidepressants (TCAs).
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: educational attainment, socio-economic group and annual income (see Table 1). Educational attainment was described by different types of education, ranging from primary school level up to university degree. These were recoded into five categories. We defined socio-economic group in the form of nine levels describing main employment status; among the employed, the ordinal ranking of occupations was based on the level of expertise required. Finally, we defined annual income in terms of quartile income groups among the source population. Subdivision of citizenship and family structure appear in Table 1.
|
From the Danish Psychiatric Central Register,14 we had information on psychiatric hospital admissions for the period 1994–1999 and out-patient activities in 1995–1999 (available from 1995). Our data included the date of admission and discharge, patient status (in- or out-patient) and diagnoses (according to ICD-1015). For each patient, information concerning psychiatric activities during 4 years prior to the first prescription was included. Main and auxiliary diagnoses were considered to be of equal importance and were classified as affective disorders (F30–39), anxiety disorders (F40–42), substance abuse (F10–19) and all others. Psychiatric history was limited to 4 years to avoid inclusion of diagnoses classified according to ICD-8, which differs fundamentally from ICD-10, introduced in Denmark on 1 January 1994.16 The psychiatric variables included are shown in Table 1. In Denmark, all psychiatric hospital treatment is free of charge.
Prescriber information
From the National Health Insurance Register we had demographic data about GPs, aggregated data on each practice's health care services, and identifiers on patients listed with each practice on 1 January 1999. Each practice's prescribing activities were covered by data from OPED. All activities covered the period June 1998–May 1999, and data on practice characteristics were considered invalid for 12 (6.5%) practices with change in physician staff in this period. Patients prescribed in those practices were therefore excluded from the study. Variables describing the GP, practice structure, workload, accessibility, use of counselling, and general and antidepressant prescribing were included (variables appear in Table 2). A counselling session comprises talking treatment of >15 min for which the GP receives a special fee. Consultations included home visits.
|
Background data on the 174 practices. The median number of consultations per GP per year was 5071 (10 and 90% percentiles: 3501–6800), the number of consultations per 100 patients per year was 339 (278–427), the number of counselling sessions per 100 consultations was 3.1 (0–9.5), and 1 year prescribing prevalence of antidepressants and all other drugs, respectively, were 3.80 (2.58–5.46) and 57.1 (50.4–63.6). Among the 110 single-handed practices, 65 (23.3%) of the GPs were female, median age 50 years (42–57) and the number of years in practice was 11.3 (3.0–22.8).
Analysis
Early discontinuation was defined as 'no purchase of any type of antidepressant during 6 months following first prescription'. We estimated rates of early discontinuation as a percentage with 95% confidence intervals (CIs). We analysed the association between early discontinuation and characteristics of patients and prescribers using logistic regression with adjustment for age, sex and antidepressant type. A potential association among patients in a practice was taken into account by using robust variance estimates (cluster option in Stata 7). We repeated the regression analyses for each type of antidepressant separately. To evaluate the variation among the practitioners, we computed 95% prediction intervals dependent on the number of patients included in the study.
For each practice, all prescribing prevalences were standardized according to age and sex by means of direct standardization, i.e. the prevalence in each of 22 age and sex strata was weighted with the fraction of this stratum in the population of patients listed within the county (n = 460 828). Continuous variables describing GPs were grouped into quintiles (replaced by scores from 1 to 5). We regarded P < 0.05 as significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
The initial drug was a new generation antidepressant for 4275 first-time users (88.0%) and a TCA for 585 (12.0%). The median age of patients was 52 years (10–90% percentiles 31–81), and 61.9% (3009) were female. Psychiatric hospital service was observed for 7% (340) during the 4 years prior to first prescription. Early discontinuation, i.e. no purchase of antidepressants during 6 months following first prescription, was observed for 33.6% (95% CI 32.3–34.9), but the rate was higher among patients initiated with TCAs [56.4% (52.3–60.5) versus 30.5% (29.1–31.9)] (Table 1). Patients' age and sex did not influence early discontinuation (Table 1). The median number of tablets prescribed on the day of first prescription was 28 (10–90% percentiles 20–30) for new generation drugs and 100 (30–100) for the TCAs. The median numbers of defined daily doses (DDD) were 28 (14–30) and 13 (6–33).
The results of the logistic regression analysis are presented in Table 2. Early discontinuation showed no difference according to age and sex, but was more frequent among people who were less educated, unemployed, receiving old age pension, in the lower income categories or of foreign citizenship. Patients prescribed their first antidepressant in practices characterized by a high number of consultations per year per 100 patients, and high prescribing rates (1 year prescribing prevalence of antidepressants and all other drugs, repectively) had a higher rate of early disconti-nuation. We observed no significant association with prescriber's age, sex, number of years in practice, practice type, workload or use of counselling (Table 2). We observed similar associations when analysing the new generation antidepressants separately. Early discontinuation was not significantly associated with psychiatric history.
For each practice, Figure 1 shows the proportion of patients who only purchased one prescription. Fourteen practices, i.e. more than the expected 4.95, fall outside the 95% prediction intervals, indicating that there are systematic differences between the practices.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
One in three first-time users of antidepressants did not purchase antidepressants in the 6 months following first prescription. Patients' age and sex did not influence early discontinuation, but patients of low socio-economic status and patients prescribed in practices characterized by high prescribing rates were at higher risk.
This study differs from many studies on antidepressant use: patients were included regardless of age, indication and co-morbidity, and the population under study was therefore very close to everyday clinical practice. Different definitions of treatment discontinuation correspond to different questions. We defined 'early discontinuation' in a simple but meaningful and transparent way. Our definition is independent of the reasons why treatment stopped, and the reasons may be intentional or non-intentional.7
A large number of patients were enrolled and all analyses were based on individual data from validated registers linked by the unique civil registration number and prescriber code. Both patients and doctors were unaware of being studied and information bias was thus avoided. We have no reason to believe that the first-time users excluded from the study show different associations between exposure variables and early discontinuation, and we therefore conclude that the selection criteria had minimal influence on the results. Classification error due to missing prescription data was likely to be minimal, because all types of antidepressants are included in OPED, and <0.5% of the total quantity of drugs was purchased at pharmacies outside the county.13 The practice variables included have been discussed in detail previously.17
An eligibility criterion of 5 years without antidepressant prescriptions is a strength of this study, because it minimizes the risk that discontinuation may be influenced by experience of previous treatment. It is a limitation that no data were available about symptoms and indications eliciting prescriptions. The area of indication is still widening beyond depression, and in the present study included dysthymia, obsessive– compulsive disorder, panic disorder, bulimia, certain types of pain, etc. Subgroup analyses of depressive patients would certainly have contributed to the interpretation of the results, and differences in discontinuation rates between subgroups may be due to differences in indications and dosage regimens. Research into a phenomenon as complex as adherence is inevitably fragmented,6 and whether differences led to differences in clinical outcome were beyond the aim of this study.
We observed that one in three first-time users of antidepressants did not purchase antidepressants in the 6 months following first prescription. Up to 6 weeks treatment is needed to reach maximum efficacy of antidepressants, and current guidelines recommend 4–9 months of treatment after resolution of the acute symptoms. We do not know if antidepressant discontinuation was in accordance or in conflict with recommendations given by the GP. Adverse effects or relief of symptoms may be the cause, but diagnostic uncertainty and a trial and error strategy used by some doctors may be part of the explanation. Different treatment strategies and a broader use of antidepressants may be hypothesized to be more prevalent among the high prescribing GPs.
Compared with the new generation antidepressants, a higher discontinuation rate among TCA users (56.4 versus 33.6%) was observed. Due to the non-randomization by initial drug, it is unclear whether this difference was due to the drugs or to other factors that influence both drug selection and discontinuation. Meta-analyses of short-term randomized clinical trials have concluded that a significant but small difference in total drop-out favours the SSRIs.2,18–20 This may reflect differences in acute tolerability. However, a Swedish study based on prescriptions in the community observed no difference according to initial drug.21
The three socio-economic status variables included in the study reflect not only economic, educational and occupational realities for each individual, but also various personality traits and individual circumstances at different stages of life.22 The three variables are associated with each other, and tend to define important differences in stress exposure and in the availability of coping resources.23 This study confirmed our hypothesis that early discontinuation of antidepressants is higher among people of low socio-economic status. Until now, social patient characteristics and practice characteristics have been missing topics in the research of adherence to antidepressant treatment. We did not observe any influence of age and sex of the patient, which is in accordance with studies of patients with depression.3,24 The Swedish study mentioned above did, however, observe higher discontinuation rates among men.21 A previous study in the County of Funen concluded that duration of treatment courses increased with age, but this study only analysed patients redeeming at least two prescriptions.25
In general practice, the prevalence of antidepressant prescribing is positively associated with the general prescribing prevalence.17 In the present study, the patients initiated in high prescribing practices had higher rates of early discontinuation. We do not know if selection of cases in the high prescribing practices is more or less appropriate than in other practices, but it seems reasonable to pay some attention to high prescribing doctors when attempting to reduce early discontinuation of antidepressants. Low adherence to prescribed medical interventions is an ever-present and complex problem. Prescribing habits are hard to change26,27 and patients make decisions about treatments that fit into their own beliefs and personal circumstances.7,28 Adherence-promoting strategies should be adapted to the stage of treatment24 and may only improve adherence if their target concerns one of the patient's worries.10 A range of strategies in the individual practice is therefore needed. We recommend that GPs are aware of the high risk of early discontinuation every time antidepressants are considered. Age and sex have no influence, but patients of low social status are at greater risk. Further studies on prescribing habits among GPs may explore the association observed between high prescribing activity and high frequency of early discontinuation of antidepressants and, furthermore, answer the question of whether this association is a feature specific to antidepressants or a general phenomenon. Qualitative methods may be used to explore the mechanisms responsible for this association.
|
Declaration |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
Funding: The study was funded by The Danish Research Foundation for General Practice, grant no. 585-148509.
Ethical approval: not applicable.
Conflicts of interest: none.
|
Acknowledgments |
|---|
We thank Secretary Lise Stark for proofreading the manuscript.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionDeclarationReferences |
|---|
1 McDonald TM. Treatment of depression: prescription for success? Primary Care Psychiatry 1997; 3: S7–S10.
2 Montgomery SA, Henry J, McDonald G et al. Selective serotonin reuptake inhibitors: meta-analysis of discontinuation rates. Int Clin Psychopharmacol 1994; 9: 47–53.[ISI][Medline]
3 Dunn RL, Donoghue JM, Ozminkowski RJ, Stephenson D, Hylan TR. Longitudinal patterns of antidepressant prescribing in primary care in the UK: comparison with treatment guidelines. J Psychopharmacol 1999; 13: 136–143.[Abstract]
4 Paykel ES, Priest RG. Recognition and management of depression in general practice: consensus statement. Br Med J 1992; 305: 1198–1202.[ISI][Medline]
5 Scott J, Eccleston D. Prediction, treatment and prognosis of chronic primary major depression. Int Clin Psychopharmacol 1991; 6 Suppl 1: 41–49.
6 Demyttenaere K. Compliance during treatment with antidepressants. J Affect Disord 1997; 43: 27–39.[CrossRef][ISI][Medline]
7 Vermeire E, Hearnshaw H, Van Royen P, Denekens J. Patient adherence to treatment: three decades of research. A comprehensive review. J Clin Pharm Ther 2001; 26: 331–342.[CrossRef][ISI][Medline]
8 Frank E, Judge R. Treatment recommendations versus treatment realities: recognizing the rift and understanding the consequences. J Clin Psychiatry 2001; 62 Suppl 22: 10–15.
9 Clark N, Jones P, Keller S, Vermeire P. Patient factors and compliance with asthma therapy. Respir Med 1999; 93: 856–862.[CrossRef][ISI][Medline]
10 Donovan JL. Patient decision making. The missing ingredient in compliance research. Int J Technol Assess Health Care 1995; 11: 443–455.[ISI][Medline]
11 Claxton AJ, Li Z, McKendrick J. Selective serotonin reuptake inhibitor treatment in the UK: risk of relapse or recurrence of depression. Br J Psychiatry 2000; 177: 163–168.
12 Maj M, Veltro F, Pirozzi R, Lobrace S, Magliano L. Pattern of recurrence of illness after recovery from an episode of major depression: a prospective study. Am J Psychiatry 1992; 149: 795–800.
13 Gaist D, Sorensen HT, Hallas J. The Danish prescription registries. Dan Med Bull 1997; 44: 445–448.[ISI][Medline]
14 Munk-Jorgensen P, Mortensen PB. The Danish Psychiatric Central Register. Dan Med Bull 1997; 44: 82–84.[ISI][Medline]
15 WHO. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva: WHO; 1992.
16 Kessing LV. A comparison of ICD-8 and ICD-10 diagnoses of affective disorder—a case register study from Denmark. Eur Psychiatry 1998; 13: 342–345.[CrossRef]
17 Hansen DG, Søndergaard J, Vach W, Gram LF, Rosholm JU, Kragstrup J. Antidepressant drug use in general practice: inter-practice variation and association with practice characteristics. Eur J Clin Pharmacol 2003; 59: 143–149.[ISI][Medline]
18 Anderson IM, Tomenson BM. Treatment discontinuation with selective serotonin reuptake inhibitors compared with tricyclic antidepressants: a meta-analysis. Br Med J 1995; 310: 1433–1438.
19 Montgomery SA, Kasper S. Comparison of compliance between serotonin reuptake inhibitors and tricyclic antidepressants: a meta-analysis. Int Clin Psychopharmacol 1995; 9 Suppl 4: 33–40.
20 Barbui C, Hotopf M, Freemantle N et al. Selective serotonin reuptake inhibitors versus tricyclic and heterocyclic antidepressants: comparison of drug adherence. Cochrane Database Syst Rev 2000; CD002791.
21 Isacsson G, Boethius G, Henriksson S, Jones JK, Bergman U. Selective serotonin reuptake inhibitors have broadened the utilisation of antidepressant treatment in accordance with recommendations. Findings from a Swedish prescription database. J AffectDisord 1999; 53: 15–22.
22 Jeynes WH. The challenge of controlling for SES in social science and education research. Educ Psychol Rev 2002; 14: 205–221.[CrossRef]
23 Turner RJ, Lloyd DA. The stress process and the social distribution of depression. J Health Soc Behav 1999; 40: 374–404.[CrossRef][ISI][Medline]
24 Demyttenaere K, Enzlin P, Dewe W et al. Compliance with antidepressants in a primary care setting, 1: beyond lack of efficacy and adverse events. J Clin Psychiatry 2001; 62 Suppl 22: 30–33.[ISI][Medline]
25 Rosholm JU, Andersen M, Gram LF. Are there differences in the use of selective serotonin reuptake inhibitors and tricyclic antidepressants? A prescription database study. Eur J Clin Pharmacol 2001; 56: 923–929.[CrossRef][ISI][Medline]
26 Oxman AD, Thomson MA, Davis DA, Haynes RB. No magic bullets: a systematic review of 102 trials of interventions to improve professional practice. Can Med Assoc J 1995; 153: 1423–1431.[Abstract]
27 Sondergaard J, Andersen M, Kragstrup J, Hansen P, Freng GL. Why has postal prescriber feedback no substantial impact on general practitioners' prescribing practice? A qualitative study. Eur J Clin Pharmacol 2002; 58: 133–136.[Medline]
28 Morris LS, Schulz RM. Medication compliance: the patient's perspective. Clin Ther 1993; 15: 593–606.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A Erlangsen, V Canudas-Romo, and Y Conwell Increased use of antidepressants and decreasing suicide rates: a population-based study using Danish register data J. Epidemiol. Community Health, May 1, 2008; 62(5): 448 - 454. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. J. Lustman, R. E. Clouse, B. D. Nix, K. E. Freedland, E. H. Rubin, J. B. McGill, M. M. Williams, A. J. Gelenberg, P. S. Ciechanowski, and I. B. Hirsch Sertraline for Prevention of Depression Recurrence in Diabetes Mellitus: A Randomized, Double-blind, Placebo-Controlled Trial. Arch Gen Psychiatry, May 1, 2006; 63(5): 521 - 529. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||