The British Journal of Psychiatry (2006) 188: 337-345. doi: 10.1192/bjp.188.4.337
© 2006 The Royal College of Psychiatrists
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit an eLetter
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in BJP
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Right arrow Citing Articles via Scopus
Google Scholar
Right arrow Articles by KENDRICK, T.
Right arrow Articles by THOMPSON, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by KENDRICK, T.
Right arrow Articles by THOMPSON, C.

Cost-effectiveness and cost-utility of tricyclic antidepressants, selective serotonin reuptake inhibitors and lofepramine

Randomised controlled trial

TONY KENDRICK, MD, FRCGP, FRCPsych

Primary Medical Care Group

ROBERT PEVELER, MA, DPhil, FRCPsych

Mental Health Group, Community Clinical Sciences Division, University of Southampton School of Medicine, Southampton

LOUISE LONGWORTH, BA, MSc

Health Economics Research Group, Brunel University, London

DAVID BALDWIN, DM, FRCPsych

Mental Health Group, Clinical Neurosciences Division, University of Southampton School of Medicine

MICHAEL MOORE, BMed Sci, Msc, MRCP, MRCGP

Three Swans Surgery, Salisbury

JUDY CHATWIN, BSc, MSc

Primary Medical Care Group, Community Clinical Sciences Division, University of Southampton School of Medicine

ANDREW THORNETT, MRCGP, FRACGP, FACPsychMed

Faculty of Health and Sciences, Staffordshire University, Stoke on Trent

JONATHAN GODDARD, MSc

Health Care Research Unit, Community Clinical Sciences Division, University of Southampton School of Medicine

MICHAEL CAMPBELL, BSc, MSc, PhD, CStat

School for Health and Related Research, University of Sheffield

HELEN SMITH, MSc, MD, MRCGP, FFPHM

Brighton and Sussex Medical School, Brighton

MARTIN BUXTON, BA

Health Economics Research Group, Brunel University, London

CHRISTOPHER THOMPSON, MD, FRCP, FRCPsych, MRCGP

Mental Health Group, Community Clinical Sciences Division, University of Southampton School of Medicine, Southampton, UK

Correspondence: Professor Tony Kendrick, Primary Medical Care Group, Community Clinical Sciences Division, University of Southampton Medical School, Aldermoor Health Centre, Southampton SO16 5ST, UK.Tel: +44 (0)23 8024 1050; fax: +44 (0)23 8070 1125; e-mail: ark1{at}soton.ac.uk

Declaration of interest T.K., R.P., D.B. and C.T. have received funds from major drug companies which are unrelated to the funding of this study.Funding detailed in Acknowledgements.


arrow
ABSTRACT
 
Background The cost-effectiveness of tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) has not been compared in a prospective study in primary care.

Aims To determine the relative cost-effectiveness of TCAs, SSRIs and lofepramine in UKprimary care.

Method An open-label, three-arm randomised trial with a preference arm. Practitioners referred 327 patients with incident depression.

Results No significant differences were found in effectiveness or cost-effectiveness. The numbers of depression-free weeks over12 months (on the Hospital Anxiety and Depression Scale) were 25.3 (95% CI 21.3–29.0) for TCAs, 28.3 (95% CI 24.3–32.2) for SSRIs and 24.6 (95% CI 20.6–28.9) for lofepramine. Mean health service costs per patient were £762 (95% CI 553–1059) for TCAs, £875 (95% CI 675–1355) for SSRIs and £867 (95% CI 634–1521) for lofepramine.Cost-effectiveness acceptability curves suggested SSRIs were mostcost-effective (with a probability of up to 0.6).

Conclusions The findings support a policy of recommending SSRIs as first-choice antidepressants in primary care.


arrow
INTRODUCTION
 
Economic modelling studies comparing selective serotonin reuptake inhibitors (SSRIs) and tricyclic antidepressants (TCAs) have usually found in favour of SSRIs (Hatziandreu et al, 1994; Stewart, 1994; Einarson et al, 1995; Canadian Coordinating Office for Health Technology Assessment, 1997; Doyle et al, 2001) but often rely on estimates of resource costs. Varying assumptions used greatly affects the conclusions (Jonsson & Bebbington, 1994; Woods & Rizzo, 1997). Three studies collected primary resource use and outcome data. A naturalistic study in the USA by Simon et al (1996) found that higher costs for fluoxetine were balanced by fewer hospital costs compared with TCA treatment, but analysed cost-consequences, not cost-effectiveness. A Czech study (Hosak et al, 2000) found similar results, but outcome was limited to days free from hospitalisation. Forder et al (1996) estimated cost-effectiveness in UK primary care, suggesting that sertraline was more cost-effective than TCAs, but used a quasi-experimental design and estimated outcomes retrospectively. There was therefore a clear need for a prospective cost-effectiveness comparison of these antidepressants in UK primary care.


arrow
METHOD
 
We aimed to compare the cost-effectiveness (cost per depression-free week) and cost-utility (cost per quality-adjusted life-year) of TCAs, SSRIs and lofepramine as recommended first-choice antidepressants in primary care, adopting a health service perspective. We included lofepramine as a third option frequently prescribed as a tolerable, safer alternative to the older TCAs. A pragmatic, open-label, three-arm randomised trial was designed, aimed at recruiting a representative sample of patients presenting to primary care with a new episode of depression.

Ethics committee approval
Ethics committee approval was granted by the South West Multicentre Ethics Committee and subsequently by local research ethics committees covering Hampshire, East Dorset, Wiltshire, West Sussex and South West Surrey.

Inclusion and exclusion criteria
All adults diagnosed with depression by their general practitioner and accepting antidepressant treatment were eligible, including those with comorbid physical or mental illness and those aged over 65 years. Those already taking antidepressants, under 18 years old, pregnant, breast-feeding, terminally ill, confused, with insufficient English language skills or temporarily resident were excluded. Patients who were prescribed antidepressants for indications other than depression (e.g. chronic pain) were excluded.

Randomisation
The researchers visited patients within a few days of referral, usually at the patient’s home, gave information about the study, and sought written consent. Consenting patients were randomly allocated to a recommended first-choice class of antidepressant, either a TCA (choice of amitriptyline, dosulepin or imipramine), or an SSRI (fluoxetine, sertraline or paroxetine), or lofepramine. Remote telephone randomisation was carried out by the University of York service. Randomisation was stratified by referring general practitioner, on the basis that one doctor’s referrals might differ systematically in severity, or in other ways, from another’s.

Partial preference design
A partial preference design was used to minimise the effect of treatment choice on recruitment, and allow assessment of the effect of receipt of preferred treatment on outcome. Patients were informed at recruitment that, following randomisation, treatment could be prescribed from a different class to the one allocated at random, if they or their doctor preferred an alternative, in a preference group. The main analysis was conducted on an intention-to-treat basis. Sensitivity analyses were carried out including only those who actually received an initial prescription from the randomised class.

Treatments
All treatments were prescribed by the general practitioners in order to keep care close to normal, and could be changed as clinically indicated, including switching class of antidepressant. Doctors were asked to prescribe using British National Formulary recommended dosages (see Appendix; British Medical Association & Royal Pharmaceutical Society of Great Britain, 2001) and to continue treatment for 6 months after remission of the depressive episode, or for at least 12 months if the patient had experienced two or more depressive episodes within the past 5 years.

Outcome measures
All outcome measures were self-completed, to avoid interviewer bias, as it was impossible to mask the researchers to group allocation. Follow-up was initially planned using 12 postal questionnaires at monthly intervals, but 13 months into recruitment the monthly questionnaires at 2, 4, 5, 7, 8 and 11 months were dropped because the response rates were low, and outcome data were collected instead through face-to-face contacts at 6 months, 9 months and 12 months, which improved response rates.

The primary outcome was the number of weeks free from depression, defined as a score of below 8 on the Hospital Anxiety and Depression Scale – Depression subscale (HADS–D; Zigmond & Snaith, 1983), the threshold for ‘possible major depressive disorder’. Linear interpolation of missing values was used, but there was no extrapolation beyond the last observation. An alternative definition of ‘depression-free’ was employed in a sensitivity analysis using a cut-off HADS–D score of below 11 (threshold for ‘probable major depressive disorder’). A wider range of psychiatric symptoms was assessed using the computerised (PROQSY) version of the Clinical Interview Schedule – Revised (CIS–R; Lewis et al, 1992). Psychiatric diagnoses were generated at baseline by means of the CIS–R algorithm used in the Office of Population Censuses and Surveys national psychiatric morbidity survey (Meltzer et al, 1994). Generic health status was measured using the 36-item Medical Outcomes Study Short Form (SF–36; Ware et al, 1993). For the estimation of quality-adjusted life-years (QALYs; see below), utility was measured using the EuroQol EQ–5D questionnaire (EuroQol Group, 1990). Patients’ use of health and social services between follow-up points was recorded on a schedule specifically designed for the study.

Cost data
Patients’ practice records were examined at the end of the 12-month follow-up period to estimate total health and social service resource use, including all medications prescribed and their duration. Additional use of resources identified from patients’ self-reports was added to the medical record data.

Analysis of clinical outcomes
Repeated-measures analysis of variance (ANOVA) using generalised linear modelling was used to test the significance of differences between groups in the number of depression-free weeks, adjusting for baseline HADS–D score. This assumed that missing data were ‘missing at random’, and used all the available data to estimate the number of depression-free weeks expected if all patients with post-baseline HADS–D scores completed all 12 months of follow-up. Other comparisons between the three groups used the Kruskal–Wallis test.

Sample size calculation
We aimed to power the study to be able to demonstrate equivalence of total costs, based on data from the study by Simon et al (1996); if equivalence of costs were found, then the comparison would simply be of clinical outcomes. To demonstrate equivalence within 5% of the expected mean log cost of 7.16 (s.d.1.0, ß=0.1, {alpha}=0.05/3), we required 260 evaluable patients per group at follow-up.

Economic analysis
The economic analysis was conducted from a health service perspective. Unit costs in pounds sterling (£) were obtained from published sources for 2001/2002, and inflated or deflated as appropriate where figures were not available (Ratcliffe et al, 1996; Brown et al, 1997; Moffett et al, 1999; British Medical Association & Royal Pharmaceutical Society of Great Britain, 2001; Netten et al, 2001). No discounting was necessary as costs and effects related to no more than 1 year. Estimated total service use costs per patient over 12 months were calculated from the medical record data, augmented by patients’ self-reports. Cost-effectiveness and cost-utility analyses included resource data only up to the time of patients’ last HADS–D or EQ–5D ratings. The QALY is a measure of a patient’s life expectancy, weighted by his or her health-related quality of life, valued on a self-reported ‘utility’ scale, where 0 represents death and 1 represents full health. The QALYs were calculated by applying a tariff of health state values, based on a representative UK sample, to the utility scores from the EQ–5D (Dolan, 1997). A multivariate generalised linear model was used to adjust for differences in baseline EQ–5D scores.

In the absence of demonstrable cost equivalence, incremental cost-effectiveness and cost-utility ratios were computed comparing TCAs with SSRIs, SSRIs with lofepramine and TCAs with lofepramine. To characterise the uncertainty around the ratios, estimates were bootstrapped with 5000 replications and presented on cost-effectiveness planes and as cost-effectiveness acceptability curves (Manly, 1997). These show the probability of each treatment strategy being cost-effective, contingent upon the value placed upon an additional depression-free week, or an additional QALY.


arrow
RESULTS
 
Recruitment
Referral rates to the study were lower than anticipated, and recruitment of practices had to be extended in five waves, corresponding to the five ethics committee areas. Between October 1999 and April 2002, a total of 87 general practitioners from 55 practices referred 388 patients, of whom 327 were randomly allocated to the three recommended classes (Fig. 1).


Figure 1
View larger version (37K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 1 Flow of participants through the trial (LOF, lofepramine; SSRI, selective serotonin reuptake inhibitor; TCA, tricyclic antidepressant).

Allocation
Of 327 patients randomised, 92 patients were prescribed a different class of antidepressant to that of the allocated treatment. Doctor preference was the stated cause for this in 48 cases and patient preference in 24 cases (no cause stated in 20). The proportions prescribed a different class differed significantly between allocated classes: TCAs, 47/113 (42%); lofepramine, 28/105 (27%); SSRIs, 17/109 (16%); {chi}2=18.7, d.f.=2, P<0.001. Following initial prescription from the allocated drug class, 81 patients later switched class. The proportions switching also differed significantly: lofepramine, 36/77 (46%); TCAs, 23/66 (35%); SSRIs 22/92 (24%); {chi}2=9.71, d.f.=2, P=0.008. Median daily dosages prescribed were amitriptyline 50 mg, dosulepin 75 mg, imipramine 100 mg, fluoxetine 20 mg, paroxetine 20 mg, sertraline 50 mg and lofepramine 140 mg.

Follow-up rates
Assessments of outcome were completed with 254 patients at 3 months (78%), 203 (62%) at 6 months, 188 (58%) at 9 months and 171 (52%) at 12 months, with no significant difference in completeness between groups (Fig. 1). Two patients died, but their deaths were not related to depression or its treatment. One patient was removed from the study for personal reasons, not related to the protocol.

Participants recruited
The demographic characteristics of the 327 randomised patients are given in Table 1, and show that almost all were White, and there were more female than male patients, as expected. Table 2 shows the ICD–10 diagnoses (World Health Organization, 1992) generated by the CIS–R. Of the 327 patients randomised, 239 (73%) received a primary diagnosis of a depressive disorder, 40 (12%) a primary diagnosis of an anxiety disorder and 48 (15%) no identifiable psychiatric diagnosis.


View this table:
[in this window]
[in a new window]

 
Table 1 Characteristics of patients randomised

View this table:
[in this window]
[in a new window]

 
Table 2 Diagnoses at baseline derived from the Clinical Interview Schedule – Revised

Clinical outcomes
Table 3 shows the values obtained for the clinical outcomes over 12 months for the three groups, on an intention-to-treat basis. All three groups improved to a similar extent, with most improvement occurring in the first 3 months. No significant difference between groups was demonstrated. The sensitivity analysis, including only those who received an initial prescription from the class to which they were randomised, also demonstrated no significant difference.


View this table:
[in this window]
[in a new window]

 
Table 3 Outcome measures at baseline and over 12 months of follow-up (all patients randomised, regardless of whether they received a drug of the allocated class)

Practice records of resource use and at least one post-baseline HADS–D score were obtained for 264 patients. The mean numbers of depression-free weeks for these patients were not significantly different between groups: 25.3 weeks (95% CI 21.3–29.0) for the TCA group, 28.3 weeks (95% CI 24.3–32.2) for the SSRI group and 24.6 weeks (95% CI 20.6–28.9) for the lofepramine group (Kruskal–Wallis test: {chi}2=2.23, P=0.327). The repeated-measures ANOVA gave estimated mean numbers of depression-free weeks over the full 12 months of 36.6 weeks for SSRIs, 35.5 weeks for TCAs and 34.8 weeks for lofepramine. These differences were not statistically significant: SSRIs {nu}. TCAs, 1.1 depression-free weeks (95% CI –4.0 to 6.3); TCAs {nu}. lofepramine, 0.7 week (95% CI –4.6 to 5.9); and SSRIs {nu}. lofepramine, 1.8 weeks (95% CI –3.5 to 7.1). No significant difference was demonstrated through sensitivity analyses, either including only those who received the randomised class, or when using the higher HADS–D cut-off score of 11.

Resource use
There was no statistically significant difference in non-drug resource use over 12 months (Table 4). The most frequent contacts were with general practitioners at the surgery, followed by contacts with practice nurses. Contacts with community psychiatric nurses and psychiatrists in out-patient clinics were much less frequent. Day centre attendances and in-patient stays were relatively uncommon.


View this table:
[in this window]
[in a new window]

 
Table 4 Non-drug resource use in the 12 months following randomisation

Costs
Table 5 summarises cost data over 12 months. Costs were skewed, and so mean and median values are presented. Mean non-drug service use costs were around £650 per patient, with no statistically significant difference demonstrated between groups. Statistically significant differences in costs were found for all drug prescriptions, and for the costs of antidepressant prescriptions alone, but not for mean total costs. However, antidepressant prescriptions accounted for less than a tenth of total costs in each group.


View this table:
[in this window]
[in a new window]

 
Table 5 Summary of total service use costs to 12 months from randomisation

Cost-effectiveness
Mean total service use costs up to the last HADS–D assessment for these 264 patients were £712 for the TCA group (95% CI 486–1062), £809 (95% CI 590–1431) for the SSRI group and £593 (95% CI 459–772) for the lofepramine group. Although differences were not statistically significant, equivalence was not demonstrated (Kruskal–Wallis {chi}2=3.76, P=0.153). Incremental costs per depression-free week were £32 more for SSRIs {nu}. TCAs, £59 more for SSRIs {nu}. lofepramine and £183 more for TCAs {nu}. lofepramine. It was not possible to calculate reliable confidence intervals around these mean ratios as the small differences led to unstable results when bootstrapped. Therefore, cost-effectiveness planes and cost-effectiveness acceptability curves were computed to illustrate the uncertainty around these estimates. The cost-effectiveness planes for each comparison included points in all four quadrants (Figs 2, 3, 4) reflecting statistically non-significant differences in outcomes and costs.


Figure 2
View larger version (21K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 2 Cost-effectiveness plane for selective serotonin reuptake inhibitors compared with lofepramine.


Figure 3
View larger version (19K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 3 Cost-effectiveness plane for tricyclic antidepressants compared with lofepramine


Figure 4
View larger version (24K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 4 Cost-effectiveness plane for selective serotonin reuptake inhibitors compared with tricyclic antidepressants.

Cost-utility
Practice and patient records of resource use and at least one post-baseline EQ–5D score were obtained for 261 patients. Among these, the mean numbers of QALYs, adjusted for baseline EQ–5D, were 0.55 (95% CI 0.48–0.61) for the TCA group, 0.59 (95% CI 0.52–0.64) for the SSRI group and 0.55 (95% CI 0.49–0.61) for the lofepramine group. Mean total service use costs for these 261 patients were £712 for the TCA group (95% CI 502–1103), £817 (95% CI 586–1486) for the SSRI group and £619 (95% CI 469–788) for the lofepramine group. Incremental costs per QALY between groups were £5686 more for SSRIs {nu}. lofepramine, £23 250 less for TCAs {nu}. lofepramine and £2692 more for SSRIs {nu}. TCAs. Cost-utility planes for each comparison again included points in all four quadrants, reflecting non-significant differences.

The cost-effectiveness acceptability curve (Fig. 5) shows (Fig. 5) shows that if an additional depression-free week were valued at less than £20, lofepramine would be likely to be most cost-effective and SSRIs least cost-effective. However, if it were valued at above £50, SSRIs would be likely to be most cost-effective and TCAs least cost-effective. However, differences between them were small. A similar curve was computed for the cost-utility results (Fig. 6). This shows that, for values placed on an additional QALY of more than £5000, the cost-utility of SSRIs was likely to be greatest, with little difference between the other two groups, although the probability of this did not rise above 0.6.


Figure 5
View larger version (17K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 5 Cost-effectiveness acceptability curve.


Figure 6
View larger version (21K):
[in this window]
[in a new window]
[as a PowerPoint slide]
 
Fig. 6 Cost-effectiveness acceptability curve. QALY, quality-adjusted life-year.


arrow
DISCUSSION
 
Main findings
Clinical outcomes
We found no significant difference in effectiveness between the three classes of antidepressant on an intention-to-treat basis, although patients allocated to treatment with TCAs were significantly more likely to receive a different class of antidepressant, usually a result of doctor rather than patient preference, and those allocated to receive lofepramine were significantly more likely to switch drug class later.

Comparison of costs
The costs of the antidepressants prescribed represented less than a tenth of total health service costs over 12 months. The initial prescribing costs of offering a TCA were lowest, but no significant difference was found in overall service costs between classes. Considerable between-patient variability and skewed cost distributions produced wide confidence intervals. No significant difference was demonstrated in sensitivity analyses including only those who received an initial prescription from the randomised class, nor when using the higher HADS–D cut-off score of 11.

Cost-effectiveness
The sample we recruited was not large enough to demonstrate equivalence, and so it was appropriate to estimate incremental cost-effectiveness and cost-utility. These were broadly similar for the three comparisons, with bootstrapped simulation estimates of cost per depression-free week or cost per QALY, plotted on cost-effectiveness planes, clearly occupying all four quadrants in each paired comparison, but the cost-effectiveness acceptability curves suggest that, for values of an additional depression-free week over £50 or for a QALY over £5000, SSRIs were most likely – and TCAs least likely – to be the most cost-effective.

Comparison with previous studies
The findings are consistent with those of Simon et al (1996) who found that over 40% of patients initially assigned to TCAs switched treatment, compared with 20% for fluoxetine. They also found no difference in effectiveness between fluoxetine and TCAs, that antidepressant costs were less than 10% of total service costs, and that the initial higher cost of fluoxetine was offset by lower out-patient and in-patient care costs, so that overall service costs were not significantly different (Simon et al, 1996). Compared with the findings of Forder et al (1996), our study suggests a more modest benefit for the SSRIs. However, the conclusions of the former study may be less reliable than those reported here owing to its non-experimental design, with masked, retrospective assessments of outcome.

Strengths of the study
The main strength of our study was its naturalistic general practice setting, which is likely to reflect usual clinical practice. Fewer than 30% of recruited patients had moderate or severe depressive episodes, which accords with other evidence that antidepressants are often prescribed for mild depression (Kendrick et al, 2001), despite a lack of evidence that they are more effective than placebo in such cases (Drug and Therapeutics Bulletin, 2003). The lower-than-recommended median dosages of TCAs prescribed also reflect usual practice. They reflect lower dosages given during the titration period, lower dosages which apparently worked and were continued, and lower dosages beyond which the patient could not tolerate the drugs and discontinued them. It is important to remember that we investigated the effect of recommending a treatment choice, as in a clinical guideline, rather than actual treatment delivery. The partial preference design permitted prescription of a different class of antidepressant, as a result of doctor or patient preference, and so we have data on patients who would not have been included if the trial had been limited to those accepting randomisation. Another strength of the study is that cost data were available for almost all patients over a 12-month period, from computerised practice records.

Limitations of the study
Sample size
The main limitation was the failure to recruit the desired sample because we received many fewer referrals per practitioner than anticipated. This suggests the patients referred might not be representative of all new patients with depression starting antidepressant therapy in general practice. We did ask the practitioners to record the number of eligible patients asked to take part in the study who declined, but this proved unsuccessful. Loss to follow-up over the 12 months further limited the power to detect differences in effectiveness between antidepressant classes. A shorter follow-up period might have reduced attrition, but risked missing important differences emerging after the initial treatment period. With the benefit of hindsight, attempting to demonstrate cost equivalence to within 5% was a rather strict criterion and led to a much larger target number of evaluable patients than we were able to recruit and follow-up. The greater uncertainty around our estimates, owing to the smaller than anticipated sample, is reflected in the relatively wide confidence intervals around costs. However, this uncertainty is taken into account in the computation of the cost-effectiveness acceptability curves.

Economic perspective
We adopted a health-service perspective for the economic analysis, and it is possible that the results might have differed had a societal perspective been included. For example, if one treatment enabled more patients to return to work, this might have altered the findings. However, given the lack of differences in effectiveness, we think this unlikely.

Implications for practice
It is difficult to judge whether a value of £50 per additional depression-free week is acceptably cost-effective. The National Institute for Clinical Excellence (2004) compares treatments with a benchmark level of around £20 000 to £30 000 per QALY, and at this level we can be approximately 60% sure that SSRIs are most cost-effective when compared with TCAs or lofepramine, whereas there is a less than 25% chance that either of the other two treatments are most cost-effective. The initial choice of an SSRI has therefore been shown to be more cost-effective at a reasonable level of probability, and this will increase as the relative cost of SSRIs falls, as their patents expire. Our results therefore tend to support the National Institute for Clinical Excellence guidelines on depression which recommend SSRIs as first-choice antidepressants in primary care (National Collaborating Centre for Mental Health, 2004).


arrow
APPENDIX
 
Recommended daily dosages of antidepressants
Tricyclic antidepressants
Age 18–65 years: 50 mg rising in 25 mg weekly steps to a maximum of 150 mg.

Age 65+ years: 25 mg rising in 25 mg weekly steps to a maximum of 120 mg.

Selective serotonin reuptake inhibitors
Fluoxetine: 20 mg daily dose throughout. Paroxetine: 20 mg increasing to 30 mg after 3 weeks and to a maximum of 40 mg after 6 weeks. Sertraline: 50 mg increasing after 3 weeks to 100 mg and after 6 weeks to a maximum of 150 mg.

Lofepramine
Daily dosage of 70 mg rising in weekly 70 mg increments in divided doses to a maximum of 210 mg.


arrow
Clinical Implications and Limitations
 
CLINICAL IMPLICATIONS
  • Differences in antidepressant costs suggested that offering a tricyclic antidepressant might be the cheapest option, but differences in overall costs, cost-effectiveness and cost-utility were not significant.
  • The cost-utility acceptability curve suggests that, for values placed on an additional quality-adjusted life-year of over £5000, selective serotonin reuptake inhibitors (SSRIs) were likely to be most cost-effective, although the probability of this did not rise above 0.6.
  • Our findings tend to support the National Institute for Clinical Excellence recommendations that SSRIs should be the first choice of antidepressant in primary care.

LIMITATIONS

  • We were unable to recruit a sample of the desired size, reducing the study’s power to detect differences in effectiveness and costs.
  • Loss to follow-up approaching 50% over12 months further limited the power.
  • Many fewer referrals per practitioner were received than anticipated, and the patients referred may not be representative of all new patients with depression being treated in primary care.


arrow
ACKNOWLEDGMENTS
 
We are grateful to all the general practitioners and patients who took part. The study was funded by the Health Technology Assessment programme of the National Health Service Research and Development Directorate. L.L. now works as a health technology analyst at the National Institute for Clinical Excellence, but this work was conducted while she was a research fellow at the Health Economics Research Group (HERG) at Brunel University. M.B.’s input was funded by HERG’s programme grant from the Department of Health’s Policy Research Programme.

T.K. has received fees for speaking at educational meetings from Pfizer, Lilly, Wyeth and Lundbeck. R.P. has received hospitality and fees for speaking at academic meetings from GlaxoSmithKline, Pfizer, Lilly, Merck, Wyeth, Lundbeck and Organon. He has also received support for research from Boots plc (makers of dosulepin). D.B. has attended advisory board meetings for Lilly, GlaxoSmithKline, Pfizer, Lundbeck and Asahi Kasei and has received fees for speaking at educational meetings from Lundbeck, Lilly, GlaxoSmithKline, Pfizer, Wyeth and Asahi Kasei. He has received grant funding for clinical trials from Lundbeck, Wyeth, Lilly, GlaxoSmithKline (and SmithKline Beecham) and Pfizer, and has received grants from Wyeth to prepare an expert report. C.T. is an executive director of the Priory Hospitals Group, and has acted as a consultant to Organon UK Pharmaceuticals, Lilly and Pfizer.

The views expressed in this paper are the authors’ and do not necessarily represent the views or stated policies of the Department of Health.


arrow
REFERENCES
 
  1. British Medical Association & Royal Pharmaceutical Society of Great Britain (2001) British National Formulary (September issue). London & Wallingford: BMJ Books & Pharmaceutical Press.
  2. Brown, J. S., Smith, R. R., Cantor, T. et al (1997) General practitioners as providers of minor surgery – a success story? British Journal of General Practice, 47, 205 –210.[Medline]
  3. Canadian Coordinating Office for Health Technology Assessment (1997) A Clinical and Economic Evaluation of Selective Serotonin Reuptake Inhibitors in Major Depression. Ottawa: CCOHTA.
  4. Dolan, P. (1997) Modelling valuations for Euroqol health states. Medical Care, 35, 1095 –1108.[CrossRef][Medline]
  5. Doyle, J. J., Casciano, J., Arikian, S., et al (2001) A multinational pharmacoeconomic evaluation of acute major depressive disorder (MDD): a comparison of cost-effectiveness between venlafaxine, SSRIs and TCAs. Value Health, 4, 16 –31.[CrossRef][Medline]
  6. Drug and Therapeutics Bulletin (2003) Mild depression in general practice: time for a rethink? Drug and Therapeutics Bulletin, 41, 64 .
  7. Einarson, T. R., Arikian, S., Sweeney, S., et al (1995) A model to evaluate the cost-effectiveness of oral therapies in the management of patients with major depressive disorders. Clinical Therapeutics, 17, 136 –153.[CrossRef][Medline]
  8. EuroQol Group (1990) EQ–5D: a new facility for the measurement of health related quality of life. Health Policy, 16, 199 –208.[CrossRef][Medline]
  9. Forder, J., Kavanagh, S. & Fenyo, A. (1996) A comparison of the cost effectiveness of sertraline versus tricyclic antidepressants in primary care. Journal of Affective Disorders, 38, 97 –111.[CrossRef][Medline]
  10. Hatziandreu, E. J., Brown, R. E., Revicki, D. A., et al (1994) Cost utility of maintenance treatment of recurrent depression with sertraline versus episodic treatment with dothiepin. Pharmacoeconomics, 5, 249 –264.[Medline]
  11. Hosak, L., Tuma, I., Hanus, H., et al (2000) Costs and outcomes of use of amitriptyline, citalopram and fluoxetine in major depression: exploratory study. Acta Medica, 43, 133 –136.[Medline]
  12. Jonsson, B. & Bebbington, P. E. (1994) What price depression? The cost of depression and the cost-effectiveness of pharmacological treatment. British Journal of Psychiatry, 164, 665 –673.[Abstract/Free Full Text]
  13. Kendrick, T., Stevens, L., Bryant, A., et al (2001) Hampshire Depression Project: changes in the process of care and cost consequences. British Journal of General Practice, 51, 911 –913.[Medline]
  14. Lewis, G., Pelosi, A. J., Araya, R. C., et al (1992) Measuring psychiatric disorder in the community: the development of a standardised assessment for use by lay interviewers. Psychological Medicine, 22, 465 –486.[Medline]
  15. Manly, B. F. J. (1997) Randomisation, Bootstrap and Monte Carlo Methods in Biology (2nd edn). London: Chapman & Hall.
  16. Meltzer, H., Gill, B. & Petticrew, M. (1994) The Prevalence of Psychiatric Morbidity Among Adults Aged 16–64, Living in Private Households, in Great Britain. London: Office for Population Censuses and Surveys.
  17. Moffett, J. K., Torgerson, D., Bell-Syer, S. B., et al (1999) Randomised controlled trial of exercise for low back pain: clinical outcomes, costs and preferences. BMJ, 319, 279 –283.[Abstract/Free Full Text]
  18. National Collaborating Centre for Mental Health (2004) Depression, Management of Depression in Primary and Secondary Care. London: National Institute for Clinical Excellence.
  19. National Institute for Clinical Excellence (2004) Guide to the Methods of Technology Appraisals. London: NICE.
  20. Netten, A., Rees, T. & Harrison, G. (2001) Unit Costs of Health and Social Care. Canterbury: Personal Social Services Research Unit, University of Canterbury.
  21. Ratcliffe, J., Ryan, M. & Tucker, J. (1996) The costs of alternative types of routine antenatal care for low-risk women: shared care vs care by general practitioners and community midwives. Journal of Health Services and Research Policy, 1, 135 –140.
  22. Simon, G. E., von Korff, M., Heiligenstein, M., et al (1996) Initial antidepressant choice in primary care – effectiveness and cost of fluoxetine vs tricyclic antidepressants. JAMA, 275, 1897 –1902.[Abstract/Free Full Text]
  23. Stewart, A. (1994) Antidepressant pharmacotherapy: cost comparison of SSRIs and TCAs. British Journal of Medical Economics, 7, 67 –79.
  24. Ware, J., Snow, K. K., Kosinski, M., et al (1993) SF–36 Health Survey, Manual and Interpretation Guide. Boston: Health Institute, New England Medical Center.
  25. Woods, S. W. & Rizzo, J. A. (1997) Cost-effectiveness of antidepressant treatment reassessed. British Journal of Psychiatry, 170, 257 –263.[Abstract/Free Full Text]
  26. World Health Organization (1992) International Statistical Classification of Diseases and Related Health Problems (ICD–10). Geneva: WHO.
  27. Zigmond, A. S. & Snaith, R. P. (1983) The Hospital Anxiety and Depression Rating Scale. Acta Psychiatrica Scandinavica, 67, 361 –370.[Medline]
Received for publication September 13, 2004. Revision received March 4, 2005. Accepted for publication March 16, 2005.


Related articles in BJP:

Highlights of this issue
KIMBERLIE DEAN
BJP 2006 188: 301-a13. [Full Text]  




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit an eLetter
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in BJP
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via CrossRef
Right arrow Citing Articles via Google Scholar
Right arrow Citing Articles via Scopus
Google Scholar
Right arrow Articles by KENDRICK, T.
Right arrow Articles by THOMPSON, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by KENDRICK, T.
Right arrow Articles by THOMPSON, C.