|Effectiveness and cost-effectiveness of antidepressant treatment in primary health care: a six-month randomised study comparing fluoxetine to imipramine
|Serrano-Blanco A, Gabarron E, Garcia-Bayo I, Soler-Vila M, Carames E, Penarrubia-Maria M T, Pinto-Meza A, Haro J M, Depressio en Atencio Primaria de Gava group
This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.
The present study compared two antidepressant strategies for persons suffering from depressive disorders. Fluoxetine (FLU), a selective serotonin reuptake inhibitor, was compared with imipramine (IMI), a tricyclic antidepressant. The mean starting dosage was 19.4 mg for FLU and 33.5 mg for IMI. The mean daily dosage at day 30 was 20.7 mg for FLU and 58.3 mg for IMI.
Type of intervention
Treatment and rehabilitation.
Economic study type
The study population comprised patients aged 18 to 65 years who were diagnosed with a depressive disorder, such as major depressive disorder (MDD), dysthymic disorder (DD), or depressive disorder not otherwise specified (DDNOS). The main exclusion criteria were:
pharmacological antidepressant treatment in the previous 60 days;
a history of alcohol or drug abuse;
psychotic symptoms or a history of bipolar disorder;
antipsychotic drug use, lithium or antiepileptic medication in the previous 6 months;
pregnancy, lactation, or women intending to become pregnant;
need of hospital admission for depression in the judgment of the general practitioner (GP); and
requirement for intensive psychotherapy or electroconvulsive therapy.
The setting was primary care. Specifically, three primary health care centres (PHCC) within the metropolitan area of Barcelona, Spain. The economic study was carried out in Spain.
Dates to which data relate
The effectiveness evidence and resource use data were from 1999 to 2001. The price year was 2001.
Source of effectiveness data
The effectiveness data were derived from a single study.
Link between effectiveness and cost data
The costing was carried out prospectively on the same sample of patients as that used in the effectiveness analysis.
The study sample initially selected comprised 110 patients, of which 103 were included. Two patients declined to participate in the evaluation and 5 could not be assessed in the week after the initial visit. The patients were randomised to treatment with either FLU (53 patients) or IMI (50 patients). There were no significant differences in baseline characteristics between the two treatment groups. Of the 103 patients, 94 (91.3%) completed at least one of the follow-up visits and were included in the statistical analyses. Sample size calculations to detect cost-differences were reported. These were based on an assumption that the effects were equal and that the costs for depression in primary care in Spain were about EUR 1,000.
This analysis was based on a multi-centre, randomised prospective naturalistic study. Randomisation was stratified by PHCC in 10-patient blocks. The investigators created a series of numbered envelopes that contained the group assignment for each patient. The GP was responsible for all patient care following his or her usual clinical practice and was only instructed to note in the patient's chart any change in antidepressant medication. Follow-up evaluations were scheduled at 1-, 3 and 6-monthly intervals after the initial evaluation. These were carried out by a clinical psychologist, blind to the patients' treatment status, who administered the assessment instruments.
Analysis of effectiveness
The outcome measures were:
the Structured Clinical Interview for DSM-IV (SCID-I),
the Montgomery Asberg Depression Rating Scale (MADRS),
the Clinical Global Impression rating scale, and
a questionnaire with basic socio-demographic, clinical and health service use data.
The analysis was conducted on an intention to treat basis.
Patients in both treatment groups improved during the follow-up period and there were no differences in effectiveness between them. However, the analysis of treatment response revealed differences amongst diagnostic sub-groups.
Patients with MDD improved more on IMI than on FLU at 30 days, as measured by the MADRS, (p<0.05).
The adjusted mean difference in clinical response between the FLU and IMI groups at 30 days using the MADRS was -6.50 (95% confidence interval, CI: -12.40 to -0.65).
There were no statistically significant differences in outcome in later evaluations (i.e. 90 and 180 days).
For patients with DD, IMI achieved better outcomes than FLU at the 3-month follow-up, (p<0.05).
The adjusted mean difference in clinical response between the FLU and IMI groups at 3 months using the MADRS was -10.06 (95% CI: -18.51 to -1.62).
Patients with DDNOS showed more improvement on FLU, as measured by the MADRS during the first month of treatment, (p<0.05).
The adjusted mean difference in clinical response between the FLU and IMI groups at 1 month using the MADRS was 4.12 (95% CI: 0.15 to 8.10).
Sixty-four per cent of patients in the FLU and IMI groups were on antidepressant therapy at the end of the study.
During the first 6 months of treatment in primary care for a depressive disorder, IMI appears to have had similar effectiveness to FLU. However, these findings varied according to the disorder being treated.
Measure of benefits used in the economic analysis
No summary measure of benefit was used and the cost and effects were left disaggregated. In effect, a cost-consequences analysis was performed.
The direct costs were for psychotropic drug treatment (antidepressants and benzodiazepines), office visits, nursing care and social work, visits to specialist practitioners, psychiatric and general medicine emergency-room care, and psychiatric and general medicine hospital admissions. The cost of the drugs was taken from the International Vademecum (Red Book) of 2001, and included value-added tax. The costs of the patients' use of health care services were obtained from a local published cost database, 2001. Discounting was not carried out, which was appropriate given the short-term horizon of the study. The quantities and the costs were reported separately, and their estimation was based on actual data. The price year was 2001.
Statistical analysis of costs
A multivariate economic data analysis was performed using a logarithmic transformation of costs to reduce the skewness of the distribution. Treatment and total costs in the 2 months before baseline were included as covariates.
The indirect costs were calculated by multiplying authorized sick leave days by the minimum daily wage in Spain (EUR 24.04) for 2001. Discounting was not carried out, which was appropriate given the short-term horizon of the study. The quantities and the costs were reported separately, and their estimation was based on actual data. The price year was 2001.
A sensitivity analysis of FLU price was conducted using current drug acquisition costs according to the International Vademecum (Red Book) of 2003, with value-added tax included.
Estimated benefits used in the economic analysis
See the 'Effectiveness results reported previously.
A comparison of the economic data showed significant differences in total costs between treatment groups after 90 and 180 days of follow-up. The FLU group showed higher total costs after both 90 and 180 days of follow-up. The total costs were EUR 784.26 (standard deviation, SD=869.91) for the FLU group versus EUR 489.11 (SD=693.54) for the IMI group after 90 days, (p<0.05), and EUR 1,330.38 (SD=1,499.07) versus EUR 691.81 (SD=1,071.62) after 180 days, (p<0.001). A separate comparison of direct and indirect costs showed that the FLU group had higher indirect costs after 180 days of follow-up (EUR 999.21; SD=1,409.91) than the IMI group (EUR 431.44; SD=82.02; p<0.05). No statistically significant differences were observed between FLU and IMI direct costs.
Comparisons among diagnostic groups revealed that in MDD, the FLU group showed significantly higher cumulative total costs after 30 days, (p<0.01), 90 days, (p<0.05) and 180 days, (p<0.001). A separate comparison of direct and indirect costs showed that the FLU group had higher direct costs after 30, 90 and 180 days of follow-up, (p<0.05 in each measure). No statistically significant differences were observed between FLU and IMI indirect costs.
Patients with DD in the FLU group showed significantly higher total costs after 30 days, (p<0.05), 90 days, (p<0.01) and 180 days, (p<0.05) than patients in the IMI group. A separate comparison of direct and indirect costs showed that the FLU group had higher direct costs after 90 days of follow-up, (p<0.05), and higher indirect costs after 90 and 180 days of follow-up, (p<0.05 and p<0.01, respectively). Patients with DDNOS did not show differences in total costs between the treatment groups.
Synthesis of costs and benefits
The costs and benefits were not combined. The sensitivity analysis showed that the statistical significance of total costs remained the same, though that of some direct costs had changed. Thus, for all patients, there were still statistically significant differences in total costs at 90 and 180 days, (p<0.05 and p<0.001, respectively). In patients with a diagnosis of MDD, statistically significant differences between treatment groups remained at 30, 90 and 180 days, as they also did in patients diagnosed with DD. The patients diagnosed with DDNOS continued as before, without statistically significant differences between treatment groups.
Imipramine (IMI) might be a more cost-effective treatment than fluoxetine (FLU) since the similar effectiveness was balanced by lower cost during the first 6 months of treatment in primary care for a depressive disorder.
CRD COMMENTARY - Selection of comparators
A justification was given for the comparator used. It reflected standard practice in the authors' setting. You should judge whether these drugs and treatments are relevant in your own setting, or whether other comparators from other drugs and strategies could also have been relevant.
Validity of estimate of measure of effectiveness
The analysis was based on a randomised prospective study with naturalistic follow-up, which was adequate for the study question. The study sample was representative of the study population and power calculations were reported, although only for costs. In addition, the patient groups were shown to be comparable at analysis. The authors stated that the small sample size of the study limited their ability to detect differences by diagnostic group, so the results of the sub-group analysis should be interpreted with care. Appropriate statistical analyses were undertaken to ensure the comparability of patients groups at baseline and to account for potential biases and confounding factors.
Validity of estimate of measure of benefit
The authors did not derive a measure of health benefit. The reader is thus referred to the comments in the 'Validity of estimate of measure of effectiveness' field (above).
Validity of estimate of costs
The study had a detailed table with direct, indirect and cumulative total costs included from the adopted societal perspective. The costs were treated stochastically and a sensitivity analysis of the prices was conducted. Discounting was not necessary since the study had a very short-term time horizon. The costs and the quantities were reported separately. The price date was also reported. These facts should aid any future reflation or re-analysis undertaken.
The authors compared their findings with those from other studies. The issue of generalisability to other settings was addressed and the conclusions appear to reflect the scope of the analysis. The authors recognised some limitations of the study. First, physicians participated in the study on a voluntary basis; this group might be less representative of all GPs. Second, the requirement of informed consent might have influenced the selection of patients. Third, randomisation affected the physician's choice of treatment and thus distanced the setting from normal clinical practice. Fourth, the measurement of the indirect costs associated with sick leave used the same cost per sick day for all patients. Fifth, sick leave days might reflect physical as well as mental disorders. Finally, no data on the incidence of side effects were recorded.
Implications of the study
The authors stated that one interesting finding of the study was that IMI might be a more cost-effective treatment than FLU at a lower mean dosage than the dosage recommended in clinical guidelines (1993 APA treatment guidelines for MDD). Some authors think that the high dosage level established for randomised controlled trials might not be necessary in primary care.
Source of funding
Supported by the Catalan Agency for Health Technology Assessment and Research.
Serrano-Blanco A, Gabarron E, Garcia-Bayo I, Soler-Vila M, Carames E, Penarrubia-Maria M T, Pinto-Meza A, Haro J M, Depressio en Atencio Primaria de Gava group. Effectiveness and cost-effectiveness of antidepressant treatment in primary health care: a six-month randomised study comparing fluoxetine to imipramine. Journal of Affective Disorders 2006; 91(2-3): 153-163
Subject indexing assigned by NLM
Adult; Antidepressive Agents, Tricyclic /economics /therapeutic use; Cost-Benefit Analysis; Depressive Disorder, Major /drug therapy /economics; Double-Blind Method; Female; Fluoxetine /economics /therapeutic use; Follow-Up Studies; Humans; Imipramine /economics /therapeutic use; Male; Primary Health Care /economics /methods; Prospective Studies; Serotonin Uptake Inhibitors /economics /therapeutic use
Date bibliographic record published
Date abstract record published