|
Efficacy and cost-effectiveness of drug and psychological treatments for common mental disorders in general health care in Goa, India: a randomised, controlled trial |
Patel V, Chisholm D, Rabe-Hesketh S, Dias-Saxena F, Andrew G, Mann A |
|
|
Record Status 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. Health technology The use of fluoxetine, an antidepressant, for patients suffering from common mental disorders. Fluoxetine was given at a standard dose of one capsule per day. The dose was doubled if the patients did not improve in their health outcomes after 2 months.
Economic study type Cost-effectiveness analysis.
Study population To be included, the participants were required to be aged 17 or older, and be resident in the vicinity of a hospital participating in the study. They were also required to speak Konkani, Marathi or English, to not be acutely medically ill, nor psychotically ill or in psychiatric care. Patients who satisfied these criteria and who scored 15 or more points on the Revised Clinical Interview Schedule (CISR) were eligible to enter the trial.
Setting The study was set in the general medical outpatient clinics of main district general hospitals in Goa, west India. The authors described the outpatient department as being used as a first point of contact for health care. Thus, the study was effectively set in primary care.
Dates to which data relate The effectiveness and resource use data were gathered between December 1998 and May 2000. The price year was 1999.
Source of effectiveness data The evidence for the final outcomes was derived from a single study.
Link between effectiveness and cost data The costing was performed retrospectively on the same patient sample as that used in the effectiveness analysis.
Study sample A power calculation performed during the planning phase of the study estimated that each of the three groups would need to contain 94 participants to achieve 90% power. The actual group size was set at 150 to allow for an attrition rate of 50%.
Of 485 consecutive patients who satisfied the inclusion criteria, 24 (4.95%) were excluded because they scored less than 15 on the CISR and 11 (2.27%) withdrew their consent. The remaining 450 were randomised equally to the antidepressant, placebo and psychological therapy groups, each of which contained 150 individuals. The mean age of all patients included in the study was 48.6 (standard deviation, SD=14.1), 66% were married and more than 80% were females.
Study design The study was a randomised controlled trial that was carried out in two centres. Randomisation was performed using random number tables in blocks of 9 patients. The follow-up was undertaken at 2, 6 and 12 months.
At 2 months, 125 participants in the antidepressant group were assessed. Fourteen had dropped out of the study, 2 had died and 9 were unavailable for this assessment. For the placebo group, 135 were assessed, of which 7 had dropped out and 8 were unavailable. For the psychological therapy group, 121 were assessed, of which 16 had dropped out, 2 had died and 11 were unavailable.
At 6 months, 118 participants in the antidepressant group were assessed. One had died and 15 were unavailable. For the placebo group, 127 were assessed, of which 4 had dropped out, 2 had died and 10 were unavailable. For the psychological therapy group, 117 were assessed, of which 2 had dropped out, 1 had died and 12 were unavailable.
At 12 months, 116 participants in the antidepressant group were assessed. One had died and 16 were unavailable. For the placebo group, 128 were assessed, of which 2 had dropped out, 2 had died and 5 were unavailable. In the psychological therapy group, 13 of the 116 assessed were unavailable.
None of the deaths resulted from suicide.
Antidepressants and placebo were identical in appearance and the patient, and the medication dispenser and the follow-up investigator were all unaware of the treatment allocation. The effectiveness of the masking procedure was also assessed. There was no attempt to mask patients in the psychological therapy group.
Analysis of effectiveness The analysis of the clinical study was conducted on an intention to treat basis, including all participants who completed at least one outcome assessment. The primary health outcome was the psychiatric morbidity, as measured by the CISR. The secondary health outcomes were disability (as measured by the BDQ), adherence and satisfaction with the treatment received. The three groups were compared in terms of age, gender, marital status, religion, years of formal education and CISR score.
Effectiveness results At 2 months, the mean difference in the CISR score adjusted for baseline was -2.6 (95% confidence interval, CI: -4.82 - -0.38; p=0.02) for fluoxetine compared with placebo, and 0.21 (95% CI: -2.03 - 2.46) for psychological treatment compared with placebo. The mean difference in the BDQ score (i.e. disability) was -1.28 (95%vCI: -2.41 - -0.15; p=0.03) between the fluoxetine and placebo groups, and 0.63 (95% CI: -1.77 - 0.51; p=0.28) between the psychological therapy and placebo groups. The differences in CISR and BDQ scores between fluoxetine and placebo were not statistically significant when estimated at 6 months and 12 months
The adherence rates were significantly different between the fluoxetine group (46%) and placebo group (56%) at 6 months, (p=0.05).
The main reason given for stopping treatment was feeling better.
Satisfaction with treatment was similar between all three groups, ranging from 67 to 73%, as was perception of benefit of treatment.
Clinical conclusions Compared with placebo, drug therapy gave slightly better results for clinical and disability outcomes in the short term, but this difference was not significant in the longer term (1 year). This is consistent with findings in developed countries. Placebo and psychological treatment gave similar results in terms of the clinical outcomes.
Modelling A regression model was used to analyse the CISR and Brief Disability Questionnaire (BDQ) scores for the intention to treat analyses.
Measure of benefits used in the economic analysis The authors used the CISR scores as a measure of benefit. These were obtained directly from the effectiveness analysis.
Direct costs Discounting was not relevant and was not applied for the majority of the cost items. The costs and the quantities of resources used were not reported separately. The health care costs were for treatment, primary care and outpatient visits, inpatient admissions, medications and investigations. The travel costs of the patients were also reported. A questionnaire completed at each follow-up was used to gather data on health care use and associated costs, and the costs incurred by service users and their families in the preceding 3 months. The unit costs for services were calculated, applied to each patient's use of services, and then converted into monthly estimates. The unit costs for services were based on an estimate of the 1-year cost of personnel, revenue, overheads and capital. A discount rate of 5% was applied to lands and buildings, and 10% to equipment and vehicles. The quantities of resources used were estimated for the same sample of patients as that used in the effectiveness analysis. The costs of antidepressant and placebo were based on consultations with a hospital outpatient doctor. The resource use data were gathered between December 1998 and May 2000. The price year was 1999.
Statistical analysis of costs The cost data were analysed by regressing the costs at each follow-up on baseline cost, group, time and group-by-time. To avoid the assumptions of parametric tests, bootstrapping with samples of 5,000 were used to estimate the 95% CIs for the mean costs.
Indirect Costs The estimated time taken to travel to and wait for services, and the workdays lost by patients and their informal caregivers, were measured through a questionnaire completed at the follow-up. No rationale was given for the inclusion of productivity costs. Two rates (semi-skilled and unskilled) were applied to estimate the costs associated with the lost days. These rates were apparently conservative measures. The quantities and the costs were not analysed separately. Discounting was irrelevant and was not applied. The price year was 1999.
Sensitivity analysis No sensitivity analysis of either the costs or benefits was performed.
Estimated benefits used in the economic analysis The benefit measure used in the economic analysis was the incremental mean CISR scores, as reported in the effectiveness results.
Cost results Over one year, the total health costs for fluoxetine therapy were R2,270 (SD=4,342) and the patient and family costs were R11,208 (SD=18,011).
The corresponding values for placebo were R3,722 (SD=6,267) for the total health costs and R12,906 (SD=20,275) for the patient and family costs. For psychological treatment, the values were R3,090 (SD=4932) and R16,951 (SD=21,792), respectively.
Synthesis of costs and benefits The benefits were reported at 2 months and from 2 to 12 months. At each time either fluoxetine or the psychological treatment was compared with placebo. The two incremental cost-effectiveness ratios considered were the incremental health care cost per CISR score and the patient and family incremental cost per CISR score. At both times, and in both cost categories, fluoxetine dominated placebo. However, when the psychological treatment was compared with placebo it resulted in lower health care costs, but worse CISR scores at both times. When the patient and family costs per CISR score were considered, psychological treatment was dominated by placebo at both times. Follow-up beyond 12 months was not reported.
Authors' conclusions Fluoxetine gave slightly better results than placebo in the short term, as measured by the clinical and disability outcomes, but no significant difference in the long term. However, fluoxetine also decreased the health care costs during follow-up, as it was a dominant strategy compared to placebo. The lack of effectiveness of the psychological treatment was unexpected.
CRD COMMENTARY - Selection of comparators The choice of placebo as the comparator for both the fluoxetine and psychological treatments was explicitly justified. Normal practice in India is for the patient to receive some physical intervention, such as vitamins, which could function as a placebo. You should decide if this is appropriate in your own setting
Validity of estimate of measure of effectiveness Considerable attention was devoted to the design of the study. The analysis used a randomised controlled trial, which was appropriate for the study question. The study sample was representative of the study population and the groups were shown to be comparable at baseline. Appropriate statistical analyses were performed to take potential biases and confounding factors into consideration. Power calculations were performed and blinding procedures were applied for placebo versus fluoxetine. The study population came from low socio-economic groups and had moderate levels of morbidity. The basis of the analysis was intention to treat, which was appropriate for the study.
Validity of estimate of measure of benefit The estimate of benefit was obtained directly from the effectiveness analysis. This choice of estimate was justified.
Validity of estimate of costs All the categories of cost relevant to the societal perspective appear to have been included in the analysis. The costs and the quantities of resource used were not reported separately, thus reducing the reproducibility of the study in other settings. A statistical analysis of the cost data was performed using regression analysis with bootstrapping. However, no sensitivity analyses were performed. The sources of the unit costs were reported and justified. Currency conversions were not performed. Discounting was applied to capital resources. As the study was for one year only, discounting was neither required nor applied to the other resource costs.
Other issues The authors made appropriate comparisons of their findings with those from other studies. It was clear that the study was looking at results in a developing country in comparison with published studies set in developed countries. The authors did not present their results selectively. The authors acknowledged that one possible reason for the lack of efficacy of the psychological treatment was the severe social problems and life circumstances faced by the patients. The authors also underlined the fact that the dominance of fluoxetine might be due to the low cost of the drug in their country, which would not be applicable to other settings.
Implications of the study The authors noted that the cost-effectiveness of fluoxetine depends on its price, which varies between countries. They suggested that further studies would need to be undertaken in other developing countries to corroborate their results.
Source of funding Funded by the Wellcome Trust.
Bibliographic details Patel V, Chisholm D, Rabe-Hesketh S, Dias-Saxena F, Andrew G, Mann A. Efficacy and cost-effectiveness of drug and psychological treatments for common mental disorders in general health care in Goa, India: a randomised, controlled trial. Lancet 2003; 361: 33-39 Indexing Status Subject indexing assigned by NLM MeSH Antidepressive Agents, Second-Generation /economics /therapeutic use; Cost-Benefit Analysis; Double-Blind Method; Female; Fluoxetine /economics /therapeutic use; Humans; India; Logistic Models; Male; Mental Disorders /classification /drug therapy /therapy; Middle Aged; Psychotherapy /economics; Treatment Outcome AccessionNumber 22003008018 Date bibliographic record published 31/05/2004 Date abstract record published 31/05/2004 |
|
|
|