A systematic review and meta-analysis of trials of treatment of depression from India.

BACKGROUND: Antidepressants hold the center stage in the treatment of depression in current clinical practice. However, it is also well-known that the treatment response and dosage requirement are influenced by ethnic variations. Although many efficacy studies have evaluated the efficacy of antidepressants, there is lack of systematic reviews and meta-analysis of the existing literature from India.
OBJECTIVE: To systematically review the efficacy of treatment of depression in the Indian context.
MATERIALS AND METHODS: We searched PubMed, Psychinfo, Medknow and Google scholar to identify studies published in peer-reviewed English language journals. All controlled trials from India evaluating the clinical efficacy of antidepressants, electroconvulsive therapy (ECT), and repetitive transcranial magnetic stimulation (rTMS) for management of depression were evaluated. Data were extracted using standard procedures and risk of bias was evaluated. Effect sizes were computed for the individual studies.
RESULTS: Effect sizes were computed from 35 clinical trials. Overall, medications were superior to placebo for treatment of depression (mean effect size (ES) of 0.87, confidence intervals (CI of 0.71-1.02). The effect was greatest for tricyclic antidepressants (ES of 1.00, CI of 0.80-1.21) followed by monoamine oxidase inhibitors (ES 0.54, CI of 0.40-0.67). ECT was superior to antidepressants (ES 0.32, CI of - 0.21 to 0.86) and active rTMS was found to be superior to sham rTMS with mean effect size of 0.74 (CI 0.39-1.08). Risk of bias was found to be considerable. However, the review literature suggests that most of the studies have not been powered adequately and have been limited to small sample sizes.
CONCLUSIONS: Although there is some data from India with respect to efficacy of antidepressants, most of the trials have been of shorter duration have been inadequately powered. The available data support the superiority of antidepressants over placebo and that of ECT over antidepressants.

INTRODUCTION

Depression is one of the most common psychiatric conditions in the general population.[1234] Not only depression imposes a great burden upon the health services, it also results in excess mortality through suicide and increased frequency of medical morbidity.[567] Although many effective treatment options are available for treatment of depression,[89] pharmacotherapy is the most commonly used modality.

Ethnic and regional variations are present in the manifestations of depression.[1011] This should also logically translate into differences in treatment options which are found to be effective and acceptable.[12] Also, pharmacogenomic data suggest that genetic make-up of different racial groups vary to some degree. This has implications on the pharmacokinetics and pharmacodynamics of the medications.[13] The pharmacological agents used for treatment of depression found effective in one population with a particular genetic make-up may not be as effective or become intolerable in another genetic stock. With the recent advances in the pharmacogenetics, it is meaningful to ascertain how well do the medications work in the Indian context.[1415]

Much of the studies in relation to the treatment of depression have been conducted in the Western countries. The results obtained from Western studies may not hold substantially true in the Indian context as the services for patients with depression are organized differently.[16] Further, the perspectives of patient in the form of type of treatment required may also differ. In view of this, there is a need to rely on studies conducted in India to evaluate what kind of interventions might work for depression.

Hence, this systematic review was conducted with the objective of evaluating the studies assessing the usefulness of antidepressants and other treatment modalities in management of depression in the Indian context, especially in terms of efficacy/effectiveness and the quality of methodologies used in these studies. As meta-analysis is widely accepted method of summarizing data about treatment options, a systematic review and meta-analysis of various studies was conducted with the objective of assessing the efficacy of antidepressants and other treatment modalities in management of depression in the Indian context.

MATERIALS AND METHODS

Search strategy: Electronic searches for published trials were carried out using PubMed, Psychinfo, and Google Scholar search engines. Additionally, the web site of Indian J Psychiatry and Medknow was used to search studies published in other Indian Journals like Indian Journal of Psychological Medicine, Indian Journal of Medical Research, Indian Journal of Medical Sciences, Journal of Postgraduate Medicine, Indian Journal of Pharmacology, and so on. The keywords were “depression,” “India,” “antidepressant” (and names of individual antidepressants), “repetitive transcranial magnetic stimulation (rTMS),” and “electroconvulsive therapy (ECT).” These key words were used in various combinations. The multiple searches carried out in January 2013 yielded 1147 abstracts. Further studies were identified from the cross references and reference list of included studies. Unpublished work was not sought as a part of this review, as till recently there was no registry for recording all the drugs trials in India. Hence, it is not possible to assess all the unpublished data.

Study selection: Studies published in peer-reviewed English language journals were included. The selection criteria for inclusion of various studies into this review and meta-analysis were: Controlled studies evaluating an antidepressant for depression, the diagnosis of depression made in accordance to any nosological system or through clinician's interview, studies having at least two treatment arms and reporting efficacy/effectiveness. Studies evaluating the antidepressants in animal models and those evaluating the efficacy/effectiveness antidepressants in other conditions like enuresis, anxiety disorders, and sexual dysfunction were excluded. Studies with less than five participants with depression in an individual treatment arm or had reported results in manner from which effect sizes could not be calculated were excluded from the meta-analysis. Further, studies which have evaluated an alternative and complementary medicine for treatment of depression (e.g., ayurvedic preparations) were not included. Recently, India has been a hub for evaluating the efficacy and tolerability of various newer antidepressants. However, we did not include these multinational trials in which patients were recruited from India but the country-specific data were not analyzed separately.

Data extraction: All the abstracts were reviewed by two investigators independently. After initial evaluation, 85 abstracts were selected for further evaluation. Full texts of all these articles were searched. We could not assess the full text of two articles;[1718] hence, these articles could not reviewed and resultantly were not included in the analysis. The full text of all these articles were reviewed independently by both the investigators for the study characteristics (e.g., nature of the study, manner of randomization, blinding, duration of study, intention to treat analysis) and clinical information (number of subjects, age range or mean, gender distribution, diagnoses made, medication groups, past treatment, efficacy measure, outcome, side effects). There was a high degree of concordance between the evaluators of the studies. A total of 41 controlled trials were identified after the review of the data. However, six controlled studies were not included in the meta-analysis as the effect sizes could not be calculated.[192021222324] Of these two studies had mentioned Mann-Whitney U values,[1924] two had limited information for calculation of effect sizes,[2023] one had only two participants in one arm[21] and one had reported P values range.[22]

Hence, on the basis of availability of information 37 studies were selected for final analysis. Any lack of concurrence was resolved by mutual consensus. The studies not included are depicted in Supplementary Table S1.

Studies which had reported more than one outcome measure, primary efficacy measure were used for calculation of effect size. In studies which had explicitly not mentioned a primary efficacy measure, the proportion of patient's population which improved or the first variable mentioned for efficacy was used for calculation of effect size. Data from intention to treat analysis were used wherever possible. The number needed to treat (NNT) was also calculated for placebo controlled studies.

Risk of bias: Risk of bias was evaluated for the studies included in the meta-analysis. The elements that were studied for risk of bias included random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and dropouts and selective reporting of results. The risk of bias was quantified using the Jadad scale,[25] which is a five-point scale for assessing the quality of randomized controlled trials (RCTs). The rating is done on the basis of reporting of randomization, blinding, and reporting of withdrawals and drop-outs. The scale has been widely used and shows good validity and reliability.[26]

Statistical analysis: Effect sizes were calculated for conducting the meta-analysis by using the standardized mean difference (d).[27] The effect size is a measure of the efficacy of an intervention. This allows easy comparison of studies using disparate methodology and efficacy measures. The greater the value of effect size, the greater is the efficacy. The confidence interval (CI) of effect size gives a measure of the impact of findings. Larger studies usually have narrower CIs. The standardized mean difference (d) was used for calculation of effect sizes, because this gives a robust measure for both categorical and continuous measures. For dichotomous variables of efficacy, logit method was used for deriving effect size and CIs. In this study, random effects model was used for computing the mean effect sizes and I2 test was used for assessing heterogeneity. Random effects model has been reported to be useful in computation of effect sizes, especially when disparate studies are combined for analysis, which was expected for this meta-analysis. The I2 test of heterogeneity is expressed as a percentage and reflects the degree of heterogeneity between the studies. It does not need to take into account the degrees of freedom for comparison of meta-analyses and makes inferences easier than Cochran's Q.[28]

In studies which had more than two groups, effect sizes were calculated for individual comparisons between different treatment modalities or between a treatment method and placebo. Mean effect sizes and CIs were calculated for comparison of tricyclic antidepressants (TCA) versus placebo, monoamine oxidase inhibitors (MAOIs) versus placebo, all antidepressants versus placebo, active versus sham rTMS and ECT versus antidepressants.

RESULTS

Table 1 lists the studies included in the meta-analysis. The 37 studies included in the meta-analysis have presented data in relation to 43 comparisons of different treatments.

Table 1

Studies included in meta-analysis

Of the included comparisons, 20 comparisons followed open-label RCTs design, 15 involved double-blind RCTs, 3 were double blind controlled trials, 2 involved open-label nonrandomized design, and 2 were cross-over trials. Two of the studies had included patients with specific medical illnesses.[5565] Seven comparisons (reported in five studies)[2930313233] involved comparisons of efficacy of TCAs and MAOIs with placebo. Studies evaluating rTMS have done so by comparing add on active rTMS intervention to antidepressant treatment with antidepressants and sham rTMS.[606162]

In nearly half of the comparisons (20 comparisons from 18 trials) imipramine has been evaluated against a placebo, another antidepressant or ECT. Besides imipramine other drugs which have been evaluated include amitriptyline (5 trials), sertraline (4 trials), citalopram (3 trials), sintamil (3 trials), venlafaxine (3 trials), mirtazapine (3 trials), escitalopram (3 trials), milnacipran (2 trials), ECT (2 trials), and rTMS (3 trials). Two trials used Selective Serotonin Reuptake Inhibitors (SSRIs) as a class for evaluation of efficacy of “add-on” antidepressants[6364] to various SSRIs and one trial has compared sertraline in depressed postmyocardial infarction patients with no therapy.[65] One trial each evaluated moclobamide, duloxetine, fluoxetine, bupropion, nortriptyline, fluphenazine add-on to nortriptyline, 5 hydroxytryptophan, duloxetine, niamide, phenelzine, centpropazine, iprindole, protriptyline, alprazolam, and sudarshan kriya.

Most of the studies had used Hamilton Depression Rating Scale (HDRS) as one of the instruments to assess efficacy. In most studies, HDRS was the primary outcome measure. The other instrument that was frequently used was Montgomery-Asberg Depression Rating Scale. Most of the studies were of short duration, with 13 comparisons of 4 weeks duration, 11 of 6 weeks duration, 4 of 8 weeks duration, and 2 of 12 weeks duration. Only one trial was of 6 months duration. Trials which evaluated the efficacy of rTMS have reported so after 6-10 sessions. Of the 43 comparisons, 30 had less than 30 patients in both the comparison arms and 2 trials had less than 30 patients in at least one arm.

The risk of bias in the included studies is shown in Table 2. The Jadad scores ranged from 0 to 5 with a median of 2 (mean of 1.97, mode 2). Newer studies had explicitly used intention-to-treat analysis,[34505458596364] while the older studies had not relied on such statistical procedures.[293032333640434748] When correlation of the year of publication was drawn with risk of bias (measured with Jadad scale), there was no statistically significant correlation (Kendall tau correlation of −0.031, P=0.813). As is evident from Table 2, most of the trials had Jadad score ranging from 0 to 2 (3 trials were rated as 0, 9 trials were rated as 1 and 12 trials were rated as having Jadad score of 12). Only five trials were rated as having Jadad score of 3 and another five were rated as 4. None of the trial was rated as having a Jadad score of 5.

Table 2

Risk of bias in the studies

Effect sizes

Seven comparisons were with placebo from the five placebo-controlled studies.[2930313233] The mean effect size of TCA versus placebo was 1.008 (CI of 0.802 to 1.214), based on a cumulative sample of 514 in either of the efficacy arm from five studies,[2930313233] with I2 test of heterogeneity at 9.13. The mean effect size of comparison of MAOI to placebo was 0.541 (CI 0.407-0.675) based on two comparisons from one study (I2=0).[30] The mean effect size of all antidepressants versus placebo was 0.871 (CI of 0.717-1.026, based upon a cumulative sample size of 714) with some degree of heterogeneity (I2=19.82). The mean effect size of comparison of ECT versus other medications was 0.265 (CI was 0.010-0.520) based upon three studies[374445] and a cumulative sample size of 58 (I2=0). Active rTMS was found to be superior to sham rTMS with effect size of 0.74 (CI 0.396-1.084) based on three studies[606162] with a cumulative sample of 104 and considerable degree of heterogeneity (I2=54.89). Bupropion,[63] mirtazapine,[64] and fluphenazine[51] were found to be effective as an add-on treatment for depression (effect sizes of 0.859, 0.232, and 0.650, respectively). The forest plot of the included studies is shown in Figure 1.

Figure 1

Forest plot of included studies

The NNT based upon the studies are depicted in Table 3. The NNT provides a relative measure of how many patients are needed to be put on medication to see one true effect of the medication. The NNT varied from 1.76 to 26.58.

Table 3

Number needed to treat in controlled studies

DISCUSSION

This systematic review evaluates the studies originating from India on efficacy of antidepressants for depression. This work extends upon a previous review by Avasthi et al.,[66] about research on antidepressants in India, inculcating the research published thereafter. This systematic review and meta-analysis estimates the quantitative effect of the antidepressants reported in the studies and presents them in a comparable manner using effect sizes. This review also systematically comments upon the risk of bias in the trials of the treatments used for depression and evaluates the NNT. However, a few studies could not be included in meta-analytic comparison.

This is to the best to our knowledge the first meta-analysis evaluating the antidepressants efficacy trials originating in India. The meta-analysis shows that TCAs seemingly have more robust data about efficacy compared to MAOIs, ECT works somewhat better than medications, and addition of rTMS to usual treatment may be beneficial. Meta-analytic evidence from other parts of the world also suggest that TCAs may be significantly more effective than SSRIs (effect size in the range of 0.30 favoring TCAs).[67] However, when only larger studies are included, the difference between TCAs and SSRIs diminish.[6768] This may be due to smaller studies of TCAs favoring these medication, which is also evident in data from India. We did not find any placebo-controlled trials for SSRIs. In the literature from other parts of the world, SSRIs have been compared among each other in terms of efficacy and acceptability, and one meta-analysis has reported advantage of escitalopram and sertraline over others.[69] Also, meta-analysis has been conducted using newer medications like venlafaxine and duloxetine.[7071] However, in the present meta-analysis, we did not calculate the effect sizes for individual comparisons between the drugs due to the lower number of studies.

The results of the present meta-analysis show that ECT to be superior to antidepressant medications. Studies from other parts of the world also suggest that ECTs is more effective than other medications and placebo in the treatment of depression.[7273] The efficacy of rTMS as an add-on treatment in this meta-analysis was found to be significant; however, it is important to note that the sample size in most of these trials have been very small. Recent meta-analysis of data from other parts of the globe suggests that active rTMS performs better than sham rTMS.[74] The NNT give an indication how well the treatment works when compared to placebo. Meta-analysis of antidepressant trials arising from other parts of the world suggest that NNTs in the range of 7 to 9.[75] The results from the present systematic review and meta-analysis suggest a comparatively low NNT for TCAs and moderate range NNT for MAOIs. This may be ascribed to lower number of studies and the differences in analyses between studies with SSRIs and TCAs (specifically intention-to-treat analysis not being conducted for older studies using TCAs).

One may wonder as to “What is the need of this kind of a review?”, when there is ample of evidence available from the studies across the world which have evaluated the efficacy of antidepressants in patients of depression. We thought that this kind of review is necessary to have a proper understanding of evidence from a country like India which is ethnically different from Western population and there may be variations in the efficacy of different antidepressants in Indian population compared those from the West. Second, in recent times, there has been lot of controversy with regard to approval of various medications by the Drug Controller General of India,[76] as to what standards and procedures should be followed prior to marketing of various medications in India. Third, meta-analysis is widely accepted as a robust method of summarizing data about treatment options. No meta-analytic study is currently available based on the efficacy data of antidepressants in Indian population. Fourth, for research to progress, it is important to look at the areas which require attention and priority so that the patients are helped. Keeping these in mind, we would try to draw some conclusions from this review and meta-analysis.

First, it can be concluded that there are limited data from Indian subcontinent evaluating the efficacy of antidepressants. Further, most of the trials have included relatively smaller sample sizes, have not be adequately powered to evaluate the efficacy of antidepressants or other treatment modalities and have been of 4-6 weeks durations. Further, the Jadad score for most of the studies ranges from 0 to 2 indicating the poor quality of conducting/reporting of these trials. Further, when one evaluates the Jadad score of various studies over the last 3-4 decades, it is evident that overall the methodology of conducting and reporting various trials has not improved significantly. As majority of the studies had a moderate to high risk of bias, this can impact the findings and portend toward positive result. All these characteristics of the studies weaken the evidence and application of the same in the routine clinical practice. Second, this review suggests that there are limited data about the efficacy of SSRIs and other newer antidepressants, especially from placebo controlled, trials arising from India. This is disconcerting given the fact that SSRIs (especially escitalopram, sertraline, and fluoxetine) are among the most commonly prescribed antidepressants in India.[777879] Third, surprisingly these is lack of data from India in special populations like children and adolescents, elderly and medically-ill persons, with only two comparisons included in this review evaluated antidepressants in medically-ill patients. Fourth, as discussed earlier that most of these trials have been limited to duration of 4 weeks; thus, the evidence generated cannot be generalized, because evidence from the West do suggests that some patients do have worsening of symptoms after initial response.[80] The shorter duration of trials also suggest that there is lack of data with respect to efficacy and effectiveness of antidepressants in the continuation phase of antidepressants. Taking all these limitations into consideration, it can be said that there is need to conduct large scale rigorous, pragmatic, preferably multicentric trials of longer duration (i.e. evaluating the efficacy in acute and continuation phase), which evaluate the whole range of symptoms of depression and include patients of all age groups.

Certain limitations should be taken into consideration while drawing interpretations from the results of this systematic review and meta-analysis. The metanalysis included those studies that were published and unpublished material were not included. Standardized mean differences could not be calculated from all the controlled studies due to the differences in reporting formats. Sensitivity analysis and meta-regression were not attempted as a part of the present work due to low number of studies for a particular meta-analytic comparison. Hence, variables like dosages of medications, gender differences, and so on were not explored in depth for the comparisons.

The present findings emphasize that much work needs to be done further for establishing the evidence base for efficacy of antidepressants in the Indian population. This is particularly regarding SSRIs and newer medications under development. Future studies should move beyond management of acute phase of depression with antidepressants and other treatment modalities and should look at the efficacy of antidepressants and other treatment modalities during the continuation and maintenance phase of treatment. Long-term, multicentric studies can give a broader and more global picture of the extent of usefulness of antidepressants across the country. Methodologically, rigorous studies, involving psychotherapy as a comparison measure, can help in the future for clinical decision making. Studies exploring the pharmacogenetic linkages of the efficacy can help in devising better treatment for individuals with particular genetic background. In fact, a few studies have been conducted in India exploring the pharmacogenetic aspects of antidepressants.[8182]