| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
British Medical Bulletin 57:161-178 (2001)
© 2001 Oxford University Press
Meta-analytical studies on new antidepressants
Neuroscience and Psychiatry Unit, School of Psychiatry and Behavioural Sciences, University of Manchester, Manchester, UK
 |
Abstract |
|---|
Top
Abstract
Methodological considerationsA systematic search found 108 meta-analyses of the use of antidepressants in depressive disorders. Defining newer antidepressants as those introduced since the early 1980s, 18 meta-analyses were selected as being informative about their relative efficacy and tolerability in comparative randomised controlled studies (RCTs).
Findings with higher confidence include: little difference in efficacy between most new and old antidepressants; superior efficacy of serotonin and noradrenaline re-uptake inhibitors (SNRIs) over selective serotonin re-uptake inhibitors (SSRIs); a slower onset of therapeutic action of fluoxetine over other SSRIs; a different side effect profile of SSRIs to TCAs with superior general tolerability of SSRIs over TCAs; poorer tolerability of fluvoxamine than other SSRIs in a within group comparison; no increased the risk of suicidal acts or ideation in fluoxetine compared with TCAs (or placebo) in low-risk patients.
Findings with a lower level of confidence include: greater efficacy of TCAs than SSRIs in in-patients; greater efficacy of amitriptyline than SSRIs; better tolerability of moclobemide than TCAs; no demonstrable difference in tolerability between SSRIs and TCAs in the elderly; no better tolerability of fluvoxamine than TCAs; better tolerability of dothiepin (dosulepin) than SSRIs; better tolerability of sertraline and greater frequency of agitation on fluoxetine than other SSRIs in a within group comparison.
In general, the meta-analyses were of uneven quality, as were the studies included, which limits the confidence in many of the results. Generalising from mostly short-term randomised controlled studies to clinical practice requires caution.
|
Methodological considerations |
|---|
Randomised controlled trials (RCTs) are the ‘gold standard’ for treatment trials1
but may still be prone to bias2 and are often too small to have adequate power to show a difference between treatments (type II statistical error). False positive results can also occur by chance (type I statistical error), particularly if subgroup or secondary analysis is used, a common feature of most antidepressant studies.
Meta-analysis aims to overcome some of these problems and provide an overview and quantification of treatment effects. To be meaningful, it must to be applied with the correct methodology in the context of a systematic review (see Henry & Wilson3 and Wilson & Henry4 for non-technical reviews). Unfortunately, the quality of meta-analyses and their interpretation remains problematic with limited standardisation of methods and analysis5. A proposal for improving the quality of meta-analyses of RCTs has been published with a useful checklist for their assessment6. Meta-analyses published as part of the Cochrane Collaboration have rigorous methodology and standardised reporting methods7.
|
Making sense of meta-analyses |
|---|
Combining results requires summary measure from each study in a form that allows statistical pooling. Different summary measures are used in the literature; the commonest are defined in Table 1 and briefly discussed here. The analysis of continuous data (such as rating scale scores) can be carried out using raw scores, but more commonly uses standardised differences or effect sizes (ES). These provide more statistical information than dichotomised data, but can be difficult to interpret. Effect sizes of 0.2, 0.5 and 0.8 have been suggested to indicate small, moderate and large effects, respectively8
, and more recently it has been suggested that the number needed to treat (NNT) can be estimated from the ES and control event rate9. For dichotomous data, such as responder/non-responder or dropout/continuation, the different analyses give different information. The risk ratio (relative risk, RR) is arguably more easily understood than the odds ratio (OR)10 and the reciprocal of the absolute risk difference (RD), known as the number needed to treat (NNT) has become a popular way of interpreting the clinical significance of a finding. The 95% confidence interval (95% CI) provides information about the likely range within which the actual value lies.
|
Combining (pooling) summary outcome measures in meta-analysis traditionally weights studies by the inverse of the variance in each study which tends to give greater weight to larger studies in which the estimate is more precise. The issue of statistical heterogeneity between individual studies is beyond the scope of this article but a ‘random effects’ model is often employed to take account of this (as opposed to a traditional ‘fixed effects’ model). Increasingly, an approach using Bayesian techniques is being employed as it does not require large sample assumptions and allows the effect of predictive variables to be examined (meta-regression).
|
Scope and methodology |
|---|
This article is an overview of meta-analyses examining the use of newer antidepressants in the treatment of depression. ‘Newer’ antidepressants are taken to mean currently available antidepressants marketed since the early 1980s. The review is confined to meta-analyses of comparative studies of antidepressants as it is the relative advantages and disadvantages of antidepressants that is of most interest. Studies simply comparing a drug against placebo and single arm meta-analyses of non-comparative studies are not included.
Meta-analysis were identified from a computerised search to March 2001 of MEDLINE, EMBASE, PSYCHInfo and the Cochrane Library using the terms ‘meta-analysis’ OR ‘systematic review’ combined with ‘antidepressant’ OR ‘anti-depressant’ OR ‘antidepressive’ OR ‘anti-depressive’. This was supplemented by hand searching the references for other reviews. A total of 108 meta-analyses were identified, but many were of older antidepressants only, considered the same comparisons, were of subgroups of studies or have been superseded. Only the most recent or informative are discussed (Table 2) with some others mentioned briefly for comparison purposes. A list of the other meta-analyses is available on request.
|
|
Meta-analyses of individual antidepressant/class comparisons |
|---|
The potential number of comparisons is large although in practice most newer drugs have been compared with tricyclic antidepressants (TCAs) or more recently with selective serotonin re-uptake inhibitors (SSRIs). Many meta-analyses consider a variety of drugs, often grouped differently so the results are considered by drug/drug class drawing on the results of the meta-analyses summarised in Table 2.
Lofepramine
Lofepramine is a tricyclic antidepressant which inhibits the re-uptake of noradrenaline, differing from the older tricyclics in being safer in overdose. Kerihuel and Dreyfus20 used drug company data to obtain evaluable data from a reasonable number of studies. Interpretation is mainly limited by the quality of the studies which were small, a third were very short by current standards (3–4 weeks) and many were pre-1980 with a lack of information on important quality issues. The authors addressed the latter point by checking 34% of the 2040 original case report forms and found no major deviation from the published data. Evaluation of efficacy was based on global improvement and 24 studies (1714 patients) provided data, 16 against amitriptyline or imipramine. More patients improved on lofepramine compared with other TCAs (64% versus 56%; OR 1.4, 95% CI 1.2–1.8). Doses of the comparator TCAs were within the range 75–150 mg. The advantage was less marked and non-significant against amitriptyline and imipramine. Tolerability in terms of total number of dropouts was only given as percentages which were numerically similar for lofepramine and comparators. However, using data available from 14 studies, significantly fewer patients on lofepramine experienced side-effects (53% versus 64%; OR 0.6, 95% CI 0.5–0.7) with individual studies tending to report less dry mouth, dizziness and sedation.
No other meta-analyses provide useful information about lofepramine as Davis et al16 (Table 2) only included 4 studies (160 patients). Within the limits of the data available, meta-analysis supports at least equal efficacy for lofepramine compared with standard TCAs but with better tolerability. Comparison with other newer antidepressants is extremely limited, only 2 studies are available comparing lofepramine with SSRIs14 which does not allow conclusions to be drawn.
Amoxapine
Amoxapine is a cyclic antidepressant which inhibits the re-uptake of noradrenaline as well as having antagonist actions at dopamine and serotonin receptors. There are no good quality meta-analyses available but Davis et al16 (Table 2) report a non-significant efficacy advantage to amoxapine over older TCAs (79% responders versus 73%) in 19 studies (784 patients). It is difficult to draw any conclusions from this.
Reversible inhibitors of monoamine oxidase A (RIMAs)
This class includes moclobemide and brofaromine but only the former has been marketed. Lotufo-Neto et al21 (Table 2) pooled results from 6 brofaromine studies (659 patients) compared with tranylcypromine or imipramine in all but one case, with a high proportion of in-patients and TCA-resistant patients. With 58.6% of brofaromine patients responding, it was non-significantly favoured over comparators (RD 4.8%, 95% CI –5.9 to 15.2). In 28 studies (3220 patients) comparing moclobemide with TCAs (mostly imipramine, clomipramine and amitriptyline), 54.5% responded to moclobemide with no difference between drugs (RD 1.8%, 95% CI –2.7 to 6.3). Compared with traditional MAOIs (4 studies, 357 patients), 68.1% of moclobemide patients responded, non-significantly fewer than on MAOIs (RD –5.8%, to –22.5 to 10.9). In 7 studies (561 patients) against SSRIs, moclobemide had a response rate of 58.1% and a non-significant efficacy advantage (RD 6.5%, –1.3 to 14.3). Finally, moclobemide had a response rate of 64.7% and was more effective than a group of 7 studies (655 patients) in which it was compared with mianserin, maprotiline or amineptine (RD 10.6%, 0.8 to 20.4, NNT 10). Tolerability was not directly analysed but other reviews were cited that reported the most common side-effects (both around 20%) were insomnia and gastrointestinal symptoms such as nausea and diarrhoea with less drowsiness, anti-muscarinic and sexual side-effects than TCAs. Williams et al27 (Table 2) also found no difference efficacy comparing RIMAs with TCAs in a smaller sample of 12 studies (1967 patients) but reported fewer side effect dropouts than older TCAs (5% versus 11%, RD –6%, –7 to –3, NNT 17) although total dropouts did not differ (26% versus 27%, RD –1%, –5 to 2).
There appears to be fairly good evidence that moclobemide has comparable efficacy to older TCAs and SSRIs and is probably better tolerated than TCAs.
Selective serotonin re-uptake inhibitors (SSRIs) The SSRIs (citalopram, fluoxetine, fluvoxamine, paroxetine and sertraline) are the first antidepressants since the TCAs to make a major impact on prescribing practice. There are a considerable number of meta-analyses available so I will only present representative results and highlight areas of uncertainty.
SSRIs compared with other antidepressants
Anderson11 (Table 2) found no difference in efficacy between SSRIs and TCAs in 102 studies (10,553 patients). The effect size was –0.03 (–0.09 to 0.03, negative value favouring TCAs) in agreement with Geddes et al19 (ES 0.030, –0.018 to 0.092, positive value favouring TCAs) in an analysis of 67 studies (Table 2). Geddes et al19 also grouped SSRIs and what they termed ‘related drugs’, venlafaxine and nefazodone, and compared them with all comparators in 98 studies (9554 patients) finding a very similar result (ES 0.035, –0.006 to 0.076). Comparisons between individual SSRIs and TCAs/comparators revealed no differences in efficacy apart from an advantage to amitriptyline found by Anderson11 (ES –0.14, –0.25 to –0.03) but not Geddes et al19 (ES 0.057, –0.027 to 0.14). This difference is likely to be accounted for both by the different method of analysis (change scores versus final score) as well as the different number of studies included (30 in the former versus 23 in the latter). A further complication is that publication bias cannot be ruled out11, although other explanations such as study setting are also possible (see below) which may also explain the statistical heterogeneity that was found between studies. In support of the finding by Anderson11, Barbui and Hotopf13 (Table 2) examined the efficacy of amitriptyline compared with other antidepressants in 99 studies (6745 patients) and concluded that amitriptyline was slightly more effective than comparators (RR 1.12, 95% CI 1.01–1.24, NNT 40) with no evidence of publication bias. Compared with SSRIs, amitriptyline was more effective in the pooled effect size (21 studies; 0.106, 0.02 to 0.19) but the pooled odds ratio, although favouring amitriptyline, was not significant (17 studies; 1.14, 0.94 to 1.38).
Anderson11 also examined lower (75–100 mg) and higher (150 mg and above) TCA dose studies separately. While higher dose studies showed a slightly, but non-significantly, greater advantage to TCAs (ES –0.08, –0.21 to 0.05) compared with those using lower doses (ES 0.025, –0.04 to 0.09) there appears little support for the contention that TCAs at doses between 75–100 mg are ineffective.
There is consistent evidence that SSRIs as a group are better tolerated than TCAs and possibly second generation antidepressants. Anderson11 reported results from 95 studies (10,553 patients) and found that 27% of patients discontinued treatment on SSRIs compared with 31.4% on TCAs (RD –3.9%, –5.6 to –2.2, NNT 26). Dropouts due to side-effects appeared to account for this difference (12.5% versus 17.3%, RD –3.1%, –4.7 to –1.5, NNT 33) with no difference in dropouts due to inefficacy (6.3% versus 6%). Geddes et al19, Williams et al27 and Barbui et al14 (Table 2) found the same results. The last study also compared SSRIs with second generation antidepressants (e.g. maprotiline, mianserin, trazodone; 25 studies, 2246 patients) and divided TCAs into old (amitriptyline and imipramine; 77 studies, 9197 patients) and new (other TCAs; 33 studies, 4782 patients). With an odds ratio > 1 indicating an advantage to SSRIs, essentially similar results were found for each comparison for dropouts overall (old TCAs: OR 1.22, 1.11 to 1.34, new TCAs: OR 1.20, 1.05 to 1.38, second generation: OR 1.17, 0.95 to 1.43) and due to side-effects (old TCAs: OR 1.61, 1.40 to 1.86, new TCAs: OR 1.28, 1.02 to 1.59, second generation: OR 1.30, 0.91 to 1.85). The results for second generation antidepressants were very similar in size to those for the TCAs although non-significant, and an advantage to SSRIs cannot be ruled out. Of interest, there were fewer dropouts due to inefficacy on TCAs than SSRIs, only significant for old TCAs (OR 0.82, 0.67 to 1.00).
Comparison involving individual drugs largely follow the same pattern with two exceptions. Anderson11 found that fluvoxamine (25 studies) did not have fewer dropouts compared with TCAs either overall (RR 1.10, 0.86 to 1.19) or due to side-effects (RR 0.94, 0.72 to 1.22: an RR < 1 indicates an advantage to SSRIs). Dothiepin (dosulepin, 8 studies), alone of the TCAs, resulted in fewer dropouts than SSRIs (total: RR 1.24, 0.95 to 1.62, side effect: RR 2.64, 1.50 to 4.63, NNH 12), a result also found by Barbui et al14.
The earlier meta-analysis by the Canadian Co-ordinating Office for Health Technology Assessment26,29 did not find any difference in dropout rates between SSRIs and TCAs, presumably because many fewer studies were analysed. Of interest are the results for specific side-effects. Compared with TCAs, SSRIs caused significantly more nausea (48 studies: RD 10.3%, 7.3 to 13.3, NNT 10), diarrhoea (15 studies: RD 9%, 4 to 14, NNT 12), agitation (11 studies: RD 6%, 1 to 10, NNT 17), anorexia (11 studies: RD 5%, 0.6 to 9, NNT 20), insomnia (32 studies: RD 4%, 1 to 6, NNT 25), nervousness (14 studies: RD 4%, 0.6 to 7.4, NNT 25) and anxiety (17 studies: RD 3%, 0.8 to 5.5, NNT 34). On the other hand, TCAs caused more dry mouth (56 studies: RD –28.1%, –34.9 to –24.8, NNT 4), constipation (49 studies: RD –11%, –14 to –8, NNT 10), dizziness (37 studies: RD –9%, –12 to –6, NNT 12), sweating (27 studies: RD –3.6%, –6.7 to –0.4, NNT 28) and blurred vision (19 studies: RD –2.8%, –5.4 to –0.02, NNT 36). Williams et al27 also mention more headache on SSRIs compared with TCAs (15% versus 11%) and more tremors (11% versus 7%) and urinary disturbance (8% versus 3%) on TCAs compared with SSRIs.
SSRI are increasingly replacing TCAs as the standard comparators for newer antidepressants and these are considered separately below. In summary, SSRIs have proven to be generally as effective as older antidepressants apart from in comparison with amitriptyline against which they may be marginally less effective. Tolerability is better than with TCAs and probably second generation antidepressants although the magnitude is modest in short-term studies. The profile of SSRIs is significantly different from TCAs with less antimuscarinic but more gastrointestinal and stimulatory side-effects.
SSRIs compared with each other
Edwards and Anderson17 (Table 2) identified 20 studies (3283 patients) comparing two SSRIs and determined the efficacy of individual SSRIs against the rest; 15 of the studies (2542 patients) were against fluoxetine. The efficacy at the end of study was non-significantly less for fluoxetine (ES –0.06, –0.14 to 0.01) but significantly so at 2–3 weeks (ES –0.11, –0.18 to –0.03), suggesting a slower onset of action compared with other SSRIs. No other efficacy differences were found between other SSRIs. Of interest, Williams et al27 reported that slightly fewer patients on fluoxetine responded than on other SSRIs (RR 0.9, 0.8 to –1.0) but only 8 studies were included in the analysis. Trindade and Menon26 found no difference between SSRIs based on an analysis of 10 studies.
The comparative tolerability of SSRIs was examined in 19 studies (2999 patients)17. The overall discontinuation rate was 25.6% with 10.8% attributed to side-effects. Fluvoxamine (5 studies, 594 patients) had significantly poorer tolerability than other SSRIs with more overall (RD 7.6%, 0.7 to 14.5, NNH 14) and side-effect dropouts (RD 6.5%, –1.5 to 14.6, NNH 16). In contrast, sertraline had slightly fewer dropouts due to side-effects than other SSRIs (RD –5.5, –10.9 to –0.01, NNT 19). Most individual side-effects were reported in only a minority of studies. An exception was nausea which occurred in 21.3% of 2706 patients in 17 studies with no difference between SSRIs. Agitation occurred more often with fluoxetine than other SSRIs (RD 3.3%, 0.4 to 6.3, NNT 30) but was reported in only 6 studies (1001 patients) so publication bias cannot be excluded, particularly as most studies were sponsored by its competitors.
In summary, it appears likely that fluoxetine has a slower onset of action than other SSRIs in the dosing schedules used in these studies (20 mg for the first week) and that fluvoxamine is less well tolerated (compatible with the results of comparisons with TCAs).
Serotonin and noradrenaline re-uptake inhibitors (SNRIs) Venlafaxine and milnacipran inhibit the re-uptake of both serotonin and noradrenaline but, unlike TCAs, lack significant affinity for other receptors. Venlafaxine is widely marketed and there have been claims for its superior efficacy over SSRIs.
Williams et al27 grouped together SNRIs and mirtazapine (see below) and reported no difference in efficacy in 6 studies (1400 patients) against TCAs. However, in 3 studies (449 patients) against trazodone they were more effective (RR 1.2, 1 to 1.4). Puech et al24 examined the efficacy of milnacipran compared with imipramine in 6 studies (726 patients) and found equal efficacy (RD –1%, –8.1 to 6). Anderson12 pooled 22 studies (3925 patients) and found that more patients on SNRIs responded compared with those on SSRIs (66% versus 60%, RR 1.09, 1.02 to 1.17, NNT 18) with similar results for the 17 venlafaxine studies (3021 patients, RR 1.09, 1.02 to 1.17) and the 5 milnacipran studies (904 patients, RR 1.09, 0.90 to 1.33). In a study examining remission, Thase et al25 pooled data from 8 studies (1599 patients), 5 against fluoxetine, and found that 45% of venlafaxine-treated patients remitted compared with 35% of SSRI-treated patients (OR 1.5, 1.3 to 1.9). The numbers of patients dropping out due to side-effects were numerically, but non-significantly, greater on venlafaxine (9% versus 7%) but slightly more patients on SSRIs dropped out due to inefficacy (6% versus 4%, P < 0.05).
These results provide some support that dual re-uptake inhibition confers greater efficacy than inhibition of serotonin re-uptake alone.
Mirtazapine
Mirtazapine is related to mianserin and enhances both serotonin and noradrenaline function through antagonism of presynaptic receptors. Zivkov et al28 reported on 5 studies (814 patients) comparing mirtazapine with amitriptyline. The weighted mean difference in 17-item Hamilton Depression Rating Scale points was 0.58 (–0.57 to 1.73) with 63.9% of amitriptyline patients responding compared with 61.3% of mirtazapine patients (RR 0.95, 0.85 to 1.05). This suggests little difference in efficacy but judgement needs to be reserved over publication bias given that only half of available TCA studies are included in this analysis. A meta-analysis of 3 studies (653 patients) comparing mirtazapine with SSRIs12 found response rates of 69.9% versus 65.4% (RR 1.11, 0.87 to 1.42) which, although not significant, is very similar in size to that seen with the SNRIs. Mirtazapine may have a more rapid onset of action than SSRIs as significantly more patients had responded at 2 weeks (34% versus 24%, RR 1.38, 1.08 to 1.76, NNT 11). However, a faster onset of action was not apparent in the studies against amitriptyline28 and the small numbers make the results vulnerable to chance.
Few data on tolerability are available from meta-analysis. Zivkov et al28 found that about equal numbers dropped out on mirtazapine and amitriptyline due to lack of efficacy/adverse events or both (13.3% versus 13%). However, this might hide a difference in reasons for dropouts as Stahl et al30, in a smaller meta-analysis of 4 placebo controlled studies (580 patients), found that non-significantly fewer patients dropped out due to side-effects on mirtazapine (10.1% versus 17.2%) but significantly more due to inefficacy (15.1% versus 7.4%, P < 0.05). The pattern of side-effects confirmed a lack of antimuscarinic and stimulant effects but dry mouth, sedation and weight gain occurred significantly more than with placebo.
|
Meta-analyses and other aspects |
|---|
Clinical variables Study setting
In-patients may be a group with more severe depression in which differences between antidepressants might be more pronounced. Anderson11
found that TCAs were more effective than SSRIs in the 25 studies (1377 patients) in in-patients (ES –0.23, –0.40 to –0.05) whereas no advantage was seen in out-patients or primary care. However, Geddes et al19 in 23 studies (1347 patients) found the advantage to be smaller and non-significant (0.1, –0.07 to 0.27) using slightly different methodology and Trindade and Menon26 noted a similar finding. In comparisons between SNRIs and SSRIs, Anderson12 found a greater advantage to SNRIs in in-patients (5 studies, 1.24, 1.08 to 1.41, NNT 8) than out-patients (13 studies, RR 1.08, 0.99 to 1.18) and no difference in general practice patients (2 studies, 0.97, 0.88 to 1.08). However, the general practice result may have been explained by the low dose of venlafaxine (75 mg).
Most treatment of depression occurs in primary care but there are relative few studies available in this setting. Mulrow et al23 (Table 2) examined 28 primary care studies (5940 patients) and concluded that newer agents had equal efficacy to TCAs in this setting (RR 1.0, 0.9 to 1.1) but dropouts due to side-effects were fewer (8% versus 13%, RD –4, –7 to 0). Anderson11, comparing SSRIs with TCAs in primary care studies, found fewer total (9 studies, RR 0.78, 0.68 to 0.90) and side-effect dropouts (8 studies, 0.75, 0.61 to 0.85).
The available meta-analyses do not give consistent findings but differences in efficacy may be more important in in-patient than community settings.
Elderly patients
Two studies which considered the elderly in a subgroup analysis concluded that there were no differences in efficacy between older and newer antidepressants in 16 studies27 or between SSRIs and TCAs in 11 studies (1207 patients)11, similar to the overall analysis. In contrast tolerability, which was better on SSRIs than TCAs in the whole group analysis, did not differ between the two groups of drugs. The dropout rate was high in the elderly with 36.5% of patients stopping treatment, 21.8% due to side-effects, in comparative SSRI-TCAs studies and there was a very small non-significant difference in side-effect dropouts (21.3% versus 22.4%, RR 0.91, 0.74 to 1.11)11. A number of meta-analyses have examined studies in elderly patients alone, that by Mittmann et al22 being probably the most informative. Efficacy comparisons showed SSRIs to be more effective than second generation antidepressants (2 studies, 66 patients; RD –36.9%, –65.4 to –8.4) and non-significantly favoured over TCAs (4 studies, 423 patients; RD 3.3%, –6.3 to 12.8). Occurrence of side-effects was non-significantly less frequent on second generation antidepressants compared with RIMAs (2 studies, 105 patients; –11.8%, –30.5 to 6.8) and on SSRIs compared with second generation antidepressants (3 studies, 134 patients; –7.5%, –34.3 to 19.3) and TCAs (4 studies, 418 patients; –9.6%, –18.7 to 1.5) with a similar picture for total dropouts. Menting31, in contrast, concluded that SSRIs were better tolerated than TCAs if the analysis was limited to the 4 better quality studies. However, the sensitivity of results to differences in study inclusion makes interpretation difficult. Therefore, in the elderly the evidence suggests that antidepressants are more poorly tolerated in general than in younger patients with no confidence that any one group of drugs is superior.
Antidepressants and suicide
There has been concern that SSRIs may increase the risk of suicidal actions in depressed patients, possibly as a consequence of agitation which occurs more often than with TCAs26. In order to address this, Beasley et al15 (Table 2) analysed studies of fluoxetine compared with TCAs (12 studies, 1451 patients). They found non-significantly more patients on fluoxetine made suicidal acts (0.7% versus 0.4%, RD 0.3%, –0.4 to 1.1) but fewer experienced the emergence of suicidal ideas (1.7% versus 3.6%, RD –1.8, –4.0 to 0.4). Separate comparisons with placebo showed no increased risk for either group of antidepressants and a greater improvement in suicidal ideation on fluoxetine compared with placebo, presumably due to greater alleviation of depression. This study is important evidence against fluoxetine causing a general increase in suicidal risk, but does not exclude a role in individual cases or in patients at high risk of suicide.
Pharmacology in relation to efficacy
There is a lack of empirical evidence to link the acute pharmacology of antidepressants and efficacy. In an attempt to explore this using meta-regression analysis, Freemantle et al18 (Table 2) explored the possible contribution of combinations of pharmacological actions to efficacy compared with serotonin re-uptake inhibition alone but were unable to demonstrate an effect. This probably partly reflects limitations in our understanding of the in vivo pharmacology of many older antidepressants in humans. The results with SNRIs compared with SSRIs (see above) are at present the best evidence that dual action at serotonin and noradrenaline transporters is better than a single action on serotonin.
|
Conclusions |
|---|
The available meta-analyses provide the best summary of a bewildering variety of studies comparing antidepressants; the main findings are given in Table 3. They are, however, beset with limitations related to the number and quality of individual RCTs as well as variations in grouping and inclusion criteria. These factors limit the confidence with which many of the findings can be viewed. In addition, the results are from short-term RCTs and caution is required in extrapolating them to clinical practice in which longer term treatment is recommended32
; however, the first 6–8 weeks is a crucial time and necessary step in treatment. These analyses do provide a counterbalance to promotional claims for individual antidepressants although the loss of detail necessary for meta-analysis necessarily excludes potentially important factors in the treatment of individual patients. In the future, effectiveness studies will hopefully provide additional guidance, but for the present these results are a starting point in weighing the evidence as to choice of antidepressant. The need to tailor treatment for individual patients, however, cautions against dogmatism.
|
|
Acknowledgements |
|---|
Wyeth Laboratories provided additional details on venlafaxine studies and financial support for the venlafaxine analyses.
Declaration of interest: the author has received research funding, honoraria for speaking and advice and support for attending scientific meetings from a number of pharmaceutical companies producing antidepressants discussed in this article.
|
Footnotes |
|---|
Correspondence to: Dr Ian M Anderson, Neuroscience and Psychiatry Unit, Room G809, Stopford Building, University of Manchester, Oxford Road, Manchester M13 9PT, UK
|
References |
|---|
- Barton S. Which clinical studies provide the best evidence? The best RCT still trumps the best observational study. BMJ 2000; 321: 255–6
[Free Full Text] - Stewart LA, Parmar MKB. Bias in the analysis and reporting of randomized controlled trials. Int J Technol Assess Health Care 1996; 12: 264–75[ISI][Medline]
- Henry DA, Wilson A. Meta-analysis. Part 1: an assessment of its aims, validity and reliability. Med J Aust 1992; 156: 31–8[ISI][Medline]
- Wilson A, Henry DA. Meta-analysis. Part 2: assessing the quality of published meta-analyses. Med J Aust 1992; 156: 173–87[ISI][Medline]
- Walker A. Meta-style and expert review. Lancet 1999; 354: 1834–5[ISI][Medline]
- Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUORUM statement. Lancet 1999; 354: 1896–900[ISI][Medline]
- Bero L, Rennie D. The Cochrane Collaboration: preparing, maintaining and disseminating systematic reviews of the effects of health care. JAMA 1995; 274: 1935–8[ISI][Medline]
- Cohen J. Statistical Power Analysis for the Behavioral Sciences. Orlando, FL: Academic Press, 1977
- Furukawa TA. From effect size to number needed to treat. Lancet 1999; 353: 1680[ISI][Medline]
- Sinclair JC, Bracken MB. Clinically useful measures of effect in binary analyses of randomized trials. J Clin Epidemiol 1994; 47: 881–9[ISI][Medline]
- Anderson IM. Selective serotonin re-uptake inhibitors versus tricyclic antidepressants: a meta-analysis of efficacy and tolerability. J Affect Dis 2000; 58: 19–36[ISI][Medline]
- Anderson IM. Meta-analyses of antidepressant drugs: selectivity versus multiplicity. In: den Boer JA, Westenberg HGM. (eds) Focus on Psychiatry; Antidepressants: Selectivity or Multiplicity? Amsterdam: Syn-Thesis, 2001; In press
- Barbui C, Hotopf M. Amitriptyline v. the rest: still the leading antidepressant after 40 years of randomised controlled trials. Br J Psychiatry 2001; 178: 129–44
[Abstract/Free Full Text] - Barbui C, Hotopf M, Freemantle N et al. Selective serotonin re-uptake inhibitors versus tricyclic and heterocyclic antidepressants: comparison of drug adherence (Cochrane Review). The Cochrane Library. 1. Oxford, Update Software. 2001
- Beasley CM, Dornseif BE, Bosomworth JC et al. Fluoxetine and suicide: a meta-analysis of controlled trials of treatment for depression. BMJ 1991; 303: 685–92[ISI][Medline]
- Davis JM, Wang Z, Janicak PG. A quantitative analysis of clinical drug trials for the treatment of affective disorders. Psychopharmacol Bull 1993; 29: 175–81[ISI][Medline]
- Edwards JG, Anderson I. Systematic review and guide to selection of selective serotonin re-uptake inhibitors. Drugs 1999; 57: 507–33[ISI][Medline]
- Freemantle N, Anderson IM, Young P. Predictive value of pharmacological activity for the relative efficacy of antidepressant drugs. Meta-regression analysis. Br J Psychiatry 2000; 177: 292–302
[Abstract/Free Full Text] - Geddes JR, Freemantle N, Mason J, Eccles MP, Boynton J. SSRIs versus other antidepressants for depressive disorder (Cochrane Review). The Cochrane Library 1. Oxford, Update Publishers. 2001.
- Kerihuel JC, Dreyfus JF. Meta-analyses of the efficacy and tolerability of the tricyclic antidepressant lofepramine. J Int Med Res 1991; 19: 183–201[ISI][Medline]
- Lotufo-Neto F, Trivedi M, Thase ME. Meta-analysis of the reversible inhibitors of monoamine oxidase type A moclobemide and brofaromine for the treatment of depression. Neuropsychopharmacology 1999; 20: 226–47[ISI][Medline]
- Mittmann N, Herrmann N, Einarson TR et al. The efficacy, safety and tolerability of antidepressants in late life depression: a meta-analysis. J Affect Dis 1997; 46: 191–217[ISI][Medline]
- Mulrow CD, Williams JWJ, Chiquette E et al. Efficacy of newer medications for treating depression in primary care patients. Am J Med 2000; 108: 54–64[ISI][Medline]
- Puech A, Montgomery SA, Prost JF, Solles A, Briley M. Milnacipran, a new serotonin and noradrenaline re-uptake inhibitor: an overview of its antidepressant activity and clinical tolerability. Int Clin Psychopharmacol 1997; 12: 99–108[ISI][Medline]
- Thase ME, Entsuah AR, Rudolph RL. Remission rates during treatment with venlafaxine or selective serotonin re-uptake inhibitors. Br J Psychiatry 2001; 178: 234–41
[Abstract/Free Full Text] - Trindade E, Menon D. Selective serotonin re-uptake inhibitors (SSRIs) for major depression. Part 1. Evaluation of the clinical literature. Ottowa: Canadian Coordinating Office for Health Technology Assessment, 1997
- Williams JWJ, Mulrow CD, Chiquette E, Noel PH, Aguilar C, Cornell J. A systematic review of newer pharmacotherapies for depression in adults: evidence report summary. Ann Intern Med 2000; 132: 743–56
[Abstract/Free Full Text] - Zivkov M, Roes KCB, Pols AG. Efficacy of Org 3770 (mirtazapine) vs amitriptyline in patients with major depressive disorder: a meta-analysis. Hum Psychopharmacol 1995; 10: S135–S145
- Trindade E, Menon D, Topfer L-A, Coloma C. Adverse effects associated with selective serotonin re-uptake inhibitors and tricyclic antidepressants: A meta-analysis. Can Med Assoc J 1998; 159: 1245–52[Abstract]
- Stahl S, Zivkov M, Reimitz PE, Panagides J, Hoff W. Meta-analysis of randomized, double-blind, placebo-controlled, efficacy and safety studies of mirtazapine versus amitriptyline in major depression. Acta Psychiatr Scand 1997; 96 (Suppl 391): 22–30
- Menting JE, Honig A, Verhey FR et al. Selective serotonin re-uptake inhibitors (SSRIs) in the treatment of elderly depressed patients: a qualitative analysis of the literature on their efficacy and side-effects. Int Clin Psychopharmacol 1996; 11: 165–75[ISI][Medline]
- Anderson IM, Nutt DJ, Deakin JFW. Evidence-based guidelines for treating depressive disorders with antidepressants: a revision of the 1993 British Association for Psychopharmacology guidelines. J Psychopharmacol 2000; 14: 3–20[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  I. Anderson, I. Ferrier, R. Baldwin, P. Cowen, L Howard, G Lewis, K Matthews, R. McAllister-Williams, R. Peveler, J Scott, et al. Evidence-based guidelines for treating depressive disorders with antidepressants: A revision of the 2000 British Association for Psychopharmacology guidelines J Psychopharmacol, June 1, 2008; 22(4): 343 - 396. [Abstract] [PDF]
|
|
|
|
 |
|
 M. Olfson, A. K. Das, M. J. Gameroff, D. Pilowsky, A. Feder, R. Gross, R. Lantigua, S. Shea, and M. M. Weissman Bipolar Depression in a Low-Income Primary Care Clinic Am J Psychiatry, November 1, 2005; 162(11): 2146 - 2151. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. A. Hansen, G. Gartlehner, K. N. Lohr, B. N. Gaynes, and T. S. Carey Efficacy and Safety of Second-Generation Antidepressants in the Treatment of Major Depressive Disorder Ann Intern Med, September 20, 2005; 143(6): 415 - 426. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Arroll, S. Macgillivray, S. Ogston, I. Reid, F. Sullivan, B. Williams, and I. Crombie Efficacy and Tolerability of Tricyclic Antidepressants and SSRIs Compared With Placebo for Treatment of Depression in Primary Care: A Meta-Analysis Ann. Fam. Med, September 1, 2005; 3(5): 449 - 456. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Trick, N. Stanley, U. Rigney, and I. Hindmarch A Double-Blind, Randomized, 26-Week Study Comparing the Cognitive and Psychomotor Effects and Efficacy of 75 mg (37.5 mg b.i.d.) Venlafaxine and 75 mg (25 mg Mane, 50 mg Nocte) Dothiepin in Elderly Patients with Moderate Major Depression Being Treated in General Practice J Psychopharmacol, June 1, 2004; 18(2): 205 - 214. [Abstract] [PDF]
|
|
|
|
 |
|
 M. Dwight-Johnson, I. T. Lagomasino, and G. M. Simpson Psychopharmacology: Underuse of Evidence-Based Pharmacotherapies for Affective Disorders Psychiatr Serv, August 1, 2003; 54(8): 1076 - 1078. [Full Text] [PDF]
|
|
|
|
 |
|
 R. Hubbard, P. Farrington, C. Smith, L. Smeeth, and A. Tattersfield Exposure to Tricyclic and Selective Serotonin Reuptake Inhibitor Antidepressants and the Risk of Hip Fracture Am. J. Epidemiol., July 1, 2003; 158(1): 77 - 84. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. Goodwin Evidence-Based Guidelines for Treating Bipolar Disorder: Recommendations from the British Association for Psychopharmacology J Psychopharmacol, June 1, 2003; 17(2): 149 - 173. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||