Although many systematic reviews and meta-analyses suggest that the efficacy of currently available antidepressants is similar for most patients in primary care and psychiatric clinics, there is an ongoing debate as to whether some antidepressants are slightly better than others.
Uncertainty about what constitutes a “clinically significant difference,” selective analyses, and drug company-sponsored issues add to the difficulty of interpreting the data. In a meta-regression analysis that included 105 comparative RCTs, the investigators did not find pharmacological predictors of efficacy, but the classification of drugs is questionable; drug company sponsorship, although not statistically significant, was the strongest influencing factor.
Older drugs
Early studies by the Danish University Antidepressant Group showed that clomipramine 150 mg/d was more effective than citalopram 40 mg/d and paroxetine 30 mg/d. However, there were some problems with these studies, including the inpatient status of patients and the effect of the sleep effect of clomipramine on study outcomes. In a meta-analysis that included 100 studies, amitriptyline was slightly more effective than other tricyclic antidepressants (TCAs) and SSRIs in hospitalized depressed patients, with a number needed to treat (NNT=24), but not in nonhospitalized patients; hospitalization may reflect higher depression severity, but other factors such as type of depression and suicide risk may be equally relevant.
A meta-analysis of MAOIs showed that phenelzine and isocarbohydrazide were less effective than promethazine in treating hospitalized patients; 10 studies were included in this meta-analysis, with a 14C20% difference in response rates and a 5-7 number needed to treat. however, the quality of these studies was mixed. Another meta-analysis involving reversible monoamine oxidase A inhibitors (RIMA) suggested that the efficacy of morclobemide did not differ from that of promethazine and clomipramine in hospitalized patients, including those with more severe depression or with psychotic symptoms.
New drugs
The pharmacology of newer antidepressants is more specific. A current point of interest is the efficacy of dual-channel SNRIs (venlafaxine, duloxetine, and milnacipran) compared with SSRIs. There are two meta-analyses comparing venlafaxine with the SSRIs class of drugs, and again the conclusions differ due to differences in study entry criteria.
In contrast, the analysis by Weinmann et al. that included 17 studies used more stringent exclusion criteria, and venlafaxine was significantly better than SSRIs on only two of the four regression indicators, namely treatment response rate (NNT=27) and change in depression scale scores, with no significant differences in remission and final depression scores. Neither of these studies found evidence of publication bias.
However, given the dose-response relationship of venlafaxine and the dual mechanism of action presented only at higher doses (>150 mg), the dose of this drug in the study needs to be considered. The efficacy of duloxetine was not superior to that of SSRIs, and the proportion of patients discontinuing treatment due to adverse effects was even higher than that of venlafaxine (OR1.79).
A pooled analysis of two studies comparing venlafaxine and duloxetine showed no significant difference in efficacy, but the response rate was numerically higher with venlafaxine, while duloxetine did not meet the prespecified criteria for noninferiority. A meta-analysis of minazepam versus SSRIs showed.
In another meta-analysis, the investigators compared the efficacy of multiple mechanism drugs (SNRIs, mirtazapine, mianserin, and moclobemide) with SSRIs. The results suggested that the former was slightly better than the latter (93 studies, response rate 63.6% vs. 59.3%, NNT=24). The effect values of the above drugs were similar, with the exception of duloxetine: this drug did not differ from SSRIs. However, the above results seem to be largely contributed by the results of venlafaxine related studies. The results of the comparison of mirtazapine with SSRIs are inconclusive.
The Escitalopram Phenomenon
In 2009, Cipriani and colleagues published a widely cited study that conducted a network meta-analysis of pooled data on RCTs regarding 12 novel antidepressants. The results showed that mirtazapine, escitalopram, venlafaxine, and sertraline had higher rates of treatment response (50% reduction in clinical scores) than duloxetine, fluoxetine, fluvoxamine, paroxetine, and reboxetine, while reboxetine was significantly less effective than all other 11 antidepressants.
Using fluoxetine as a standard control, sertraline, escitalopram, and venlafaxine, the smaller the value, the better the efficacy. In addition, the treatment discontinuation rates of escitalopram and sertraline were significantly lower than those of duloxetine, fluvoxamine, paroxetine, reboxetine, and venlafaxine. Taken together, sertraline and escitalopram had the most favorable efficacy and tolerability.
A follow-up meta-analysis evaluated 234 studies, 118 of which were head-to-head studies. Among these head-to-head studies, the most significant treatment response rate comparisons included escitalopram versus citalopram (OR1.49), sertraline versus fluoxetine (OR1.42), and venlafaxine versus fluoxetine (OR1.47).
Reboxetine
Following the publication of the previous guideline, a meta-analysis that included all data related to the NE reuptake inhibitor reboxetine (also including unpublished studies) showed that it was not an effective antidepressant: remission rates were not better than placebo (OR1.17), lower than SSRIs (fluoxetine, paroxetine, and citalopram, OR0.80), and the rate of discontinuation due to adverse effects was also higher than fluoxetine (OR1.79). However, another meta-analysis showed no difference between the efficacy of reboxetine and SSRIs. An RCT published in 2014 showed that the difference in efficacy between reboxetine and SSRIs disappeared after accounting for nonadherence.
Regardless, the uncertainty of reboxetine efficacy and poor overall tolerability suggest that the routine clinical use of reboxetine should be avoided, but may be tried when patients have previously responded poorly to 5-HTergic antidepressants.
In summary
The relative efficacy between antidepressants depends on the following factors: whether the drugs are considered individually or in groups based on class or pharmacological properties; whether the doses of drugs with a quantitative-effective relationship are taken into account; whether head-to-head comparisons, meta-analyses, or indirect comparisons by network meta-analysis are used; and which drugs are used as control drugs.
The results of the head-to-head comparison showed a small advantage for the following antidepressants: clomipramine, venlafaxine, escitalopram, and sertraline; other evidence suggested a weak advantage for amitriptyline and mirtazapine. Although overall, these differences are not significant, with higher NNT values, they may be meaningful for specific patients who need to maximize efficacy (e.g., severely depressed or refractory patients).