NPC Archive Item: Evidence to support sertraline as first-line drug treatment option for major depression

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

25th February 2009

A meta-analysis which considered the comparative efficacy of 12 second-generation antidepressants, has concluded that sertraline may be the best choice when starting treatment for moderate to severe major depression in adults because it has the most favourable balance between benefits, acceptability and acquisition cost. However, the meta-analysis has many limitations: the statistical significance of the results is uncertain and the clinical difference between antidepressants is likely to be small.

Prescribers should continue to follow the NICE guideline for the management of depression. The choice of antidepressant, or indeed any drug, should be based on efficacy, side-effect profile, patient preference, past experience with treatment, and cost.  Selective serotonin re-uptake inhibitors (SSRIs) are an appropriate first-line choice for most people where the prescribing of an antidepressant is appropriate. NICE recommends fluoxetine or citalopram. However, this meta-analysis suggests that generic sertraline may also be a suitable option.

What is the background to this?
NICE published its clinical guideline on the management of depression in 2004 (updated in 2007 to reflect MHRA advice regarding venlafaxine). The guideline provides recommendations for treatment options for depression of all severities, including the use of antidepressants. The NICE evidence-based review, reported in the Full Guideline, identified no clinically important difference between any of the antidepressants with regard to efficacy, including the tricyclic and second-generation antidepressants. This guideline is currently being updated with an expected publication date of September 2009.

As reported in a MeReC Rapid Review, a recent systematic review of clinical trial data of second-generation antidepressants was prepared to inform the American College of Physicians (ACP) in the development of their guideline for the use of second generation antidepressants. This review compared the benefits and harms of the SSRIs and other second-generation antidepressants licensed for use in the USA for depression. It concluded that second-generation antidepressants did not differ substantially in efficacy or effectiveness. The subsequent ACP Guideline recommends that physicians choose second-generation antidepressants on the basis of their adverse event profile, cost and patient preference.

What does this study claim?
This meta-analysis (117 RCTs), which included 25,928 adults with unipolar major depression, found statistically significant differences between the efficacy and acceptability of 12 second-generation antidepressants over a mean duration of 8.1 weeks. In terms of response rate (the proportion of patients with a reduction of at least 50% from the baseline score on a depression rating scale), multiple-treatments meta-analysis showed that mirtazapine, escitalopram, and venlafaxine, were statistically significantly more efficacious than duloxetine▼, fluoxetine, fluvoxamine, paroxetine, and reboxetine. Sertraline was more effective than fluoxetine, paroxetine and reboxetine. Based on dropout rates in the first 8 weeks of treatment, escitalopram and sertraline appeared to be better tolerated than several other second-generation antidepressants (see Table 2 below).

The authors conclude that escitalopram and sertraline are good first choices when starting a treatment for moderate to severe major depression in adults because they have the best balance between efficacy and acceptability. Sertraline may be the most appropriate first-line treatment because of its lower cost. In terms of acceptability, reboxetine was the least tolerated agent among the 12 antidepressants and was significantly less effective than all the other 11 drugs. Therefore, reboxetine should not be used as a routine first-line acute treatment for major depression.

So what?
There are few large, high quality, independently sponsored comparative trials of antidepressants and the meta-analysis has many limitations relating to the poor quality of the included studies. As the authors point out, most studies comparing many of the antidepressants were done by the pharmaceutical companies marketing these compounds, which might be a source of bias. Indeed the results of the study show that discrepancies existed between some of the results of the multiple-treatments meta-analysis and those in the direct comparisons. For example, although direct comparison showed a difference, multiple testing showed no statistically significant difference between escitalopram and citalopram, or mirtazapine and venlafaxine.

When studies were given a quality rating according to the adequacy of allocation concealment and blinding, only 12 were rated as adequate: the majority were rated as unclear, which may have biased the results and undermined the validity of the meta-analysis. In addition, the mean sample size was small (109.8 participants per group) and the mean duration of studies was just 8.1 weeks: only 14 studies had a follow-up longer than 12 weeks. It is possible that significant differences might be seen in terms of statistical or clinical efficacy after longer term treatment. This is particularly relevant in a condition such as depression where treatment is generally required for at least six months after the patient responds.

The multiple treatment meta-analysis procedure that was used in this study, and which does not consider placebo responses, has additional limitations. For example, a key assumption behind multiple-treatments meta-analysis is that the analysed network is coherent, i.e. that direct (treatments are compared within a RCT) and indirect (treatments are compared between RCTs by combining results compared with a common comparator) evidence on the same comparisons do not disagree beyond chance. In addition, the statistical significance of the results is uncertain because, as the authors point out, adjustments were not made for multiple testing.

Although some statistically significant differences in efficacy and acceptability were found between agents, it is possible that any differences are of limited clinical significance. As the paper points out there is a large placebo response in antidepressant trials i.e. many patients will show a response to a placebo. Over and above this, the difference in effect sizes between the drugs is likely to be small across a population and cost considerations need to be taken into account. For example, the proportion of patients who do not respond to sertraline, but who respond to escitalopram is probably very small and unlikely to make escitalopram’s use as a first-line agent cost-effective. It is also worth noting that, in the meta-analysis, a response was defined as the proportion of patients who had a reduction of at least 50% from the baseline score on a depression rating scale, the clinical significance of which is unclear. Any benefits of treatment will depend on the initial severity of the depression, with those with more severe depression benefiting more. Finally, the meta-analysis does not compare second-generation antidepressants with first-generation antidepressants.

In its clinical guideline for the management of depression, NICE recommends routine, careful monitoring of symptoms, side effects and suicide risk (particularly in those aged under 30), especially when initiating antidepressants. Patients who are prescribed antidepressants should be reviewed on a regular basis, beginning within one to two weeks after initiating treatment. If the drug has been taken regularly as prescribed, and there is no response to a standard dose of antidepressant, and there are no significant side effects, then an increase in dose can be considered. If there has not been a satisfactory response after a month (six weeks if there is a partial response), or the antidepressant is poorly tolerated, alternative treatment options should be considered. If it is decided to continue with antidepressant treatment, then a switch to a single alternative antidepressant can be considered.

See the NICE guideline for alternative options for the treatment of depression and more details of the factors to be considered when prescribing antidepressants. Further information on the management of depression is also available on the depression section of NPC.

Study details

Cipriani A, Furukawa TA, Salanti G, et al. Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis. Lancet, published online 29 January 2009, doi:10.1016/S0140-6736(09)60046-5

Design: Meta-analysis of 117 RCTs from 1991 to 2007.

Patients, interventions and comparisons: 25,928 adults taking a comparable therapeutic dose of a new-generation antidepressant (bupropion, citalopram, duloxetine▼, escitalopram, fluoxetine, fluvoxamine, milnacipran, mirtazapine, paroxetine, reboxetine, sertraline, and venlafaxine) as monotherapy for the acute-phase treatment (6–12 weeks) of unipolar major depression. Placebo groups were excluded.

Outcomes: The primary outcomes were response and dropout rates because these were the most consistently reported hard outcome measures of acute-treatment efficacy and acceptability. For efficacy analysis, response rate was used instead of a continuous symptom score in order to make the interpretation of results easier for clinicians. Response was defined as the proportion of patients who had a reduction of at least 50% from the baseline score on the Hamilton depression rating scale (HDRS) or Montgomery-Åsberg depression rating scale (MADRS), or who scored much improved or very much improved on the clinical global impression (CGI) at 8 weeks. Treatment discontinuation was defined as the number of patients who terminated the study early for any reason during the first 8 weeks of treatment.

Results: 63% of trials were performed in North America and Europe. 64% of participants were women. The mean sample size was 109.8 participants per group. The mean duration of studies was 8.1 weeks and only 14 studies had a follow-up longer than 12 weeks. The overall mean baseline score at study entry was 23.47 (standard deviation [SD] 4.27) for HDRS-17, 25.72 (SD 4.62) for HDRS-21, and 30.09 (SD 4.64) for MADRS. The quality of only 12 studies was rated as adequate: most trials were rated as unclear according to the quality assessment. Results were analysed on an intention-to-treat basis.

For efficacy, direct comparisons suggested that escitalopram was statistically significantly more effective than citalopram; citalopram was statistically significantly more effective than reboxetine and paroxetine; mirtazapine than fluoxetine and venlafaxine; sertraline than fluoxetine; and venlafaxine than fluoxetine and fluvoxamine. For dropouts, fluoxetine was statistically significantly better tolerated than reboxetine, and citalopram than sertraline. The results that appear to be statistically significant are shown in Table 1. However, these results were from 42 independent analyses without adjustment for multiple testing so around two results would be expected to appear significant by chance alone. Also, it is unclear what the clinical significance of these differences is. No difference was shown for other comparisons.

Table 1: direct comparison meta-analysis
Efficacy (response rate) Odds ratio (95% CI)
Citalopram    vs paroxetine 1.54 (1.04 to 2.28)
reboxetine 1.72 (1.01 to 2.93)
Escitalopram   vs citalopram 1.47 (1.15 to 1.90)
Mirtazapine    vs fluoxetine 1.55 (1.07 to 2.23)
venlafaxine 1.53 (1.03 to 2.25)
Sertraline      vs fluoxetine 1.42 (1.13 to 1.78)
Venlafaxine    vs fluoxetine 1.36 (1.14 to 1.62)
fluvoxamine 2.36 (1.04 to 5.38)
Acceptability (dropout rate)
Citalopram    vs sertraline 0.67 (0.46 to 0.98)
Fluoxetine     vs reboxetine 0.68 (0.49 to 0.94)

A multiple-treatments meta-analysis, which allows the integration of data from direct (within RCTs) and indirect (between RCTs) comparisons, found that escitalopram, mirtazapine, and venlafaxine were statistically significantly more efficacious than duloxetine▼, fluoxetine, fluvoxamine, paroxetine, and reboxetine. Sertraline was significantly more effective than fluoxetine, paroxetine and reboxetine. Reboxetine was significantly less efficacious than all the other 11 antidepressants. Escitalopram and sertraline showed the best profile of acceptability, leading to statistically significantly fewer discontinuations than did duloxetine▼, fluvoxamine, paroxetine, reboxetine, and, for escitalopram alone, venlafaxine. The statistically significant results are shown in Table 2. However, these findings arise from 66 simultaneous comparisons and about three statistically significant findings might be expected by chance alone.

Table 2: multiple-treatments meta-analysis
Efficacy (response rate) Odds ratio (95% CI)
Escitalopram  vs duloxetine 1.33 (1.08 to 1.67)
fluoxetine 1.32 (1.12 to 1.55)
fluvoxamine 1.35 (1.02 to 1.76)
paroxetine 1.30 (1.10 to 1.53)
reboxetine 1.95 (1.47 to 2.59)
Mirtazapine    vs duloxetine 1.39 (1.06 to 1.85)
fluoxetine 1.37 (1.14 to 1.67)
fluvoxamine 1.41 (1.09 to 1.82)
paroxetine 1.35 (1.11 to 1.64)
reboxetine 2.03 (1.52 to 2.78)
Sertraline       vs fluoxetine 1.25 (1.08 to 1.45)
paroxetine 1.22 (1.04 to 1.45)
reboxetine 1.85 (1.41 to 2.44)
Venlafaxine    vs duloxetine 1.30 (1.01 to 1.67)
fluoxetine 1.28 (1.11 to 1.47)
fluvoxamine 1.30 (1.01 to 1.69)
paroxetine 1.27 (1.06 to 1.49)
reboxetine 1.89 (1.45 to 2.50)
Acceptability (dropout rate)
Citalopram     vs fluvoxamine 0.73 (0.54 to 0.99)
reboxetine 0.62 (0.45 to 0.84)
Escitalopram  vs duloxetine 0.70 (0.54 to 0.92)
fluvoxamine 0.69 (0.50 to 0.94)
paroxetine 0.76 (0.62 to 0.93)
reboxetine 0.58 (0.43 to 0.81)
venlafaxine 0.78 (0.64 to 0.97)
Sertraline      vs duloxetine 0.74 (0.55 to 0.99)
fluvoxamine 0.72 (0.53 to 0.97)
paroxetine 0.80 (0.66 to 0.96)
reboxetine 0.61 (0.45 to 0.84)
Bupropion     vs reboxetine 0.62 (0.45 to 0.86)
Fluoxetine     vs reboxetine 0.70 (0.53 to 0.92)

Sponsorship: No drug manufacturing companies were involved

Please comment on this blog in the NPC discussion rooms, or using our feedback form