Data from our study do not support our hypotheses that citalopram and St. John’s Wort would statistically differentiate from placebo-treatment on primary or secondary outcome measures for subjects with Minor Depressive Disorder (). In this study all three treatment groups responded exceedingly well to the structure of being in a clinical trial, as evidenced by clinically meaningful improvements in measures of symptom severity, quality of life, and psychological well-being. Thus, neither of our pharmacological treatment interventions provided additional benefit over placebo treatment. The sample size, although relatively small (N=79), would have had sufficient power to detect a clinically relevant advantage of either active treatment over placebo, such as was found for fluoxetine vs. placebo in the study of minor depression by Judd et al. (2004)
. In the present study, however, the placebo group consistently demonstrated clinically meaningful improvement on all primary and secondary outcome measures. The degree of benefit in the placebo cohort suggests that subjects with minor depression may be responsive to brief non-pharmacologic treatment approaches.
We employed a validated structured assessment of adverse events beginning at the randomization visit. Surprisingly, approximately 60% of the study sample endorsed symptoms that would have been construed as treatment-emergent adverse effects prior to the ingestion of any study compound. In fact, they reported an overall mean of 4.8 PRISE symptoms per subject at baseline. This suggests that it is important to identify systematically, at baseline, the presence and severity of symptoms that otherwise might be misconstrued as treatment-emergent adverse events during the course of a study. Thus, employing a systematic approach to the assessment of adverse events with a validated instrument like the PRISE, yields much more complete and informative data about both baseline levels of events and the emergence or worsening of events during the course of treatment. A second important finding is that all three groups of subjects reported a high rate of adverse events during the study. However, fewer than 40% of the St. John’s Wort and placebo-treated groups reported that any of these adverse effects were distressing as compared to 60% of subjects in the citalopram-treated group. This higher overall rate of adverse events during treatment with citalopram compared to St. John’s Wort, along with higher rates for emerging/worsening problems with increased perspiration, sexual dysfunction, and fatigue, would be likely to be statistically significant in larger study samples. Although the absolute number of adverse events during treatment was not significantly different for the St. John’s Wort and placebo-treated groups, the St. John’s Wort group, like the citalopram group, did endorse significantly more gastrointestinal adverse events than the group receiving placebo, and substantially (although not significantly) more sleep problems. This serves as an important reminder that “natural” products should not be assumed to be benign.
There may be several explanations for the fact that neither active treatment separated from placebo on efficacy outcomes in this study. One possible factor was that we used very stringent inclusion and exclusion criteria for subjects with Minor Depressive Disorder. Individuals could not meet a primary diagnosis for any other psychiatric condition, and although they could have a past history of Major Depressive Disorder, they needed to have a sustained period of remission prior to developing Minor Depressive Disorder. In addition, based on guidance from the Initial Review Group, we limited the range of depressive severity for subjects on the 17-item Hamilton Depression Rating Scale to those with scores from 10 to 17 (inclusive) over the past week. The stringent criteria for this study may have yielded a sample of subjects whose depression was too mild to show added benefit for either active treatment, beyond the general therapeutic effects of study participation evident in the placebo-treated group, let alone differentiation between the two putative active treatments. As previously mentioned, the secondary analysis performed in the Lecrubier et al. study (2002)
, and an unpublished analysis of the study by Judd, et al. (personal communication), suggested that the active treatments had greater separation from placebo only in the most severely ill groups of individuals with MinD. In a secondary analysis of our sample, we did not find that subjects with greater severity ratings were more responsive to active treatment or less responsive to placebo). Another feature that may have confounded our findings is our overall study design: This design biases recruitment to attract subjects willing to commit to a longer-term treatment trial, and by virtue of the crossover, may create an expectancy favoring treatment response. Thus, one cannot know whether the study design, which included 12 weeks of acute treatment plus 12 weeks of continuation or cross-over treatment, could have contributed to the high placebo effect.
Another challenge faced by this study was the common use of St. John’s Wort and SSRI’s for the treatment of mild Major Depressive Disorder and depressive spectrum disorders in the community. Potential subjects may have been less motivated to enter an active comparator, placebo-controlled study of St. John’s Wort. An additional factor that may have influenced recruitment was the evolution of subject compensation policies that emerged over the course of this study. When this study was conceptualized, subjects were not commonly remunerated for participation, and so this was not budgeted for in our application. By the time this study was initiated, sites in our communities were advertising and paying subjects several thousand dollars to participate in such longer-term trials. We believe this affected our recruitment of subjects into the study and may have subtly influenced the characteristics and treatment expectations in this trial. (The original plan was to recruit 300 patients -- 100 to each treatment arm.) A fourth issue that might account for the lack of separation of St. John’s Wort and citalopram from placebo was a higher than previously observed placebo-response. Although this placebo response rate is not different from that reported in recent trials of Major Depressive Disorder, it is higher than the range of placebo response rates identified in the studies reviewed by Volz and Laux (2000)
for subthreshold and mild depression. Original power analyses for this study were based on findings from the Judd et al. study (2004)
of nearly twice as large a mean change in IDS-C30
scores for MinD subjects treated with fluoxetine vs. placebo over 12 weeks [mean=7.6 (9.6) vs. 3.9 (9.0); N=77 in each group], with a standardized treatment effect size of 0.4 (P=0.012). As was noted earlier, the benefit of fluoxetine was almost exclusively in the most severely ill group, who were omitted from the present study. The 1-point difference in mean change in IDS-C30
scores found between citalopram and placebo in the present study is consistent with many more recent publications that suggest there is a more modest separation between active antidepressant treatments and placebo for subjects with MDD than previously documented (Khan et al., 2002
; Zimmerman et al., 2002
; Kirsch et al., 2008
; Turner et al., 2008
; Moreno et al., 2009
; Fournier et al., 2010
). These articles proposed that study-related and patient-related factors, similar to those identified above, may affect treatment response. Our findings are also consistent with a recent review and meta-analysis of the few available placebo controlled trials for Minor Depressive Disorder (Barbui et al., 2011
), which concluded that antidepressant treatment fails to show a clinically meaningful advantage over placebo treatment for this disorder. It is also worth noting that a secondary investigation on the Hypericum Depression Trial Study (Vitiello et al. 2005
), which compared hypericum versus sertraline for MDD, found that 17% of hypericum patients had no measurable hyperforin, and 17% of placebo patients had measurable serum hyperforin, suggesting that many patients were noncompliant and others were obtaining hypericum over the counter. We did not include metabolite measurements in our study, and so we cannot rule out the possibility that subject adherence problems may have contributed to the response patterns we observed.
The current NICE Guideline on the Treatment and Management of Depression in Adults (National Institute for Health and Clinical Excellence, 2010
) recommends that antidepressants therapy is not a first-line treatment for mild depression or subthreshold depressive symptoms, specifically because of concerns about risk-to-benefit ratio. For people with mild to moderate depression or persistent subthreshold depressive symptoms, the NICE guidelines recommend low-intensity psychosocial interventions. Antidepressant treatment should be reserved for people with persistent symptoms (particularly if unresponsive to other interventions), suicidal risk, or a past history of moderate or severe depression. This recommendation is consistent with the Canadian Network for Mood and Anxiety Treatments guidelines as well (Parikh et al., 2009
). As summarized in the recent meta-analysis by Barbui et al. (2011)
, no specific pharmacologic treatment has been demonstrated to be effective for MinD to-date. However, the best non-pharmacologic treatment for Minor Depressive Disorder is also an area that requires extensive investigation. Minor Depressive Disorder may represent a variant of a depressive spectrum subtype that might be particularly responsive to a brief, focused psychotherapeutic intervention based either on a cognitive behavioral or an interpersonal therapy model (Szigethy et al., 2007
; Allart-van Dam et al., 2007
; Cuijpers et al., 2005
; Maina et al., 2005
). We believe that this would be an important area of further research.
In conclusion, this study suggests that neither St. John’s Wort nor citalopram treatment has a clinically or statistically significant benefit for acute treatment of Minor Depressive Disorder when compared to placebo treatment. These findings were clearly due to a consistently high placebo response rate on all outcome measures. Our results suggest that more extensive investigation of Depressive Spectrum Disorders, their natural course, and most effective treatment modalities, is warranted.