By Dr. Geoffrey Modest
BMJ published an article finding the lack of benefit of either paroxetine or imipramine in adolescents with major depressive disorder (see BMJ 2015;351:h4320). To me, this study was really interesting because it resulted from an initiative called RIAT (Restoring Invisible and Abandoned Trials), an attempt by an international group to challenge the selective reporting of outcomes of randomized controlled trials, either because the results were never published or were misreported. In this case, the study (Study 329) was funded by a drug company (smithkline beecham), reported in 2001 in the Journal of the American Academy of Child and Adolescent Psychiatry, but “was largely ghostwritten, claimed efficacy and safety for paroxetine that was at odds with the data” (see Account Res 2008; 15: 152, which found that Study 329 showed “how ghostwriting of clinical trial results can contribute to the manipulation of data to favor the study medication. Study 329 of paroxetine pediatric use was negative for efficacy and positive for harm”). The concern was that this was an important article influencing the use of antidepressants, including paroxetine, in adolescents. Details of the article, referred to as “Restoring study 329”:
- The researches reanalyzed the reported study, using the drug company’s final clinical report, other publicly available documents, and 77,000 pages of de-identified individual case reports provided by the drug company. They adhered to the original protocol.
- 275 adolescents aged 12-18 [mean age 15, 57% female, 85% white, mean duration of depression 13 months, 80% with 1 previous episode, 20% with concurrent anxiety, mean Hamilton depression scale (HAM-D)score of 19] who met DSM-IV criteria for a current episode of major depression of at least 8 weeks’ duration, in 12 study sites (10 in US, 2 in Canada) from 1994-1997; patients randomized to paroxetine, imipramine, or placebo in 1:1:1 ratio
- Patients were titrated to paroxetine 20mg or imipramine 200mg over 4 weeks, independent of their response to the medications
- Nonresponders at that point were titrated to a maximum dose of paroxetine 60mg or imipramine 300mg.
- Patients had 45 minute weekly sessions of psychotherapy
- Primary outcome: changes in HAM-D in the acute phase (1st 8 weeks), with a response defined as HAM-D score <=8 or >50% reduction in baseline score.
Results:
- Mean paroxetine dose of 28 mg/d by week 8; mean imipramine dose of 206 mg/d
- Paroxetine was no more effective than placebo (mean prespecified level for clinical significance was 4 points on HAM-D scale)
- Review of the data showed no difference between placebo, paroxetine, or imipramine at any weekly assessment from week 1-8, or after multiple imputation modeling (this is a statistical technique to include missing data by different modeling methods, as opposed to just ignoring the missing data which could introduce a significant bias)
- In terms of harms of therapies: they found that by reviewing the case report forms, the reported harms in the original article did not include 14% of the adverse events in the paroxetine group:
- Paroxetine: 159 adverse events found by authors vs 136 reported; the additional events were mostly psychiatric ones (12/23)
- Imipramine: 257 found vs 240 reported; mostly cardiovascular ones (5/17) not reported
- Placebo: 77 found, 67 reported; mostly psychiatric (4/10) not reported
- Of note, in terms of suicidal and self-injurious behavior (perhaps the most concerning adverse event):
- The original paper reported 5 events with paroxetine, 3 with imipramine and 1 with placebo, and they were listed as “emotional lability”!!!
- The case reports from the drug company reported 7 events with paroxetine, 3 with imipramine and 1 with placebo
- The current RIAT analysis reported 11 events with paroxetine (i.e., more than twice that of the original paper), 4 with imipramine (3 definite, 1 possible) and 2 with placebo (1 definite, 1 possible)
- And several instances were found with bias in interpreting these results (e.g. “an investigator, knowing the patient was on placebo, declared that a suicidal event was ‘definitely related to treatment”’ but of the 11 patients on paroxetine with serious adverse events, only one ”was considered by the treating investigator to be related to paroxetine treatment”)
So, a few issues:
- The obvious one is that it is in the drug companies’ financial interests to mislead or misreport findings that might be harmful to their profits. The difficult-to-believe part of the above story was the extent of this malfeasance, with the drug company ghost-writing this article, exaggerating the positive effects of paroxetine, and hiding and minimizing the harmful ones (including suicide….). And, as in many, many of my blogs, this issue is increasingly critical: more and more articles in the most prestigious medical journals are funded by drug companies, the FDA is ceding more responsibility to the drug companies to do the studies and to do the followup postmarketing surveillance when drugs are approved (especially if approved early, or using surrogate markers), and it is clear that the FDA-required postmarketing studies are done pretty rarely by the drug companies. It really makes it very hard for me to be an early or even middle-aged adopter of new drugs. I just don’t trust the process overall. For more details on this, including rigorous assessments of post-marketing surveillance studies, see several blogs in https://stg-blogs.bmj.com/bmjebmspotlight/category/pharmacy/ .
- I essentially never use paroxetine for anything because of its relatively high frequency of potentially very serious withdrawal reactions. And I would really never use it with adolescents, who, as a group, are probably less likely to be assiduous medication takers.
- I don’t think the above study should cast too much of a pall on the role of SSRIs in adolescent depression. There are many studies finding significant efficacy, mostly done with fluoxetine. And, though the studies are mixed, most show that the combo of SSRI and psychotherapy (esp. cognitive behavioral therapy) is even more effective.