Of 24 registration trials involving eight products submitted to the FDA, four went unpublished in medical journals -- three of which found that the study drug's efficacy was equivalent to placebo or inferior to an active comparator, according to Erick H. Turner, MD, of Oregon Health and Science University in Portland, and colleagues.
Moreover, five of the 20 published trials "showed some evidence of outcome reporting bias," the researchers wrote online in PLoS Medicine.
Turner and colleagues also noted that the scale of the publication bias was relatively modest -- exaggerating the drugs' effectiveness relative to placebo or active comparators by a nonsignificant 8%. The weighted-average effect size in the published trials was 0.47 (95% CI 0.40 to 0.54), which declined only to 0.44 (95% CI 0.37 to 0.50) when the unpublished trials were included.
But overall, the researchers found the results concerning.
"Selective reporting of research results undermines the integrity of the evidence base, which ultimately deprives clinicians of accurate data for prescribing decisions," they wrote.
They also slammed the FDA for making it unnecessarily difficult for ordinary physicians to get the full picture from trial data submitted to the agency.
"At the present time, the FDA is not as transparent with its clinical trial data as it could be," Turner and colleagues wrote. "For example, we pointed out in 2004 that the reviews for several antipsychotic drugs were posted on the FDA website, but only for the original indication of schizophrenia and not for bipolar mania. More than seven years later, the mania reviews remain inaccessible."
For the current study, the researchers had to extensively maneuver the material posted on the FDA's website, which spread important data for a given drug across multiple, nonsearchable electronic documents.
Turner and colleagues examined the FDA archives for placebo-controlled premarketing trial data on aripiprazole (Abilify), iloperidone (Fanapt), olanzapine (Zyprexa), paliperidone (Invega), quetiapine (Seroquel), risperidone (Risperdal), risperidone long-acting injection (Risperdal Consta), and ziprasidone (Geodon).
The researchers then checked to see whether trials reported to the FDA were published in medical journals and, if so, how.
Trials were considered "published" if journal articles about them included enough data to be included in an ordinary meta-analysis, were in English, and "were reasonably discoverable by, and accessible to, the average clinician."
Of the 20 published trials, 15 had been rated by FDA reviewers as positive for the study drug whereas the remaining five were considered negative or questionable.
Only one of the four unpublished trials was deemed positive by FDA reviewers, the researchers found. The other three all found that the study drug was not significantly superior to placebo. Two of these were on aripiprazole and one tested ziprasidone.
The latter trial also included haloperidol as an active comparator, which proved to be statistically superior to ziprasidone.
However, the one unpublished study that favored the study drug over placebo also involved ziprasidone.
Turner and colleagues additionally found that, when trials were published, they were sometimes buffed up to show the drugs in a positive light, with unfavorable data ignored in summary descriptions or omitted altogether.
Four such reports dealt with iloperidone. "One efficacy issue that was apparent from the FDA review of iloperidone, but not from the corresponding journal articles, was that the drug frequently proved to be statistically inferior to active comparators," Turner and colleagues indicated.
They also noted that the iloperidone journal reports obscured trial findings that the drug was superior to placebo only in patients with schizoaffective disorder, not in those with schizophrenia only. Instead, the articles only reported results in the total patient population that included both diagnostic groups.
Also questioned by Turner and colleagues was the reporting of a quetiapine trial. An FDA reviewer had said it "provides marginal support for antipsychotic efficacy" and that "the data fall short of meeting the customary level of statistical proof."
In contrast, the corresponding journal article declared that the results showed "significant differences ... between treatment groups which favored [quetiapine] throughout the trial." Also, whereas the FDA review had noted no evidence of efficacy in patients who completed the scheduled treatment, this finding was not mentioned in the journal article.
One of the published ziprasidone studies also glossed over findings that did not support the drug's efficacy. For example, according to Turner and colleagues, an FDA review had noted that a 60-mg dose twice daily was superior to placebo only at certain time points and not on all outcome measures.
The journal report merely stated that the drug was superior to placebo at week four on two outcome measures, omitting mention of the lack of efficacy at other evaluations and on a third primary endpoint.
But the researchers also emphasized that the overall effect of such biases was modest, and less than other researchers had found for antipsychotic drugs (using a different set of drugs and trials). It was also less than was found in an earlier analysis of antidepressant trial reporting by Turner and a different group of colleagues.
Turner and colleagues acknowledged several limitations to the current study. The small number of trials available for analysis diminished the review's statistical power. They did not address safety endpoints, only efficacy.
Most importantly, the researchers noted that the FDA database on premarketing studies is not comprehensive. It may lack trials conducted outside the U.S. prior to starting the FDA approval process. Moreover, it does not include postmarketing analyses -- which "may represent the majority of trials conducted on a given drug," Turner and colleagues noted.
Secondary source
interesting trials.
ReplyDelete