Rapid Responses to:

LETTERS:
Jon Jureidini, Anne Tonkin, and Peter R Mansfield
TADS study raises concerns
BMJ 2004; 329: 1343-d-1344-d [Full text]
*Rapid Responses: Submit a response to this article

Rapid Responses published:

[Read Rapid Response] TADS critiques lack scientific rigor, accuracy and thus credibility
John S. March, Benedetto Vitiello   (24 December 2004)

TADS critiques lack scientific rigor, accuracy and thus credibility 24 December 2004
Top
John S. March,
Professor and Chief, Division of Child and Adolescent Psychiatry
Duke University Medical Center, Durham, NC 27710,
Benedetto Vitiello

Send response to journal:
Re: TADS critiques lack scientific rigor, accuracy and thus credibility

To the Editor: We thank Jureidini et al. for their interest in the TADS study 1, but find their critique to lack scientific rigor and accuracy, and thus credibility.

Jureidini et al. confuse randomization and masking. To imply that TADS is like a split-brain patient is to set up a straw man. TADS does not consist of “two separate randomized studies.” Rather, patients were consecutively randomized to one of the four TADS treatments using a computerized stratified randomization, a 1:1:1:1 treatment allocation ratio, permuted blocking within each stratum, and site and gender as stratification variables 2 3. Additionally, the combination group was not simply the additive sum of the CBT and fluoxetine groups, nor were the pharmacotherapists, the CBT therapists, the independent evaluators, segregated into two groups. The masking issue, however, is a serious one, which we grappled with extensively in the design phase of the study2. In comparative treatment trials that include both medication and psychotherapy, it is possible to double-blind the pills only conditions but not the psychosocial treatment conditions. Hence, as is customary, masking in TADS was maintained for the primary dependent measures by means of an independent evaluator kept blind to treatment assignment, while placebo served as a credible trial-wide control condition 4. This is a design choice not a design flaw, which we readily acknowledge constrains the question being asked 2 3. To fully eliminate expectancy effects, which impact all the treatment groups not just the combination condition, would have required a fully specified 2 x 2 factorial, with placebo paired with CBT and a sham psychosocial treatment paired with fluoxetine and also with placebo in a “double negative” control condition. There are many objections to a fully factorial design in an effectiveness setting, among them: (1) If credible, sham psychosocial treatments are active and if inactive are not credible 5, (2) expectancy is a desired contributor to the outcome in an effectiveness RCT that we intend to analyze separately, and (3) feasibility for recruitment among other things would have suffered out of all proportion to value gained. These objections are also true for adding a fifth CBT+PBO arm, but without the virtues of a true 2 x 2 design.

Similarly, in asserting that “the small or absent advantages of fluoxetine on other end points… shows that fluoxetine, like all other antidepressants, is of doubtful clinical importance for children” Jureidini et al. confuse statistical significance with the magnitude of the clinical effect and selectively and inappropriately reference the TADS to support their poor opinion about antidepressants in general 1 6. In the TADS, combined treatment with fluoxetine and CBT proved superior to placebo and to CBT on five of five measures indexing MDD outcome. Fluoxetine alone proved superior to placebo on 3 of 5 measures and to CBT on 5 of 5 measures. Effect sizes on the CDRS-R week 12 endpoints and NNTs on the dichotomized CGI-I indicated that the two fluoxetine conditions showed a moderate to large effect relative to placebo, whereas CBT did not. We further note that the mean and median duration of MDD in TADS was 70 and 42 weeks, respectively, indicating little spontaneous remission in these moderately to severely ill teens with MDD. Using a spontaneous remission rate of 10% and an averaged response rate of 65% for treatment with fluoxetine either alone or in combination with CBT, the corresponding NNT for medication conditions as compared to no treatment is 1.8. Thus, in contrast Jureidini’s unenthusiastic view of medications, we and others 7, conclude that the benefits of including fluoxetine (we make no claims about other antidepressants) in the treatment of adolescent MDD are readily apparent and clinically meaningful.

Jureidini’s comment that “more subjects leaving the study than reporting adverse effects” seems to imply that adverse events were somehow underreported. In fact, the TADS reporting of adverse event data, which largely conforms to new CONSORT recommendations 8, goes way beyond what is typical for a report of this type 9. We clearly describe what happens to subjects through the course of trial with respect to patient attrition, show that attrition does not differ by treatment assignment and primarily reflects lack of benefit rather than adverse events, present fully intent- to-treat analyses of adverse events, document and comment upon the fact that we found more adverse events in fluoxetine-treated subjects than in PBO or CBT treated subjects, and highlight data on harm to self that shows a slightly increased risk in fluoxetine-treated patients. We are preparing a paper that will further explicate non-psychiatric, psychiatric and self and other harm-related adverse events. However, it is apparent from the data presented already that the TADS treatments proved both acceptable and reasonably well tolerated.

With respect to the balance between benefit and harm, we calculated the number needed to harm (NNH) for the outcome of harm to self and the number needed to treat (NNT) for each treatment condition and for the two fluoxetine conditions combined. We presented the resultant NNT to NNH ratios in an invited presentation at the FDA hearings on the safety of antidepressants in the pediatric population10. For fluoxetine relative to placebo, the absolute benefit increase is 27%, the NNT is 4, the absolute risk increase is 4.7%, the NNH is 21 and the NNT to NNH ratio is 5.3. For combined treatment versus placebo, the absolute benefit increase is 37%, the NNT is 3, the absolute risk increase is 2%, the NNH if 50 and the NNT to NNH ratio is 16.7. For any fluoxetine condition versus the two non- medication conditions, the absolute benefit increase is 31%, the NNT is 3, the absolute risk increase is 3%, the NNH is 34 and the NNT to NNH ratio is 11.7. Thus, for all harm-to-self related AEs, the NNH to NNT ratio favors the inclusion of fluoxetine. Furthermore, the NNHs for comparably defined suicide attempts in the fluoxetine conditions from TADS and in the recent FDA metaanalysis are both about 50 11. We note therefore that the ratio of benefit to risk also is favorable for strictly defined suicide attempts. Fortunately, completed suicides are very rare—the case fatality rate for boys is 0.21% and for girls is 0.024% 12-14, with corresponding NNHs for completed suicide in the tens of thousands, confirming that the balance of benefit to risk continues to favor treatment with medication even when the outcome of concern is completed suicide. Lastly, it can be shown using standard clinometric analyses that the NNH is higher in the untreated than the treated state even allowing for a constant odds ratio of 1.5 favoring a small drug-induced increase in suicidality.

Jureidini and others 15 who would deny moderately to severely ill teens with MDD access to fluoxetine (preferably fluoxetine + CBT) should ask themselves the following question: “What is the risk of overstating the case for harm due to fluoxetine?”

For the TADS Team:

John S. March, MD, MPH Duke University Medical Center

Benedetto Vitiello, MD National Institutes of Mental Health

References

1. Jureidini J, Tonkin A, Mansfield PR. TADS study raises concerns. BMJ 2004;329(7478):1343-d-1344.

2. TADS. Treatment for Adolescents With Depression Study (TADS): rationale, design, and methods. J Am Acad Child Adolesc Psychiatry 2003;42(5):531-42.

3. TADS. Fluoxetine, cognitive-behavioral therapy, and their combination for adolescents with depression: Treatment for Adolescents With Depression Study (TADS) randomized controlled trial. Jama 2004;292(7):807-20.

4. Klein D. Control groups in psychotherapy and pharmacotherapy evaluations. Treatment 1997;1(1):http://journals.apa.org/treatment/vol1/97_a1.html.

5. Parloff M. Placebo controls in psychotherapy research: a since qua non or a placebo for research problems? Journal of Consulting and Clinical Psychology 1986;54:79-87.

6. Jureidini JN, Doecke CJ, Mansfield PR, Haby MM, Menkes DB, Tonkin AL. Efficacy and safety of antidepressants for children and adolescents. Bmj 2004;328(7444):879-83.

7. Brent DA. Antidepressants and pediatric depression--the risk of doing nothing. N Engl J Med 2004;351(16):1598-601.

8. Ioannidis JP, Evans SJ, Gotzsche PC, O'Neill RT, Altman DG, Schulz K, et al. Better reporting of harms in randomized trials: an extension of the CONSORT statement. Ann Intern Med 2004;141(10):781-8.

9. Glass RM. Treatment of adolescents with major depression: contributions of a major trial. Jama 2004;292(7):861-3.

10. March J. Treatment for Adolescents with Depression Study: Stage I ITT Outcomes (FDA). Food and Drug Administration Pspychopharmacologic Drugs Advisory Committee and the Pediatric Advisory Committee: http://www.fda.gov/ohrms/dockets/ac/04/slides/2004-4065S1_04_FDA- March.ppt, 2004.

11. Hammad TA. Results of the Analysis of Suicidality in Pediatric Trials of Newer Antidepressants. Food and Drug Administration Pspychopharmacologic Drugs Advisory Committee and the Pediatric Advisory Committee: http://www.fda.gov/ohrms/dockets/ac/04/slides/2004- 4065S1_08_FDA-Hammad_files/frame.htm, 2004.

12. Anderson RN, Smith BL. Deaths: leading causes for 2001. National Vital Statististics Report 2003;52(9):1-85.

13. Spicer RS, Miller TR. Suicide acts in 8 states: incidence and case fatality rates by demographics and method. Am J Public Health 2000;90(12):1885-91.

14. Grunbaum JA, Kann L, Kinchen SA, Williams B, Ross JG, Lowry R, et al. Youth risk behavior surveillance--United States, 2001. MMWR Surveill Summ 2002;51(4):1-62.

15. Lenzer J. Specialists challenge claim that fluoxetine plus talk therapy works best for depressed adolescents. BMJ September 4, 2004;329(7465):529.

Competing interests: Dr March has served as a consultant to and has received research support from Lilly, Pfizer, and Wyeth. The TADS was supported by contract RFP-NIH-NIMH 98-DS-0008 from the National Institute of Mental Health. Lilly Inc providedfluoxetine and matching placebo under an independent educational grant to Duke University.