 |
 |
|
|
|
Epidemiologic studies have shown that major depressive disorder (MDD) has a prevalence of 2% in children and 4% to 8% in adolescents. The American Academy of Child and Adolescent Psychiatry has recommended psycho- pharmacological intervention for this condition (1). In the pediatric age group, selective serotonin reuptake inhibitors (SSRIs) and venlafaxine, a serotonin noradrenalin reuptake inhibitor (SNRI) that acts as an SSRI in smaller doses (2), have been preferred to tricyclic antidepressants (TCAs), presumably because TCAs did not demonstrate efficacy in a metaanalysis of 12 placebo-controlled trials involving children aged 6 to 18 years (3). Of the SSRIs and venlafaxine, fluoxetine is the only drug that meets FDA-outlined criteria for use in pediatric MDD (4). In a US study, Jenson and others found that SSRIs were the second most commonly prescribed psychotropic medications in children and adolescents, second only to stimulants (5). Given the prevalence of MDD in youth and the extent to which SSRIs and venlafaxine are being prescribed in this population, it is important to consider the available evidence for and against using these drugs to treat MDD in children and adolescents. The greatest safety concern regarding SSRIs and venlafaxine is that they may induce or aggravate suicidal ideation or suicidal gestures. As recently as February 2004, Health Canada issued an advisory warning that citalopram, fluvoxamine, paroxetine, sertraline, and venlafaxine are not proven to be safe and effective in pediatric populations (6). According to the advisory, the main concern is a possible associated increased risk of suicide-related events when these antidepressants are used in patients under age 18 years. A Web site presented by the Medicines and Healthcare products Regulatory Agency (MHRA) of the UK lists the data supporting this decision (7). (The MHRA report complements and overlaps with the present review.) Paroxetine has been particularly targeted in this regard because, in unpublished studies, treatment groups have demonstrated significantly higher rates of suicidal thoughts and suicidal behaviour than have placebo groups (8,9). There were no cases of completed suicides in these studies. The idea that SSRIs may induce suicidal ideation is not new. Healy supports this view, citing evidence of this relation in challenge, dechallenge, and rechallenge studies; in meta-analyses of randomized controlled trials (RCTs); and in healthy volunteer studies and epidemiology studies (10). Most of these studies, however, involve adult populations. Conversely, a research team sponsored by Eli Lilly and GlaxoSmithKline conducted epidemiologic research examining the relation between rates of adolescent suicide and rates of antidepressant medication use across 588 US ZIP code regions (11). The results indicated that antidepressant use correlates with decreased suicide rates. However, it is important to note that causation cannot be implied through correlation studies; in certain ZIP code areas, for example, both anti depressant use and decreased suicide rates may be associated with income levels and availability of psychotherapy. RationaleCurrently, SSRI and venlafaxine use in children and adolescents with MDD is highly controversial. This controversy is based on 3 questions: Are SSRIs and venlafaxine effective in the treatment of children and adolescents with MDD? Does SSRI or venlafaxine administration increase the risk of suicidal ideation and suicidal gestures in children and adolescents with MDD? How many children and adolescents prescribed antidepressant medication actually meet DSM-IV criteria for MDD? This review critiques the evidence of efficacy and safety available from published RCTs of SSRI or venlafaxine use in children and adolescents. It is important to note that, after this article was submitted for publication, Jureidini and others published a review with a nearly identical purpose (12). Their paper examines the identical list of articles but emphasizes different critiques of the research, some of which overlap with the discussion below. Unless otherwise mentioned, the conclusions below are independent of the Jureidini article. MethodsCriteria for article inclusion were that the article must study an SSRI or venlafaxine and the design must be a randomized, placebo-controlled trial. The subjects had to be children aged 6 to 11 years, adolescents aged 12 to 18 years, or these age groups combined. Subjects had to be diagnosed as having MDD in accordance with criteria set out by the DSM-III, DSM-III-R, or DSM-IV. The following studies were excluded: unpublished studies; paper presentations; studies involving patients aged 19 years or over; studies that were not published in English; studies wherein nefazadone, mirta- zipine, or TCAs were the primary drug being investigated; case studies; and open-label trials. The Medline database from January 1, 1996, to January 17, 2004, was the main search engine. MeSH terms used included the following: depressive disorders, antidepressive agents, fluoxetine, paroxetine, sertraline, citalopram, fluvoxamine, venlafaxine, child, and adolescent. From this search, recent review articles on the pharmacologic treatment of pediatric depression were read and cross-referenced. The search was also limited to RCTs, and the resulting list was examined closely for articles fitting the inclusion criteria. A senior child and adolescent psychiatrist was consulted for other trials not found in the search. Figure 1 summarizes the method of article finding.  ResultsIn 1990, Simeon and others performed a double-blind study (13). This study involved randomizing 32 subjects, aged 13 to 18 years, to a fluoxetine arm (60 mg daily) and a placebo arm. In each arm, 15 subjects completed the study. Patient evaluations included the Hamilton Depression Rating Scale (HDRS) and the Clinical Global Impressions (CGI) scale. Results did not reach clinical significance on any of the measures used. The study funding source was not listed. In 1997, Emslie and others published an RCT involving 96 subjects, aged 7 to 17 years (14). The treatment arm received fluoxetine 20 mg daily for 8 weeks, and the control arm received a placebo for 8 weeks. This was a single-centre trial sponsored by the National Institute of Mental Health (NIMH). The 2 primary outcome measures were the Clinical Global Impressions Improvement scale (CGI-I) and the Children’s Depression Rating Scale-Revised (CDRS-R). A responder was defined as a subject having a CGI-I rating of 1 or 2 (“very much improved” or “much improved,” respectively) at the end of 8 weeks. In addition, mean CDRS-R scale scores were compared on a week-by-week basis. Remission was defined as a CDRS-R score of < 29. According to the CGI-I, 56% of the fluoxetine arm and 33% of the placebo arm were considered to be responders (P < 0.05). At the end of the trial, differences in mean CDRS-R scores were significant (P < 0.01). The study thus demonstrated significant efficacy on both primary outcomes. Of the fluoxetine subjects, 31% were considered in remission at the end of the 8 weeks, compared with 23% in the placebo arm (a nonsignificant difference). It is important to note that other outcome measures, such as scales for global functioning and general psychiatric symptoms, did not demonstrate fluoxetine’s efficacy. This was attributed to the trial’s short duration. In addition, self-rating scales (specifically, the Children’s Depression Inventory [CDI], the Beck Depression Inventory [BDI], and the Weinberg Screening Affective Scale) did not show significant effects of fluoxetine. From the 2 significant findings in the context of several nonsignificant results, Emslie and others conclude that fluoxetine is effective in treating pediatric MDD—a statement that would be more convincing if global functioning scores were significantly improved in the treatment group. Further, Jureidini and others (13) note that Emslie and colleagues expanded the definition of recovery after the actual study and before the publication of their paper, which shifts the primary outcome results from nonsignificant to significant. Emslie’s second study investigating fluoxetine was a multicentre trial sponsored by Eli Lilly (15). This 8-week study involved 219 subjects aged 8 to 18 years, randomized under the supervision of 15 different investigators. A categorical measure based on CDRS-R scores was the primary outcome implemented. In this trial, a response was defined a priori as a > 30% decrease in scores. For secondary outcomes, scales used included the CGI-I, the Hamilton Anxiety Rating Scale, the BDI, the CDI, and the Global Assessment of Functioning Scale. In this trial, 52% of the subjects on fluoxetine were considered CGI-I responders, compared with 37% of the subjects on placebo. These results are similar to Emslie’s previous study and are significant. Upon completion of the trial, a mean 22-point decrease in CDRS-R scores was noted in the fluoxetine group, compared with a mean 14.9-point decrease recorded for the placebo arm. This difference is significant (P < 0.001)—in fact, a significant difference between mean scores arose after 1 week of the trial. The authors state that the primary outcome, CDRS-R–defined responders, approached but did not achieve significance when the treatment and placebo arms were compared. In post hoc analysis, the authors demonstrate that the difference between groups would have been significant if they had defined responders as achieving > 20%, > 40%, > 50%, or > 60% decrease in scores. They also note that the number of responders (that is, those with a > 30% decrease in scores) would have been significant if they had modified their calculation of percentage change to account for the minimum score of 17 on the CDRS-R scale. The authors conclude that fluoxetine is effective in the treatment of child and adolescent depression. Given that the post hoc analysis involved more suitable calculations, the statistical claims of efficacy are convincing. In keeping with Emslie’s previous study, global functioning scales and self-rating scales did not demonstrate significant differences between treatment group and placebo group, calling into question the clinical utility of fluoxetine. Keller and others performed an RCT for paroxetine that was sponsored by GlaxoSmithKline (16). The subject group (n = 275) included only adolescents, aged 12 to 18 years. The subjects were randomized throughout 12 centres: 1 treatment group (n = 93) received paroxetine titrated from 20 to 40 mg daily; another group (n = 95) took imipramine; and the remainder (n = 87) took placebo. The primary outcomes were measured by the HDRS. The first primary outcome was categorically sorted into “response,” defined as an end-of-trial score of < 8 or a > 50% reduction in score, and “nonresponse,” wherein neither of these criteria were met. The second primary outcome was significant change in mean HDRS score. The research group also measured CGI scores (using Emslie and others’ definition of response) and scores on the Schedule for Affective Disorders and Schizophrenia for Adolescents-Lifetime version (K-SADS-L). The results could not demonstrate a difference between treatment and placebo arms in either of the primary outcomes defined a priori. Post hoc analysis showed that a definition of response requiring only an HDRS score of < 8 would have manifested a significant relation between treatment and remission from depression. In addition, post hoc analysis showed that the depression item scores on both the HDRS and the K-SADS-L were affected by paroxetine administration. Moreover, at 65.6%, the number of CGI-defined responders in the treatment group was significant, compared with 48.3% in the placebo group. The authors conclude that paroxetine is effective in treating adolescent MDD. This conclusion is inappropriate, however, because it is based on secondary outcomes and liberal use of post hoc analysis, whereas the primary outcome showed nonsignificant results. Wagner and others have performed the most recent published controlled trial of SSRIs in child and adolescent MDD (17). Sponsored by Pfizer, this investigation originally comprised 2 separate controlled trials. After aggregation of the studies, 376 subjects were randomized. Subjects were drawn from 53 centres in 5 different countries, although the vast majority were recruited in the US. The age range was 6 to 17 years. The primary outcome measure was the mean change from baseline in the CDRS-R Best Description of Child total score, adjusted for the minimum score of 17. Other outcome measures included a categorical use of the CDRS-R (with responders defined as having a > 40% decrease in score), the CGI-I (with Emslie and others’ definition of responders), the Multidimensional Anxiety Scale for Children, the Children’s Global Assessment Scale, and the Pediatric Quality of Life Enjoyment and Satisfaction Questionaire. When all patients who had been randomized were considered, the changes in mean CDRS-R Best Description of Child total scores differed significantly between treatment and placebo groups (–22.84 vs –20.19, respectively; P < 0.007). Moreover, a significantly higher number of CGI-I responders and CDRS-R–defined responders were found in the treatment arm, compared with the placebo arm (P < 0.05 in both cases). For efficacious cognitive-behavioural therapy (CBT) in adolescents with mild-to-moderate depression, the number needed to treat is 4 subjects (18); by contrast, Wagner and others note that their analysis demonstrates a need for 10 subjects to be treated to demonstrate efficacy in 1 subject. They attribute this high number needed to treat to a high placebo response rate (59% for CDRS-R placebo responders and 53% for CGI placebo responders). It should be noted that, when broken down into age groups, the treatment effect is only demonstrated in the adolescent group (P < 0.01) and is not significant in the child group. The authors note that the trial was not designed to determine age effects. The study does demonstrate efficacy; however, the high number needed to treat calls into question sertraline’s clinical utility. In addition, the MHRA claims that reanalysis of this data demonstrates that sertraline lacks efficacy, compared with placebo (8). Jureidini and others point out that lack of efficacy is demonstrated when the 2 identical RCTs are treated separately (13). If the studies are amalgamated, a small difference is noted, owing to a large sample size. Jureidini and others question the clinical significance of this small difference in outcomes between the treatment and control groups. Mandoki and others have conducted the only published placebo-controlled trial of venlafaxine use in children and adolescents (19). The article does not clearly indicate the funding source. The sample size was limited to 33 subjects at a single site. The research group measured symptoms with the CDI, the Child Behavior Checklist, the HDRS, and the CDRS. Subjects were randomized to placebo or venlafaxine. For children aged 8 to 12 years, the dosage was 12.5 mg titrated from once daily to 3 times daily; for adolescents aged 13 to 17 years, the dosage was 25 mg titrated from once daily to 3 times daily. CBT was also administered to all subjects. On all measures, the results demonstrated no significant difference between treatment and placebo groups. The authors suggest that the negative findings may be due to low dosages and high rates of hepatic metabolism in pediatric populations, to short duration of treatment, or to the fact that CBT may have distorted any effect from the medications. CBT has been shown to be effective in improving depressive symptoms in 62% of child and adolescent subjects, whereas the rate of improvement in placebo groups was 36% (18), making this last point very plausible. Safety With Regard to Suicide DiscussionPublished studies of RCTs yield inconclusive evidence to determine the efficacy and safety of SSRIs and SNRIs in treating pediatric MDD. Current research projects are underway, which will further our understanding of pharmacologic treatment of pediatric depression. One such project is the Treatment for Adolescents with Depression Study (TADS) (20). This study’s purpose is to examine the effects of fluoxetine, CBT, and a combination of the 2, compared with each other and with placebo. The following discussion addresses the shortcomings of the studies described in the results section, namely, unjustified claims of efficacy, lack of improvement in global functioning scores, inadequate adverse-effects data collection, exclusion of suicidal subjects, age grouping, small sample size, conflict of interest through pharmaceutical company sponsorship, and publishing bias. With regard to efficacy, only 2 of the 6 studies claimed efficacy on a priori–defined outcome measurements: Emslie’s 1997 study on fluoxetine (14) and Wagner’s 2003 study on sertraline (17). As mentioned above, both of these were contested in further analysis. A few studies claimed efficacy on the basis of post hoc analysis, which allowed for the influence of hindsight bias (21), and secondary measures (Emslie’s 2002 study on fluoxetine [15] and Keller’s 2001 study on paroxetine [16]). Given that a P value of 0.05 represents a 1-in-20 probability that the result of a single outcome measure is owing to chance, increasing the number of outcome measures also increases the probability that at least one result will be deemed significant when it is in fact owing to chance. It is important to limit primary outcome measures to 1 or 2 components or to use protected P values. Claims of efficacy should be based on the condition that primary measures demonstrate significant differences between groups, ideally with secondary outcome support. In the ongoing TADS study, by contrast, the primary outcome measures have been published and have been clearly defined as a categorical CGI rating and the mean CDRS score. It should be noted that none of the 6 studies demonstrated drug effects on global functioning or quality of life scales, although the rationale behind the pharmacotherapy research is that MDD causes significant morbidity and mortality. The studies’ limited time course (no more than 10 weeks) may be a factor in these negative results. The TADS study will examine effects in the acute phase (that is, at 12 weeks) as well as longer-term effects (that is, results at 36 weeks). This may offer more information on significant gains in daily functioning resulting from treatment. In reporting results, only 2 studies directly addressed the issue of safety in relation to suicidal thoughts and behaviours. No studies detailed a systematic method of defining and determining suicidal ideation or suicide attempts. The TADS study has indicated the use of a manual-outlined method for specifically inquiring about suicidal ideation and attempts. Only subjects who were nonsuicidal to begin with were enrolled in these studies—a fact that needs to be considered. Thus these trials could not address the question of whether SSRIs or SNRIs aggravate preexisting suicidal ideation. Although it can be presumed that a significant number of clinical pediatric patients who may be prescribed SSRIs or SNRIs have suicidal ideation or behaviours, the above studies cannot be generalized to this population. For ethical reasons, it is not appropriate to enroll these subjects into placebo-controlled studies of medications believed to be efficacious. The TADS study does not specifically address this issue; however, it refers to another ongoing study sponsored by the NIMH: the Treatment of Adolescent Suicide Attempters (TASA) study. It should also be considered that 4 of the 6 studies reviewed do not differentiate between prepubertal and postpubertal depression. Investigators need to acknowledge that developmental factors may play a role in MDD within the pediatric age group; that is, children may respond differently to treatment than do adolescents. The TADS study examines adolescent depression only. A distinct study should examine prepubertal depression and the respective appropriate treatments. Another important consideration is the studies’ relatively small sample size. In Joffe and others’ metaanalysis of the effectiveness of antidepressants, it is possible to compute an effect size of r = 0.24 (22). To capture an effect of that size reliably, a sample size of 120 (60 per group) is required to find significant results 80% of the time (K Parker, personal communication, March 2004). By this standard, only 3 of the 6 studies had enough subjects. Conversely, the TADS study accounted for statistical power when determining the number of subjects to enroll. Of the 6 studies, the 3 with the largest n values were sponsored by drug companies. While the industry likely needs to perform initial studies on medications, studies sponsored by pharmaceutical companies should not ultimately determine whether drugs are approved for widespread use. As recently suggested by Bhandari and others, a substantial conflict of interest introduces some doubt regarding the integrity of these studies’ findings (23). By contrast, because the TADS study is sponsored by the NIMH, which is a government agency, the industry impact is minimized. Lastly, there is reason to believe a publishing bias has influenced this review. When Laughren conducted a thorough search of the FDA database, 12 RCTs of SSRIs or SNRIs in children and adolescents were revealed (24). However, where Laughren found 12, this review found only 6 such trials. If we assume that unpublished studies are those least favourable to the marketing of the medication, this demonstrates a possible significant publishing bias. The TADS project alleviated this problem by publishing its purpose and methods while subject recruitment and analysis were beginning to get underway. In summary, the current literature on SSRI and SNRI use in pediatric MDD is clouded by assumptions of efficacy based on post hoc analysis and secondary outcomes, lack of evidence of change in quality of life, inadequate measures of suicidal ideation and attempts, the exclusion of suicidal subjects, the assumption that pubertal changes are not a factor, underpowered studies, drug industry sponsorship, and publishing bias. From an examination of the RCTs, high rates of SSRI and venlafaxine prescription in the pediatric population are unjustified. The design of future studies on SSRI and SNRI use in children and adolescents should take the shortcomings of the reviewed studies into account. AcknowledgementsThis article was supervised and reviewed by Dr N Roberts, Associate Professor, Department of Psychiatry, Queen’s University, and Dr K Parker, Director of Child and Adolescent Psychiatry Research, Queen’s University, Kingston, Ontario. References1. American Academy of Child and Adolescent Psychiatry. Practice parameters for the assessment and treatment of children and adolescents with depressive disorders. J Am Acad Child Adolesc Psychiatry 1998;37(Suppl):63S–83S. 2. Arana GW, Rosenbaum JF. Handbook of psychiatric drug therapy. Philadelphia (PA): Lippenscott Williams & Wilkins; 2000. p 86. 3. Hazell P, O’Connell D, Heathcote D, Robertson J, Henry D. Efficacy of tricyclic drugs in treating child and adolescent depression: a meta-analysis. BMJ 1995;310:897–901. 4. Food and Drug Administration: Psychopharmacologic Drugs Advisory Committee and the Anti-Infective Drugs Advisory Committee. Pediatric program summary statistics. Available: http://www.fda.gov/ohrms/dockets/ac/04/ briefing/4006B1_04_Ped%20Exclusivity-OCD_wr_template.pdf. Accessed 2004 July 1. 5. Jenson PS, Bhatara VS, Vitiello B, Hoagwood K, Feil M, Burke LB. Psychoactive medication prescribing practices for US children: gaps between research and clinical practice. J Am Acad Child Adolesc Psychiatry 1999;38:557–65. 6. Health Canada. Warnings/Advisories 2004. Available: http://www.hc-sc.gc.ca/ english/protection/warnings/2004.htm. Accessed 2004 July 1. 7. Medicines and Healthcare products Regulatory Agency. Overview of regulatory status and CSM advice relating to major depressive disorder (MDD) in children and adolescents including a summary of available safety and efficacy data. Available: http://www.mhra.gov.uk/news/2003.htm#ssri. Accessed 2004 July 1. 8. Wooltorton E. Paroxetine (Paxil, Serotax): increased risk of suicide in pediatric patients. CMAJ 2003:169;446. 9. Waechter F. Paroxetine must not be given to patients under 18. BMJ 2003;326:1282. 10. Healy D. Lines of evidence on the risks of suicide with selective serotonin reuptake inhibitors. Psychother Psychosom 2003;72:71–9. 11. Olfson M, Shaffer D, Marcus SC, Greenberg T. Relationship between antidepressant medication treatment and suicide in adolescents. Arch Gen Psychiatry 2003;60:978–82. 12. Jureidini JN, Doecke CJ, Mansfield PR, Haby MM, Menkes DB, Tonkin AL. Efficacy and safety of antidepressants for children and adolescents. BMJ 2004;328:879–83. 13. Simeon JG, Dinicola VF, Ferguson HB, Copping W. Adolescent depression: a placebo-controlled fluoxetine treatment study and follow-up. Prog Neuropsychopharmacol Biol Psychiatry 1994;14:791–5. 14. Emslie GJ, Rush AJ, Weinburg WA, Kowatch RA, Hughes CW, Carmody T, and others. A double-blind, randomized, placebo-controlled trial of fluoxetine in children and adolescents with depression. Arch Gen Psychiatry 1997;54:1031–7. 15. Emslie GJ, Heiligenstein JH, Wagner KD, Hoog SL, Ernest DE, Brown E, and others. Fluoxetine for acute treatment of depression in children and adolescents: a placebo-controlled, randomized clinical trial. J Am Acad Child Adolesc Psychiatry 2002;41:1205–15. 16. Keller MB, Ryan ND, Strober M, Klein RG, Kutcher SP, Birmaher B, and others. Efficacy of paroxetine in the treatment of adolescent major depression: a randomized, controlled trial. J Am Acad Child Adolesc Psychiatry 2001;40:762–72. 17. Wagner KD, Ambrosini P, Rynn M, Wohlberg C, Yang R, Greenbaum M, and others. Efficacy of sertraline in the treatment of children and adolescents with major depressive disorder: two randomized controlled trials. JAMA 2003;290:1033–41. 18. Harrington R, Whitaker J, Shoebridge P, Campbell F. Systematic review of efficacy of cognitive behaviour therapies in childhood and adolescent depressive disorder. BMJ 1998;316:1559–63. 19. Mandoki MW, Tapia MR, Tapia MA, Sumner GS, Parker JL. Venlafaxine in the treatment of children and adolescents with major depression. Psychopharmacol Bull 1997:33:149–54. 20. March J, Silva S, Petrycki S, Curry J, Wells K, Fairbank J, and others. Treatment for Adolescents With Depression Study (TADS): rationale, design, and methods. J Am Acad Child Adolesc Psychiatry 2003;42:531–42. 21. Gilovich T. Something out of nothing. How we know what isn’t so. New York: The Free Press; 1993. p 19–21. 22. Joffe R, Sokolov S, Streiner D. Antidepressant treatment of depression: a metaanalysis. Can J Psychiatry 1996;41:613–6. 23. Bhandari M, Busse JW, Jackowski D, Montori VM, Schünemann, Sprague S. Association between industry funding and statistically significant pro-industry findings in medical and surgical randomized trials. CMAJ 2004;170:477–80. 24. Laughren T. Brief regulatory history of antidepressants and suicidality and update on current plans for analysis of pediatric suicidality data from controlled trials. Presented at joint meeting with the Anti-Infective Drugs Advisory Committee. Available: http://www.fda.gov/ohrms/dockets/ac/04slides/4006s1.htm. Accessed 2004 July 1. Author(s)Manuscript received March 2004, revised, and accepted April 2004. 1. Formerly, Fourth-Year Medical Student, Queen’s University, Kingston, Ontario; now, First-Year Resident, Department of Psychiatry, University of Ottawa, Ottawa, Ontario. Address for correspondence: Dr D Courtney, Postgraduate Education, Department of Psychiatry, c/o TOH–General Campus, Room 4418, Box 400, 501 Smyth Road, Ottawa, ON K1H 8L6 e-mail: darren.courtney@sympatico.ca
1 | 2
|
||||||