Healthy Skepticism International News

April 2006

Healthy Skepticism about antidepressants for children and adolescents -an Australian perspective

Discussion of the benefits and harms of selective serotonin reuptake inhibitors (SSRIs) for major depression in children and adolescents has been amongst the most interesting debates in medicine so far this century [1-3]. The great variation in conclusions appears to arise from different approaches to evaluation of evidence, so the debate has important implications for decision making in all areas of medicine.

In April 2005 the Australasian colleges representing psychiatrists, physicians and general practitioners released a clinical guideline on the use of SSRIs to treat children and adolescents with depression [4]. This paper presents a chronological summary of recent Australian contributions to the debate leading up to the release of that guideline, and a discussion of factors that may have influenced its development. This summary places the Australian contributions in their chronological and international context from the perspective of Healthy Skepticism Inc. Healthy Skepticism is an international organisation aiming to improve health by reducing harm from misleading drug promotion [5].

Chronology

In October 2002 Chan et al. [6] proposed changes to the 1997 NHMRC guidelines for treating young people with depression, [7] recommending that “SSRIs, particularly fluoxetine and paroxetine, should also be considered as a first-line treatment” for major depression in young people, on the basis of evidence from two randomised controlled trials (RCTs) [8,9]. Two of us (JNJ, ALT) pointed out that the authors of those RCTs had claimed to have shown efficacy despite there being no significant difference from placebo on their pre-specified primary endpoints [10,11].

We assembled a multidisciplinary team to perform a systematic review of newer antidepressants for depressed children and adolescents. Our review, published in the BMJ in April 2004 [12], found statistically significant but small benefit, with stronger evidence of harms.

The average benefit from newer antidepressants was only about 3.9 points (95% CI 1.4-5.6) on the Children’s Depression Rating Scale Revised (CDRS-R) (possible range 17-113). The average additional benefit from antidepressants was much smaller than the average improvement on placebo.  This is illustrated by the point estimates for one trial where the CDRS-R improved by 20.19 in the placebo group and by 22.84 in the sertraline group [13].

RCTs are usually powered to detect statistically significant differences in primary efficacy endpoints but not less frequent adverse events. Consequently, it is noteworthy that two trials had statistically significant differences in adverse events, but not in pre-specified efficacy endpoints. However the investigators’ conclusions highlighted differences in other efficacy endpoints without mentioning the adverse event findings [13,14].

Our conclusion was that the magnitude of benefit of the newer antidepressants is unlikely to be sufficient to justify the harms. Consequently, confidently recommending these drugs as a treatment option, let alone as first-line treatment, would be inappropriate.  We also concluded that the design and reporting of the trials were biased. (See our full conclusions in Box 1.)

Box 1

Conclusions of the Jureidini et al. (2004) systematic review

“The trials consistently found large improvements in placebo groups, with statistically significant additional benefits for active drug on some measures only. These results make a major benefit from newer antidepressants unlikely, but a small benefit remains possible. Randomised controlled trials usually underestimate the serious adverse effects of drugs. The fact that serious adverse effects with newer antidepressants are common enough to be detected in randomised controlled trials raises serious concerns about their potential for harm. The magnitude of benefit is unlikely to be sufficient to justify risking those harms, so confidently recommending these drugs as a treatment option, let alone as first line treatment, would be inappropriate.

We are concerned that biased reporting and overconfident recommendations in treatment guidelines may mislead doctors, patients, and families. Many will undervalue non-drug treatments that are probably both safer and more effective. Accurate trial reports are a foundation of good medical care. It is vital that authors, reviewers, and editors ensure that published interpretations of data are more reasonable and balanced than is the case in the industry dominated literature on childhood antidepressants. This is particularly true in the light of the increasing reliance on online abstracts by doctors who lack the time or the skills for detailed analysis of complete trial reports.”

Our review was followed a week later by a similar review in The Lancet by Whittington et al., who later won that journal’s paper of the year award [15]. Whittington et al.’s study was more comprehensive than ours in that it included unpublished trials obtained from the UK Committee on Safety of Medicines. Whittington et al. meta-analysed each newer antidepressant separately and concluded that “Published data suggest a favourable risk-benefit profile for some SSRIs; however, addition of unpublished data indicates that risks could outweigh benefits of these drugs (except fluoxetine) to treat depression in children and young people.”

The Whittington et al. review’s only disagreement with our conclusion that antidepressants are likely to do more harm than good was their judgment that fluoxetine may be moderately effective. Both reviews analysed the same two fluoxetine trials [8,9]. The difference in conclusions arises from different methods. Firstly, Whittington et al. did not examine the quality of the trials. We did, and found several flaws, including exclusion of initial responders in the placebo group and switching of primary endpoints. Secondly, we meta-analysed continuous endpoints for all published trials together, whereas Whittington et al. meta-analysed both continuous and binary endpoints for each drug separately. They put more emphasis on a binary endpoint: the remission rate, i.e. the percentage of subjects whose depression improved enough to get over a somewhat arbitrary cut-off point so as to be classified as no longer depressed. They found that the relative risk of non-remission with fluoxetine versus placebo was 0·78 (95% CI 0·67–0·90) suggesting moderate effectiveness. In contrast, their meta-analysis of continuous scale endpoints for fluoxetine showed only “a small reduction in depressive symptoms”. The magnitude of that reduction was little more than we had reported for all newer antidepressants (see Box 2). Dichotomising continuous data can create distorted impressions and exaggerate small differences in effect [16]. In this instance, the discrepancy between the impression of moderate efficacy on the binary endpoint and the impression of minimal efficacy on the continuous endpoint can be explained as a “squeezing the middle” effect [17]; i.e. some children and adolescents do better on fluoxetine than on placebo but almost as many do worse on fluoxetine than on placebo. Consequently the binary remission endpoint makes fluoxetine look better than it really is by missing information about those who do not do well on fluoxetine [18]. These two fluoxetine trials were also criticised in an FDA statistical review [19].

In May 2004 Hall et al. published a trend analysis showing a 12-fold increase in use of antidepressants by 15-24 year old males in Australia from 1991 to 2000, with no change in their suicide rate, and a 10-fold increase in use of antidepressants by 15-24 year old females, with a 20% increase in their suicide rate [20]. They did not present data for younger children. The suicide rate declined among older adults. Hall et al. cautiously suggested that increased antidepressant prescribing might have been one factor contributing to this decline. However, such ecological data are difficult to interpret because there are many influences on suicide, and this conclusion was vigorously debated in rapid responses and letters.

Box 2

Effect size findings in the two meta-analyses

Many of the trials of antidepressants for children used multiple endpoints based on several different continuous scales used to measure depression. Our Jureidini et al. (2004) meta-analysis incorporated results from different scales by calculating the average Hedges g effect size: the difference between the mean continuous scale scores for the treatment and placebo groups standardised divided by the pooled population standard deviation. We found that the average efficacy effect size of newer antidepressants for children and adolescents was a Hedges g of 0.26 (95% CI 0.13-0.40).

Whittington et al. (2004) meta-analysed all published and unpublished studies of newer antidepressants separately and found similar results for all of them except fluoxetine. They found that the efficacy effect for fluoxetine was a standardised mean difference (SMD) of 0.43 (95% CI 0.20-0.65).  The SMD is a generic term for effect size calculated with a range of techniques including Hedges g so these numbers are approximately equivalent.

Effect sizes are more meaningful if expressed in natural units such points on the CDRS-R scale. Standardised effect sizes can be converted to natural units if the standard deviation is known. The only standard deviation for CDRS-R scores disclosed by a published trial was 14.8 in the treatment group.8

Assuming a standard deviation of 15 points on CDRS-R then:

* Our Jureidini et al. (2004) meta-analysis found that the average efficacy of the newer antidepressants was 3.9 CDRS-R points (95% CI 2.0-6.0).

* The Whittington et al. (2004) meta-analysis found that the efficacy of fluoxetine was 6.5 CDRS-R points (95% CI 3.0-9.8).

Conclusion: It is possible that fluoxetine is more effective than other antidepressants for children and adolescents but there have not been head to head trials and there is considerable overlap between the 95% confidence intervals from the meta-analyses. Furthermore it remains unlikely that the benefit is enough to justify the harms even for fluoxetine.

In August 2004 the Treatment for Adolescents with Depression Study (TADS) results were published [21]. The main finding was no significant difference in the double-blind comparison of fluoxetine and placebo on the most informative primary endpoint, the CDRS-R, but this was not disclosed in the abstract.  TADS also compared cognitive behaviour therapy alone and in combination with fluoxetine. These comparisons are of dubious validity because they were not blinded. We pointed out these problems in a letter in the BMJ [22]. The TADS investigators responded aggressively to our criticisms [23].

In September 2004, US Food and Drug Administration hearings confirmed that antidepressants increase “suicidality” (suicide related thoughts and behaviours) in children and adolescents (17,24,25). It is possible that SSRIs have a “space shuttle” type safety profile, i.e. moderately safe during stable use but more dangerous when starting or stopping [26].

In August 2004 Haby et al. published an economic comparison of SSRIs and cognitive behaviour therapy (CBT) that found that CBT by psychologists is the most effective and cost-effective option for the first-line treatment of major depression in children and adolescents [27]. Haby et al. were able to locate only two other studies of the cost-effectiveness of CBT for depression, neither of which focused on children and adolescents. This illustrates the need for more economic analysis of non-pharmacological treatments for depression.

In October 2004 the Australian Adverse Drug Reactions Advisory Committee (ADRAC) stated correctly that “Assessment of the published and unpublished data available for SSRI use in children and adolescents indicates that there is evidence of an increased risk of suicidality, including suicidal ideation, suicide attempts and self-harm events, associated with each of the SSRIs” [28]. ADRAC also stated, incorrectly, that “In a recent study [TADS], at the completion of therapy fluoxetine was beneficial for the treatment of depression in adolescents with moderate to severe symptoms”.  ADRAC asked the colleges representing psychiatrists, physicians and GPs to produce a clinical guideline.  An offer by Healthy Skepticism to assist development of that document was declined for reasons that were not divulged to us.

In December 2004 Rowe et al. published a non-systematic review in the Australian Family Physician, asserting that “Although the case against SSRIs in the management of adolescent depression is as yet unproved, on the balance of current information and according [to] the recommendations of NHMRC, ADRAC and pharmaceutical companies, in cases where CBT has failed or the depression is life threatening, Australian GPs may currently prescribe fluoxetine” [29].

We fear that prescribing a near ineffective drug in those circumstances may dangerously distract from the need to seek further assistance.

There were three important events in April 2005. Firstly, the European Medicines Agency warned that, because of the potential risk of suicidal behaviour, SSRIs and serotonin-norepinephrine reuptake inhibitors (SNRIs) should not be used in children and adolescents except in their approved indications, which do not include depression and anxiety [30].

Secondly, an MJA editorial recommended that “Until the ambiguities are resolved, based on our clinical experience, we believe that SSRIs (chiefly fluoxetine) can be considered, but only for severe depression, when it produces serious impairment and fails to respond to psychosocial treatment over a few weeks” [31]. The authors explicitly stated that they were not discussing the evidence regarding efficacy in making this recommendation.

Thirdly, the three Australian Colleges released the clinical guideline that ADRAC had requested, with a media release asserting that “Clinicians recognise that antidepressants can be very safe and effective when used as part of the comprehensive management of moderate to severe depression in the young and in many instances they can be life-saving” [32]. This guideline does not disclose the names of the authors, their methods or any competing interests.

Discussion

There are a number of possible reasons why the April 2005 media release and guideline are so positive about antidepressants despite RCT evidence of greater harm than benefit.

Firstly, many doctors place greater trust in clinical experience than RCT data [33]. Many may have observed improvement in children using these drugs without understanding that more than 80% of the apparent benefit also occurs on placebo [34]. Furthermore, adverse effects may have been misattributed to the disease instead of the drugs.

Secondly, many doctors fear that it would be negligent to not “do something” [35] to prevent suicide [32]. However, there is no good evidence that antidepressants reduce suicide rates in children and adolescents. By contrast, there is strong evidence that antidepressants increase suicidal behaviour [24]. The increased suicide rate among young Australian females and the stable rate among young Australian males at the same time as dramatically increased use of antidepressants is not conclusive but does not support optimism about suicide prevention for young people [20].

Thirdly, we cannot exclude the possibility of influence from drug companies because the guideline writers did not disclose their identities or any competing interests. The UK House of Commons Health Committee Report on the influence of the Pharmaceutical Industry (House of Commons, 22 March 2005) recommended that ‘a register of interest be maintained by the relevant professional bodies…detailing all substantial gifts, hospitality and honorary received by members.  The Register should be made available for public inspection.’ We urge Australian authorities to follow this advice.

We are concerned that the media release and guideline will lead to significant unnecessary unintended harm to young Australians. We believe the best way forward would be for Australia’s NHMRC to commence a high quality guideline development process and to fund research on improving the dissemination of evidence and on alternatives to antidepressants.

Peter R Mansfield, BMBS

Research Fellow, Department of General Practice

University of Adelaide

Jon N Jureidini, MBBS, FRANZCP, PhD

Head, Department of Psychological Medicine

Child Youth & Women’s Health Service, Adelaide

Melissa K Raven MPsych, MMedSci (Clin Epid)

Lecturer, Department of Public Health

Flinders University, Adelaide

Anne L Tonkin BSc, BMBS, PhD, FRACP

Associate Professor, Department of Clinical and Experimental Pharmacology

University of Adelaide

Statement of competing interests:
All authors are members of Healthy Skepticism Inc. PRM is the founder, JNJ is Chair of the Management Group and MKR & ALT are members of the Management Group.

Funding:

There was no specific funding for this article. PRM is supported by a National Health and Medical Research Council Public Health Postgraduate Scholarship.

View/Hide References

HS Int News index

Page views since 15 March 2010: 6060

Comments

Our members can see and make comments on this page.