<!DOCTYPE html>
<style>body{font-family: arial; font-size:150%}</style>
<body bgcolor="#ffffff">

<p align="justify">This site is designed by child and adolescent psychiatrists for child and adolescent psychiatrists.<br>
If you are a CAP please take 5 minutes to review the information below.<br>
We believe that, when necessary, information important to us, to our community, can be spread effectively peer to peer. <br>
You can help make this a reality at our annual meeting: If you find this site worthwhile, please pass allcapsunlocked.com on to at least 2 other CAPs

who have not yet seen it.<p>

<h2 align="justify"><u>Background</u></h2>
<p align="justify">In 2001 JAACAP editor Myna Dulcan published “Efficacy of Paroxetine in the Treatment of Adolescent Major Depression: A Randomized,

Controlled Trial” (reference 1). Based on findings from a clinical trial, Study 329, led by Drs. Neal Ryan and Martin Keller, and sponsored by

GlaxoSmithKline. The last information that JAACAP has published about Study 329 was in 2003, a response from Ryan and Keller defending the

representation of information from the trial (reference 2).</p>

<p align="justify">In 2008 a great deal of new information regarding Study 329, previously not public, emerged (reference 3). This new information has

raised serious questions about what had been published in JAACAP about the study. It is information that many of us feel is important be shared within

our community. Among the reasons for this: the new information not only radically alters our understanding of the original study, but also highlights

broader issues about the literature we rely upon as clinicians.</p>

<p align="justify">Since 2008 there have been multiple requests, from both within AACAP and without, to JAACAP editor Andres Martin to allow this

information to be shared with readership. There have also been multiple requests to successive AACAP presidents to allow this information to be shared

some with members. After more than 5 years, it is time we share directly with each other the information below. This context needs to be available to

each of us, as well as to our community as a whole. That is the purpose of this site.</p>

<h2 align="justify"><u>A Synopsis</u></h2>
<h2 align="center"><strong>Representation of Study 329 2&deg; Endpoints</strong></h2>
<table cellspacing="0" cellpadding="0" border="0" align="center">           
<td rowspan="2"><img border="0" align="left" height="50" src="http://1boringoldman.com/images/uri-1.jpg" /></td>
<td>negative, Paxil did not separate from placebo</td>
<td>positive p&lt;.05</td>
<table cellspacing="0" cellpadding="4" border="0" align="left">                     
<td valign="top">Representation in JAACAP</td>
<td><img border="0" align="left" height="50" src="http://1boringoldman.com/images/uri-2.jpg" /></td>
<td align="center" colspan="2"><img border="0" height="36" src="http://1boringoldman.com/images/uri-7.jpg" /></td>
<td valign="top">Actual total of 2&deg; endpoints analyzed (based on the Study 329 trial documents themselves)</td>
<td><img border="0" align="left" height="150" src="http://1boringoldman.com/images/uri-3.jpg" /></td>
<td align="center" colspan="2"><img border="0" height="36" src="http://1boringoldman.com/images/uri-7.jpg" /></td>
<td valign="top">2&deg; endpoints prespecified in the original written protocol. Remainder added later.</td>
<td><img border="0" align="left" height="150" src="http://1boringoldman.com/images/uri-4.jpg" /></td>
<td align="center" colspan="2"><img border="0" height="36" src="http://1boringoldman.com/images/uri-7.jpg" /></td>
<td valign="top">Findings if statistical correction applied for testing of multiple endpoints.</td>
<td><img border="0" align="left" height="150" src="http://1boringoldman.com/images/uri-5.jpg" /></td>
<td align="justify" colspan="2">To be clear. This is what Ryan and Keller stated in JAACAP:<br />
                           &quot;Since our two primary outcome measures did not reach a p&lt;0.05 level of statistical significance, the more complex

question that  remains is whether or not we fairly interpreted the pattern of significant p values across a range of secondary endpoints as indicating

that paroxetine is better than placebo for adolescent depression.</td>
<td valign="top">Accompanied by this...</td>
<td><img border="0" align="left" height="50" src="http://1boringoldman.com/images/uri-2.jpg" /></td>
<td align="justify" colspan="2">With no other 2&deg; endpoints presented. This is the only representation that has appeared in JAACAP.</td>
<td valign="top">This is the actual pattern of p values across the range of 2&deg; endpoints analyzed in the study - never represented in JAACAP.</td>
<td><img border="0" align="left" height="150" src="http://1boringoldman.com/images/uri-6.jpg" /></td>
<td align="justify" colspan="2">That Ryan and Keller describe thus: &quot;Since we found a clear pattern of significant p values across secondary

analyses... we thought and still think this provides significant evidence of efficacy of paroxetine compared to placebo in adolescent


<p align="center">.................................</p>

<p align="justify">Please pass this site on to at least 2 other CAPs who have not yet seen it. By simply doing that you will have a real impact.
Through such concerted individual action we can highlight that we are a close knit group of professionals and information does not have to only travel

vertically within our community. Can highlight that, when considered worthwhile by many of us, we can communicate rapidly and effectively through

direct grassroots efforts. Any medical students, residents, or fellows interested in presenting the information on this site as part of a journal club

are welcome to contact us directly for supporting materials. We can be reached at infoallcapsunlocked*AT*gmail.com (replace the *AT* with @ to email


<h2 align="justify"><u>Bottom Line</u></h2>

<p align="justify">The original authors, the JAACAP editors, and AACAP governance, have assessed that the information summarized above not be passed on

to our CAP community. It is time we in the community make each other aware.</p>

<h2 align="justify"><u>Questions & Answers</u></h2>

<p align="justify"><strong><em>Should the original paper be retracted?</em></strong> That question is beyond the scope of this site. We believe

regardless of one’s views on retraction, the central question is whether CAPs should be even made aware of the information on this site.</p>

<p align="justify"><strong><em>Have serious concerns also been raised about the representation of the adverse event data?</em></strong> Yes. Equally

serious concerns have raised about the representation of the adverse event data.</p>

<p align="justify"><strong><em>Was the original paper ghostwritten?</em></strong> That depends on one’s definition of ghostwritten.  Perhaps more

useful is to detail how the manuscript was prepared: GSK had contracted with a medical writer to prepare a manuscript based upon Study 329. The first

draft of that manuscript to be submitted to GSK prior to distribution to the academic authors (reference 5,6,7). Draft was then provided to Martin

Keller (lead academic author along with Neal Ryan), and then distributed to the other authors (reference 8). The presentation of the secondary

endpoints, and adverse events, appears in the medical writer’s initial drafting largely as it does in the manuscript eventually submitted to JAACAP.


<p align="justify"><strong><em>Does all this concern me if I know that SSRIs are safe and effective for adolescents with depression?</em></strong>

That, if anything, they are underutilized? We would suggest very much YES. If one is looking to make use of an intervention as effective as possible,

the representation in our literature of information from clinical trials clearly remains important to us. This can be seen in how the original

representation of Study 329 information in JAACAP was used. At the time it was published the two main SSRI choices for adolescents with depression were

fluoxetine and paroxetine. The original representation was used to promote paroxetine as first line, in preference to fluoxetine, for adolescents with

depression (reference 9).</p>

<p align="justify"><strong><em>Does all this concern me if I know that treatment is highly individualized and heavily based on clinical experience?

</em></strong> We would continue to suggest very much YES. As an illustration: Even prior to the original representation of the adverse effect data in

JAACAP, clinicians had, based on their clinical experience, raised questions about the use of paroxetine over fluoxetine. Particularly within the

context of paroxetine’s very short half-life.  The original representation of adverse event data in JAACAP was used to attempt to counter these

concerns (reference 10).</p>

<p align="justify"><strong><em>What exactly do you mean that, for over 5 years, information has not been passed on to readers?</em></strong> Over the

past six years multiple letters regarding the selective representation of Study 329 have been submitted to Andres Martin for publication in JAACAP.

JAACAP has consistently declined to publish them. The editor in chief has explicitly stated that, aside from the question of retraction, that there is

no indication for any editorial action (Editorial action would seem to extend to allowing readership to be made aware of any of this information).

Correspondence is available upon request. The editor in chief has presented to the AACAP council that JAACAP editors have, on 4 separate occasions,

comprehensively reviewed the representation of Study 329 in JAACAP and determined that no editorial action was indicated.
Within AACAP, authority is invested in the president and council. The assembly, committees, and task forces all act within a purely advisory capacity.

The president and council have similarly, on multiple occasions, assessed to not allow information to be passed on regarding the selective

representation of Study 329.  This has included declining to provide ROCAPs the direct contact information for other ROCAPs. Correspondence is

available on request.</p>

<p align="justify"><strong><em>Is there a challenge to the editorial independence of the JAACAP editor in chief?</em></strong> NO. Editorial discretion

within JAACAP is clearly invested in the editor in chief and those decisions are independent of AACAP governance. Decisions regarding publication and

retraction are at his/her discretion, and not subject to change by AACAP governance. This is the standard of practice for professional


<p align="justify"><strong><em>So what are you requesting of JAACAP editors and/or AACAP governance?</em></strong> This website is designed directly

for the CAP community itself. To be reviewed and shared peer to peer. Effective dialogue across our entire national CAP community should not be solely

at the discretion of the JAACAP editor-in-chief and/or AACAP council.</p>

<p align="justify"><strong><em>All this seems hard to take at face value?</em></strong> We are not aware of any information that would call into

question the synopses provided above. The annotated references provided can clarify any points that seem uncertain, and we would encourage their review

by any interested CAPs. JAACAP has not shared with readers, or the broader CAP community, the details of any of their four reviews of the handling of

Study 329. We would welcome any additional information they would have to share. Additional information can be sent to infoallcapsunlocked*AT*gmail.com

(replace *AT* with @). Much more importantly, we would encourage broad review and discussion across our CAP community. It is the raison d'être of this

website. And, we would suggest, the best assurance of fair assessment.</p>

<p align="center">.................................</p>


<li><div align="justify">The Paper JAACAP published on Study 329 in 2001:<br>
<a target="_blank" href=" http://www.justice.gov/opa/documents-and-resources-july-2-2012-glaxosmithkline-gsk-press-conference">href="

See “complaint exhibit #2”</div></li>

<li><div align="justify">The last thing JAACAP published about Study 329 (in 2003). First draft from Ryan and Keller essentially as it later appeared

in JAACAP:<br>
<a target="_blank" href="http://dida.library.ucsf.edu/tid/zru38h10">href="http://dida.library.ucsf.edu/tid/zru38h10</a></div></li>

<li><div align="justify">2008 publication detailing some of the selective reporting in Study 329. Based on proprietary trial documents:
<a target="_blank" href="http://iospress.metapress.com/content/k36834543w9063rr/">http://iospress.metapress.com/content/k36834543w9063rr/</a><br>
<a target="_blank"



<li><div align="justify">2012 analysis of the impact of applying Bonferroni statistical correction:<br>
<a target="_blank" href="http://1boringoldman.com/index.php/2012/08/26/the-lesson-of-study-329-conventions-and-


<li><div align="justify">Proposal from medical writer for preparing drafts of Study 329 for journal submission:<br>
<a target="_blank" href="http://dida.library.ucsf.edu/tid/uru38h10">http://dida.library.ucsf.edu/tid/uru38h10</a></div></li>

<li><div align="justify">Submission of first draft from medical writer.<br>
<a target="_blank" href="http://dida.library.ucsf.edu/tid/otu38h10">http://dida.library.ucsf.edu/tid/otu38h10</a></div></li>

<li><div align="justify">Medical writer states she did not contact Keller during preparation of first draft:<br>
<a target="_blank" href="http://1boringoldman.com/index.php/2012/08/30/the-lesson-of-study-329-the-

<a target="_blank" href="http://1boringoldman.com/index.php/2012/11/27/paxil-study-352-whats-ghost-


<li><div align="justify">Draft from medical writer approved by Keller to be sent to the other authors for review:<br>
<a target="_blank" href="http://dida.library.ucsf.edu/tid/anu38h10">http://dida.library.ucsf.edu/tid/anu38h10</a></div></li>

<li><div align="justify">Attempting to market Paxil for depression in adolescents to get a head start on the competition (i.e., other SSRIs that would

be prescribed in this population):<br>
<a target="_blank" href="http://www.justice.gov/opa/documents-and-resources-july-2-2012-glaxosmithkline-gsk-press-

See “US complaint”. Most notably highlighted in p13 section 53 and p14-16 sections 55-62.</div></li>

<li><div align="justify">It is suggested that the shorter half-life of paroxetine results in a more favorable side effect profile than fluoxetine. This

within the same meeting where the same speaker was presenting the adverse effect data from Study 329 favorably.<br>
<a target="_blank" href="http://www.justice.gov/opa/documents-and-resources-july-2-2012-glaxosmithkline-gsk-press-

See “complaint exhibit #2” bottom of p2 “side effects” </div></li></ol>

<h2><u>Additional References</u></h2>

<div align="justify">A humorous illustration of why the selective presentation of analysis of multiple endpoint data can be problematic. Done with

stick figures and jelly beans:</div>
<div align="justify"><a target="_blank" href="http://xkcd.com/882/">href="http://xkcd.com/882/</a></div>
<div align="justify">Nice 2011 summary of why retraction might actually be warranted:</div>
<div align="justify"><a target="_blank" href="http://blogs.discovermagazine.com/neuroskeptic/2011/01/20/retract-that-

<div align="justify">Declaration in the British Medical Journal that the original data from Study 329 will be obtained, analyzed, and

<div align="justify"><a target="_blank"

<div align="justify">When the paper appears, it will likely be the first major clinical trial republished this way.</div>