To the Membership of the APA

I sent the letter that begins on page 4 to the Trustees of the APA on April 8, 2010. It contains an urgent plea that the Trustees move immediately to correct the increasingly wayward course of DSM5. The DSM5 Task Force is about to begin a field trial that is a complete mistake:

I sent the letter to the Trustees of the APA on April 8, 2010. It contains an urgent plea that the Trustees move immediately to correct the increasingly wayward course of DSM5. The DSM5 Task Force is about to begin a field trial that is a complete mistake:

1. It is unnecessarily expensive (a self-financed $2.5 million or more).

2. It is unnecessarily complicated and will cause additional delays.

3. It fails to address the most important questions concerning the impact of proposed DSM5 changes on prevalence rates and on false-positive diagnoses.

4. It is so poorly designed that the results are very likely to discredit psychiatric diagnosis.

I write directly to you now because this may be a last chance tipping point to save DSM5-and because the APA leadership has so far proved itself consistently unable to provide adequate supervision. You need to be informed of the many crippling DSM5 problems before it will be too late for you to influence your leaders to take the decisive actions needed to solve them.

Dear Trustees:

Nine months ago, Dr Robert Spitzer and I wrote to alert you that DSM5 had gone badly off track. We warned that its process was unsupervised, poorly planned, secretive, disorganized, and was falling far behind schedule. You took the appropriate steps of appointing an Oversight Committee and delaying for 1 year the target dates for field trials and for the publication of DSM5.

Unfortunately, this extra year is being used very inefficiently and a major disaster lurks just ahead. The first DSM5 draft is of surprisingly poor quality, making it questionable whether a usable manual can emerge even within the expanded time line. And the news gets worse. The recently posted plans for field trials are a prescription for disaster. The project will cost a fortune, cause further delays, ask the wrong questions, and produce embarrassing results.

If you are to save the day, you must first fully understand just how serious the defects are in the DSM5 products and future plans:

1. By now, the criteria sets should have been carefully edited to ensure complete clarity and consistency. Inexplicably, after all this time, they remain in a rough and disorganized state, not close to being fit for the field testing that is about to begin. DSM5 urgently needs a thorough and painstaking editing that will eliminate all its many errors, ambiguities, and inconsistencies. It is apparent that no one working on DSM5 knows how to write or edit clean criteria. [Editor’s Note: Please see “The DSM5 Draft: Can the Poor Writing Be Salvaged?” which is posted on the DSM5 section of]

2. The failure to do any risk/benefit analysis has allowed the survival of many quite far out suggestions that would cause disastrous unintended consequences. These are an ongoing distraction and put DSM5 and the field in a bad light. The worst of the proposals should have been eliminated much earlier and should be washed out now. [Please see “The Missing Risk/Benefit Analyses for DSM5” on our Web site.]

3. The failure to do a forensic review has allowed the survival of a number of changes that will create nightmares for forensic psychiatrists and for the legal system. [Please see]

4. Everything about the proposed DSM5 field trial design is completely wrongheaded. The project is shockingly overpriced (my guess is at least between $2 and $3 million). It creates an unnecessarily complex logistic quagmire (3000 subjects, 3 assessments per subject, 10 different centers with 10 different IRBS, 3 separate clinicians involved in the evaluations, videotaping 20% of the interviews, etc) If the field trials get very far behind schedule (trust me, they definitely will), the APA will likely yet again be forced to delay publication of DSM5. But worst of all, the field trial is a risky red herring that will answer a question that nobody is raising-ie, what is the reliability of psychiatric diagnosis? As Dr Kupfer himself stated (in his 2008 article “On the Road to DSM-V and ICD-11” in the European Archives of Psychiatry and Clinical Neuroscience), “The reliability of DSM as a clinical tool has been upheld but less emphasis has been given to its validity.”

The DSM5 field trial design will unnecessarily reopen the reliability question and come up with an embarrassing answer-low reliabilities caused by its unfocused design and ambiguously written criteria sets. Meanwhile, the questions that do count have been completely ignored-ie, what are the effects of DSM5 changes on rates, false positives, and diagnostic validity? The failure to include comparisons with DSM-IV makes it impossible to know what impact the changes will have on caseness and prevalence. The failure to do these assessments in nonclinical populations (which could be done by telephone in an inexpensive random-digit-dialing paradigm) makes it impossible to assess the risk of false positives.

These are the burning questions that must be answered if the APA wants to be perceived by the field and the public as a responsible and careful custodian of the official diagnostic system. Unless revised, this mammoth endeavor will be a huge mistake-a waste of time and money, expended on a largely irrelevant question that will likely produce embarrassing results, and put the whole enterprise at risk. [Please see “The DSM5 Field Trial Proposal-An Expensive Waste of Time” on our Web site.]

5. This one golden opportunity to fully harmonize with ICD-11 has not been given any priority and may soon be lost.

6. The hardest work looms ahead . . . the enormously labor-intensive task of writing the DSM5 text. The prior DSM5 failures in writing, in editing, and in meeting deadlines suggest a poor prognosis for a readable text, done well and done on time. The DSMs have become popular educational texts-but who will want to buy a manual written in the obscure language that makes the existing DSM5 publications so painful to read?

7. Although the projected publication date for DSM5 is 3 years away, time is quickly running out given the many things that have yet to be done and the extremely slow rate of progress in the work that has been done to date. If action is not taken now, further publication delays are likely.

How can you save DSM5?

1. Ensure that there is an immediate, thorough, and skillful editing of the proposed DSM5 criteria sets. Because no one currently working on DSM5 knows how to write or edit diagnostic criteria, it is imperative that outside help be recruited without further delay. Among all the many problems facing DSM5, developing clean criteria is the highest priority. If the criteria remain a mess, DSM5 will not be useful to clinicians, researchers, educators, forensic evaluators, or the wider public.

2. Appoint 3 subcommittees that report to the Oversight Committee and that are assigned, respectively, to risk/benefit analysis, forensic review, and supervising the field trials. There needs to be a mechanism to guarantee that subcommittee input is given sufficient weight, since some work groups may be emotionally invested in pet ideas that cannot stand up to a searching external review.

3. The field trial design needs to be reviewed and radically revised. My understanding is that there was no prior external review of this misguided design by the DSM5 Task Force, the Oversight Committee, or experts in the field. Developed in isolation and at the last minute, the field trial design has the wrong goals, the wrong methods, the wrong settings, the wrong time lines, the wrong price-and it will obtain the wrong results. If done as currently planned, these field trials could very likely sink DSM5 by achieving artificially low reliabilities. The Oversight field trial subcommittee should be put in charge of the redesign and its implementation.

4. Abandon the rating scales project. There is no chance that clinicians will ever use these ad hoc, untested, and cumbersome scales. Instead of wasting time and money in hurried field testing, those few scales that have some promise should be included in the appendix and test-ed properly over the next few years. This project never made sense and clearly requires time, money, and skill far beyond what is available. Continuing to do it poorly is a wasteful distraction and a drain on already overstretched resources that should be devoted to all the really essential tasks that to date have received so little careful attention. [Please see “Rating Scales: DSM5 Bites Off Far More Than It Can Chew”on our Web site.]

5. Ensure that work begins immediately on the DSM5 text. In our DSM-IV experience, writing the text was by far the hardest and most time-consuming task in the entire project. A last minute rush will re-sult in predictable poor quality, more publication delays, or more probably both.

6. Make a serious attempt now to harmonize DSM5 with ICD11. This is a surprisingly easy task, would be the most useful legacy of DSM5, and could spare the world a lot of unnecessary confusion. The previous facile lip service on this issue is no substitute for a real effort.

The worried predictions Dr Spitzer and I made last year did not require any special gifts of clairvoyance-they were perfectly obvious projections based on our prior experiences. At the time, the DSM5 leadership angrily contradicted the validity of our concerns and provided misleading assurances that all was well with DSM5. It soon became clear that they were either unaware of, or were actively ignoring, the many serious and stubbornly recurrent problems in planning and execution. Not surprisingly, all our predictions were soon and convincingly confirmed by the continued secrecy, consistent delays, disorganization, and extremely poor-quality product that have characterized every aspect of the work on DSM5. Recognizing this, you stepped in to appoint the Oversight Committee, postponed field trials until the draft criteria could be reviewed by the field, and finally delayed the publication date.

All were tough but necessary decisions. Unfortunately, they were not enough. The state of DSM5 was not promising last year, but it is in much more trouble now. I am convinced that my current pessimistic projections are well grounded and will almost certainly come to sad fruition unless you rise to the occasion and take immediate and decisive action. Without your firm input, DSM5 is likely to be a continuing embarrassment to psychiatry and to the APA.

Extrapolating forward from what has been accomplished so far, a quality DSM5 will not be produced on time and may never be produced at all. Your fiduciary responsibility to the Association, to our field, and (most of all) to our patients requires that you quickly assume a much more active supervisory role. This is probably the last available nodal point when your action can save the day. Once the totally misguided field trials are under way, there will be a strong inertial momentum to press forward and turn out what will quite likely be an unusable and unread- able manual. The riskiest position you can possibly take is to remain complacent and let things drift further over the cliff. The power and the responsibility lie in your hands. Please use it before it is too late.

Best wishes and good luck in your difficult deliberations.

Allen Frances, MD Professor Emeritus of Psychiatry Duke University School of Medicine



1. Rettew DC, Lynch AD, Achenbach TM, et al. Meta-analyses of agreement between diagnoses made from clinical evaluations and standardized diagnostic interviews. Int J Methods Psychiatr Res.2009;18:169-184.