ICD,DSM and The Tower of Babel

ICD Insight

In the best of worlds, psychiatry would have just one diagnostic system and it would be reliable, validated, derived from a deep understanding of psychopathology, and supported by biological tests. Instead, we have two different and competing diagnostic systems: the International Classification of Diseases (ICD) and the Diagnostic and Statistical Manual of Mental Disorders (DSM). Both are useful languages for communication – bridging the clinical/research interface and guiding treatment decisions. But both are also deeply flawed – unreliable in average everyday practice, largely unvalidated, and lacking in biological tests. And consensus has been impossible to achieve despite the obvious futility of competition.

The disconnect between the actual and the ideal is partly inherent, partly man made. Establishing diagnostic thresholds has been a surprisingly difficult conundrum across all medical specialties. Defining what is and what is not ‘disease’ is a puzzle, even when it comes to seemingly simple things like diabetes, hypertension, osteoporosis, and cancer. How much greater then the challenge for psychiatry given that the brain is so much more complicated than the body and that the boundary between normal and disorder is so frustratingly fuzzy when it comes to human behavior and symptoms.

But there is also no denying that those of us who have worked on the diagnostic systems could have done a better job of making them compatible. It would have been highly desirable to have one integrated system that is safe and scientifically sound – rather than two that are venturesome and subject to expert whim. We are stuck with the two methods because of an unhappy combination of differing bureaucratic priorities between the American Psychiatric Association and the World Health Organization; the financial conflict of interest occasioned by the enormous DSM publishing profits; and the clashing egos of the experts working on both sides. In a better world, these all could have easily been overcome – in the untidy world we live in, they posed what have turned out to be impossible obstacles.

The break between systems started in 1980 when DSM-III departed sharply from ICD-9. Like all international consensus documents, the successive ICDs had been rigidly conservative, producing sketchy definitions so inadequate they were largely ignored. In stark contrast, the revolutionary DSM-III captured the imagination of mental health professionals and the public; had an instantaneous and international impact on clinical practice and research; and became a perennial and profitable best seller. Admittedly, DSM-III also had its major downsides: it greatly oversold the reach of descriptive psychiatry; distracted attention away from the whole patient by focusing only on symptoms; disregarded psychological factors, social context, and family history; encouraged adventurism and diagnostic exuberance; and gave ammunition to misleading drug company marketing.

The World Health Organization tried to play catch-up in its next revision (ICD-10) by including its own sets of diagnostic criteria which were more or less equivalent to, but also, in minor but frustrating ways, different from those in DSM-III. By design, we began our own work on DSM-IV shortly after work began on ICD-10 – the goal being to bring the two systems into closer alignment. For the most part, we failed. Although the ICD and DSM experts met often, their respective decisions were not as aligned as hoped or anticipated. Partly this was due to their separate starting points of reference – ICD was modeled on DSM-III which was then widely used around the world, whereas DSM-IV was meant to stay close to DSM-III-R, the system in place in the United States. Partly the failure to integrate was due to differing institutional priorities. ICD-10 wanted to be useful to a much more heterogeneous set of providers across different countries, health systems and cultures; to use experts from all over the world; and to make its own mark on psychiatric diagnosis. DSM-IV was meant to stabilize a diagnostic system that was changing too fast. Some of the differences were purposeful (i.e. where there was an honest difference of opinion and there was hope that future research might resolve the issues). Unfortunately, there were also many minor but confusing differences between systems that were the result of the stubborn expert groups on each side who stuck to their guns in defending their differing preferred wordings.

Another recent opportunity to harmonize systems arose with the more or less simultaneous development of DSM-5 and ICD-11. Once again, the opportunity has been lost. The proper path toward what would have been a happy integration was completely obvious. Each system had a basic flaw that the other could help correct – working in concert the two together would be much more effective than either could possibly be alone. The ICD-10 flaw was the failure of its research version containing diagnostic criteria sets to attract any interest. Journals throughout the world consistently threw their weight behind DSM-IV and researchers everywhere routinely used it as their system of choice. Responding appropriately to this fact, ICD-11 has opted to drop any effort to provide definite criteria for defining disorders and will instead substitute a simpler approach that it hopes will be more useful for those busy clinicians who are frustrated by the DSM flaw of being very complicated for use in routine practice. The ideal melding of the two systems would have been to make them modular – the simple ICD-11 definitions would capture the same constructs that were elaborated in the more detailed DSM-5 criteria sets. The systems could have been completely compatible – with ICD-11 more useful where speed and simplicity were necessary; DSM-5 more useful for research and for clinical settings that allowed for more precise diagnosis. Although there were superficial gestures toward ‘harmonizing’ DSM-5 and ICD-11, the same bureaucratic, financial, and ego forces have prevented any meaningful integration of systems.

DSM-5 has been widely regarded as a fiasco. Will ICD-11 learn from its mistakes and do any better? It is still too early to tell – but I have my doubts. It is hard to believe that ICD-11 can produce an excellent product because it suffers from a terrible resource problem. If DSM-5 was cursed by being ridiculously over-funded, ICD-11 is cursed by being ridiculously underfunded. The bloated investment of an astounding US$25 million dollars spent on DSM-5 (for comparison, DSM-IV cost US$5 million) promoted excessive ambition, overselling, and a premature rush to press to rescue the American Psychiatric Association from its budget deficit. ICD-11 has suffered from the opposite problem of having almost no funding. It is mostly a volunteer effort that may be at the whim of its own experts and could easily suffer from the same idiosyncratic decision making that has doomed DSM-5.

What’s wrong with allowing the experts to decide what goes into the diagnostic system? A lot. Experts have an inherent intellectual conflict of interest that inclines them to want to expand their pet area of study. They always worry about the harms to the missed, false-negative patient and never worry about the harms to the mislabeled, false-positive non-patient. Experts imagine the diagnostic system will be used in the careful and expert way they themselves would use it and are remarkably naïve about how badly it will be misused in actual clinical practice by clinicians who are much less expert, rushed for time, and influenced by powerful external forces (especially aggressive and misleading drug company marketing). You certainly do need experts to contribute to the development of the diagnostic system, but you can’t give them free rein to expand it with their unrealistic and untested dream lists. The diagnostic manual should shoot for safe rather than sorry and should protect itself from new ideas that are not ready for prime time.

So what is my advice for ICD-11? First, learn from the swirl of controversy surrounding the changes introduced by DSM-5. Don’t turn normal grief into Major Depressive Disorder; normal temper tantrums into Disruptive Mood Dysregulation Disorder; normal forgetting in old age into Minor Neurocognitive Disorder; poor eating habits into Binge Eating Disorder; and expectable worry about physical symptoms into Somatic Symptom Disorder. Second, recognize that we are suffering from diagnostic inflation and a terrible misallocation of scarce resources – over-diagnosing and over-treating normals who don’t need it, while shamefully neglecting the seriously ill who desperately do and third, tighten the diagnostic system rather than loosen it and provide black box warnings for those disorders that are difficult to diagnose accurately and are currently being carelessly over-diagnosed (e.g. Attention Deficit Disorder, Autism, Bipolar Disorder, Mild Major Depressive Disorder, and Post-Traumatic Stress Disorder).

Will any of this advice be heeded? I am not holding my breath. ICD-11 is probably too far along its way, too little motivated, and too little resourced for any last minute save. The experts will most likely include all sorts of untested and unsafe suggestions and we will be stuck with two systems even more flawed and unsafe than their respective predecessors.

Is there hope that future psychiatric diagnosis will come to its senses and become more useful? Yes, but it will be a slow and uphill slog. Don’t count on any quick and pervasive breakthroughs on the neuroscience front. If translational research from basic science to clinical practice were easy, the remarkable neuroscience and genetics revolutions of the past 30 years would have already made some important leaps across the gap. Instead, despite the exponential increase in knowledge of how our brains and our genes work, psychiatric practice is about the same as it was 30 years ago. A diagnostic system based on biological understanding and biological tests will take decades of very gradual, piecemeal advances. We may soon have a battery of tests for Alzheimer’s, but schizophrenia will take many decades to crack and likely will eventually be divided into dozens, or even hundreds, of different disorders.

In the near future, the main hope for taming diagnostic inflation in psychiatry is the growing reaction against over-diagnosis in all of medicine sponsored by the British Medical Journal (and strongly supported in Australia by work done at Bond University). Psychiatric diagnosis has followed the pack of the rest of medicine in becoming far too exuberant. As medicine comes to its senses, so hopefully will psychiatry.