We can take one further step toward finding common ground in my ongoing debate with Drs. Pies and Zisook. Dr. Pies has helpfully pointed the way in his latest piece, which can be found in the comments section here.
We agree that the problem is the fuzzy boundary between normal grief and Major Depressive Episode (MDE). It is at this mild end that our dispute lies. There is no dispute at the severe end -- severe depressive symptoms during bereavement are already meant to be diagnosed as MDE using the current guidelines in DSM-IV-TR. So the test case is someone who has lost a spouse or child and has just two weeks of sadness and loss of interest, appetite, sleep, and energy. Such a person would have to be diagnosed with MDE if we were to follow the DSM-5 suggestion to simply remove the Bereavement exclusion.
Dr. Pies and I both disagree with the DSM-5 suggestion that two weeks is a long enough duration. There is no research suggesting that the distinction between normal grief and mild MDE can be reliably and validly made so early after the loss of a loved one in a griever with such mild and ubiquitous symptoms. This level of mild symptoms for this short a duration occurring in the immediate aftermath of a loss are just too common and too compatible with normal grief to be considered MDE.
Dr. Pies and I also agree that there is no need to medicate this griever having mild and completely expectable symptoms for so short a period. We would both instead recommend a combination of commiseration, empathy, support, and watchful waiting to see if the person goes on to have a normal evolution of grief (most will) or goes on to have enduring or more severe symptoms compatible with MDE.
Dr. Pies worries more than I do that the current DSM-IV-TR criteria are too stringent and create a false negative problem- i.e. missing MDE during bereavement and withholding appropriate treatment from those who need it. He would make it easier to get an MDE diagnosis during bereavement than is currently possible using DSM-IV-TR, but would require much more stringent standards for its diagnosis than are being suggested for DSM-5. Dr. Pies suggests a one-month duration (not two weeks) and perhaps a higher threshold of symptom severity.
Dr. Pies' suggestions are a reasonable way to balance the risks of false negatives vs. false positives. I worry much more than he does about false positive overdiagnosis and overtreatment. So I would prefer to stay where we are, but I could not strongly disagree with the Pies/Zisook suggestion. I would add to it just two exclusions. The diagnosis of MDE may be inappropriate if the individual's culture calls for a more profound expression of grief or if the individual has a personal history of a deep, but self-limited, grief in the past.
So, where do we continue to disagree? I believe that to remove the Bereavement exclusion for MDE would disastrously open the floodgates to the misdiagnosis and overtreatment of normal grief (especially by hurried primary care physicians who do much of the prescribing). To me, the DSM-5 suggestion, as it stands now simply doesn't fly at all and is a public health threat. In contrast, Dr. Pies is willing to hold his nose and swallow a DSM-5 suggestion that he acknowledges will confuse grief with MDE. He invokes the image of saved suicides to defend his persevering effort to not miss any MDE patient, regardless of the negative consequences.
I analyze the benefits and risks of the DSM-5 proposal quite differently. I don't see much benefit because I don't think the current rules result in many missed cases. DSM-IV-TR already allows extremely wide play for clinical judgment in deciding between grief and MDE. A clinician is welcomed to make an MDE diagnosis whenever he believes the specific circumstances warrant it, even with relatively mild symptoms in a grieving person (say, someone who has had previous depressions or previous prolonged grief). If the grief symptoms are severe, DSM-IV-TR insists that MDE be diagnosed. Clinicians do not find impediments to the appropriate diagnosis under the current system. It is an unproven and unlikely red herring that the DSM-5 change will save lives (or for that matter have any beneficial effect at all).
But, as detailed in several previous posts, I believe the risks of the suggested change are great. I won't go through all the specific arguments yet again, but DSM-5 is promoting a needless expansion of psychiatric diagnosis that would reduce the dignity of grief, create insurance and job stigma, and result in unnecessary, expensive, and potentially harmful treatment.
There is no proven efficacy for medication treatment given after just two weeks of mild symptoms in grievers. My guess is that the placebo response rates would be so high in this population that no active medication efficacy could ever be demonstrated. But medications do have side effects (including increasing suicidal symptoms in some people).
My suggestion to Pies/Zisook is to not give up the fight with DSM-5 for an appropriate one-month duration requirement for milder MDE -- if not in all situations (my choice and theirs) then at least for MDE during bereavement. Swallowing a spoiled half loaf can be bad for the nation's health. We can assume that drug advertising to patients and marketing to doctors will result in a flood of treatment that they would also question. This should worry Pies/Zisook more than it does and stiffen their resolve to fight the good fight (with an admittedly unreasonable DSM-5) to require a one-month duration of symptoms.
So let's be clear about a reasonable compromise. Reasonable people can disagree about the precise duration requirement before grief can be considered mild depression. Pies/Zisook suggest one month. I am OK with the DSM-IV-TR two months. Some people think it should be longer. No one but DSM-5 is suggesting two weeks, which seems like a really bad idea.
This ongoing Talmudic debate with Pies/Zisook has been fruitful. A similarly careful risk/benefit analysis should be, but is not, occurring for all of the many questionable DSM-5 proposals.
One of the most important distinctions in all of psychiatry is often also the most difficult to make: Is the patient's depression part of a bipolar or a unipolar course of mood disorder? This is so consequential a decision because treating the depression of bipolar disorder with antidepressants can trigger problematic irritability, mood swings, and rapid cycling. To reduce this risk, patients receiving antidepressants for bipolar depression usually also receive either a mood stabilizer or an antipsychotic (or too often both). But the reduced risk of mood swings conferred by the covering medication comes at a potentially heavy cost in side effects and complications (especially dangerous weight gain and diabetes). The tough question is where to draw the diagnostic line between bipolar and unipolar mood disorder in a way that best balances the risks of taking vs. not taking the covering medication.
The trend in diagnostic habits over the past 25 years has been clear. The boundaries of bipolar disorder have expanded widely at the expense of unipolar; the prevalence ratio between them is now about 1:3 as opposed to the previous 1:6. This has caused a markedly increased use of mood stabilizing and antipsychotic drugs; they have protective effects in those who need them, but harmful side effects in those who may not.
Two factors account for this move toward bipolar (and away from unipolar) disorder. The first was the introduction by DSM-IV of a new official category (Bipolar II) that expanded the bipolar category into unipolar territory. Bipolar II describes patients who have depressions that alternate with hypomanic episodes (rather than with the manic episodes of classic Bipolar Disorder). Hypomanic episodes are milder and often briefer versions of classic mania and are therefore very much more difficult to diagnose. This is complicated further by the fact that brief periods of elevated mood can be provoked by drugs or medications. And it is often hard to distinguish the elevated mood of hypomania from a temporary return to normal mood in someone who has become accustomed to always feeling depressed.
Patients with depression and hypomania are at the boundary separating bipolar and unipolar disorder. They could have been classified in either camp. We made the decision to describe them as Bipolar II in DSM-IV because the weight of the course, family history, and treatment evidence suggested that they sorted better with Bipolar Disorder. This was not an easy decision. We had to balance: 1) concerns that patients with bipolar tendencies would be harmed iatrogenically if they received antidepressants without coverage, against 2) concerns that some patients identified as Bipolar II were really unipolar and would receive the added burden of unnecessary and potentially harmful medication. This was a close call, but on balance it seemed safer to include Bipolar II as a new category. I still think this made sense, although the extent of the bipolar fad that followed was certainly surprising and remains a serious concern.
We perhaps should have, but did not, anticipate how intense and effective the marketing and "education" campaign mounted by the pharmaceutical industry would be, alerting psychiatrists, primary care doctors, other mental health workers, and patients about the perils of previously "missed" bipolar disorder. This led to more accurate diagnosis and safer treatment for many patients, but like all fads, it overshot and has led to unnecessary medication for others who are now misdiagnosed as bipolar on very flimsy grounds.
This brings us to two problematic changes being considered by the DSM-5 mood disorders work group: 1) allowing hypomania to be diagnosed just on the basis of increased energy/activity (no longer requiring the presence of elevated mood or irritability), and 2) reducing the duration requirement for a hypomanic episode (now set at four days by DSM-IV). As is usual with all the changes being considered for DSM-5, the lowered thresholds are meant to avoid missed diagnoses, but this seems far outweighed by the risks of overdiagnosis of Bipolar II Disorder and the consequent overtreatment with harmful medications.
We must accept that there is no currently available method that infallibly distinguishes bipolar from unipolar disorder at their boundary. Until we have a much deeper understanding of the different pathogeneses of unipolar vs. bipolar disorder, this crucial distinction must be based exclusively on clinical judgment using the admittedly arbitrary DSM rules of thumb defining the symptoms and duration of a hypomanic episode. Any decisions lowering the symptom and duration requirements for a hypomanic episode will feed what is already a bipolar fad.
Sometimes you spot a serious problem and figure out a very well-intended solution, only to discover eventually that your solution created as much trouble as the original problem. The workers on DSM-5 have spotted an enormously worrisome problem: the wild overdiagnosis of childhood Bipolar Disorder, which has led to a massive increase in the use of antipsychotic and mood stabilizing medications in children and teenagers.
Something clearly needs to be done to correct this. The proposed DSM fix -- a new diagnosis called "Temper Dysregulation with Dysphoria" -- is meant to provide a less risky diagnostic home for the kids currently misdiagnosed as bipolar. Unfortunately, temper dysregulation is just the kind of unhappy solution that you later come to greatly regret. It is a makeshift proposal, with considerable risks, and a disqualifying conceptual problem.
How is it makeshift? As the DSM-5 work group candidly admits, the research evidence on this new diagnosis is remarkably thin, based on the very recent work of just one research group. Virtually nothing is known about its likely prevalence in the general population, whether it can be distinguished reliably from normal developmental or situational irritability, its relation to other disorders that present with temper outbursts, its course, its preferred treatments, or the tradeoff between treatment response and adverse complications.
The criteria set for temper dysregulation require: 1) severe recurrent temper outbursts in response to common stressors; 2) occurring at least three times a week; 3) for at least a year; 4) in someone who is persistently irritable.
The definition was created largely ad hoc with no systematic testing (outside of that one research program) of the performance characteristics of the items to determine how they would play if in wide general use. It is loosely written and in fact contains no exclusion for use in adults (which I assume is an oversight).
Why is such a makeshift solution being given any serious consideration? The work group freely admits that the scientific rationale is completely inadequate. Their proposal rests exclusively on two real and pressing clinical needs: 1) to reduce the overdiagnosis and overtreatment of bipolar disorder; and, 2) to do something to help the considerable suffering that these temper outbursts cause the children themselves, their parents and teachers, and society at large.
But this attempted fix is itself highly risky and likely to set off its own cascade of unfortunate unintended consequences. The biggest problem with the proposal is that it is not nearly restrictive enough. While trying to rescue kids who are now misdiagnosed as bipolar, it will undoubtedly open the door to the misdiagnosis of normal kids who happen to be temperamental or in difficult family circumstances. Every effort must be made to distinguish mental disorder-level temper problems (that cause clinically significant distress and impairment) from those that are within the limits of a normal, but difficult temperament experiencing the aches and pains of growing up.
This is hard to do. First off, there is enormous variability in what are considered appropriate expressions of temper across kids, across developmental periods, across families, and across subcultures. The definition of "severe" will likely vary greatly depending on the tolerance of the clinician, family, school, and peer group. The "stressors" that trigger the episodes may be minimal in some cases, remarkably provoking of readily understandable temper reactions in others. Family fights that are based in interpersonal problems will be translated into individual psychopathology. Finally, in the heat of battle, it will be forgotten that kids often do outgrow a developmentally or situationally triggered temperamental period in their lives.
In expert hands, it is conceivable that the diagnosis might be contained to achieve just its intended goal of reducing the diagnosis of bipolar disorder, but in the real world many diagnoses are made by primary care clinicians who have limited expertise in psychiatry, little time with each patient, are dealing with harried family members who want a quick solution to a pressing problem, and are influenced by drug company salespeople. My experience tells me that this makeshift diagnosis may well become very popular and will spread to normal kids who would do a lot better without treatment.
Temper dysregulation also shares very difficult-to-define boundaries with about a dozen other mental disorders. It would be excluded in the differential diagnosis with Major Depressive, Dysthymic, Bipolar, Schizophrenic, Autistic, Separation Anxiety, and Post-traumatic Stress Disorders. It would be allowed to coexist with Oppositional, Conduct, Attention Deficit, and Substance Use, creating an artificial comorbidity with them and raising the risk of unnecessary and inappropriate treatment. For example, many drug-abusing teenagers will meet the criteria for temper dysregulation. Most often, the preferred intervention should be dealing with the substance problem, not adding a medication.
Which brings us to the risks of treatment for this prematurely concocted diagnosis. Unfortunately, it is inevitable that this will often consist of atypical antipsychotic drugs because these are heavily marketed and may be helpful in reducing some forms of explosive temper outbursts. But their beneficial effects in some must be balanced against their very great dangers when widely used for the many. These medications often cause enormous and rapid weight gains, increasing the risk of diabetes, medical complications, and reduced life span. Their use in severely disturbed kids raises its own set of serious clinical and ethical questions, but it can be justified in extremely exigent circumstances. Their use in kids who are having disturbing (but essentially "normal") developmental or situational storms or are irritable for other reasons (e.g. substance use, ADD) would be disastrous, but it will happen and probably often.
What is the conceptual problem? Temper dysregulation describes a single symptom, not a complete syndrome. It is at the same level of abstraction as fever or suicide -- it can occur as part of the presentation of any number of disorders, but is not a separate disorder in and of itself. In the distant past, psychiatry used to classify by symptoms, but found this to be cumbersome and much less satisfying than our current serviceable, if limited, syndromal approach.
What is the solution? Severe, recurring, and persistent temper outbursts are a crucially important clinical and societal problem and deserve special attention. But given the limited state of current knowledge, it is premature to regard Temper Dyregulation as an independent coherent syndrome that would solve more problems than it would cure. There are two far better ways for DSM-5 to attempt to address this problem.
DSM-5 could add a specifier, "With Explosive Outbursts," whenever this is a prominent part of the presentation of all of the dozen or so conditions in the differential diagnoses. Or it could add the specifier just to Oppositional Defiant Disorder and attempt to use this as a default away from Bipolar Disorder. Or "temper problems" could be one of the cross cutting dimensions, if these make it into DSM-5. Each of these possible conventions would make the point that temper outbursts are (again, like fever or suicidal thoughts) nonspecific symptoms of many disorders, not a specific and separate disease.
The criteria and text of DSM-5 should be as commendably clear as the drafts are that there are serious problems caused by the current practice of over diagnosis of childhood bipolar disorder. Specific guidelines should be recommended that would attempt to preempt the loose non-criteria-based diagnostic habits that have recently gained such unwarranted purchase.
These steps alone will, of course, not solve the dangerous "epidemic" of childhood bipolar disorder. There are limits to what DSM-5 can do in this regard. The NIMH and FDA may have to step up their involvement. There is a pressing need for an educational campaign to professionals and to the public to highlight the risks of overuse of atypical antipsychotics and to recommend caution in diagnosis and treatment of kids with temper outbursts. Much more research funding should be directed to this area. Funding for the study of explosive behavior has been inadequate in the past because it is an inherently difficult research topic, but the huge public health significance of widespread antipsychotic use should now give it a priority.
We can take one further step toward finding common ground in my ongoing debate with Drs. Pies and Zisook. Dr. Pies has helpfully pointed the way in his latest piece, which can be retrieved here.
We agree that the problem is the fuzzy boundary between normal grief and Major Depressive Episode (MDE). It is at this mild end that our dispute lies. There is no dispute at the severe end - severe depressive symptoms during bereavement are already meant to be diagnosed as MDE using the current guidelines in DSM-IV-TR. So the test case is someone who has lost a spouse or child and has just two weeks of sadness and loss of interest, appetite, sleep, and energy. Such a person would have to be diagnosed with MDE if we were to follow the DSM-5 suggestion to simply remove the Bereavement exclusion.
Dr. Pies and I both disagree with the DSM-5 suggestion that two weeks is a long enough duration. There is no research suggesting that the distinction between normal grief and mild MDE can be reliably and validly made so early after the loss of a loved one in a griever with such mild and ubiquitous symptoms. This level of mild symptoms for this short a duration occurring in the immediate aftermath of a loss are just too common and too compatible with normal grief to be considered MDE.
Dr. Pies and I also agree that there is no need to medicate this griever having mild and completely expectable symptoms for so short a period. We would both instead recommend a combination of commiseration, empathy, support, and watchful waiting to see if the person goes on to have a normal evolution of grief (most will) or goes on to have enduring or more severe symptoms compatible with MDE.
Dr. Pies worries more than I do that the current DSM-IV-TR criteria are too stringent and create a false negative problem- ie missing MDE during bereavement and witholding appropriate treatment from those who need it. He would make it easier to get an MDE diagnosis during bereavement than is currently possible using DSM-IV-TR, but would require much more stringent standards for its diagnosis than are being suggested for DSM-5. Dr. Pies suggests a one month duration (not two weeks) and perhaps a higher threshold of symptom severity.
Dr. Pies' suggestions are a reasonable way to balance the risks of false negatives vs false positives. I worry much more than he does about false positive overdiagnosis and overtreatment. So I would prefer to stay where we are, but I could not strongly disagree with the Pies/Zisook suggestion. I would add to it just two exclusions. The diagnosis of MDE may be inappropriate if the individual's culture calls for a more profound expression of grief or if the individual has a personal history of a deep, but self limited, grief in the past.
So, where do we continue to disagree. I believe that to remove the Bereavement exclusion for MDE would disastrously open the floodgates to the misdiagnosis and overtreatment of normal grief (especially by hurried primary care physicians who do much of the prescribing). To me, the DSM-5 suggestion, as it stands now simply doesn't fly at all and is a public health threat. In contrast, Dr. Pies is willing to hold his nose and swallow a DSM-5 suggestion that he acknowledges will confuse grief with MDE. He invokes the image of saved suicides to defend his persevering effort to not miss any MDE patient, regardless of the negative consequences.
I analyze the benefits and risks of the DSM-5 proposal quite differently. I don't see much benefit because I don't think the current rules result in many missed cases. DSM-IV-TR already allows extremely wide play for clinical judgment in deciding between grief and MDE. A clinician is welcomed to make an MDE diagnosis whenever he believes the specific circumstances warrant it -- even with relatively mild symptoms in a grieving person (say someone who has had previous depressions or previous prolonged grief). If the grief symptoms are severe, DSM-IV-TR insists that MDE be diagnosed. Clinicians do not find impediments to the appropriate diagnosis under the current system. It is an unproven and unlikely red herring that the DSM-5 change will save lives (or for that matter have any beneficial effect at all).
But, as detailed in several previous posts, I believe the risks of the suggested change are great. I won't go through all the specific arguments yet again, but DSM-5 is promoting a needless expansion of psychiatric diagnosis that would reduce the dignity of grief, create insurance and job stigma, and result in unnecessary, expensive, and potentially harmful treatment.
There is no proven efficacy for medication treatment given after just two weeks of mild symptoms in grievers. My guess is that the placebo response rates would be so high in this population that no active medication efficacy could ever be demonstrated. But medications do have side effects (including increasing suicidal symptoms in some people).
My suggestion to Pies/Zisook is to not give up the fight with DSM-5 for an appropriate one month duration requirement for milder MDE -- if not in all situations (my choice and theirs) then at least for MDE during bereavement. Swallowing a spoiled half loaf can be bad for the nation's health. We can assume that drug advertising to patients and marketing to doctors will result in a flood of treatment that they would also question. This should worry Pies/Zisook more than it does and stiffen their resolve to fight the good fight (with an admittedly unreasonable DSM-5) to require a one month duration of symptoms.
So let's be clear about a reasonable compromise. Reasonable people can disagree about the precise duration requirement before grief can be considered mild depression. Pies/Zisook suggest one month. I am OK with the DSM-IV-TR two months. Some people think it should be longer. No one but DSM-5 is suggesting two weeks- which seems like a really bad idea.
This ongoing Talmudic debate with Pies/Zisook has been fruitful. A similarly careful risk/benefit analysis should be, but is not, occurring for all of the many questionable DSM-5 proposals.
On August 15, I published an op-ed piece in The New York Times expressing the view that normal grief is normal and should not be confused with Major Depressive Disorder (MDD). The DSM-5 suggestion to remove the bereavement exclusion for MDD would convert grief after losing a loved one into mental disorder. Two short weeks of expetable symptoms like sadness, insomnia, difficulty working, and loss of interest, appetite, and energy would qualify for an MDD diagnosis. This mislabeling would then often trigger stigma and unnecessary medication treatment. More details can be found in the op-ed piece itself or on previous numbers of this blog.
On August 20, the Times published a number of letters taking all sides on the issue. There were two rejoinders to my view that I believe are misleading enough to require comment:
Counter Argument 1: Patients experiencing a well-established Major Depressive Episode (MDE) beginning during bereavement are no different in presentation and treatment response than those whose MDE follows after other severely stressful life events.
Reply: True enough, but totally irrelevant to my concern. Well established MDD is not in question (it is already diagnosable in DSM- IV-TR). The respondents continue to confuse the issue by focusing only on the already well established cases of MDD with a duration in studies usually greater than two months These are the true positives and there is no controversy whatever regarding their diagnosis. Well established (i.e. severe or enduring) MDD during bereavement has never been the issue.
It is the false positives I worry about -- those with normal and time-limited grief that will remit in the natural course of things without diagnosis or treatment. Two weeks is far too short a duration when we are considering relatively mild symptoms that are so intrinsic to grieving. Rushing to judgment that a mental disorder is present will lead to remarkably high false positive rates and transform normal grief into a medical disorder.
Counter Argument 2: The respondents claim that the DSM-5 intention is only to diagnose MDD, not to include normal grief.
Reply: The crucial and clinching point is that these are clinically completely indistinguishable at frequently encountered levels of normal grief. Prospective studies show that almost half of all the bereaved reach MDE two-week symptom thresholds sometime during the first year after their loss, usually within the first two months. I challenge anyone to distinguish clinically between two weeks of normal grief and two weeks of mild MDD under these circumstances. I certainly can't make this distinction, I very much doubt that my respondents can, and I feel sure that primary care physicians can't manage it while seeing a grieving patient in a seven minute evaluation.
Distinguishing grief from MDD is no problem when symptoms become severe or are enduring. DSM-IV-TR already recognizes this. It allows the diagnosis of MDD anytime during bereavement when there is suicidality, psychosis, morbid worthlessness, psychomotor retardation, or inability to function. This is meant to encourage early diagnosis and active psychiatric intervention whenever this is needed. There is no compelling problem that needs fixing. Grieving patients who need psychiatric help already get it.
Before jumping the gun to a premature and potentially harmful diagnosis, why not watchfully wait a few more weeks to determine if the grief is severe and enduring enough to warrant the label of mental disorder. To do as DSM-5 suggests would instead mislabel a substantial portion of normal grievers and would inappropriately stretch the boundary of psychiatry by medicalizing grief.
The furor surrounding the recently proposed Alzheimer's Guidelines was provoked by their premature attempt to introduce early diagnosis, well before accurate tools are available. The same laudable but currently clearly unrealistic ambition has propelled two of the worst suggestions for new diagnoses in DSM-5: Psychosis Risk, Mild Neurocognitive.
The concept of early identification and intervention is understandably appealing. The problems that eventually blossom into full-fledged psychiatric disorders do not arise suddenly and de novo. Undoubtedly, they have had a long history of gradual stages with changes that at first cause no symptoms whatever, followed by mild premonitory symptoms, followed by the full blown disorder. Clearly, it would be wonderful to prevent the progression and its consequent mounting damage by intervention at the earliest possible moment. Accurate early diagnosis followed by effective early treatment would reduce the direct burden of illness and also its secondary negative consequences.
Optimists among the proponents of preventive psychiatry point to the trend throughout medicine to catch disease earlier and intervene more aggressively. Without going into the merits and risks of early screening in medicine (which remains a mixed and highly controversial issue), the analogy simply doesn't fly. Early diagnosis in psychiatry currently lacks any tools to be helpful and may instead, in its well-intentioned and unwitting way, be extremely harmful both to the individual patient and to public policy.
Preventive psychiatry would have to rest on six foundations: 1) a method of diagnosis that is accurate even in the early stages of the disorder; 2) a treatment that is effective in improving early symptoms and in preventing their progression; 3) a treatment that is safe even if provided over the necessary course of what may be many decades; 4) a manageable degree of stigma, worry, and disadvantage from gaining a label that implies risk and progressive impairment; 5) a favorable risk/benefit analysis regarding clinical utility; and 6) a reasonable public policy cost/benefit analysis. Let's see how the psychosis risk and mild cognitive disorders stack up on these necessary benchmarks:
On diagnostic accuracy: neither proposed disorder has a diagnostic measure that is accurate. Psychosis risk has a false-positive rate of 70-90%. Laboratory studies for mild cognitive are still in very early stages of testing.
On treatment efficacy: None proven for either disorder.
On treatment safety: antipsychotic medications likely to be used for psychosis risk frequently produce enormous weight gain and its dire complications.
On stigma and worry: considerable for both. The power to label could here be the power to destroy.
On clinical utility: none for either. It is all risk and no current gain.
On public policy cost/benefit: especially unfavorable for minor cognitive disorder given the very expensive imaging studies and the lack of any clinical benefit.
Before their suggestions will make any sense, the experts in schizophrenia and dementia who are pushing for earlier diagnosis need first to do the research to fill in all the above blanks. Most likely this research enterprise will take a decade (and possibly much more). Until then, caution is safer than wishful thinking.
In July, panels sponsored jointly by the National Institute of Aging and the Alzheimer's Association presented controversial proposed guidelines for diagnosing Alzheimer's at three different stages of its progression: 1) preclinical, 2) mild cognitive impairment, and 3) classic dementia.
The preclinical panel stated that laboratory testing (i.e. PET or MRI scans, spinal taps, or blood tests) before the appearance of symptoms was meant to be purely for research. But, the other two panels seemed to suggest that laboratory testing was ready, or soon would be ready, to be used in routine clinical practice in diagnosing mild cognitive impairment or dementia. Faced with widespread skepticism, the panels held a conference call yesterday to clarify their position. As reported by Gina Kolata in The New York Times, there is reassuring new information. The panels recognize that laboratory testing is still only a research tool and will not be recommending that it be included as part of current clinical diagnosis. This makes great sense. All the available tests are at an early stage of development and are not nearly ready for routine use.
Rapid strides are being made in the study of Alzheimer's disease, with powerful new methods leading us closer to understanding its causes and mechanisms. But let's not jump the gun and mislead ourselves or the public into the false beliefs that a diagnostic breakthrough has already been made and that a treatment breakthrough is possible in the near future.
It is pretty easy to show that a promising laboratory procedure yields different group mean values when comparing Alzheimer's to a control group. It is very difficult to prove that it has sufficient reliability, accuracy, clinical utility, and cost effectiveness to become a useful diagnostic test worthy of use in routine clinical practice. It will require years of testing in very varied populations before we will learn if any of the currently available candidates is indeed the long awaited diagnostic test for Alzheimer's.
It is understandable that Alzheimer's experts have a strong desire to become preventively proactive. Can amyloid be the early marker of Alzheimer's, in analogy to cholesterol and heart disease? Can early identification and early intervention prevent the ravages of the disease? The problem is that you simply cannot skip the middle steps. Do the research first, then publish the guidelines.
And we should also be cautious in our expectations for a treatment breakthrough. It is possible that learning more about the mechanisms of Alzheimer's may eventually lead to the development of a rational cure or preventive, but it is equally possible that it will not. The general experience in medicine over the past three decades is that an exponential explosion in knowledge about a disease does not often lead to any immediate miracle cure. Moreover, the lack of success to date in developing medications for Alzheimer's does not inspire confidence. The available drugs -- although they have been highly profitable to the drug companies -- have little, if any, efficacy for patients. Attempts to develop a new generation of effective drugs have so far failed despite considerable investment. There does not appear to be any low-hanging fruit.
We should have and encourage reasonable hope regarding advances in Alzheimer's, but should avoid hype and hoopla. Progress will be steady, but probably much slower than suggested by the recent excitement.
The DSM-5 first draft has proposed many new diagnoses that would create enormous problems (especially false-positive "epidemics" and forensic misuse). Two perceived needs have driven the DSM-5 Work Groups in this unhappy direction:1) therapeutic zeal not to miss patients who might benefit from treatment; and 2) an aversion toward using the Not Otherwise Specified (NOS) categories. I will argue that these NOS categories impart a great deal of useful clinical information and are essential to the flexible and effective use of the manual. Giving every presentation a specific name and code in order to reduce the use of NOS would create much worse problems than it would solve.
The common prejudice against NOS diagnosis is that it puts psychiatry in a bad light. Why should as many as a third of our patients not qualify for anything more definitive? How do we explain this to them, their families, to referral sources, and to ourselves? How can we plan a specific treatment if the patient doesn't have a specific diagnosis? And so on. It may be useful to answer these questions in the act of exploring the different ways patients actually qualify for a NOS diagnosis:
1) There is simply not enough information to be more specific. Sometimes, this occurs because there was insufficient time for a complete evaluation or the patient is uncooperative and there is no informant or chart. Often, though, it comes from the inherent uncertainties of the situation. I have, for example, rarely felt comfortable with any label other than Psychotic Disorder NOS for psychotic teenagers who have only short track records. There is usually just too much uncertainty about the etiology (e.g., role of drugs) and their future course to be more definitive. There is nothing to be defensive about in using NOS in these situations. The designation Psychotic Disorder NOS conveys a great deal of information, while keeping tentative what deserves to be kept tentative. The immediate treatment target is clear without imposing a premature closure on long term treatment needs or prognosis. This can easily and productively be explained to patients and families.
2) The presentation clearly belongs in the section, but does not fit the prototype of any of the specific disorders defined there. For example, in DSM-IV we included binge eating disorder as an example of Eating Disorder NOS, rather than elevating it to a separate coded category. This allows the clinician the flexibility to diagnose an individual patient when this is deemed necessary without prematurely reifying a category that has yet to pass its risk-benefit test and might have unfortunate unintended consequences.
3) The condition is subthreshold to the specific criteria sets, but nonetheless causes obvious clinically significant distress or impairment. There is no bright line between mental disorder and normality. The decision whether a mental disorder is or is not present inherently has to be made on a case by case basis. The NOS categories provide needed flexibility in diagnosing the many people who present at the boundary with normality. Clinicians can use the appropriate NOS category for the early diagnosis of subthreshold conditions (e.g. "prepsychotic risk") when this clearly warranted for that particular person. This is far preferable to introducing a specific category for "psychosis risk" that would inevitably misidentify many individuals who would be much better off without diagnosis and treatment.
4) The condition presents a mixture of symptoms from different specific disorders that are individually subthreshold but jointly causative of clinically significant distress or impairment. The proposal for a Mixed Anxiety Depressive Disorder is a perfect example and is best handled as an NOS. If made an official category, it would immediately become one of the most popular diagnoses in DSM-5 without any proof that treatment would provide more good than harm for the millions of people who would get the diagnosis.
In all these ways, the NOS categories are indispensable. They should be celebrated, rather than denigrated, and used whenever they are the best description of the less than typical patient. The designation NOS is never really nonspecific or noninformative because it places the patient in a suitable section of the manual without providing more certainty or specificity than the situation allows.
Advice to DSM-5:
1) Accept the fact of life that a certain degree of diagnostic uncertainty and heterogeneity is inherent in the definition of mental disorder. Do not seek to attain an unattainable and pseudoprecise total specificity.
2) Appreciate that each NOS designation provides considerable information (for example, Psychotic Disorder NOS is very different in its treatment and prognostic connotations from Mood Disorder NOS or Eating Disorder NOS).
3) List the most common examples under each NOS category and allow these to be subtypes of that NOS (e.g., "Eating Disorder NOS, binge eating presentation" or "Mood Disorder NOS, premenstrual dysphoric presentation."
4) Clinicians using the NOS diagnoses are dealing with nonprototypical boundary cases. They must therefore be especially careful in determining that the presentation is accompanied by sufficient clinically significant distress or impairment to warrant a diagnosis of mental disorder.
5) Do not create new diagnoses in a vain attempt to replace NOS. The suggestions for new DSM-5 diagnoses should instead be available as examples under the most appropriate NOS ("minor neurognitive" under Cognitive Disorders NOS, etc). There are two exceptions among the proposed new diagnoses -- "paraphilic coercive" and "hypersexuality" -- both of which are harmful constructs whose use should be discouraged altogether, even within the NOS rubric.
Previously, I have been quite critical of the DSM-5 suggestion to introduce a new diagnosis, Minor Neurocognitive Disorder, on the grounds that it would create a large false-positive problem and would lead to unnecessary worry and cost with no useful intervention. Even more ambitious and dangerous are the recently suggested diagnostic guidelines for Alzheimer's created by a panel jointly sponsored by the National Institute on Aging and the Alzheimer's Association. The proposal is a clear case of narrowly focused experts getting far ahead of the available technology to suggest what will be an enormously costly public health experiment with dire unintended consequences.
The goal of the proposed guidelines is laudable: to identify those at risk for Alzheimer's even before they have developed clinical symptoms and to intervene preventitively before the damage is done. The suggested guidelines would divide Alzheimer's into three groups of ascending severity and clarity of presentation: 1) preclinical (i.e. no symptoms, but positive laboratory findings); 2) mild impairment; and 3) classic dementia. The guidelines would recommend laboratory studies to make the diagnosis in the first two groups, neither of which is currently considered an official diagnosis.
If we had well established diagnostic tools to identify preclinical and mild presentations, the guidelines would make great sense. Unfortunately, however, we do not yet have proven tests, and guidelines that pretend we do are premature and reckless. Laboratory studies for Alzheimer's are of recent vintage, are tested only in small selected samples, will probably have huge false-positive rates in the general population, are expensive, and carry medical risks. None is near ready to be used in routine clinical practice, particularly in the general population.
To make matters worse (and the suggested guidelines even more ridiculous), there is no effective treatment for Alzheimer's in any of its stages. So, finding out that you are (only possibly) at risk for developing Alzheimer's would provide little or no benefit -- but would create needless worry, testing, treatment, expense, risk, and insurance and disability issues. The attempt to provide early identification with fallible tests and no effective treatment serves no useful purpose and can cause great harm not only to individuals, but also to public health policy. Scarce health dollars should not be wasted on what would amount to a frivolous public health experiment. First, let's do the research necessary to prove the tests are sufficiently specific and to find medications that work.
How could such a bad idea be forwarded by renowned experts sponsored by august organizations. I have in earlier pieces written on the tunnel vision of experts in any given area and their natural enthusiasm for pushing the boundaries of their disorder of interest. No doubt the premature emergence of these guidelines results from the great frustration we all feel at the slow pace of development of diagnostic and treatment tools for Alzheimer's. Most of us expected there to be a well established laboratory test by now and drug discovery has also been disappointingly slow. My guess is that the guideline makers hope to jumpstart the field by highlighting the potential of early identification. But this is definitely putting the cart before horse. Guidelines that will have great influence on how people live their lives and how the country will spend limited healthcare dollars must follow well established science and an inclusive public policy debate, not lead it.
I am convinced from my experience with experts that they act from naïve good faith and that they think expanding their field of interest will be good for patients. They tend to be blind to false-positive problems and societal costs because they are not trained to think in these terms, not because of conflicts of interest. But such naïve goodwill does not motivate the corporations that market drugs and diagnostic tests. There will be an explosion of testing and treatment if these guidlines are approved, much or all of it unnecessary and expensive, some of it downright harmful. The medical/industrial complex will have a field day.
The suggested guidelines for Alzheimer's are not yet official, so there is still hope. Given the great impact they will have on public health policy, they should not become official until there is a wide public policy debate with input and monitoring that reaches beyond the narrow group of experts in the field. Decisions on Alzheimer's are too important to patients, and public policy decisions should be made exclusively by experts on Alzheimer's.
Every month or so, someone (usually very smart and passionate) sends me a detailed proposal for a new diagnostic system offered as an alternative to the jumbled, pedestrian, atheoretical, and purely descriptive method used in DSM. The new system is invariably theory driven, clever, neat, and plausible. Surely, it is quite easy to be more coherent than a DSM that consists of a jumble of disorders gathered together largely through a historical accreting process based mostly on clinical observation and descriptive research -- without an underlying theory or deep knowledge of causality.
The new systems come in three types: 1) Brain biology: these used to be based on correlates with neurotransmitters, but recently neural networks of various kinds are much more popular; 2) Psychological dimensions: hundreds of scales have been developed and carefully tested; and 3) Systems based on psychodynamic, ethological, and developmental models -- less popular now than they once were.
Unfortunately, none of these approaches, however elegant, is remotely ready for inclusion in the official system of psychiatric nomenclature. DSM must by its very nature be a conservative document that follows and never leads the field. The problem with all of the suggestions to replace the admitted DSM jumble is that there are so many contenders, none of which has been proven or has attained wide acceptance from the field. It is also not possible to choose one from among so many plausible, but necessarily parochial systems, when most clinicians have absolutely no interest in any of them and the proponents of rival systems can make about equally valid claims for their respective pet methods.
The DSM-IV experience with the personality disorders was a rude and disheartening awakening. I very much hoped to include at least an optional dimensional personality rating scale. We were able to gather together in one room the proponents of all the competing dimensional systems to attempt the selection of one or some compromise among them. It didn't work; we could not forge a consensus because each participant remained wedded to his own scale (however minimally different it was from its near neighbors). Without wide agreement, it is impossible to force a field to accept changes that represent one necessarily narrowly defined perspective. The DSM-5 effort to include personality dimensions will also undoubtedly fail -- for this reason as well as for its unbelievably byzantine complexity.
I feel sure that our clumsy descriptive classification may not be the only, or even the optimal way, to sort things for future research. But I feel equally certain that DSM remains necessary to carry forth the current, everyday, practical clinical and administrative work that are its first priority. Once we have attained a widely accepted, etiological understanding of at least some forms of psychopathology, the new insights will gradually replace our clumsy, but nonetheless now still useful system.
At this stage in this arena, the wisdom of the philosopher Vico trumps the much greater and better-known Descartes. Descartes sought to use what we now call Cartesian rationality and mathematical order to sort what were previously seemingly disorderly phenomena. This turned out to be a screaming success in the mathematical, physical and chemical worlds, but has (as Vico predicted) much less purchase in understanding the sloppy complicatedness of human affairs, including psychiatric diagnosis.
Greetings Dr. Frances,
I read Good Grief published in the NYT on Saturday, August 14, 2010 and am writing to inquire if there is anything someone such as myself and anyone I may recruit, can do to ass...
What is your opinion about the use of drugs to help kids focus so that they can truly learn in the classroom. What about the control of unruly behavior via the use of drugs?
Is it possible to benefit...
Dr. Allen Frances is professor emeritus at Duke University, where he was previously chair of its department of psychiatry and behavioral sciences. He was also chair of the DSM-4...Read More
How To Avoid Medicalizing Normal Grief In DSM-5