interpreting dsm-5
On publication, it contained errors. Not surprising in such a massive undertaking, especially one that is the work of many committees. On the DSM-5 website, you can read corrections for the few coding errors that were caught either just before or just after actual printing. (In accord with its change from postpartum to peripartum onset for mood and other disorders, I suppose DSM-5 would call that the peripublication period.) Beyond the numbers game, there are other, less readily resolved problems. I hasten to point out that these are not all errors. Some represent the fallout from changes in definitions. Some are probably just oversight. And a few resulted from the absence of adequate central coordination, which may well have been a downstream consequence of insufficient time to get the job done.
Some of these inconsistencies (and outright errors) may seem a little abstruse to the casual reader. Perhaps they smack a bit too much of “Inside Baseball” for the taste of many. However, for a book that hews so tightly to exact diagnostic prescriptions, even little glitches of exposition make for problems of diagnosis. I’ll present some of these issues and then explain as best I can how we are to understand them. They are also covered in my new book, DSM-5 Made Easy (it is the successor to DSM-IV Made Easy). From Guilford Publications.
Onset after medication use
DSM-5 mentions “With onset after medication use” as an optional specifier for substance-induced anxiety disorder, obsessive-compulsive and related disorder, and sexual dysfunctions, but not for psychotic, mood, or sleep disorders. This seems, at best, inconsistent.
My view. The problem stems from the fact that different subcommittees wrote different parts of the criteria, and the coordination wasn’t always perfect. What I would do, will do, is to use the after medication use specifier whenever it seems appropriate, regardless of what the manual says. To do so is, after all, accurate, appropriate, and good patient care.
Substance-induced disorders
As regards substance-induced psychotic, mood, anxiety, &c disorders, the manual doesn’t say whether "with onset after medication use" can be used simultaneously with "with onset during intoxication" or "with onset during withdrawal," but there will obviously be such opportunities.
My view. Again, do what’s right for the patient. I’ve corresponded with some on the substance-related disorders committee on this issue; there’s agreement that clinicians do have some free will, after all.
Narcolepsy numbers
The criteria for narcolepsy specify ICD-10 and ICD-9 numbers for specific types. However, they ignore two types that may be among the most commonly encountered: narcolepsy with cataplexy with hypocretin deficiency, and narcolepsy with cataplexy (where hypocretin deficiency is not known). What’s a clinician to do?
My view. I’ve corresponded with a member of the sleep–wake disorders committee who referred me to Dr. Regier and Dr. Narrow at the APA. Here is their answer:
For both of these disorders, use the codes G47.8 [780.59] (for ICD-10 and ICD-9 respectively). The verbiage we’d use would be
Other specified sleep–wake disorder, narcolepsy with cataplexy with hypocretin deficiency, and
Other specified sleep–wake disorder, narcolepsy with cataplexy with unknown hypocretin status
But isn’t it just a bit peculiar not to include numbers for such popular and important types?
With atypical features
Any major depressive episode can incorporate atypical features. However, with atypical features is not a specifier listed for bipolar II disorder (which must have major depressive episodes).
My view. Actually, the DSM-5 staff did get back to me on this one, with the statement that it was omitted from the bipolar II text when it should have been included. So, I’ll go right ahead and use the atypical features specifier if the patient’s symptoms warrant.
Specifiers for pedophilic disorder
Pedophilic disorder is the only paraphilia that does not contain specifiers for in a controlled environment or in remission.
My view. Surely, a pedophile who is incarcerated is less likely to re-offend than is, say, a sexual sadist. I’d go right ahead and use it, if warranted by the facts. But research is pretty clear on the fact that pedophiles tend not to remit.
Caffeine intoxication code
The ICD-9 code for caffeine intoxication is given as 305.90. That is the same as for several forms of substance use disorder (PCP, inhalants, other). All other ICD-9 intoxication codes are 292.89. Surely, this is an error.
My view. Well, if it is an error, no one seems willing to cop to it. Actually, this one may be due to problems matching ICD-9 codes with DSM-5. I’ll go right ahead and use it as written, and try to lose no sleep over it. After all, by October of 2015, ICD-9 will be history and we can all use the more rational ICD-10 codes.
By the way, there is a place on the DSM-5 website to write in with questions about issues. I’ve written in more than a dozen queries; so far, I’ve gotten back several answers. One of them was to this question, and it more or less ignored the point that I was trying to make (which was that, if disorders have to share numbers, they should at least share numbers with similar disorders). Nonetheless, I do get the impression that people there are trying to cope with queries coming in about use of DSM-5. They're just coping slowly.
Age-related Cognitive decline (ARCD)
DSM-IV included the useful “Age-related cognitive decline” for older patients who report trouble remembering things but test normal. ARCD doesn’t appear in DSM-5. [I admit, the ICD-9 code we used for it, 780.93, wasn’t all that reassuring (“memory loss”).] Have we just abandoned this useful category?
My view. With the creation of neurocognitive disorder (NCD), DSM-5 now embraces multiple levels of seriousness for cognitive difficulty, defined by whether or not the patient (and family) are being materially impaired by the decline. What is left not quite so well defined is the borderland between mild NCD and normal. There is still room for us to identify a patient who complains about memory problems but who tests normal as “one of the worried well.” But no longer do we have a special term for this particular brand of normal. I suppose that we don’t actually need to code this on a form to use it; it remains an extremely useful concept for communicating with (and reassuring) our patients.
Neurocognitive disorder due to multiple causes
OK, this one will take some explaining. Imagine you have a patient who has Alzheimer’s disease. You’ve got radiological evidence, you’ve noted that there’s been gradual onset and steady progression. According to the DSM-5 criteria, you should diagnose “probable NCD due to Alzheimer’s disease.” But now, your patient has a stroke, and suddenly, diagnostic chaos reigns. Although the criteria for “NCD due to multiple etiologies” gives the particular example of NCD due to Alzheimer's plus subsequent vascular NCD, one of the criteria for Alzheimer’s NCD includes the proscription against multiple causes. Thus, these two criteria sets seem to be mutually incompatible.
My view. The clinical resolution is to ignore the DSM-5 criteria, and just go ahead and diagnose what you see. And that’s apparently pretty much what the committee intended when they discussed this specific example (according to my informant). We should go ahead and diagnose Alzheimer’s and diagnose vascular NCD, but omit the rubric of multiple causation. It’s just that this solution didn’t quite make it into the DSM-5 manual. Too bad. And confusing.
Personality disorder criteria
The subcommittee recommended a wholesale change in the DSM-5 criteria for personality disorder. Those clinicians wanted to see enacted an entirely different scheme that involved not just categorical criteria (as was the case in the previous DSMs), but add to it the ability to recognize that personality pathology exists on a continuum from nearly normal to very much affected, indeed. But this was not accepted by the governing body of the American Psychiatric Association, and the old DSM-IV criteria were simply carried over to the new edition, more or less intact (some of the wording in the accompanying text has been updated). One practical fallout from this decision has been that there is suddenly no diagnosis to indicate a patient who has particular personality traits, but not enough of them to diagnose, say, narcissistic personality disorder.
My view. I think this is too bad, but not a tragedy. We simply have to remember to mention in the diagnostic summary that these traits were identified; only we cannot put it into the diagnosis and give it a number. I suspect this will all come out in the wash that will occur eventually, though probably not next Monday. Meantime, in order to help those of us who are not experts in the field of personality pathology, we desperately need some examples of how to use the Part III alternative criteria for personality disorders. It would help us with our patients and it would (probably) help speed the acceptance of the dimensional personality disorder criteria.
No mental disorder and diagnosis deferred
Using DSM-5 and ICD-10, how are we to say 1. No mental disorder? 2. Diagnosis deferred? These diagnoses were a part of DSM-IV; they do not appear in DSM-5.
My view. The first part of this, not mentally ill, is relatively simple. There is an ICD-10 code you can use, but the name is incredibly clunky. Here it is: Z03.89 (V71.09] Encounter for Observation for Other Suspected Diseases and Conditions Ruled Out. It’s long and, frankly, silly sounding, but it’s what we’ve been given to work with. What I’d do is to use the code, but just write down “No mental illness.” The record room will never know what hit them.
As for "diagnosis deferred," there isn’t really anything that’s been specified that will do the job exactly. F99 is for Mental Illness Unspecified, and that will have to do. For me, this represents a significant issue. I feel that our ability to call a patient “Mentally ill, but undiagnosed” is one of the most powerful statements we as clinicians can make. It allows us to acknowledge two things: that, after examination, we have cause to consider a patient ill; and it allows us to state that we do not have enough information (sometimes, too much information) to allow a fully informed diagnosis that will assist us in planning treatment, prognosis, and so forth.
There is one other possibility: the even less specific R69, Unspecified illness. But, wouldn’t you think we'd have at least enough information to know that it’s mental?
Psychosis severity
For many psychotic disorders, DSM-5 encourages us to rate severity. The system it would like us to use is this: score symptoms on a 5-point scale. The symptoms we are to consider are: hallucinations, delusions, disorganized speech, abnormal psychomotor behavior, negative symptoms, impaired cognition, depression, and mania. Each of these is to be rated as 0 = not present, 1 = equivocal, 2 = mild, 3 = moderate, 4 = severe. That’s 8 ratings per patient per evaluation. However, the manual does not state how the severity specifier is to be reported. Is it attached to each diagnosis, with each component (hallucinations, delusions, &c.) given a number, or should these ratings be spelled out (mild, moderate). Or, should the clinician specify rather an average, overall score?
My view. These ratings appear in Section III, Emerging Measures and Models. The manual acknowledges that the rating measures described here have not been approved formally, that “the scientific evidence is not yet available to support widespread clinical use.” Each relevant criteria set in the Schizophrenia Spectrum and Other Psychotic Disorders chapter clearly (and kindly, in my view) states that diagnosis can be reported “without using this severity specifier.” If I decide to report severity, it will probably be a single, encompassing mild, moderate, or severe, and I’ll tack it on to the code descriptor. When I do rate each symptom separately, it’ll go into the body of the discharge (or admission) summary.
Accommodation in neurocognitive disorder
Criterion B for mild neurocognitive disorder states that the patient may require "greater effort, compensatory strategies, or accommodation..." The last term, accommodation, is nowhere defined in the text. What does this mean?
My view. One of the committee members wrote to me that it means the same as taking compensatory strategies, but I don’t think that’s actually what’s intended. I would guess it means that someone else has made allowances for the person who is having just a little trouble—like the classic sign on the door, “Wendy’s room.” But I’m still looking for an official answer. At APA, mum's the word.
Scoring of WHODAS 2.0
DSM-5 has deep-sixed the GAF (Global Assessment of Functioning) in favor of the World Health Organization Disability Assessment Schedule, a lengthy (36 questions) form that is designed to be less arbitrary than the GAF and to allow better comparisons with non-mental disabilities. One of its attractions is the promise of a complex scoring method that takes into account multiple levels of difficulty for the different items. This should allow conversion into a metric that ranges from 0 (no disability) to 100 (full disability). I say "should" because the computer program that is supposed to be available on the WHO website simply isn't.
I've written repeatedly to APA requesting further information; so far, there has been no response other than a very nice lady who says words to the effect of, "I'm trying to refer this to the appropriate person." The nice lady has since left the employ of APA, a tragic loss.
Seems to me, if we are to be denied use of the (relatively straightforward) GAF, we should at least have the advertised benefits of the new scale. Stay tuned.
My view. I'm still using the GAF. OK, it's a tad arbitrary, but it was simple, straightforward, and addressed the issues that matter to mental health care folk.
So, that's part of my laundry list of quibbles. For now. I'd be interested in hearing from others who have had problems interpreting DSM-5.
Some of these inconsistencies (and outright errors) may seem a little abstruse to the casual reader. Perhaps they smack a bit too much of “Inside Baseball” for the taste of many. However, for a book that hews so tightly to exact diagnostic prescriptions, even little glitches of exposition make for problems of diagnosis. I’ll present some of these issues and then explain as best I can how we are to understand them. They are also covered in my new book, DSM-5 Made Easy (it is the successor to DSM-IV Made Easy). From Guilford Publications.
Onset after medication use
DSM-5 mentions “With onset after medication use” as an optional specifier for substance-induced anxiety disorder, obsessive-compulsive and related disorder, and sexual dysfunctions, but not for psychotic, mood, or sleep disorders. This seems, at best, inconsistent.
My view. The problem stems from the fact that different subcommittees wrote different parts of the criteria, and the coordination wasn’t always perfect. What I would do, will do, is to use the after medication use specifier whenever it seems appropriate, regardless of what the manual says. To do so is, after all, accurate, appropriate, and good patient care.
Substance-induced disorders
As regards substance-induced psychotic, mood, anxiety, &c disorders, the manual doesn’t say whether "with onset after medication use" can be used simultaneously with "with onset during intoxication" or "with onset during withdrawal," but there will obviously be such opportunities.
My view. Again, do what’s right for the patient. I’ve corresponded with some on the substance-related disorders committee on this issue; there’s agreement that clinicians do have some free will, after all.
Narcolepsy numbers
The criteria for narcolepsy specify ICD-10 and ICD-9 numbers for specific types. However, they ignore two types that may be among the most commonly encountered: narcolepsy with cataplexy with hypocretin deficiency, and narcolepsy with cataplexy (where hypocretin deficiency is not known). What’s a clinician to do?
My view. I’ve corresponded with a member of the sleep–wake disorders committee who referred me to Dr. Regier and Dr. Narrow at the APA. Here is their answer:
For both of these disorders, use the codes G47.8 [780.59] (for ICD-10 and ICD-9 respectively). The verbiage we’d use would be
Other specified sleep–wake disorder, narcolepsy with cataplexy with hypocretin deficiency, and
Other specified sleep–wake disorder, narcolepsy with cataplexy with unknown hypocretin status
But isn’t it just a bit peculiar not to include numbers for such popular and important types?
With atypical features
Any major depressive episode can incorporate atypical features. However, with atypical features is not a specifier listed for bipolar II disorder (which must have major depressive episodes).
My view. Actually, the DSM-5 staff did get back to me on this one, with the statement that it was omitted from the bipolar II text when it should have been included. So, I’ll go right ahead and use the atypical features specifier if the patient’s symptoms warrant.
Specifiers for pedophilic disorder
Pedophilic disorder is the only paraphilia that does not contain specifiers for in a controlled environment or in remission.
My view. Surely, a pedophile who is incarcerated is less likely to re-offend than is, say, a sexual sadist. I’d go right ahead and use it, if warranted by the facts. But research is pretty clear on the fact that pedophiles tend not to remit.
Caffeine intoxication code
The ICD-9 code for caffeine intoxication is given as 305.90. That is the same as for several forms of substance use disorder (PCP, inhalants, other). All other ICD-9 intoxication codes are 292.89. Surely, this is an error.
My view. Well, if it is an error, no one seems willing to cop to it. Actually, this one may be due to problems matching ICD-9 codes with DSM-5. I’ll go right ahead and use it as written, and try to lose no sleep over it. After all, by October of 2015, ICD-9 will be history and we can all use the more rational ICD-10 codes.
By the way, there is a place on the DSM-5 website to write in with questions about issues. I’ve written in more than a dozen queries; so far, I’ve gotten back several answers. One of them was to this question, and it more or less ignored the point that I was trying to make (which was that, if disorders have to share numbers, they should at least share numbers with similar disorders). Nonetheless, I do get the impression that people there are trying to cope with queries coming in about use of DSM-5. They're just coping slowly.
Age-related Cognitive decline (ARCD)
DSM-IV included the useful “Age-related cognitive decline” for older patients who report trouble remembering things but test normal. ARCD doesn’t appear in DSM-5. [I admit, the ICD-9 code we used for it, 780.93, wasn’t all that reassuring (“memory loss”).] Have we just abandoned this useful category?
My view. With the creation of neurocognitive disorder (NCD), DSM-5 now embraces multiple levels of seriousness for cognitive difficulty, defined by whether or not the patient (and family) are being materially impaired by the decline. What is left not quite so well defined is the borderland between mild NCD and normal. There is still room for us to identify a patient who complains about memory problems but who tests normal as “one of the worried well.” But no longer do we have a special term for this particular brand of normal. I suppose that we don’t actually need to code this on a form to use it; it remains an extremely useful concept for communicating with (and reassuring) our patients.
Neurocognitive disorder due to multiple causes
OK, this one will take some explaining. Imagine you have a patient who has Alzheimer’s disease. You’ve got radiological evidence, you’ve noted that there’s been gradual onset and steady progression. According to the DSM-5 criteria, you should diagnose “probable NCD due to Alzheimer’s disease.” But now, your patient has a stroke, and suddenly, diagnostic chaos reigns. Although the criteria for “NCD due to multiple etiologies” gives the particular example of NCD due to Alzheimer's plus subsequent vascular NCD, one of the criteria for Alzheimer’s NCD includes the proscription against multiple causes. Thus, these two criteria sets seem to be mutually incompatible.
My view. The clinical resolution is to ignore the DSM-5 criteria, and just go ahead and diagnose what you see. And that’s apparently pretty much what the committee intended when they discussed this specific example (according to my informant). We should go ahead and diagnose Alzheimer’s and diagnose vascular NCD, but omit the rubric of multiple causation. It’s just that this solution didn’t quite make it into the DSM-5 manual. Too bad. And confusing.
Personality disorder criteria
The subcommittee recommended a wholesale change in the DSM-5 criteria for personality disorder. Those clinicians wanted to see enacted an entirely different scheme that involved not just categorical criteria (as was the case in the previous DSMs), but add to it the ability to recognize that personality pathology exists on a continuum from nearly normal to very much affected, indeed. But this was not accepted by the governing body of the American Psychiatric Association, and the old DSM-IV criteria were simply carried over to the new edition, more or less intact (some of the wording in the accompanying text has been updated). One practical fallout from this decision has been that there is suddenly no diagnosis to indicate a patient who has particular personality traits, but not enough of them to diagnose, say, narcissistic personality disorder.
My view. I think this is too bad, but not a tragedy. We simply have to remember to mention in the diagnostic summary that these traits were identified; only we cannot put it into the diagnosis and give it a number. I suspect this will all come out in the wash that will occur eventually, though probably not next Monday. Meantime, in order to help those of us who are not experts in the field of personality pathology, we desperately need some examples of how to use the Part III alternative criteria for personality disorders. It would help us with our patients and it would (probably) help speed the acceptance of the dimensional personality disorder criteria.
No mental disorder and diagnosis deferred
Using DSM-5 and ICD-10, how are we to say 1. No mental disorder? 2. Diagnosis deferred? These diagnoses were a part of DSM-IV; they do not appear in DSM-5.
My view. The first part of this, not mentally ill, is relatively simple. There is an ICD-10 code you can use, but the name is incredibly clunky. Here it is: Z03.89 (V71.09] Encounter for Observation for Other Suspected Diseases and Conditions Ruled Out. It’s long and, frankly, silly sounding, but it’s what we’ve been given to work with. What I’d do is to use the code, but just write down “No mental illness.” The record room will never know what hit them.
As for "diagnosis deferred," there isn’t really anything that’s been specified that will do the job exactly. F99 is for Mental Illness Unspecified, and that will have to do. For me, this represents a significant issue. I feel that our ability to call a patient “Mentally ill, but undiagnosed” is one of the most powerful statements we as clinicians can make. It allows us to acknowledge two things: that, after examination, we have cause to consider a patient ill; and it allows us to state that we do not have enough information (sometimes, too much information) to allow a fully informed diagnosis that will assist us in planning treatment, prognosis, and so forth.
There is one other possibility: the even less specific R69, Unspecified illness. But, wouldn’t you think we'd have at least enough information to know that it’s mental?
Psychosis severity
For many psychotic disorders, DSM-5 encourages us to rate severity. The system it would like us to use is this: score symptoms on a 5-point scale. The symptoms we are to consider are: hallucinations, delusions, disorganized speech, abnormal psychomotor behavior, negative symptoms, impaired cognition, depression, and mania. Each of these is to be rated as 0 = not present, 1 = equivocal, 2 = mild, 3 = moderate, 4 = severe. That’s 8 ratings per patient per evaluation. However, the manual does not state how the severity specifier is to be reported. Is it attached to each diagnosis, with each component (hallucinations, delusions, &c.) given a number, or should these ratings be spelled out (mild, moderate). Or, should the clinician specify rather an average, overall score?
My view. These ratings appear in Section III, Emerging Measures and Models. The manual acknowledges that the rating measures described here have not been approved formally, that “the scientific evidence is not yet available to support widespread clinical use.” Each relevant criteria set in the Schizophrenia Spectrum and Other Psychotic Disorders chapter clearly (and kindly, in my view) states that diagnosis can be reported “without using this severity specifier.” If I decide to report severity, it will probably be a single, encompassing mild, moderate, or severe, and I’ll tack it on to the code descriptor. When I do rate each symptom separately, it’ll go into the body of the discharge (or admission) summary.
Accommodation in neurocognitive disorder
Criterion B for mild neurocognitive disorder states that the patient may require "greater effort, compensatory strategies, or accommodation..." The last term, accommodation, is nowhere defined in the text. What does this mean?
My view. One of the committee members wrote to me that it means the same as taking compensatory strategies, but I don’t think that’s actually what’s intended. I would guess it means that someone else has made allowances for the person who is having just a little trouble—like the classic sign on the door, “Wendy’s room.” But I’m still looking for an official answer. At APA, mum's the word.
Scoring of WHODAS 2.0
DSM-5 has deep-sixed the GAF (Global Assessment of Functioning) in favor of the World Health Organization Disability Assessment Schedule, a lengthy (36 questions) form that is designed to be less arbitrary than the GAF and to allow better comparisons with non-mental disabilities. One of its attractions is the promise of a complex scoring method that takes into account multiple levels of difficulty for the different items. This should allow conversion into a metric that ranges from 0 (no disability) to 100 (full disability). I say "should" because the computer program that is supposed to be available on the WHO website simply isn't.
I've written repeatedly to APA requesting further information; so far, there has been no response other than a very nice lady who says words to the effect of, "I'm trying to refer this to the appropriate person." The nice lady has since left the employ of APA, a tragic loss.
Seems to me, if we are to be denied use of the (relatively straightforward) GAF, we should at least have the advertised benefits of the new scale. Stay tuned.
My view. I'm still using the GAF. OK, it's a tad arbitrary, but it was simple, straightforward, and addressed the issues that matter to mental health care folk.
So, that's part of my laundry list of quibbles. For now. I'd be interested in hearing from others who have had problems interpreting DSM-5.