DSM 5 and Diagnostic Inflation
Reply To Misleading Comments From The Task Force
Posted Jan 23, 2012
My biggest concern regarding DSM 5 is that it will dramatically increase the rates of mental disorder and cheapen the currency of psychiatric diagnosis. The DSM 5 proposals do this two ways: 1) by reducing thresholds for existing disorders; and, 2) by introducing new high prevalence disorders at the boundary with normality. Unless corrected, DSM 5 may create millions of newly mislabeled 'patients,' with resulting unnecessary and potentially harmful treatment, stigma, and wasteful misallocation of scarce resources.
In a recent commentary in the American Journal of Psychiatry, the DSM 5 leadership defend their opposite position. They admit that they are indifferent to the manual's impact on rates and justify this on the grounds that no one knows for sure what the true or optimal rates should be. In an earlier blog, I chided the Task Force for ignoring the real world harmful unintended consequences that will follow the dramatic increase in prevalence rates caused by their untested and risky proposals.
The Task Force has come back with the following Q and A which popped up at the APA website, apparently in response to my warnings about diagnostic inflation. You can find it at:
"Q: Was prevalence estimated in the DSM-5 Field Trials?
A: The prevalence of every target diagnosis evaluated in the field trial was estimated.
Q: Will the prevalence of DSM-5 disorders be very much higher than the
prevalence of DSM-IV disorders?
A: In general, the prevalence rates of the diagnoses evaluated in the Field
Trials are slightly lower than DSM-IV prevalence rates."
The wording is remarkably misleading. Note that the DSM IV rates in the field trial were "estimated" by chart review, but that the DSM 5 rates were ""evaluated" by systematic interview. This results in a totally meaningless comparison of apples and oranges. The DSM IV and DSM 5 rates should have been systematically compared (as is customary) using common data gathered in the field trial diagnostic interviews. This is absolutely standard research operating procedure- always compare apples to apples, don't switch assessment methods. It is beyond understanding why this simple step was omitted in the DSM 5 field trials and why chart diagnosis is offered now as a lame substitute.
The Q/A prediction that DSM 5 prevalence rates will be lower than DSM IV is wrong, impossible, even laughable. It is obvious that most changes suggested for DSM 5 will increase prevalence rates above those in DSM IV, often quite dramatically. And most clearly, this is the case for the new diagnoses whose high rates could not possibly be estimated given the limitations of the field trials. The DSM 5 team should most certainly know better than to claim that DSM 5 won't raise prevalences. I am not sure which of the two possible interpretations is worse- that they are being deliberately misleading or that they are terminally self deluded. Either way, the failure to measure prevalences in the field trial is an unaccountable error and the failure to reckon the risky consequences of DSM 5 proposals is just plain reckless.
As I first pointed out before the DSM 5 field trials began, the proper design should have included:
1) For existing disorders- Ratings of DSM IV, ICD 10, and DSM 5 criteria items to allow comparison of rates across the three systems.
2) For new disorders- sampling their likely rates in general psychiatric settings, in primary care, and (by telephone) in the general population.
The academic centers that were selected for DSM 5 field testing are ivory towers that don't generalize well to the real world. Indeed, most psychiatric diagnosis and medication treatment is now done by primary care doctors and the impact of DSM 5 must be tested where it will most be used.
The whole purpose of field testing is to identify and correct problems in preliminary DSM suggestions before they are set in stone as official guides to diagnostic practice. The design of DSM 5 field trial unaccountably left out the most important question (its impact in rates) and the most important settings (routine clinical practice). The DSM 5 leadership now provides a fudged, incorrect, and belated reply to the risks of diagnostic inflation- don't worry, trust us, it won't happen. This is nonsense- diagnostic inflation will most certainly happen unless DSM 5 is corrected before publication. Such willful blindness is a sure prescription for bad surprises and serious unintended consequences.