Comparing the diagnostic criteria for the DSM-5 and ICD-10
Both the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), and the International Statistical Classification of Diseases and Related Health Problems, 10th Edition (ICD-10) have established diagnostic criteria for alcohol use disorders (AUDs). While the DSM is widely used by clinicians, the U.S. Centers for Medicare and Medicaid Services has recently called for providers to bill for services using ICD-10 designations. Given the ramifications for who will and will be not eligible for treatment, this study compares the two diagnostic approaches, finding discrepancies between the two for more mild and moderate cases of AUDs.
Results will be published in the April 2015 online-only issue of Alcoholism: Clinical & Experimental Research and are currently available at Early View.
"Clinicians and administrators may be most interested in the differences between the two diagnostic approaches for the ways in which they define who is and is not eligible for treatment," explained Norman G. Hoffmann, adjunct professor of psychology at Western Carolina University as well as corresponding author for the study. "The broader version of the ICD, for instance, can make treatment available for patients who have not yet developed the most serious AUDs, which could help stem the development of a more serious disorder."
"Although the clinical use of the DSM classification system remains commonplace - at least in the U.S. - with the forthcoming federal mandate that all U.S. health care settings transition to ICD-10 billing codes, empirical evidence is necessary to determine if the DSM-5 does in fact facilitate a cross-walk to the new ICD-10 coding system," added Steven L. Proctor, a psychology postdoctoral fellow with the Addictive Disorders Treatment Program at the G.V. (Sonny) Montgomery VA Medical Center. "The implications derived from such work not only inform future research efforts, but they directly impact all facets of substance use disorder treatment - whether assessment, diagnosis, selection of appropriate treatment interventions, billing or reimbursement."
"The most likely reason for this transition is that the ICD-10 will be used for all health conditions, so it seems reasonable to require ICD-10 diagnosis for behavioral health as well," said Hoffmann. "The DSM-5 does not cover conditions outside of behavioral health. Thus, avoiding having to use one criteria for one set of conditions and another for all else was probably the reason for the decision to go with the ICD-10 as the universal diagnostic criteria."
The researchers used data from 6,871 male and 801 female admissions to a state prison system to compare the DSM-5 severity index for alcohol use disorder to the ICD-10 clinical and research formulations for harmful use and dependence. All inmates were between 18 and 65 years of age, and slightly more than half were white, with the largest proportion of minorities being African-American (31.5% and 21.5% respectively for males and females), followed by Native American (7.7% and 13.2% respectively by gender).
"The ICD-10 and DSM-5 converge for cases who would not receive a diagnosis and those who manifest the most severe forms of alcohol use disorder," said Hoffmann. "There is more discrepancy between the two, however, for more mild and moderate cases of alcohol use disorder. This has significant implications not only for diagnosis, but also for the development and application of treatment services."
Proctor agreed. "Roughly one-third of DSM-5 mild cases would not receive a diagnosis per the ICD-10 clinical version, which in turn translates to reduced access to treatment services for a fairly large number of individuals," he said. "When the research version criteria are applied, we see this number rise to nearly one-half of all mild cases."
"This could ultimately have a significant impact not only on how much of the treatment is reimbursed, but also how much treatment is provided - especially if only so much treatment will be covered by Medicare and Medicaid - and treatment outcomes," said Hoffmann. "If patients are misdiagnosed to a lesser degree, they are likely to receive inadequate treatment, and this could delay the recovery process."
Hoffmann and Proctor described various implications of these findings for different audiences.
"Clinicians need to know that the version of the ICD-10 which is applied, that is, clinical versus research, is going to impact diagnoses and related treatment services," said Hoffmann. "Clinicians need to document the nature and extent of positive criteria and tie those findings to treatment response and outcomes to determine if the diagnostic distinctions relate to reality."
"From a clinical standpoint," added Proctor, "providers can have confidence in their diagnostic determinations for those patients with a severe AUD or those without a diagnosis. Similarly, in terms of billing and reimbursement for treatment services, patients with a severe DSM-5 alcohol use disorder will remain largely unaffected when insurers encounter ICD-10 dependence codes."
"The implication for researchers is that they need to explore the empirical justification for diagnostic criteria as they pertain to substance use disorders," noted Hoffmann. "Neither the DSM-5 nor the ICD-10 criteria have irrefutable evidence for their validity. Both are consensus formulations based on expert judgment rather than rigorous empirical evidence. Finally, the average reader needs to know that diagnostic approaches vary, and this has significant implications for treatment services rendered and reimbursement of those services. They need to understand that getting a second opinion would probably be a good idea to verify that the diagnostic determination seems consistent between clinicians."
"Although there appears to be a generally high level of agreement between the two diagnostic classification systems in that nearly all individuals with a severe AUD per the DSM-5 received an ICD-10 dependence diagnosis," added Proctor, "there are likely some very important individual differences between the two groups in terms of their clinical presentation. Considering the wide variation in the specific criteria or symptoms used to arrive at a diagnosis between systems, additional research is necessary to identify relevant prognostic factors and determine the clinical course for these two groups."