Characterizing and Coding Psychiatric Diagnoses Using EHR Data

In Reply Our recent article elicited letters raising several issues. Wright and colleagues provide important criticisms of the framing of our article. We agree that diagnosis is a more accurate description of phecodes derived from International Statistical Classification of Diseases, Ninth and Tenth Revisions, Clinical Modification (ICD-9/10-CM), billing codes in electronic health records (EHRs), and prevalence of diagnoses in EHRs is more accurate than prevalence. Though some recent work has shown that using 2 or more EHR codes has good sensitivity for a confirmed diagnosis of certain severe disorders, we cannot ensure this is the case across all diagnoses and populations. There is likely rampant underdiagnosis of heavily stigmatized disorders, as conditions like alcohol use disorders are generally missed in routine medical care, and few patients with alcohol use disorders go on to receive treatment. The goal of our analysis was to provide an overview of available information in All of Us, not to infer that these rates reflect patterns in the broader population.
Source: JAMA Psychiatry - Category: Psychiatry Source Type: research