Defining common mental health disorders among adolescents using electronic primary care data: an external validation
Rates of common mental health disorders appear to be increasing among children and adolescents. Epidemiological research could clarify this, identify its causes, and inform interventions to ameliorate it. Individuals with mental health disorders may be less likely to participate in research projects but will be represented in routine data. However, routine data, such as electronic primary care records, may also be subject to bias. The extent of this bias and the validity of epidemiological research based on routine data can be investigated using linkage between information collected in research projects and routine patient records.
Subjects included in this study were 1,562 adolescents who had completed the Revised Clinical Interview Schedule (CIS-R) in the Avon Longitudinal Study of Parents and Children (ALSPAC) at age 17-18 years, had linkage established to their electronic primary care records, and for whom we had a GP record up to at least 6 months after the time they completed the CIS-R. We used lists of Read codes corresponding to diagnoses, symptoms and treatments to create twelve definitions of common mental health disorders (CMHD) as well as depression alone. CMHD consisted of depression and anxiety disorders; the latter included generalised anxiety disorder, mixed anxiety and depression, panic disorders, and phobic disorders. We calculated sensitivities and specificities of these definitions, using the CIS-R as the gold standard.
Of the 1,562 individuals included in this study, 115 (7.4%) had a diagnosis of depression according to the CIS-R and 213 (13.6%) had a CMHD. Sensitivities of the definitions based on primary care data ranged from 5.2% to 24.3% for depression and from 3.8% to 19.2% for CMHD. The specificities of all definitions were above 98% for depression and above 96% for CMHD. For both depression and CMHD, the definition that included current diagnosis, treatment or symptoms identified the highest proportion of cases.
Case definitions for common mental health disorders based on information available in routine patient records are highly specific but have low sensitivity when compared to definitions based on standard survey instruments that record symptoms. Several considerations including help seeking behaviour and the recording practice of clinicians are likely to underlie these differences. Routine health data may underestimate the prevalence of some psychological disorders but may still be valid for studying risk factors and clinical outcomes. Further research is needed to find ways to minimise bias in studies where both self-reported and linked electronic primary care data are available.