JMIR Publications

ABSTRACT

Background:

Norwegian Child and Adolescent Mental Health Services (CAMHS) use World Health Organization’s (WHO) multiaxial diagnostic system based on the International Classification of Diseases, Tenth Revision (ICD-10); however, analysis of prescribing patterns among axes I-III is underexplored in Electronic Health Records (EHRs) with intertwined patient, episode of care, and contact information.

Objective:

Develop and demonstrate an analytic pipeline for mining and mapping information from EHRs to facilitate understanding of clinical processes and support informed decision-making. This study utilizes Norwegian CAMHS EHR data to identify common diagnoses, comorbidities, and medication use across axes I–III per individual contact.

Methods:

We extracted records of patients ≤19 years old with a primary mental health diagnosis on axes I–III and ≥1 medication per individual contact. Diagnoses were categorized according to the International Classification of Diseases, Tenth Revision (ICD-10), and medications by the Anatomical Therapeutic Chemical (ATC) classification system. Descriptive analyses quantified contact counts, diagnosis frequency, comorbidity rates, and medication frequency within each diagnostic category. Next, we mapped the medications used across all the contacts and non-comorbid contacts separately along each axis.

Results:

Of 7,214 prescribing contacts (Axis I: 7,179; II: 821; III: 65), comorbidity was present in 12.1% (I), 96.1% (II), and 96.9% (III). Leading diagnoses were, in axis I, behavioral-emotional disorders (F90–F98), in axis II, school skills & learning difficulties (F81), and in axis III, mild mental retardation (F70). Most observed comorbidities were F90-F98 with speech & language development disorder (F80), F81, and mixed specific skills development disorder (F83). Psychostimulants predominated across all diagnosis axes, with methylphenidate being the most common. For other ATC categories, the most commonly prescribed medications were: antidepressants (sertraline and fluoxetine); antipsychotics (risperidone and aripiprazole); hypnotics and sedatives (melatonin); antiepileptics (lamotrigine); anxiolytics (diazepam), and non-psychotropics (laxatives, vitamins, and supplements). Medication profiles varied minimally by axis or comorbidity status.

Conclusions:

We demonstrated a mining and mapping analytic pipeline for EHRs to analyze diagnoses, comorbidities, and prescribing practices at the individual contact level. In Norwegian CAMHS, axis I diagnoses were common, often behavioral-emotional disorders. Among the medications, psychostimulants and antidepressants were common. Beyond characterizing diagnoses and medication prescribing patterns, the study presents an approach for mining and mapping EHR data to analyze and provide service-level metrics, as well as clinical practice insights.