Claims data from the IBM MarketScan Commercial Database (Q1/2014–Q4/2018) were used. The database consists of medical and drug data of over 200 million individuals covered by employer-sponsored private health insurance and includes records of inpatient services, inpatient admissions, outpatient services, and prescription drug claims. Data are de-identified and comply with the patient requirements of the Health Insurance Portability and Accountability Act. Therefore, no institutional review board exemption nor informed consent was required.

A retrospective claims-based analysis was conducted to assess treatment patterns of patients with ADHD newly initiating a Food and Drug Administration (FDA)-approved ADHD pharmacological treatment as previously described [19, 20]. The index date was defined as the first observed ADHD prescription fill date that was preceded by at least a 6-month treatment gap (i.e., washout period to capture newly initiated treatment episode). The baseline period was the 6 months before the index date. The study period was the 12 months after index date. The follow-up period was the 6 months following the study period; the additional 6 months allowed for sufficient time to determine treatment changes within the entirety of the 12-month study period.

Details on patient selection criteria have been described previously [19, 20]. Briefly, patients were included if they had two or more diagnoses of ADHD on distinct dates (International Classification of Diseases, Ninth/Tenth revision, Clinical Modification [ICD-9-CM codes 314.0x; ICD-10-CM codes F90.x]), had one or more prescription fill for an FDA-approved ADHD treatment (i.e., stimulants and non-stimulants) on or after the first observed ADHD diagnosis, and received no ADHD treatment 6 months prior to the index date. All patients included in this study were required to have 18 months of continuous health plan enrollment following the index date (i.e., 12-month study period and 6-month follow-up period).

Patients were stratified by comorbidity profiles of interest (i.e., anxiety, depression, and anxiety and/or depression), which were assessed during the 6-month baseline period using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) ICD-9/10-CM coding recommendations (see Supplementary Table 1 for the list of codes) [5]. Patients with ADHD who had no comorbid anxiety or depression observed during the baseline period were considered to have no comorbidity profile of interest (i.e., the without comorbidity cohort).

For the cost analysis, patients with a comorbidity profile who had at least one treatment change following the ADHD diagnosis observed during the 12-month study period were further stratified into three mutually exclusive cohorts based on the number of treatment changes observed (i.e., 1, 2, or ≥ 3).

Study measures and outcomes included baseline patient characteristics (e.g., age, gender, comorbidities), treatment characteristics (e.g., types of index pharmacological treatment at the class and agent levels, use of treatment combinations [i.e., two or more ADHD-related agents] and/or psychotherapy, and treatment duration of the first regimen observed), treatment changes, and total annual direct healthcare costs associated with treatment changes.

Definitions for a treatment regimen and treatment changes were as previously published [19, 20]. Briefly, treatment discontinuation was defined as no ADHD-related agents for at least 180 consecutive days after the last day of supply of the treatment regimen; treatment switch was defined as initiation of a new ADHD-related agent with no prescription fills from the previous treatment regimen within the 30 days following initiation; treatment add-on was defined as initiation of a new ADHD-related agent with one or more other prescription fill from the previous treatment regimen within the 30 days following initiation; and treatment drop was defined as discontinuation of an ADHD-related agent while other agent(s) were not discontinued. The frequency and odds of experiencing a treatment change were examined during the 12-month study period.

Total annual direct healthcare costs included medical (i.e., inpatient, outpatient, and emergency department visits) and pharmacy costs and were measured from the index date until the end of the 12-month study period. Costs were assessed from the payers’ perspective and reported in 2019 US dollars.

All measures and outcomes were reported separately by age group (i.e., children [aged 6–12 years], adolescents [aged 13–17 years], and adults [aged  18+ years]) and by comorbidity profile (i.e., anxiety, depression, and anxiety and/or depression).

Baseline patient and treatment characteristics as well as frequency of treatment changes were described using means, standard deviations, and medians for continuous variables and frequencies and percentages for categorical variables.

Odds ratios (ORs) of experiencing a treatment change were estimated between patients with and without a comorbidity profile of interest using logistic regressions. Total annual direct healthcare costs were compared between patients with and without treatment changes using ordinary least-squares regression models.

ORs and cost differences were adjusted for the following a priori selected demographic and clinical characteristics that had a standardized difference of at least 0.1 between cohorts: age, gender, region, health plan, type of ADHD diagnosis, year of index date, and number of patients who had at least one psychotherapy visit (children only). Adjusted effect size was reported along with 95% confidence intervals (CIs), using robust standard errors, and p values.