Opportunities for Pharmacogenomics in Pediatrics: Prescribing Trends of Psychiatric Medications With Pharmacogenomic Implications at a Multistate Pediatric Health System

Introduction

The Clinical Pharmacogenetics Implementation Consortium (CPIC) has published guidelines providing pharmacogenomic (PGx) recommendations for more than 100 drugs¹; however, limited data exist describing prescribing patterns of these medications in pediatric populations.² With increasing evidence describing the benefits of PGx testing to tailor drug therapy in psychiatric conditions,³ along with a worsening mental health crisis in pediatrics,⁴ it is vital to assess the prevalence of medication prescribing and potential impact of implementing PGx testing in this population. In this study, we describe the prescribing patterns of psychiatric medications with PGx implications over the past decade across Nemours Children’s Health (NCH), a multistate pediatric health care system.

Methods

Prescription data for psychiatric drugs classified as CPIC level A/B⁵ (Table 1, based on CPIC classifications as of August 1, 2023), hereinafter referred to as “PGx psych” medications, were extracted from the PEDSnet database⁶ for all NCH locations from January 1, 2010, through December 31, 2020. These data included medication name, dose, route, frequency, date prescribed, prescribing provider specialty/department, site, and patient demographics. Only the first prescription for each unique drug per patient was included, unless otherwise specified. For prescribing by gene, patients were counted as based on the first prescription for each unique gene. For PGx psych medication prescribing by department, patients were counted as based on the first prescription for each unique patient prescribed by each department, therefore patients could be counted for more than 1 department. Drugs with genetic implications related to specific disorders (i.e., valproic acid, divalproex sodium), orders for topical administration, and patients older than 21 years were excluded. All data were analyzed in R_4.3.0 or GraphPad Prism_9.4.1. Descriptive statistics were used as appropriate. Year-over-year changes in prescribing patterns were calculated by using Supplemental Equation S1.

Table 1.Patient Demographics^*

View Fullscreen

Results

A total of 21,442 unique patients received at least 1 PGx psych medication between 2010 and 2020 across NCH (Table 1). The number of unique PGx psych medications prescribed for each patient was 1 (n = 17,726; 82.7%), 2 (n = 2901; 13.5%), 3 (n = 656; 3.1%), and 4 or more (n = 159; 0.7%). The most frequently prescribed medications were amitriptyline and sertraline (Table 2). Overall prescribing was highest in the departments of neurology, primary care, and psychiatry (Figure A and B; Supplemental Tables S1 and S2).

View Figure

• Download as Powerpoint
• Download Figure

Table 2.PGx Psych Medication Prescribing by Unique Patient 2010–2020

View Fullscreen

To assess the prescribing of new PGx psych medications across the study period, we selected unique first prescriptions for all included medications for each patient from 2011–2020 (Supplemental Table S3). The most frequently prescribed medications by class were selective serotonin reuptake inhibitors (SSRIs) and tricyclic antidepressants (TCAs); serotonin-norepinephrine reuptake inhibitors (SNRIs) and anticonvulsants were infrequently prescribed (Figure C). SSRI and TCA prescriptions demonstrated a 10-year increase of 142.3% and 43.7%, respectively. Medications prescribed to treat attention-deficit/hyperactivity disorder (ADHD), antipsychotics, and SNRIs demonstrated no significant change in prescribing over time, while anticonvulsants demonstrated a decrease in prescribing rates over time.

Given the increase in SSRI prescriptions, we further explored their prevalence over the study period (Supplemental Table S4). The most frequently prescribed SSRIs included sertraline, escitalopram, and citalopram (Figure D). Over time, escitalopram and sertraline demonstrated the greatest increase in prescription of 319.5% and 212.4%, respectively. Citalopram prescriptions were notably reduced over time with a 10-year decrease in prescribing of 66.7%. As the total number of SSRI prescriptions increased over time by 142.3%, we also reviewed the share of each SSRI relative to the total number of SSRIs prescribed (Figure E). Relative prescribing of both escitalopram and sertraline increased from 20.0% to 34.7% and 46.1% to 59.5%, respectively. Citalopram prescriptions declined from 28.3% to 3.9%. From 2017 onward, escitalopram and sertraline made up >88% of all newly prescribed SSRIs.

Discussion

Across a large multicenter pediatric cohort, we found increasing use of psychiatric medications that could be guided by PGx data. Identification of high-prescribing departments and specific medications prescribed will help focus PGx implementation and education efforts.⁷ Previous reports suggest certain SSRIs, notably citalopram and escitalopram, are some of the most commonly prescribed CPIC level A medications for pediatric patients.⁸ We found SSRIs to be among the most frequently prescribed medications in this cohort; however, the most frequently prescribed SSRIs here were sertraline and escitalopram. This difference is likely due to the recent inclusion of sertraline to CPIC level A and exclusion of sertraline from prior studies.⁷ While some previous studies only examine CPIC level A medications, we chose to include medications with CPIC level B ratings because these medications have at least 1 recommendation for change in prescribing based on genetics.⁵ While the evidence is rated as weaker than for CPIC level A medications, CPIC level B gene-drug pairs are often included in clinical implementation.⁹

There are several key limitations to this study. First, this study did not include analysis of medication dosing, duration of use, indication, or existing PGx results.^10–12 Secondly, as a single center, retrospective, cross-sectional study, these results are subject to biases that could affect both the internal and external validity of our findings. Our findings may not apply to institutions in other geographic areas or serving different patient populations. Internal changes at NCH, including provider changes, may have had an effect on our findings. Additionally, we used CPIC guidelines to define potential actionability in this study. While the CPIC guidelines may be used for both adults and pediatrics, the data supporting their use in pediatrics are less well established. Finally, we did not include psychiatric medications without PGx implications, which prevents an accurate estimation of opportunities for PGx testing, because testing may be considered prior to medication use or in scenarios where a CPIC-rated medication was not ultimately prescribed. This also underestimates the overall increase in psychiatric medication prescriptions.

Given the worsening mental health crisis and increasing evidence demonstrating benefits of PGx testing for mental health conditions, the data presented here suggest an opportunity to provide PGx-guided care. Further research is necessary to optimize implementation of PGx testing into the care of pediatric patients.