Care Team Attributes Predict Likelihood of Utilizing Pharmacogenomic Information During Inpatient Prescribing

Abstract

Medication prescribing is imperfect, and unintended side effects complicate patient care. Pharmacogenomics (PGx) is an emerging solution that associates genotypes with personalized drug-related outcomes, but it has not been widely adopted. We hypothesize that patient and provider attributes may predict and promote PGx utilization. We studied PGx using data from the ACCOuNT study, a multi-institutional prospective trial that implemented broad preemptive PGx result delivery for African American inpatients [Clinicaltrials.gov NCT03225820]. Patients were genotyped, and their PGx information was made available within an integrated informatics portal. Utilization of PGx data (defined as the active choice to review PGx information) was left to the enrolled provider's discretion. Our primary endpoint was to identify patient and care team attributes associated with PGx use. We identified statistically significant univariate predictors and utilized logistic regression to compare relative predictiveness. This study included 187 patients (60.4% female, median age 55, 75.4% treated at the University of Chicago, 17.6% at Northwestern University, and 7.0% at the University of Illinois Chicago) and 188 providers (63.8% MD, 22.3% PharmD, 6.4% PA, and 7.4% APN). In multivariate analysis, we found that the use of PGx information in a prior admission significantly predicted the use in subsequent admissions (OR 7.62, p < 0.05). Similarly, pharmacist participation on care teams significantly predicted PGx use (OR 4.52, p < 0.05). In the first systematic analysis of the impact of patient and care team factors on inpatient PGx clinical decision support (CDS) adoption, we found that actionable care team attributes, such as pharmacist participation or successful initial adoption measures, predict PGx CDS use.

Keywords: personalized medicine; pharmacogenomics; precision medicine.

FIGURE 1

Predicted likelihood of PGx CDS accession to view pharmacogenomic results during inpatient care. The base case scenario consists of a median‐length admission (6 days) for a median‐age patient (55 years old) cared for by one enrolled study physician prescribing a median number of medications (6) at the low‐adoption institution (Institution A).