Table 4.

Summary of challenges and potential solutions

Challenge		Potential solution/lesson learned
Trial design	Randomization	• Randomizing at the patient level provides more similar baseline characteristics between comparison groups versus randomization by enrollment site (e.g., cluster randomization)
	Questionnaire completion	• Patients are willing to complete questionnaires so long as they are not overly burdensome or too frequent • To assess association between outcome and use of pharmacogenetics, the majority of questionnaires need to be completed, and a high percentage of patients with actionable phenotype need to have drug changed
Provider education	Prescriber knowledge gaps	• Prescribers learn best from case-based education; providing this type of education upfront as well case discussion sessions throughout is ideal • Having prescribers undergo personal genotyping is a beneficial educational method • Offering online CME is not enough incentive to complete education • In-person education methods (e.g., noon conference/grand rounds) are well attended
Test order and interpretation	Identifying who to test	• Prescribers would like electronic decision support tools to identify potentially appropriate patients to test
process	Children do not like blood draws	• Offering noninvasive genetic sample collections is key for younger populations • Blood draws do not seem problematic in adults, however may not be the most convenient option as not all clinic locations have phlebotomy stations
	Turnaround time for pharmacogenetic test results	• Genotype should be available during patient encounters to optimize prescriber’s ability to act on it • Prescribers, patients, and parents are willing to wait for results for drug therapy decisions in some settings
	Phenotypes resulting as a range or indeterminate	• Having a testing platform that can detect which CYP2D6 allele is duplicated is ideal • Blood samples may result in fewer undetermined phenotypes as compared with buccal samples
	Phenotype results	• Normal metabolizers, although not a classic “actionable” phenotype, are clinically informative • Concomitant medications (e.g., CYP2D6 inhibitors) must be considered to evaluate true phenotype, especially in patients who are genotypically normal metabolizers
	Prescriber interpretation and integration of pharmacogenetic results	• Prescribers highly depend on consults from pharmacists • Clear concise guidance must be provided through active alerts or consults in the EHR • Prescribers value eventual availability of pharmacogenetic results for their patients enrolled in control armsa • Genotype data availability at the time of prescriber–patient encounter, and clarity for clinician of availability of the genotype data and recommended actions is likely important for high levels of adherence to genotype-guided recommendations
EHR integration	Prescribers cannot recall every patient who was genotyped	• Prescribers do not necessarily look in the chart for pharmacogenetic results, especially if they are buried with other lab results  ◦ An ideal solution is a section of the patient’s chart for genetic results, and a quick indicator to note if there are results in there  ◦ A workaround is ensuring the prescriber notes in their encounter note that they are ordering a pharmacogenetic test; then, they will review prior to patient’s next visit and will know to look for results • Prescribers want alerts to tell them exactly what to do if results are available
Patient perceptions	Patient perceptions and knowledge of pharmacogenetics	• Patient education to manage patient expectations on what information pharmacogenetic results will provide is essential • Having brief patient-friendly educational materials for before and after testing is ideal • Patients value receiving pharmacogenetic results at the end of the study

CME continuing medical education, EHR electronic health record.

^aAll but the CYP2D6-opioid cancer pain trial offered control participants pharmacogenetic testing.