This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Antidepressants are used by primary care providers to treat a variety of conditions, including (but not limited to) depression and anxiety. A trial-and-error approach is typically used to identify effective therapy, as treatment efficacy and safety can vary based on the response, which is affected by certain gene types. Pharmacokinetic pharmacogenomic (PGx) testing provides phenotypic classification of individuals as poor, intermediate, extensive, and ultrarapid CYP450 metabolizers, providing information for optimal drug selection.
The objective of this pilot study is to examine the feasibility, acceptability, and preliminary effectiveness of PGx testing when used after starting a new antidepressant medication.
We are conducting a pilot study with physicians from 6 Department of Family Medicine clinics at the University of Michigan who are willing to use PGx test results to manage antidepressant medication use. From enrolled physicians, patients were recruited to participate in a 6-month randomized, wait-list controlled trial in which patient participants newly prescribed an antidepressant had PGx testing and were randomized equally to have the results released to their primary care physician as soon as results were available or after 3 months. Patients were excluded if they had been taking the antidepressant for more than 4 weeks or if they had undergone PGx testing in the past. Physician participants completed a baseline survey to assess demographics, as well as knowledge, feasibility, and acceptability of PGx testing for this population. At the conclusion of the study, physician participants will complete a survey to assess knowledge, satisfaction, feasibility, acceptability, perceived effectiveness, and barriers to widespread adoption of PGx testing. Patient participants will complete a baseline, 3-month, and 6-month assessment, and control patient participants will have an additional 9-month assessment. Data collected will include the reason for antidepressant use, self-reported medication adherence, side effects, patient health questionnaire 8-item depression scale, generalized anxiety disorder 7-item scale, 12-Item Short-Form Health Survey, work status or changes, and physician and emergency department visits. PGx knowledge and perceptions (including acceptability and feasibility) as well as demographic information will also be obtained.
We recruited 23 physician participants between November 2017 and January 2019, and 52 patient participants between January 2018 and April 2019. Currently, all physician and patient participants have been recruited, and we expect data collection to conclude in January 2020.
This study will examine the preliminary effectiveness of PGx testing after treatment initiation and determine the feasibility and acceptability of PGx testing for use in primary care. Through this study, we expect to demonstrate the benefit of PGx testing and lay the foundation for translating this approach into use within primary care.
ClinicalTrials.gov NCT03270891; https://clinicaltrials.gov/ct2/show/NCT03270891
RR1-10.2196/13848
Antidepressant medications are the third most commonly prescribed drug type in the United States and are taken by 10.7% of American adults [
Antidepressant medication treatment and achievement of a positive clinical outcome are complicated by several factors. It can take 2 to 4 weeks for symptoms to improve [
Pharmacokinetic pharmacogenomic (PGx) testing of antidepressant drug metabolism by CYP450 genes and the functional variants (CYP2D6, CYP2C9, CYP2C19, and CYP1A2) may potentially assist with determining drug effectiveness and the possibility of side effects for an individual patient and to ensure the medication(s) prescribed are the best choice for that patient. Individuals can be phenotypically classified as poor, intermediate, extensive/normal or ultrarapid CYP450 metabolizers [
Barriers to adopting PGx testing into clinical practice include issues such as gaps in evidence-based data (insufficient or lacking usefulness), lack of genome-based medicine education, low patient awareness of utility, ethical concerns, inadequate support from prescribers, and inconsistent reimbursement [
The objective of this pilot study is to examine the feasibility, acceptability, and preliminary effectiveness of PGx testing when available just after, or 3 months after, initiating a new antidepressant medication. In terms of feasibility, we are asking physician participants to self-report barriers to PGx testing, as well as whether they would continue to recommend PGx testing. For acceptability, we are measuring physician and patient participants’ perceptions of PGx testing and whether they think the information is valuable. Finally, the preliminary effectiveness is assessed using antidepressant medication changes, as well as changes in medication adherence, patient symptoms, and health care utilization.
We are conducting a pilot study with primary care physicians and their patients. We enrolled physician participants who agreed to participate in a 1-group pre-post design study, using PGx testing with their enrolled patients. We then recruited patient participants seen by 1 of our participating physicians to enroll in a 6-month, randomized, wait-list controlled trial in which all patient participants newly prescribed (within the last 4 weeks) an antidepressant had PGx testing. The wait-list controlled study design allows us to both have a true control comparison group and also allows us to look at longitudinal effects. Patient participants were randomized equally to 2 groups; one group had their PGx results released to their physician immediately after the baseline visit (intervention), and the other group had results sent to the physician 3 months after their baseline visit (control). All patients enrolled in this trial were recently prescribed one of the target medications by their participating physician. Although PGx test results and any resulting treatment recommendations are made to participating physicians by the study pharmacist, it is up to the physician’s discretion to determine if, and how, test results will be used in clinical care.
The primary outcome for effectiveness is the proportion of patients who were prescribed antidepressant medications which are not contraindicated based on PGx test results. Secondary effectiveness endpoints include the change in symptoms or symptom severity as indicated by changes in the 8-item Patient Health Questionnaire (PHQ-8) depression scale [
PGx test results are provided to the clinical pharmacist through the
We provided primary care physicians their patients’ PGx test results, interpretation, and antidepressant medication therapy recommendations from our trained clinical pharmacist through a Research Subjective, Objective, Assessment, and Plan (SOAP) note in the electronic medical record (EMR; see
This pilot study is being conducted with physicians and patients from 6 Department of Family Medicine (DFM) clinics at our academic medical center in 5 neighboring communities.
To conduct this study, we recruited 2 groups of participants (physician and patient participants).
We used convenience sampling to identify physician participants. To be eligible for the study, physician participants (1) must be practicing at 1 of the 6 UM DFM clinics, (2) self-report that they were willing to prescribe antidepressants, (3) must be willing to use PGx test results for the management of antidepressant medications, and (4) must be willing to use PGx test results in the treatment of their patient participants enrolled in the study. Physician participants were recruited through a variety of methods, including presentations at faculty business meetings and clinical site staff meetings, as well as targeted emails or letters and follow-up phone calls. All physician participants completed the consent process and will receive no incentives for their participation in this study.
Once physician participants were enrolled, we recruited participants from among their patients (
We used purposive sampling to identify patient participants. Potential patient participants were identified each week through an automated report from the EMR. Potential participants received 3 targeted recruitment points of contact—initially an email or postal letter, followed by 2 phone calls, emails, or text messages. Patient participants were also recruited through direct provider referral, as well as through posted flyers in the DFM clinics. If a potential participant was eligible to participate in the study, except for the fact that their primary care physician was not yet participating, we attempted to recruit the physician through emails and phone calls.
Both physician and patient participants were screened for inclusion and exclusion criteria before obtaining consent and enrollment. Once screened as eligible and enrolled, physician participants completed baseline assessments (
Potential patient participants who were screened as eligible gave consent and then completed baseline assessments, as well as had their blood drawn (3 mL) for PGx testing. Patient participants were then randomized to have PGx test results sent to their enrolled physician as soon as the results were available (intervention) or after 3 months (control;
Initially, randomization of patient participants was set to be stratified by physician to ensure balance within provider; however, given the small patient-to-provider enrollment ratio, stratification by provider was not deemed necessary. Instead, we used a block randomization approach, randomizing patients into 1 of the 2 arms using a block size of 6. The randomization schedule was unknown a priori to researchers enrolling patients and revealed to the patient after completion of the baseline survey.
PGx test results were released to physician participants as a SOAP-formatted research note in the EMR for which the physician participant was alerted through electronic notification. The PGx test results and recommendations provided were for the specific antidepressant—enzyme pair (see
Patient participant flowchart. *Patient was withdrawn from the study 3 months after enrollment because it was discovered that the physician participant listed in the electronic medical record had no previous contact with the patient, and the physician who prescribed the antidepressant did not want to enroll in the study.
Physician participant study flow. PGx: pharmacogenomic.
Patient participant study flow. PGx: pharmacogenomic.
No study staff member, including the clinical pharmacist, will provide PGx results to patients; it is up to the physician participants if they choose to share this information with their patients. All intervention patient participants are followed for 6 months, with phone-based follow-up at 3 and 6 months. All control patients are followed for 9 months, with phone-based follow-up at 3, 6, and 9 months.
Physician participants completed a self-administered investigator-developed paper survey at baseline. The baseline survey assessed demographics and physician participants’ characteristics, knowledge of PGx testing, and perceptions of the feasibility and acceptability of PGx testing in primary care (
At the end of the study, physician participants will self-administer a paper survey. This end of study survey contains similar questions as the baseline survey, with the addition of open-ended questions asking whether physician participants recommended PGx testing to other physicians or whether they ordered PGx testing on patients not enrolled in this study. For the open-ended questions, if the physician agrees and would rather give verbal responses to the questions, their responses will be audio recorded for subsequent qualitative analysis.
Feasibility will be assessed by responses to the poststudy questionnaire, which asks physicians to indicate the extent to which they agree with the statements “I think that the process of learning how to use PGx testing in my daily practice would be easy” and “I believe that I could easily incorporate PGx testing into my clinical practice.” Acceptability will be assessed by responses to the poststudy questionnaire, which asks physicians to indicate the extent to which they agree with the statements “I see the potential benefit of using PGx testing in my clinical practice” and “If I were to recommend that patients undergo PGx testing, I believe that my patients would choose to be tested.”
Physician participant measures of feasibility, acceptability, and preliminary effectiveness of pharmacogenomic testing within the investigator developed semistructured interview and survey.
Construct | Assessment time | |
|
Baseline | Poststudy |
Demographic information and characteristics of physicians | ✓a | —b |
Knowledge of PGxc testing | ✓ | ✓ |
Acceptability of PGx testing | ✓ | ✓ |
Utility of PGx testing | ✓ | ✓ |
Effectiveness of PGx testing | ✓ | ✓ |
Feasibility of PGx testing | ✓ | ✓ |
Barriers to PGx testing | ✓ | ✓ |
Satisfaction with PGx testing | ✓ | ✓ |
a✓ indicates that the construct was assessed.
bNot assessed.
cPGx: pharmacogenomic.
All patient participants completed a paper baseline survey, had their blood drawn for the PGx test, and had a pill count of all prescription medications completed at baseline. All patient participants are asked to complete the 3- and 6-month follow-up survey and pill count by phone, and control patients have an additional 9-month follow-up survey and pill count. The baseline survey assessed demographics, participants’ characteristics, and health history. All follow-up assessments completed after baseline are telephonic. The baseline and follow-up phone surveys contained questions from the PHQ-8, GAD-7, 12-Item Short Form Survey, Work Productivity and Activity Impairment Questionnaire, the Antidepressant Side-Effect Checklist, and the ARMS. The baseline and follow-up surveys also contained questions from the 3-item self-rated adherence scale, questions about health care utilization, prescribed medications, and ascertained PGx knowledge from an investigator developed survey [
All survey data, including patient and provider demographics and characteristics, will be analyzed using descriptive statistics to ensure data quality. Independent samples
Effectiveness using patient data will be assessed using statistical models. Whether the patient participant is currently prescribed an antidepressant that is appropriate (not contraindicated) based on the PGx test result will be assessed either as soon as results are available or 3 months after the baseline visit, depending on randomization. Logistic regression models, using a generalized estimating equation framework to account for within-subject dependence, will assess the effect of the intervention on the likelihood of being prescribed an appropriate medication by including time, study group, and time by study group interaction as predictors. Secondary effectiveness measures (adherence, depression severity and symptoms, health status, and health care utilization) will be analyzed similarly using linear mixed models, given the continuous nature of the measures. These models will be used to compare the average change in outcomes between study group by including time, study group, and time by study group interaction as the primary predictors of interest for each outcome. For all models, time will be measured both as survey time and time since test results were released to physicians. In the latter, data from the baseline to 6 months period for the intervention group and from 3 to 9 months for the control group will be assessed. All models will be adjusted for patient demographics.
Patient participant measures of feasibility, acceptability, and preliminary effectiveness of pharmacogenomic testing.
Construct | Measure | Assessment time | ||||
|
|
Baseline (through paper survey) | 3 months (through phone) | 6 months (through phone) | 9 months (control only; through phone) | |
|
||||||
|
Demographic information | Gender, age, and income | ✓a | —b | — | — |
|
||||||
|
Depression symptoms and severity | Eight-item Personal Health Questionnaire Depression Scale | ✓ | ✓ | ✓ | ✓ |
|
Anxiety symptoms and severity | Generalized Anxiety Disorder 7-Item Scale | ✓ | ✓ | ✓ | ✓ |
|
Functional health status | 12-Item Short Form Survey | ✓ | ✓ | ✓ | ✓ |
|
Missed days of work | Work Productivity and Activity Impairment Questionnaire | ✓ | ✓ | ✓ | ✓ |
|
Health history | Smoking history, body mass index | ✓ | — | — | — |
|
Health care utilization | Physician visits and hospital and emergency room admissions | ✓ | ✓ | ✓ | ✓ |
|
Adverse drug reactions | Antidepressant Side-Effect Checklist | ✓ | ✓ | ✓ | ✓ |
|
Medication and medication changes | Medications prescribed and number of medication changes | ✓ | ✓ | ✓ | ✓ |
|
Medication adherence | Adherence to Refills and Medication Scale | ✓ | ✓ | ✓ | ✓ |
|
Medication adherence | Pill count | ✓ | ✓ | ✓ | ✓ |
|
Medication adherence | Three-Item Self-rated Adherence Scale | ✓ | ✓ | ✓ | ✓ |
|
||||||
|
Pharmacogenomic knowledge, perceptions, and experiences | Investigator Developed | ✓ | — | ✓c | ✓ |
a✓ indicates that the construct was assessed.
bNot assessed at that time.
cIntervention arm only.
This study is ongoing. Physician enrollment occurred between November 2017 and January 2019, and we recruited 23 physician participants. Patient participant enrollment occurred between January 2018 and April 2019, and we recruited 52 patient participants. We had to withdraw 1 patient because their physician had not seen the patient clinically, nor was the physician the prescriber (see
At the outset of this study, we sought to recruit patient participants who had recent diagnoses of depression and/or anxiety recorded in their medical record. However, the initial pool of potentially eligible participants was quite small, and enrollment rates were poor. In May 2018, the IRB approved protocol changes to widen the inclusion criteria to any patient with a new prescription for an antidepressant medication from one of the enrolled physicians, regardless of clinical diagnosis, as reflected in this revised protocol. This expansion in inclusion criteria resulted in some participants not having mental health–related diagnoses; however, it is likely that many participants may have undocumented depression or anxiety.
This paper outlines the protocol of a pilot study, which includes a wait-list randomized controlled trial investigating PGx testing for antidepressant medications in primary care. Despite commercial availability of PGx testing, institutions have been slow to adopt these tests [
Despite underutilization, previous work has suggested that PGx test results could improve patient care by reducing the number of side effects, failed medication trials, and time to achieve treatment response [
One of the perceived barriers to utilization of PGx testing in primary care is that clinical utility of the test has not been well established [
This is a pilot study with a small sample size of patient participants, which limits sample genetic variability (there are more than 50 CYP450 enzymes), the number of abnormal metabolizers, and differences in participant demographics. As previously described, we focused on CYP2C19 and CYP2D6, which metabolize the majority of the most commonly prescribed FDA-approved antidepressant medications and have CPIC guidelines to support therapeutic recommendations. In addition, because this is a pilot study, we are also limited by our relatively small sample size of patient participants who were recruited from a convenience sample of DFM physicians within our institution. Future work is needed with larger sample sizes adequate to detect a difference between groups, ensure demographic saturation, and provide a broader representation of genetic variability.
To increase enrollment, we expanded our protocol inclusion criteria from a diagnosis of depression and/or anxiety to anyone taking targeted antidepressants. The broadening of possible diagnoses increases variability in comorbidities and may change certain patient outcomes. Future studies should limit the diagnoses to avoid these variations.
As with any PGx test, there are limitations to the assay. Direct DNA testing will not detect all variants that result in decreased or increased enzyme activity. Absence of a detectable gene mutation or polymorphism does not rule out that the patient has ultrarapid, intermediate, or poor metabolizer; reduced activity; or reduced response phenotypes for these genes. More research is needed to identify all genetic variations and their possible phenotypes.
This study is being conducted in primary care clinics from 1 academic health center in the United States, which limits the generalizability of the results. Only physicians who were willing to use PGx testing for the selection of antidepressant medications were enrolled, which may be a potential bias, and limits assessment of feasibility and physician willingness to use PGx testing in clinical practice. Ideally, we would have enrolled patients before starting the antidepressant; however, delaying antidepressant therapy while waiting for PGx test results would be unethical.
At our institution, clinical pharmacists have full access to the EMR, are able to write notes, and contact prescribers using direct messaging through the EMR. These interactions are recorded within the EMR. DFM clinics are a Patient-Centered Medical Home and have clinical pharmacists practicing under a collaborative practice agreement within each clinic. Physicians are accustomed to receiving and sending information through the EMR to clinical pharmacists to facilitate patient care. Although not unique to our clinics, this practice model is not commonly used throughout the United States.
This pilot study is designed to assess the feasibility, acceptability, and preliminary effectiveness of using PGx testing in a primary care setting among persons newly prescribed an antidepressant. Physicians’ and patients’ perceptions of PGx testing, as well as changes in patient medication use and outcomes, will be evaluated. Results from this study may yield positive effects for all patients on antidepressant medications, not just those with anxiety and/or depression.
Pharmacogenomic subjective, objective, assessment, and plan note template.
Adherence to Refills and Medication Scale
Clinical Pharmacogenetics Implementation Consortium
Department of Family Medicine
electronic medical record
Food and Drug Administration
Generalized Anxiety Disorder 7-item scale
odds ratio
pharmacogenomic
Patient Health Questionnaire 8-item depression scale
Subjective, Objective, Assessment, and Plan
University of Michigan
This research was funded by MCubed, an internal funding mechanism at the University of Michigan (UM). MCubed is part of the UM’s Third Century Initiative, established by the Offices of the President and Provost, and housed in the UM’s Office of Research. Support for the project was also provided by the Michigan Institute for Clinical & Health Research, specifically grant UL1TR002240 for data systems support. The authors would like to thank Anisa Bici, Shivang Danak, Erica Godley, Joseph Hubbard, Emily Kaip, Dana Roberson, Markie Silverman, Fatima Waheeli, and Abigail Wilson for assistance with this project and Jill Bowdler, Judy Connelly, Lilly Pritula, and Rania Ajilat for administrative support. They would also like to thank Arina Bierdz from the Data Office for Clinical & Translational Research for the automated reports for the project. Finally, the authors would like to thank Progenity, LLC (in particular, Jeffrey Buis, Paul Bien, and Jay Stoerker) for performing the PGx testing.
LRB is obliged to disclose a conflict of interest, as her spouse is an employee and stock option holder at Progenity.