PRIME Care providers’ use of pharmacogenetic testing to inform antidepressant prescribing: Results of qualitative interviews

Abstract

Objective:

Pharmacogenetic (PGx) testing in the treatment of depression is associated with modest improvements. However, little is known about providers’ experiences, including how, and for which patients, they incorporate test results into prescribing decisions. The study goal was to qualitatively: 1) explore how providers used PGx testing in prescribing decisions for patients with major depressive disorder during a PGx pragmatic trial, 2) highlight implications of providers’ experiences for future implementation.

Methods:

Interviews were conducted with providers participating in the trial (N=61) using questions informed by the Consolidated Framework for Implementation Research. Analysis used a rapid analytic approach.

Results:

Analysis identified two main themes: 1) perceptions of patients best suited to benefit from PGx testing; 2) approaches to using PGx test results in prescribing. Most providers expressed positive experiences with using PGx testing to inform depression medication prescribing. There was variation in how providers used the test in clinical decision-making, including uncertainty about the extent to which results should be relied upon. Providers had varied perceptions of which patients are suitable for this testing and in the application of test results to prescribing.

Conclusions:

Future implementation of PGx should establish policy and procedure around test usage, as well as mechanisms to support ongoing provider education.

Introduction

Pharmacogenetic (PGx) testing is a novel practice with a growing literature to support its use in psychiatric prescribing.¹ In theory, PGx testing provides information on how a patient’s genetic makeup may affect medication experiences, enabling providers to align medication selection with individual genetic profiles and potentially avoid side effects. Such information may be especially beneficial in the treatment of depression, as poor medication experiences are common. Patients often try multiple antidepressants before finding an effective one, while chances of treatment drop-out increase with each failed trial.² Further, poor gene-medication matches may affect a significant proportion of patients. One study found that approximately one quarter of patients treated in primary care were prescribed an antidepressant that may not have been a good fit based on their pharmacogenomic profile.³ Other recent studies of PGx for depression prescribing found significant gene-medication interactions in approximately 20% of patients.^4–6 Data collected from trials thus far demonstrate that use of PGx testing in the treatment of depression is associated with a modest increase in response and remission rates,^1,4,6 although additional studies are needed to determine which tests perform best for which groups of patients.¹

The Precision Medicine in Mental Health Care (PRIME Care) study was designed to add to the evidence base by examining the utility of PGx in the treatment of depression while exploring provider perceptions.⁷ PRIME Care was a pragmatic randomized clinical trial, assessing the effectiveness of using a commercially available PGx test in the treatment of major depressive disorder in the Veterans Health Administration (VA).^6,7 This 22-site study assessed whether treatment guided by the PGx results would lead to a higher rate of depression remission and decreased use of medications with potential gene-drug interactions. As a type 2 hybrid trial,⁸ the PRIME Care design included implementation science methods to understand how PGx was used in clinical practice during the trial and to inform future implementation.⁷ This approach, guided by the Consolidated Framework for Implementation Research (CFIR; ⁹ included detailed exploration of providers’ perceptions of the intervention and recommendations for implementation at the beginning of the trial,^10,11 as well as an examination of providers’ experiences after using the PGx testing in psychiatric prescribing.

As the evidence for PGx effectiveness in depression care has developed, investigators have begun to study provider and patient perceptions of its use e.g., ^12,13–15. While perceptions are overall positive, very little is known about providers’ actual experiences of using the test in psychiatric prescribing, including how, and for which patients, they incorporate the test result information into prescribing decisions. This paper uses qualitative data to explore how providers described their use of PGx test results in clinical decision-making for patients with major depressive disorder during the PRIME Care study. We also highlight the implications of providers’ experiences for future implementation.

Methods

Study Context

All PRIME Care study procedures, including the qualitative interviews with provider participants, were approved by the VA Central IRB, as well as the regulatory bodies at each of the participating sites. Providers gave informed consent to participate in the trial and the interviews.

PRIME Care used a commercially available combinatorial genetic test Genesight^®4 which produces a report consisting of a cover page that sorts antidepressant medications into three columns, based on the patients’ genotype: green or “use as directed”, yellow “moderate gene-drug interaction”, and red “significant gene-drug interaction”. The table is footnoted to include the specific polymorphisms that are associated with a higher risk of drug-gene interactions and the direction or functional nature of the predicted interaction (i.e., result in slower or faster metabolism of the drug), to help with dosing decisions. Similar tables were provided for additional psychotropic medication classes (e.g., antipsychotics). Additional pages of the report provide detailed information on specific genes. Each individual test costs approximately $1000 (undiscounted, retail price), the costs of which were covered by the testing company for purposes of the trial.

Recruitment for Qualitative Interviews

Interview recruitment focused on VA primary care and mental health providers enrolled in the clinical trial. Recruitment and interviews began approximately two years after the first patient was randomized to allow sufficient time for providers to experience using PGx results.

Providers who had not yet received a test result were excluded. To ensure a range of experiences, providers were divided into primary care or mental health, and then three groups based on the number of patients they had referred who were randomized into the study: low-referring (1–2 patients), mid-referring (3–7 patients), and high-referring (8 or more patients). Interview subjects were then recruited from each group via email.

Over time, as interviews were completed, recruitment efforts were purposively targeted to increase participation from needed categories (primary care vs. mental health providers, low vs. high referring, providers from particular sites, etc.). These efforts resulted in final participation that was representative of the larger study in terms of percentage of primary care vs. mental health providers, with approximately equal numbers across referral groups, and at least one interviewee from each of the participating trial sites.

Data collection

An interview guide was developed using the CFIR domains of Individual Characteristics, Inner Setting, and Outer Setting to inform the questions, and meet the aims of the larger implementation study (the Intervention Characteristics domain was addressed earlier in the study¹¹). Questions were written with the goal of eliciting thoughts that would map back to constructs within these domains to ensure the interviews would cover an array of factors likely to be important in future implementation of PGx. The guide was adjusted in an iterative fashion after an initial set of interviews to eliminate identified redundancies in the questions.

All interviews were conducted via telephone and audio-recorded. Participants were informed of the interview purpose and reminded of consent information and the voluntary nature of their participation before beginning. All participants verbally reaffirmed their consent to participate and to be recorded.

Interviews were conducted by two PhD-level medical anthropologists with a research assistant present to facilitate recording. All interviews were conducted between August 2019 and January 2021, with a pause from March-June of 2020 during the initial phase of the COVID-19 pandemic. Interviews generally ranged from 30–60 minutes (mean≈35 minutes).

Data analysis

Interview recordings were transcribed by a professional service, and transcripts were analyzed using a rapid analytic approach.¹⁶ This approach is an efficient method for processing large amounts of qualitative information with results comparable to those of more traditional line-by-line coding approaches.^17,18

Specifically, a template matrix was designed to organize the data into primary categories based on the interview question guide, summarize the content of each interview, and capture supporting quotations. One qualitative analyst was responsible for summarizing each individual interview based on the complete transcript. The summary and supporting details were then reviewed by a second analyst for accuracy, completeness, and consistency. The analysts then met to review the summaries and make minor adjustments to the content based on the review process.

The descriptive content from each individual interview was then transferred to a matrix to facilitate integration and comparison of content from all the transcripts. The matrix contained rows for each category and a column for the summarized content of each interview. One analyst then reviewed the content across each category from the entire sample of interviews and drafted a summary of the content for each category. This summary was then reviewed by a second analyst for accuracy of interpretation and characterization of the data. Importantly, the summaries noted not only common themes and ideas, but also exceptions and observations of differences between the transcripts and participants.

Results

Table 1 displays the demographics and practice characteristics of the 61 interviewed providers. Most (80%) worked in specialty mental health; the remainder were primary care providers. The majority (60%) were MDs, 30% were Advanced practice nurses (n=18) or Physician Assistants (n=1), and the remainder were PharmDs.

Table 1.

Participant Demographics (N=61)

	N	%
Gender
Male	27	44.3
Female	34	55.7
Age
Under 30	2	3.3
31–40	14	23
41–50	13	21.3
51–60	24	39.4
Over 60	7	11.5
Missing/ Unknown	1	1.6
Race*
White	43	70.5
Black/African-American	3	4.9
Asian	11	18.0
Prefer not to answer/missing	5	8.2
Hispanic (Yes)	6	9.8
Year Completed Training
Before 1980	2	3.3
1981–1990	3	4.9
1991–2000	11	18.0
After 2000	44	72.1
Missing	1	1.6
Referral Group
Low (1–2)	17	27.9
Mid (3–7)	24	39.4
High (8 or more)	20	32.9
Specialty
Primary Care	12	19.4
Mental Health	49	80.3
Provider Type
MD	37	60.7
Nurse Practitioner/  Physician Assistant	19	31.1
PharmD	5	8.3

^*Note: totals may not equal 100% because participants could choose more than one option

Qualitative analysis resulted in identification of two main themes: 1) perceptions of patients best suited to benefit from PGx testing and 2) approaches to using PGx test results in prescribing (Box 1). The range of themes and sub-themes identified were remarkably consistent regardless of provider type, setting (PC vs. MH), or use of the intervention (low vs. mid vs. high) and therefore are presented in aggregate.

Box 1. Provider Use of PGx Testing: Identified Themes and Sub-Themes.

Theme 1: Perceptions of patients best suited to benefit from PGx testing

• Patients with previous unsuccessful medication trials for depression/treatment-resistant depression

• Patients who were resistant to/ uncomfortable with the idea of medication to treat depression

Theme 2: Approaches to using PGx test results in prescribing

• Test results may only be influential for some patients

  • Sometimes all medications in green “use as directed” column
  • Results did not always seem to match patient experience of medications
  • Results often not specific enough (i.e., had long lists of medications in each category)

• When results were influential, they were exciting

• Strategies for use of PGx test varied

  • Some relied on cover page & tried to use green column meds
  • Some prescribed meds in the yellow or red “use with caution” columns but used detailed information pages to adjust dosing

• Comfort with test varied especially with the more detailed information provided

Perceptions of patients best suited to benefit from PGx testing

Most providers saw PGx testing as most suitable for patients who had already unsuccessfully tried medications for depression, with some specifically mentioning treatment resistant depression:

“Generally, patients that were included were patients that maybe hadn’t done well on previous therapy and still needed help…basically patients that I thought would benefit if we had some additional information in choosing which therapy to go with next.” [PC Provider, #46]

Related to this, many providers discussed using the testing to help patients feel more comfortable if they were resistant to the idea of medication for their depression. The testing was seen as fulfilling this purpose regardless of whether patients had previously tried medication.

A lot of times I feel that patients have had multiple medication trials and they’re afraid to or they’re reluctant to try another medication. They’ve tried two or three and maybe four… some of them have tried more than that, seven or eight, and combinations, and they’re just reluctant. They’re not doing well and they’re reluctant, and so I think the study sort of gives them a little bit of hope like, oh, maybe rather than just throwing something at the wall and seeing if it sticks, we’ll have some data. [MH Provider, #30]

I used it for patients that I knew were very depressed that would not previously go on medication or had a lot of side effects with medications.…It’s like the equivalent of someone saying… “I’m gonna give you the exact medication that fits you.” They were more likely to try it, so in my resistant patients it was very useful, because they felt more secure. [PC Provider, #25]

Approaches to using PGx test results in prescribing

Frequently, providers felt that the test did not influence the medication decision for many of their patients. However, they were excited by cases where the test was especially impactful.

Well, it didn’t really change treatment decisions notably except for one person…The one time it did make a difference was when it showed that he would be a slow metabolizer, and so we would need to use a lower dose of the antidepressant than typically he’s done. [MH Provider, #46]

I got the results, which indicated a certain type of medications so I can clearly see which will be the better and which medications might not work as well, and quite often I saw medications that I offered the patient were actually in the column that were suggested to use, and sometimes they weren’t… I had one patient…one of them really made significant progress after I switched from one medication to another, quite remarkable change in his symptoms reduction…results were quite fascinating. [MH Provider, #28]

Some providers expressed frustration or disappointment, based on their perception that the test results simply indicated medications to avoid, without more specific guidance.

I only have a few folks. But often the results were not incredibly specific. Meaning, that there was a long list of medications that were in each category and so it didn’t narrow it down greatly that you should use this pill versus that pill. It was use this list and stay away from that list. [MH Provider, #2]

Most providers discussed primarily relying on the cover page of the results and trying to use medications in the “use as directed” (i.e., green) column first. Many providers also discussed reasons why they may not have selected a green medication, including patients’ comorbidities, associated medications, or preferences; medications not on the VA formulary; patients doing well on a medication in the yellow or red columns; and the provider’s clinical judgment. As one primary care provider expressed:

At least one time I picked one that was in the yellow. I’ve never picked one in the red column. I think there have been times when what I was using was in the yellow column. And I think at least one time maybe in the green column, it was just medications that I was not necessarily comfortable with. [PC Provider, #16]

One mental health provider offered detailed insight into the rationale for keeping patients on yellow or red category medications:

Oftentimes, people have different responses to medications anyways. I would get the results back …and then by that time, they would be like “Oh, I’m doing better,” but the medication that I started them on when I made the change before I got the results [as occurred with patients in the delayed results condition], it was working. It might have been in the yellow or red category and then the patient would get confused and I would explain… a lot of it looks at your metabolism and these medications. In psychiatry, sometimes the medications are not so clear…you start somebody on a medicine and something that you don’t think it’s going to work based upon how you think of it ends up working really well or something that you think is going to be a slam dunk doesn’t work for a reason that you can’t explain. There’s such a subjective experience of taking psychotropic medications that sometimes the results would make you think one way, but that just wasn’t the way the patient was experiencing it and then they would ask me “What would you do? It’s yellow. Should we change?” and I’m like “Well, you’ve been on this medication for a few months now and you’re doing well… I don’t think we should change it.” [MH Provider, #50]

Some providers said they used the test to inform dosing considerations, rather than to completely switch medications, using the medication they already had in mind for a patient regardless of which column it fell in, but adjusting the dosing to be either higher or lower than usual based on the genetic results. This approach was particularly common among mental health providers, some of whom saw these dosing adjustments as a particularly valuable aspect of the testing.

Dosing. That is one of the things I appreciate for the medications that do have gene drug interactions, like the recommendations for dosing, whether to dose higher or lower…. But yeah, I think that’s the biggest benefit I’ve had from that so far when there isn’t too many options that are, “Use as directed”…that has been the most useful aspect of the results or testing. [MH Provider, #44]

Some providers also discussed experiences where the test results did not match the patient’s past medication experiences, and how those conversations were handled:

I would say about maybe 75% [results matched patient experience], but there were definitely a couple of cases that it was not. I can think of at least two or three that they swore that they could not take certain medicines and they made them really sick, and they were in the use as directed column, or that they felt it was really, really helpful and they tolerated it well, and they’re doing excellent on it, but it was in that moderate gene where it says that they may be more likely to have side effects. Again, I just had that conversation with them that this isn’t an exact science yet, but it just helps to guide the providers, and they were usually fine with it. [MH Provider, #12]

Most providers felt comfortable using the test and reported that their confidence grew with use of the test. However, there were some who felt less confident and expressed a desire for more training and education on the test, especially regarding additional details beyond the first page. For example:

I remember asking the study people to just make sure I was understanding the study correctly with one of the patients, because interestingly he was on a medication that he found helpful, and the study said just to be cautious because there are too many-- or not too many, but there were enough interactions that it’s just “use with caution” type thing, and I just wanted to make sure, and then I think I read the bottom part, the little I guess subscript of notes, and it kind of made sense to me. I’ll say when it got to the nitty-gritty of the pharmacogenetic stuff it sort of made sense, but it was a little bit more above my head than I would’ve liked. [MH Provider, #48]

There were a few providers who felt comfortable with these more detailed results and found them to be very valuable. One provider, referring to the longer report, said:

I basically got all the information from… all those annotations in the pages… the information that I had in front of me was I think I feel it was sufficient to make a decision and learn. [MH Provider, #28]

Discussion

This manuscript presents the results of qualitative interviews with primary care and mental health providers participating in the PRIME Care trial of pharmacogenetic testing in Veterans receiving treatment for depression. There was significant variation in how providers described their use of the test in their clinical decision-making, including some uncertainty about the extent to which the results should be relied upon for medication decisions. These findings are consistent with previous studies of providers’ PGx test perceptions and experiences.^13–15,19 Our findings move beyond the extant literature by providing detailed descriptions of providers’ experiences using PGx when prescribing medications for depression. Similar to Frigon et al.,¹⁹ most providers in our study had limited awareness of PGx testing prior to participating in PRIME Care, and some expressed concerns over a need for clearer guidelines for PGx utilization. After using PGx testing during PRIME Care, many providers in our sample described instances where they felt it made a significant difference in a patient’s treatment. Others remained somewhat skeptical, unsure if the testing benefitted the majority of their patients.

These results have several important implications for future clinical implementation and research. First, most providers in PRIME Care felt that PGx testing was best suited to patients with a past history of unsuccessful medication use. This approach could be warranted, as those patients who have already tried several medications without success may have increased likelihood of genetic involvement in the metabolism of antidepressant medications, and therefore may be more likely to have actionable PGx results. However, treatment-resistant depression may also result from complicating factors (i.e., substance use, medical or psychiatric comorbidities) that could limit the utility of the test results.²⁰

Alternatively, the full impact of PGx test results on treatment outcomes may be more effectively demonstrated on first treatment episode patients.²⁰ PGx testing could lead to earlier identification of patients who are likely to experience adverse medication reactions to antidepressants. Finding an effective and well-tolerated medication sooner, could avoid subjecting patients to potential side effects or ineffective medications, thereby reducing the likelihood that patients will drop out of treatment.² Waiting until a patient has already tried a few medications before ordering a PGx test may limit the impact of the test results,²⁰ and result in greater missed opportunities for improving patients’ depression.

Second, while providers’ experiences of using the PGx test varied, many discussed having one or two patients for whom the testing had marked effect in their treatment decisions and outcomes. These positive experiences may make providers more likely to use the PGx test; however, with only approximately 20% of patients having actionable test results, providers may need to use PGx multiple times before experiencing that benefit. Our findings highlight providers’ concerns around patient-level criteria (all patients, or only those with certain characteristics), and timing of PGx testing (at the initial prescription, or after trying a few medications) to yield the greatest benefit. More research is needed to understand these issues and identify criteria that maximize the cost-effectiveness of PGx treatment relative to treatment as usual.

Third, providers in our study discussed using the test results in varying ways, using more or less of the information provided. Some mentioned only using the cover page with the three columns, while others felt more comfortable using the footnotes and more detailed information to make dosing and other adjustments. This variation indicates that some providers may not have made use of all information available to inform their decisions, while others may have over-interpreted the detail in the footnotes. Ongoing provider education is needed to ensure the test is being used appropriately and to the greatest benefit.

Finally, providers’ discussions of how they used the test in their clinical decision making has implications for future research on PGx testing. In particular, some providers discussed using the dosing adjustments guided by the moderate and significant gene-drug interaction (i.e., the yellow and red) columns, as well as more specific genetic information, while others reported relying primarily on the no-gene-drug interaction column (i.e., the green). These different approaches to incorporating the test results have implications for future clinical trials, in terms of assessing and understanding the use of the test, and what constitutes “adherence”. Second, providers in our study discussed the potential “placebo” effect for patients, possibly increasing patient confidence in their medications, while others expressed concerns in the opposite direction, worrying that if a patient sees their medication is yellow or red, they may be negatively disposed towards it, even if there are valid clinical reasons for its use. Empirically exploring these impressions, the extent to which a placebo effect plays a role in the effectiveness of PGx-guided treatment, and exploring patient perceptions, may be an important next step for research.

Limitations:

This work is subject to limitations. Because the study was conducted in the VA, unique patient populations and site-specific structural factors may contribute to providers’ experiences and perceptions and may not be applicable to PGx testing in mental health care in community settings. It is also not possible to determine how site-specific variation within the VA may have influenced providers’ perceptions as the study methods did not allow for that level of analysis. Additionally, interview participants were a convenience sample of providers who volunteered to participate in the PRIME Care trial; as such they may have different views about PGx testing from the general population of providers. However, the wide range of PGx uptake and perceptions in our sample is reassuring in this regard, and likely to be comparable to the range in a broader population. Further, provider experiences during a research study may be different than in regular practice, (i.e., imposed limitations on patients’ eligibility, and presence of supportive research coordinators and local site investigators encouraging PGx use). However, since this was a pragmatic trial, providers had significant autonomy in how they used the test in their clinical decision-making. Finally, we were not able within the scope of this study to also explore the patient experience of the PGx testing. Understanding the patient experience and how it contributes to shared decision-making around depression medication usage is critical to further understanding the use of PGx,²¹ and should be examined in future studies.

Conclusions

In conclusion, providers in our study generally expressed positive experiences with using PGx testing to inform prescribing of medications for depression. However, their experiences highlighted variations in perceptions of which patients are best suited to this type of testing and in the application of the testing results to prescribing decisions. Future programs to facilitate broader implementation and adoption of PGx should consider establishing policy and procedure around the use of the test, as well as mechanisms to support ongoing provider education.