Abstract
Aim:
To elicit preferences for pharmacogenomic (PGx) testing in polypharmacy patients.
Materials & methods:
A face-to-face discrete choice experiment survey was designed and administered to adult polypharmacy patients recruited at a local retail pharmacy in Albuquerque (NM, USA).
Results:
A total of 128 eligible polypharmacy patients completed the discrete choice experiment survey and significantly preferred a PGx test with lower cost, better confidentiality and higher certainty of identifying best medication/dose and side effects and one that can be used to advocate for their treatment needs (all p < 0.01).
Conclusion:
This is the first study eliciting preferences for PGx testing among polypharmacy patients. The study found most polypharmacy patients were willing to take a PGx test and their preferences were mostly influenced by test cost.
Keywords: DCE, discrete choice experiment, patient preferences, PGx, pharmacogenomic test, polypharmacy
Plain language summary:
Patients who concurrently take five or more medications are at a higher risk of experiencing side effects related to drug–drug/drug–gene interactions. ‘One size doesn’t fit all’ – individuals may respond differently to the same dose of a medication. Pharmacogenomic (PGx) testing identifies individual genetic information that may help explain better or worse outcomes or potential problems with drug therapies and eventually may help optimize patient treatment. The authors conducted a face-to-face survey to assess preferences for PGx testing in polypharmacy patients and found that most polypharmacy patients were willing to take a PGx test and their preferences were mostly influenced by test cost and performance, as well as the confidentiality of test results.
Polypharmacy is generally defined as the concomitant use of multiple prescription medications. A WHO report on medication safety in polypharmacy indicates that “although there is no standard definition, polypharmacy is often defined as the routine use of five or more medications. This includes over-the-counter, prescription and/or traditional and complementary medicines used by a patient” [1].
It is estimated that 10% of the US population and 30% of the elderly US population are exposed to polypharmacy (five or more medications simultaneously) [2–4]. As the average age of the US population increases, polypharmacy continues to be a growing public health concern [5]. Overprescribing occurs in polypharmacy and there is evidence of overprescribing in older patients with complex medical conditions, though balancing the benefits and risks of medication therapies can be challenging [6]. Also, adequate nutritional intake can be difficult to achieve for many elderly individuals and the age-related changes in the elderly may impact not only nutrient metabolism and kinetics but also drug responses [7]. Polypharmacy in the elderly is a particular concern because of the potential altered pharmacokinetic and pharmacodynamic response to medications as well as the presence of chronic diseases in this population [8]. Coexisting conditions that are common in the elderly population may cause impairment in renal and cardiac function, which may also affect drug response. Coexisting conditions may also lead to multidrug therapy, exposing patients to a greater risk of adverse drug events (ADEs) [9]. As age increases, organ function and systems decline, which also influences the metabolism of particular drugs that may lead to unexpected drug reactions [10]. As the number of medications taken by polypharmacy patients increases, the risk of ADEs from potential drug–drug/drug–gene interactions also increases [11]. Drug–drug interactions are usually monitored in clinical practice, while drug–gene/drug–drug–gene interactions are not routinely evaluated, despite the fact that they account for 15–60% of all-type interactions [11].
One method of reducing the potential risk of overprescribing and unnecessary healthcare resource utilization related to ADEs is to identify patient genotypes and drug metabolizer status through pharmacogenomic (PGx) testing. PGx evaluates how individuals’ genetic makeup may affect their response to medications, which allows for the prevention of ADEs and for personalized medication treatment [12,13]. In a study by Brixner et al., a prospective cohort of CYP450-tested patients was compared with its propensity-score-matched cohort of CYP-untested patients on healthcare resource utilization [11]. The authors reported a significant decrease in emergency room visits and hospitalizations, as well as high satisfaction rates, among providers with PGx testing availability.
Previous studies have reported that PGx can improve the identification of additional drug–drug interactions and reduce potential ADEs [14,15]. ADEs previously considered nonpreventable may now be predicted and intervened with actionable PGx relationship information. ADEs may be reduced or prevented by choosing an alternative treatment or adjusting drug dosing in patients with genetic variation in drug-metabolizing enzymes. A systematic literature review by Phillips et al. quantitatively examined the potential role of PGx in decreasing ADEs [16]. The literature review results suggest an association between genetic variability in enzymes involved in drug metabolism and the incidence of ADEs, indicating the potential prevention of ADEs by having individuals’ PGx information. Patients’ medication adherence may also be improved by introducing PGx testing prior to medication prescribing by engaging patients in the selection and dosing of medications while improving patients’ confidence and trust in the safety and efficacy of prescription medications [17].
The integration of PGx testing into routine medical care will be based on multiple stakeholders’ (patient, practitioner, payer) acceptance of this technology. Patient perspectives and preferences for, and acceptance of, PGx testing are particularly important to explore when striving to improve and achieve an optimal medication treatment strategy. Preferences for various aspects of PGx testing may be directly related to patients’ expectations, concerns and satisfaction. By eliciting patient preferences toward PGx testing, researchers and providers can better understand patients’ priorities and develop efficient methods to address patient acceptance of PGx testing to improve their medication therapy.
Previously reported studies have elicited preferences for PGx testing among the general public, patients with a specific disease condition such as depression or cancer and patients on a specific medication [18–20]. However, no published study has reported or evaluated preferences for PGx testing among patients with polypharmacy, a patient group at a high risk of ADEs. The purpose of this study is to investigate polypharmacy patients’ preferences for PGx testing.
Materials & methods
The study used a cross-sectional observational study design and a discrete choice experiment (DCE) to assess preferences for PGx testing among patients with polypharmacy. DCEs have the advantages of 1) simultaneously measuring trade-offs between attributes and allowing investigators complete control of attributes and choice sets and 2) providing choice alternatives and information that are otherwise impossible to reveal in real life [21]. In the present study, PGx testing was defined as preemptive testing used to categorize individuals’ genetic profiles to predict potential drug–drug/drug–gene interactions, reduce ADEs and optimize current and future medication therapies. The authors followed the framework from the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) good practice in “constructing experimental designs for discrete-choice experiments” to ensure DCE design quality [22]. The development and reporting of the DCE survey also conformed to the “Conjoint Analysis Applications in Health – A Checklist: A Report of the ISPOR Good Research Practices for Conjoint Analysis Task Force” [23].
Study population
The study’s target population was the group of patients exposed to polypharmacy for chronic conditions who resided in the metropolitan areas in NM, USA. The authors included a convenience sample of patients with self-reported polypharmacy (≥5 maintenance medications, which could include prescription, over-the-counter and/or complementary and alternative medicine therapies) and patronizing a local retail chain pharmacy in Albuquerque, NM, USA. Patients were approached in the pharmacy to determine their willingness to participate in the study. Snowball sampling was also used to maximize patient recruitment; approached patients were encouraged to forward the study information to anyone who was potentially eligible.
Patients were considered eligible for the study if they met all of the following criteria: were ≥18 years of age, could read and understand English and were taking ≥5 maintenance medications concurrently (including prescription, over-the-counter medications and/or complementary and alternative medicine therapies for chronic conditions). Patients were excluded from the study if they presented evidence of potential cognitive impairment as determined with a six-item screener questionnaire score of <4 [24]. The patient recruitment and DCE survey administration were conducted from December 2020 through April 2021, a period dominated by the COVID-19 pandemic. Study investigators fully complied with the social distancing requirements and public health orders issued by the state of NM during the patient recruitment and interviews [25].
DCE design
Identification & selection of DCE attributes & attribute levels
The attributes and attribute levels of PGx testing in the DCE survey were identified based on the results from a systematic literature review, a focus group discussion, an expert panel review and a pilot survey. A systematic literature review was conducted in PubMed, Web of Science (all databases excluding MEDLINE) and EBSCO (CINAHL Complete, APA PsycInfo and Academic Search Complete) as of March 2020 to identify and evaluate peer-reviewed articles on original qualitative or quantitative studies reporting patient attitudes or preferences toward PGx testing. Articles were reviewed for attributes (key factors) affecting patient attitudes or preferences toward PGx testing. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were followed and the process of identifying eligible articles and selecting included studies is presented in Supplementary Figure 1 [26]. The search terms utilized included ‘(pharmacogenomic or pharmacogenetic) and (preference or attitude or acceptance or belief) and patient.’ Studies reporting key factors related to attitudes, acceptance or concerns regarding PGx testing were included. Studies were excluded if the study purpose was not relevant, the study was not conducted from a patient perspective, the study was not written in English or it was not an original research article (e.g., commentaries, editorials and conference abstracts). The authors limited the initial search with the filtering options available in the databases – “English,” “peer-reviewed,” “research article” and “journal article” – and then further applied the inclusion/exclusion criteria when reviewing the titles and abstracts of the search results. The initial search generated 691 articles from PubMed, 31 from Web of Science and 139 from EBSCO. Twenty-seven articles remained after applying the inclusion/exclusion criteria. The reference list of each included article was examined, and no additional eligible articles were identified.
Findings from the systematic literature review on key factors affecting patient attitudes or preferences toward PGx testing were categorized to inform attributes in the DCE using the unified theory of acceptance and use of technology (UTAUT) as a framework [27,28]. The UTAUT is a validated theoretical model that was developed based on a synthesis of eight theoretical models and designed to explain behaviors related to the acceptance and use of new technologies. The UTAUT consists of four core determinants (constructs) and up to four moderators [27]. The use of the UTAUT in the present study ensured capturing crucial attributes to understand patients’ preferences for the PGx testing technology. An online focus group discussion with four polypharmacy patients was conducted to review the initial attributes from the literature and to identify additional attributes. The initial attributes from the literature and attributes identified from the focus group discussion were then reviewed and modified by four experts in PGx or PGx testing services and by the authors. The final list of attributes, attribute levels and descriptions, as well as related UTAUT constructs in the study context, are presented in Table 1. A pilot survey of the DCE survey in ten eligible patients did not result in revisions to the attributes or attributes levels.
Table 1.
Related Unified Theory of Acceptance and Use of Technology construct interpretations and pharmacogenomic testing attributes, attribute definitions and attribute levels used in the discrete choice experiment.
| UTAUT construct - performance expectancy: the degree to which the patient believes that PGx testing will help improve treatment effectiveness and reduce adverse events | ||
|---|---|---|
| PGx attributes | Attribute definitions | Attribute levels |
| Identify best medication/dose | The degree to which the test will help identify the best medication and correct dose for your treatment | 1) Definitely 2) Likely |
| Identify/prevent potential side effects | The degree to which the test will help identify/prevent side effects from your medications or treatment | 1)Definitely 2) Likely |
| Privacy/access | To whom your PGx test results will be available | 1) Healthcare providers (physician, pharmacist) 2) Other third parties (health insurance companies, life insurance companies, data banks etc.) |
| Involvement and advocacy for my treatment | The degree to which you can utilize the PGxtest results in terms of getting involved in and advocating for your medication treatment decisions with your physician | 1) Definitely 2) Likely |
| UTAUT construct - facilitating conditions: the degree to which the patient believes that there are enough support and guidance to use the PGx tests | ||
| PGx attribute | Attribute definition | Attribute levels |
| Communication of the test results | The type of healthcare providers who will explain the results for you | 1) A pharmacist explains the test results 2) A physician explains the test results 3) Either a pharmacist or a physician explains the test results |
| UTAUT construct - effort expectancy: the degree of ease to use the PGx testing | ||
| PGx attribute | Attribute definition | Attribute levels |
| Cost/insurance | Insurance coverage/out-of-pocket cost | 1) US$50, not covered by insurance 2) US$500, not covered by insurance 3) Fullycovered by insurance |
Note that due to the complexity of the discrete choice experiment design, only the most important attributes were included. Attribute 'provider of test (the type of healthcare providers that recommend or schedule a PGx test for you)' related to the UTAUT construct “social influence” was considered initially but was not ranked among the most important attributes by the patient focus group.
PGx: Pharmacogenomics; UTAUT: Unified theory of acceptance and use of technology.
DCE choice question generation
Based on the final list of attributes and attribute levels, a DCE survey was generated using experimental-design macros in SAS 9.4 (NC, USA) [22,29]. The survey was designed to have two alternatives in each choice set, and each alternative consisted of six attributes. Table 2 presents an example of a DCE choice set. Considering the complexity of the DCE study, two alternatives, instead of three or more alternatives, were chosen to avoid excessive cognitive burden to participating patients, to prevent the likelihood of patients choosing a status quo option compared with the two-alternative design and to prevent issues with nonparticipation data when including an ‘opt-out’ option [30]. A previous review of DCE design features found that most DCEs use no more than six attributes [31]. The ISPOR good practice also suggests that the number of levels for each attribute should be limited to three or four [23]. Therefore, the maximum number of levels per attribute in the present DCE study was limited to four. A fractional factorial design was utilized in the DCE. The design included a fraction of all possible choice alternatives and was appropriate for evaluating the main effects of attributes with a feasible sample size. Efficient choice designs have common traits, including orthogonality (attributs levels vary independently), level balance, minimal overlap between attribute levels and the smallest variance matrix [23]. The relative D-efficiency was measured to examine the design efficiency, with values ranging from 0% to 100%. The authors achieved a relative D-efficiency of 85%, which is considered good/acceptable [29]. The complete DCE survey is presented in Supplementry Appendix B.
Table 2.
Example discrete choice experiment choice set.
| Attribute | PGx test A | PGx test B |
|---|---|---|
| Identify best medication/dose | Likely | Definitely |
| Identify/prevent potential side effects | Definitely | Likely |
| Privacy/access | Other third parties (health insurance companies,data banks etc.) | Healthcare providers (physician, pharmacist) |
| Communication of the test results | A physician | A pharmacist |
| Involvement and advocacy for my treatment | Definitely | Likely |
| Cost/insurance (payment type) | $50, not covered | Fully covered by insurance |
| Which PGx test do you prefer? | □ | □ |
PGx: Pharmacogenomics.
Survey administration
Study participants were asked to complete the self-administered, paper DCE survey. The investigator provided participants with detailed explanations of each attribute and attribute level to ensure the attribute definitions were clear to the respondents before they completed the choice sets. In consideration of possible cognitive uncertainty or heuristics-related bias, a repeated choice set was included as a rationality check question to examine the internal consistency of patients’ responses [32]. The investigator was present with each patient during the entire survey to answer any questions the participants had. After the DCE survey, the participants were asked about their general willingness to take a PGx test (yes/no), previous use of PGx testing (yes/no) and patient demographic characteristics. Each patient was compensated with a $20 merchandise card for completing the entire survey. A $5 merchandise card was given to patients who failed the cognitive screening or failed to complete the entire survey as compensation for their time.
Data analysis
Patient demographic and clinical characteristics were summarized with proportions for categorical variables and means and standard deviations for continuous variables. A conditional logit model was used to assess preference weights for attribute levels of PGx testing. McFadden’s pseudo R2 value was used as an indicator of goodness of fit of the model; values of 0.2–0.4 are usually considered to indicate a good model fit [33]. Preference weights (parameter coefficients, indicating the relative value patients put on a certain level compared with the reference level within an attribute) were reported with derived odds ratios (exponentiated coefficients, ratios of the probabilities of choosing alternatives) and p-values. A positive preference weight represents an increase in value or utility for a PGx test with the feature of the particular attribute level compared with the reference level feature; a negative preference weight represents a decrease in utility (negative preference) for a PGx test with the feature in the particular attribute level compared with the feature in the reference level [34]. The utility/disutility associated with levels of an attribute indicates the benefits/satisfaction that polypharmacy patients may get from having a PGx test with the characteristics in the attribute level(s). A p value of ≤ 0.05 was considered statistically significant. Subgroup analyses using conditional logit models were conducted by gender, age, income, race/ethnicity, education, marital status, annual household income, number of prescription medications and number of chronic conditions. Between-group differences in preference weights were investigated by including interaction terms of attributes and the covariate being examined. Preference weights for each attribute level in each subgroup and differences in preference weights for each attribute level with p-values were estimated.
Results
Patient characteristics
A total of 135 patients (including the ten pilot survey patients) agreed to participate in the study. Three patients did not pass the study prescreening questions and four failed the DCE consistency check, leaving 128 qualified responses for the final analyses.
Table 3 presents a summary of the demographic and clinical characteristics of the study sample (n = 128). The mean (standard deviation) age of the study sample was 57.4 (±14.7) years, with 28 patients (21.9%) aged 18–44 years, 55 patients (43%) aged 45–64 years and 44 patients (34.4%) aged 65 years or more. Female patients were 62.5% (n = 80) of the study sample. The majority of the patients were non-Hispanic White (n = 77; 60.2%), with others being Hispanic or Latino (n = 34; 26.6%), African–American (n = 6; 4.7%), Native American (n = 2; 1.6%), Asian (n = 2; 1.6%) and multiethnic (n = 6; 4.7%). Among the 128 patients, 67 patients (52.3%) had an undergraduate or higher-level degree and 72 patients (56.3%) had an annual household income greater than $35,000. Fifty-one patients (39.8%) were married when surveyed, with the others being single (n = 33; 25.8%), widowed (n = 20; 15.6%) or divorced/separated (n = 24; 18.8%). On average, the patients had 3.8 (±1.8) chronic conditions, with the lowest number being one and the highest being ten. The mean number of prescription medications in the study sample was 5.8 (±2.8), with 40 patients (31.3%) having one to four prescription medications, 79 (61.7%) patients having five to ten prescription medications and nine patients (7%) having 11 or more prescription medications. Overall, most patients (n = 113; 88.3%) were willing to have any kind of PGx tests, while the others primarily expressed concerns about PGx test cost and privacy issues.
Table 3.
Demographic and clinical characteristics of the final study sample.
| Characteristic | n (total = 128) | Percentage (%) |
|---|---|---|
| Age (years) | ||
| Mean (SD) | 57.4 | (14.7) |
| 18–44 | 28 | 21.9 |
| 45–64 | 55 | 43.0 |
| ≥65 | 44 | 34.4 |
| Missing | 1 | 0.8 |
| Gender | ||
| Male | 48 | 37.5 |
| Female | 80 | 62.5 |
| Race/ethnicity | ||
| Non-Hispanic White | 77 | 60.2 |
| Hispanic or Latino | 34 | 26.6 |
| Native American | 2 | 1.6 |
| African-American | 6 | 4.7 |
| Asian | 2 | 1.6 |
| Two or more races | 6 | 4.7 |
| Missing | 1 | 0.8 |
| Annual household income | ||
| <US$20,000 | 26 | 20.3 |
| $20,000–835,000 | 30 | 23.4 |
| $35,001-$50,000 | 22 | 17.2 |
| $50,001–875,000 | 23 | 18.0 |
| >$75,000 | 27 | 21.1 |
| Education | ||
| Less than high school | 3 | 2.3 |
| High school graduate | 22 | 17.2 |
| Some college | 36 | 28.1 |
| College graduate | 46 | 35.9 |
| Graduate school | 21 | 16.4 |
| Marital status | ||
| Single | 33 | 25.8 |
| Married | 51 | 39.8 |
| Widowed | 20 | 15.6 |
| Divorced/separated | 24 | 18.8 |
| Health insurance | ||
| Yes | 126 | 98.4 |
| No | 2 | 1.6 |
| Number of chronic diseases | ||
| Mean (SD) | 3.8 | (1.8) |
| 1–3 | 61 | 47.7 |
| ≥4 | 67 | 52.3 |
| Number of prescription meds | ||
| Mean (SD) | 5.8 | (2.8) |
| 1–4 | 40 | 31.3 |
| 5–10 | 79 | 61.7 |
| ≥11 | 9 | 7.0 |
| Cancer | ||
| Yes | 13 | 10.2 |
| No | 115 | 89.8 |
| Overall attitude toward PGx testing | ||
| Yes | 113 | 88.3 |
| No | 14 | 10.9 |
| Missing | 1 | 0.8 |
| Previous PGx experience | ||
| Yes | 2 | 1.6 |
| No | 125 | 97.6 |
| Missing | 1 | 0.8 |
PGx: Pharmacogenomics; SD: Standard deviation.
Patient preferences for PGx testing
Overall population
Response data from 128 patients who completed all 12 of the DCE choice sets (each with two alternatives) were used for the analysis of preferences for PGx testing. A total of 3072 (128 patients × 12 sets × 2 alternatives) observations were used for the conditional logit model. Table 4 presents the regression results of patient preferences for PGx tests using the response data from the entire study sample. The result of the likelihood ratio test for testing the global null hypothesis (all preference weights are zero) suggested a statistically significant relationship between patient choices and attributes (likelihood ratio χ2: 659.64; p < 0.001). The model with predictors (attributes) resulted in a McFadden’s pseudo R2 value of 0.31, suggesting a good model fit. All six attributes had significant preferences, except the ‘communication of the test results’ attribute, with the ‘cost/insurance’ attribute having the largest absolute value for the level ‘$500, not covered’ when compared with the level ‘fully covered by insurance’. Patients would gain an increase of 0.74 in utility if a PGx test could definitely help identify the best medication/dose compared with a PGx test that could only likely help identify the best medication/dose (odds ratio [OR]: 2.09; p < 0.001). Similarly, patients would attain an increase of 0.23 in utility if a PGx test could definitely help identify/prevent potential side effects compared with a PGx test that could only likely help identify/prevent potential side effects (OR: 1.26; p = 0.002).
Table 4.
Patient preferences for pharmacogenomic testing: overall study sample (n = 128).
| Pharmacogenomic test attributes and levels | Preference weight | Standard error | Odds ratio (95% CI) | p-value |
|---|---|---|---|---|
| Identify best medication/dose | ||||
| Likely | Reference | |||
| Definitely | 0.74 | 0.07 | 2.09 (1.83; 2.39) | <0.001 |
| Identify/prevent side effects | ||||
| Likely | Reference | |||
| Definitely | 0.23 | 0.07 | 1.26 (1.09; 1.45) | 0.002 |
| Privacy/access | ||||
| Other third parties | Reference | |||
| Healthcare providers | 0.73 | 0.07 | 2.09 (1.82; 2.39) | <0.001 |
| Communication of the test results | ||||
| Either a pharmacist or a physician | Reference | |||
| A pharmacist | −0.12 | 0.09 | 0.88 (0.74; 1.06) | 0.18 |
| A physician | 0.11 | 0.10 | 1.12 (0.92; 1.36) | 0.25 |
| Involvement and advocacy for my treatment | ||||
| Likely | Reference | |||
| Definitely | 0.24 | 0.07 | 1.27 (1.11; 1.44) | <0.001 |
| Cost/insurance | ||||
| Fully covered by insurance | Reference | |||
| US$50, not covered | −0.43 | 0.10 | 0.65 (0.54; 0.78) | <0.001 |
| $500, not covered | −1.50 | 0.10 | 0.22 (0.19; 0.27) | <0.001 |
| Respondents (n) | 128 | Likelihood ratio χ2 (p-value) | 659.64 (<0.001) | |
| Observations (n) | 3072 | Pseudo R2 | 0.31 | |
In terms of ‘privacy/access’, patients would attain an increase of 0.73 in utility if a PGx test only allowed healthcare providers to access the test results compared with a PGx test that also allowed other third parties to access test results (OR: 2.09; p < 0.001). ‘Definitely’ was preferred over ‘likely’ for the degree to which the patients could utilize PGx test results to advocate for treatment needs and involvement in treatment decisions (OR: 1.27; p < 0.001). Compared with a PGx test that was fully covered by insurance, patients would lose 0.43 utility if a PGx test cost $50 and was not covered by insurance (OR: 0.65; p < 0.001), and the loss of utility would increase to 1.50 if a PGx test cost $500 and was not covered by health insurance (OR: 0.22; p < 0.001). Physicians alone were preferred (not statistically significant) over ‘either a pharmacist or a physician’ for explaining the PGx test results (OR: 1.12; p = 0.25), while pharmacists were less preferred (not statistically significant) when compared with ‘either a pharmacist or a physician’ (OR: 0.88; p = 0.18).
Subgroup comparisons
In subgroup analyses, it was observed that compared with the 18–64 age group, the ≥65 age group would obtain less utility if the PGx test results could only be accessed by healthcare providers than if the PGx test also allowed other third parties to access test results and would lose a significantly larger unit of utility for a PGx test that cost $500 and was not covered by insurance than for a PGx test that was fully covered by insurance. Compared with the higher annual household income (>$35,000) group, the lower annual household income (≤$35,000) group would have a significantly higher loss of utility if a PGx test cost $50 or $500 and was not covered by insurance than if a PGx test were fully covered by insurance. Compared with single/divorced/widowed/separated polypharmacy patients, married polypharmacy patients would have a significantly higher utility for a PGx that only allowed healthcare providers to access the test results than for a PGx test that also allowed other third parties to access test results. Tables summarizing subgroup comparisons are presented in Supplementary Appendix A (Tables 1–24).
Discussion
This study elicited polypharmacy patients’ preferences for PGx testing with a DCE survey and found that all attribute levels were significantly associated with patients’ preferences for a PGx test, except the ‘communication of the test results’ attribute levels. The ‘cost/insurance’ attribute contained the largest preference weight. This finding is consistent with previous qualitative and non-DCE quantitative studies on patient attitudes toward and perceptions of PGx testing, which reported that test cost is an influential patient uptake factor of general PGx testing [35–44]. A recent systematic literature review on DCE and conjoint analysis studies on genetic/genomic tests also found that the cost attribute was the most important and most studied attribute, appearing in 76.9% of the 292 studies they reviewed [45].
The attribute ‘involvement and advocacy for my treatment’ was novel and was identified in this study’s patient focus group discussion. It was defined in this study as “the degree to which the patient can utilize the PGx test results in terms of getting involved in and advocating for your medication treatment decisions with your physician”. Little has been reported on or described for this attribute in PGx or genetic/genomic testing literature. This finding is important, as patient-centered care and patient involvement in their own care are the foundations of healthcare services. A review by the UK’s NICE on improving patient experiences reiterated that information is a prerequisite for involvement in decision-making and patient care and that healthcare professionals should provide an environment in which patients feel able to share and participate in decisions [46]. Today, patients are likely to be involved in treatment decision-making, as they are more willing to express their thoughts on and preferences for treatment [47–49]. The finding related to the ‘involvement and advocacy for my treatment’ attribute further highlights the importance of patient-centered care and suggests that polypharmacy patients may favor autonomy in their medication treatment decisions and a preference to utilize their PGx test results to advocate for their medication treatment needs.
The ‘identify best medication/dose’ attribute was associated with the second largest preference weight in this study, and this finding is consistent with previous qualitative studies that found improving medication prescribing, helping identify effective drugs and helping treatment efficiency were the key factors in patients’ positive attitudes toward PGx testing [36,37,50–53]. The ‘privacy/access’ attribute represented the third largest preference weight. This was consistent with the study’s focus group discussion results, in which ‘privacy/access’ was one of the highest-ranked attributes. The ‘privacy/access’ attribute was not evaluated in previously reported DCEs on patient preferences for PGx testing but was among the key factors identified in qualitative studies and non-DCE quantitative studies. Previous studies expressed patient concerns about the privacy/confidentiality of test results, misuse of test results and potential discrimination from employers or insurance companies [36,37,41,42,44,50–58].
The ‘communication of the test results’ attribute was the only attribute without a statistically significant difference in preferences for a PGx test between attribute levels. In Payne et al.’s DCE study on patient preferences for PGx testing to identify the risk of azathioprine-related neutropenia, the attribute ‘Explanation (who explains the result to the patient)’ was examined, with its attribute levels being ‘general practitioner,’ ‘pharmacist,’ ‘hospital doctor’ and ‘nurse’ [59]. When compared with ‘general practitioner,’ ‘pharmacist’ was associated with a significant decrease in preference weight (β = −0.385; p = 0.039), and ‘hospital doctor’ was associated with a significant increase in preference weight (β = −0.264: p = 0.040). The authors also observed a nonsignificant decrease in preference weight when comparing ‘a pharmacist’ with ‘either a pharmacist or a physician’ for ‘the communication of the test results’ attribute, as well as an increase in preference weight if ‘a physician’ explains the PGx test results. Previous DCE studies on patient preferences for PGx testing did not examine this attribute; however, qualitative studies and surveys on patient perceptions of PGx testing have reported the importance of explanation of the test results [39,43].
The noted differences between these findings and previous studies’ findings could be due to multiple reasons. First, the type of PGx tests evaluated in this study was different from the types in previously reported DCE studies. The majority of previous DCE studies focused on PGx tests for identifying a single gene, a disease or a single medication, while the present DCE study focused on generic, pre-emptive PGx testing. Additionally, the attributes and attribute levels in this study varied compared with previous DCE studies in the type of PGx testing evaluated. For example, Herbild et al.’s study evaluated preferences for a CYP2D6 screening test for guiding the use of antidepressants; therefore, the attributes in their study were directly related to the effectiveness, side effects and cost of antidepressants [18]. The present study focused on generic PGx testing and included attributes covering broader aspects of a PGx test. Second, this study used different designs in terms of the numbers of alternatives, attributes and attribute levels compared with previously reported DCE studies. The present study examined six attributes with two or three attribute levels, while previous studies examined varied numbers of attributes (four to eight), with the number of levels ranging from two to eight. The present study included only two alternatives in each choice set without an opt-out option, while two previous DCE studies allowed an opt-out option in each choice set [19,60]. Third, the characteristics of the study populations differed between previous DCE studies and the present study. This study included polypharmacy patients, while previous DCE studies included either patients with a specific disease or patients on a single treatment.
The present study’s subgroup analysis findings on the ‘privacy/access’ attribute suggest a concern, especially in younger and middle-aged patients, about potential confidentiality issues with PGx testing. Further investigation is needed into the factors driving younger patients’ concerns about privacy issues and whether they are associated with sociodemographic attributes (e.g., education level, income level, political affiliation). Future studies may also examine how PGx test results are being utilized by third parties, such as insurance companies or laboratories conducting PGx testing. This study’s findings suggest there may be a need for policies regarding the security and privacy of PGx test results and third parties’ access to patient genetic information.
Significant differences were noted between income groups on the ‘cost/insurance’ attribute levels. A recent study on the utilization of PGx testing in a US managed-care population found that the percentage of insured patients receiving a PGx test is still only 0.12% and, among those with a test, slightly more than 50% was covered by Medicare/Medicaid [61]. Prior reviews have noted that when economic evaluations have been conducted, a majority have been found to be cost effective and a minority cost saving [62,63]. While PGx testing has gained attention in the last decade and this study’s findings indicate patients’ overall willingness to have PGx tests, support for reimbursement for PGx testing is challenged by the lack of evidence of its clinical and economic utility [64]. It has been found that payers recognize value in PGx information, but barriers persist – for example, payers have expressed a desire for more determinations in guidelines about the value of testing for clinical outcomes [65].
To the authors’ knowledge, this is the first reported study eliciting polypharmacy patients’ preferences toward PGx testing in a retail pharmacy setting. Although six previously published DCE studies elicited patient preferences for PGx testing, those studies primarily focused on either general outpatient patients or patients with a specific disease or medication [18,19,59,60,66,67]. The present study is unique in its focus on a patient group potentially at higher risk of experiencing insufficient treatment responses and drug–drug/drug–gene interactions due to the additive effects of the concurrent use of five or more medications. Polypharmacy patients would conceivably receive a large benefit from having PGx testing that examined patient genotypes and phenotypes to optimize medication therapy.
The study findings on patient preferences for PGx testing may inform future researchers and decision-makers about the key aspects (attributes) of a PGx test that may affect patient preferences or choice for a PGx test and may help facilitate the development and implementation of PGx testing services. This study is also the first DCE study in the PGx testing literature utilizing a theoretical/behavioral model (UTAUT). Previous studies identified PGx test attributes primarily from focus group discussions and expert panel reviews only. The benefit of incorporating the UTAUT is that it provides a conceptual framework for understanding patient behavior related to the adoption of new technologies and guides the research to consider and include key PGx testing attributes from a patient perspective. The six attributes tested in this DCE were well aligned with the four core constructs (performance expectancy, effort expectancy, social influence and facilitating conditions) of the UTAUT.
This study followed the ISPOR Good Research Practice recommendations for DCE studies, but the study results should be considered in light of several limitations. The results may be subject to unmeasured attributes that might affect a patient’s preference for a PGx test. However, the impact is likely limited because of the incorporation of an extensive attribute identification process, including a patient focus group discussion, an expert panel review, a research team review and pilot testing. The study results may also be subject to patient understanding of survey questions. However, the impact of patient misunderstanding was minimized, as most patients expressed that the survey was easy to understand and the investigator was available with/to the patient to provide detailed definitions, explanations and clarifications of key concepts during the survey. The sample size needed in this study was based on Orme’s rule of thumb and took into account the need for stratified analyses. Although a sample of over 100 respondents is usually considered suitable for modeling preferences, the sample size of this study may be considered insufficient by other criteria or sample size calculation methods [18,63,64].
This study utilized a convenience sample of patients with polypharmacy (concurrent use of ≥5 medications, including prescriptions, over-the-counter medications and/or complementary and alternative medicine therapies), and all the study participants were recruited at one local retail pharmacy store in one large metropolitan city in NM. This may limit generalizability to other patient populations. The study sample, though, has the strength of comprising patients utilizing community-based services and the findings may inform future studies focused on improving PGx testing-related services in that setting. Readers need to be cautious when generalizing these findings to other polypharmacy patients, as the definition of polypharmacy used in this study’s inclusion/exclusion criteria may be different from some other common polypharmacy definitions, such as ‘concurrent use of ≥5 prescription medications’.
Finally, this study used self-reported patient information on demographic and clinical characteristics, which may have led to information bias and potential misclassification, as patients may not have correctly recalled information. Most of the included patients were aware of the $20 study participation incentive, which may have led to misreporting of patient polypharmacy status. However, the misreported information on polypharmacy status should have been minimal because the authors used a questionnaire on specific chronic conditions and the number of medications in the prescreening phase to double-check patient eligibility.
Conclusion
Our study found that polypharmacy patients’ preferences for PGx testing were mostly influenced by the test’s cost, performance and confidentiality of results. Our study findings may inform PGx test developers to facilitate the design and implementation of PGx testing. Our findings may also help policymakers better understand patient uptake of PGx testing and help with the implementation of PGx testing-related services as well as patient education and communication on key aspects of PGx testing.
Our study suggested significant differences in preferences for PGx test attributes ‘cost/insurance’ and ‘privacy/access’ between different age groups and income groups. This has potential policy implications, particularly for access to PGx tests for low-income and elderly patients and for ensuring the confidentiality of patient test results. Our study also added additional value by identifying a new attribute, ‘involvement and advocacy for my treatment,’ which incorporated patient centricity and advocacy, and may help attract more attention to patient involvement and empowerment in healthcare decision-making related to the adoption of new technologies. Future studies need to examine further the underlying reasons for the differences in preferences for certain attribute levels by patient characteristics.
Healthcare providers and policymakers may consider incorporating PGx testing into the standard care for polypharmacy patients, especially patients experiencing unusual drug-related problems. As technologies advance, we may observe a decrease in the cost of having a PGx test, enabling PGx testing to become available to more patients at an earlier treatment phase. However, the potential overuse of PGx testing also needs to be considered together with its uptake to prevent unnecessary healthcare resource utilization. It is also important to conduct cost–effectiveness analyses of PGx testing in polypharmacy or other patient populations at different treatment phases to inform payers on the value of adopting and reimbursing PGx tests.
Supplementary Material
Summary points.
Polypharmacy patients in the study had an overall high interest in taking a pharmacogenomic (PGx) test.
Patients significantly preferred a PGx test with lower cost, better confidentiality and higher certainty of identifying the best medication/dose and side effects and a PGx test that can be used to advocate for their treatment needs (all p < 0.01).
The study suggested significant differences in preferences for PGx test attributes ‘cost/insurance’ and ‘privacy/access’between different age groups and income groups.
This provides policy implications for reimbursing PGx tests for low-income and elderly patients and ensuring the confidentiality of patient test results.
These findings also help policymakers better understand patient uptake of PGx testing and help with the implementation of PGx testing-related services as well as patient education and communication about PGx testing and its key aspects.
Healthcare providers and policymakers may consider incorporating PGx testing into the standard care for polypharmacy patients, especially patients experiencing unusual drug-related problems.
Financial & competing interests disclosure
This study was funded by the New Mexico Research Grant ($3,000, #14369) from the Graduate & Professional Student Association at The University of New Mexico. Partial funding for manuscript completion was provided by grants from the National Human Genome Research Institute (R01HG011792). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Footnotes
Ethical conduct of research
This study was reviewed and approved by the Institutional Review Board (IRB) at The University of New Mexico (study ID: 20–246). Informed consent has been obtained from the participants involved.
Supplementary data
To view the supplementary data that accompany this paper please visit the journal website at: www.futuremedicine.com/doi/suppl/10.2217/pme-2022-0056
Data sharing statement
The manuscript reports the original results of a face-to-face discrete choice experiment survey in patients. Deidentified data from survey responses are available upon request.
References
Papers of special note have been highlighted as: • of interest
- 1.WHO, Geneva, Switzerland. Medication Safety in Polypharmacy: Technical Report (2019). [Google Scholar]
- 2.Gu Q, Dillon CF, Burt VL. Prescription drug use continues to increase: U.S. prescription drug data for 2007–2008. NCHS Data Brief (42), 1–8 (2010). [PubMed] [Google Scholar]
- 3.CDC Atlanta, GA USA. Multiple Chronic Conditions among US Adults: a 2012 Update. (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Quinn KJ, Shah NH. A dataset quantifying polypharmacy in the United States. Sci. Data 4, 170167 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bushardt RL, Massey EB, Simpson TW, Ariail JC, Simpson KN. Polypharmacy: misleading, but manageable. Clin. Interv. Aging 3(2), 383–389 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Steinman MA, Hanlon JT. Managing medications in clinically complex elders: “There’s got to be a happy medium.” JAMA 304(14), 1592–1601 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jyrkka J, Mursu J, Enlund H, L¨ onnroos E. Polypharmacy and nutritional status in elderly people.¨ Curr. Opin. Clin. Nutr. Metab. Care 15(1), 1–6 (2012). [DOI] [PubMed] [Google Scholar]
- 8.Fulton MM, Allen ER. Polypharmacy in the elderly: a literature review. J. Am. Acad. Nurse Pract. 17(4), 123–132 (2005). [DOI] [PubMed] [Google Scholar]
- 9.Doos L, Roberts EO, Corp N, Kadam UT. Multi-drug therapy in chronic condition multimorbidity: a systematic review. Fam. Pract. 31(6), 654–663 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hughes SG. Prescribing for the elderly patient: why do we need to exercise caution? Br. J. Clin. Pharmacol. 46(6), 531–533 (1998). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Brixner D, Biltaji E, Bress A et al. The effect of pharmacogenetic profiling with a clinical decision support tool on healthcare resource utilization and estimated costs in the elderly exposed to polypharmacy. J. Med. Econ. 19(3), 213–228 (2016). [DOI] [PubMed] [Google Scholar]
- 12.Aneesh TP, Sekhar SM, Jose A, Chandran L, Zachariah SM. Pharmacogenomics: the right drug to the right person. J. Clin. Med. Res. 1(4), 191–194 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Pharmacogenomics FAQ. National Human Genome Research Institute (2020). www.genome.gov/FAQ/Pharmacogenomics
- 14.Cabrera M, Finkelstein J. A use case to support precision medicine for frequently hospitalized older adults with polypharmacy. AMIA Jt Summits Transl. Sci. Proc. 2016, 16–21 (2016). [PMC free article] [PubMed] [Google Scholar]
- 15.Finkelstein J, Friedman C, Hripcsak G, Cabrera M. Potential utility of precision medicine for older adults with polypharmacy: a case series study. Pharmgenomics Pers. Med. 9, 31–45 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Phillips KA, Veenstra DL, Oren E, Lee JK, Sadee W. Potential role of pharmacogenomics in reducing adverse drug reactions: a systematic review. JAMA 286(18), 2270–2279 (2001). [DOI] [PubMed] [Google Scholar]
- 17.Haga SB, LaPointe NMA. The potential impact of pharmacogenetic testing on medication adherence. Pharmacogenomics J. 13(6), 481–483 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Herbild L, Gyrd-Hansen D, Bech M. Patient preferences for pharmacogenetic screening in depression. Int. J. Technol. Assess. Health Care 24(1), 96–103 (2008). [DOI] [PubMed] [Google Scholar]
- 19.Najafzadeh M, Johnston KM, Peacock SJ et al. Genomic testing to determine drug response: measuring preferences of the public and patients using discrete choice experiment (DCE). BMC Health Serv. Res. 13, 454 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Payne K, Fargher EA, Roberts SA et al. Valuing pharmacogenetic testing services: a comparison of patients’ and health care professionals’ preferences. Value Health 14(1), 121–134 (2011). [DOI] [PubMed] [Google Scholar]
- 21.Kjaer T A review of the discrete choice experiment – with emphasis on its application in health care. Proceedings of the University of Southern Denmark. 1 (2005). [Google Scholar]
- 22. Reed Johnson F, Lancsar E, Marshall D et al. Constructing experimental designs for discrete-choice experiments: report of the ISPOR Conjoint Analysis Good Research Practices Task Force. Value Health 16(1), 3–13 (2013). • Provides an overview of the methods of constructing experimental designs for discrete choice experiments.
- 23. Bridges JFP, Hauber AB, Marshall D et al. Conjoint analysis applications in health – a checklist: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force. Value Health 14(4), 403–413 (2011). • Outlines the good research practices for conjoin analysis.
- 24.Callahan C, Unverzagt F, Hui S, Perkins A, Hendrie H. Six-item screener to identify cognitive impairment among potential subjects for clinical research. Med. Care 40(9), 771–781 (2002). [DOI] [PubMed] [Google Scholar]
- 25.New Mexico Department of Health. Public Health Orders and Executive Orders. NM, USA: (2021). [Google Scholar]
- 26.Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann. Intern. Med. 151(4), 264–269 (2009). [DOI] [PubMed] [Google Scholar]
- 27.Venkatesh V, Morris MG, Davis GB, Davis FD. User acceptance of information technology: toward a unified view. MIS Quarterly 27(3), 425–478 (2003). [Google Scholar]
- 28.Venkatesh V, Thong J. Unified theory of acceptance and use of technology: a synthesis and the road ahead. J. Assoc. Inf. Syst. 17(5), 328–376 (2016). [Google Scholar]
- 29.Kuhfeld WF. SAS Institute Inc., Cary, NC. Marketing research methods in SAS. (2010). [Google Scholar]
- 30.Ryan M, Gerard K, Amaya-Amaya M. Using Discrete Choice Experiments to Value Health and Health Care. Springer Science & Business Media, Antilles, The Netherlands: (2008). [Google Scholar]
- 31.Marshall D, Bridges JF, Hauber B et al. Conjoint analysis applications in health – how are studies being designed and reported? Patient 3(4), 249–256 (2010). [DOI] [PubMed] [Google Scholar]
- 32.WHO, Geneva, Switzerland. How to Conduct a Discrete Choice Experiment for Health Workforce Recruitment and Retention in Remote and Rural Areas: A User Guide with Case Studies. (2012). [Google Scholar]
- 33.Hauber AB, Gonzalez JM, Groothuis-Oudshoorn CGM´ et al. Statistical methods for the analysis of discrete choice experiments: a report of the ISPOR Conjoint Analysis Good Research Practices Task Force. Value Health 19(4), 300–315 (2016). [DOI] [PubMed] [Google Scholar]
- 34.Adams J, Bateman B, Becker F et al. Discrete choice experiment. NIHR Journals Library (2015). www.ncbi.nlm.nih.gov/books/NBK326957/
- 35.De Marco M Views on personalized medicine: do the attitudes of African American and White prescription drug consumers differ? Public Health Genomics 13(5), 276–283 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Issa AM, Tufail W, Hutchinson J, Tenorio J, Baliga MP. Assessing patient readiness for the clinical adoption of personalized medicine. PHG 12(3), 163–169 (2009). [DOI] [PubMed] [Google Scholar]
- 37.Lee YM, McKillip RP, Borden BA, Klammer CE, Ratain MJ, O’Donnell PH. Assessment of patient perceptions of genomic testing to inform pharmacogenomic implementation. Pharmacogenet. Genomics 27(5), 179–189 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Issa AM, Hutchinson JF, Tufail W, Fletcher E, Ajike R, Tenorio J. Provision of personalized genomic diagnostic technologies for breast and colorectal cancer: an analysis of patient needs, expectations and priorities. Per. Med. 8(4), 401–411 (2011). [DOI] [PubMed] [Google Scholar]
- 39.Fargher EA, Eddy C, Newman W et al. Patients’ and healthcare professionals’ views on pharmacogenetic testing and its future delivery in the NHS. Pharmacogenomics 8(11), 1511–1519 (2007). [DOI] [PubMed] [Google Scholar]
- 40.Lee YM, Manzoor BS, Cavallari LH, Nutescu EA. Facilitators and barriers to the adoption of pharmacogenetic testing in an inner-city population. Pharmacotherapy 38(2), 205–216 (2018). [DOI] [PubMed] [Google Scholar]
- 41.Frigon M-P, Blackburn M-E, Dubois-Bouchard C, Gagnon A-L, Tardif S, Tremblay K. Pharmacogenetic testing in primary carè practice: opinions of physicians, pharmacists and patients. Pharmacogenomics 20(8), 589–598 (2019). [DOI] [PubMed] [Google Scholar]
- 42.Lee I-H, Kang H-Y, Suh HS, Lee S, Oh ES, Jeong H. Awareness and attitude of the public toward personalized medicine in Korea. PLOS ONE 13(2), e0192856 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Gibson ML, Hohmeier KC, Smith CT. Pharmacogenomics testing in a community pharmacy: patient perceptions and willingness-to-pay. Pharmacogenomics 18(3), 227–233 (2017). [DOI] [PubMed] [Google Scholar]
- 44.O’Daniel J, Lucas J, Deverka P et al. Factors influencing uptake of pharmacogenetic testing in a diverse patient population. Public Health Genomics 13(1), 48–54 (2010). [DOI] [PubMed] [Google Scholar]
- 45. Ozdemir S, Lee JJ, Chaudhry I, Ocampo RRQ. A systematic review of discrete choice experiments and conjoint analysis on genetic testing. Patient 15(1), 39–54 (2022). • A systematic literature review assessing features and outcomes of discrete choice experiments and conjoint analysis on genetic/genomic testing.
- 46.Royal College of Physicians (UK). Enabling Patients to Actively Participate in Their Care. (2012). [Google Scholar]
- 47.Balint J, Shelton W. Regaining the initiative: forging a new model of the patient–physician relationship. JAMA 275(11), 887–891 (1996). [DOI] [PubMed] [Google Scholar]
- 48.Schain WS. Patients’ rights in decision making: the case for personalism versus paternalism in health care. Cancer 46(Suppl. 4), S1035–S1041 (1980). [PubMed] [Google Scholar]
- 49.Rosenbaum L The paternalism preference – choosing unshared decision making. Obstet. Gynecol. Surv. 70(12), 739–740 (2015). [DOI] [PubMed] [Google Scholar]
- 50.Trinidad SB, Coffin TB, Fullerton SM, Ralston J, Jarvik GP, Larson EB. “Getting off the bus closer to your destination”: patients’ views about pharmacogenetic testing. Perm. J. 19(3), 21–27 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Gray SW, Hicks-Courant K, Lathan CS, Garraway L, Park ER, Weeks JC. Attitudes of patients with cancer about personalized medicine and somatic genetic testing. J. Oncol. Pract. 8(6), 329–335 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Avey JP, Hiratsuka VY, Beans JA, Trinidad SB, Tyndale RF, Robinson RF. Perceptions of pharmacogenetic research to guide tobacco cessation by patients, providers and leaders in a tribal healthcare setting. Pharmacogenomics 17(4), 405–415 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Shaw JL, Robinson R, Starks H, Burke W, Dillard DA. Risk, reward, and the double-edged sword: perspectives on pharmacogenetic research and clinical testing among Alaska native people. Am. J. Public Health 103(12), 2220–2225 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Lachance K, Korol S, O’Meara E et al. Opinions, hopes and concerns regarding pharmacogenomics: a comparison of healthy individuals, heart failure patients and heart transplant recipients. Pharmacogenomics J. 15(1), 13–19 (2015). [DOI] [PubMed] [Google Scholar]
- 55.Rogausch A, Prause D, Schallenberg A, Brockmoller J, Himmel W. Patients’ and physicians’ perspectives on pharmacogenetic testing.¨ Pharmacogenomics 7(1), 49–59 (2006). [DOI] [PubMed] [Google Scholar]
- 56.Chan SL, Low JJW, Chia KS, Wee H-L. Attitudes on warfarin pharmacogenetic testing in Chinese patients and public. Int. J. Technol. Assess. Health Care 30(1), 113–120 (2014). [DOI] [PubMed] [Google Scholar]
- 57.Kobayashi E, Satoh N. Public involvement in pharmacogenomics research: a national survey on public attitudes towards pharmacogenomics research and the willingness to donate DNA samples to a DNA bank in Japan. Cell Tissue Bank 10(4), 281 (2009). [DOI] [PubMed] [Google Scholar]
- 58.Pereira NL, So D, Bae J-H et al. International survey of patients undergoing percutaneous coronary intervention and their attitudes toward pharmacogenetic testing. Pharmacogenet. Genom. 29(4), 76–83 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Payne K, Fargher EA, Roberts SA et al. Valuing pharmacogenetic testing services: a comparison of patients’ and health care professionals’ preferences. Value Health 14(1), 121–134 (2011). [DOI] [PubMed] [Google Scholar]
- 60.Dong D, Ozdemir S, Mong Bee Y, Toh S-A, Bilger M, Finkelstein E. Measuring high-risk patients’ preferences for pharmacogenetic testing to reduce severe adverse drug reaction: a discrete choice experiment. Value Health 19(6), 767–775 (2016). [DOI] [PubMed] [Google Scholar]
- 61.Anderson HD, Crooks KR, Kao DP, Aquilante CL. The landscape of pharmacogenetic testing in a US managed care population. Genet. Med. 22(7), 1247–1253 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Phillips KA, Sakowski JA, Trosman J, Douglas MP, Liang S-Y, Neumann P. The economic value of personalized medicine tests: what we know and what we need to know. Genet. Med. 16(3), 251–257 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Ginsburg GS, Phillips KA. Precision medicine: from science to value. Health Aff. (Millwood) 37(5), 694–701 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Trosman JR, Van Bebber SL, Phillips KA. Health technology assessment and private payers’ coverage of personalized medicine. JOP 7(Suppl. 3), S18–S24 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Keeling NJ, Rosenthal MM, West-Strum D, Patel AS, Haidar CE, Hoffman JM. Preemptive pharmacogenetic testing: exploring the knowledge and perspectives of US payers. Genet. Med. 21(5), 1224 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Chan SL, Wen Low JJ, Lim YW et al. Willingness-to-pay and preferences for warfarin pharmacogenetic testing in Chinese warfarin patients and the Chinese general public. Per. Med. 10(2), 127–137 (2013). [DOI] [PubMed] [Google Scholar]
- 67.Powell G, Holmes EAF, Plumpton CO et al. Pharmacogenetic testing prior to carbamazepine treatment of epilepsy: patients’ and physicians’ preferences for testing and service delivery. Br. J. Clin. Pharmacol. 80(5), 1149–1159 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The manuscript reports the original results of a face-to-face discrete choice experiment survey in patients. Deidentified data from survey responses are available upon request.