Effect of co-payment on behavioral response to consumer genomic testing

Individuals who paid out-of-pocket for consumer genomic testing were more likely to share results with their physician and obtain follow-up health screenings, compared to those who received fully subsidized testing.

Abstract

Existing research in consumer behavior suggests that perceptions and usage of a product post-purchase depends, in part, on how the product was marketed, including price paid. In the current study, we examine the effect of providing an out-of-pocket co-payment for consumer genomic testing (CGT) on consumer post-purchase behavior using both correlational field evidence and a hypothetical online experiment. Participants were enrolled in a longitudinal cohort study of the impact of CGT and completed behavioral assessments before and after receipt of CGT results. Most participants provided a co-payment for the test (N = 1668), while others (N = 369) received fully subsidized testing. The two groups were compared regarding changes in health behaviors and post-test use of health care resources. Participants who paid were more likely to share results with their physician (p = .012) and obtain follow-up health screenings (p = .005) relative to those who received fully subsidized testing. A follow-up online experiment in which participants (N = 303) were randomized to a “fully-subsidized” versus “co-payment” condition found that simulating provision of a co-payment significantly increased intentions to seek follow-up screening tests (p = .050) and perceptions of the test results as more trustworthy (p = .02). Provision of an out-of-pocket co-payment for CGT may influence consumer’s post-purchase behavior consistent with a price placebo effect. Cognitive dissonance or sunk cost may help explain the increase in screening propensity among paying consumers. Such individuals may obtain follow-up screenings to validate their initial decision to expend personal resources to obtain CGT.

Implications

Practice: Price incentives for consumer genomic testing may serve as a mechanism that can be used to motivate post-test health behaviors.

Policy: Stakeholders, including genetic testing companies and insurers, should consider the effect of pricing schemes, particularly out-of-pocket consumer costs, on behavioral response to consumer genomic testing.

Research: Future research should explore the effect of different pricing schemes across different types of consumer genomic tests, as well as across a diverse range of populations.

INTRODUCTION

With consumer genetic testing (CGT) becoming increasingly available for everyday consumers, a growing body of research has examined the effect of CGT on health and behavioral outcomes [1–4]. This body of research, including a recent meta-analysis, has generally found minimal or no effect on lifestyle [3] or psychological status [5, 6]. On the other hand, there is evidence that CGT does prompt some fraction of those who obtain it to share their results with a physician or healthcare provider [7], which has the potential to result in downstream impacts on health outcomes. Importantly, however, existing research has generally not discussed nor examined the role of marketing variables such as price in post-CGT behavior and outcomes. This is despite research in behavioral economics showing that the way products and services are marketed, including how they are priced, can significantly impact post-acquisition behavior [8, 9].

Over the past decade, providers of CGT have experimented with different pricing schemes. For instance in December of 2012, the CGT company 23andMe decreased the price of their service from $299 to $99 [10] in order to reach their goal of recruiting 1 million customers. Moreover, companies have also sometimes collaborated with employers such that the out-of-pocket cost of CGT to employees can be as low as zero [5]. The pricing of CGT is an important policy question for public and private entities including government agencies, employers, and insurance companies. Interestingly, research in health behavior suggests in addition to impacting the adoption of the test, different pricing strategies may also influence health-related behaviors post testing [8], which in turn can influence the realized health benefits of CGT. In particular, these studies find that when consumers pay (versus do not pay) for a treatment such as a pain medication, they expect the product to work better, and this higher expectation can translate into better actual outcomes (e.g., if the person believes the product will work, they will also align their own behavior to aide its efficacy). Building on this literature, this research examines the possibility that the out-of-pocket price paid by consumers for CGT can positively affect consumer post-test expectations, intentions, and actions.

The potential effect of different pricing strategies is important for CGT because like many other health and medical products, the realized health benefits of genetic testing, if any, critically depend on the post-test efforts of the consumer. Because CGT offers information but not treatment, for it to translate into actual improvements in health outcomes, it is arguably the case that the consumer must act on it in some way. For instance, the consumer may end up doing nothing with the information, and hence realize little value from the test, or the consumer may expend the effort to change their health behaviors or seek further professional medical advice or services, which may or may not lead to health benefit. Furthermore, from a health research perspective, if pricing affects post-test behavior, then all research studying the effect of CGT health outcomes should take care to discuss whether and how study participants paid for CGT—for example, if participants received CGT for free in the trial and later no post-test health benefit is observed, one might consider whether the conclusions can be generalized to contexts where participants obtained CGT at a cost.

Despite the importance of understanding the effect of pricing on consumer response to genetic testing, little research exists on this issue. In the current study, we examine the effect of an out-of-pocket co-payment for CGT on consumer’s post-purchase behavior using both correlational field evidence and a hypothetical online experiment.

FIELD STUDY

Methods

The Scripps Genomic Health Initiative (SGHI) is a longitudinal cohort study originally designed to determine the impact of CGT on consumers. See previously published papers [11, 12] for details of the original study design and results for the primary endpoints. Briefly, the SGHI was launched in 2008 with the goal of investigating the behavioral and psychological impact of CGT. Participants in the SGHI were offered the Navigenics Health Compass genetic test, which was a commercially available CGT of genetic risk for common diseases. Study participants were recruited, in part, through health and technology company employers. They completed a baseline survey at enrollment to record their demographic, health, and behavioral information, then received the Navigenics Health Compass CGT and subsequent results. An average of 6 months later, they were again surveyed on their health outcomes and behaviors.

Study groups

An interesting feature of the SGHI, but not a topic that was explored in prior publications [11, 12], is that participants were offered different pricing of CGT when they were recruited. The amount of co-payment required, in part, depended on how different employers subsidized the cost of CGT, and in part, on when the participant enrolled. The initial co-payment for participants who enrolled at the beginning of the study was $150. To encourage early enrollment, the co-payment increased over time with the highest amount, charged in the final months of recruitment, being $470. Thus the co-payment group participants paid anywhere from $150 to $470 for CGT. Another group of participants (we will refer to them as the fully subsidized group) received CGT for free because their employer paid the full cost of the test. All participants who received a fully subsidized test were employees of Sempra Energy, whereas a range of employers were represented among those in the co-payment group. Although this characteristic of our study group renders assessment of bias related to potential employer effects impossible, it is important to note that the consumers themselves did not choose their payment program, but were assigned to a pricing condition.

Measures

See earlier publications [4, 5] for the full list of survey measures administered at baseline and at the 6-month post-test follow-up. Relevant to the current research, participants reported the following behaviors 6 months after they received their CGT results: (i) the number of follow-up screening tests done and (ii) whether they had shared the test results with their physician. These two measures are the dependent measures of interest to the current research. Participant’s demographic and health information taken at baseline, as well as their propensity to see a physician at baseline, were also assessed.

Data analysis

Potential differences in the number of completed medical screenings between pricing groups were assessed using analysis of covariance. Logistic regression was used to assess the association between pricing group and whether results were shared with a physician. In all cases, the following covariates were included in the analysis: age, gender, health-related occupation (yes/no), ethnicity, income, education, follow-up survey interval in days, and the type of survey completed (i.e., some participants completed a shorter version of the follow-up survey, which was used to increase response rate). All statistical analyses were performed in SPSS, and all reported p values are uncorrected.

Results and Discussion

Sample characteristics, including demographic information by pricing group, are presented in Table 1. There were 369 individuals who received a fully subsidized test (18%) and 1,668 individuals who provided an out-of-pocket co-payment (82%). There were no statistically significant differences in baseline propensity to see a doctor between payment groups (p = .812). There were small but statistically significant differences between the groups with respect to gender, income, education, ethnicity, occupation, and follow-up interval. Thus, all analyses incorporated these factors as covariates. We also note that there was differential attrition as a function of pricing group with a follow-up rate of 54% in the fully subsidized group and 63% in the co-payment group (p < .0005). This difference is consistent with the view that paying a price for the test may increase one’s commitment to the test and hence downstream activities related to it, including participating in follow-up research related to the test. Our next analysis on downstream health behavior is thus limited to subjects engaged enough with CGT to participate in the follow-up survey. Among the research-engaged subjects, tested whether paying a co-pay still produced differential behavioral outcomes compared with those who received the test for free.

Table 1.

Sample characteristics

Demographic/baseline variable	Fully subsidized	Out-of-pocket Co-Payment
N	369	1,668
Age, Mean (SD)	47.3 (10.3)	46.6 (12.4)
Gender (female, %)	49.3	56.7
Annual income (Median category)	100–149k	100–149k
Education (Median category)	Graduated 4-year college	Completed some post-college
Ethnicity (Caucasian, %)	74.0	86.5
Health-related occupation (SH Employee, %)	<1	28.6
Long-term follow-up interval in days, Mean (SD)	143.7 (41.8)	174.1 (76.4)

A logistic regression on whether one shared one’s CGT results with a physician as the dependent variable, and with pricing group and the above covariates as predictors showed that participants in the co-payment group were significantly more likely to share their results with their physician post-testing (odds ratio = 1.5, p = .012; Table 2). Specifically, 27.9% of the co-payment group reported sharing their results with their physician, versus only 21.3% of the fully subsidized group. Furthermore, analysis of covariance (ANCOVA) shows that participants who provided a co-payment also completed a greater number of health screening tests during the follow-up period (3.3 tests done), relative to those who received fully subsidized testing (2.6 tests done, p = .005; Table 2). As stated above, at baseline, the groups did not differ in their propensity to visit a physician.

Table 2.

Impacts of fully subsidized versus out-of-pocket co-payment for CGT on study outcomes at an average of 6-months post testing (n = 369 subsidized, n = 1668 paid)

Outcomes	Fully Subsidized	Co-Payment	Effect Size	p Value
Number of screenings completed, Mean (SD)	2.6 (2.5)	3.3 (2.7)	η2 < .004	.005a
Shared results with Physician (Yes, %)	21.3	27.9	OR = 1.5	.012b

Eight covariates were age, gender, health-related occupation, ethnicity, income, education, follow-up interval in days, and short versus original follow-up survey.

CGT consumer genomic testing; OR odds ratio.

^aAnalysis of covariance, effect of payment group.

^bLogistic regression, effect of payment group.

We note that although we see these differences between individuals who either did or did not provide a co-payment, we found no further price effects within the co-pay range of $150 to $470. One possible explanation is that this may be due to a ceiling effect, whereby $150 is sufficiently motivating to get a maximum shift in follow-up actions. Furthermore, given the self-selected enrollment nature of this field study, we cannot disentangle whether the differences in post-test behaviors as a function of co-payment merely captured greater importance placed on the test a priori (i.e., selection of consumers based on price; [13]), or if the co-payment actually caused the difference in post-test behavior (i.e., price placebo effect, [8]). Therefore, we conducted the experiment described below to examine whether provision of a co-payment can cause shifts in consumers’ perceptions and intentions after a CGT scenario simulation.

EXPERIMENTAL PROCEDURE

Methods

A total of 405 subjects were recruited on MechanicalTurk (U.S. users only) to participate in a decision-making study and completed the study (another 27 people, evenly distributed between conditions, started the survey but did not complete it, and were excluded from analysis). At the start of the study, participants were given information about CGT and were asked to imagine that the test normally costs $200, but that their employer is offering the test at a subsidized rate of $30. Participants were asked whether they would purchase the test for $30 (yes, no). We planned, a priori, to only study the subset of subjects who responded yes to this question, to eliminate any price-based selection effect (i.e., all subjects analyzed placed at least a $30 value on genetic testing). Our analysis was thus based on 303 individuals (75% of the initial subject pool). These participants had a mean age of 34 years, and 41% were female. For more demographic information pertaining to the general MechanicalTurk population, see the study by Paolacci and Chandler [14]. While U.S. MechanicalTurk participants tend to be racially and ethnically diverse, this population is also over-represented by people who tend to use the Internet (e.g., younger, more educated). Thus our results should be interpreted with these sample characteristics in mind.

After participants indicated they were willing to pay $30 for genetic testing, they were randomly assigned to one of two conditions. In the co-payment condition, they were asked to imagine that they purchased the test for $30. In the fully subsidized condition, subjects were told, alternatively, that their employer has elected to cover the full cost of testing, and therefore that the test would be free of charge to them. All participants were then guided through a simulated experience. Specifically, they went through screens where they agreed to testing, entered payment information (co-payment condition only), and received and then sent back the test kit (see Supplemental Material).

At the end of this simulated transaction process, all participants were told that they have now received the test results, and they will read part of the results. They then saw test results for three common conditions, including colon cancer, type 2 diabetes, and macular degeneration. In all three, they were told they had higher than average genetic risks, and were given recommendations to see a doctor to receive follow-up screening tests. After reading the information, they were asked “How likely are you to call your doctor in the next a few days to make an appointment to talk about all of the above?” (1 = Not likely; 7 = Very likely), as well as about their perceptions of the test (“How trustworthy [accurate, informative, high quality] are the test results?”). These measures were the key outcome variables for the experimental procedure. We also included a lifestyle intentions question (Intention to take Omega-3, which has been found to be beneficial to the diseases). In addition, we measured their general attitude toward the test result (“How seriously do you take the test results?” “How interested are you to learn more about your test results?”), and their perceptions of the cost (e.g., “How costly (money-wise) is the test to you?” see Table 3 for details of all measures).

Table 3.

Experimental study measures, and results by payment group (n = 152 free, n = 151 copay)

Measures	Fully Subsidized, Mean (SD)	Co-Payment, Mean (SD)	F (1, 299) for Payment Group	p Valuea
Follow-up intentions
How likely are you to call your doctor in the next a few days to make an appointment to talk about all of the above? (1=Not likely; 7 = Very likely)	4.39 (1.89)	4.77 (1.84)	3.88	.050**
Test result perceptions
How trustworthy are the test results? (1=Not at all; 7 = Very much)	5.13 (1.10)	5.41 (1.04)	5.19	.02**
How accurate are the test results? (1=Not at all; 7 = Very much)	5.03 (1.04)	5.27 (1.12)	3.86	.050**
How informative are the test results? (1=Not at all; 7 = Very much)	5.59 (1.28)	5.59 (1.21)	0.004	.95
How high quality is this test? (1=Not at all; 7 = Very much)	5.11 (1.15)	5.37 (1.11)	4.02	.046**
Lifestyle change intention
Omega-3 has been found to have many health benefits related to the conditions above. How likely are you to start taking (or continue to take) an Omega-3 supplement daily? (1=Not likely; 7 = Very likely)	5.44 (1.57)	5.52 (1.57)	.17	.68
General attitude towards test results
How seriously do you take the test results? (1=Not at all; 7 = Very much)	5.53 (1.26)	5.68 (1.26)	.94	.33
How interested are you to learn more about your test results? (1=Not at all; 7 = Very much)	5.90 (1.24)	5.91 (1.19)	.04	.85
Cost perceptions
How costly (money-wise) is this test to you? (1=Not at all; 7 = Very much)	2.37 (1.98)	2.96 (1.78)	7.09	.008***
Is this test worth the cost (money-wise)? (1=Not at all; 7 = Very much)	6.03 (1.40)	5.78 (1.31)	2.44	.12
How costly (effort-wise) is this test to you? (1=Not at all; 7 = Very much)	2.98 (2.02)	3.34 (2.07)	2.08	.15
Is this test worth the cost (effort-wise)? (1=Not at all; 7 = Very much)	6.01 (1.32)	5.91 (1.30)	.30	.58

ANCOVA, analysis of covariance.

^aANCOVA with payment group as independent variable, gender, and age as covariates.

**Significant at the p = .05 level.

***Significant at the p = .01 level.

Results and discussion

An ANCOVA with price as the factor and gender and age as covariates showed that those who simulated having provided a co-payment (vs. not providing a co-payment) for CGT indicated significantly greater interest to see a doctor soon for follow-up tests (M_co-pay = 4.77, SD = 1.84; M_free = 4.39, SD = 1.89; p = .050). Age also had a positive effect, whereby older individuals were more likely to indicate greater interest in scheduling follow-up tests (p = .01).

ANCOVA’s with gender and age as covariates also showed that those who simulated having provided a co-payment (vs. not providing a co-payment) for CGT perceived the test as higher in quality (M_co-pay = 5.37, SD = 1.11; M_free = 5.11, SD = 1.15; p = .046), more accurate (M_co-pay = 5.27, SD = 1.12; M_free = 5.03, SD = 1.04; p = .050), and more trustworthy (M_co-pay = 5.41, SD = 1.04; M_free = 5.13, SD = 1.10; p = .02). Interestingly, co-payment participants did not perceive the test to be more informative than fully subsidized participants (M_co-pay = 5.59, SD = 1.20; M_free = 5.59 also, SD = 1.28; p = .95). This suggests that payment did not influence the perception of the amount of useful information contained in the test results; however, payment did significantly influence people’s trust and perception of the accuracy and quality of the information.

Finally, payment did not significantly change people’s intention to use Omega-3 supplements (M_co-pay = 5.52, SD = 1.57; M_free = 5.44, SD = 1.57 also; p = .68). This suggests that the payment effect may be limited to follow-up behaviors that are immediate and direct next steps, but may not readily impact broad lifestyle change intentions. Payment also did not affect general attitude toward the test. Unsurprisingly, it did affect the perception of the costliness of the test, whereby those who simulated a copay felt the test was more costly than those who simulated receipt of CGT for free. See Table 3 for full results on all measures.

DISCUSSION

Our findings from both a field study of consumers of CGT and a hypothetical online experiment suggest that provision of a co-payment by consumers for CGT may influence consumers’ post-test perceptions of the test and their post-test behaviors. Specifically, field data revealed a correlation between provision of a co-payment for CGT and post-test completion of follow-up health screening tests and sharing of CGT results with a physician. In addition, a hypothetical online experiment showed that simulating receipt CGT with a co-payment (vs. fully subsidized CGT) significantly increased the intention to see a physician for follow-up screening tests, as well as the perceived quality and trustworthiness of the test results. This research suggests the that price placebo effect may occur for CGT products, and therefore may impact down-stream medical decision making and behavior.

Previous research in behavioral economics and psychology shows that the price paid often influences downstream usage behavior and performance through changing a person’s expectations about the product [8]. For example, a price discount may lead consumers to expect a product to work less well, which can change the person’s subjective experience and subsequent performance using the product. Another reason paying versus not paying for a product may change downstream behavior is that of cognitive dissonance [15]. Specifically, people want to view themselves as being consistent in their actions; hence paying for a product, and expecting the product to be of high quality, are consistent behaviors. Another theory that might explain the observed effect, however, is the sunk cost bias [16]. Specifically, when deciding on a current action, people tend to consider how much cost has already been spent on the course of action, and they will try to recoup the cost already spent on the course of action. Thus individuals may obtain follow-up screenings as a way of recouping the cost of the test and to validate their initial decision to expend personal resources to obtain it. The current research found evidence consistent with the expectations account. Specifically, in the simulated experience study, subjects in the copay condition perceived the CGT results to be more accurate and trustworthy, suggesting that they expect the CGT results to offer better predictions and suggestions regarding their health. Such expectations may, in turn, motivate them to seek further medical services as suggested by the CGT results. Future research may explore the psychological underpinnings of the observed price effect on CGT in greater depth.

Future research can build on our evidence and conduct prospective trials in the field where price is randomized and consumer selection under different prices is well-controlled. Ultimately, increased understanding of the impacts of CGT co-payment on behavioral response to genetic testing may help shape public policy in this rapidly evolving area of health care. Such research may also generalize to other health and medical tests and products and thus provide even greater value to the field.

Marketing and public policy implications

While advances in genetic testing have been significant, their value when offered to the consumer as a non-prescription, over-the-counter service remains unclear. Medical researchers are continuing to study the behaviors of consumers after undergoing genetic testing. A study using hypothetical scenarios to assess willingness to pay for genetic testing found that participants prefer to take predictive tests, even when a direct treatment does not exist [17]. Thus this research examining consumer response to genetic testing may offer a significant contribution to the medical and public policy domain by providing a marketing and consumer behavior dimension into the medical literature.

Furthermore, understanding what influences consumers to exert effort in obtaining health screenings is an important topic due to benefits from potential early detection of disease. One of the key benefits of genetic testing is that it may allow consumers to engage in targeted disease screening based on one’s personal risk profile. Getting health screenings can often lead to earlier diagnosis of disease, where benefits include prolonged survival and better maintenance of quality of life. Early detection may also allow health policy experts to predict and plan for utilization of health resources. The relationship between consumer out-of-pocket expense of CGT and the extent to which CGT motivates adherence to accepted health screening guidelines is, therefore, an important area of study. Moreover, a critical additional area of study relates to how pricing of CGT may result in possible exacerbation of existing health disparities. Specifically, research that explores the relationship between socioeconomic status, disposable income, purchase/adoption of CGTs, and health outcomes is critical. The broad availability and marketing of CGT could potentially further divide access to state-of-the-science medicine across socioeconomic communities.

Limitations

The sample included in our field study likely represents “early adopters” of CGT and includes primarily Caucasian, college educated individuals that have an above average socioeconomic status. Therefore, this sample is not generalizable to the larger public. In addition, the longitudinal cohort design did not include a control group. Moreover, because all of the individuals in the fully subsidized CGT group came from a single employer (vs. from multiple employers as in the co-payment group), there may be bias related to employer effects for which we were unable to account. We are also unable to speak to the subsample who did not respond to our follow-up survey, and the fact that we lost more subjects to attrition in the free condition than in the co-pay condition. Strengths of the study include the conduct of both a large sample field study in addition to a separate hypothetical online experiment with a rigorous randomized design.

CONCLUSION

This research adds to existing literature on CGT during an important time when the industry is evolving and the nature of future offerings are uncertain given the U.S. regulatory environment. For instance, 23andMe (https://www.23andme.com) was issued a warning letter by the FDA in Fall 2013 and discontinued returning genetic risk information [18]. Recently, however, the company received approval to, once again, begin offering a subset of 10 specific genetic risk assessments to consumers. In this light, as well as regardless of whether consumers will purchase the test independently or through a physician, this study provides useful insights to understand the impact of pricing, including out-of-pocket consumer expense, on consumer behavioral response to CGT.

Supplementary Material:

Supplementary material is available at Translational Behavioral Medicine online.

Compliance with Ethical Standards

Conflict of Interest: None declared.

Primary Data: Findings reported have not been previously published and this manuscript is not being simultaneously submitted elsewhere. Data have not been previously reported elsewhere. The authors have full control of all primary data and agree to allow the Journal to review data if requested.

Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This article does not contain any studies with animals performed by any of the authors.

Informed Consent: Informed consent was obtained from all individual participants included in the study.