Consumer Responses to Price Disclosure in Direct-to-Consumer Pharmaceutical Advertising

Abstract

This survey-based behavioral experiment was conducted to understand how consumers are likely to respond to including the drug price in any direct-to-consumer pharmaceutical advertising.

In the “American Patients First”¹ blueprint released in May 2018, the Trump administration proposed including the drug price in any direct-to-consumer pharmaceutical advertising (DTCPA) as an approach to lower prescription drug prices. In October 2018, the Centers for Medicare & Medicaid Services proposed requiring that television DTCPA disclose drug prices.² We conducted a behavioral experiment to understand how consumers are likely to respond to the price disclosure.

Methods

We recruited participants using Amazon’s Mechanical Turk,³ an online job board commonly used to enlist experiment participants. Participants were instructed to assume they had recently been diagnosed as having type 2 diabetes. They were randomly assigned to view 1 of 5 advertisements for a fictitious diabetes prescription drug (“Mayzerium”). Representing the current practice, the advertisement in the control condition made no mention of the drug’s price. The remaining 4 advertisements disclosed either a low ($50 per month) or high ($15 500 per month) price, representing the 1st and 99th percentiles, respectively, of the average wholesale price in 2016 of diabetic prescription drugs.⁴ In 2 “modifier” conditions, the advertisement included a modifying statement with the disclosed price indicating that “eligible patients may be able to get Mayzerium for as little as $0 per month” (the advertisements are available in the eAppendix in the Supplement). Such language commonly appears on advertisements for drug coupons and co-pay/coinsurance assistance cards.

This study received institutional review board approval from Clemson University. All participants provided implied consent by participating in the survey.

After viewing the advertisement, participants completed a questionnaire to measure their likelihood (ranging from 1 [highly unlikely] to 7 [highly likely]) of asking their physician about the drug, asking their insurer about the drug, researching the drug online, and taking the drug. Pairwise comparisons of responses were conducted using Wilcoxon rank sum tests. In addition, participants were asked about the expected out-of-pocket cost and the perceived effectiveness of the drug (ranging from 1 [highly ineffective] to 7 [highly effective]). They also answered demographic questions, and each received $0.50 as compensation for participation.

Results

Our sample included 580 participants, representing a wide range of ages, household incomes, education, insurance coverage, and health (Table 1). For the low-priced drug, the price disclosure, with or without the modifier, did not alter consumer responses. For the high-priced drug, the price disclosure significantly reduced the likelihood of participants asking their physician about the drug (5.12 vs 2.90; P < .001), asking their insurer about the drug (5.01 vs 4.09; P = .003), researching the drug online (5.94 vs 4.92; P < .001), and taking the drug (4.93 vs 3.24; P < .001) (Table 2). However, these results were significantly mitigated when the modifier was included: asking their physician about the drug (2.90 vs 4.48; P < .001), asking their insurer about the drug (4.09 vs 4.85; P = .01), researching the drug online (4.92 vs 5.74; P = .003), and taking the drug (3.24 vs 4.36; P < .001). Participants did not perceive the low-priced drug as significantly less effective than the high-priced drug. Results were robust when controlling for all demographic variables listed in Table 1.

Table 1. Characteristics of 580 Study Participants^a.

Characteristic	No. (%)
Age, y
18-25	74 (12.8)
26-34	218 (37.6)
35-44	150 (25.9)
45-54	70 (12.1)
55-64	43 (7.4)
65-74	25 (4.3)
Sex
Male	317 (54.7)
Female	263 (45.3)
Household income, $
<25 000	102 (17.6)
25 000-34 999	105 (18.1)
35 000-49 999	100 (17.2)
50 000-74 999	133 (22.9)
75 000-99 999	71 (12.2)
100 000-149 999	47 (8.1)
150 000-199 999	15 (2.6)
200 000-249 999	5 (0.9)
≥250 000	2 (0.3)
Education
Some high school	3 (0.5)
High school	69 (11.9)
Some college	144 (24.8)
Associate’s degree	83 (14.3)
Bachelor’s degree	206 (35.5)
Some postgraduate study	15 (2.6)
Master’s degree	45 (7.8)
Doctor’s degree	15 (2.6)
Has health insurance
Yes	464 (80.0)
No	116 (20.0)
Has high-deductible health plan
Yes	103 (17.8)
Nob	477 (82.2)
Has prescription drug coverage
Yes	402 (69.3)
Noc	178 (30.7)
History of type 2 diabetes
Yes	36 (6.2)
No	544 (93.8)
History of other chronic condition
Yes	113 (19.5)
No	467 (80.5)
Use of prescription drugs
Yes	203 (35.0)
No	377 (65.0)

^aTo participate in the experiment, individuals must have been located in the United States, have an Amazon’s Mechanical Turk³ approval rating of at least 95%, have completed at least 500 previous tasks on Amazon’s Mechanical Turk, and have passed an attention check question included at the beginning of the experimental instrument.

^bParticipants who indicated they did not have health insurance or were unsure whether they had a high-deductible health plan were considered as not having a high-deductible health plan.

^cParticipants who indicated they did not have health insurance or were unsure whether they had prescription drug coverage were considered as not having prescription drug coverage.

Table 2. Questionnaire Responses Among 580 Study Participants Exposed to Different Direct-to-Consumer Pharmaceutical Advertising Formats.

Questionnaire Item	Mean (SD)
	No Price (n = 116)	Low Price		High Price
		Without Modifier(n = 115)a	With Modifier (n = 116)b	Without Modifier (n = 116)a	With Modifier (n = 117)c
Ask your doctor about the drugd	5.12 (1.73)	5.29 (1.51)	5.33 (1.57)	2.90 (2.21)e	4.48 (1.95)e
Ask your insurer about the drugd	5.01 (1.82)	5.17 (1.75)	5.19 (1.80)	4.09 (2.27)f	4.85 (1.99)g
Research the drug onlined	5.94 (1.32)	6.12 (1.17)	6.12 (1.29)	4.92 (2.02)e	5.74 (1.47)f
Use the drugd	4.93 (1.52)	4.91 (1.43)	4.92 (1.49)	3.24 (2.05)e	4.36 (1.63)e
Expected out-of-pocket cost, $	77.76 (148.94)	49.75 (43.62)	52.60 (77.97)	2787.08 (5209.57)e	1355.39 (3624.00)f
Perceived effectivenessh	5.14 (1.17)	5.10 (1.32)	5.23 (1.11)	5.29 (1.17)	5.14 (1.15)

^aTwo-tailed Wilcoxon rank-sum tests were conducted to determine whether the median value of the column is statistically different from the median value of the “No Price” column.

^bTwo-tailed Wilcoxon rank-sum tests were conducted to determine whether the median value of the column is statistically different from the median value of the “Low Price: Without Modifier” column.

^cTwo-tailed Wilcoxon rank-sum tests were conducted to determine whether the median value of the column is statistically different from the median value of the “High Price: Without Modifier” column.

^dRanging from 1 (highly unlikely) to 7 (highly likely).⁵

^eP < .001.

^fP < .01.

^gP < .05.

^hRanging from 1 (highly ineffective) to 7 (highly effective).⁵

Our study had some limitations. First, actual patients might respond differently than experiment participants. Second, our results might not be generalizable to drugs of other therapeutic classes, in different price ranges, or using other marketing strategies in DTCPA. Third, clinician responses to price disclosures were outside the scope of this study.

Discussion

While price disclosure had little influence on consumer responses to the low-priced drug, it substantially decreased consumer interest in the high-priced drug. However, this finding weakened if the advertisement included a modifier indicating that consumers’ out-of-pocket cost might be zero. Although many challenges remain in designing the ultimate US Food and Drug Administration regulation,⁶ our results suggest that requiring pharmaceutical companies to disclose the price in DCTPA can be potentially effective in reducing consumer interest in high-priced drugs, but the inclusion of modifiers in these disclosures can reduce or eliminate the influence of disclosure.

Supplement.

eAppendix. Supplemental Appendix