Using information provision to increase interest in genetic research participation

The trial will focus on how different types of information provision influence participant behavior. This study will comprise of three arms:
● Arm 1: Baseline genomics research recruitment information with no extra information provision.
● Arm 2: Baseline genomics research recruitment information and additional information targeted at improving trust in genomics research. This arm aims to increase the participant's trust in genomics research by explaining how their privacy will be protected, how they can communicate with the study coordinators or support center if they have any questions and how the research study would protect them against risks.
● Arms 3: Baseline genomics research recruitment information and additional information targeted at improving the perceived benefits of genomics research. In this arm, various additional benefits are explained including: how joining the research study could improve their health, how joining could benefit their family and community and what they could learn about their own health.

2020-11-03

2020-12-16

The primary endpoint is the participants' interest in participating in a pharmacogenomics research program, signaled through the provision of the person’s email address. This is coded as a binary measure, with a value of 1 if the participant leaves their email address and 0 otherwise.

Participants will be offered the opportunity to provide their email address to learn more about participating in a research program that would offer genomics results for free. After the study, participants who provided their email addresses would receive an invitation to an actual nationwide research program. The hypothetical recruitment materials, which the participant read during the experiment, are based on this actual research program. As a participant in the research program, they can submit a biosample and receive genomics results including pharmacogenomic, ancestry and traits and ACMG 59 information free of charge.

The secondary endpoints include:
1. Trust in the genomics research program
2. Perception of benefits of participating in genomics research
3. Comprehension of PGx testing and results, research safeguards, and research benefits

The first two secondary endpoints will be measured by adapting previously published and validated scales to assess trust in and perceived benefits of the hypothetical program described in the participant facing materials [1-3]. The comprehension endpoint will be measured using a series of multiple-choice questions (4 or 8 questions, depending on the treatment). For participants who do not receive the additional information covered in the trust or benefit conditions, the corresponding comprehension questions will be explicitly presented as opinion questions and are not counted in the comprehension score.

References:
1. Platt, J., & Kardia, S. (2015). Public trust in health information sharing: implications for biobanking and electronic health record systems. Journal of Personalized Medicine, 5(1), 3-21.
2. Platt, J. E., Jacobson, P. D., & Kardia, S. L. (2018). Public trust in health information sharing: a measure of system trust. Health Services Research, 53(2), 824-845.
3. Ulrich, C. M., Zhou, Q., Ratcliffe, S. J., Knafl, K., Wallen, G. R., Richmond, T. S., & Grady, C. (2018). Development and preliminary testing of the perceived benefit and burden scales for cancer clinical trial participation. Journal of Empirical Research on Human Research Ethics, 13(3), 230-238.

This study will take place on Predictiv (www.predictiv.co.uk), an online platform for running behavioral experiments built by the Behavioral Insights Team. Predictiv provides access to a large panel, including over 2 million individuals in the US, as well as the functionality to run a range of online experiments, such as choice simulations and comprehension tests.

Participants will be recruited from the general public via a network of panel providers to the Predictiv platform. They will be able to view a brief description of the study before entering the platform, to help them decide if they want to take part. These participants have already consented with their respective panel provider to take part in online research and be contacted about studies for which they are eligible. In addition, they will be explicitly asked if they want to take part in this study after receiving general information about the study, including the task, duration of the experiment and compensation.

We will aim to recruit 4200 participants from the general public via the panel providers connected to the Predictiv platform. We will use sample quotes to recruit a socio-demographically diverse sample as follows:
● Race : 48% white, 52% non-white
● Age : 85% 18-64 years, 15% 65+ years
● Sex : 50% male, 50% female
● Household income : 31% income under $25,000, 69% income over $25,000
● Education : 88% high school or more, 12% less than high school
● Geography : 94% urban/metropolitan, 6% rural (note: We will aim to oversample participants from rural areas so that they make up 10% of the total sample, in order to be able to provide more precise descriptive statistics about them.)

Participants will be pre-screened on the above demographics before being able to access the Predictiv platform. Once we have reached a quota, new participants with that characteristic will not be able to enter the experiment.

Once recruited from the Predictiv platform, the participant is shown a welcome screen which orients them to the experiment tasks and the duration of the study. After giving consent, the participant is presented with a summary screen detailing the various parts of the study, the duration (8-10 minutes). Participants receive a fixed compensation from their panel provider based on the experiment duration (estimated at around $1.50 to $2.50 depending on their specific demographic - this variability is due to the fact that participant compensation is determined dynamically using a marketplace format). In addition, the participant can earn additional variable compensation of a maximum of $1.00 (in the control condition) or $2.00 (in the trust and perceived benefits conditions). Information on variable compensation is provided to participants during the instructions.

To begin the study, the participant answers demographic questions about themselves including age, gender, insurance status, and ethnicity. After answering the demographic questions, participants are asked to review recruitment materials for a hypothetical research program. The recruitment materials differ depending on the treatment condition the participant has been assigned to. After seeing the recruitment materials, the participant is asked whether they would join a similar research program. If answered affirmatively, the participant has an opportunity to provide their email address to receive an invitation to join a similar research program. Those who answered they would not join the research program are asked their reasons for not joining. The participant is then asked to complete a survey with their opinions about the research program which aims to assess their level of trust and their perceived value of participating in genomic research. A comprehension quiz is then conducted to assess their understanding of the research program, earning them an additional $0.25 for each correct response. The final section of the study asks the participant to answer a series of belief questions about what they think the research program offers (corresponding to the program components the participant did not see recruitment materials about). These are questions are not incentivized.

The study will be active in the Predictiv system for an estimated four weeks allowing for individuals to participate in the study.

Once they consent, participants are randomly assigned a number between 1 and 3, which corresponds to one of the three trial arms.

The unit of randomization is at the individual participant level. This happens when participants enter the experiment and individuals are not re-randomized.

No

N/A

Approximately 4200 persons

Approximately 1400 persons per arm.

Alpha (significance level): 5% Power: 80% Total planned sample size: 4200 Number of trial arms: 3 Base rate or SD: 50% This data is derived from another experiment, preregistered here: https://osf.io/k5ywc/. Expected MDES for this trial based on power calculations: Cohen’s h = -0.116; ~ 5.8pp Multiple comparison corrections: Yes, we have corrected for two comparisons using the Benjamini-Hochberg step-up procedure, assuming that we will be contrasting both of the intervention arms with the control arm. Covariate adjustment: The power calculation doesn’t take into consideration the fact that there will be a rich set of covariates available in our dataset. These power calculations were run in R using the following code: library(magrittr) baselines <- c(0.40, 0.50, 0.60) n_comparisons <- 2 # only treatments against control for(one_baseline in baselines){ power.prop.test(n = 4200/3, p1 = one_baseline, power = 0.80, sig.level = 0.05/2) %>% print() } # note: with covariate adjustment, our power should improve # cohen's h for MDES 2*(asin(sqrt(0.5)) - asin(sqrt(0.5581604))) # anticipated cohen's h for effect size of 6% 2*(asin(sqrt(0.5)) - asin(sqrt(0.56)))