Increasing participant comprehension and autonomy throughout the consent process.

Status: Incomplete (prior to departure) / Role: Lead UX Researcher

A regulatory requirement for human subjects research (HSR) was recently revised and required all research organizations to update their informed consent to be in compliance. Even though our main research study was technically grandfathered into compliance since it existed before the regulatory requirement, our IRB still expected us to make the appropriate updates due to the sheer capacity of what this research was covering.

Since the consent's content was going to be updated, we saw an opportunity to also update the consent's design to revamp the entire process from start to finish to -

• Address new and existing concerns about what our users were consenting to,

• Increase consent rate of users, and

• Meet brand design standards.

Throughout this process, we planned to ensure the following -

• Easy accessibility and navigation for users to identify and read information that mattered to them,

• Easy digestibility for users to actually understand what they were consenting to.

A Blast from the Past!

As a result of unethical practices in human subjects research (e.g., the Tuskegee Syphilis Study), informed consent became a regulatory requirement to protect those that participated in research to increase our scientific knowledge. The current regulations have the minimum requirements on what to share with prospective participants, but lack any guidance on how.

The Common Rule (45 CFR 46) was revised for the first time in 2018-2019 and established new requirements for informed consent to adapt to the rapid improvements in scientific research. The lack of regulatory guidance provided an opportunity to explore how we wanted to implement said regulatory requirements. Some of the main areas that we dug into were focused on -

• What did potential research participants value when participating in genetic research?

• What were some hesitations of research participation?

• What were some existing issues that negatively impacted research consent/participation?

Research Process

Baseline

Based on the media alone, the company had "an inkling" about the negative sentiment that the public had about our product - lots of folks had this conception that 23andMe sold data to health insurance companies to raise premiums, but that wasn't the case at all! Despite that, we wanted to start with a clean slate and establish a baseline understanding of what new users expected to see in a consent and how.

Since this was meant to be a baseline that supplemented what we already knew from various feedback channels, we opted for some unmoderated sessions to be quick, but to also replicate the environment that a person might be reading the consent (by themself...for the most part).

• For the consent content: participants were asked questions about what they expected to read about in a consent and what they wanted to read in a consent.

• For the consent design: participants were shown the consent and were asked to give their initial thoughts and reactions on how it impacted their interpretation and processing.

Define and Refine

We then created a list of topics that a prospective customer might expect and want to read in an informed consent document for a research study.

With the list above, we launched another round of research to refine how we addressed one of the new regulatory requirement updates: the key information summary. The key information summary is meant to be a TLDR before the full consent to give participants a brief overview of what the study entailed, but there was no regulatory definition of what was to be included in it.

To do so, we had participants do a card sorting activity with a Likert-scale rating with the different topics that other participants had identified as essential to an informed consent document. We chose "very important" as the topics that would be included in the key information summary. And of course, participants were encouraged to provide open-ended feedback about their rating.

The other part of the first round of research was to explore potential informed consent document designs - the main consensus was that "the consent was a huge wall of text." It looked intimidating, overwhelming, and had legalese, which contributed a bit of anxiety due to not knowing if they properly understood what it meant. The scrolling never stopped and each section was just full of text that participants couldn't be bothered to look at closely - and not looking at it closely meant they might not fully comprehend what our research was like, which fueled the negative sentiment that 23andMe was already experiencing.

To get some initial concepts started, we visited a few online resources and did a brief literature review to see what other folks had already done in this space. SageBionetworks was one of the resources that inspired our early concept - a recurring sentiment was the lack of navigation and transparency in the content, making it seem like it was purposefully hiding something in the massive wall of text. By breaking down the different sections to emphasize personal narrative and progression, participants could feel empowered to read at what they deemed most appropriate for themself.

Next Steps

Although this was one of my favorite projects, it unfortunately came to end as my time at 23andMe was coming to an end. We were on track to meet with our assigned product team to get this into the roadmap and establish proper resourcing and were in early discussions with design, engineering, and legal to tease apart what we could and couldn't do.

Prior to leaving 23andMe:

• The consent language was in its final stages of revision by internal stakeholders and our IRB. We wanted to include our IRB in on this project because ultimately, they would be the ones reviewing and approving it. We also wanted to take this chance to show them how UX research could be used in this manner to make participant-centric decisions for our study.

• The final consent design was still being discussed based on existing backend infrastructure and designer resourcing. As always, we're limited by what is feasible for engineering to do and that discussion still needed to happen.