#1370 2010 · 23andMe · personal genomics
Customers Pay to Join the Research Cohort They Become
the problem
Genome-wide studies starved for huge surveyed cohorts: clinical recruitment cost too much to ask enough questions fast enough.
background
Traditional genetic epidemiology recruited cohorts through clinics: slow, expensive, and rarely funded for trivial-seeming traits. 23andMe, the personal genomics company, sold consumers their own genotyping and, in the same flow, invited them to consent to research.
For a study published in PLoS Genetics in 2010, the company drew on customers who consented to research use of anonymized genotypes and survey answers: 13 web-based surveys covering 22 traits, from hair curl and freckling to asparagus anosmia and the photic sneeze reflex.
what everyone would do
Write grants to fund clinical recruitment, one questionnaire at a time.
what they saw
Recruiting participants is a cost center only when participants get nothing. Pay them in their own data and they fund the study, answer surveys willingly and come back for recontact: the cohort is the business model.
the move
The cohort pays for itself. Customers buy the kit to learn their own genetics; consenting to research costs them a click, and answering web surveys deepens engagement with the product they already paid for. The result is a study sample of 9,126 unrelated customers of northern European ancestry with genotypes and phenotypes gathered at near-zero marginal cost. The paper highlights 'the low marginal cost of asking additional questions over the web and the speed of broadcasting recruitment messages in parallel online', plus cheap recontact, because participants keep returning to see their own data.
why it works
Curiosity about oneself is the strongest survey incentive; the product supplies genotypes as a byproduct; consent and questions ride on engagement customers already want; the marginal cost of each additional question approaches zero.
the payoff
Novel genome-wide associations for hair curl, freckling, asparagus anosmia and photic sneeze reflex, from 9,126 consenting customers (PLoS.
where it breaks
It fails on representation (early cohorts skewed toward paying customers of European ancestry, limiting transferability), on regulator and public trust in health-data consent, and on any trait that needs clinical measurement a survey cannot capture.
what came after
The participant-funded cohort became 23andMe's second business, therapeutic partnerships on consented data, and the template for consumer research panels across health apps.
references
- [1]Web-Based, Participant-Driven Studies Yield Novel Genetic Associations for Common TraitsPLoS Genetics, 2010journals.plos.org
- [2]Genentech, 23andMe Seek Participants for Study on Consumer-Driven GWAS ModelGenomeWeb, 2015genomeweb.com