A couple of months ago I (with Sarah Greer Rodock) published a paper, linked HERE, about the Spit for Science program at Virginia Commonwealth University. That program, originated by Kenneth Kendler (who later dissociated himself from it) and Danielle Dick, genotyped more than 12,000 students at VCU and administered comprehensive assessments of their alcohol use and related behaviors.
In about twenty peer-reviewed papers reporting genetically-informed results, something amazing happened: they didn’t find anything. The study, founded on an assumption that the heritability of alcohol use and abuse is close to .5, estimated the heritability to be less than .1, and more often than not, not significantly different from zero. No replicable gene or SNP associations were identified. The median R2 for the polygenic scores they reported was around one half of one percent; again, most of the time the scores did not predict better than statistical chance.
As we documented in our paper, the authors of the reports described the results clearly in a numerical sense, but never acknowledged that the results seemed to suggest that something wasn’t working at a theoretical level. Papers started with an assertion of the primary importance of genetic variance, reported null results, then concluded with a triumphant re-assertion of the importance of genetics. In recent years Danielle Dick has written a series of over-the-top theoretical papers in which she endorses the movie Gattaca as a model for the future of psychiatric genetics. I’m not kidding.
The one thing that saved Spit for Science from being a complete ethical disaster is that no feedback was provided to the individual students. Although they were certainly led to believe that some of them had “genetic predispositions” that caused them to drink, no individual predictions were made on the basis of the individual polygenic scores.
Now that has changed. Danielle Dick has moved on from VCU to Rutgers, where she is Director of the Addiction Research Center. Here she has started what looks to be Spit for Science 2.0, recruiting people age 18-25 to give their DNA that: “provides individuals with personalized risk profiles that combine genetic, behavioral, and environmental information.“
Danielle Dick has spent the last 15 years demonstrating beyond any reasonable doubt that there is no way this information, at least the genetic part of it, will be valid. Although I am willing to be told about some new piece of evidence that hasn’t been reported, I think manipulating young people’s impressions about their “risk for addiction” on the basis of these data is clearly unethical. My concerns, to be clear, are not based on some esoteric refiguring of what S4S reported, or a philosophical analysis of the meaning of polygenic scores: it is the plain outcome of what this researcher has reported. Have a look at our paper for details.
I’m not a bioetchics cop, that is someone else’s problem. And all I know about the details of this program is what is described on this web page. My deeper concern is with the fact that the theorizing of psychiatric genetics seems have come unmoored from their own data. This is supposed to be a scientific enterprise, but the field seems completely unable to acknowledge that the paradigm isn’t working. I get it: twin studies made it seem as though everything had a genetic basis, and once GWAS came along everyone thought it would just be a matter of time before scientists unraveled it. But it hasn’t happened, not for anything. Spit for Science was just the one program that made it explicit.
And, as we document in our paper, refusal to come to grips with your own results has consequences. The road to Gattaca is lined with overenthusiastic scientists who can’t admit when the tools they have developed are invalid. Now we have college students being given invalid information about their “genetic risk” for addiction based on a polygenic score that accounts for maybe 1% of the variance. Real harm will be done. They should stop.
I will start by mentioning that I am not a part of the above mentioned project (preregistered here: https://osf.io/b3swh/, ClincialTrials.gov ID: NCT06287203).
From what I understand this project involves using the work we did previously (https://www.nature.com/articles/s41380-022-01801-6), with the goal of understanding whether receiving feedback about one's risk (broadly conceived) has any impact on real world behavior change related to substance use.
Importantly, while the project website may be framed in the language of genetics, the preregistration notes that: "Our research team will combine the PGS with the behavioral/environmental data to generate risk estimates." So my guess is that the environmental and behavioral/clinical risk index will account for the lion's share of the variance (same as it did in the original paper). Should the focus on social and environmental risk factors be more apparent in the website? Probably. But the point about student's being given information about their "genetic risk" is inaccurate.
There a lot of valid criticisms to level at the field of psychiatric genetics, but those criticism should be about what is actually being done. That is not the case for the project listed here.