Recovery Capital Predicts Remission at Almost Every Level of Genetic Risk. The Exception Is the Interesting Part.

The clinical intuition that a stable job, a bed and a working phone matter as much as a family history now has a number attached — one that holds up across nearly the whole range of inherited risk. Among 7,257 U.S. adults treated for substance use problems, moving from the 10th to the 90th percentile of a recovery capital index raised the projected probability of 12-month remission by 7.9 percentage points in people with low polygenic risk and by 10.6 points in those at median risk.

Among people at high polygenic risk, the same jump in recovery resources bought 6.4 points — with a confidence interval that crossed zero.

The study, published online in Addiction in August and publicized by the University of Illinois Urbana-Champaign this week, is among the first to put social resources and a genome-wide risk score into the same model of remission. Janaka Kosgolla, a research scientist at Illinois's School of Social Work, and center director Douglas Smith drew on the NIH All of Us Research Program, following adults treated for non-nicotine substance use problems between December 2020 and October 2023.

What the Index Measured

Recovery capital is a term of art for the sum of what a person has to work with: housing, employment, transport, mental health, and at the community level enough treatment slots and support meetings. Kosgolla and Smith built a 21-item index from All of Us survey responses, scored from 0 to 1, spanning human, physical, social and cultural resources.

Against it they set a polygenic risk score built from roughly 585,000 genetic variants, standardized and adjusted for ten ancestry principal components. Remission came from electronic health records rather than self-report — a month counted if it carried a remission diagnostic qualifier or a negative toxicology result, with episodes as contiguous runs. Of the sample, 2,413 people contributed at least one remission episode, 5,124 in total; alcohol use disorder accounted for 33.6 percent of episodes and opioid use disorder for 38 percent, while cannabis accounted for 1.9 percent.

The analysis weighted relapse hazards for treatment and censoring, then bootstrapped 12-month remission estimates for each combination of the two indices.

Where the Advantage Thinned

The relationship was not a simple gradient. Remission probability was lowest among people in the low-to-middle range of recovery capital who also sat in the middle-to-high range of genetic risk, and highest among those with high recovery capital and low genetic risk. The authors call this non-monotonic. It implies something practical: the least favorable prognosis belongs not to the worst genes but to modest resources combined with elevated risk — a population screening could actually identify.

Smith's reading is deliberately hopeful. People at high genetic risk with a family history of addiction, he said, can and do go into remission given the right resources. The paper's own conclusion is more measured, noting that the contrasts are modest and sensitive to assumptions about censoring.

Three cautions travel with it. The design is observational, so recovery capital may partly proxy for severity rather than cause better outcomes. Remission was measured through records, which rewards people who stay in contact with health systems — itself a form of recovery capital. The cohort also skews toward those who sought treatment. For clinicians the takeaway is narrower: housing, social support and daily structure deserve the same documentation as a family history, particularly for patients carrying several vulnerabilities at once, and integrated dual diagnosis care is where those resources are most often assembled. The genetic score, on this evidence, adds modest information on top of what a good social assessment already captures.