Risk-Taking Fluctuations Predict Next-Month Drinking Rises in Mild AUD, One-Year Smartphone Study Shows

Researchers deployed four smartphone tasks every four weeks and drinking questionnaires every two days over one year, yielding an average 7.4 assessments per participant. Increases in risk-taking under mixed gambles and reduced information sampling bias each predicted higher subsequent consumption; response inhibition, working memory, and reward-driven risk-taking showed no such links after controlling for prior drinking levels.

This temporal specificity challenges prevailing models that treat cognitive deficits mainly as consequences of heavy use. It aligns with prior observational data from the NESARC-III cohort showing decision-making traits precede escalation in sub-clinical drinkers and extends cross-sectional findings in Biological Psychiatry that linked AUD severity to loss-context biases without establishing sequence.

The design cannot rule out unmeasured third variables such as sleep or stress that might jointly drive both risk-taking and intake. Next steps require RCTs testing whether real-time task-based alerts can blunt consumption spikes by at least 15 % over six months in independent samples.

Evidence quality is moderate: the within-person longitudinal design supports directional inference better than cross-sectional work yet remains observational and relies on self-reported drinking without biomarker validation.