Drinking to Cope or Coping to Drink? Behavioral Profiles of Stress Management and Alcohol Use Risk Among Medical Students: A Cross-Sectional Study.

Background/Objectives: Alcohol misuse among medical students is commonly attributed to academic stress, yet the specific role of coping mechanisms in this relationship has received limited attention. We investigated whether substance use coping, rather than stress exposure itself, drives alcohol use risk in Romanian medical students, and whether distinct coping-based subgroups can be identified through cluster analysis. Methods: We conducted a cross-sectional survey among 244 medical students (mean age 21.95 ± 3.27 years; 67.2% female) at the University of Medicine and Pharmacy of Craiova, Romania. Alcohol use was measured with the AUDIT and coping strategies with the Brief COPE. Analyses included Mann-Whitney U tests, Spearman correlations, multiple linear and binary logistic regression, and k-means clustering. Results: At-risk drinking (AUDIT ≥ 8) was identified in 19.7% of participants. The tendency to use substances to cope with stress (substance use coping) was the strongest predictor of AUDIT scores in both linear regression (B = 2.090, p < 0.001, R2 = 0.513) and logistic regression (OR = 2.026, p < 0.001). Male sex independently predicted at-risk status (OR = 2.572, p = 0.025), while planning was protective in both models (B = -0.657, p = 0.005; OR = 0.691, p = 0.029). Humor also emerged as a significant risk factor (OR = 1.638, p = 0.005). K-means analysis (k = 5) revealed five coping profiles with significantly different AUDIT distributions (Kruskal-Wallis H = 47.26, p < 0.001). The Substance-Oriented cluster (13.1% of students) had a mean AUDIT of 12.66, compared with 3.00-4.13 in other clusters. Conclusions: In a subgroup of medical students, alcohol use appears integrated into the coping repertoire rather than merely being a consequence of stress. The identified coping profiles should be interpreted as prototypical configurations with overlapping boundaries rather than discrete categorical types, given the low silhouette coefficient (0.094) of the cluster solution. The strong predictive effect of substance use coping should be interpreted with the caveat that the Brief COPE Substance Use subscale and the AUDIT share content related to alcohol use behavior, which may inflate the observed association. These findings point to the need for coping-specific interventions. Planning skills training and a more nuanced understanding of humor's role in drinking contexts may offer avenues for prevention. However, the logistic model's sensitivity of 50.0% indicates that coping-based identification alone would miss approximately half of at-risk students, underscoring the need for further refinement before clinical application.