Fig. 4: Antibiotic resistance gene load predicts long-term mortality risk.

a Total ARG load predicts long-term mortality risk in a 17-year follow-up (N = 7095 participants, probabilistic multivariate Cox proportional hazards model; See Supplementary Table 6). The points indicate the median, while the bars represent the 95% credible interval (CI). The model is adjusted for Enterobacteriaceae relative (log10) abundance, age, smoking, sex, diabetes, antineoplastic and immunomodulating agents, body mass index, self-reported antihypertensive medication, systolic blood pressure, recent antibiotics use (six months before baseline), household income, and raw vegetable and salad consumption. The median hazard ratio (HR) is shown per unit increase of each variable, along with the 95% CIs; variables whose CI overlaps with 1 (no association) are excluded from the graph. b Cumulative incidence of all-cause mortality during the follow-up period for individuals stratified by high (red; >458 RPKM) and conventional (blue) ARG load. The shaded area represents the 95% CI. High-ARG load is associated with significantly higher mortality among women (P = 0.006; log-rank test; multivariate Cox). A similar but non-significant trend is observed in men (P = 0.06). Associations of ARG load with sepsis and cause-specific mortality are shown in Supplementary Fig. 8. Here, in a representative cohort of 7095 Finnish adults, the authors reveal that gut antibiotic resistance is shaped not only by antibiotic use but also by the microbiome, diet, lifestyle, household income—and is linked to higher long-term mortality.