Age and sex disparities in the prevalence of common lumbar MRI abnormalities: a hospital-based cross-sectional study of 1381 cases

Common lumbar MRI abnormalities have largely been studied one lesion at a time, in different populations, with the literature dominated by single-lesion series, asymptomatic cohorts and Western populations. What is missing is a simultaneous, age- and sex-stratified description of all of them within one cohort, allowing their relative frequencies and co-occurrence to be compared directly. We retrospectively analysed 1,381 patients aged 20–99 years undergoing lumbar MRI at a single tertiary hospital in 2025; those with prior lumbar surgery or incomplete records were excluded. Ten abnormalities were coded as binary variables from routine clinical reports issued under two-tier senior review, using NASS/ASSR/ASNR-based criteria. Prevalence is reported with Wilson 95% CIs. Age and sex associations were quantified by Cramér’s V and φ, with age-adjusted odds ratios (ORs) and adjusted prevalence ratios (PRs) from logistic and modified Poisson regression. The 45 pairwise comparisons were FDR-corrected at 5% (Benjamini–Hochberg). Degenerative findings were the norm: lumbar degenerative disease(LDD, 73.1%), lumbar disc herniation(LDH, 62.5%), lumbar disc bulge(LDB, 55.6%); lumbar disc extrusion (LDE, 3.8%) and suspicious haemangioma (SH, 3.5%) were least common. Age association varied ten-fold (V 0.074–0.759), strong for LDD(15.2% to 97.9%) and moderate for LDB, vertebral compression fracture (VCF) and fasciitis, but weak for Modic changes, lumbar spinal stenosis(LSS), Schmorl’s nodes, SH and LDH, and non-significant for LDE. Sex differences were small throughout (all |φ|<0.1), clinically meaningful only for LDE (adjusted OR 2.22, 95% CI 1.25–3.94) and LSS (1.78, 1.27–2.51). Of 45 pairwise comparisons, 13 survived adjustment and FDR correction: eight formed a positive disc–endplate–canal network with Modic changes as hub (strongest: LDE–LSS, adjusted OR 9.91), and five were inverse, all involving VCF (LDB–VCF OR 0.15). Adjustment reversed the crude VCF–LDD association and removed all crude associations of fasciitis and SH. Lumbar MRI abnormalities should not be treated as a single age-driven entity, only LDD, LDB, VCF and fasciitis are strongly age-dependent, LDH is common at all ages, and sex differences are largely confined to LDE and LSS. Co-occurrence is non-random: a disc–endplate–canal network dissociated from a compression-fracture phenotype. These estimates describe imaging findings in a referred population, not symptomatic disease. Not applicable.