Associations between photograph-derived postural deviations and self-reported pelvic pain locations during pregnancy

Photograph-derived posture assessment may provide a low-burden way to describe musculoskeletal alignment during pregnancy, but its relationship with pelvic pain location is unclear. This exploratory retrospective observational study linked standardized early-pregnancy photographs with pre-randomization questionnaire data from 118 pregnant women. An automated photograph-based system classified 13 postural features into three ordinal levels. Outcomes were self-reported pain in the sacroiliac joint (SIJ) and symphysis pubis (SP) regions and the Oswestry Disability Index (ODI). Firth logistic and median quantile regression models adjusted for age, BMI, work position, gestational age at photography, and parity were used. Multiplicity was evaluated by Benjamini–Hochberg q values. Multivariable models were post-selection sensitivity analyses. SIJ‑region pain was reported by 62 women (52.5%) and SP‑region pain by 41 (34.7%). Across the 26 adjusted feature‑wise models, no posture‑outcome association survived Benjamini–Hochberg correction. Knee hyperextension was the most recurring nominal signal. It showed a nominal positive association with SP‑region pain in the primary feature‑wise model and a nominal inverse association with SIJ‑region pain in the post‑selection sensitivity analysis, while the exploratory multinomial analysis suggested that the pain‑location signal was mainly concentrated in the isolated SP‑region pain group (n = 19). In the PCA, the second principal component score was nominally associated with isolated SP‑region pain. No posture feature was associated with ODI in the 13 feature‑wise median quantile models. Given the small sample size, these findings should be interpreted as preliminary evidence. Larger prospective studies using validated pregnancy-specific posture measurements and clinically verified pain outcomes are needed.