Inter-reader reliability in multi-plane and multi-time point Hallux valgus correction evaluation on post-operative weight-bearing radiographs from a prospective multi-center trial and correlation with patient reported outcome measures.

Abstract

The primary purpose of this study was to determine interreader reliability (IRR) of hallux valgus (HV) related parameters on different time-points in dorsoplanar and coronal planes, i.e. hallux valgus angle (HVA), metatarsal rotation, sesamoid subluxation, metatarsal sesamoid osteoarthritis, tibial sesamoid point, and lateral round sign. The secondary purpose was to correlate these measurements with patient-reported outcome measures (PROMs) at the initial presentation for HV surgery and at 6 weeks, 4 months, 6 months, 12 months, and 24 months. Radiographic imaging and PROMs were collected at 5 different postoperative time points. Two musculoskeletal radiologists independently performed all readings and foot measurements blinded to each other's reads and the clinical information. Intraclass coefficient and kappa were obtained for interreader analysis. A partial spearman rank order was used to correlate radiographic foot measurements and PROMs. Across all 5 follow-up time points (6 weeks, 4 months, 6 months, 12 months, and 24 months), we found excellent IRR for HVA and poor reliability for lateral round sign, sesamoid subluxation, and metatarsal-sesamoid OA. We found excellent IRR for TSP at all follow-up time points except for at 6 weeks. There was a weak, positive correlation between sesamoid subluxation and PROMIS depression score at 24-month follow-up (R=0.21), and a negative correlation between metatarsal-sesamoid OA and PROMIS social score at 24-month follow-up (R=-0.21). We report excellent reproducibility for HVA in post-operative radiographs but variable- poor to excellent IRR for a range of HV-related parameters assessed on the axial view. There were no major trends in the correlation between the quantitative radiographic foot measurements and PROMs except for some weak correlations.