Validity and reliability of the Manchester Oxford Foot Questionnaire (MOXFQ) in one-year postoperative ankle fracture...

Ankle fracture patients are a heterogenous group with differences in age, sex, fracture morphology, and treatment provided. With the increased focus on patient-centered treatment, patient-reported outcome measures (PROMs) are increasingly adopted by clinicians to facilitate best clinical practice. The Manchester Oxford Foot Questionnaire (MOXFQ) has demonstrated good measurement properties when used in patients with foot or ankle disease. The PROM has three domains: (1) Pain; (2) Walking/Standing; and (3) Social Interaction. One study found sufficient content validity for the Pain and Walking/Standing domains when used in the evaluation of ankle fracture patients. Another validation study demonstrated acceptable structural validity and reliability for the MOXFQ in ankle fracture patients 12 weeks after injury. The aim of this study is to assess the structural validity and reliability of the Norwegian version of the MOXFQ in the context of an ankle fracture patients one year after surgery and provide patient acceptable symptom state (PASS) estimates.

A pragmatic cross-sectional study design was used to collect the one-year MOXFQ follow-up data from patients surgically treated for an ankle fracture in the period 2017 to 2020 at (Stavanger University Hospital). The structural validity and internal consistency were assessed using confirmatory factor analysis. A separate test-retest study including patients at least one year since ankle surgery was used in the assessment of reliability and measurement error.

A confirmatory factor analysis of the three-factor model of the MOXFQ had a good model fit (TLI 0.94; CFI 0.95; RMSEA 0.094; SRMR 0.039). However, the measurement model demonstrated poor discriminant validity of the three factors. A unidimensional model of the 16 items had worse model fit, while a second-order factor model demonstrated strong factor loadings for a second-order factor. A bi-factor model also revealed a strong general factor but also unique variance in the Pain and Social Interaction domain. The domains had good internal consistency (McDonald's omega 0.80 to 0.95) and test-retest reliability (ICC 0.80 to 0.92). The standard errors of measurements for the three domains were between 6.5 and 7.5, and 5.5 for the MOXFQ-Index (scale 0 to 100). PASS estimates for the (sub)scales were: Pain 45; Walking/Standing 39; Social Interaction 19; and MOXFQ-Index 34.

The MOXFQ with three domains demonstrated sufficient structural validity and reliability when used in the evaluation of a one-year postoperative ankle fracture population. Reporting the scores of the Pain and Walking/Standing domains was best supported.