Cross-cultural Adaptation and Validation of the Ankle Osteoarthritis Scale (AOS) for Use in French-speaking Populations

Power and sample size calculation The required number of patients has been estimated based on qualities to assess. The reproducibility (test-retest) of the questionnaire is one of the main properties and assess the lower acceptable limit is .85. However, knowing that the reproducibility of the English version was .97, the investigators estimated that the French version would be at least .90. A sample of 10 patients would be sufficient to obtain a power of 99% with a reproducibility of .901,9. Regarding validity, the investigators wish to obtain a statistically significant correlation with the three selected tests. The investigators hypothesized that the lowest correlation between the AOS-Fr and the SF-36 questionnaire would provide the lowest correlation, because the SF-36 is not specific to ankle OA. By estimating an average correlation of .50, 30 patients are required to obtain a significant correlation with a power of 80% and p < .051,9.

Translation The study was approved by a local ethics committee (protocol #09-130). Following the guidelines for the cross cultural adaptation process written by the American Association of Orthopaedic Surgeons (AAOS) in 2000, the translation was performed using this six-step process 2: 1) Initial translation; 2) Synthesis; 3) Back translation; 4) Expert committee; 5) Test of the Pre-final version; and 6) Submission of document to the developer. For the "Initial translation", two independent translators whose mother tongue was French translated the scale from English to French. The second step of "Synthesis" required these two translators to meet, discuss and compose a synthesised version of the AOS. Two independent translators, blind to the original scale, whose mother tongue was English, translated the synthesised version back to English. An "Expert committee" composed of a linguist, two orthopaedic surgeons and the four translators revised the whole process and consolidated the pre-final version. The "Test of the Pre-final version" was performed using 60 subjects, with no ankle osteoarthritis, who answered the questionnaire and evaluated their comprehension of each item. From the data collected in this step, some sections of the test were changed to improve comprehension and readability. For the final step the investigators submitted the corrected version to Dr. Charles Saltzman, the developer of the English version, who approved its translation and use (Supplementary data 1).

Evaluation of psychometric properties In order to validate the use of the French AOS version, the investigators set out to evaluate the psychometric properties of the test on ankle osteoarthritis patients using the same process as the original study validating the English version of the AOS 6. Patients with isolated ankle degenerative changes at the orthopaedic outpatient clinic of the Centre Hospitalier Universitaire de Sherbrooke (CHUS) were included. To be included, patients had to consider French as their mother tongue and be able to read and write in French. Patients under 18 years old and those with additional foot and/or ankle pathologies were excluded.

At the first visit, socio-demographic data (age, sex, BMI, occupation) and pertinent past medical history (Diabetes, neuropathy, ankle or foot fractures)) were collected and patients were asked to perform single heel lift tests of both affected and unaffected side. The participants completed three questionnaires: AOS-French version, SF-36, and WOMAC. Criterion validity was assessed by comparing the AOS to the WOMAC and the SF-36 scores. Construct validity was established by the examination of the correlation between the AOS scores and the single heel lift test. Finally, to measure test-retest reliability, the patient was asked to complete the AOS a second time and a small questionnaire detailing any modification to the treatment (shoes, orthotics, medication, surgery, injection) one week following the first visit and return it by mail. Any modification to their treatment between the completion of the two AOS questionnaires would nullify the test-retest reliability, so these patients would be excluded in the analysis.