Article

The primary objective was to produce a French-Canadian translation of
AMSTAR (a measurement tool to assess systematic reviews) and to examine the
validity of the translation's contents. The secondary and tertiary objectives
were to assess the inter-rater reliability and factorial construct validity of
this French-Canadian version of AMSTAR. Methods: A modified approach to
Vallerand's methodology (1989) for cross-cultural validation was used.1 First, a
parallel back-translation of AMSTAR2 was performed, by both professionals and
future professionals. Next, a first committee of experts (P1) examined the
translations to create a first draft of the French-Canadian version of the AMSTAR
tool. This draft was then evaluated and modified by a second committee of experts
(P2). Following that, 18 future professionals (master's students in
physiotherapy) rated this second draft of the instrument for clarity using a
seven-point scale (1: very clear; 7: very ambiguous). Lastly, the principal
co-investigators then reviewed the problematic elements and proposed final
changes. Four independent raters used this French-Canadian version of AMSTAR to
assess 20 systematic reviews that were published in French after the year 2000.
An intraclass correlation coefficient (ICC) and kappa coefficient were calculated
to measure the tool's inter-rater reliability. A Cronbach's alpha coefficient was
also calculated to measure internal consistency. In addition, factor analysis was
used to evaluate construct validity in order to determine the number of
dimensions. Results: The statements on the final version of the AMSTAR tool
received an average ambiguity rating of between 1.0 and 1.4. No statement
received an average rating below 1.4, which indicates a high level of clarity.
Inter-rater reliability (n=4) for the instrument's total score was moderate, with
an intraclass correlation coefficient of 0.61 (95% confidence interval [CI]:
0.29, 0.97). Inter-rater reliability for 82% of the individual items was good,
according to the kappa values obtained. Internal consistency was excellent, with
a Cronbach's alpha coefficient of 0.91 (95% CI: 0.83, 0.99). The French-Canadian
version of AMSTAR is a unidimensional tool, as confirmed by factor analysis and
community values greater than 0.30. Conclusion: A valid French-Canadian version
of AMSTAR was created using this rigorous five-step process. This version is
unidimensional, with moderate inter-rater reliability for the elements overall,
and with excellent internal consistency. This tool could be valuable to
French-Canadian professionals and researchers, and could also be of interest to
the international Francophone community.

Langue : FRANCAIS