Article

Systematic reviews are used to inform practice, and develop guidelines and protocols.
A questionnaire to quantify the risk of bias in systematic reviews,
the review paper assessment (RPA) tool, was developed and tested. A search of
electronic databases provided a data set of review articles that were then
independently reviewed by two assessors using the RPA. The inter-rater
reliability was between moderate and good (kappa scores 0.46-0.95). Many reviews
did not describe the purpose in terms of population, intervention, comparator and
outcome measure (i.e. PICO format), making inter-rater agreement on this question
difficult. The RPA discriminated between high-quality reviews and those with a
risk of bias (e.g. inadequate reporting of search terms, lack of independent
reviewing or inclusion of non-randomized-controlled trials).
The RPA
questionnaire was revised to ensure that questions (on the basis of clarity of
purpose, extent of search, independence of reviewers, randomized-controlled trial
inclusion and availability of data) had dichotomous answers so that the positive
responses scored one. The risk of bias increases as the score reduces.

Langue : ANGLAIS