Checklist for Neurological Review Article

Tags: Poster
Cantisani T, Celani M, Cannata A, Cusi C, Filippini G, Munari L, Russo R, Taus C, Candelise LU

Introduction:

Objectives:

Methods:

Results:

Discussion: Neurological reviews of interventions can be a useful tool for neurologists in improving their clinical practice. Neurologists need reviews of a good methodological quality, excluding any significant bias, but also reviews that provide information relevant to their specific clinical work. With this in mind, an experimental database of all types of neurological reviews has been set up (Possible Cochrane Neurologic Network). The database includes 587 reviews selected between Jan 1996 and June 1997. The present study was carried out to identify a reliable method to assess the methodological and informative quality of neurological reviews. Secondary end points were a) to standardise a structural abstract and b) to evaluate the degree of agreement between different categories of neurologists. Three pairs of neurologists from different clinical backgrounds (trainees, consultants, senior consultants) are currently assessing 40 reviews, selected from the experimental database using the list of casual numbers. The authors' names, publication dates and the names of journals were deleted from copies of all the articles being assessed. Each reader, without receiving specific training, independently fills in a standardised checklist. A legend is provided to help complete each item properly. The checklist is divided into two parts. One part appraises the methodological quality according to Oxman and Guyatt, including nine single items regarding the identification and selection of trials, data synthesis and coherence between data and conclusions followed by an overall evaluation. The other measures the informative quality through seven items (e.g. type of intervention, number of trials included, inclusion of clinically important outcomes) and a overall evaluation of information. The informative part of the checklist had been previously validated by two independent readers. The items will be analysed either by the proportion of potential interjudge agreement beyond change (K index) and its 95% confidence limits or by the percentage of agreement between the two readers with its confidence limits, depending on the response required. The two overall evaluation items will be analysed by the Chi-square test. Reliability will be tested in each single couple and between the three different couples to determine the level of agreement.