Article type
Year
Abstract
Background: Good methodological quality is necessary to reduce risk of bias in randomized controlled trials (RCTs) and in meta-analyses. As part of a review of clinical and methodological characteristics of malaria RCTs conducted in Africa, we assessed the methodological quality of trials conducted after the publication of the original CONSORT statement in 1996. This is a novel analysis that can highlight training needs for clinicians conducting trials in potentially resource-limited settings.
Objectives: To analyse the methodological quality of malaria RCTs conducted in Africa between 1997 and 2007.
Methods: We run systematic searches for malaria RCTs in electronic databases (Medline, Embase, the Cochrane Library), and applied an African geographic search filter to identify trials conducted in Africa. Results were exported to the statistical package STATA 8 to obtain a random sample from the overall data set. We evaluated 60 trial reports published between 1997 and 2007 for risk of bias according to 4domains (randomized sequence generation, allocation concealment, blinding, and loss to follow-up).
Results: Sequence generation was considered adequate (as done by using a random numbers table or electronically generated) in 35 reports, but was not clearly reported in 21trials (Table1). Many RCTs did not mention methods of allocation concealment or blinding of participants or intervention providers. In contrast, loss to follow up was accounted for in most RCTs (49 out of 60).
Conclusions: The quality of malaria trials' reports was not consistent among the 4domains analysed: a large proportion of RCTs had a high risk of bias for blinding and allocation concealment, whereas loss to follow-up was mostly well reported. Similar suboptimal reporting has been widely reported for RCTs in different healthcare areas, potentially affecting the validity of trial results and the estimates of treatment effects, and is not associated with trials conducted in resource-poor settings.
Objectives: To analyse the methodological quality of malaria RCTs conducted in Africa between 1997 and 2007.
Methods: We run systematic searches for malaria RCTs in electronic databases (Medline, Embase, the Cochrane Library), and applied an African geographic search filter to identify trials conducted in Africa. Results were exported to the statistical package STATA 8 to obtain a random sample from the overall data set. We evaluated 60 trial reports published between 1997 and 2007 for risk of bias according to 4domains (randomized sequence generation, allocation concealment, blinding, and loss to follow-up).
Results: Sequence generation was considered adequate (as done by using a random numbers table or electronically generated) in 35 reports, but was not clearly reported in 21trials (Table1). Many RCTs did not mention methods of allocation concealment or blinding of participants or intervention providers. In contrast, loss to follow up was accounted for in most RCTs (49 out of 60).
Conclusions: The quality of malaria trials' reports was not consistent among the 4domains analysed: a large proportion of RCTs had a high risk of bias for blinding and allocation concealment, whereas loss to follow-up was mostly well reported. Similar suboptimal reporting has been widely reported for RCTs in different healthcare areas, potentially affecting the validity of trial results and the estimates of treatment effects, and is not associated with trials conducted in resource-poor settings.