Article type
Year
Abstract
Background: Adherence to good methodological quality is necessary to minimise bias in randomised controlled trials (RCT). Previous studies have shown the importance of certain characteristics on trial quality, but we are not aware of any studies of the association between methodological quality and characteristics specific to HIV/AIDS trials or African trials. We speculated that location might have a negative impact on a trial's conduct and quality in regions where resources and training are scarce.
Objectives: 1) To identify and describe the quality of all HIV/AIDS RCTs conducted wholly or partly in Africa, reported before 2004; 2) to identify a random sample of HIV/AIDS RCTs conducted wholly or partly in North America before 2004; and 3) to compare the quality of trials in these two sets and to assess whether location is predictive of trial quality.
Methods: African trials were identified from electronic bibliographic databases (MEDLINE, EMBASE, CENTRAL, LILACS), electronic conference databases (AIDSearch, NLM Gateway), and handsearching of African medical journals not indexed on MEDLINE. After determining eligibility, we extracted information on the characteristics of each trial, including the quality domains of allocation concealment, generation of the random sequence, and masking of the assessor. North American trials were identified from searching the same electronic bibliographic databases (excluding LILACS) and a random sample was selected for the comparison. Univariate and multivariate logistic regression analyses were conducted to evaluate the association between location (Africa vs North America) and allocation concealment, generation of the random sequence, and masking of the assessor.
Results: We identified 99 African trials (80 from electronic bibliographic databases, 18 from conference databases and one from handsearching). Trials were conducted in 20 sub-Saharan African countries, with 51 treatment and 48 prevention trials. Most studies were funded by government agencies based outside Africa (50). The pharmaceutical industry sponsored 49 trials. Based on the reports, allocation concealment was judged to be adequate in 51% of trials, generation of the sequence was adequate in 41% and assessors were masked in 33%. The random sample of North American trials included 114 trials, out of a total of 785 eligible studies. Based on the reports for these 114 trials, allocation concealment was adequate in 22%, generation of the sequence was adequate in 20% and assessors were masked in 26%. Multivariate logistic regression models of the 190 trials identified from the electronic bibliographic databases showed that allocation concealment and generation were significantly associated with location, after adjusting for other trial characteristics (intervention type, number of centres, presence of a CONSORT flow diagram, clear definition of a primary outcome, approval by a local ethics committee, type of randomization, US government or pharmaceutical funding, sample size and the year of the primary report). African trials were three times as likely to report adequate allocation concealment (OR = 3.24; 95%CI: 1.59 to 6.59; p = 0.001) and twice as likely to report adequate generation of the sequence (OR = 2.36; 95%CI: 1.20 to 4.67; p = 0.013). We did not find an association between location and masking of the assessors.
Conclusion: The higher quality of African trials is unexpected. As most African trials are externally funded, it is possible that stricter requirements of funding agencies when leading trials in other countries, and greater experience and training of principal investigators of an international stature, may account for this difference. Our findings show that trial location should be considered when evaluating trial quality in systematic reviews. Further research is needed to establish whether our findings are replicated in other regions and in other healthcare areas.
Objectives: 1) To identify and describe the quality of all HIV/AIDS RCTs conducted wholly or partly in Africa, reported before 2004; 2) to identify a random sample of HIV/AIDS RCTs conducted wholly or partly in North America before 2004; and 3) to compare the quality of trials in these two sets and to assess whether location is predictive of trial quality.
Methods: African trials were identified from electronic bibliographic databases (MEDLINE, EMBASE, CENTRAL, LILACS), electronic conference databases (AIDSearch, NLM Gateway), and handsearching of African medical journals not indexed on MEDLINE. After determining eligibility, we extracted information on the characteristics of each trial, including the quality domains of allocation concealment, generation of the random sequence, and masking of the assessor. North American trials were identified from searching the same electronic bibliographic databases (excluding LILACS) and a random sample was selected for the comparison. Univariate and multivariate logistic regression analyses were conducted to evaluate the association between location (Africa vs North America) and allocation concealment, generation of the random sequence, and masking of the assessor.
Results: We identified 99 African trials (80 from electronic bibliographic databases, 18 from conference databases and one from handsearching). Trials were conducted in 20 sub-Saharan African countries, with 51 treatment and 48 prevention trials. Most studies were funded by government agencies based outside Africa (50). The pharmaceutical industry sponsored 49 trials. Based on the reports, allocation concealment was judged to be adequate in 51% of trials, generation of the sequence was adequate in 41% and assessors were masked in 33%. The random sample of North American trials included 114 trials, out of a total of 785 eligible studies. Based on the reports for these 114 trials, allocation concealment was adequate in 22%, generation of the sequence was adequate in 20% and assessors were masked in 26%. Multivariate logistic regression models of the 190 trials identified from the electronic bibliographic databases showed that allocation concealment and generation were significantly associated with location, after adjusting for other trial characteristics (intervention type, number of centres, presence of a CONSORT flow diagram, clear definition of a primary outcome, approval by a local ethics committee, type of randomization, US government or pharmaceutical funding, sample size and the year of the primary report). African trials were three times as likely to report adequate allocation concealment (OR = 3.24; 95%CI: 1.59 to 6.59; p = 0.001) and twice as likely to report adequate generation of the sequence (OR = 2.36; 95%CI: 1.20 to 4.67; p = 0.013). We did not find an association between location and masking of the assessors.
Conclusion: The higher quality of African trials is unexpected. As most African trials are externally funded, it is possible that stricter requirements of funding agencies when leading trials in other countries, and greater experience and training of principal investigators of an international stature, may account for this difference. Our findings show that trial location should be considered when evaluating trial quality in systematic reviews. Further research is needed to establish whether our findings are replicated in other regions and in other healthcare areas.