Mega-analysis of meta-analysis: an examination of meta-analysis with an emphasis on once-daily aminoglycoside comparative trials

Meta-analysis is increasingly performed to try and answer clinical questions around which controversy exists. We performed meta-analyses using three different methods as examples to illustrate problems and points to consider when doing such work. The questions of interest were comparing frequency of nephrotoxicity and clinical effectiveness with once-daily aminoglycoside regimens versus multiple-daily dose aminoglycoside regimens. Inclusion and exclusion criteria were established. Studies were obtained from literature searches, and included published studies and abstracts. The overall odds ratio (95% confidence interval) for nephrotoxicity and clinical effectiveness meta-analyses of these studies using the Mantel-Haenszel-Peto method were 0.70 (0.51-0.94, p = 0.042) and 0.79 (0.59-1.07, p = 0.09), respectively. Analyzing the data by Fisher's combined test and the Mean-P method showed agreement with the other methods' conclusions for nephrotoxicity but not for clinical effectiveness. Meta-analysis is not an exact procedure and contains various problems and inconsistencies. Combining studies of poor quality is not ideal for answering unresolved clinical questions.