Sir:FigureWe are writing in response to the article, “A Meta-Analysis of Human Acellular Dermis and Submuscular Tissue Expander Breast Reconstruction,” by Kim et al. (Plast Reconstr Surg. 2012;129:28–41).1 In this study, the authors performed two separate analyses. In the first, pooled estimates of complication rates for each procedure were calculated from 48 studies, 42 of which analyzed only one of the procedures. The second was a meta-analysis of six of these studies that compared both procedures. Kim et al. concluded that acellular dermal matrix reconstruction causes more complications, but their analyses do not support this conclusion. Although the meta-analysis showed significant differences for four of the six measured outcomes, the other analysis showed no significant differences between complication rates for any of the six outcomes, indicating that the studies included in the meta-analysis are not representative of all available data. The authors used uncontrolled cohort studies in the analysis, which has a high potential to introduce bias in the analysis, and an assessment of study quality was not conducted to address this. Likewise, sensitivity analyses were not conducted to address whether removing studies of lower quality, such as those with a lot of missing data, affected results. Although Kim et al. stated that they conducted a sensitivity analysis, they did not present the results. This is of particular concern given the significant amount of heterogeneity among the studies. Without results of sensitivity analyses to address this, it cannot be known whether pooling results was appropriate.2,3 Finally, although the authors present their search terms, they do not explicitly state the Boolean operators they used. The authors mention using the reference list of publications to manually identify 22 articles suitable for their analysis, but they do not report the details of this strategy. Moreover, MEDLINE was the only database used. These issues raise concern about the comprehensiveness of the search according to the established guidelines.4,5 Based on the foregoing limitations, results of this study and their interpretations should be viewed with caution. Kaveh Hajifathalian, M.D. David G. Zacharias Luis A. Gonzalez-Gonzalez, M.D. Julie Goodman, Ph.D. Departments of Biostatistics and Epidemiology, Harvard School of Public Health, Harvard University, Boston, Mass. DISCLOSURE The authors have no financial interest in any of the products or devices mentioned in this communication.
Read full abstract