Sir: Thank you for the opportunity to respond to the comments of Dr. Liu et al. They reference a press release citing a study published in The Journal of the American Medical Association using a national registry of Denmark. The actual study evaluates the bleeding risk of nonsteroidal antiinflammatory drugs in post–myocardial infarction patients on formal systemic anticoagulation, including clopidogrel and warfarin sodium.1 In this anticoagulated population, there is an increased risk of bleeding (gastrointestinal, intracranial, or symptomatic anemia) with nonsteroidal antiinflammatory drugs. However, this study bears little resemblance to our meta-analysis, which evaluates surgical patients on no other anticoagulation. It is well known that nonsteroidal antiinflammatory drugs can interfere with platelet aggregation, but there is no evidence that suggests nonsteroidal antiinflammatory drugs lyse previously formed clot. In a surgical setting with adequate intraoperative hemostasis, nonsteroidal antiinflammatory drug use seems to be relatively safe.2 Dr. Liu and colleagues further note that funnel plots are less useful when fewer studies are included. This, unfortunately, is a common problem in analyses restricted to the plastic surgery literature because there are few randomized controlled trials. We included the plot for completeness, but acknowledge the limits of the bias analysis.3 Regarding the number of included studies, it is noteworthy that we intentionally restricted inclusion to studies that would be sensitive to postoperative bleeding (small, soft-tissue surgical sites). The attrition analysis demonstrates that we screened 881 total primary studies, including 29 full-text randomized controlled trials, before narrowing to the four studies that were eventually selected. To reply to their third note, we agree that homogeneity is critical to meta-analyses. We limited the population to surgical patients who received ibuprofen postoperatively for pain management. This created a generally homogenous population, but some heterogeneity exists because this is not restricted to a single operation. Thus, the population size of individual studies could skew toward the effect in that operative population. The largest primary study (n = 170) concerned reconstruction of Mohs defects. The other studies included herniorrhaphy, breast surgery, and facial aesthetic surgery (combined, n = 117). However, the results may still be considered useful with caution. In addition, the individual studies each found no difference in bleeding incidence between ibuprofen and control, which is concordant with our findings. Their fourth point concerns use of the Jadad scale, which we included in addition to other forms of bias analysis. Although the authors are correct to note that this is not the most up-to-date system for bias analysis, it was included as an adjunct for completeness. We also performed a Grading of Recommendations Assessment, Development and Evaluation analysis. The Grading of Recommendations Assessment, Development and Evaluation guidelines, with which the authors do not seem to be familiar, is a process for rating the quality of scientific evidence. Figure 2 is a graphic representation using the RevMan software suite from the Cochrane Collaboration.4 Grading of Recommendations Assessment, Development and Evaluation analysis is a component of the software package. Further discussion of the use of the Grading of Recommendations Assessment, Development and Evaluation guidelines can be reviewed in the plastic surgery literature.5,6 DISCLOSURE None of the authors has a financial interest in any of the products, devices, or drugs mentioned in this communication. Brian P. Kelley, M.D.Jeffrey H. Kozlow, M.D., M.S.Section of Plastic and Reconstructive SurgeryUniversity of Michigan Health SystemAnn Arbor, Mich.
Read full abstract
Feb 26, 2018
Open Access