HSC Section 3 - Trauma, Critical Care and Sleep Medicine

M OUBAYED ET AL .

FIGURE 2. (Continued). [Color figure can be viewed at wileyonlinelibrary.com]

the incidence of VTEs for all studies is presented in Fig- ure 2A, with 95% CI for each study. The overall inci- dence is 0.4% (95% CI, 0.3% to 0.55%; I 2 , 99%). A funnel plot and Duval and Tweedie’s trim and fill (Figure 2B) did not demonstrate any publication bias and, thus, would not have changed our result. A classic Fail Safe N demonstrated that 3791 missing studies would be required to significantly change our findings. The incidence of bleeding complications is presented in a forest plot in Figure 2C, and showed an overall inci- dence of 0.9% (95% CI, 0.6% to 1.2%; I 2 , 98%). A fun- nel plot and Duval and Tweedie’s trim and fill (Figure 2D) did not demonstrate any publication bias and, thus, would not have changed our result. A classic Fail Safe N demonstrated that 6118 missing studies would be required to significantly change our findings. We a priori expected a high degree of heterogeneity, given that we were summarizing the incidence of VTEs (Figure 3A) and bleeding complications in OHNS across different populations and in different healthcare settings.

Therefore, we performed the analysis using a number of moderators as subgroup analyses to assess whether this might explain some of the heterogeneity (study type, OHNS subgroup, and observation period for outcome). The OHNS subgroup did significantly impact the I 2 , whereas the remaining features did not explain the heterogeneity. Despite decreasing the I 2 , the point estimates and CIs did not significantly change and are, therefore, not reported. Similarly, meta-regression on publication year did not change our results nor did it explain the heterogeneity. Effect of chemoprophylaxis Figure 3 shows a forest plot of studies that reported the incidence of VTEs (Figure 3B) and bleeding (Figure 3B) comparing OHNS patients receiving chemoprophylaxis or not. The forest plots included 3 studies. Bahl et al 6 reported that 19% of patients received intraoperative ketorolac, and 37.6% of patients had sequential compres- sive devices (sequential compressive devices). The

HEAD & NECK—DOI 10.1002/HED JUNE 2017

168