Box 1.

Resolution of Clinical Scenario

Armed with your new critical appraisal skills, you assess the results of the open tibial shaft NMA [7]. Included trials vary in their risk of bias (summarized, using the Cochrane risk of bias tool [modified] [11]). You note that, in general, the evidence base is at high risk of bias: only one of the trials being assessed is assessed as low risk of bias. Further, there are only two treatment comparisons that have multiple direct comparison trials. Heterogeneity is low where there are multiple comparisons; however, most of the direct comparisons are informed by only a single trial. The authors present forest plots for each comparison, along with absolute event rates, ORs, and 95% CIs. As you assess the results from direct, indirect, and combined network comparisons, you note there is incoherence in some of the comparisons (eg, reamed vs unreamed intramedullary nailing). However, you are able to discern the certainty of treatment effect owing to the authors’ use of GRADE assessments. Unfortunately, the highest ranking is moderate, and most are low. The authors present SUCRA score rankings in which unreamed intramedullary nailing is most likely to be the best treatment (SUCRA = 86%), reamed intramedullary nailing is ranked second (SUCRA = 54%), external fixation is third (SUCRA = 24%), and plating is most likely to be the worst treatment (SUCRA = 4%). The authors have excluded trials with very low confidence to avoid misleading rankings, although the low certainty of most estimates undermines the trustworthiness of the rankings. Based on the information, you conclude that evidence from randomized trials shows that your initial decision to use an unreamed intramedullary nail was likely the best treatment option for preventing reoperation in your patient with an open tibial shaft fracture. You recognize, however, that certainty in all the estimates is moderate or lower. Higher-quality evidence is needed and you are able to see where evidence is lacking.