No abstract

Background Outcome studies have reported the safety of computed tomographic pulmonary angiography (CTPA) as a stand-alone imaging technique to rule out pulmonary embolism (PE). Whether this can be applied to all clinical probabilities remains controversial. Objectives We performed a meta-analysis to determine the proportion of patients with venous thromboembolic events (VTE) despite a negative CTPA according to pretest PE prevalence. Methods We searched MEDLINE, EMBASE and the Cochrane Library (January 1990 to May 2017) for outcome studies recruiting patients with suspected PE using CTPA as a diagnostic strategy. The primary outcome was the cumulative occurrence of VTE at 3 months following a negative CTPA. Results Twenty-two different studies were identified. VTE was confirmed in 2.4% of patients (95% CI, 1.3-3.8%) either at the time of the index event or in the 3 months follow-up. Subgroup analyses suggested that the cumulative occurrence of VTE was related to pretest prevalence of PE, as VTE occurred in 1.8% (95% CI, 0.5-3.7%), 1.4% (95% CI, 0.7-2.3%), 1.0% (95% CI, 0.5-1.8%) and 8.1% (95% CI, 3.5-14.5%) of subgroups of patients with a PE prevalence < 20%, 20-29%, 30-39% and ≥ 40%, respectively. This was further confirmed using meta-regression analysis. Conclusions The negative predictive value of CTPA for VTE varies according to pretest prevalence of PE, and is likely to be insufficient to safely rule out VTE as a stand-alone diagnostic test amongst patients at the highest pretest probability of VTE. Prospective studies are required to validate the appropriate diagnostic algorithm for this subgroup of patients.