Clinical Decision Rules for Pulmonary Embolism in Hospitalized Patients: A Systematic Literature Review and Meta-analysis

Abstract

Background Clinical decision rules (CDRs) for pulmonary embolism (PE) have been validated in outpatients, but their performance in hospitalized patients is not well characterized. Objectives The goal of this systematic literature review was to assess the performance of CDRs for PE in hospitalized patients. Methods We performed a structured literature search using Medline, EMBASE and the Cochrane library for articles published on or before January 18, 2017. Two authors reviewed all titles, abstracts and full texts. We selected prospective studies of symptomatic hospitalized patients in which a CDR was used to estimate the likelihood of PE. The diagnosis of PE had to be confirmed using an accepted reference standard. Data on hospitalized patients were solicited from authors of studies in mixed populations of outpatients and hospitalized patients. Study characteristics, PE prevalence and CDR performance were extracted. The methodological quality of the studies was assessed using the QUADAS instrument. Results Twelve studies encompassing 3,942 hospitalized patients were included. Studies varied in methodology (randomized controlled trials and observational studies) and reference standards used. The pooled sensitivity of the modified Wells rule (cut-off ≤ 4) in hospitalized patients was 72.1% (95% confidence interval [CI], 63.7-79.2) and the pooled specificity was 62.2% (95% CI, 52.6-70.9). The modified Wells rule (cut-off ≤ 4) plus D-dimer testing had a pooled sensitivity 99.7% (95% CI, 96.7-100) and pooled specificity 10.8% (95% CI, 6.7-16.9). The efficiency (proportion of patients stratified into the 'PE unlikely' group) was 8.4% (95% CI, 4.1-16.5), and the failure rate (proportion of low likelihood patients who were diagnosed with PE during follow-up) was 0.1% (95% CI, 0-5.3). Conclusion In symptomatic hospitalized patients, use of the Wells rule plus D-dimer to rule out PE is safe, but allows very few patients to forgo imaging.