Sixteen studies were included in the review. Seven studies (n=278 patients) assessed the diagnosis of right ventricular dysfunction using brain natriuretic peptide or N-terminal pro-brain natriuretic peptide. All of the studies enrolled consecutive patients and described the inclusion/exclusion criteria, reference standard, definition of right ventricular dysfunction and participant characteristics. Five studies reported blinding.
Diagnostic accuracy for right ventricular dysfunction:
The diagnostic odds ratio for studies evaluating brain natriuretic peptide, using a reference standard of echocardiography, ranged from 28.44 to 293.91. The pooled diagnostic odds ratio was 39.45 (95% CI 15.54 to 100.12; four studies), with no evidence of statistically significant heterogeneity. The area under the summary receiver operating characteristic curve was 0.91; a perfect diagnostic test would have a value of 1.00, whilst a test with no value would have a value of 0.50.
The diagnostic odds ratios for the two studies evaluating N-terminal pro-brain natriuretic peptide, using a reference standard of echocardiography, ranged from 8.32 and 96. The pooled diagnostic odds ratio was 24.73 (95% CI 2.02 to 302.37), with statistically significant heterogeneity (p=0.043).
The single study using computed tomography as the reference standard found a diagnostic odds ratio for brain natriuretic peptide of 6.56 (95% CI 1.92 to 22.39) and 5.67 (95% CI 1.69 to 18.91) for N-terminal pro-brain natriuretic peptide.
Prognostic value for patients with pulmonary embolism:
Fourteen studies assessed the prognostic value of brain natriuretic peptide or N-terminal pro-brain natriuretic peptide (n=1,045 patients). The risk of death was increased in patients with an elevated brain natriuretic peptide, using a cut-off of 100pg/ml; the pooled odds ratio for all-cause in-hospital mortality was 6 (95% CI 1.31 to 27.43) compared to patients with a normal level.
The risk of death was also increased in patients with an elevated N-terminal pro-brain natriuretic peptide, using a cut-off of 1,000ng/L, the pooled odd ratio for all-cause in-hospital mortality was 7.34 (95% CI 1.85 to 29.03), compared with patients with normal levels. With a cut-off of 600ng/L, the pooled odds ratio was 16.12 (95% CI 3.1 to 83.68). There was no statistically significant heterogeneity in any analysis.
Results of subgroup analyses were also reported.