The prognostic value of troponin release after adult cardiac surgery — a meta-analysis
Lurati Buse, Giovanna A. ; Koller, Michael T. ; Grapow, Martin ; Bolliger, Daniel ; Seeberger, Manfred ; Filipovic, Miodrag
In: European Journal of Cardio-Thoracic Surgery, 2010, vol. 37, no. 2, p. 399-406
Ajouter à la liste personnelle- Summary
- To assess the accuracy of increased troponin (Tn) concentrations for the prediction of mid-term (≥12 months) mortality after coronary artery bypass graft (CABG) and valve surgery, we performed a systematic review identifying all studies reporting on the association between postoperative troponin release and mortality after cardiac surgery. Studies were identified through 30 April 2008 by electronic searches of the MEDLINE, EMBASE and BIOSIS databases. Two reviewers independently selected studies, assessed methodological quality and extracted the data. We primarily considered mid-term (≥12 months) and secondarily short-term (≤30 days) all-cause mortality. A bivariate random-effects model was used to study determinants and to pool measures of prognostic accuracy of Tn. Seventeen studies fulfilled the inclusion criteria with a total of 237 mid-term deaths in 5189 patients and 296 short-term deaths in 9703 patients. The diagnostic odds ratio of increased Tn concentrations was 5.46 (95% confidence interval (CI) 2.0-14.6) for mid-term mortality and 6.57 (95% CI 4.3-10.1) for short-term mortality after adult cardiac surgery. Alternatively expressed, for troponin elevation, the sensitivity was 0.45 (0.26-0.67) and the specificity 0.87 (0.73-0.90) to predict mid-term mortality. The sensitivity was 0.59 (0.48-0.69) and the specificity 0.82 (0.72-0.89) for short-term mortality. Between-study variability was high. In conclusion, this meta-analysis provides evidence for an association between postoperative Tn release with mid- and short-term all-cause mortality after adult cardiac surgery. However, differences in populations, timing of Tn testing, Tn subunit and Tn assays make definitive conclusions about effect size and cut-off values difficult