• Vanessa Martins de Oliveira, Rafael Barberena Moraes, Airton Tetelbom Stein, and Eliana Márcia Wendland.
• Intensive Care Unit, Hospital de Clinicas de Porto Alegre, Ramiro Barcelos Street 2350, Porto Alegre, Brazil. Electronic address: vmoliveira@hcpa.edu.br.
• J Crit Care. 2017 Dec 1; 42: 129-137.

BackgroundThere is a need for a better understanding of the role of C-reactive protein (CRP) as a valid marker for the detection of bacterial infections in critically immunosuppressed patients. A high negative predictive value of CRP is also needed to rule out sepsis and bacterial infections in immunocompetent patients. However, few studies have evaluated the performance of CRP in immunocompromised hosts. The aim of the present study was to evaluate the performance of CRP as a marker of infection in critically immunosuppressed patients.MethodsThe inclusion criterion was immunosuppression for which CRP was used as a bacterial infection marker. Searches were performed in the Cochrane Register, MEDLINE, EMBASE, SCOPUS, Web OF Science, LILACS and CINAHL databases. We applied the Quality Assessment of Diagnostic Accuracy Studies tool 2 (QUADAS 2) to evaluate the quality of the articles and evaluated the test accuracy parameters using hierarchical summary receiver operating characteristic (HSROC) curves and bivariate random effect models.ResultsOnly 13 of 21 studies produced quantitative results. We analyzed all studies using the random effects method (restricted maximum likelihood) and obtained a joint diagnostic odds ratio (DOR) of 3.04 (95% confidence interval [CI] 1.71-5.40) with heterogeneity (I2=91%, Q=181.48, p<0.001). Therefore, a bivariate model was applied. Analyzing the tuberculosis carrier, steroid user, or presence of opportunistic infection subgroups, as described in the proposal, was not possible due to the lack of information on these topics included in the articles.ConclusionsCRP appears to be a good screening tool for sepsis in critically immunosuppressed patients. Submitted PROSPERO 2015: CRD42015019329.Copyright © 2017 Elsevier Inc. All rights reserved.

27 keywords...

hide…