Methods used to assess the performance of biomarkers for the diagnosis of acute kidney injury: a systematic review and meta-analysis

Methods used to explore biomarkers for acute kidney injury (AKI) might have a major impact on the results and the use of these biomarkers. We evaluated the methods used to investigate biomarkers of AKI.A systematic review and meta-analysis were performed using a computerized search of the MEDLINE and the EMBASE databases (PROSPERO CRD42017059618). Articles reporting biomarker's performance to diagnose AKI were included. The outcome included a description of the methods used to assess the performance of biomarkers to diagnose AKI.Among the 295 included studies, assessment of biomarkers was the primary endpoint in 284 with sample size calculation in only 8% of cases. Eighty-five percent of the studies summarized the performance of biomarkers with receiver operating characteristic (ROC) curves; however, 74 studies (25%) did not provide the threshold, sensibility or specificity. A total of 176 studies evaluated more than one biomarker, and only 25% combined biomarkers to increase diagnostic performance. We determined that the definition of AKI and study design impacted the diagnostic performance using uNGAL (urinary neutrophil gelatinase-associated lipocalin) as an example. Major publication bias was identified.Most articles that reported biomarkers of AKI performance present methodological weaknesses. Basic rules should be provided to increase the quality of reporting in this area.