Reproducibility of C-reactive protein analyses

The aim of this study was to measure the reliability of different nephelometric techniques for measuring C-reactive protein (CRP). One hundred and twenty samples were obtained from 40 patients. All 120 samples were divided in three parts to measure CRP using three different methods. Reliability was determined by the kappa index and intraclass correlation coefficient. The intraclass correlation coefficient ranged from 0.78 to 0.94. When CRP values were categorized in four groups, the kappa index reached 75-86% and percentage of agreement varied from 95% to 97%. When CRP values were divided into two groups, the kappa index was 73% to 78% and the percentage of agreement was 86% to 89%. We found that, CRP determinations with different nephelometric methods were highly reproducible, even when different analysts were involved. Ultrasensitive techniques are needed only if the clinical objective is to obtain a CRP measurement under 0.3 mg/dl.