Dupont C, Rodenbach J, Flachaire E
Service de MPR, hôpital Renée-Sabran, boulevard Edouard-Herriot, Giens, 83400 Hyères, France.
Ann Readapt Med Phys. 2008 Jun;51(5):348-57. doi: 10.1016/j.annrmp.2008.01.014. Epub 2008 May 27.
Determination of the utility of C-reactive protein (CRP) levels when measured 21 days after hip and knee arthroplasties for early diagnosis of infectious complications.
This study was performed in two parts: establishment of a reference curve by measurement of CRP levels once a week in a cohort of 94 patients (50 total hip arthroplasties and 44 total or unicondylar knee arthroplasties); study of the diagnostic value of two different CRP cut-offs (25mg/l, the mean CRP level and two standard deviations; 18 mg/l, mean and one standard deviation) at D21 postoperative in a population of 48 patients, of whom 12 presented septic complications (four surgical site infections [SSIs] and eight intercurrent infections).
We observed very high interindividual variations in CRP values two weeks after arthroplasty. These variations decreased strongly in the third week postoperative. In the seven patients with a CRP level above 25mg/l at D21, there were no false-positives. In the 41 patients with a CRP level below 25mg/l at D21, there were five false-negatives and no false-positives. With the CRP threshold set at 18 mg/l, we observed four false-positives and four false-negatives.
DISCUSSION-CONCLUSION: A CRP level threshold of 25mg/l is not sufficiently reliable for early detection of postoperative infections (whether at the surgical site or elsewhere), as judged by a sensitivity of 58.3% and a negative predictive value of 87.8%. However, the 25mg/l threshold displays first-rate specificity and positive predictive values (both 100%). A CRP threshold at 18 mg/l is no better because even though it yields slightly a higher sensitivity value (66.7%), it strongly decreases specificity (88.9%). CRP is an important tool for postoperative monitoring but often appears to be difficult to use. The diagnosis of septic complications is based on clinical and paraclinical arguments. Local discharge, fever over 38 degrees C and local/persistent pain and stiffness are more informative indicators of postoperative infection.
测定髋关节和膝关节置换术后21天测量C反应蛋白（CRP）水平对感染并发症早期诊断的效用。
本研究分两部分进行：在94例患者（50例全髋关节置换术和44例全膝关节或单髁膝关节置换术）队列中每周测量一次CRP水平以建立参考曲线；在48例患者群体中研究术后第21天两种不同CRP临界值（25mg/l，即CRP平均水平加两个标准差；18mg/l，即平均水平加一个标准差）的诊断价值，其中12例出现败血症并发症（4例手术部位感染[SSI]和8例并发感染）。
我们观察到置换术后两周CRP值个体间差异非常大。这些差异在术后第三周大幅下降。在术后第21天CRP水平高于25mg/l的7例患者中，无假阳性。在术后第21天CRP水平低于25mg/l的41例患者中，有5例假阴性且无假阳性。将CRP阈值设定为18mg/l时，我们观察到4例假阳性和4例假阴性。
讨论 - 结论：以58.3%的灵敏度和87.8%的阴性预测值判断，25mg/l的CRP水平阈值对术后感染（无论是手术部位还是其他部位）的早期检测不够可靠。然而，25mg/l的阈值显示出一流的特异性和阳性预测值（均为100%）。18mg/l的CRP阈值也好不到哪里去，因为尽管它产生的灵敏度值略高（66.7%），但其特异性大幅下降（88.9%）。CRP是术后监测的重要工具，但似乎往往难以使用。败血症并发症的诊断基于临床和辅助临床依据。局部引流、体温超过38摄氏度以及局部/持续性疼痛和僵硬是术后感染更具信息性的指标。
