外周血白细胞计数和（或）血糖早期不能识别手足口病重症重型病例

目的评价外周血WBC、CRP和血糖在早期识别手足口病重症重型病例中的价值。方法通过电子病历检索系统，回顾性分析2009年1月1日至2010年12月31日复旦大学附属儿科医院收治的手足口病普通病例（1期组）681例和重症重型病例（2期组）239例，比较两组的WBC、CRP及血糖值并绘制上述3项指标的受试者工作特征曲线（ROC），计算曲线下面积（AUC）及约登指数，确定最佳临界值，并探索联合检测方法的可行性。结果①两组WBC、CRP和血糖值均为非正态分布数据。1期组和2期组WBC的中位数分别为11．4和11．3×10^9·L^-1，CRP中位数均为8mg·L^-1（CRP〈8mg·L^-1计为8mg·L^-1），1期组和2期组血糖的中位数分别为5．0和5．6mmol·L^-1。②采用Stata10．0软件进行统计分析，1期组和2期组WBC经log转换后呈正态分布，但方差不齐，采用t检验，差异无统计学意义（P=0．427）。1期组和2期组CRP和血糖均采用非参数统计方法，两组CRP和血糖差异均有统计学意义（P均〈0．001）。③根据WBC、CRP及血糖绘制ROC曲线，3项指标的AUC值分别为0．512、0．405和0．625。WBC的最佳临界值为7．85×10^9·L^-1，敏感度和特异度分别为88．7％和18．4％。血糖的最佳临界值为5．25mmol·L^-1，敏感度和特异度分别为60．7％和59．0％。④将WBC和血糖的最佳临界值进行串联检测的敏感度为37．3％，特异度为81．2％。结论WBC和（或）血糖在早期识别手足口病重症重型病例的准确性较低，而CRP不能作为识别手足口病重症重型病例的指标。WBC和血糖串联检测可提高诊断试验的特异度，但未提高诊断试验的价值。

White blood cell count and glucose of peripheral blood can t indicate severe hand,foot and mouth disease with neurological involvement

Objective To evaluate the value of white blood cell count （WBC）, C reactive protein （CRP） and glucose of peripheral blood in detection of severe hand, foot and mouth disease （HFMD） with neurological involvement , and to provide the reference for clinical diagnosis of severe HFMD. Methods The sick children with HFMD admitted to Childreng Hospital of Fudan University from Jan 2009 to Dec 2010, were recruited and divided into the mild HFMD （stage 1 ） and the severe HFMD （stage 2 ） according to HFMD clinical criterion. All the cases were reviewed by Electronic Medical Record. Severe HFMD in this research were those with neurological involvement but not patients with cardiopulmonary dysfunction. WBC, CRP and glucose of all the patients were collected on the first day of admission, and compared by using stata 10.0 software. Receiver operating characteristic （ROC） analysis was performed. The best diagnostic value was found by calculating their area under the curve （AUC） and Youden index. Results （·）A total of 920 patients were recruited into this study. 681 patients were categorized into stage 1 and 239 patients were categorized into stage 2. @WBC, CRP and glucose of these two groups were all in abnormal distribution. The median WBC was 11.4 × 10^9 · L^-1 （ range 3.1 × 10^9 · L^- 1- 39.8 ×10^9 · L^- 1 ） and 11.3 × 10^9 · L^-1 （ range 4.9 ×10^9 · L^-1 - 26.3 × 10^9 · L^-1 ） for stage-I subjects and stage-2 subjects, respectively.. The median CRP of two groups was both 8mg · L 1 （stage 1 range： 8 - 160mg · L^-1 , stage 2 range ： 8 - 77 mg · L^- 1 ）. The median glucose of stage 1 was 5.0mmol · L^- 1 （ range 3.5 - 11.7 mmol · L^-1）, and of stage2 was 5.6 mmol · L^-1（range 3.5-15.7 mmol · L^-1）. （3） Using student-t test to analyze WBC after log transformtaion, there was no significant difference in WBC between two groups （ P = 0. 427 ）. Non-parametric test was used to analyze CRP and glucose, the results showed significant difference （P 〈 0.001 ）. （4） AUC of WBC, CRP and glucose was 0.512, 0. 405 and 0.625. The best diagnostic value of WBC was 7.85 x 10^9 · L^-1 （ sensitivity 88.7% , specificity 18.4% ）. The best diagnostic value of glucose was 5.25 mmol · L^-1 （sensitivity 60.7%, specificity 59.0% ）. （5） Combining two best diagnostic values and using diagnostic testing, achieving sensitivity of 37.3% and specificity of 81.2%. Conclusion WBC counts and glucose levels showed low validity in detection of severe HFMD with neurological involvement. CRP levels could not to be used to predict severe HFMD. Serial test in this research did not increase the diagnostic validity. Clinical doctors should pay more attention to clinical features and signs.