Consistency of peripheral whole blood and venous serum procalcitonin in children: a multicenter parallel controlled study

魏永越,陈红兵,陈炜刚,张泓,吴国荣,沈照波,朱叶飞,郑跃杰,陆权,赵德育,江咏梅,董晓艳,刘成桂,张冲林,艾涛,郑爱斌,邹映雪,田礼军,陈运生,魏良敏,李羚,顾猛,刘瀚旻,杨俊梅,沈永明
DOI: https://doi.org/10.3760/cma.j.cn112140-20210224-00153
2021-01-01
Abstract: To explore the consistency of peripheral whole blood and venous serum procalcitonin (PCT) levels, and the value of peripheral whole blood PCT in evaluating pediatric bacterial infection. This multicenter cross-sectional parallel control study was conducted in 11 children's hospital. All the 1 898 patients older than 28 days admitted to these hospitals from March 2018 to February 2019 had their peripheral whole blood and venous serum PCT detected simultaneously with unified equipment, reagent and method. According to the venous serum PCT level, the patients were stratified to subgroups. Analysis of variance and chi-square test were used to compare the demographic characteristics among groups. And the correlation between the peripheral blood and venous serum PCT level was investigated by quantitative Pearson correlation analysis.The PCT resultes were also converted into ranked data to further test the consistency between the two sampling methods by Spearman's rank correlation test. Furthermore, the ranked data were converted into binary data to evaluate the consistency and investigate the best cut-off of peripheral blood PCT level in predicting bacterial infection. A total of 1 898 valid samples were included (1 098 males, 800 females),age 27.4(12.2,56.7) months. There was a good correlation between PCT values of peripheral whole blood and venous serum (0.970.01). The linear regression equation was PCT=0.135+0.929×PCT. However, when stratified to 5 levels, PCT results showed diverse and unsatisfied consistency between the two sampling methods (0.51-0.92, all 0.01). But after PCT was converted to ordinal categorical variables, the stratified analysis showed that the coincidence rate of the measured values by the two sampling methods in each boundary area was 84.9%-97.1%. The dichotomous variables also showed a good consistency (coincidence rate 96.8%-99.3%, Youden index 0.82-0.89). According to the severity of disease, the serum PCT value was classified into 4 intervals(<0.5、0.5-<2.0、2.0-<10.0、≥10.0 μg/L), and the peripheral blood PCT value also showed a good predictive value (AUC value was 0.991 2-0.997 9). The optimal cut points of peripheral whole blood PCT value 0.5、1.0、2.0、10.0 μg/L corresponding to venous serum PCT values were 0.395, 0.595, 1.175 and 3.545 μg/L, respectively. There is a good correlation between peripheral whole blood PCT value and the venous serum PCT value, which means that the peripheral whole blood PCT could facilitate the identification of infection and clinical severity. Besides, the sampling of peripheral whole blood is simple and easy to repeat.
What problem does this paper attempt to address?