大规模人群筛检的假阴性分析

徐宁; 童懿昕; 蒋鸿琳; 周艺彪; 姜庆五

doi:10.19428/j.cnki.sjpm.2022.22031

摘要:

目的探讨传染病在低感染率流行状态下，筛检试验的阴性预测值和假阴性数量在不同人群感染率、灵敏度和特异度情况下的变化趋势。

方法通过数据模拟，假定人口数为2 000万，分别计算人群疾病感染率为0.1%、1.0%和5.0%的情况下，不同灵敏度（75.0%、80.0%、85.0%、90.0%、95.0%、99.0%）和特异度（90.0%、95.0%、99.0%、99.9%）组合的阴性预测值、真阴性数和假阴性数。

结果当人群感染率为0.1%时，灵敏度≥75.0%和特异度≥90.0%的筛检试验在2 000万的人群中可发现的真阴性数约为1 798.20万~1 996.00万人。当灵敏度为75.0%时，阴性预测值为99.972%~99.975%，假阴性人数为0.50万人；当灵敏度提高至99.0%时，阴性预测值为99.999%，假阴性人数减少至200人。当人群感染率为1.0%时，灵敏度≥75.0%和特异度≥90.0%的筛检试验在2 000万的人群中可发现的真阴性数约为1 782.00万~1 978.02万人。灵敏度为75.0%时，阴性预测值为99.720%~99.748%，假阴性人数为5.00万人；当灵敏度提高至99.0%时，阴性预测值升高至99.990%，假阴性人数减少至2 000人。当人群感染率为5.0%时，灵敏度≥75.0%和特异度≥90.0%的筛检试验在2 000万的人群中可发现的真阴性数约为1 710.00万~1 898.10万人。当灵敏度为75.0%时，阴性预测值为98.559%~98.700%，假阴性人数可达25.00万人；当灵敏度达到99.0%时，阴性预测值升高至99.942%~99.947%，假阴性人数减少至1.00万人。人群感染率越低，筛检中出现的假阴性者人数越少。

结论开展大规模筛检中的假阴性者人数随着感染率的升高而成倍的增加，应尽可能在传染病流行的早期开展筛检，以尽快控制疾病的流行。

Abstract:

Objective To explore the changing trend of negative predictive value and number of false negatives in screening tests under the condition of low infection rate of infectious diseases.

Methods Assuming that the population is 20 million， to calculate the negative predictive value， numbers of true negatives and false negatives of the combination of different sensitivity （75.0%， 80.0%， 85.0%， 90.0%， 95.0%， 99.0%） and specificity （90.0%， 95.0%， 99.0%， 99.9%） when the disease infection rate of the population is 0.10%， 1.0% and 5.0% respectively.

Results When the population infection rate is 0.1%， with the screening test sensitivity ≥75.0% and specificity ≥90.0%， the number of true negatives in 20 million people is about 17.98‒19.96 million. When the sensitivity is 75.0%， the negative predictive value is 99.972%‒99.975%， and the number of false negatives is 5 000； When the sensitivity increases to 99.0%， the negative predictive value is 99.999%， and the number of false negatives decreases to 200. When the population infection rate is 1.0%， a screening test with sensitivity ≥75.0% and specificity ≥90.0% can detect about 17.82‒19.78 million true negatives in 20 million population. When the sensitivity is 75.0%， the negative predictive value is 99.720%‒99.748%， and the number of false negatives is 50 000； When the sensitivity increases to 99.0%， the negative predictive value increases to 99.990%， and the number of false negatives decreases to 2 000. When the population infection rate is 5.0%， with sensitivity ≥75.0% and specificity ≥90.0%， the number of true negatives in 20 million people is about 17.10‒18.98 million； when the sensitivity is 75.0%， the negative predictive value is 98.559%‒98.700%， and the number of false negatives can reach 250 000； When the sensitivity is 99.0%， the negative predictive value increases to 99.942%‒99.947%， and the number of false negatives decreases to 10 000. The lower the infection rate of the population， the fewer false negatives will appear in the screening.

Conclusion The number of false negatives in large-scale screenings increases exponentially with the increase of infection rate. Screenings should be carried out as early as possible in a pandemic of infectious diseases， so as to control the spread of the pandemic as soon as possible.

大规模人群筛检的假阴性分析

False negatives of screening in large-scale population