Finnish Cancer Registry, Helsinki, Finland.
Cancer Epidemiol Biomarkers Prev. 2010 Feb;19(2):381-7. doi: 10.1158/1055-9965.EPI-09-1038.
We assessed the performance and validity of cytology in the Finnish screening program by considering high-grade neoplasia and cervical cancer (CIN3+) rates as detected in the program and by reevaluating cases observed after a negative screening test.
This retrospective study included 915 screen-detected CIN3+ cases and 421 cases observed after a negative screen. Randomized and blinded reevaluation of potential false-negative screening tests covered 345 archival case smears from women without a referral to colposcopy, as well as 689 control smears for estimating performance and validity measures.
The false-negative rate at the cutoff of low-grade squamous intraepithelial lesion or worse was 35% (95% confidence interval, 30-40%). In the subpopulation with original screening result of Pap I, the false-negative rate was 23% (18-28%). Sensitivity of screening laboratory rereading for detecting low-grade lesions or worse as atypical was 75% (67-82%) and specificity 93% (91-94%). Reproducibility of specific cytologic diagnoses was only fair. False negatives constituted 11% of all CIN3+ diagnoses in the screened population; those false negatives with an original Pap I screening result constituted 5%.
Although screen failures in the form of diagnostic false negatives occur within the Finnish screening program, their effect on cancer incidence is fairly small and cannot be readily decreased without sacrificing the high specificity of screening or without high incremental costs. Feedback for the screening laboratories is needed, however, to improve the reproducibility of cytologic diagnoses to optimize the burden of intensified follow-up and treatment of precancerous lesions.
我们通过考虑该方案中检测到的高级别肿瘤和宫颈癌（CIN3+）的发生率，并通过重新评估阴性筛查试验后的病例，评估了细胞学在芬兰筛查计划中的性能和有效性。
这项回顾性研究包括 915 例筛查发现的 CIN3+病例和 421 例阴性筛查后观察到的病例。对潜在假阴性筛查试验进行随机和盲法重新评估，涵盖了 345 例无转诊行阴道镜检查的女性的存档病例涂片，以及 689 例用于估计性能和有效性的对照涂片。
在低级别鳞状上皮内病变或更高级别病变的截止值处，假阴性率为 35%（95%置信区间，30-40%）。在原始筛查结果为巴氏 I 级的亚人群中，假阴性率为 23%（18-28%）。筛查实验室重新阅读检测低级别病变或更高级别不典型病变的敏感性为 75%（67-82%），特异性为 93%（91-94%）。特定细胞学诊断的重复性仅为中等。假阴性构成筛查人群中所有 CIN3+诊断的 11%；原始巴氏 I 级筛查结果的假阴性构成 5%。
尽管在芬兰筛查计划中出现了诊断假阴性的筛查失败，但它们对癌症发病率的影响相当小，如果不牺牲筛查的高特异性，或者不增加高成本，就很难降低其影响。然而，需要为筛查实验室提供反馈，以提高细胞学诊断的重复性，从而优化对癌前病变的强化随访和治疗负担。
