Risk factors of cervical cancer (CC) development are well investigated, however, those influencing the risk of a potential false negative cytology preceding diagnosis of an invasive CC are not. We have aimed to explore these factors according to the data from Organised Cervical Cancer Screening Programme (OCCSP) in Poland. A total of 2.36 million of Pap tests sampled in 2010–2012 within OCCSP were merged with the Polish National Cancer Registry to identify CC cases after abnormal cytology and after normal cytology within 3 years of screening. Of 1460 invasive CCs, 1025 were preceded by abnormal and 399 by normal cytology result. Multivariate logistic analysis indicated that the presence of microorganisms in the Pap (OR = 2.18, 95% CI 1.65–2.87), evaluation by smaller (below 9000 slides processed per year) laboratories (OR = 1.60, 95% CI 1.22–2.09) and non‐squamous histology of cancer increased the odds for a potential false negative result (OR = 3.39, 95% CI 2.37–4.85 for adenocarcinoma, OR = 1.99, 95% CI 1.11–3.55 for other types of carcinoma), whereas cervical ectropion, other macroscopic changes on the cervix and smoking decrease the odds for a potential false negative Pap test result preceding CC (OR = 0.61, 95% CI 0.45–0.82, OR = 0.41, 95% CI 0.25–0.67, OR = 0.60, 95% CI 0.46–0.78, respectively). Proper triage of women with microscopic signs of microorganisms in the Pap smear should be reconsidered and cytology should be assessed in laboratories processing over 9000 slides annually to decrease the odds for negative Pap test result in 2 years before CC diagnosis. Information on macroscopic changes on the cervix provided to cytomorphologist may reduce the risk of a potential false negative cytology result.

We have aimed to study reasons for reporting false-negative cytology results preceding diagnosis of interval cervical cancers (CC) in Poland. Data on all Pap smears collected in the organised screening in 2010-2015 were retrieved from the electronic database and linked with Polish National Cancer Registry (PNCR) data. False-negative results were defined as those sampled and assessed normal up to 3.5 years before diagnosis of invasive CC. False-negative slides were then seeded among twice as many randomly selected slides from the same lab and reviewed independently by three expert cytomorphologists. New diagnosis was established when experts agreed on a result. Of 48 selected false-negative slides, 1 case was diagnosed as a low-grade abnormality, 22 cases as a high-grade abnormalities, 3 cases as unsatisfactory for evaluation and 5 as no intraepithelial lesion of malignancy (NILM) by all three experts. There was no agreement in 17 cases. Percentages of agreement between experts was 64.6. Interobserver agreement rate was moderate with Fleiss' κ values. Our pilot study indicates evaluation errors as the main reason of false-negative cytology preceding interval CC in the organized screening programme in Poland. True lack of abnormal cells on the slide is the next reason.

Background False‐negative (FN) results in cervical cancer (CC) screening pose significant risk for participants and should be audited. The aim of the study was to analyse the results of audit of FN slides collected in 2010–2013 in Polish Cervical Cancer Screening Program (CCSP) and to seek for risk factors of obtaining true‐negative result (TN; not containing abnormal cells as confirmed in audit) before CC diagnosis. Methods Screening database was merged with National Cancer Registry to identify negative slides preceding histologically confirmed CC diagnosis up to 42 months. Two blinding slides were randomly assigned per each FN. The whole set was reassessed independently by three pathologists with 30 years of experience in cytology evaluation. Final audit result was established in the case of ≥2 coherent reports. Agreement rates and kappa (κ) coefficients were calculated. Logistic analysis of risk factors for obtaining TN result was performed. Results Of 374 included FNs, 204 were considered abnormal (54.6%) and 91 were confirmed negative for intraepithelial neoplasia (24.3%). Agreement between experts was moderate for FNs (κ = 0.266) and fair for blinding slides (κ = 0.142) when grouping abnormal slides. Adenocarcinoma diagnosis elevated the risk of TN result (OR = 3.83); detection of macroscopic changes on the cervix and smoking lowered the risk (OR = 0.39, OR = 0.40 respectively). Conclusions Misinterpretation was the main reason for FN cytology in the CCSP which indicated the need of further personnel training to increase screening quality. Rather low agreement between auditors requires further insight. A standardised process of auditors' selection should be planned to increase audit quality.

In our pilot study we have aimed to assess interlaboratory variability of cytological diagnoses in selected laboratories participating in the Polish Cervical Cancer Screening Programme (CCSP) to establish grounds for certification system for cytodiagnosticians and to monitor the quality of services. Set of 50 selected Pap smears, previously reassessed by an expert on the grounds of clinical, colposcopic and histological data was blinded and sent to 15 laboratories in Poland with request for evaluation according to routine practice according to the Bethesda 2001 system. Concordance with expert diagnoses reached a median of 82% (range: 66% to 92%), with median unweighted κ coefficient at κ = 0.67 (range 0.40 to 0.86) depending on laboratory. This indicates substantial agreement among laboratories, however with essential differences in proper evaluation in some outlying laboratories. Agreement was highest in samples with high-grade, lower for low-grade abnormalities. Slides with ASC-US and ASC-H expert diagnoses were most troubling for cytodiagnosticians. Sets of highly selected cytological slides with expert diagnoses may serve as a tool in the process of comprehensive periodic recertification of cytodiagnosticians in the screening programme. A benchmark level of agreement with expert diagnoses should be established to guide corrective actions for cytodiagnosticians with lowest agreement.