Fernández Montejo María Del Pilar, Saghir Zaigham, Bødtger Uffe, Jepsen Randi, Lynge Elsebeth, Lophaven Søren
Centre for Health Research, Zealand University Hospital, Nykøbing Falster, Denmark
Department of Medicine, Gentofte Hospital, Hellerup, Denmark.
BMJ Open Respir Res. 2024 Dec 25;11(1):e002499. doi: 10.1136/bmjresp-2024-002499.
We assessed the impact of recruitment criteria on lung cancer detection in a future Danish screening programme with low-dose CT.
We combined data from two Danish population-based health examination surveys with eligibility criteria from seven randomised controlled trials on lung cancer screening. Incident lung cancers were identified by linkage with the National Pathology Data Bank (Patobank). For an average of 4.4 years of follow-up, we calculated sensitivity, specificity, efficient frontier and number needed to screen (NNS) for lung cancer detection.
When applying the different eligibility criteria to the 48 171 persons invited to the two surveys, the number of lung cancer cases identified in the target groups varied from 46 to 68. The National Lung Screening Trial (NLST) criteria had the highest sensitivity of 62.6% (95% CI 52.7 to 71.8) and the Dutch-Belgian NEderlands-Leuvens Screening ONderzoek (NELSON) criteria had the highest specificity 81.6% (95% CI 81.0 to 82.1). Sensitivity was higher for men than for women (NLST criteria 71.7% (95% CI 57.7 to 83.2) and 53.7% (95% CI 39.6 to 67.4), respectively). The NLST criteria identified the target population obtaining the lowest NNS with 46.3. The application of the NLST criteria showed that the higher the sensitivity, the lower the number of false-negative cases and, thus, the lower the NNS.
This study highlights the impact of the definition of the at-risk population on lung cancer screening efficacy. We found lower sensitivity among women regardless of screening criteria used. This should be carefully addressed in a possible screening programme.
我们评估了在未来丹麦低剂量CT筛查项目中，招募标准对肺癌检测的影响。
我们将两项丹麦基于人群的健康检查调查数据与七项肺癌筛查随机对照试验的纳入标准相结合。通过与国家病理数据库（Patobank）联动识别新发肺癌病例。在平均4.4年的随访期内，我们计算了肺癌检测的灵敏度、特异度、有效前沿和筛查所需人数（NNS）。
对受邀参加两项调查的48171人应用不同的纳入标准时，目标组中识别出的肺癌病例数从46例到68例不等。国家肺癌筛查试验（NLST）标准的灵敏度最高，为62.6%（95%可信区间52.7至71.8），荷兰-比利时荷兰-鲁汶筛查研究（NELSON）标准的特异度最高，为81.6%（95%可信区间81.0至82.1）。男性的灵敏度高于女性（NLST标准分别为71.7%（95%可信区间57.7至83.2）和53.7%（95%可信区间39.6至67.4））。NLST标准识别出的目标人群的NNS最低，为46.3。NLST标准的应用表明，灵敏度越高，假阴性病例数越低，因此NNS也越低。
本研究强调了高危人群定义对肺癌筛查效果的影响。我们发现，无论使用何种筛查标准，女性的灵敏度都较低。在可能开展的筛查项目中应谨慎对待这一问题。
