number needed to screen and prevalence
- related: Biostats
- tags: #literture #pulmonary
Links to this note
Optimal screening strategies depend on diagnostic tests that are readily available, have high sensitivity, and offer an intervention that can improve outcomes. The NLST was the first major lung cancer screening trial to demonstrate a mortality reduction in subjects who underwent screening with annual low-dose chest CT scanning over the course of 3 years. Follow-up modeling suggested that ongoing CT screening beyond 3 years would prevent even more deaths.
Success in screening depends on the prevalence of the disease, with less prevalent diseases requiring more subjects screened to prevent one death. When the USPSTF urged expansion of the screening eligibility to a broader range of patients who are younger with less tobacco consumption, epidemiologic principles would suggest the disease prevalence in the expanded population would be lower. Therefore the NNS should go up in response to screening a population with less lung cancer prevalence. As treatment effectiveness improves, the benefit from early detection of lung cancer should be greater, and the NNS to prevent a death should decrease. Results from a subsequent Dutch-Belgian Randomized Lung Cancer Screening Trial (NELSON) published in 2020 demonstrated an NNS of just 130 (down from 320 in NLST trial). These principles may be discussed during the shared decision-making process with patients eligible for LCS, along with the potential harms from screening (eg, false-positive results).1