Characterization of time to diagnosis indicates shorter interval for screenable versus symptom-driven cancers.

Abstract

10526 Background: Cancer screening provides a straightforward diagnostic path compared to non-screening, which often requires complex evaluation for patients with non-specific symptoms. The time to diagnosis (from clinical presentation or screening event to definitive diagnosis) captures the clinical efficiency of the journey. The study characterizes time to diagnosis for cancers with and without USPSTF Grade A/B screening recommendations. This could inform the value of screening in diagnostic management and care efficiency improvements. Methods: The Geisinger Health System Phenomics Initiative Database provides deidentified EHR and tumor registry data. Patients aged 50-74 with encounters at least biennially for six years prior to a cancer diagnosis were frequency-matched to a non-cancer control group by diagnosis year. From 18 months prior to diagnosis, equally-sized random samples were drawn (with replacement) from a time-bounded sliding window at monthly intervals. Aggregated procedure and diagnostic code counts for cancer and non-cancer samples were used to calculate the Bray-Curtis dissimilarity (BCd) statistic. The start of the diagnostic path was defined as the earliest date with a monotonically-increasing statistically-significant difference in BCd (95% CI), and was stratified into cancers with and without screening and by stage. Results: 8188 new cancer cases were identified between 2010-2019; 40% had an established screening modality (breast, lung, colorectal, cervical), 55% of which were screened prior to diagnosis. The time to diagnosis for non-screenable cancers was 6 months versus 3 months for screenable cancers. The BCd for procedures differed by 0.029 (95% CI: 0.021-0.037) between the two groups at 6 months prior to diagnosis. Stages I and IV had a time to diagnosis of 3 months across all cancers versus 7 and 6 months for stages II and III, respectively. Conclusions: We report the first known quantification of time to diagnosis across multiple cancers. Screenable cancers were associated with shorter time to diagnosis than non-screenable cancers. Stage II and III cancers took longer to diagnose than stage I and IV, likely related to early detection by screening in the former and severe symptom presentation in the latter. These novel data highlight improved time to diagnosis as another achievable benefit from expanded, multi-cancer screening.