Diagnostic Timing and Ovarian Cancer Survival
Menée aux Etats-Unis à partir de données du registre des cancers de Caroline du Nord portant sur 2 359 patientes atteintes d'un cancer de l'ovaire (âge médian : 71 ans), cette étude de cohorte évalue l'association entre le délai avant le diagnostic de la maladie et la mortalité toutes causes confondues à 5 ans
Importance : While earlier diagnosis improves outcomes for many cancers, studies report that earlier diagnosis does not improve ovarian cancer (OC) survival, discouraging investment in diagnostic improvements. One possible explanation is confounding due to the wait time paradox, in which patients with severe disease are diagnosed more quickly but have poorer outcomes, masking potential benefits of earlier diagnosis.
Objective :To visualize how nonlinear modeling of time to diagnosis alters observed associations with OC survival, with particular attention to the wait time paradox.
Design, Setting, and Participants : This cohort study used data from the North Carolina Central Cancer Registry with linked multipayer claims from Medicare, Medicaid, and private insurance. Participants included female patients diagnosed with OC from January 1, 2009, to December 31, 2019, in North Carolina. Follow-up was completed June 30, 2021. Data were analyzed from May 1, 2024, to November 1, 2025.
Exposure : Time to diagnosis, defined as the interval from the earliest symptom-related health care encounter to cancer diagnosis.
Main Outcomes and Measures : All-cause 5-year mortality.
Results : Among 2359 women included in the analysis (median age, 71 [IQR, 64-78] years), the median diagnostic interval was 33 (IQR, 10-149) days. Compared with all other quartiles, patients in the second-shortest quartile of the diagnostic interval were more likely to be diagnosed at localized (83 of 586 [14.2%] vs 180 of 1773 [10.2%]) or regional (134 of 586 [22.9%] vs 330 of 1773 [18.6%]) stages (P = .001) and had a median survival exceeding 5 years (lower 95% CI bound, 4.1 years), compared with median survivals of 2.5 years (95% CI, 2.0-3.1 years) in the shortest quartile and 2.9 years (95% CI, 2.5-3.8 years) in the longest quartile. When modeled using restricted cubic splines, the adjusted association between time to diagnosis and mortality was U-shaped; compared with a reference interval of 80 days, intervals of 10 and 360 days were associated with higher mortality (hazard ratios, 1.29 [95% CI, 1.09-1.53] and 1.13 [95% CI, 0.88-1.46], respectively).
Conclusions and Relevance: In this cohort study of patients with OC, very short diagnostic intervals were associated with poorer survival, consistent with residual confounding by the wait time paradox, while longer diagnostic intervals were also associated with worse outcomes. Conventional linear approaches may obscure this important confounding by severity in the association between diagnostic timing and survival, contributing to null findings in prior studies and potentially discouraging efforts to improve diagnosis of this difficult-to-detect disease.
JAMA Network Open , article en libre accès, 2026