Epidemiology and risk factors for ovarian cancer incidence in the USA: a multilevel analysis

Ovarian cancer (OC) has the worst prognosis and highest death rate of all gynaecological cancers in the USA. We examined the independent effects of individual-, neighbourhood-, and state-level factors on ovarian cancer incidence using a multilevel analytical framework.

In this retrospective cohort study, we analysed de-identified data from the All of Us research database, identifying women ≥18 years without prior ovarian cancer before January 2017. Participants were followed from 1 January 2017 through October 2023 (median follow-up: 6.6 years). Mixed-effects Cox regression models examined data on 85 388 individuals nested within ZIP-code areas and states, analysing individual-level risk factors and neighbourhood-level socioeconomic determinants, while accounting for geographic clustering. We fitted four progressive models: a null (random effects only), individual-level factors, neighbourhood-level factors, and full model with all covariates.

Among 85 388 women followed for a total of 569 847 person-years, 419 (0.49%) developed OC. Age demonstrated the strongest associations, with significantly elevated risks of developing OC among women aged 50-59 years (adjusted hazard ratio (aHR) = 1.83; 95% confidence interval (CI) = 1.28-2.61), 60-69 years (aHR = 2.01; 95% CI = 1.39-2.90), and ≥70 years (aHR = 1.67; 95% CI = 1.07-2.59) compared to those <40 years. Retired women had increased risk of OC compared to employed women (aHR = 1.39; 95% CI = 1.04-1.86). Non-Hispanic Black women demonstrated lower risk of OC than non-Hispanic White women (aHR = 0.63; 95% CI = 0.45-0.88). Regional variations showed 53% lower risk in the South vs. Northeast (aHR = 0.47; 95% CI = 0.25-0.86). Hormone replacement therapy was associated with increased risk of OC (aHR = 2.46; 95% CI = 1.07-5.67). Significant geographic clustering of OC was observed at neighbourhood and state levels.

Individual-level factors, particularly age and employment status, are the primary determinants of OC risk, while apparent geographic disparities reflect population composition, rather than unmeasured environmental factors. The complete explanation of geographic clustering through measured covariates could provide important insights for targeted prevention strategies and future epidemiological research.