Causal inference, cancer registry data, and a single state policy change: Evaluating Virginia's HPV vaccine mandate

Cancer epidemiologists increasingly contribute evidence evaluating the causal effect of policies on cancer outcomes. The validity of such evidence relies on attention to conceptual, analytic, and data aspects of causal inference research. To highlight these aspects and contribute policy-relevant evidence, we evaluated Virginia's Human Papillomavirus (HPV) vaccination mandate on cervical cancer incidence. HPV vaccination mandates were intended to increase vaccine uptake and reduce cervical cancer incidence, yet causal evidence on cancer incidence remains scarce. Virginia implemented an HPV vaccine mandate in 2008, providing a natural experiment within a causal inference framework.

After motivating why traditional methods may be invalid, we estimated the effect of Virginia's HPV vaccine mandate on cervical cancer incidence among females aged 20-24 with a synthetic control triple differences design. Our post-period begins in 2016, when the first mandate-exposed cohort reached ages 20-24. State-level cancer registry data (2003-2019) were used to compare Virginia's incidence trends to a weighted combination of control states, while adjusting for within-state trends across age groups exposed and not exposed to the mandate.

Cervical cancer incidence among 20-24-year-old females declined over time in both Virginia and comparison states. Ignoring within-state differential age-group trends, the synthetic control difference-in-differences design showed no evidence of any effect. The synthetic control triple difference design estimated that Virginia's mandate was associated with 1.6 additional cervical cancer cases per 100,000 population (CI = 0.4, 2.7). Alternative specifications, including adjustments for screening and multiple sensitivity analyses, produced consistent results in direction and inference.

Compared to states with no mandate, Virginia's HPV vaccine school-entry mandate was not associated with reduced cervical cancer incidence. Our results should be interpreted as evidence about the implementation and effectiveness of Virginia's mandate policy, not as evidence against the established effectiveness of HPV vaccination as a cancer prevention tool.