Optimizing Cervical Cancer Screening: Investigating the Implementation and Impact of Home-Based HPV Self-Sampling in the United States

Abstract

Cervical cancer (CC) screening and human papillomavirus (HPV) testing programs have been successful in reducing CC incidence and mortality in the United States. However, screening adherence has declined over the past 20 years, in part due to knowledge gaps, stigma, and inaccessibility driven by structural determinants of health. This has established the need for alternative screening strategies, such as home-based HPV self-sampling (HPV-SS). HPV-SS is nascent in the US but gaining traction, having received recent approvals from the Food and Drug Administration. Additionally, policies like the June 2022, Dobbs v. Jackson Women's Health Organization (Dobbs) Supreme Court decision targeted reproductive care access among those eligible for screening and may be spilling over and influencing CC screening uptake in ways not yet evident. As HPV-SS and a shifting policy environment shape screening in the US, this dissertation aims to understand this evolving landscape and the broader implications through three distinct but related Chapters. Chapter 1 used data from STEP (Self-Testing options in the Era for Primary HPV screening for CC), a pragmatic, randomized trial within Kaiser Permanente Washington of ~31,000 participants to assess the effectiveness of mailed, home-based HPV-SS for increasing CC screening uptake compared to usual care and educational resources across a range of screening histories (adherent, overdue, unknown). We used binomial and multinomial regression models with generalized estimating equations to investigate if the effectiveness of offering home-based HPV-SS differed by residential neighborhood socioeconomic status (nSES), operationalized as Social Vulnerability Index (SVI) and affluence. This work involved the deterministic linkage of electronic health records with publicly available census tract nSES data. We demonstrated that mailing home-based HPV-SS kits increased screening completion overall, but effects were lower among those living in low nSES tracts. Among overdue participants, who are of highest priority, the directly mailed HPV-SS intervention increased screening completion by 11.9% (95%CI: 6.2, 17.7) for those in low vs. 17.9% (95%CI: 15.0, 20.8) in high nSES SVI areas. Individuals in low nSES areas were less likely to choose self-sampling over in-clinic screening. Chapter 2 also used data from STEP, but restricted to those with a history of screening adherence (~13,000). In this chapter, we used modified Poisson regression and propensity score stratification for confounding to investigate if offering and choosing to use HPV-SS was associated with subsequent flu vaccine uptake, an important and annually recommended preventive service. There were no differences in flu vaccine uptake comparing those randomized to receive mailed HPV-SS kits compared to usual care screening procedures or educational materials (relative risk (RR)=1.01; 95%CI: 0.96, 1.07), and no differences in vaccine uptake comparing those who utilized HPV-SS compared to those who in-clinic screened (RR=0.99; 95%CI: 0.88, 1.10). However, those who did not complete any screening were less likely to receive a flu vaccine compared to those who did (self-sample or in-clinic), and were characteristically different regarding past healthcare behaviors. Data for Chapter 3 came from the Behavioral Risk Factor Surveillance System (BRFSS), a national telephone survey of US residents. With data from ~41,000 respondents, we employed a difference-in-differences (DID) design to investigate if abortion bans enabled by the Dobbs decision were associated with changing trends in CC screening in states that did vs. did not restrict abortion, as well as a difference-in-difference-in-differences (DDD) design to investigate if this relationship was stronger in populations with historically less access to screening. DID and DDD were fit both unadjusted and adjusted as linear regression models, directly estimating prevalence differences, with state and year fixed effects. All states in our analysis experienced an increase in CC screening from 2022 to 2024. However, while nonsignificant, there was preliminary evidence that Dobbs-enabled abortion bans influenced CC screening adherence prevalence over this period (adjusted DID= -2.20%; 95%CI: -7.02, 2.62), and that Dobbs may be influencing some populations’ screening more than others (adjusted DDD for public vs. private insurance= -4.8%; 95%CI: -16.0, 6.3). This dissertation demonstrated that HPV-SS may be less effective for improving screening rates in low nSES areas, which has implications for existing CC screening disparities. However, HPV-SS is also unlikely to influence flu vaccination, providing evidence that this novel screening tool may not have unintended consequences on uptake of other care. Additionally, Dobbs could be contributing to the already challenging screening situation in the US, and may reveal populations in greatest need of targeted interventions, such as HPV-SS. This dissertation provides valuable context for healthcare officials and stakeholders looking to make evidence-based decisions around the implementation and utility of HPV-SS in the US, with implications for increasing CC screening uptake, addressing disparities, and improving care delivery, considering individual, community, and policy-level factors.