Adapting a model of cervical carcinogenesis to self-identified Black women to evaluate racial disparities in the United States

Abstract

Background

Self-identified Black women in the United States have higher cervical cancer incidence and mortality than the general population, but these differences have not been clearly attributed across described cancer care inequities.

Methods

A previously established microsimulation model of cervical cancer was adapted to reflect demographic, screening, and survival data for Black US women and compared with a model reflecting data for all US women. Each model input with stratified data (all-cause mortality, hysterectomy rates, screening frequency, screening modality, follow-up, and cancer survival) was sequentially replaced with Black-race specific data to arrive at a fully specified model reflecting Black women. At each step, we estimated the relative contribution of inputs to observed disparities.

Results

Estimated (hysterectomy-adjusted) cervical cancer incidence was 8.6 per 100 000 in the all-race model vs 10.8 per 100 000 in the Black-race model (relative risk [RR] = 1.24, range = 1.23-1.27). Estimated all-race cervical cancer mortality was 2.9 per 100 000 vs 5.5 per 100 000 in the Black-race model (RR = 1.92, range = 1.85-2.00). We found the largest contributors of incidence disparities were follow-up from positive screening results (47.3% of the total disparity) and screening frequency (32.7%). For mortality disparities, the largest contributor was cancer survival differences (70.1%) followed by screening follow-up (12.7%).

Conclusion

To reduce disparities in cervical cancer incidence and mortality, it is important to understand and address differences in care access and quality across the continuum of care. Focusing on the practices and policies that drive differences in treatment and follow-up from cervical abnormalities may have the highest impact.

The United States has made incredible progress in cervical cancer prevention and treatment, with incidence and mortality rates declining more than 50% since the 1970s (1). Yet, although improvements have been seen for all groups, these improvements are not observed equally across racial and ethnic groups, resulting in racial disparities in incidence and mortality that have persisted, or even increased (2,3).

Reasons behind these disparities are complex, but evidence increasingly supports that they are fundamentally the product of inequities in care access and quality resulting from what it means to be racialized as Black in the United States (2,4,5). Self-identified Black (Black) individuals in the United States are more likely to live in contexts with narrow Medicaid eligibility and with lower access to economic opportunity, health-promoting community resources, and primary care (6-9). Further, the history of reproductive care in the United States is fraught with racism, including experimentation without consent, forced sterilization, and high maternal mortality (10-12); the legacy and current landscape of this mistreatment reverberate through health-care systems and impact cervical cancer prevention and treatment (13).

Unpacking the numerous contributors to existing cervical cancer disparities is a challenging but necessary step toward correcting these injustices. Simulation modeling methods have guided population-level clinical and policy recommendations around human papillomavirus (HPV) vaccination, cervical cancer screening, follow-up, and treatment (14–16). The development of models specifically reflecting Black populations may help deconstruct process failures in the cancer care continuum, demonstrate long-term implications of these failures on health outcomes, and identify effective intervention points to reduce disparities (17). We adapted and validated a well-established Cancer Intervention and Surveillance Network model of cervical cancer with data specific to Black women and used the model to estimate the contribution of observable differences in practice across the continuum of cancer care to disparities in cervical cancer incidence and mortality. The results are intended to inform discussions about clinical and policy approaches to reduce racial disparities in cervical cancer incidence and mortality.

Methods

Model overview

We adapted an existing Cancer Intervention and Surveillance Network (CISNET) microsimulation model of HPV infection and cervical carcinogenesis in the United States (15,18,19) to incorporate data specific to Black women (see Table 1). Our model simulates a female birth cohort from age 9 years until death with transitions through health states reflecting clinically relevant stages of HPV infection and cervical carcinogenesis, including acquisition of HPV, clearance of HPV, progression of HPV infection to cervical precancer, invasion of precancerous lesions to cervical cancer, mortality from cervical cancer, and mortality from noncervical cancer causes. The model reflects key health-care processes including hysterectomy, screening, diagnosis, and treatment of precancerous lesions and cancer.

Table 1.

Summary of model input parameters^a

Inputs	Data source	Years of data	Race definition
Demographics
Annual mortality rate	NCHS	1960-2017	Black by death certificate
Hysterectomy rate	BRFSS (25)	2011, 2018	Non-Hispanic Black by self-report
Cervical cancer screening (Table 2)
Proportion never screened	NHIS	2019	Non-Hispanic Black by self-report
Screen frequency	METRICS (28)	2016-2019	Non-Hispanic Black by medical record
Screen method	METRICS (28)	2016-2019	Non-Hispanic Black by medical record
Follow-up from abnormal finding	METRICS (28)	2016-2019	Non-Hispanic Black by medical record
HPV—calibration
HPV prevalence (Supplementary Figure 1, available online)	NMHPVR	2007-2009	Not stratified by Race
HPV prevalence - Sensitivity (Supplementary Figure 1, available online)	NHANES	2002-2006	Non-Hispanic Black by self-report
HPV type distribution - Sensitivity	Meta-analysis (34)	1993-2015	Non-Hispanic Black by medical record
Cervical cancer outcomes
Stage-specific survival by age (Supplementary Table 1, available online)	SEER 13	2002-2016	Black by SEER registry data
Incidence by age (Figure 1, A)	SEER 13	2000-2018	Black by SEER registry data
Mortality by age (Figure 1, B)	SEER 13	2000-2017	Black by SEER registry data

^aBRFSS = Behavioral Risk Factor Surveillance Survey; METRICS = Multilevel Optimization of the Cervical Cancer Screening Process in Diverse Settings and Populations; NCHS = National Center for Health Statistics; NHANES = National Health and Nutrition Examination Survey; NHIS = National Health Interview Survey; NMHPVPR = New Mexico HPV Pap Registry; SEER = Surveillance Epidemiology and End Results.

Conceptualization of Race

To adapt our model, we obtained input data across multiple sources for Black women in the United States, relying on nationally representative data when possible. The majority of our data sources ask participants to self-report race; exceptions to this are noted in Table 1. When possible, we defined Black inputs specifically as non-Hispanic Black individuals. Race is a social construct and is not meant to imply biological differences between populations (20). In our model, the measure of Black race is meant to reflect individual and community-level experiences of discrimination (individualized racism) as well as the impact of current and historic policies on health and health care (structural racism) that may affect cervical cancer rates and outcomes (4,5,21). We compare the Black model with our existing model (all-race model), an approach that is inherently attenuated relative to the full Black–White disparity in the United States but allows us to demonstrate the gap between existing approaches using population-level data and a fully specified model reflecting the multilevel impacts of racism.

Model input parameters

Demographics

Data on all-cause mortality by age were obtained for the 1970 birth cohort using life tables from the National Center for Health Statistics, with race defined by death certificate as previously described (22-24). For years beyond available data (2017), mortality was projected by modeling historic National Center for Health Statistics mortality data in single-year ages. Data on hysterectomy rates for the 1970 birth cohort for Black women and all women were obtained from a published study from the 2010 and 2018 Behavioral Risk Factor Surveillance Survey (25).

Cervical cancer screening

Cervical cancer screening practices for all women and Black women were obtained from 2 sources: the 2019 National Health Interview Survey and the Multilevel Optimization of the Cervical Cancer Screening Process in Diverse Settings and Populations (METRICS) consortium (26,27). To estimate the proportion of the population that is unscreened, we used Black-race–specific data on self-report of ever having received cervical cancer screening from the National Health Interview Survey (see Table 2) (28). To assess differences in the modality of screening test by Black race, we evaluated use of Pap test only vs Pap and HPV DNA co-testing among a cohort of average-risk individuals aged 30-65 years enrolled in 1 of 3 sites of the METRICS consortium, using the race reported in the patients’ medical record (29). To assess screening frequency, we used screening interval data from the METRICS consortium to estimate the proportion of individuals testing at 1-, 2-, 3-, 4-, or 5-year intervals by Black race and screening modality, after conditioning for those who ever screen. Finally, among those who had an abnormal result from a Pap smear or co-test, we estimated the Black-race–specific and all-race probability of follow-up colposcopy or treatment of precancerous lesions within 6 months (see Table 2). Sensitivity and specificity of Pap and HPV testing were based on data from published trials and did not vary by race.

Table 2.

Cervical cancer screening inputs

Input	All-race model, %	Black-race model, %
Test method
Pap only (vs HPV co-test)	47.0	56.0
Screen frequency
Never	12.6	14.6
Pap test only
Annual Pap	9.1	7.2
every 24 month Pap	10.6	9.7
every 36 month Pap	41.5	35.2
every 48 month Pap	12.2	14.4
every 60 month Pap	14.0	18.8
Co-test
Annual co-test	6.1	4.7
every 24 month co-test	7.9	6.7
every 36 month co-test	27.1	27.0
every 48 month co-test	28.8	26.8
every 60 month co-test	17.5	20.2
Follow-up to colposcopy and treatment of precancer	73.0	65.0

Cervical cancer survival

Conditional relative survival by race (for Black- and all-race women), age at diagnosis (younger than 40 years, 40-59 years, 60 years and older), and stage at diagnosis were obtained from the Surveillance Epidemiology, and End Results (SEER) 13 database using invasive squamous cell carcinoma cases from 2002 to 2008 with follow-up data through 2016 (30,31) (Supplementary Table 1, available online). These data were converted to a monthly conditional probability of death by race, age, and stage at diagnosis for input into the model.

HPV prevalence

HPV prevalence data were obtained from 2 sources. First, we obtained data from the New Mexico HPV Pap Registry (NMHPVPR) from 2007 to 2009, also used in the development and calibration of our original all-race model (32). HPV genotype data from NMHPVPR were not available by race, therefore we also analyzed the 2003-2006 National Health and Nutrition Examination Surveys (NHANES), using linear array to assess HPV prevalence from vaginal swabs (33). We compared HPV prevalence by race, genotype (HPV-16, HPV-18, and all other high-risk types), and age. Though a relatively small sample, NHANES does not show statistically significant differences in HPV prevalence by Black race (Supplementary Figure 1, available online). Further, a recent meta-analysis showed that although there are some differences in HPV type distribution in precancerous disease, type distribution by race in squamous cell carcinoma is not statistically significant (34). Therefore, our primary analysis uses NMHPVPR data, assuming no racial differences in HPV prevalence or type distribution (Supplementary Figure 1, available online ). In sensitivity analysis, we repeat the analysis using the model recalibrated to fit race-stratified NHANES data.

Calibration and comparison to SEER

Because parts of the carcinogenesis process are unobservable and cannot be estimated directly, we use a process of calibration: identifying multiple plausible values for uncertain natural history parameters that fit produce a good model fit to target data. For our primary analysis, we used an existing calibration of our all-race model, taking the top 25 best-fitting sets using previously described calibration methods (19). We then fix all cervical cancer natural history variables (HPV acquisition and clearance, progression and regression, and natural immunity) to ensure an identical disease process in the all-race and Black-race models and run all comparisons in matched model pairs, incorporating Black-race inputs of all-cause mortality by age, hysterectomy by age, cancer screening parameters (frequency, modality, and follow-up), and cancer survival by stage and age of diagnosis.

We compared model outputs (median, minimum, and maximum of the 25 parameter sets) for age-adjusted cervical cancer incidence and mortality (squamous cell carcinoma) for the all-race and Black-race models against SEER 13 (1992-2017), using 5-year averages across this period and comparing with the median, minimum, and maximum of this range. For comparison to SEER, we did not adjust for hysterectomy in the denominator, which is consistent with SEER’s inclusion of all women in the at-risk population, regardless of hysterectomy. However, given racial differences in hysterectomy rates, we hysterectomy adjusted in our final model estimates.

In sensitivity analysis, we independently recalibrated models to fit to HPV prevalence values reported from NHANES. Using an existing repository of alternative natural history parameter sets, generated for a prior calibration (19), we compared outcomes estimated by each to NHANES data on HPV prevalence and race-specific estimates of HPV type distribution in grade 3 cervical intraepithelial neoplasia (CIN3) and cancer. We identified the top 25 parameter sets fitting to the all-race data and the top 25 sets fitting to the Black-race data. We compared the model estimates of absolute and relative disparities between the Black-race and all-race models using these parameter sets.

Analysis

To understand the relative contribution of each input, we started with our all-race model and sequentially substituted Black input variables across each of the 25 natural history parameter sets. We reported model projections of cervical cancer incidence and mortality in each step and compared the incremental change in the disparity after introduction of each input with the total disparity observed when comparing the full all-race and Black-race models. In our base case, we replaced inputs in the following sequential manner: all-cause mortality, hysterectomy rates, screening variables for modality, frequency, follow-up, and cancer survival. We then reported a mean, minimum, and maximum value across the 25 natural history parameter sets. In sensitivity analysis, we replicated the replacement process 500 times, randomizing the order of input replacement in each run to remove the impact of input ordering. For each of these 500 replications, we calculated the mean proportion of the disparity attributable to each input (averaging the 25 parameter sets) and then reported the mean, minimum, and maximum value across the 500 replications. The model inputs, calibration data and simulated results that support the findings of this study are available from the corresponding author upon reasonable request.

Results

Model validation

Our model demonstrated reasonable comparison against historic SEER data on incidence, with our all-race model estimating 8.6 (range = 6.1-9.2) cases per 100 000 and our Black-race model estimating 10.7 (range = 7.5-11.6) cases per 100 000, both within the bounds of age-adjusted SEER estimates from 1992 to 2017. (Figure 1, A). We found a similar pattern for mortality, with a death rate of 2.9 (range = 2.0-3.1) per 100 000 in our all-race model and 5.5 (range = 4.0-5.8) per 100 000 in our Black-race model (Figure 1, B). After adjusting the at-risk population denominator for hysterectomies, our final estimates of incidence were 8.6 (range = 6.1-9.2) and 10.8 (range = 7.6-11.7) cases per 100 000 for all and Black women, respectively, and estimates of mortality were 2.9 (range = 2.0-3.1) and 5.5 (range = 4.0-5.8) deaths per 100 000 for all and Black women, respectively.

Figure 1.

Comparison of model outputs of cervical cancer incidence and mortality to Surveillance, Epidemiology, and End Results (SEER; age standardized). Model data reflects the individual estimates (points) and distribution (boxplots) of age-adjusted incidence and mortality across 25 natural history parameter sets. SEER data reflect age-adjusted incidence and mortality for squamous cell carcinoma in 5-year intervals.

Sequential replacement

In our sequential replacement of input variables (Figure 2; Table 3), the model projected little difference from the base all-race model when adjusting for all-cause mortality by race. Median estimates also remained stable when adding Black women hysterectomy rates. Lower use of co-testing relative to Pap-only testing among Black women increased disparities in both incidence (from 8.6 to 9.0 per 100 000) and mortality (from 2.9 to 3.0 per 100 000). Differences in screen frequency further increased estimated incidence (to 9.7 per 100 000) and mortality (to 3.3 per 100 000). When adding in differences in receipt of follow-up care, the estimated incidence increased to 10.8 per 100 000 and estimated mortality increased to 3.6 per 100 000. Differences in stage-specific survival on mortality was large, increasing the final estimated mortality for Black women to 5.5 per 100 000. Comparing the Black model with the all-race model produced a median relative risk for cervical cancer incidence of 1.24 (range = 1.23-1.27) and for cervical cancer mortality of 1.92 (range = 1.85-2.00; Figure 2).

Figure 2.

Estimated cervical cancer incidence and mortality in sequentially replaced models. Reflects variation across 25 paired sets of model parameters; each model adds an additional Black model parameter, keeping all previous replacements for incidence (A and C) and mortality (B and D). CC = cervical cancer.

Table 3.

Attribution of cervical cancer disparities^a

Estimate	Incidence per 100 000 (Range)	Percent attributed - Incidence (Range)	Morality per 100 000 (Range)	Percent attributed - Mortality (Range)
Model estimate, all	8.58 (6.15-9.20)		2.86 (2.04-3.10)
Demographics (all-cause mortality)	8.61 (6.16-9.27)	1.0% (−1.8%, 2.9%)	2.86 (2.03-3.08)	0.0% (−0.9%, 1.0%)
Demographics (hysterectomy rates)	8.65 (6.20-9.32)	2.2% (0.6%-3.5%)	2.90 (2.06-3.14)	2.1% (0.7%-2.8%)
Screening (modality)	9.01 (6.42-9.72)	16.4% (14.1%-20.1%)	3.04 (2.14-3.30)	4.9% (4.1%-5.8%)
Screening (frequency)	9.71 (6.85-10.52)	33.7% (28.8%-35.5%)	3.28 (2.30-3.55)	9.8% (8.0%-11.9%)
Screening (follow-up)	10.77 (7.57-11.67)	47.3% (43.9-51.1%)	3.59 (2.53-3.93)	12.7% (11.0%-14.4%)
Treatment (stage-specific survival)	—	—	5.49 (4.02-5.84)	70.3% (67.0%-75.3%)
Model estimate, Black	10.77 (7.57-11.67)		5.49 (4.02-5.84)

^aShows the mean, minimum, and maximum estimate for estimated cervical cancer incidence and mortality and proportional attribution across 25 paired parameter sets. Proportional attribution reflects the percent (of the total estimated excess cases per 100 000 women); each component is estimated to contribute.

Attribution of disparities

Using these estimated differences to calculate the relative contribution of each input to the observed disparities, we found the largest driver of cervical cancer incidence disparities was screening follow-up, accounting for 47.3% (range = 43.9%-51.1%) of the total disparity (Table 3). Screening frequency accounted for 32.7% (range = 28.7%-35.5%) of the disparity, and modality of screening used (Pap only vs co-test) accounted for 16.4% (range = 14.1%-20.1%).

For mortality disparities, the vast majority of differences was driven by stage-specific survival (70.3%, range = 67.0%-75.3%) followed by follow-up from screening (12.7%, range = 11.0%-14.4%) and screening frequency (9.8%, range = 8.0%-11.9%). In sensitivity analysis, we recalculated these percentages in randomly re-ordered versions of input replacement to remove the effect of order. Across 500 randomly re-ordered runs, we found the magnitude of the impact was similar for all variables, suggesting ordering has minimal impact on conclusions (Supplementary Table 2, available online).

Sensitivity analysis: NHANES data

When using models that were calibrated to HPV prevalence data from NHANES, projected cervical cancer incidence and mortality rates increased in all-race and Black-race models, with estimated incidence at 12.6 (range = 9.1-15.7) per 100 000 in the all-race model and 14.7 (range = 10.6-18.6) per 100 000 in the Black-race model and mortality at 4.1 (range = 2.7-4.8) per 100 000 in the all-race and 7.7 (range = 5.1-9.4) per 100 000 in the Black-race model (Supplementary Figure 2, available online ). However, the relative disparities estimated by the models using NHANES data were similar to our base-case model that was fitted to HPV prevalence data from the NMHPVPR, with a relative risk of 1.21 for incidence (vs 1.24 in our base case) and 1.91 for mortality (vs 1.92 in our base case).

Discussion

Using our adapted all-race model of cervical carcinogenesis to reflect inputs specific to Black women in the United States, we were able to replicate observed racial disparities in cervical cancer incidence and mortality with no changes in natural history assumptions or model structure, underscoring the importance of process failures in driving observed disparities in cervical cancer by Black race. Incidence disparities were largely driven by differences in screening frequency and timely follow-up by race, whereas mortality disparities were primarily driven by differences in stage-specific survival, emphasizing the need to improve care access and quality along the full continuum of cervical cancer care.

Our measure of stage-specific survival from SEER provides an indirect measure of treatment access and quality, including delays in receipt of care or differences in receipt of guideline-recommended treatment, which have been described previously (35-37). In addition to addressing the timeliness and quality of cancer treatment, it is important to address barriers to follow-up from abnormal findings (38,39), which can also result in delayed treatment of cancer or precancerous lesions. Finally, we note that differences in hysterectomy rates by race are important to consider, not as risk factors themselves but as an important source of bias in our national estimates. As SEER data do not account for hysterectomy rates, these estimates are underestimating not only the risk of cervical cancer in the at-risk population but also racial disparities in cervical cancer outcomes by as much as 40% (5,40).

Inequities in care access and quality by Black race are products of systematic racism including social policies and disinvestment in communities, producing gaps in insurance coverage, funding, and support within the health-care system and access to health-promoting resources (4,5,41). Ongoing research seeks to quantitatively link these larger structural factors to specific care process measures such as screening usage and follow-up. Estimates of effects derived from these analyses can be incorporated into models to help generate data on actionable strategies at a system level (42,43). It is important to acknowledge that system-level failures are best addressed by system-wide solutions. Addressing inequities in insurance coverage and investing in underresourced health systems and communities should be a top priority for equity efforts.

However, in the absence of political will for system-level interventions, disparities may be reduced through more proximal interventions to temporarily overcome specific structural barriers. Targeted screening programs and novel screening methods like HPV self-collection may help improve screening uptake (39,44-46). Disparities in follow-up and treatment may be reduced through individual-level interventions such as patient navigation (44,47), but larger interventions targeting inequities within or across clinics may be required, particularly as differences in time to treatment initiation and multiple measures of treatment quality are well-documented for Black women (35,48).

Our model estimates the impact on squamous cell carcinoma—the histological type responsible for the majority of all cervical cancers. However, evidence suggests that adenocarcinoma is increasing as a proportion of diagnosed cervical cancers over time (49) and that patterns of incidence and mortality from adenocarcinoma may differ for Black women in important ways (50). Future work should seek to better understand these patterns and their implication for reducing disparities in cervical cancer incidence and mortality.

Our model is able to reasonably replicate historic SEER data on incidence and mortality, but cervical cancer incidence and mortality are both declining in the United State (1), and all evidence suggests future rates will further decline as a result of HPV vaccination (51). We do not incorporate HPV vaccination into this analysis as available incidence and mortality data primarily reflect unvaccinated cohorts. It is promising that Black adolescents have similarly high or higher rates of HPV vaccination than non-Hispanic White adolescents (52), however, prior work suggests that HPV vaccination alone is unlikely to close racial disparities in cervical cancer incidence or mortality (53). Additionally, there may be differences in HPV type distribution by race (34,54). Our sensitivity analysis incorporating the best available data on HPV type distribution in prevalent infections, precancerous lesions, and invasive squamous cell carcinoma among Black women provided evidence that potential differences in type distribution are not large contributors to existing disparities, but better data are needed to understand where differences are meaningful and whether this will change in the context of HPV vaccination.

Our model reflects a comparison between Black-race and all-race models, which is an underestimate of the full disparity between Black individuals and non-Hispanic White individuals. Development of a White-race model, along with a model reflecting Hispanic ethnicity and other race or ethnic groups, may provide additional insights on reasons for disparities between these groups. Our current analysis seeks through comparison to the standard all-race model, emphasizing specifically what is missed for self-identified Black individuals when using population-level data.

Our study has several limitations that should be considered in evaluating our results. First, our model data come from a variety of sources reflecting different populations and time periods. Our mortality and hysterectomy data reflect a 1970 birth cohort, but high-quality historic data on HPV prevalence and time-varying screening patterns for this birth cohort are not available, nor do we have cancer incidence or mortality data reflective of this cohort’s risk over their full lifetime. Therefore, we do not suggest our model output is predictive for a specific cohort or time window but rather is broadly reflective of racial differences across the lifespan, using the best data available. Our estimates of hysterectomy rates do not specify reasons for the procedure and therefore may include those who received a hysterectomy as a treatment for cervical cancer or precancer. Data suggest a malignant indication for approximately 10% of hysterectomies (55), so although we may overestimate hysterectomy-adjusted cervical cancer rates as a result, any overestimation is likely small. Similarly, our estimates of screening come from electronic medical records reflecting 3 sites, which may not be generalizable to screening behavior in the United States. Although these sites reflect a variety of insurance statuses and location of care (28), they may reflect populations with more regular access to care than the general US population.

Although a single model cannot capture the multilevel, multigenerational factors influencing cervical cancer outcomes in the United States, our adapted model provides insight on key process failures (screening, follow-up, and treatment) driving cervical cancer disparities among self-identified Black individuals in the United States. To make progress in cancer equity, it is crucial to address differences in cancer care underlying these disparities, as well as take action to dismantle the larger structural factors that produce differential care access. Decision models can help identify where and how policy, practice, and care-focused interventions may most effectively and efficiently reduce these disparities, but more quantitative work is needed to inform efforts to connect structural inequities and large-scale policy interventions to cancer health equity.

Author contributions

Jennifer Spencer, PhD (Conceptualization; Formal analysis; Writing—original draft), Emily A Burger, PhD (Conceptualization; Formal analysis; Writing—review & editing), Nicole G Campos, PhD (Conceptualization; Formal analysis; Writing—review & editing), Mary Caroline Regan, MS (Data curation; Methodology; Project administration; Writing—review & editing), Stephen Sy, MPH (Conceptualization; Formal analysis; Methodology; Software; Writing—review & editing), and Jane J Kim, PhD (Conceptualization; Formal analysis; Funding acquisition; Supervision; Writing—original draft; Writing—review & editing).

Funding

This work was supported by the National Cancer Institute (U01CA253912).

Monograph sponsorship

This article appears as part of the monograph “Reducing Disparities to Achieve Cancer Health Equity: Using Simulation Modeling to Inform Policy and Practice Change,” sponsored by the National Cancer Institute, National Institutes of Health ([Comparative Modeling to Inform Cervical Cancer Control Policies; 3 U01 CA253912-03S1]).

Conflicts of interest

Authors have no conflicts of interest to report.