Trends and Racial/Ethnic Differences in Predictors of Cervical Cancer Screening among US Women Ages 30–64 Years

Cervical cancer is one of the most preventable female cancers. Population-based screening for cervical cancer has been shown to reduce incidence and mortality from the disease (1–3). Unfortunately, despite scientific advancements in prevention and treatment, an estimated 604,000 new cases of cervical cancer were diagnosed, and 342,000 women died from the disease globally in 2020 (4). In the United States (US), approximately 14,480 women were diagnosed with cervical cancer, and more than 4,200 died from the disease in 2021 (5). Screening for high-risk human papillomavirus (HPV) infection and precancerous or early neoplastic changes in the cervix using HPV-based and cytology-based (Papanicolaou/Pap) tests presents invaluable opportunities for interventions to reduce incidence and mortality from the disease (1–3).

In 2018, the World Health Organization (WHO) made a global call to eliminate cervical cancer by 2030 using multiple strategies, including efficient screening of 70% of all women by 35 and 45 years of age (6). In addition, the 2018 US Preventive Services Task Force (USPSTF) guidelines recommendations include: (i) women ages 21 to 29 years—a Pap test alone every three years; and (ii) women ages 30 to 65 years Pap test alone every three years, an HPV test alone every five years, or HPV/Pap co-test every five years (7). Whereas in 2020, the American Cancer Society (ACS) updated its cervical cancer screening (CCS) recommendations (8). The new guidelines differ from the previously published 2012 ACS guidelines. Specifically, ACS now recommends starting CCS at age 25 instead of 21 and an HPV test every five years as the preferred screening for women ages 25 to 65 years. However, if this is not available, an HPV/Pap co-test every five years or a Pap test alone every three years can be used (8).

Although CCS rates increased in the late 1980s and 1990s (9, 10), rates have decreased over the last two decades (11). Overall, screening rates are lower among racial minority women, and this is believed to contribute to the increased mortality among racial minority groups (12–15). In addition, factors affecting minority women, such as higher prevalence of socioeconomic disadvantages and lack of access to preventive or medical services, may contribute to CCS disparities for this vulnerable population (16–18). However, trends in screening based on recently updated guidelines (7, 8) and determinants of CCS among specific racial/ethnic subgroups remain unknown. Hence, the primary aim of this study was to examine the most recent trends in self-reported CCS utilization from 2016 to 2020 and determine predictors of CCS participation across racial/ethnic subgroups. In addition, because there are different CCS methods, we also sought to understand changes in Pap and HPV DNA testing over time. The findings of this study could help identify persisting disparities in CCS completion among groups and inform the development of race-specific and targeted interventions to increase CCS among at-risk populations.

In this cross-sectional study, we analyzed data from the National Behavioral Risk Factor Surveillance System (BRFSS) 2016, 2018, and 2020. BRFSS is the largest annual population-based, random-digit–dialed telephone health survey of non-institutionalized, US civilian adults ages 18 years or older (19). The surveys are conducted by the Centers for Disease Control and Prevention (CDC) in collaboration with the state health departments in all 50 states, the District of Columbia, and participating US territories (19). BRFSS uses a complex sampling weighting method to address potential selection bias, including non-response and non-coverage. The median response rate of the surveys included in this analysis was 47.0% in 2016, 49.9% in 2018, and 47.9% in 2020. Additional details on the sampling method, response rates, data quality, weighting, methods of analysis of BRFSS have been published elsewhere (19, 20).

BRFSS collects CCS screening-related data nationally on even-numbered years. To assess trends and predictors of CCS utilization among women ages 30–64 years without hysterectomy, we pooled and analyzed data from the 2016 (n = 472,318), 2018 (n = 418,474), and 2020 (n = 389,826) BRFSS surveys. Survey weight for each year was provided by BRFSS, and we calculated a final weight for the pooled data based on the proportion of sample size from each survey year using the methodology described previously by the CDC (19). We included 235,713 women, ages 30–64 years, without previous hysterectomies, or missing outcome or selected variables (Supplementary Fig. S1). This study was excluded from the institutional review board because it involved the analysis of publicly available, deidentified data, with no direct human subject contact.

The primary dependent variable was up-to-date CCS. To determine CCS utilization, respondents were initially asked: Have you ever had a Pap test? Then those with an affirmative answer to this question were asked: How long has it been since you had your last Pap test? To assess HPV testing participation, respondents were asked: An HPV test is sometimes given with the Pap test for CCS. Have you ever had an HPV test? Those with a positive response were asked: How long has it been since you had your last HPV test? (19).

We then considered respondents as: (i) Having up-to-date CCS if they reported a Pap test within the past three years or an HPV test within the past five years, (ii) having received a Pap test, if they reported Pap testing within the past three years, and (iii) having received an HPV test if they reported an HPV test within the past 5 years. Respondents who refused to answer the questions, were unsure, or had missing response to the main outcome (up-to-date CCS) were excluded from this analysis (Supplementary Fig. S1). The outcomes were measured as binary variables (Yes/No).

We examined predictors of CCS within racial/ethnic subgroups, categorized as non-Hispanic Whites (NHW), non-Hispanic Blacks (NHB), Hispanic, non-Hispanic American Indians or Alaskan Natives (NH AI/AN), non-Hispanic Asians (NH Asian), and non-Hispanic Other (NHO). In addition, the NH Asian group included non-Hispanic Asians and other Pacific Islanders or Native Hawaiian, while NHO comprises non-Hispanic other races and multiracial participants.

Potential predictors of CCS were: Age (30–39, 40–49, 50–64 years); marital status (married/cohabiting, not married); educational attainment (less than high school degree, high school degree or equivalent, some college or more); employment status (employed and unemployed); annual household income [less than $15,000, $15,000-$49,999, $50,000 or more, and missing (due to high proportion of missing income value)]; health insurance coverage (yes, no); general health (good/better, poor/fair); personal physician (yes, no); routine check-up attendance (less than one year, one year to less than five years, greater than five years); and had a mammogram (yes, no), and survey year.

We described sample characteristics, both overall and stratified by race/ethnicity. To examine the trends in recommended CCS (up-to-date CCS, Pap testing, and HPV testing) completion, we first estimated the weighted proportion of self-reported CCS, overall, by survey year, and by race/ethnicity. Second, we examined the changes in CCS completion from 2016 to 2020, overall and by racial/ethnic groups, using simple linear regression with t-statistics testing the null hypothesis that there was no change in screening over time. Survey year was used as the independent variable (2016 as “1,” 2018 as “2,” and 2020 as “3”), and we adjusted for age, education, income, and health insurance that are known to influence cancer screening (21, 22). We report the corresponding t-statistic and P value. Third, we used multivariable logistic regression modeling to assess race/ethnicity differences in predictors of CCS utilization, while controlling for survey year. We included only variables that were significant (P < 0.05) from the bivariate logistic regression in the multivariable models. Potential interactions were assessed by fitting the interaction term between race/ethnicity and SES variables (income, education, and employment) into the model. Likelihood ratio tests of nested models with and without interaction term were performed. To assess statistical significance of differences in associations across the strata of host characteristics, we assessed the P value for the type III analysis of effects for the interaction term. All analyses were conducted in Stata 17.0, and accounted for the complex survey design with the calculated final survey weights, sampling units, and strata using a survey package in Stata. The significance level was set at 5% throughout the study.

The data analyzed in this study were obtained from BRFSS and are publicly available at: https://www.cdc.gov/brfss/index.html.

Table 1 shows the descriptive characteristics and weighted proportions of US women ages 30–64 years in BRFSS 2016 to 2020 overall and by race/ethnicity. Our weighted sample was 60.8% NHW, 12.2% NHB, 18.7% Hispanic, 0.9% AI/AN, 5.7% NH Asian, and 1.8% identified as other race. The majority was 40 years or older. Hispanic women had the highest proportion of women with low education, unemployment, low income, and no personal doctor. In addition, a larger proportion of NH Asian women reported having higher educational attainments compared with other racial/ethnic subgroups.

Table 2 depicts the temporal trends in weighted proportions of women with up-to-date CCS and by specific screening methods, from 2016 to 2020 overall and by race/ethnicity. There was minimal change in CCS rates over the 5-year study period. Overall, up-to-date CCS rates remained unchanged from 2016 to 2018 (84.2% vs. 84.6%), though rates showed a little downtick from 2018 to 2020 (84.6% vs. 83.3%). Although Pap testing declined over time (82.7% vs. 80.6%) from 2016 to 2020, HPV-based testing increased substantially from 2016 to 2020 overall (43.4% vs. 52.7%), and across subgroups, despite a little dip in 2020. Also, we found disparities in up-to-date CCS and by screening methods among NH Asian women (up-to-date CCS 74.4%, Pap testing 73.0%, and HPV testing 44.1%) compared with NHW women (up-to-date CCS 83.2%, Pap testing 80.8%, HPV testing 51.8%) in 2020.

Crude odds ratios (Table 3) suggested that across all racial/ethnic groups, household income, health insurance, having a personal physician, and routine check-up were associated with having up-to-date CCS. Whereas the relationships between up-to-date CCS and age, marital status, education, employment, and general health varied across groups. In the multivariable analysis, including all racial/ethnic groups, we found evidence for interaction between race/ethnicity and SES variables (income, P < 0.001; education, P < 0.001; and employment, P = 0.044). In the adjusted multivariable logistic regression models (Table 4), age and attending routine health check were consistent predictors of up-to-date CCS regardless of race/ethnicity. In addition, having a personal physician was linked to CCS completion for most subgroups except among NHO women. Other important findings from our data include, across all race/ethnic group, older women were less likely to have up-to-date CCS. On the contrary, women who attended wellness checks within the past year had approximately three to fourteen times the odds of following CCS guidelines.

Furthermore, previous mammogram use was associated with greater odds of up-to-date CCS in NHW [aOR, 2.97; 95% confidence interval (CI), 2.72–3.25], NHB (aOR, 2.75; 95% CI, 2.18–3.47), Hispanic (aOR, 2.74; 95% CI, 2.22–3.37), and NH Asian women (aOR, 3.78; 95% CI, 2.62–5.45; all P < 0.001). In contrast, there were variations in the link between socioeconomic status variables and up-to-date CCS across groups. In addition, predictors of HPV-based testing are available in the Tables 5 and 6. Likewise, correlates of receiving HPV-based testing varied but routine health examinations and age were consistent predictors for all racial/ethnic subgroups. We tested for collinearity among our predictor variables and the mean variance inflation factor for collinearity test was less than 2.

We examined recent trends in CCS and predictors by race/ethnicity among a nationally representative sample of US women ages 30 to 64 years, who had not undergone a hysterectomy, from 2016 to 2020. Overall, up-to-date CCS rates were stable from 2016 to 2018, with a slight, but significant decline from 84.6% in 2018 to 83.3% in 2020. Though Pap test rates dwindled overall and across racial/ethnic subgroups, there was an upsurge in HPV-based testing over time. These findings are consistent with previous studies (23–26).

An analysis of county-level data reported a similar decline in Pap test uptake (23). In addition, population-based studies using the National Health Interview Survey reported declines in Pap testing (25, 27). Silver and colleagues (24) examined yearly data on HPV and Pap tests participation in an academic medical center and observed a significant increase in HPV/Pap co-testing uptake. Another analysis using healthcare claims data reported a decline in overall CCS rates but an upsurge in HPV/Pap co-testing rates (26). The shift toward HPV-based screening may be due to evolving scientific knowledge on the higher accuracy of the method compared with Pap testing (28), in detecting high-risk HPV in cervical cells identified as a major causal factor in cervical cancer (29, 30).

Our analyses showed racial/ethnic differences in rates and predictors of up-to-date CCS and HPV-based testing. Studies have documented disparities in cancer-screening participation, including CCS among racially minoritized populations (11, 31). Furthermore, across all racial/ethnic subgroups, older women were less likely to receive timely screening. Past studies have documented low CCS participation among older women (14, 18, 21, 32). Unlike older women, younger women are more likely to have contact with health care providers for reproductive health reasons, including pregnancy care, family planning services, and thus may receive CCS (33, 34). In addition, older women's perceived susceptibility to cervical cancer and need for screening may be low (35). Yet older women are at higher risk of cervical cancer (14, 21), and poor screening behavior may reduce the opportunity to detect easily treatable cervical cancer. In our study, we found that women who engaged with healthcare services, particularly routine general examinations were more likely to be up to date with CCS. Understanding and addressing barriers to preventive services may help mitigate delayed or no participation in CCS and increase the opportunity to achieve the WHO global cervical cancer elimination goals (6).

In the present study, SES variables, and general health were not consistent predictors of up-to-date CCS and HPV testing for all subgroups. For example, NHW, NH AI/AN, and NH Asian women with higher incomes were more likely to receive timely CCS, whereas greater income was not a determinant for NHB, Hispanic, and NHO women. Notwithstanding, higher income was linked to HPV testing among NHB, Hispanic, and NH Asian women. In addition, the association between up-to-date CCS or HPV testing and income was stronger among NH Asian women compared with other race/ethnicity. Many studies in diverse settings, including in the US, United Kingdom, Italy, Spain, and Sweden (21, 32, 36–39), have suggested a link between higher education attainment and CCS. However, in our study, this was not true among Hispanic and NH AI/AN women. Our results suggest a need to understand both implicit and explicit barriers to timely CCS participation, including cultural, religious, and psychosocial factors across socioeconomic classes.

Our study has several limitations. First, all variables used in the analysis were measured with self-reported surveys and could not be verified by medical record review. Thus, it could be subject to recall bias and social desirability. Second, the BRFSS uses a cross-sectional survey, and we cannot determine clear temporality or make causal inferences from our findings. Third, we cannot rule out the possibility that HPV testing was under-reported. Many participants reported being unsure of receiving the test and were excluded from the HPV-testing analysis. Providers could inform and educate their clients about specific CCS tests and potential advantages of the tests being administered. Fourth, many national surveys, including BRFSS, have reported lower participation rates in recent years (40). Non-response bias is a potential limitation in the current study; however, we used weighted data in our analyses to minimize potential biases. Fifth, the time frame of our data from 2016 to 2020 may not reflect changes based on the most recently updated guidelines (7, 8), future studies could investigate changes before and after these updates. Finally, the 2020 data were collected during the COVID-19 pandemic, and due to the lockdown, people might have delayed or canceled preventive care utilization. Moreover, recent studies have suggested a substantial decline in cancer prevention services utilization during the COVID-19 pandemic, particularly during the 2020 lockdown order (41, 42). Despite these limitations, BRFSS is a large sample and has been shown to be reliable and valid (20) and outcomes of interest were measured for multiple years.

In conclusion, this population-based study provides essential updates on recent CCS utilization. The slight decrease in CCS rates in 2020 may be attributable to disruptions in preventive healthcare delivery during the COVID-19 pandemic lockdown order. The increase in HPV testing overall and across all racial/ethnic subgroups may reflect evolving scientific evidence for HPV as a preferred screening method. There is persisting disparity in CCS participation by age and race/ethnicity. It is of serious concern that older women at the greatest risk of being diagnosed with cervical cancer are less likely to be screened. There is a need for more exploratory research to understand CCS participation barriers that may be peculiar to these groups of women and their mechanisms of influence. In addition, though the link between CCS and socioeconomic factors varied by race/ethnicity, having regular healthcare providers and preventive services utilization were consistent predictors of up-to-date CCS. Policies funding inclusive and culturally relevant interventions and optimizing access to routine health checks could increase CCS participation across all groups.

No disclosures were reported.

The funders had no role in the design of the study, the collection, analyses or interpretation of the data, the writing of the article, or the decision to publish the results.

I.O. Sokale: Conceptualization, resources, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. J.R. Montealegre: Conceptualization, resources, funding acquisition, writing–review and editing. A.O. Oluyomi: Resources, writing–review and editing. A.P. Thrift: Conceptualization, resources, supervision, funding acquisition, methodology, project administration, writing–review and editing.

This work was funded by a research training award from the Cancer Prevention and Research Institute of Texas (CPRIT) for the Systems Epidemiology of Cancer Training (SECT) Program (RP210037; to I.O. Sokale, PI: A.P. Thrift). This research was also supported by a grant from the National Institute on Minority Health and Health Disparities (NIMHD, R01MD01375; to PI: J.R. Montealegre). This study was also supported by the facilities and resources of the Dan L. Duncan Comprehensive Cancer Center P30 CA125123.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC Section 1734.