U.S. national trends in cervical cancer screening by sexual orientation and race/ethnicity in cisgender women

Abstract

Introduction:

Changes in up-to-date cervical cancer screening (CCS) over time by sexual orientation and race/ethnicity were estimated to identify trends in screening disparities.

Methods:

This 2024 retrospective, cross-sectional analysis of National Health Interview Survey data (years 2013, 2015, 2019 and 2021) included 40,818 cisgender women aged 21–65 without hysterectomy. Joinpoint analysis was performed to calculate the annual percent change (APC) of up-to-date CCS from 2013 to 2021. Logistic regression (years 2019 and 2021) was used to describe the relationship between up-to-date screening and sexual orientation, race/ethnicity, and the interaction thereof.

Results:

The adjusted odds of up-to-date CCS were 22 % lower for lesbian, gay, and/or bisexual (LGB) compared to heterosexual women (OR = 0.78, p = 0.01). Up-to-date CCS fell significantly from 80.50 % in 2013 to 75.00 % in 2021 for heterosexual respondents (APC = −0.97 %, p < 0.01), but was stable across years for LGB respondents. Up-to-date CCS decreased for Hispanic (APC = −1.52, p < 01) and non-Hispanic White only heterosexual women (APC = −0.63, p = 0.02). It also decreased for non-Hispanic Black/African American only LGB women (APC = −2.67, p < 0.01) falling from 85.22 % in 2013 to 67.91 % in 2021. By multiplicative interaction, LGB Hispanic women were more up-to-date than their heterosexual counterparts (p = 0.05).

Conclusions:

In 2021 there were approximately 19.72 million women aged 21–65 who were not up-to-date with CCS. 1.76 million LGB women were not up-to-date for CCS, and a greater proportion of these women identified as non-Hispanic Black/African American. CCS must be improved for all cisgender women, and specific attention should be given to those who identify as LGB and/or Black/African American.

1. Introduction

Early detection of cervical cancer via cervical cancer screening is highly effective at preventing mortality and is recommended by the US Preventive Services Task Force (US Preventive Services Task Force, 2018). Despite the wide availability of cervical cancer screening in the United States, systemic racism and discrimination in the healthcare system, poor cervical cancer screening insurance coverage, inequitable healthcare access, and anti-LGBTQIA+ stigma have caused cervical cancer screening services to be underutilized by marginalized populations (Ramsey et al., 2022; Spencer et al., 2023; Baumann et al., 2024). Black and Hispanic/Latina women are disproportionately burdened by cervical cancer incidence and mortality compared to non-Hispanic or Latina White women (CDC, 2020; Watson et al., 2017; State Cancer Profiles, 2024). LGBTQIA+ populations also face unique barriers to cervical cancer screening, since both patients and providers often underestimate the risk of contracting high-risk human papillomavirus (HPV) and therefore do not screen as often as recommended (Tracy et al., 2010; Matthews et al., 2004; Curmi et al., 2014).

Unfortunately, up-to-date screening prevalence has been declining in the United States in recent decades across all age groups (Watson et al., 2017), and in a 2021 report by the National Cancer Institute, only 72.4 % of women age 21–65 were up-to-date (National Cancer Institute, 2024), well below the Healthy People 2030 goal of 84.3 % (Ward et al., 2019). The objective of this present study was to estimate changes in guideline-concordant, up-to-date cervical cancer screening over time by sexual orientation and race/ethnicity using National Health Interview Survey (NHIS) data from 2013, 2015, 2019 and 2021 (full cohort) to better understand national trends in cervical cancer screening for populations marginalized on the basis of race and sexual orientation. Note that there was not sufficient data to explore these trends for gender minority populations, however inequities in cervical cancer screening may exist.

2. Methods

Data from the NHIS, a robust database of health information designed to monitor the health of the U.S. noninstitutionalized population, were used in this retrospective, population-based, cross-sectional analysis (NHIS, 2023). Survey years 2013, 2015, 2019, and 2021 were included because they contained the required cervical cancer screening and demographic variables for the analysis.

Respondents included cisgender (denotes gender identity corresponding with sex assigned at birth) women aged 21–65 without history of total hysterectomy. The total cohort of respondents was 40,818. Respondents who self-identified as lesbian, gay, or bisexual or an identity other than straight, lesbian, or bisexual were coded as lesbian, gay, and/or bisexual (LGB). Respondents who self-identified as straight were coded as heterosexual. Race and ethnicity (Hispanic, non-Hispanic White only, non-Hispanic Black/African American only, non-Hispanic Asian only, non-Hispanic other single or multiple races) were self-reported by survey respondents. Self-reported covariates also included age (categorized as age 21–29, 30–39, 40–49, and 50–65) and insurance status (Private, Medicaid and other public, Other insurance, Uninsured, Missing). “Medicaid and other public” insurance status includes Medicare and state-sponsored health insurance. “Other insurance” includes Military, Indian Health Service, single service insurance, and Medi-Gap, which is supplemental insurance coverage that is purchased by Medicare beneficiaries for out-of-pocket expenses not covered by Original Medicare.

The prevalence of up-to-date cervical cancer screening and odds ratios between sexual orientation and race/ethnicity categories were calculated. Up-to-date cervical cancer screening was determined based on NHIS questions “When did you have your most recent Pap test?” and “Did you have a human papillomavirus (HPV) test with your most recent Pap test?”. If the most recent screening test was within the intervals defined by the US Preventive Services Task Force screening guidelines for respective ages and modalities, they were considered up-to-date (US Preventive Services Task Force, 2018). If the respondent indicated having ever received a Pap or HPV test by the survey question, “Have you ever had a test to check for cervical cancer?”, they were considered ever-screened.

SAS Survey procedures were utilized to estimate the unweighted and weighted prevalence of up-to-date screening and ever-screening among the LGB and heterosexual populations by year and further analyzed the data by race/ethnicity subgroups. Unweighted results represent raw observation counts in the dataset and corresponding frequencies, while weighted results represent estimates and proportions representative of the U.S. population (N = 308,897,709). Data were weighted using pooled weights to adjust for unequal probability of selection and nonresponse.

Piecewise log-linear time calendar trends in the proportion of up-to-date cervical cancer screening were modeled and the annual percent change was calculated from 2013 to 2021 (Joinpoint Regression Program, 2024). Logistic regression models with survey years 2019 and 2021 (n = 17,944) were also generated to describe the relationship between up-to-date screening and year, sexual orientation, race/ethnicity, age, and insurance status. Unfortunately, years 2013 and 2015 could not be combined with years 2019 and 2021 due to significant changes in survey weighting starting in 2019 (IPUMS Center for Data Integration. IPUMS NHIS, 2024; Bramlett et al., 2020).

Additive and multiplicative interaction was assessed using logistic regression modeling and methods described by Knol and VanderWeele (Knol and VanderWeele, 2012). The model described the relationship between up-to-date cervical cancer screening and sexual orientation, race/ethnicity, and the interaction thereof, adjusted by year, age, and insurance status. Multiplicative interaction was assessed by the significance of the interaction (product) term in the regression model, and additive interaction was assessed by calculation of relative excess risk due to interaction (RERI), proportion attributable to interaction (AP) the synergy index using methods described by Hosmer and Lemeshow (Hosmer and Lemeshow, 1992).

All p-values were from two-sided tests and results were considered statistically significant at p < 0.05. The data report followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for cross-sectional studies (Field et al., 2014). This study was approved by the University of Illinois Chicago Institutional Review Board under non-human subjects determination. Data analysis was performed in the Joinpoint Regression Program (National Cancer Institute) Joinpoint Trend Analysis Software (Joinpoint Regression Program, 2024), version 5.2.0, and SAS Studio, version 3.81, using SAS survey procedures in the year 2024.

3. Results

The analytical sample included 40,818 cisgender women aged 21–65 with no history of hysterectomy. 4.33 % of the unweighted sample and 4.48 % of the weighted sample (N = 308,897,709) identified as LGB (Table 1; Table A.2). Overall, 79.16 % of the unweighted sample was up to date for cervical cancer screening at the time of survey and 90.41 % of respondents had received cervical cancer screening at least once in their lifetimes. The prevalence of up-to-date screening among heterosexual cisgender women (78.91 %) was greater than the prevalence of up-to-date screening among LGB cisgender women (73.29 %; Table 1). Of the weighted sample, 70.64 % of the LGB respondents reported up-to-date cervical cancer screening in contrast to 78.92 % of the heterosexual respondents. The prevalence of ever-screening was also greater for heterosexual cisgender women (90.60 %) than LGB cisgender women (86.25 %) across years (Table A.2; additional year-by-year data in Tables A.3–A.6).

Table 1.

Descriptive Characteristics of Cisgender Women Respondents by Sexual Orientation, United States, 2013–2021 (n = 40,818).

		Sexual orientation
		Lesbian, gay, and/or bisexual		Heterosexual
Characteristic	Unweighted frequency	Unweighted frequency, %		Unweighted frequency, %
Total	40,818	1767	4.33 %	39,051	95.67 %
Cervical cancer screening up-to-date
Yes	32,111	1295	73.29 %	30,816	78.91 %
No	8455	471	26.66 %	7984	20.45 %
Missing	252	1	0.06 %	251	0.64 %
Age, years
21–29	8290	657	37.18 %	7633	19.55 %
30–39	10,760	509	28.81 %	10,251	26.25 %
40–49	8632	285	16.13 %	8347	21.37 %
50–65	12,329	295	16.69 %	12,034	30.82 %
Missing	807	21	1.19 %	786	2.01 %
Race/ethnicity
Hispanic	7623	252	14.26 %	7371	18.88 %
Non-Hispanic White only	23,742	1126	63.72 %	22,616	57.91 %
Non-Hispanic Black/African American only	5776	253	14.32 %	5523	14.14 %
Non-Hispanic Asian only	2828	57	3.23 %	2771	7.10 %
Non-Hispanic Other single/multiple racesa	849	79	4.47 %	770	1.97 %
Insurance status
Private	8107	288	16.30 %	7819	20.02 %
Medicaid and other
publicb	2714	98	5.55 %	2616	6.70 %
Other insurancec	535	20	1.13 %	515	1.32 %
Uninsured	2140	75	4.24 %	2065	5.29 %
Missing	27,322	1286	72.78 %	26,036	66.67 %

All significant Rao-Scott Chi-Square tests at p < 0.01.

^a.Includes American Indian or Alaska Native.

^b.Includes Medicare and state-sponsored insurance.

^c.Includes Medi-Gap (supplemental insurance coverage that is purchased by Medicare beneficiaries for out-of-pocket expenses not covered by Original Medicare), Military, Indian Health Service, and single service insurance.

Logistic regression analysis showed disparities in cervical cancer screening between LGB and heterosexual cisgender women, even when adjusting by year, age, race/ethnicity, and insurance status, with LGB cisgender women reporting 22 % lower odds of up-to-date cervical cancer screening compared to their heterosexual counterparts (OR 0.78, 95 % CI 0.64 to 0.95; Table 2). Among both heterosexual and LGB cisgender women, racial and ethnic minorities (Hispanic, non-Hispanic Black/African American only, non-Hispanic Asian only, non-Hispanic other single/multiple races) also reported lower odds of up-to-date cervical cancer screening compared to non-Hispanic White only heterosexual individuals (Table 2).

Table 2.

Logistic Regression Model of Up-to-Date Cervical Cancer Screening among Cisgender Women Respondents, United States, 2019 & 2021 (n = 17,944).

Independent variable	Odds Ratioa	95 % confidence interval
Year
Year-to-year change (continuous)	0.95	(0.91, 0.99)
Sexual orientation
Lesbian, Gay, and/or Bisexual	0.78	(0.64, 0.95)
Heterosexual	1.0
Race/ethnicity
Hispanic	0.75	(0.66, 0.84)
Non-Hispanic White only	1.0
Non-Hispanic Black/African American only	0.91	(0.78, 1.06)
Non-Hispanic Asian only	0.47	(0.40, 0.55)
Non-Hispanic Other single and multiple racesb	0.74	(0.51, 1.09)
Age, years
21–29	0.50	(0.44, 0.58)
30–39	1.0
40–49	0.77	(0.67, 0.88)
50–65	0.55	(0.49, 0.61)
Insurance status
Private	1.0
Medicaid and other publicc	0.63	(0.56, 0.72)
Other insuranced	0.73	(0.58, 0.91)
Uninsured	0.33	(0.28, 0.37)

^a.Odds ratio estimates were adjusted by all other variables in the table.

^b.Includes American Indian or Alaska Native.

^c.Includes Medicare and state-sponsored insurance.

^d.Includes Medi-Gap (supplemental insurance coverage that is purchased by Medicare beneficiaries for out-of-pocket expenses not covered by Original Medicare), Military, Indian Health Service, and single service insurance.

In both logistic regression and trend analysis, there was an overall decrease in up-to-date screening across years. In the adjusted logistic regression model, cisgender women in 2021 showed a 5 % decrease in odds of up-to-date screening compared to 2019 (OR 0.95, 95 % CI 0.91 to 0.99). In trend analysis among heterosexual cisgender women, there was a 0.97 % annual percentage decrease in up-to-date screening, declining from 92.39 % in 2013 to 85.17 % in 2021 (p < 0.01; Fig. 1). There was no significant trend in up-to-date screening for LGB cisgender women in trend analysis (Fig. 1).

Fig. 1.

Trends in Proportion of Up-to-Date Cervical Cancer Screening among Cisgender Women by Sexual Orientation, United States, 2013–2021.

Abbreviations: APC, Annual Percent Change; LGB, Lesbian, Gay, and/or Bisexual.

*p < 0.01.

Hispanic and non-Hispanic White only heterosexual cisgender women in particular saw a decrease in up-to-date screening across years in trend analysis. There was a 1.52 % annual percentage decrease in up-to-date screening for Hispanic heterosexual individuals (p < 0.01; Fig. 2), falling from 77.15 % in 2013 to 69.63 % in 2021. And there was a 0.63 % annual percentage decrease among non-Hispanic White only heterosexual individuals (p = 0.03; Fig. 2), falling from 82.78 % in 2013 to 79.53 % in 2021. Among LGB cisgender women, non-Hispanic Black/African American only individuals saw a 2.67 % annual percentage decrease in up-to-date screening from 2013 to 2021 (p < 0.01; Fig. 2), falling from 85.22 % in 2013 to 67.91 % in 2021.

Fig. 2.

Trends in Proportion of Up-to-Date Cervical Cancer Screening among Cisgender Women by Sexual Orientation and Race/Ethnicity, United States, 2013–2021.

Abbreviations: APC, Annual Percent Change; NH, Non-Hispanic.

*p < 0.01.

In preliminary interaction analysis, heterosexual Hispanic cisgender women (OR 0.71, 95 % CI 0.64 to 0.82), heterosexual non-Hispanic Asian only cisgender women (OR 0.47, 95 % CI 0.40 to 0.55), LGB non-Hispanic White only cisgender women (OR 0.69, 95 % CI 0.55 to 0.86), and LGB non-Hispanic Black/African American only cisgender women (OR 0.37, 95 % CI 0.20 to 0.68) had decreased odds of cervical cancer screening compared to heterosexual non-Hispanic White only cisgender women (Table A.1). Only LGB Hispanic cisgender women had significant multiplicative interaction (OR 0.83, 95 % CI 0.52 to 1.31; p = 0.05) and no subgroup had significant additive interaction (Table A.1).

4. Discussion

The overall declining trends in up-to-date cervical cancer screening prevalence in the United States from 2013 to 2021 reported in this study are of public health concern and track with previously-reported data from 2000 to 2015 (Watson et al., 2017). These trends are statistically significant among heterosexual cisgender women and their non-Hispanic White only and Hispanic subgroups. Despite the unchanging prevalence of up-to-date cervical cancer screening between 2013 and 2021 among LGB cisgender women, this prevalence is significantly lower than that of heterosexual cisgender women. This finding is corroborated by previous studies and represents a significant disparity in cervical cancer screening between LGB and heterosexual populations (Baumann et al., 2024; Tracy et al., 2010; Matthews et al., 2004; Lin et al., 2024).

When adjusting for factors like sexual orientation, age, insurance status, and survey year, all racial/ethnic minorities (Hispanic, non-Hispanic Black/African American only, non-Hispanic Asian only, non-Hispanic other single/multiple races) also showed lower prevalence of up-to-date cervical cancer screening in the present study, which is concordant with a 2023 study of cervical cancer screening by race/ethnicity (Spencer et al., 2023), a 2024 study of Behavioral Risk Factor Surveillance System (BRFSS) data (Lin et al., 2024), and the authors’ 2023 analysis of Healthy Chicago Survey data (Baumann et al., 2024), but deviates from the aforementioned 2000–2015 trend report that showed greater screening among Hispanic and non-Hispanic Black compared to non-Hispanic White cisgender women (Watson et al., 2017). This could be the result of differences in how up-to-date screening was defined in the present versus the previous study, differences in model adjustment factors, or changes in screening patterns since 2015.

There was notable, highly significant intersection between sexual orientation and race for LGB non-Hispanic Black/African American only cisgender women in the present study. This subgroup saw the greatest annual percentage decrease in up-to-date cervical cancer screening, even with a relatively small sample size (n = 253). The magnitude of the odds ratio comparing up-to-date cervical cancer screening for this group to heterosexual non-Hispanic White only cisgender women was also striking, showing 63 % decreased odds in up-to-date cervical cancer screening, the largest of any subgroup. A similar finding was also shown in Healthy Chicago Survey data, with Black/African American LGB individuals reporting a 15 % lower prevalence of up-to-date cervical cancer screening compared to their heterosexual counterparts, the greatest prevalence difference among all race/ethnic groups. The 2024 study of BRFSS data showed significantly lower prevalence of up-to-date screening among Black individuals, due mostly to low prevalence of ever-screening in this group (Lin et al., 2024). This intersection warrants greater exploration and expansion on previous qualitative work that demonstrated that patients experience compound effects of discrimination on the basis of race and sexual orientation that inhibits receiving cervical cancer screening (Agénor et al., 2015; Joudeh et al., 2021; Turpin et al., 2021).

Interestingly, LGB Hispanic women showed significant interaction between ethnicity and sexual orientation with LGB cisgender women reporting greater odds of up-to-date cervical cancer screening compared to their heterosexual counterparts. While this trend was not significant in Joinpoint analysis, LGB Hispanic cisgender women also showed increased screening from 2013 to 2021. This echoes previous findings in the Healthy Chicago Survey data, where Hispanic/Latina cisgender women reported only a 5 % lower prevalence of up-to-date cervical cancer screening compared to 11 % lower among non-Hispanic or Latina cisgender women (Baumann et al., 2024). A 2014 study saw no difference in the odds of screening between lesbian and heterosexual Latina women (Agénor et al., 2014). These findings may require more investigation, but show a promising lack of LGB disparity among Hispanic/Latina women.

4.1. Limitations

There are several limitations to this study, primarily due to the data originating from a cross-sectional survey. Incidence, and therefore likelihood, of up-to-date cervical cancer screening in the analytical sample cannot be determined. The results do not represent individual cervical cancer screening experience. Although weighting was used in the statistical methods, selection and non-response bias may have impacted calculated prevalence estimates. Because all NHIS variables studied required self-report, the effects of social desirability, random bias, and recall bias should be considered and may have particularly impacted the respondents’ report of their most recent cervical cancer screening test (Rauscher et al., 2008). Calculations of measures of additive interaction (like RERI) may be unreliable when using odds ratios that over-estimate prevalence ratios for common outcomes like cervical cancer screening (VanderWeele and Knol, 2014).

There are also limitations in the data. The COVID-19 pandemic and related social and hospital policies disrupted routine cancer screening for most residents of the United States. Cervical cancer screening was no exception, and a temporary decline in cervical cancer screening service utilization has been shown in the year 2020 (Oakes et al., 2023). Because cervical cancer screening is provided in multi-year intervals, this may have impacted the overall prevalence of up-to-date screening in 2021, however this year followed the trends of 2013–2019. For unknown reasons, 2019 was a relatively “good” year for cervical cancer screening among LGB cisgender women that may represent an outlier, with 72.49 % of respondents reporting up-to-date screening, compared to 72.06 % in 2015 and 68.86 % in 2021. This may have affected trends in screening prevalence, leading to an insignificant annual percent change in trend analysis. Small sample sizes, particularly in combined sexual orientation and race/ethnicity subgroups, also may not have allowed for statistical significance in what appeared to be downward trends in up-to-date cervical cancer screening among heterosexual racial/ethnic minorities. More robust data across additional years would potentially elucidate trends not found in the present study.

However, considering these intersections, it is imperative for providers to understand their patients’ unique identities and experiences, particularly for procedures like cervical cancer screening, to combat these barriers. The health system as a whole should also work to increase the representativeness of the healthcare workforce and foster the careers of those underrepresented in medicine, particularly to increase rates of follow-up after abnormal screening results (Agénor et al., 2015; Jetty et al., 2022). Newly FDA-approved HPV self-sampling could also help address practical and emotional barriers to screening for this population (Garg et al., 2024; Scarinci et al., 2021).

5. Conclusion

In 2021 there were approximately 19.72 million cisgender women aged 21–65 who were potentially not up to date with cervical cancer screening. Of these, 1.76 million LGB women were not up to date for cervical cancer screening, and a greater proportion of these women identified as non-Hispanic Black/African American Only. There is an alarming overall downward trend in up-to-date cervical cancer screening, which is especially concerning for marginalized populations. Future efforts should improve screening in cisgender populations regardless of sexual orientation, and specific attention should be aimed at LGB and Black/African American identifying cisgender women.

Glossary

Cervical Cancer Screening (CCS)

Medical procedures used to detect early signs of cervical cancer, including Pap tests and HPV testing

Human Papillomavirus (HPV)

A virus that can cause lesions and cervical cancer, among other health issues, and is often screened for during CCS

Up-to-Date (UTD) Screening

A term referring to individuals who have had a cervical cancer screening within the recommended time intervals set by health guidelines. The US Preventive Services Task Force defines this as Pap testing every 3 years for individuals with a cervix age 21–29 and Pap testing with HPV testing every 5 years for individuals age 30–65

Lesbian, Gay, Bisexual (LGB)

Terms referring to individuals’ sexual orientations. Lesbian and gay refer to women who are physically, romantically, and/or emotionally attracted to other women. Bisexual refers to individuals who are physically, romantically, and/or emotionally attracted to both men and women

Cisgender

A term for individuals whose gender identity corresponds with the sex they were assigned at birth, in contrast to transgender individuals

Joinpoint Regression Program

A statistical software used to analyze trends in prevalence of disease or screening over time

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ypmed.2025.108223.

Data availability

The data is publicly available on the NHIS website.