Journal Pre-proof Geographic and sociodemographic differences in cervical cancer screening modalities
Ann Goding Sauer, Priti Bandi, Debbie Saslow, Farhad Islami, Ahmedin Jemal, Stacey A. Fedewa PII:
S0091-7435(20)30038-4
DOI:
https://doi.org/10.1016/j.ypmed.2020.106014
Reference:
YPMED 106014
To appear in:
Preventive Medicine
Received date:
30 July 2019
Revised date:
29 January 2020
Accepted date:
1 February 2020
Please cite this article as: A.G. Sauer, P. Bandi, D. Saslow, et al., Geographic and sociodemographic differences in cervical cancer screening modalities, Preventive Medicine(2018), https://doi.org/10.1016/j.ypmed.2020.106014
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2018 Published by Elsevier.
Journal Pre-proof Geographic and Sociodemographic Differences in Cervical Cancer Screening Modalities
Ann Goding Sauer1, Priti Bandi1, Debbie Saslow2, Farhad Islami1, Ahmedin Jemal1, Stacey A. Fedewa1
1
Intramural Research Department, American Cancer Society Cancer Control Department, American Cancer Society
re lP
ur Jo
Word Counts: Abstract=247 Main Text=3,392
na
Ann Goding Sauer, MSPH 250 Williams Street NW Atlanta, GA 30303 Phone: 1 (404) 329-7989 Fax: 1 (404) 321-4669 Email:
[email protected]
-p
Contact Information for Corresponding Author:
ro
of
2
Conflict of Interest: All authors are employed by the American Cancer Society, which receives grants from private and corporate foundations, including foundations associated with companies in the health sector for research outside of the submitted work. The authors are not funded by or key personnel for any of these grants.
1
Journal Pre-proof
2
Abstract Cervical cancer screening recommendations for women aged 30-65 years include co-testing (high-risk human papillomavirus [hrHPV] with Pap testing) every five years or Pap testing alone every three years. Geographic variations of these different screening modalities across the United States have not been examined.
of
We selected 82,426 non-pregnant women aged 30-65 years from the 2016 Behavioral Risk Factor Surveillance System with data on sociodemographics, hysterectomy, and cervical cancer screening, representing 42 states and the District of Columbia. Logistic regression models with predicted marginal probabilities were used to calculate state-level prevalence estimates of recent cervical cancer screening and uptake of co-testing, Pap testing, and hrHPV testing among those who were recently screened. Analysis was conducted in 2018-2019.
lP
re
-p
ro
Recent screening prevalence ranged from 80.0% (Idaho) to 92.2% (Massachusetts), with more state-level geographic variability in co-testing than Pap testing alone. Uptake of co-testing ranged from 27.5% (Utah) to 49.9% (District of Columbia); compared to the national estimate, co-testing was lower in 12 states and higher in six states. Overall, Midwestern and Southern states had the lowest uptake of co-testing whereas Northeastern states had the highest. Sociodemographic, healthcare, and behavioral factors accounted for some but not all state-level variation in co-testing.
Jo
ur
na
There was substantial state-level variability in co-testing prevalence, which was lowest in Midwestern and Southern states; the variation was not entirely explained by individual sociodemographic, healthcare, and behavioral factors. Future studies should monitor the impact of geographic variations in screening modalities on state-level differences in cervical cancer incidence, survival, and mortality.
Keywords: Uterine Cervical Neoplasms; Behavioral Risk Factor Surveillance System; Early Detection of Cancer; Vaginal Smears
Journal Pre-proof
3
Introduction In the United States (US), an estimated 13,000 women will be diagnosed with cervical cancer in 2019.1 While overall cervical cancer incidence has declined since the 1970’s, reductions in squamous cell cervical cancers have stalled in some groups and incidence of adenocarcinomas have increased in younger women in recent years.2 Historical declines in cervical cancer incidence are attributed to the Papanicolaou (Pap) smear test, which has long been a recommended cervical cancer screening modality. For the first time in 2002, the American Cancer Society recommended high-risk HPV (hrHPV) testing in conjunction with Pap testing
of
(co-testing) no more frequently than every three years for women aged 30 years and older as a
ro
method of cervical cancer screening.3 In 2012, for women aged 30-65 years, the US Preventive Services Task Force (USPSTF) and the American Cancer Society recommended co-testing every
-p
five years or Pap testing alone every 3 years.4, 5 Co-testing is the preferred screening method for this age group due to its improved sensitivity in detecting glandular cervical cancers as well as
lP
re
the high negative predictive value.
Organized cervical cancer screening programs in some countries in the European Union may help more seamlessly implement co-testing and primary hrHPV testing as means of cervical
na
cancer screening and reduce barriers to screening.6, 7 In contrast, cancer screening in the US is more heterogeneous with a variety of factors, many based on individual women and their
ur
healthcare providers, influencing whether screening is received and which mode is utilized.
Jo
Results from previous studies indicate that the prevalence of cervical cancer screening in the US via Pap testing has decreased in recent years but the prevalence of co-testing may have increased.8-10 Previous investigations also indicate that receipt of Pap testing differs by sociodemographic, healthcare, and behavioral factors such as race/ethnicity, educational attainment, smoking status, body mass, and having a personal healthcare provider.9-13 Yet, few studies have examined variations in receipt of co-testing according to sociodemographic factors and none have examined potential state-level differences. The aim of this study was to examine contemporary state-level differences in cervical cancer screening modalities among women aged 30-65 years. The secondary aim was to examine whether cervical cancer screening varied by sociodemographic, healthcare, and behavioral factors and if these factors accounted for potential state-level differences.
Journal Pre-proof
4
Methods Study Population Non-pregnant women aged 30-65 years with intact uteri from 42 states and the District of Columbia were selected from the 2016 Behavioral Risk Factor Surveillance System (BRFSS). The BRFSS is an annual state-based telephone survey of non-institutionalized adults supported by the Centers for Disease Control and Prevention that collects cancer screening data in evennumbered years.14 The overall median response rate was 47.0%; the median response rates for
of
the Midwest, Northeast, South, and West regions were 54.0%, 42.2%, 43.0%, and 51.8%,
ro
respectively. Women from Arkansas, Arizona, Connecticut, Maryland, Rhode Island, Vermont, New Hampshire, and Washington were not included as hysterectomy data were not available in
-p
2016. Women with missing data for race/ethnicity (n=1,270), educational attainment (n=144), and cervical cancer screening (n=4,545) were excluded, resulting in an analytic population of
lP
re
82,426.
Measures
In accordance with USPSTF and American Cancer Society cervical cancer screening
na
recommendations,4, 15 recent cervical cancer screening was defined as 1) co-testing within the past five years, 2) Pap testing alone within the past three years, or 3) hrHPV testing alone within
ur
the past five years. Among women who were recently screened, we examined the proportions of
Jo
each of these three mutually exclusive categories. Although hrHPV testing alone every five years among women aged 30-65 years was not recommended by the USPSTF until 2018, primary hrHPV testing was used for cervical cancer screening in some health systems beginning in 2015 given approval by the US Food and Drug Administration and interim guidance from other organizations for primary hrHPV testing no more frequently than every three years.16 Herein we do not present results for hrHPV testing alone due to low overall prevalence and unstable estimates for most states.
Self-reported data on cervical cancer screening were based on a series of questions regarding the respondent’s Pap testing history followed by questions regarding HPV testing and hysterectomy: “A Pap test is a test for cancer of the cervix. Have you ever had a Pap test?,” “How long has it
Journal Pre-proof
5
been since you had your last Pap test?,” “Now, I would like to ask you about the Human Papillomavirus or HPV test. An HPV test is sometimes given with the Pap test for cervical cancer screening. Have you ever had an HPV test?,” “How long has it been since you had your last HPV test?,” and “Have you had a hysterectomy? (A hysterectomy is an operation to remove the uterus (womb).”14
Data on covariates included age [30-39, 40-49, 50-65 years]; race/ethnicity [non-Hispanic white [NHW], non-Hispanic black [NHB], Hispanic, and non-Hispanic other/multiracial]; educational
of
attainment [
ro
status (uninsured/insured), having a personal healthcare provider (yes, no), cigarette smoking status (current, former, never), and body mass index (BMI) category (underweight, healthy
-p
weight, overweight, or obese).
re
Statistical Analysis
lP
Descriptive statistics for sociodemographic variables were calculated by state and summarized by region. Logistic regression models with predicted marginal probabilities were used to
na
calculate the prevalence of each outcome by state. To further examine state-level differences in outcomes, unadjusted (PR) and adjusted (aPR) prevalence ratios and associated 95% confidence
Jo
regression models.
ur
intervals (CI) were calculated using an aggregated national estimate as the referent in logistic
To determine if state-level differences in recent cervical cancer screening and co-testing were due to state-level differences in respondent sociodemographic, healthcare, or behavioral factors as described above, we ran a series of adjusted models to examine possible associations. The first adjusted model (Model 1a) included age, race/ethnicity, and educational attainment; the second model (Model 1b) was additionally adjusted for insurance status and identification of a personal healthcare provider. The third adjusted model (Model 1c) further included smoking status and BMI category. As a sensitivity analysis, we ran a model (Model 1d) that included age, race/ethnicity, educational attainment, and insurance status.
Journal Pre-proof
6
We also used multilevel logistic regression models to assess the magnitude of geographic (statelevel) variability in recent cervical cancer screening and co-testing and whether individual factors accounted for potential state-level differences in screening. We began with an unconditional (intercept-only) model (UCM model) that estimated between state variability without individual factors before adding the individual factors of interest: Model 2a (UCM+ age, race/ethnicity, educational attainment), Model 2b (age, race/ethnicity, educational attainment, insurance status, personal healthcare provider), and Model 2c (age, race/ethnicity, educational attainment, insurance status, identification of healthcare provider, smoking status, and BMI
of
category). Because we cannot compare variance components directly in the case of binary outcome variables, we used the Intraclass Correlation Coefficient (ICC)17 – the proportion of the
ro
total variation in the outcome that is attributable to between-state variance – to quantify the
-p
magnitude of the variation in outcomes between states. The ICC was estimated via the latent
re
response formulation.
lP
Data analyses were performed in 2018 and 2019 using SAS-callable SUDAAN release 11.0.1 and STATA 15.1 and accounted for the complex survey design of BRFSS. Institutional review
Descriptive statistics
ur
Results
na
board approval was not needed as deidentified, publicly-available data were used.
Jo
The distribution of age, race/ethnicity, and educational attainment varied by geographic region (Supplemental Figure 1a-c). For example, the Northeast had the highest proportion of women age 50-65 years (45%) compared to 38% in the South and West. More than 80% of women in the Midwest were NHW versus 64% in the South. About 39% of women in the Northeast were college graduates compared to 31% in the South. Healthcare and behavioral factors also varied (Supplemental Figure 2a-c). Approximately 13% of women in the South reported no health insurance versus 8% in the Midwest and Northeast. About 20% in the West reported no personal healthcare provider compared to 8% in the Northeast. Approximately 19% of women in the South were current smokers versus 14% in the Northeast and West. About 36% of women in the South were obese compared to 27% in the Northeast.
Journal Pre-proof
7
Recent Cervical Cancer Screening Among women aged 30-65 years, the prevalence of recent cervical cancer screening was 87.2% and was lower among women aged 40-65 years than those aged 30-39 years (40-49 years aPR=0.96, 95%CI: 0.95-0.98; 50-65 years aPR=0.91, 95%CI: 0.90-0.92) (Table 1). Recent screening prevalence was also lower among women with lower educational attainment, the
of
uninsured, and those without a personal healthcare provider. Additionally, current smokers had a lower prevalence of recent cervical cancer screening than never smokers (aPR=0.94, 95%CI:
Jo
ur
na
lP
re
-p
Hispanics (aPR=1.06, 95%CI: 1.04-1.07) than NHWs.
ro
0.93-0.96). The prevalence was higher among NHBs (aPR=1.03, 95%CI: 1.01-1.04) and
Journal Pre-proof
8
Table 1. Prevalence and adjusted prevalence ratios for recent cervical cancer screening overall and co-testing in the past five years among women age 30-65 years, BRFSS 2016 Recent Screening
Co-testing
Prev
Prev LL 95% CI
Prev UL 95% CI
aPR
aPR LL 95% CI
aPR UL 95% CI
Prev
Prev LL 95% CI
Prev UL 95% CI
aPR
aPR LL 95% CI
aPR UL 95% CI
87.2
86.8
87.7
---
---
---
37.8
37.1
38.5
---
---
---
30-39 years
91.8
91.1
92.5
referent
referent
referent
49.0
47.6
50.4
referent
referent
referent
40-49 years
88.5
87.5
89.3
0.96
0.95
0.98
39.6
38.2
41.1
0.81
0.77
0.85
50-65 years
83.5
82.7
84.3
0.91
0.90
0.92
27.8
26.8
28.9
0.57
0.54
0.60
NH white
86.6
86.0
87.2
referent
referent
referent
36.0
35.1
36.9
referent
referent
referent
NH black
88.9
87.4
90.2
1.03
1.01
1.04
46.0
43.8
48.2
1.28
1.21
1.35
Hispanic
91.6
90.4
92.7
1.06
1.04
1.07
43.0
40.7
45.4
1.20
1.12
1.27
NH-Other/Multi
83.1
80.6
85.4
0.96
0.93
34.9
31.5
38.5
0.97
0.88
1.07
Educational Attainment
84.3
82.3
86.1
0.93
29.3
26.7
32.1
0.72
0.65
0.80
84.6
83.4
85.6
0.93
Some College
88.6
87.8
89.4
0.98
College Graduate
90.8
90.1
91.4
referent
l a
0.95
HS
Uninsured
79.5
77.6
81.3
Insured
89.1
88.6
No HCP
77.4
75.8
J
HCP
89.9
Current Former Never
Overalla Age
o r p
Race/Ethnicity
0.91
0.95
35.9
34.3
37.6
0.88
0.84
0.93
0.97
0.99
40.6
39.1
42.1
1.00
0.95
1.05
referent
referent
40.7
39.5
41.8
referent
referent
referent
0.89
0.87
0.91
35.5
32.9
38.3
0.92
0.85
1.00
referent
referent
referent
38.6
37.8
39.5
referent
referent
referent
78.9
0.86
0.84
0.88
35.2
33.2
37.2
0.90
0.85
0.96
89.4
90.3
referent
referent
referent
38.9
38.0
39.7
referent
referent
referent
83.5
82.4
84.6
0.94
0.93
0.96
47.4
45.5
49.2
1.32
1.26
1.39
89.1
88.2
90.0
1.01
0.99
1.02
39.7
38.1
41.3
1.11
1.06
1.16
88.6
88.0
89.2
referent
referent
referent
35.8
34.9
36.8
referent
referent
referent
83.2
79.8
86.2
0.95
0.91
0.99
43.6
37.8
49.5
1.12
0.97
1.28
Personal HCP
89.5
rn
u o
Insurance Status
0.92
e
r P 0.99
f o
Smoking Status
Body Mass Index Underweight
Journal Pre-proof
9
Healthy Weight
87.9
87.0
88.6
referent
referent
referent
39.1
37.8
40.3
referent
referent
referent
Overweight
89.0
88.1
89.7
1.01
1.00
1.03
38.3
36.9
39.6
0.98
0.93
1.03
Obese
86.8
85.9
87.6
0.99
0.97
1.00
37.3
36.0
38.6
0.95
0.91
1.00
Prev: prevalence LL: lower limit UL: upper limit
f o
CI: confidence interval aPR: adjusted prevalence ratio
o r p
NH: non-Hispanic HS: high school HCP: healthcare provider a
e
Estimates are unadjusted. Denominator for recent=82,426; co-testing=70,990.
r P
Note: Recent screening includes co-testing within the past five years, Pap testing alone within the past three years, or hrHPV testing alone within the past five years. Multivariable model for adjusted prevalence estimates included age, race/ethnicity, education, insurance status, identified personal healthcare provider, smoking status, and BMI category. Model for co-testing are among those who reported recent screening. Respondents included in model for recent=74,954; there were 64,781 women included in model for co-testing.
l a
Source: Behavioral Risk Factor Surveillance System, 2016.
Jo
n r u
Journal Pre-proof
10
By state, the unadjusted prevalence of recent cervical cancer screening ranged from 80.0% in Idaho to 92.2% in Massachusetts (Supplemental Table 1). The unadjusted prevalence of recent screening in 31 states was similar to the national estimate; prevalence in five states (Idaho, Indiana, North Dakota, Texas, and Wyoming) was lower (Figure 1). In contrast, the prevalence in seven states (California, Colorado, Massachusetts, Michigan, Minnesota, North Carolina, and Wisconsin) was higher than the national estimate. Figure 1. Unadjusted prevalence ratio of recent cervical cancer screening by state among women ages
Jo
ur
na
lP
re
-p
ro
of
30-65 years, BRFSS 2016
Note: Recent screening includes co-testing within the past five years, Pap testing alone within the past three years, or hrHPV testing alone within the past five years. Prevalence ratios calculated relative to an aggregate national estimate. Categories determined by whether the confidence interval for the prevalence ratio spanned 1. Source: Behavioral Risk Factor Surveillance System, 2016.
Journal Pre-proof
11
The state-level prevalence of recent cervical cancer screening accounting for sociodemographic factors (Model 1a) was generally comparable to the unadjusted estimates ranging from 80.1% to 91.8% and was similar to the national estimate in 30 of the 43 states assessed (Supplemental Table 1). When healthcare factors were also incorporated (Model 1b), estimates ranged from 81.4% to 90.6% and were similar to the national estimate in 34 states; prevalence was also similar to the national estimate in 34 states when behavioral factors were added (Model 1c; range: 81.6% to 90.6%). Based on our sensitivity analysis (Model 1d, adjusting for sociodemographic factors and health insurance status) prevalence ranged from 81.8% to 91.2%
ro
of
and was similar to the national estimate in 33 states.
Furthermore, the unconditional intercept-only multilevel model indicated that the magnitude of
-p
state-level variability in recent cervical cancer screening (variance: 0.036, SE: 0.008) was small; the ICC of 0.0132 (SE: 0.0034) indicated that only about 1.3% of the total variation could be
re
attributed to between-state differences. The ICC was comparable with the sequential addition of
lP
individual factors (Model 2a: 0.0130, SE: 0.0031; Model 2b: 0.0087, SE: 0.0023; Model 2c:
na
0.0091, SE: 0.0026).
Co-Testing among Women Who Were Recently Screened for Cervical Cancer Among women who were recently screened for cervical cancer, 37.8% received co-testing
ur
(Table 1). Compared to women aged 30-39 years, the prevalence of co-testing was lower among
Jo
older women (40-49 years aPR=0.81, 95%CI: 0.77-0.85; 50-65 years aPR=0.57, 95%CI: 0.540.60). Co-testing was more common among NHBs (aPR=1.28, 95%CI: 1.21-1.35) and Hispanics (aPR=1.20, 95%CI: 1.12-1.27) than NHWs. The prevalence of co-testing was markedly lower among women with lower educational attainment, those without a personal healthcare provider and never smokers. Co-testing prevalence did not vary by health insurance status or BMI.
Among women who were recently screened, the unadjusted prevalence of co-testing ranged widely from 27.5% in Utah to 49.9% in the District of Columbia (Supplemental Table 2) and, in general, Midwestern and Southern states had lower prevalence of co-testing than Northeastern and Western states. Compared to the national estimate, the unadjusted prevalence of co-testing was lower in 12 states, and similar to the national estimate in 25 states (Figure 2). The
Journal Pre-proof
12
prevalence of co-testing was higher than the national estimate in 6 states (Colorado, the District of Columbia, Florida, Oregon, Maine, and New York), three of which are located in the Northeast.
Jo
ur
na
lP
re
-p
ro
of
Figure 2. Unadjusted prevalence ratio of co-testing in the past 5 years by state among women ages 3065 years, BRFSS 2016
Note: Estimates for co-testing are among those who were recently screened. Prevalence ratios calculated relative to an aggregate national estimate. Categories determined by whether the confidence interval for the prevalence ratio spanned 1. Source: Behavioral Risk Factor Surveillance System, 2016.
Journal Pre-proof
13
When sociodemographic factors were included in the model (Model 1a), prevalence ranged from 26.3% to 45.3% and was similar to the national estimate in 25 states (Supplemental Table 2). When also incorporating healthcare factors (Model 1b), prevalence ranged from 26.5% to 45.2% and was similar to the national estimate in 26 states. Results from Model 1c (including behavioral factors) and our sensitivity analysis (Model 1d, sociodemographic factors and insurance status) were comparable to those of the other models.
Again, the unconditional intercept-only multilevel model indicated that the magnitude of state-
of
level variability in co-testing (variance: 0.044, SE: 0.011) was small; the ICC of 0.011 (SE:
ro
0.0025) indicated that only about 1.1% of the total variation could be attributed to between-state differences and was similar with the addition of individual factors (Model 2a: 0.0134, SE:
-p
0.0026; Model 2b: 0.0133, SE: 0.0027; Model 2c: 0.0130, SE: 0.0024).
re
Pap Testing Alone among Women Who Were Recently Screened for Cervical Cancer
lP
Given the mutual exclusivity of the defined cervical cancer screening modalities and the very low prevalence of hrHPV testing alone, the results for Pap testing alone are, in general, the
na
complement of the co-testing results presented above. Among women who were recently screened, the overall prevalence of Pap testing alone was 61.6% and was lower among younger women, NHBs and Hispanics, and college graduates (data not shown). Pap testing prevalence
ur
was also lower among women who reported having a personal healthcare provider and among
Jo
current and former smokers. Pap testing prevalence did not vary by health insurance status or BMI. The unadjusted prevalence of Pap testing alone ranged from 48.7% in the District of Columbia to 71.8% in Utah (data not shown).
Discussion In this contemporary state-based study of cervical cancer screening modalities among US women aged 30-65 years, co-testing prevalence varied widely by state ranging from 28% in Utah to 50% in the District of Columbia and was generally lower in Midwestern and Southern states than
Journal Pre-proof
14
Northeastern states. Co-testing prevalence also differed by sociodemographic, healthcare, and behavioral factors with lower uptake among older women, NHWs, those with lower educational attainment, those without a personal healthcare provider, and never smokers. Adjusting for individual sociodemographic factors accounted for some but not all state-level differences in prevalence and incorporating healthcare and behavioral factors did not account for any additional state-level variation. Furthermore, little variation in co-testing prevalence can be attributed to between-state differences and thus is likely reflective of individual factors not assessed here.
of
Our findings were consistent with previous studies that found that co-testing was more common
ro
among those with higher levels of education and younger women.8, 10, 12, 18 Furthermore, the uptake of co-testing was higher in Northeastern states where there is a higher proportion of
-p
college-educated women. By age, younger women may be more likely than older women to accept the relatively novel screening tests (co-testing or hrHPV testing alone) and its longer
re
screening interval.19 Our results are consistent with another study using BRFSS data showing
lP
that NHWs had a lower prevalence of cervical cancer screening compared to NHBs20 while estimates based on data from the National Health Interview Survey (NHIS) have shown that NHBs and NHWs had similar uptake of cervical cancer screening.10, 12, 18, 20 Some of the
ur
the different survey modes.20
na
variability by race/ethnicity may be due to differences in overreporting21, 22 and/or the impact of
Jo
We initially hypothesized that potential variations in co-testing prevalence across states could be attributed to state-level differences in individual sociodemographic, healthcare, and behavioral factors. However, adjusting for these factors accounted for only a portion of the variation in sequentially adjusted models suggesting that there are differences in the adoption of co-testing. Such differences may be accounted for by individual factors unmeasured in the current study as well as provider, healthcare system, and policy-related factors. For example, we did not assess rurality, and Blake et al. reported that those in rural areas were less likely to be aware of the HPV-cervical cancer association than their urban counterparts.23 Furthermore, a woman’s general awareness of HPV and the HPV vaccine has been shown to influence her acceptance of HPV testing as a means of cervical cancer screening.19 Awareness of the patient’s HPV vaccination status has also been shown to influence providers’ recommendation for hrHPV
Journal Pre-proof
15
testing.24 These associations may, in part, explain our results of a higher proportion of co-testing in Northeastern states where HPV vaccination uptake is also relatively high.25 In addition, provider specialty has been shown to influence their recommendation for cervical cancer screening.26, 27 Access to specialist providers and provider recommendation for co-testing may vary across states. The diffusion of innovations concept28, 29 may play a role in geographic differences of providers and healthcare systems adopting more novel screening modalities (cotesting or primary hrHPV testing). Healthcare systems may also differ with regards to when they begin offering hrHPV testing and co-testing and to whom it is offered.30 Such differences may
of
be, in part, due to differences in payment structure and population characteristics across the
ro
healthcare systems.
-p
We also hypothesized that insurance coverage may account for geographic differences and our results are consistent with others,10, 18 showing that insured women had a higher prevalence of
re
recent cervical cancer screening compared to the uninsured. However, there was no difference in
lP
co-testing or Pap testing prevalence by insurance status. This may suggest that existing services and programs that provide low income and un- or under-insured women access to cervical cancer screening (e.g., National Breast and Cervical Cancer Early Detection Program [NBCCEDP]31)
na
help offset state-level differences in insurance options.32 Furthermore, the NBCCEDP aims to implement more evidence-based interventions at the provider- and system-level such as patient-
ur
provider reminders and increasing access to timely, appropriate screening.33 Evidence suggests
Jo
that low-income women in states that have expanded Medicaid32 experienced an increased uptake in cervical cancer screening.34 While Pap testing is available in all states regardless of Medicaid eligibility pathway, coverage of follow-up procedures and other cervical cancer screening tests may vary within and across states based on eligibility pathway.35 However, for women aged 30-65, Pap testing alone and co-testing are covered by many other insurance plans, including Medicare.36, 37
Limitations of this study include reliance on self-reported data. While the sensitivity of selfreported Pap testing is high, screening frequency may be overestimated21, 22 due to social desirability bias in this telephone-based survey.38 Furthermore, the accuracy of self-reported hrHPV testing (and co-testing) is not known and may differ by race/ethnicity. Some women may
Journal Pre-proof
16
be unaware if/when an hrHPV test was administered. Saraiya et al. previously reported that about 65% of women were unsure how the HPV test was administered and about one in three women reported being unsure how often they had HPV testing.19 Based on nationally-representative data, Watson et al. reported that about 17% of women were unsure if they had received an HPV test.10 Among women in our sample, a median of about 21% (range: 11% to 28%) indicated they were unsure if they had ever had an HPV test compared to <1% of women who were unsure if they had ever had a Pap test (data not shown). Watson et al.10 also note several factors pertaining to the HPV testing questions in the NHIS that also apply to the BRFSS. For example, 2016 was
of
the first time that HPV testing was included in the core BRFSS questionnaire. Additionally,
ro
implementation of focus groups and more field testing of questionnaire content may provide more insight regarding general knowledge of HPV and HPV testing that could enrich the
-p
questionnaire and resulting data in the future. Compared to our results, those of a study using medical record data had a lower prevalence of overall recent cervical cancer screening and a
re
higher prevalence of co-testing.9 This difference may, in part, be due to the limited information
lP
available in the BRFSS, knowing only when the most recent HPV and Pap tests were administered but unable to discern if they were administered simultaneously. There may also be
na
non-response bias associated with data used for the present study and there were regional differences in the percent of women excluded from our analytic population (Supplemental Table 3), but estimates were weighted to mitigate non-response. In addition, women from eight states
ur
were excluded from this study because hysterectomy data were not available in the 2016 BRFSS.
Jo
Despite these limitations, this study provides valuable state-level estimates of co-testing among women who were recently screened for cervical cancer.
Conclusion There was substantial state-level variability in co-testing prevalence, which was lowest in Midwestern and Southern states. Sociodemographic characteristics and other selected individual factors accounted for some but not all state-level variability in the prevalence of co-testing. Cotesting and hrHPV testing alone is likely to play a more prominent role in cervical cancer screening moving forward through the post-HPV-vaccination era.39, 40 Future studies should aim to assess additional individual factors that may influence state-level differences in co-testing and
Journal Pre-proof monitor the impact of these geographic variations in screening modalities on state-level
Jo
ur
na
lP
re
-p
ro
of
differences in cervical cancer incidence, survival, and mortality.
17
Journal Pre-proof
18
Jo
ur
na
lP
re
-p
ro
of
Acknowledgements: The authors’ salaries for this project are solely funded through American Cancer Society funds.
Journal Pre-proof
19
References
Jo
ur
na
lP
re
-p
ro
of
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019: 1-28. 2. Islami F, Fedewa SA, Jemal A. Trends in cervical cancer incidence rates by age, race/ethnicity, histological subtype, and stage at diagnosis in the United States. Prev Med. 2019: 316-323. 3. Saslow D, Runowicz CD, Solomon D, et al. American Cancer Society guideline for the early detection of cervical neoplasia and cancer. CA Cancer J Clin. 2002;52: 342-362. 4. Saslow D, Solomon D, Lawson HW, et al. American Cancer Society, American Society for Colposcopy and Cervical Pathology, and American Society for Clinical Pathology screening guidelines for the prevention and early detection of cervical cancer. CA Cancer J Clin. 2012;62: 147-172. 5. Moyer VA. Screening for cervical cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2012;156: 880-891, W312. 6. International Agency for Research on Cancer. Cancer Screening in the European Union: Report on the implementation of the Council Recommendation on cancer screening. Lyon, France, 2017. 7. Chrysostomou AC, Stylianou DC, Constantinidou A, Kostrikis LG. Cervical Cancer Screening Programs in Europe: The Transition Towards HPV Vaccination and Population-Based HPV Testing. Viruses. 2018;10: pii: E729. 8. Watson M, Benard V, Flagg EW. Assessment of trends in cervical cancer screening rates using healthcare claims data: United States, 2003-2014. Prev Med Rep. 2018;9: 124-130. 9. MacLaughlin KL, Jacobson RM, Radecki Breitkopf C, et al. Trends Over Time in Pap and Pap-HPV Cotesting for Cervical Cancer Screening. J Womens Health. 2019: 244-249. 10. Watson M, Benard V, King J, Crawford A, Saraiya M. National assessment of HPV and Pap tests: Changes in cervical cancer screening, National Health Interview Survey. Prev Med. 2017;100: 243-247. 11. Nelson W, Moser RP, Gaffey A, Waldron W. Adherence to cervical cancer screening guidelines for U.S. women aged 25-64: data from the 2005 Health Information National Trends Survey (HINTS). J Womens Health. 2009;18: 1759-1768. 12. Goding Sauer A, Siegel RL, Jemal A, Fedewa SA. Current Prevalence of Major Cancer Risk Factors and Screening Test Use in the United States: Disparities by Education and Race/Ethnicity. Cancer Epidemiol Biomarkers Prev. 2019;28: 629-642. 13. MacLaughlan SD, Lachance JA, Gjelsvik A. Correlation between smoking status and cervical cancer screening: a cross-sectional study. J Low Genit Tract Dis. 2011;15: 114-119. 14. Centers for Disease Control and Prevention (CDC). Behavioral Risk Factor Surveillance System Survey Data, 2016. Available from URL: https://www.cdc.gov/brfss/data_documentation/index.htm Accessed August 29, 2017. 15. Curry SJ, Krist AH, Owens DK, et al. Screening for Cervical Cancer: US Preventive Services Task Force Recommendation Statement. Jama. 2018;320: 674-686. 16. Huh WK, Ault KA, Chelmow D, et al. Use of primary high-risk human papillomavirus testing for cervical cancer screening: interim clinical guidance. Gynecol Oncol. 2015;136: 178-182. 17. Austin PC, Merlo J. Intermediate and advanced topics in multilevel logistic regression analysis. Statistics in Medicine. 2017: 3257-3277. 18. White A, Thompson TD, White MC, et al. Cancer Screening Test Use - United States, 2015. MMWR Morb Mortal Wkly Rep. 2017;66: 201-206. 19. Saraiya M, Kwan A, Cooper CP. Primary HPV testing: U.S. women's awareness and acceptance of an emerging screening modality. Prev Med. 2018;108: 111-114. 20. Sauer AG, Liu B, Siegel RL, Jemal A, Fedewa SA. Comparing cancer screening estimates: Behavioral Risk Factor Surveillance System and National Health Interview Survey. Prev Med. 2017: 94-100.
Journal Pre-proof
20
Jo
ur
na
lP
re
-p
ro
of
21. Rauscher GH, Johnson TP, Cho YI, Walk JA. Accuracy of self-reported cancer-screening histories: a meta-analysis. Cancer Epidemiol Biomarkers Prev. 2008;17: 748-757. 22. Howard M, Agarwal G, Lytwyn A. Accuracy of self-reports of Pap and mammography screening compared to medical record: a meta-analysis. Cancer Causes Control. 2009;20: 1-13. 23. Blake KD, Ottenbacher AJ, Finney Rutten LJ, et al. Predictors of human papillomavirus awareness and knowledge in 2013: gaps and opportunities for targeted communication strategies. Am J Prev Med. 2015;48: 402-410. 24. Cooper CP, Saraiya M. Primary HPV testing recommendations of US providers, 2015. Prev Med. 2017;105: 372-377. 25. Walker TY, Elam-Evans LD, Yankey D, et al. National, Regional, State, and Selected Local Area Vaccination Coverage Among Adolescents Aged 13-17 Years - United States, 2017. MMWR Morb Mortal Wkly Rep. 2018;67: 909-917. 26. Saraiya M, Berkowitz Z, Yabroff KR, Wideroff L, Kobrin S, Benard V. Cervical cancer screening with both human papillomavirus and Papanicolaou testing vs Papanicolaou testing alone: what screening intervals are physicians recommending? Arch Intern Med. 2010;170: 977-985. 27. Corbelli J, Borrero S, Bonnema R, et al. Differences among primary care physicians' adherence to 2009 ACOG guidelines for cervical cancer screening. J Womens Health. 2014;23: 397-403. 28. Rogers ED. Diffusion of preventive innovations. Addict Behav. 2002;27: 989-993. 29. Finney Rutten LJ, Nelson DE, Meissner HD. Examination of population-wide trends in barriers to cancer screening from a diffusion of innovation perspective (1987-2000). Prev Med. 2004;38: 258-268. 30. Kamineni A, Tiro JA, Beaber EF, et al. Cervical cancer screening research in the PROSPR I consortium: Rationale, methods and baseline findings from a US cohort. Int J Cancer. 2019;144: 1460-1473. 31. Center for Disease Control and Prevention. National Breast and Cervical Cancer Early Detection Program (NBCCEDP): About the Program. Available from URL: https://www.cdc.gov/cancer/nbccedp/about.htm. 32. The Henry J Kaiser Family Foundation. Status of State Action on Medicaid Expansion Decision. Available from URL: https://www.kff.org/health-reform/state-indicator/state-activity-aroundexpanding-medicaid-under-the-affordable-careact/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D Accessed May 17, 2019. 33. Wong FL, Miller JW. Centers for Disease Control and Prevention's National Breast and Cervical Cancer Early Detection Program: Increasing Access to Screening. J Womens Health (Larchmt). 2019;28: 427-431. 34. Sabik LM, Tarazi WW, Hochhalter S, Dahman B, Bradley CJ. Medicaid Expansions and Cervical Cancer Screening for Low-Income Women. Health Serv Res. 2018;53 Suppl 1: 2870-2891. 35. The Henry J. Kaiser Family Foundation. Medicaid Coverage of Family Planning Benefits: results from a state survey, 2016:47. 36. Centers for Medicare & Medicaid Services. Decision Memo for Screening for Cervical Cancer with Human Papillomavirus (HPV) Testing (CAG-00442N). Available from URL: https://www.cms.gov/medicare-coverage-database/details/nca-decision-memo.aspx?NCAId=278 Accessed July 17, 2019. 37. Centers for Medicare & Medicaid Services. Preventive care benefits for women. Available from URL: https://www.healthcare.gov/preventive-care-women/ Accessed July 17, 2019. 38. Holbrook AL GM, Krosnick JA,. Telephone versus Face-to-Face Interviewing of National Probability Samples with Long Questionnaires: Comparisons of Respondent Satisficing and Social Desirability Response Bias. Public Opinion Quarterly. 2003;67: 79-125.
Journal Pre-proof
21
Jo
ur
na
lP
re
-p
ro
of
39. El-Zein M, Richardson L, Franco EL. Cervical cancer screening of HPV vaccinated populations: Cytology, molecular testing, both or none. Journal of clinical virology : the official publication of the Pan American Society for Clinical Virology. 2016;76 Suppl 1: S62-S68. 40. Lees BF, Erickson BK, Huh WK. Cervical cancer screening: evidence behind the guidelines. Am J Obstet Gynecol. 2016;214: 438-443.
Journal Pre-proof
22
Author contributions
Jo
ur
na
lP
re
-p
ro
of
Ann Goding Sauer: conceptualization, methodology, formal analysis, writing- original draft, writing – review & editing; Priti Bandi: methodology, formal analysis, writing – review & editing; Debbie Saslow: writing- original draft, writing – review & editing; Farhad Islami: writing- original draft, writing – review & editing; Ahmedin Jemal: writing- original draft, writing – review & editing; Stacey Fedewa: conceptualization, methodology, writing- original draft, writing – review & editing, supervision
Journal Pre-proof Ms. No.: PM-19-1239 Title: Geographic and Sociodemographic Differences in Cervical Cancer Screening Modalities Highlights:
ur
na
lP
re
-p
ro
of
Substantial variability in co-testing prevalence across US states Co-testing prevalence was lowest in Midwestern and Southern states Co-testing prevalence variation not entirely explained by individual-level factors
Jo
23