Correlates of Cervical Cancer Screening Adherence among Women in the U.S.: Findings from HINTS 2013-2014

Abstract

Following the latest update of cervical cancer screening guidelines in 2012, we estimate the prevalence of guideline adherent cervical cancer screening and examine its associated factors among a nationally representative sample of US women aged 21-65 years. Our study was based on cross-sectional data from Cycles 3 (2013) and 4 (2014) of the Health Information National Trends Survey (HINTS). The final analytic sample consisted of 2,822 women. Guideline adherent cervical cancer screening was defined as having a Pap test within the last three years. Correlates of guideline adherent cervical cancer screening included socio-demographic and health-related characteristics and HPV/cervical cancer-related beliefs and knowledge items. Multivariable logistic regression analyses were used to estimate prevalence of guideline adherent screening. An estimated 81.3% of women aged 21-65 years reported being screened for cervical cancer within the last three years. Controlling for sociodemographic and health-related characteristics and survey year, women aged 46-65 years were less likely to be guideline adherent than those aged 21-30 years (aPR = 0.89; 95% CI: 0.82 – 0.97). The adjusted prevalence of adherence was significantly higher among married/partnered than among not married women (aPR =1.13; 95% CI: 1.05 – 1.22), and those with one to three medical visits (aPR = 1.30; 95% CI: 1.14 – 1.48), and four or more visits in the past year (aPR = 1.26; 95% CI: 1.09 – 1.45) compared to those with no medical visits. Differences in unadjusted prevalence of guideline adherent screening depending on women’s beliefs and knowledge about HPV and cervical cancer were not significant in adjusted analyses. Lack of interaction with a healthcare provider, being not married/partnered and increasing age continue to be risk factors of foregoing guideline adherent cervical cancer screening.

Introduction

Cervical cancer affects the lives of many women in the United States. In 2014, 12,578 women received a cervical cancer diagnosis, and 4,115 women died from the disease (Centers for Disease Control and Prevention, 2017). The human papillomavirus (HPV) is the primary cause of cervical cancer, and the HPV vaccine is the best method for primary prevention against the types of HPV most likely to cause cervical, vaginal, and vulvar cancers. For secondary prevention, the Papanicolaou (Pap) test examines any cell changes in the cervix to prevent cervical cancer from developing. The American Cancer Society (ACS) currently recommends women between 21 and 29 years receive a Pap test every 3 years, and women between 30 and 65 combine regular Pap testing with HPV DNA testing (“cotesting”) every 5 years or complete Pap testing alone every 3 years (Moyer & U. S. Preventive Services Task Force, 2012; R. A. Smith et al., 2015). The 2012 ACS guidelines - which recommend that women who are at average risk for cervical cancer not pursue annual testing - are consistent with the current recommendations of the American College of Obstetricians and Gynecologists (ACOG) and the United States Preventive Services Task Force (USPSTF; ACOG, 2016; Saslow et al., 2012; U.S. Preventive Services Task Force, 2012).

Following these recommendations, there has been some research on whether patients are receiving provider recommendations on screening interval guidelines, and what types of interventions could affect patients’ screening behaviors (Ashok, Berkowitz, Hawkins, Tangka, & Saraiya, 2012; Roland, Benard, Greek, Hawkins, & Lin, 2016). False positive tests resulting from overscreening have been suggested to produce unnecessary anxiety in women and add to medical costs (Kauffman, Griffin, Lund, & Tullar, 2013; Saraiya et al., 2010). Another concern is that while recommendations may change in terms of the recommended screening interval, patients may be unaware of such changes, not have access to recent medical information, react negatively to new recommendations, or simply ignore any changes in recommendations, as reported in one study of medically underserved, immigrant women following changes in breast cancer screening guidelines (Nagler, Lueck, & Gray, 2017).

A national survey study concluded acceptance of an extended cervical cancer screening interval was more likely among women who had received the HPV vaccine and had received a prior Pap test; however, only 6% of participants in the sample reported they were currently being screened according to the recommended interval (Cooper, Saraiya, & Sawaya, 2015). Alternatively, in another study conducted with community health center patients in Illinois, the majority believed that because longer screening intervals would lead to cancer worry, only one-third were either “somewhat” or “very likely” to adhere to their provider’s recommendation for a 3-year screening interval. Another study reported patients believed in the importance of annual screening, and an intervention using the HPV DNA test and an educational brochure to encourage women to consider long screening intervals had little effect on changing women’s perceptions (Roland et al., 2016).

While there has been a precipitous decline in cervical cancer mortality rates over the last 30 years, some groups of women including Hispanics, African-Americans, newer immigrants, and the uninsured are at higher risk because they are either not receiving quality cervical cancer screening or do not have adequate access to follow-up care after an abnormal screening result (Sabatino et al., 2015). Approximately 50% of cervical cancer cases have been identified in women who are not up-to-date with screening, and 13% of cases have been found in women who did not receive follow-up care after an abnormal result (Akers, Newmann, & Smith, 2007; Eggleston, Coker, Das, Cordray, & Luchok, 2007; Spence, Goggin, & Franco, 2007; Sung et al., 2000). Most cases of cervical cancer are both highly preventable – through HPV vaccination and screening – and treatable if patients follow appropriate follow-up procedures after an abnormal screening result (Eggleston et al., 2007; J. S. Smith et al., 2013; Spence et al., 2007).

We explored possible factors associated with adherence to recommended screening intervals (last Pap test ≤ 3 years) in a nationally representative sample of women aged 21-65 years following the latest update of cervical cancer screening guidelines in 2012. We sought to determine the association between sociodemographic characteristics, insurance status, frequency of contact with healthcare professionals, self-reported health status, and HPV/cervical cancer knowledge with adherence to cervical cancer screening recommendations, since these variables were correlated with screening in previous studies using HINTS data (Ashok et al., 2012; Ojeaga, Alema-Mensah, Rivers, Azonobi, & Rivers, 2017). Based on prior research, possible variables were tested for association with screening; for example, women who self-identified as Hispanic or non-Hispanic Black, were uninsured, had less contact with physicians, and were less knowledgeable about HPV would be less likely to adhere to cervical cancer screening recommendations than women without these characteristics (Cowburn, Carlson, Lapidus, & DeVoe, 2013; Shoemaker & White, 2016; Walsh & O’Neill, 2015). Analyses and findings resulting from national survey research studies can inform both health policy and service delivery to better address and ultimately reduce cervical cancer health disparities.

Methods

Data Source

We used data from the Heath Information National Trends Survey (HINTS), Cycles 3 (2013) and 4 (2014; National Cancer Institute, 2016). HINTS is a nationally representative cross-sectional survey sponsored by the National Cancer Institute. The survey collects data routinely about the public’s use of cancer-related information and how the public perceives cancer risks, including preventive screenings. Data used in this study were collected by mail from September through December 2013 (Cycle 3) and from August through November 2014 (Cycle 4). During the first stage of a two-stage sampling design, a stratified sample of addresses was selected from a file of residential addresses. During the second stage, one adult from each sample household responded to the survey. The overall household response rates were 35.2% and 34.4% for Cycles 3 and 4, respectively, with lower response rates in both cycles for high minority areas (Westat, 2014, 2015). However, stratifying addresses in areas with high concentrations of minority populations permitted oversampling of this stratum.

There were 6,862 observations in the two cycles combined; with sex and age data missing for 373 observations. Consistent with the guidelines from ACOG, ACS, and USPSTF, we focused on 2,930 women aged 21 years to 65 years because these women are recommended to receive cervical cancer screening. We then excluded 55 women who indicated they had a history of cervical cancer. Among the 2,875 remaining women, 53 observations were missing information on being up-to-date with cervical cancer screening. Our final analytic sample consisted of 2,822 women who responded to the screening history question.

Cervical Cancer Screening

Female adults aged 18 and older were asked a categorical response question, “how long ago did you have your most recent Pap test to check for cervical cancer?” The response options were: “a year ago or less”; “more than 1, up to 2 years ago”; “more than 2, up to 3 years ago”; “more than 3, up to 5 years ago”; “more than 5 years ago”; and “Never.” Additionally, there was a new question introduced in Cycle 3 which asked whether a doctor had ever told their patients they could choose whether to have a Pap test, given the update in screening interval guidelines. Based on responses to the screening history question and consistent with the nationally recommended screening intervals, we dichotomized participants aged 21-65 for adherence based on the 3-year interval.

Correlates of Cervical Cancer Screening

Consistent with prior research (Cowburn et al., 2013; Shoemaker & White, 2016; Walsh & O’Neill, 2015), we considered several sociodemographic and health-related variables associated with cervical cancer screening. These variables included: age; race and ethnicity; education level; health insurance coverage; having a regular doctor, nurse, or other health care professional; number of medical office visits in the past year; marital status; Body Mass Index (BMI) category; smoking status; family history of cancer; and self-reported health status.

The health beliefs around disease prevention, HPV/cervical cancer and cervical cancer screening were examined using responses to seven questions related to (1) the likelihood of looking for information about cancer from any source; (2) beliefs about one’s lifetime chance of getting cancer; (3) confidence in one’s ability to take good care of one’s health; (4) knowledge of ever having heard of HPV; (5) memory of a doctor ever telling them they could choose to have a Pap test; (6) belief about whether HPV could cause cervical cancer; and (7) belief about whether HPV was a sexually transmitted disease.

Statistical Analysis

First, we calculated the weighted percentages of guideline adherent cervical cancer screening, sociodemographic characteristics, and items capturing women’s knowledge and beliefs on HPV/cervical cancer and screening. We then examined differences in distribution of these characteristics and items by adherence to recommended cervical cancer screening intervals using chi-square tests and contrasts of marginal linear predictions from binary logistic regressions.

Next, because guideline adherent screening is a non-rare outcome, we calculated adjusted prevalence ratios (aPR) instead of odds ratios of guideline adherent cervical cancer screening by controlling for all the sociodemographic characteristics and survey year. We included survey year to control for temporal trends in guideline adherent cervical cancer screening. We used the PREDMARG statement in SUDAAN’s PROC RLOGIST procedure (Bieler, Brown, Williams, & Brogan, 2010). With an additional seven models, we estimated aPR of cervical cancer screening given each of the belief or knowledge items adjusting for sociodemographic characteristics and survey year (results not shown).

Stata/SE 14.1 for Windows (StataCorp LP, College Station, TX) was used for all descriptive and bivariate analyses. PREDMARG option in SAS-Callable SUDAAN (version 9.4; Research Triangle Institute, Research Triangle Park, NC) logistic regression procedure was employed for the model-adjusted PRs. All analyses were weighted to account for the complex sampling design used in HINTS and nonresponse and incorporated the jackknife replicate weights needed to compute accurate standard errors. Respondents who had missing values for relevant variables used in any statistical analyses were excluded from the analyses. Statistical significance level was set at α = 0.05. All tests were two-tailed.

Results

Participant demographic characteristics are shown in Table 1. In the sample, 22.1% of women were aged 21-30 years, 35.4% were aged 31-45 years, and 42.6% were aged 46-65. The majority of women in the sample were non-Hispanic White (64.0%), married or living as married (60.3%), had a regular health care provider (62.9%), and had health insurance coverage (81.7%). Approximately 52% reported one to three medical office visits and 32.9% reported four or more medical office visits in the past year. According to their BMI, over 25% of women were overweight and 33.7% were obese. Approximately 76% had a history of cancer in their immediate family.

Table 1.

Sociodemographic and Health-Related Characteristics, HINTS 2013-2014 (n ^a = 2,822)

Characteristic	n a	% b	95% CI
Adherence to cervical cancer screening c
Pap > 3 years or never	515	18.9	16.4 – 21.2
Pap ≤ 3 years	2,307	81.3	78.8 – 83.6
Age group (years)
21-30	360	22.1	19.8 – 24.5
31-45	818	35.4	33.0 – 37.8
46-65	1,644	42.6	41.1 – 44.1
Race/ethnicity
Hispanic	491	15.3	13.8 – 16.9
Non-Hispanic White	1,409	64.0	62.6 – 65.3
Non-Hispanic Black	514	13.2	11.8 – 14.7
Other	226	7.55	6.37 – 8.93
Marital status
Not married	1,346	39.7	37.7 – 41.8
Married/living as married	1,449	60.3	58.2 – 62.3
Education
Less than high school/high school	698	28.4	26.7 – 30.2
Some college	839	31.9	29.8 – 34.0
College graduate or higher	1,272	39.8	38.5 – 41.1
BMI category
Underweight/normal	992	39.2	36.6 – 42.2
Overweight	761	27.1	24.7 – 29.7
Obese	953	33.7	31.2 – 36.3
Health insurance
No	426	18.3	16.5 – 20.3
Yes	2,367	81.7	79.7 – 83.5
Have regular health professional
No	959	37.1	34.4 – 39.8
Yes	1,815	62.9	60.2 – 65.6
Times visiting healthcare professional during the past 12 months
None	403	15.5	13.5 – 17.7
1-3 times	1,386	51.7	48.8 – 54.5
≥ 4 times	980	32.9	30.4 – 35.4
Smoking status
Never smoker	1,826	63.7	60.9 – 66.4
Former smoker	552	18.5	16.6 – 20.7
Current smoker	433	17.8	15.5 – 20.4
Has any family member had cancer?
No	654	24.5	22.2 – 27.1
Yes	1,950	75.5	73.0 – 77.8
Self-reported health status
Excellent/very good	1,317	47.3	44.4 – 50.2
Good	976	38.5	35.7 – 41.4
Fair/poor	469	14.2	12.5 – 16.2

^aunweighted number of observations.

^bweighted to reflect national estimates.

^cAmerican Cancer Society (ACS), American Society for Colposcopy and Cervical Pathology (ASCCP), and American Society for Clinical Pathology (ASCP) 2012 (Saslow et al., 2012); U.S. Preventive Services Task Force (USPSTF) 2012 (USPSTF, 2012); American College of Obstetricians and Gynecologists (ACOG) 2016 (ACOG, 2016).

As reported in Table 2, almost 60% of women reported looking for cancer information from any source. Just over one-third considered themselves to be either “likely” or “very likely” to develop cancer in their lifetime. Over two-thirds were “completely confident” or “very confident” in their ability to take care of their health. Over 80% had heard of HPV, and over 75% believed HPV caused cervical cancer. Almost two-thirds of participants believed HPV was a STD, and the other one-third of participants either disagreed or were not sure.

Table 2.

Description of HPV and Cervical Cancer Related Beliefs and Knowledge Items: HINTS 2013-2014

Characteristic	n a	% b	95% CI
Have looked for cancer information from any source?
No	1,082	41.6	38.7 – 44.5
Yes	1,514	58.4	55.5 – 61.3
Do you think your chance of cancer is
Very unlikely/unlikely	553	17.4	15.5 – 19.4
Neither unlikely nor likely	1,202	47.2	44.4 – 50.1
Likely/very likely	872	35.4	32.4 – 38.5
How confident are you in your ability to take care of your health
Completely/very confident	1,918	67.5	64.6 – 70.3
Somewhat confident	691	27.1	24.4 – 30.0
Little/not confident	155	5.4	4.4 – 6.7
Have you heard of HPV?
No	588	19.7	17.4 – 22.2
Yes	2,217	80.3	77.8 – 82.6
Has a doctor ever told you that you could choose whether or not to have a Pap test?
No	1,681	58.0	54.8 – 61.1
Yes	1,129	42.0	38.9 – 45.2
Do you think HPV can cause cervical cancer?
No	63	2.2	1.5 – 3.1
Yes	1,576	75.7	72.9 – 78.4
Not sure	552	22.1	19.6 – 24.8
Do you think HPV is an STD?
No	423	18.5	16.3 – 20.8
Yes	1,331	63.5	60.3 – 66.5
Not sure	438	18.1	15.9 – 20.5

^aunweighted number of observations;

^bweighted to reflect national estimates;

CI – confidence interval.

An estimated 81.3% of women aged 21-65 years reported being screened for cervical cancer within the last three years. In unadjusted analyses, guideline-adherent cervical cancer screening was significantly associated with age, marital status, education, health insurance, medical office visits, and self-reported health status, as described in Table 3. A significantly smaller percentage of women aged 46-65 years (77.1%) reported guideline-adherent screening compared to women aged 31-45 years (87.3%; p < 0.001). Fewer women who reported not being married (76.3%) than married (84.8%) were guideline adherent (p < 0.001). Higher educational levels were significantly associated with a higher unadjusted prevalence of guideline adherent screening (p = 0.002): from 75.8% among women with high school or less education to 80.5% among women with some college and 85.9% among college graduates or higher. There were no significant differences by race (Black/White) or ethnicity (Hispanic/non-Hispanic). A higher percentage of women with any type of health insurance reported guideline adherent screening compared to those who were uninsured (p < 0.001). A significantly lower percentage of women with no medical office visits reported guideline adherent screening (60.6%) than those with one to three visits (85.5%; p < 0.001) and 4 or more visits (84.3%; p < 0.001). A significantly higher percentage of women in “excellent/very good health” reported guideline adherent screening (85.0%) than those in “good” (77.8%; p = 0.005) and “fair/poor health” (78.4%; p = 0.047).

Table 3.

Prevalence of Guideline-Adherent Cervical Cancer Screening by Sociodemographic and Health-Related Characteristics, HINTS 2013-2014

	Unadjusted		Adjusted
Characteristic	% a	95% CI	PR b	95% CI
Age group (years)*
21-30	79.8	71.2 – 86.4	Ref.
31-45	87.3	83.1 – 90.6	1.03	0.93 - 1.13
46-65	77.1	73.7 – 80.2	0.89	0.82 - 0.97
Race/ethnicity
Hispanic	83.7	76.3 – 89.1	Ref.
Non-Hispanic White	81.4	77.9 – 84.5	0.94	0.86 – 1.04
Non-Hispanic Black	86.8	79.3 – 91.9	1.02	0.92 – 1.13
Other	75.5	64.1 – 84.2	0.87	0.74 – 1.03
Marital status*
Not married	76.3	71.4 – 80.5	Ref.
Married/living as married	84.8	82.2 – 87.2	1.13	1.05 – 1.22
Education*
Less than high school/high school	75.8	70.4 – 80.5	Ref.
Some college	80.5	74.6 – 85.2	1.01	0.93 – 1.11
College graduate or higher	85.9	78.8 – 83.6	1.07	0.97 – 1.17
BMI category
Underweight/normal	79.7	74.8 – 83.9	Ref.
Overweight	84.4	79.8 – 88.1	1.06	0.98 – 1.15
Obese	80.9	76.7 – 84.5	1.03	0.94 – 1.12
Health insurance*
No	68.9	60.8 – 76.1	Ref.
Yes	84.1	81.4 – 86.5	1.14	0.99 – 1.32
Have regular health professional*
No	75.7	70.9 – 80.0	Ref.
Yes	84.7	81.8 – 87.2	1.06	0.98 – 1.14
Times visiting healthcare professional during the past 12 months*
None	60.6	52.8 – 67.8	Ref.
1-3 times	85.5	82.3 – 88.3	1.30	1.14 – 1.48
≥4 times	84.3	79.6 – 88.1	1.26	1.09 – 1.45
Smoking status
Never smoker	82.6	79.4 – 85.4	Ref.
Former smoker	81.0	74.5 – 86.2	1.00	0.92 – 1.10
Current smoker	79.7	69.4 – 82.7	1.02	0.92 – 1.12
Has any family member had cancer?
No	81.5	75.6 – 86.3	Ref.
Yes	81.3	78.0 – 84.3	1.02	0.95 – 1.10
Self-reported health status*
Excellent/very good	85.0	81.2 – 88.0	Ref.
Good	77.8	73.8 – 81.3	0.95	0.88 – 1.02
Fair/poor	78.4	71.9 – 83.7	0.96	0.87 – 1.07

^aweighted to reflect national estimates;

^bPR – prevalence ratio, adjusted for all the characteristics listed in this table and survey year. Ref. – reference group.

^*Statistically significant bivariate associations. Statistically significant differences in adjusted estimates are in bold, p < 0.05.

When adjusting for all sociodemographic and health-related characteristics and survey year (Table 3), statistically significant associations remained between guideline adherent screening and each of the three characteristics: age, marital status, and number of medical office visits. The adjusted prevalence of guideline-adherent cervical cancer screening was significantly lower among women aged 46-65 years (aPR = 0.89; 95% CI: 0.82 – 0.97) than those aged 21-30 years. Women who were married or living as married had a higher adjusted probability of having a Pap within the last three years compared to women who were not married (aPR =1.13; 95% CI: 1.05 – 1.22). The adjusted prevalence of guideline adherent screening was significantly higher among women with one to three medical visits and four or more medical visits than those who had not visited a healthcare professional during the past year: aPR = 1.30 (95% CI: 1.14 – 1.48) and aPR = 1.26 (95% CI: 1.09 – 1.45).

Out of seven HPV/cervical cancer related belief and knowledge items, three were significantly associated with guideline adherent cervical cancer screening based on unadjusted analyses, as shown in Table 4. A significantly higher percentage of women who had heard of HPV reported guideline adherent screening (84.0%) than those who had not heard of HPV (70.4%; p < 0.001). A significantly lower percentage of women who were not sure whether HPV caused cervical cancer reported guideline adherent screening (78.6%) than those who knew (85.4%; p = 0.039). A significantly higher percentage of women who thought HPV was an STD reported guideline adherent screening (86.5%) than women who were not sure (77.9%; p = 0.013). These differences in unadjusted prevalence of guideline adherent screening depending on women’s beliefs and knowledge were not significant in analyses adjusted for women’s sociodemographic and health-related characteristics and survey year (results not shown).

Table 4.

Description of HPV and Cervical Cancer Related Beliefs and Knowledge Items by Adherence to Cervical Cancer Screening Guidelines: HINTS 2013-2014

Characteristic	Unadjusted
	% a	95% CI
Have looked for cancer information from any source?
No	79.6	75.5 – 83.2
Yes	82.4	78.8 – 85.5
Do you think your chance of cancer is
Very unlikely/unlikely	84.2	78.9 – 33.4
Neither unlikely nor likely	83.1	79.5 – 86.2
Likely/very likely	78.0	72.9 – 82.5
How confident are you in your ability to take care of your health?
Completely/very confident	82.8	79.8 – 85.4
Somewhat confident	78.9	72.9 – 83.8
Little/not confident	74.3	64.0 – 82.4
Have you heard of HPV? *
No	70.4	63.5 – 76.5
Yes	84.0	81.3 – 86.4
Has a doctor ever told you that you could choose whether or not to have a Pap test?
No	79.7	76.3 – 82.6
Yes	83.6	79.7 – 86.8
Do you think HPV can cause cervical cancer? *
No	89.2	72.6 – 96.3
Yes	85.4	82.3 – 88.0
Not sure	78.6	72.2 – 83.8
Do you think HPV is an STD? *
No	81.5	75.2 – 86.4
Yes	86.5	83.3 – 89.2
Not sure	77.9	70.7 – 83.7

^aweighted to reflect national estimates.

^*Statistically significant differences, p < 0.05.

Discussion

Factors associated with cervical cancer screening adherence were estimated in a US population-based sample of women between 21 and 65 years. This study examined sociodemographic, healthcare utilization, and cervical cancer/HPV belief and knowledge variables associated with adherence to cervical cancer screening in previous studies using HINTS data (Ashok et al., 2012; Nelson, Moser, Gaffey, & Waldron, 2009). The overall proportion of women receiving guideline adherent screening in this sample was 81.3%, which is slightly lower than the 82.6% survey estimate for all U.S. States and Washington, D.C. from the Behavioral Risk Factor Surveillance System (Centers for Disease Control and Prevention, 2015). These screening estimates are lower than the 2008 baseline set for Healthy People 2020 of 84.5%, and well below the 93% 2020 target objective (U.S. Department of Health and Human Services, 2016). Therefore, the current data from HINTS indicates that more work is needed to increase screening rates by connecting women with a regular health care provider to provide opportunities for cancer education and screening, especially for middle aged women between 46 to 65 years of age.

For sociodemographic and health characteristics, results of the unadjusted analyses indicated age, marital status, education, health insurance status, having a regular healthcare provider, frequency of healthcare visits, and self-reported health status were associated with adherence to cervical cancer screening recommendations. For the belief and knowledge items, having heard of HPV, believing that HPV caused cervical cancer, and thinking that HPV was a STD were all positively associated with adherence to screening recommendations. In a previous study with HINTS data, Pap test adherence using the one-year screening interval was associated with older age groups between 31-64, being non-Hispanic Black or non-Hispanic White, being a college graduate, having at least one visit to a healthcare provider in the last year, and HPV awareness (Ashok et al., 2012). Moreover, in other past studies with HINTS data, researchers reported differences in adherence to cervical cancer screening by race and ethnicity (Hirth, Laz, Rahman, & Berenson, 2016; Nelson et al., 2009). Therefore, the results of this study were surprising in not identifying any disparities in screening rates by race and ethnicity.

In the adjusted analyses, only the variables for age, marital status, and frequency of healthcare visits were associated with adherence to screening recommendations. Women who were between 46-65 years, not married, and had not visited their healthcare provider in the last year were significantly less likely to be adherent. The finding that age was inversely related to having had a recent Pap test could possibly be explained by a history of negative Pap tests. Patients could interpret negative Pap smear results as giving them license to engage in less frequent screening or no screening at all. A study based on a multivariate analysis of 2007 HINTS data reported women in that older age category were less likely to have a Pap test in the last year and more likely to have received one over three years ago (Ashok et al., 2012). The same trend for screening by age category was also reported in a study of 2005 HINTS data, but age was not significant in either bivariate analyses or multivariate models (Nelson et al., 2009). A community-based intervention study in the Appalachian region of Kentucky reported lower uptake of screening by women between 55-59 years compared to women 40-44 years (Studts et al., 2012). While women might perceive that they are at decreased risk for cervical cancer as they grow older, incidence and mortality rates for cervical cancer are not significantly lower in upper age brackets, especially when correcting for women who have received a hysterectomy (Rositch, Nowak, & Gravitt, 2014). Adequate screening at ages 50 to 64 years is significantly associated with substantially lower risk of cervical cancer after the age of 65 years (Castanon, Landy, Cuzick, & Sasieni, 2014). A study based on data from the 2013 and 2015 National Health Interview Surveys (NHIS) reported incidence rates for cervical cancer among older women not up-to-date with screening did not decline until at least the age of 85 years. The proportion of women who reported they never had a Pap test or that their most recent screening had been over five years ago increased with age, from 12.1% for women aged 41-45 years to 18.4% for women aged 61-65 years. Premature discontinuation of routine screening among women younger than 65 years old could contribute to preventable cases of invasive cervical cancer and deaths (White, Shoemaker, & Benard, 2017). These studies highlight the importance of increasing screening adherence prior to the age of 65 years and reinforce the importance of our finding that women are being screened less as they age.

The finding that marital status was associated with better adherence rates (85% adherent in the “married” category vs. 76% in the “not married” category) has been reported in a study based on Behavioral Risk Factor Surveillance System (BRFSS) data as an independent predictor of breast, cervical, and colorectal cancer screening (Hanske et al., 2016). For cervical cancer screening, married participants had higher odds of cervical cancer screening (OR = 1.29, 95% CI: 1.16-1.43). Another study using BRFSS data reported men and women who were either married or living as married had significantly higher odds of being up-to-date with colorectal cancer screening than other individuals (El-Haddad, Dong, Kallail, Hines, & Ablah, 2015). A chart review study of low-income Hispanic women in central Florida reported significantly more married (84%) than unmarried women (69%) were adherent to screening guidelines (Luque et al., 2011). Few studies have examined the spousal or partner involvement or support in the utilization of cervical cancer screening. A lack of male involvement has been suggested by some studies as an overlooked risk factor for non-adherence (Rosser, Zakaras, Hamisi, & Huchko, 2014). Further quantitative and qualitative research into the effect of spousal or partner support on women’s cervical cancer screening behaviors is needed.

In a prior study of 2005 HINTS data, adherence to screening within the last three years was 84%, (Nelson et al., 2009) and in our study of 2013 and 2014 HINTS data, prevalence of adherence was 81%. The only sociodemographic variable identified as a significant predictor for screening adherence was health insurance coverage in the former study (Nelson et al., 2009). In our analysis, health insurance coverage was only significant in the unadjusted analysis, but number of medical office visits was significant for both categories (1-3; ≥4 times) compared to no visits. This variable was not included in that prior study. However, another study using 2007 HINTS data reported an association between more visits and receiving a Pap test within one year (Ashok et al., 2012). Our finding is consistent with studies that report women with more frequent medical visits have greater contact with the healthcare system and are more likely to be encouraged or reminded to receive a Pap test (Katz et al., 2015). While the periodic well-woman visit includes an annual pelvic exam, many physicians continue to recommend annual Pap testing and may be concerned that they would lose regular contact with their patients if they recommended less frequent screening intervals (Roland et al., 2013; Saraiya et al., 2010). An intervention study with providers to recommend the longer screening interval of 3 to 5 years reported providers were more likely to make the updated recommendation and were less concerned that patients would have less contact with the health care system (Benard et al., 2016). Because we were unable to ascertain if women received their Pap test as a routine visit or a special appointment, it was not possible to determine the nature of these medical visits.

Limitations

There were several limitations to our study. First, because the question on non-routine reasons for visiting a health care provider was not asked in the latest iterations of HINTS, we were not able to identify and exclude women who might have had their most recent cervical cancer screening as a retest for a previous abnormal test or follow-up after a hysterectomy. Similarly, we were not able to exclude women who might have had a hysterectomy since this question was not asked in the current version. Such exclusion criteria have been applied in studies using earlier HINTS data (Ashok et al., 2012). The prevalence of hysterectomy in the total population of women aged 20 years and older has been estimated at 20.1% (Rositch et al., 2014). Based on the most recent estimates of the proportion of the U.S. female population aged 41 to 70 years with no hysterectomy who had not been recently screened for cervical cancer (defined as never having a Pap test or the last Pap test was over 5 years ago), 12.1% were aged 41 to 45 years, 11.7% were aged 46 to 50 years, 11.3% were aged 51 to 55 years, 17.1% were aged 56 to 60 years, and 18.4% were aged 61 to 65 years (White et al., 2017). Hence, correcting for hysterectomy, the proportion of guideline-adherent women is similar to the one reported in our study.

Second, self-reported HINTS data might not have accurately captured screening practices of women. Concordance rates for medical records and self-reported data suggest the prevalence of cancer screening is overestimated by self-report, whereas the receipt date of the last test is underestimated (Rauscher, Johnson, Cho, & Walk, 2008). Moreover, it is not possible to identify frequent users of Pap testing since HINTS no longer includes a question on Pap testing interval prior to the most recent screening. Therefore, while the issue of screening more frequently than necessary is a problem worthy of study, the elimination of this question precludes analysis of whether screening intervals are changing for some women.

Lastly, given the cross-sectional nature of our data, we caution naïve readers against making inferences on different causal pathways in which sociodemographic, health-related, belief and knowledge variables may influence screening. Many of the factors studied, such as age, education, demographic characteristics, marital status, and insurance status are correlated, so it is problematic to isolate individual variables as specific predictors of screening. It is possible for beliefs and knowledge, for example, to influence screening practice, particularly if screening recommendation or other patient education is delivered during a healthcare provide visit.

Conclusions

In conclusion, our study provides an update on cervical cancer screening patterns in the US population following the 2012 consensus updates on screening by major organizations. Regardless, many women may have been engaging in a three-year screening interval before the consensus update based on a doctor’s recommendation. Our findings demonstrate screening rates in 2013-14 are below the target rates according to Healthy People 2020; therefore, continued efforts to encourage underscreened women to schedule a well-woman exam are needed. Future studies will be able to study the relative rates of overuse of screening and potential harms from both the individual (e.g., patient anxiety) and societal (e.g., additional healthcare costs) perspectives (Coronado et al., 2013).