RESEARCH AND PRACTICE

Influence of Hormonal Contraceptive Use and Health

Beliefs on Sexual Orientation Disparities in Papanicolaou
Test Use
Brittany M. Charlton, MSc, Heather L. Corliss, MPH, PhD, Stacey A. Missmer, ScD, A. Lindsay Frazier, MD, Margaret Rosario, PhD, Jessica A. Kahn, MD,
MPH, S. Bryn Austin, ScD

Reproductive health screenings, such as Papa-

Objectives. Reproductive health screenings are a necessary part of quality
nicolaou (Pap) tests, are a necessary part of health care. However, sexual minorities underutilize Papanicolaou (Pap) tests
quality health care.1 These screenings can de- more than heterosexuals do, and the reasons are not known. Our objective was
tect precancerous conditions that may lead to to examine if less hormonal contraceptive use or less positive health beliefs
cervical cancer if not treated. Despite the health about Pap tests explain sexual orientation disparities in Pap test intention and
benefits, sexual minority (e.g., lesbian, bisexual) utilization.
female adolescents and young adults under- Methods. We used multivariable regression with prospective data gathered
utilize Pap tests.2---7 In a previous study with the from 3821 females aged 18 to 25 years in the Growing Up Today Study (GUTS).
cohort employed in this study, sexual minori- Results. Among lesbians, less hormonal contraceptive use explained 8.6% of
the disparities in Pap test intention and 36.1% of the disparities in Pap test
ties, compared to heterosexuals, were less likely
utilization. Less positive health beliefs associated with Pap testing explained
to have had a Pap test within the last year or in
19.1% of the disparities in Pap test intention. Together, less hormonal contra-
their lifetime.8 We now extend this finding by
ceptive use and less positive health beliefs explained 29.3% of the disparities in
exploring how hypothesized determinants of Pap test intention and 42.2% of the disparities in Pap test utilization.
Pap testing intentions and behaviors help to Conclusions. Hormonal contraceptive use and health beliefs, to a lesser
explain sexual orientation disparities. A more extent, help to explain sexual orientation disparities in intention and receipt of
thorough understanding of factors that explain a Pap test, especially among lesbians. (Am J Public Health. 2014;104:319–325.
underutilization may enable targeted clinical doi:10.2105/AJPH.2012.301114)
care and public health messages.
It has been proposed that lesbians are less
likely than heterosexuals to regularly visit test.12 Sexual minority adolescents may be use and HBM constructs (i.e., risk perception,
a gynecologist because of their lower likelihood particularly at risk because negative percep- attitudes and beliefs, cues to action) as
of hormonal contraceptive use.6,7 Other factors tions and attitudes about HPV are asso- they relate to Pap testing among adolescents
that may explain screening disparities include ciated with lower screening rates among and young adults. We hypothesized that
lower risk perception3 of cervical cancer and adolescents.13 unlike completely heterosexuals, sexual
negative attitudes and beliefs toward Pap tests.9 The Health Belief Model (HBM) is a psycho- minorities
Previous studies report that lesbians, in con- logical model that aims to explain and predict
trast to heterosexuals, perceive themselves to health behaviors. To our knowledge, it has 1. use hormonal contraceptives less frequently
be at lower risk for contracting the human yet to be applied to Pap test use within various (and therefore may not see a health care
papillomavirus (HPV) and subsequently devel- sexual orientation groups but has effectively provider as frequently who could provide
oping cervical cancer. Common reasons in- explained health disparities in other popula- a Pap test);
clude the belief that sex between women is tions.14 The model includes 4 constructs fo- 2. perceive themselves to be at lower risk for
intrinsically safe because of limited fluid trans- cused on an individual’s attitudes and beliefs: cervical cancer;
mission, feeling invisible in gay male-focused susceptibility, severity, benefits, and barriers.15 3. perceive themselves to be at lower severity
sexual health promotions and therefore feeling An additional concept, cues to action, is of cervical cancer, if a diagnosis did occur;
“safe” because of a lack of messaging, and the thought to further motivate behavior,16 4. believe there are fewer benefits to Pap
emphasis in the “safer sex” discourse that one as are normative beliefs. test use;
needs to only worry about penile penetrative The Institute of Medicine states that more 5. believe there are more barriers to Pap
sex.10 Physician recommendations have proved data on Pap test use among sexual minority test use;
to be the strongest factor associated with Pap women are needed to better inform medical, 6. hold fewer normative beliefs about Pap
test utilization among all women.11 However, governmental, and educational policies.17 test use;
10% of bisexuals and lesbians report being Therefore, we examined sexual orientation 7. receive fewer cues to action for Pap test use;
discouraged by physicians from getting a Pap group disparities in hormonal contraceptive and

February 2014, Vol 104, No. 2 | American Journal of Public Health Charlton et al. | Peer Reviewed | Research and Practice | 319
 RESEARCH AND PRACTICE

8. perceive fewer other individuals utilizing screened, given the transient nature of HPV, Hormonal contraception. Hormonal contra-
Pap tests. slow growth of cervical cancer, and subsequent ception use was assessed in the 2007 ques-
unwarranted treatment. tionnaire wave with 3 questions: “During the
Additionally, we hypothesized that these past year did you use 1. . .birth control pills,
factors would partially explain (i.e., mediate) Measures patch (Ortho-Evra), ring (Nuvaring), or injectable
sexual orientation group disparities in Pap test Sexual orientation. Sexual orientation was estrogen for any reason; 2 . . . Depo Provera; or
intention and utilization. measured with a question from the 2005 3 . . . Implanon?” Responses included “yes”
questionnaire wave that was adapted from and “no.” If a participant responded “yes” to
METHODS the Minnesota Adolescent Health Survey19,20 any of these questions, they were considered
asking about identity and feelings of attraction: to be a hormonal contraceptive user.
The Growing Up Today Study (GUTS) is “Which of the following best describes your HBM constructs. HBM constructs consisting
a longitudinal cohort of US adolescents and feelings?” Responses included “1. completely of 4 broad domains were assessed on the
young adults that started in 1996 to assess heterosexual (attracted to persons of the op- 2005 questionnaire wave with 4- and 5-level
various health topics. Women from the Nurses’ posite sex), 2. mostly heterosexual, 3. bisexual Likert scales: 1. risk perception; 2. attitudes/
Health Study II, who were the mothers of the (equally attracted to men and women), 4. beliefs; 3. cues to action; and 4. perception of
GUTS participants, provided consent and contact mostly homosexual, 5. completely homosexual others’ Pap test utilization.
information for their children, aged 9 to 14 years. (gay/lesbian, attracted to persons of the same The first domain, risk perception regarding
Questionnaires were mailed to more than sex), 6. not sure.” The “mostly homosexual” cervical cancer, included 2 areas: susceptibility
25 000 potential GUTS participants, of whom, and “completely homosexual” responses were and severity. Perceived susceptibility was
9039 girls (68%) and 7843 (58%) boys combined to form a lesbian category to in- measured by using: “How likely is it that
returned completed questionnaires, indicating crease statistical power. A second question you will get cervical cancer sometime in your
their consent. Questionnaires are mailed an- measured sex of sexual contacts in 2005 with life?” while perceived severity was assessed
nually or biennially, and more detailed infor- a question reading: “During your life, the by using: “If you were to get cervical cancer,
mation on the initial recruitment is available person(s) with whom you have had sexual how serious a problem would it be?”
elsewhere.18 Detailed reproductive health and contact is (are). . .” Responses included “I have The second domain, attitudes and beliefs
sexual risk behavior questions were assessed not had sexual contact with anyone,” “females,” regarding Pap testing, included 3 areas: per-
on the 2005 GUTS questionnaire, thus this “males,” or “female(s) and male(s).” The above ceived benefits, perceived barriers, and norma-
analysis was restricted to female participants questions were combined to form the following tive beliefs. Perceived benefits were modeled
responding in that questionnaire year (n = 6494). 5 sexual orientation groups: “completely using a mean score from a multipart question:
We included participants who reported their heterosexual without any same-sex contact,”
If I were to get a Pap test. . . 1. I would be less
sexual orientation and who met the American “completely heterosexual with same-sex
likely to get cervical cancer; 2. My doctor or
College of Obstetricians and Gynecologists contact,” “mostly heterosexual,” “bisexual,” nurse might find something wrong that I can’t see
(ACOG) Pap test guidelines in 2005 of at least and “lesbian.” We applied an indicator variable or don’t know about; 3. It would give me peace
of mind to know what the result is; 4. It would
3 years since first sexual intercourse or being to missing data on sex of sexual contacts (n = 20).
help me take control of my health.
21 years old or older (n = 4224). Analyses did Sexual history. All items on sexual history
not include participants who were unsure of were collected in the 2005 questionnaire Perceived barriers were modeled using
(n = 6) or missing (n = 14) their sexual orien- wave. Three variables assessed initiation of a mean score from that same multipart ques-
tation or those missing geographic information sexual intercourse as well as the age of first tion: “If I were to get a Pap test. . . 1. It would be
(n = 7). Multiple items constituted each HBM sexual intercourse and number of sexual part- painful; 2. It would be embarrassing.” Norma-
construct so a mean for each construct was ners: “Have you ever had sexual intercourse? tive beliefs were modeled using a mean score
computed. Therefore, we excluded any partic- (By sexual intercourse we mean vaginal or anal from a multipart question that read: “Do these
ipant who did not respond to at least 1 item in sex).” Responses included “yes,” “no,” and “not people think you should get a Pap test? 1.
each of the 7 HBM constructs (n = 402). sure,” and if a participant indicated “yes” or Your friends; 2. Your mother; 3. Your health
Thus, the final analytic sample was 3821. “not sure” they were prompted to answer: care provider.”
We conducted sensitivity analyses restricted “During your life, with how many people have The third domain, cues to action, was
to participants who were eligible for a Pap test you had sexual intercourse?” and “How old assessed with the following question: “How
by the current ACOG guidelines set in 2009 were you when you had sexual intercourse for often has your mother talked to you about
(n = 2471). These updated guidelines recom- the first time?” When the age at first sexual protecting yourself from cervical cancer?”
mend screening begin at age 21 years, regard- intercourse (n = 29) or number of partners Finally, the last domain, perception of others’
less of sexual history,19 rather than at age (n = 38) were missing among individuals who Pap test utilization, was assessed using the
21 years or younger if sexual intercourse has ever had sexual intercourse, the respective mean of 2 questions: “How often do you think
started. Guidelines were changed so that means within each sexual orientation category your friends get a Pap test?” and “How often do
adolescent girls would not be unnecessarily was used. you think your mother gets a Pap test?”

320 | Research and Practice | Peer Reviewed | Charlton et al. American Journal of Public Health | February 2014, Vol 104, No. 2
 RESEARCH AND PRACTICE

Pap test intention and utilization. To assess statistical methods in regression analyses to completely heterosexuals with same-sex con-
an individual’s Pap test intention, the 2005 account for correlations resulting from sibling tact (70%).
questionnaire included an item: “How likely is groups in the cohort.23 We used linear re- Table 2 displays the HBM constructs and
it that you will get a Pap test in the next year?” gression models to determine parameter esti- Pap test intention variables by sexual orienta-
followed by a 5-level Likert scale response mates for the intention outcome and estimated tion. Unlike completely heterosexuals with no
ranging from extremely unlikely to extremely odds ratios (OR) for the utilization outcome same-sex contact, completely heterosexuals
likely, which was modeled linearly. To assess using completely heterosexuals without any with same-sex contact perceived more benefits
Pap test utilization in the following year, the same-sex contact as the reference group. We to Pap tests, normative views, and other in-
2007 questionnaire included the question: also evaluated the potential mediating effect of dividuals utilizing Pap tests and reported being
“Have you had a Pap test in the past year?” hormonal contraceptive use, sexual history, more likely to get a Pap test in the next year.
followed by “yes,” “no,” and “not sure.” and HBM constructs on the association be- Mostly heterosexuals perceived a higher risk of
Covariates. Additional covariates included tween sexual orientation and Pap test intention cervical cancer, although lower severity, and fewer
age, race, and geographic region. Age and racial and utilization using the publicly available barriers to Pap tests while they also reported being
information were collected on the 1996 base- %mediate macro.24 more likely to get a Pap test in the next year.
line questionnaire. An indicator variable for Bisexuals also perceived a higher risk for cervical
missing data was created for race (n = 17). RESULTS cancer. Finally, lesbians, in contrast to completely
Geographic region was accessed each year so heterosexuals, perceived a lower severity for
this analysis used data from 2005. Among the 3821 female participants in- cervical cancer and fewer other individuals utiliz-
cluded in our analyses, 81% (n = 3,090) de- ing Pap tests and this group reported being less
Statistical Analysis scribed themselves as completely heterosexual likely to get a Pap test in the next year.
We conducted descriptive statistics and with no same-sex contact, 2% (n = 71) as To examine associations between sexual
multivariate regression analyses with SAS sta- completely heterosexual with same-sex part- orientation and Pap test intentions as well as
tistical software version 9.2.21 Analyses using ners, 13% (n = 501) as mostly heterosexual, utilization before and after adjusting for hor-
2005 outcome data were cross-sectional, and 3% (n = 107) as bisexual, and 1% (n = 52) as monal contraceptive use, sexual history, and
analyses using 2007 outcome data were lon- mostly homosexual or completely homosexual HBM constructs, we restricted the remaining
gitudinal. We began by conducting a factor (lesbian). Table 1 displays the study population analyses to the subsample who reported all of
analysis on the HBM constructs to ensure our characteristics by sexual orientation. The racial these items (n = 3242). Table 3 shows the
scale supported the model.22 Although the composition of the cohort was largely White adjusted parameter estimates and ORs from 4
HBM is well explored, we wanted to ensure the (93%), and participants were 18 to 25 years of models. Model 1 adjusts for sociodemographic
questions we used were valid and that we age (mean = 21.5 years) when they returned variables, model 2 further adjusts for sexual
placed each question into the proper construct. the 2005 questionnaire. In contrast to com- history variables, model 3a further adjusts
In particular, this factor analysis was necessary pletely heterosexuals with no same-sex contact, model 2 for HBM constructs, model 3b further
because of the new questions about normative individuals who identified as completely het- adjusts model 2 for hormonal contraception
beliefs. Using a minimum Eigen-value of 1, erosexual with same-sex contacts, mostly het- use, and model 4 further adjusts model 2 for
the factor analysis identified 2 constructs: at- erosexual, bisexual, or lesbian were more likely HBM constructs and hormonal contraception
titudes and beliefs as well as the perception of to report that they had ever had sexual in- use. Although not all of the findings were
others’ Pap test utilization. We therefore com- tercourse (this could include opposite- and statistically significant, completely heterosex-
bined the necessary variables to model these same-sex intercourse). Mostly heterosexuals, uals with same-sex contact, bisexuals, and
2 constructs. Because of the limited number bisexuals, and lesbians reported a younger age lesbians were less likely to intend to get
of remaining variables, the other constructs of first sexual intercourse and more sexual a Pap test in the next year, and all sexual
(risk perception and cues to action) were partners. Completely heterosexuals with same- minority groups, except completely heterosex-
grouped from variables that were considered sex contact reported an older age of first sexual uals with same-sex contact, had lower odds
theoretically related. For example, the per- intercourse but also had more sexual partners. of receiving a Pap test in the following year
ceived risk construct included only 1 item on A high percentage of participants, regardless of compared to their completely heterosexual
susceptibility and 1 item on severity so we sexual orientation, reported having male sexual peers. After adjusting for sociodemographics,
grouped these 2 variables into a single risk- contact: 63% of lesbians, 95% of bisexuals, sexual history, HBM constructs, and hormonal
perception construct. 96% of mostly heterosexuals, 68% of com- contraceptives, lesbians (b = –0.46, P < .001)
We investigated sexual orientation group pletely heterosexuals with same-sex partners, were less likely to intend to get a Pap test
patterns in the HBM constructs. We then and 90% of completely heterosexuals. Finally, in the next year compared to completely
examined multiple covariates that may con- lesbians were the least likely to report using heterosexuals.
found or explain associations between sexual hormonal contraception (25%), followed by Compared with model 2, where we adjusted
orientation and Pap test utilization. We used bisexuals (45%), mostly heterosexuals (58%), for sociodemographics and sexual history, the
generalized estimating equation (GEE) completely heterosexuals (60%), and HBM constructs among lesbians in model 3a

February 2014, Vol 104, No. 2 | American Journal of Public Health Charlton et al. | Peer Reviewed | Research and Practice | 321
 RESEARCH AND PRACTICE

explained 19.1% of the disparities in Pap test

< .001

< .001
intention. Similarly, compared with model 2,

Note. P values were estimated by linear, logistic, and multinomial generalized estimating equation models with completely heterosexual as the referent group. The sample size was n = 3821 and was restricted to those aged 18–20 years
.79

.08
.11

.16
.15
P
Lesbian (n = 52)
hormonal contraceptive use among lesbians
in model 3b explained 8.6% of the disparities in

Mean 6SD

21.6 61.3

16.8 62.1
or % (No.)

3.9 62.0
92 (48)
88 (46)
63 (33)

25 (13)
Pap test intention and 36.1% of the disparities
in Pap test utilization. Together, less positive
health beliefs and less hormonal contraceptive
TABLE 1—Demographic and Sexual Health Characteristics by Sexual Orientation in a Cohort of US Adolescent and Young Adult Females Aged 18–25 Years

16.3 62.1 < .001

< .001
.003
Bisexual (n = 107)

.04
.13
.03
.06
P

use explained 29.3% of the disparities in Pap test

intention and 42.2% of the disparities in Pap test
Mean 6SD

21.3 61.5
or % (No.)

4.5 61.8
95 (102)
91 (97)
89 (94)

45 (48)
utilization. No other mediation effects were
significant. The subanalysis among participants
eligible for Pap tests under the new guidelines
< .001
< .001

< .001
< .001
returned similar results.
Mostly Heterosexual

.04
.02

.54
P
(n = 501)

DISCUSSION
Mean 6SD

21.4 61.4

16.6 61.9
or % (No.)

4.5 61.8
93 (466)
91 (454)
96 (483)

58 (291)

In this study, we examined whether less

positive health beliefs about Pap tests and
Completely Heterosexual With

cervical cancer as well as less hormonal con-

< .001

< .001
< .001
Same-Sex Contact (n = 71)

.03
.54
.27

.07
P

traceptive use partially explained sexual orien-

tation group disparities in Pap test intention
and utilization. We found that less positive
Mean 6SD

21.6 61.3

16.4 61.9
or % (No.)

4.6 61.7
93 (66)
97 (69)
68 (48)

70 (50)

health beliefs associated with Pap testing ex-

plained part of the disparities in Pap test
intention among lesbians relative to completely
heterosexuals. Additionally, less hormonal
Completely Heterosexual (n = 3090),

contraceptive use explained some of the dis-

Mean 6SD or % (No.)

parities in Pap test intention among lesbians as

well as a large proportion of the disparities in
21.5 61.3

17.3 62.0
83 (2561)
94 (2890)
90 (2767)

60 (1840)

3.5 61.9

receiving a Pap test among lesbians.

In the mediation analysis, less positive health
beliefs associated with Pap testing explained
19.1% of the disparities in Pap test intention
among lesbians relative to completely hetero-
Assessed in 2007 and included oral contraceptives, Depo-Provera, and Implanon.
Mean 6SD or % (No.)

sexuals. On average, lesbians do not believe

All (n = 3821),

they are as susceptible to cervical cancer as do

21.5 61.3

17.1 62.0
85 (3238)
93 (3553)
90 (3433)

55 (3517)

3.7 62.0

completely heterosexuals, and these findings

support the limited existing literature.6,9,25
Ben-Natan et al. used the HBM model to
with ‡ 3 years since coitarche or those aged ‡ 21 years.

examine Pap test use in 108 lesbians, aged 18

Age at first sexual intercourse, y (range £ 13– ‡ 21)

to 41 years, and found that the HBM constructs

contributed 22% to predicting lesbian’s
Subsample who had initiated sexual intercourse

Sexual intercourse partners (range = 1–‡ 6)

Pap test intention.3 However, the HBM model

(defined as vaginal or anal sex; n = 3242)
Sexual intercourse (same- or opposite-sex)

is limited in that its premise is economical-

Hormonal contraception use in last ya

rational.3 The model captures an individual’s

decision based only on their cost-benefit per-
ceptions and does not account for the effect
Male sexual contact ever
Age, y (range = 18–25)

of other factors such as health insurance,

proximity to a clinic, etc. Nonetheless, previous
research demonstrates that components of the
HBM, including perceived benefits, are some of
White

the strongest predictors of Pap test utilization.26

a

Intention reflects beliefs more than behavior,

322 | Research and Practice | Peer Reviewed | Charlton et al. American Journal of Public Health | February 2014, Vol 104, No. 2
 RESEARCH AND PRACTICE

TABLE 2—Health Belief Model Constructs and Papanicolaou (Pap) Test Intention Variables by Sexual Orientation in a Cohort of US Adolescent
and Young Adult Females Aged 18–25 Years

Completely Heterosexual Completely Heterosexual With Mostly Heterosexual

(n = 3 090) Same-Sex Contact (n = 71) (n = 501) Bisexual (n = 107) Lesbian (n = 52)
Constructs and Variables Mean (SE) Mean (SE) P Mean (SE) P Mean (SE) P Mean (SE) P

Risk perception
Perceived susceptibilitya 2.7 (0.02) 2.9 (0.11) .06 2.8 (0.04) .003 3.0 (0.09) < .001 2.8 (0.14) .51
Perceived severityb 3.3 (0.01) 3.3 (0.09) .46 3.2 (0.03) .009 3.3 (0.07) .47 3.1 (0.10) .03
Attitudes/beliefs
Perceived benefitsc 4.2 (0.01) 4.4 (0.07) .004 4.1 (0.03) .44 4.1 (0.07) .27 4.0 (0.11) .1
Perceived barriers d 2.5 (0.02) 2.3 (0.14) .26 2.3 (0.05) .002 2.4 (0.10) .72 2.7 (0.17) .22
Normative beliefse 4.9 (0.02) 5.3 (0.14) .02 4.9 (0.06) .99 4.9 (0.13) .63 4.6 (0.19) .12
Cues to actionf 2.4 (0.02) 2.5 (0.12) .44 2.5 (0.04) .22 2.5 (0.09) .36 2.4 (0.12) .61
Perceptions of others’ utilizationg 3.3 (0.02) 3.5 (0.08) .008 3.2 (0.05) .21 3.1 (0.10) .16 2.9 (0.14) .02
Intention to get Pap test in the next yearh 4.6 (0.02) 4.7 (0.08) .05 4.7 (0.04) .01 4.5 (0.10) .26 4.1 (0.14) .001

Note. SE and P values were estimated by linear generalized estimating equation models with completely heterosexual as the referent group. The sample size was n = 3821 and was restricted to
those aged 18–20 years with ‡ 3 years since coitarche or those aged ‡ 21 years.
a
Five-level Likert scale for 1 question (range: 1 = extremely unlikely to 5 = extremely likely).
b
Four-level Likert scale for 1 question (range: 1 = not serious to 4 = extremely serious).
c
Five-level Likert scales for 4 questions (range: 1 = not beneficial to 5 = very beneficial).
d
Five-level Likert scales for 2 questions (range: 1 = not a barrier to 5 = very much a barrier).
e
Five-level Likert scales for 3 questions (range: 1 = others believe should not have Pap test to 5 = others believe should have Pap test).
f
Four-level Likert scale for 1 question (range: 1 = less cues to 4 = more cues).
g
Four-level Likert scale for 2 questions (range: 1 = infrequent practice to 4 = frequent practice).
h
Five-level Likert scale (range: 1 = extremely unlikely to 5 = extremely likely), reported in 2005/2006.

which may be why the HBM better captures The present study is among the first to our young adults, but we were limited by the
the Pap test intention than the utilization. knowledge to apply the HBM to Pap test use cohort’s homogenous racial composition.
We found that a lower prevalence of hor- among various sexual orientation subgroups. Because of the self-report nature of the data,
monal contraceptive use explained 8.6% of the Our analysis contributed in new ways and had any kind of gynecological exam might be
disparities among lesbians relative to com- a number of strengths such as the large, misreported as a Pap test, which could lead to
pletely heterosexuals in Pap test intention and national sample. Previous studies were limited overreporting.27 Any such misclassification is
36.1% of the disparities among lesbians in in power and therefore combined sexual mi- likely to be nondifferential (e.g., participants are
receiving a Pap test. Many physicians perform norities into a single category.12 We were able as likely to misreport a gynecological exam
a Pap test before providing contraceptive pre- to more precisely observe different sexual as a Pap test, regardless of their sexual orien-
scriptions, which may explain why this medi- orientation groups by distinguishing among tation) and therefore would bias the results
ating effect was relatively large. For women completely heterosexuals with no same-sex toward the null.28 There may have been mis-
with male sexual partners, standard health care contact, completely heterosexuals with same-sex classification of ever having had sexual inter-
practice maximizes the chances that they will contact, mostly heterosexuals, bisexuals, and course as the sex item ambiguously defined
get Pap tests, but this does not hold true for lesbians in addition to controlling for sexual “sexual intercourse” as “vaginal or anal.”
women with only female partners. Further- history. Clinicians should be aware that there is In conclusion, regular Pap tests are needed
more, physicians often are not aware of les- substantial heterogeneity across sexual minority by all women who meet current guidelines,
bians’ reproductive health needs.4 Previous subgroups and that this heterogeneity may regardless of sexual orientation. Physicians
studies have shown that lesbians use less pre- affect Pap test utilization in important ways. should be aware of disparities by sexual ori-
ventive medical services5 and are more likely Our study had several limitations. All of the entation, in part attributed to less hormonal
to delay seeking care for various health needs, participants are children of nurses and are not contraceptive use and less positive health
including reproductive health, compared with a representative sample, which could limit beliefs among sexual minorities, and should
heterosexual women.4 Lesbians report having generalizability. Daughters of nurses may have take care to communicate the importance of
gaps in knowledge about cervical cancer more health care access than the general Pap tests with sexual minority patients. Health
(e.g., risk factors)3 and the importance of population. The study population was younger behavior can be improved by changing
preventive screening,5 and they utilize fewer than those of other studies addressing similar HBM factors such as perceived benefits of
Pap tests than do heterosexuals.2,6---8 issues, so we could examine adolescents and Pap tests,29 so clinical encounters of different

February 2014, Vol 104, No. 2 | American Journal of Public Health Charlton et al. | Peer Reviewed | Research and Practice | 323
 RESEARCH AND PRACTICE

TABLE 3—Multivariable Linear and Logistic Regression Results for Papanicolaou (Pap) Test Variables in a Cohort of US Adolescent and Young
Adult Females Aged 18–25 Years

Pap Test Intention in the Next Year (Reported: 2005/2006) Pap Test Utilization in the Last Year (Reported: 2007/2008)
Proportion of Effect Mediated Proportion of Effect Mediated
b (SE) Pa by Additional Variables, % (P) OR (95% CI) Pa by Additional Variables, % (P)

Model 1b
Completely heterosexual (Ref) 1.00
Completely heterosexual with same-sex contact –0.01 (0.09) .89 1.34 (0.53, 3.41) .54
Mostly heterosexual –0.01 (0.03) .8 0.71 (0.52, 0.98) .03
Bisexual –0.20 (0.09) .03 0.56 (0.31, 0.99) .05
Lesbian –0.68 (0.15) < .001 0.24 (0.12, 0.47) < .001
Model 2c
Completely heterosexual (Ref) 1.00
Completely heterosexual with same-sex contact –0.03 (0.10) .79 NAd 1.89 (0.68, 5.26) .22 NAd
d
Mostly heterosexual –0.05 (0.03) .16 NA 0.68 (0.49, 0.95) .02 NAd
Bisexual –0.24 (0.09) .009 NAd 0.59 (0.33, 1.07) .08 NAd
Lesbian –0.66 (0.14) < .001 NSd 0.37 (0.17, 0.81) .01 NSd
e
Model 3a
Completely heterosexual (Ref) 1.00
Completely heterosexual with same-sex contact –0.08 (0.10) .43 NAf 1.67 (0.62, 4.49) .31 NSf
f
Mostly heterosexual –0.02 (0.03) .56 NS 0.72 (0.52, 1.01) .06 NSf
f
Bisexual –0.18 (0.09) .05 NS 0.64 (0.36, 1.15) .13 NSf
f
Lesbian –0.49 (0.13) < .001 19.1 (.004) 0.42 (0.19, 0.94) .03 NSf
Model 3bg
Completely heterosexual (Ref) 1.00
Completely heterosexual with same-sex contact –0.04 (0.10) .63 NAf 1.49 (0.60, 3.72) .4 NSf
f
Mostly heterosexual –0.04 (0.03) .19 NS 0.72 (0.51, 1.02) .07 NSf
f
Bisexual –0.22 (0.09) .02 NS 0.74 (0.41, 1.32) .31 NSf
f
Lesbian –0.60 (0.14) < .001 8.6 (.001) 0.67 (0.29, 1.55) .35 36.1 (.02)f
h
Model 4
Completely heterosexual (Ref) 1.00
Completely heterosexual with same-sex contact –0.09 (0.10) .37 NAf 1.36 (0.56, 3.30) .5 NSf
f
Mostly heterosexual –0.02 (0.03) .6 NS 0.75 (0.53, 1.07) .11 NSf
f
Bisexual –0.16 (0.09) .07 NS 0.81 (0.46, 1.41) .45 NSf
f
Lesbian –0.46 (0.13) < .001 29.3 (< .001) 0.76 (0.31, 1.85) .55 42.2 (.007)f

Note. CI = confidence interval; NA = No mediation; NS = No significant mediation; OR = odds ratio. The sample size was n = 3242 and was restricted to subsample of those aged 18–20 years
with ‡ 3 years since coitarche or aged ‡ 21 years who reported sex of sexual contacts, age at coitarche, and number of sexual intercourse partners.
a
P values estimated by linear and logistic generalized estimating equation regression with completely heterosexual as the referent group.
b
Adjusted for sociodemographics (age, race, geographic region of residence).
c
Adjusted for sociodemographics and sexual history (any male sexual contacts, age of coitarche, and number of sexual intercourse partners), as well as intention to get Pap test in the next year.
d
Compared to model 1.
e
Adjusted for sociodemographics, sexual history, and Health Belief Model constructs (risk perception [perceived susceptibility, perceived severity], attitudes/beliefs [perceived benefits, perceived
barriers, normative beliefs], cues to action, and perceptions others’ utilization), as well as intention to get Pap test in the next year.
f
Compared to model 2.
g
Adjusted for sociodemographics, sexual history, and hormonal contraception use, as well as intention to get Pap test in the next year.
h
Adjusted for sociodemographics, sexual history, Health Belief Model constructs, and hormonal contraception use, as well as intention to get Pap test in the next year.

types, not only visits to receive hormonal practitioners and physicians can work to re- targeted education programs. Increasing Pap
contraception, are an important opportunity duce emotional barriers and increase per- test use will not only improve reproductive
for health care providers to intervene and ceived benefits of Pap tests for everyone, health and overall wellness, but also reduce
improve Pap test utilization among sexual especially lesbians. Interventions might include preventable cervical cancers and death in
minorities. For example, public health physician trainings, media campaigns, or sexual minorities. j

324 | Research and Practice | Peer Reviewed | Charlton et al. American Journal of Public Health | February 2014, Vol 104, No. 2
 RESEARCH AND PRACTICE

About the Authors 7. Marrazzo JM, Stine K. Reproductive health history of cancer screening programme in urban Sweden.
Brittany M. Charlton is with the Department of Epidemi- lesbians: implications for care. Am J Obstet Gynecol. Psychooncology. 2001;10(1):76---87.
ology, Harvard School of Public Health, Boston, MA. 2004;190(5):1298---1304. 27. Biro FM, Rosenthal SL, Rymarquis L, Kollar LM, Hillard
Heather L. Corliss and S. Bryn Austin are with the Division 8. Charlton BM, Corliss HL, Missmer SA, et al. Re- PJ. Adolescent girls’ understanding of Papanicolaou smear
of Adolescent and Young Adult Medicine, Boston Children’s productive health screening disparities and sexual ori- results. J Pediatr Adolesc Gynecol. 1997;10(4):209-- 212.
Hospital, Harvard Medical School, Boston, MA. Stacey A. entation in a cohort study of US adolescent and young 28. Gullen WH, Bearman JE, Johnson EA. Effects of
Missmer, is with Department of Obstetrics, Gynecology, and adult females. J Adolesc Health. 2011;49(5):505---510. misclassification in epidemiologic studies. Public Health
Reproductive Biology, Brigham and Women’s Hospital and
9. Price JH, Easton AN, Telljohann SK, Wallace PB. Rep. 1968;83(11):914---918.
Harvard Medical School, Boston, MA. A. Lindsay Frazier
Perceptions of cervical cancer and Pap smear screening 29. Park S, Chang S, Chung C. Effects of a cognition-
is with the Dana-Farber Cancer Institute, Boston, MA.
behavior by women’s sexual orientation. J Community emotion focused program to increase public participation
Margaret Rosario is with the City University of New York,
Health. 1996;21(2):89---105. in Papanicolaou smear screening. Public Health Nurs.
City College and Graduate Center, New York, NY. Jessica A.
Kahn is with the Division of Adolescent Medicine, Cincin- 10. Power J, McNair R, Carr S. Absent sexual scripts: 2005;22(4):289---298.
nati Children’s Hospital Medical Center, Cincinnati, OH. lesbian and bisexual women’s knowledge, attitudes and
Correspondence should be sent to Brittany M. Charlton, action regarding safer sex and sexual health information.
Department of Epidemiology, Harvard School of Public Cult Health Sex. 2009;11(1):67---81.
Health, 677 Huntington Avenue, 9th Floor, Boston, MA 11. Lubetkin EI, Santana A, Tso A, Jia H. Predictors of
02115 (e-mail: [email protected]). Reprints cancer screening among low-income primary care patients.
can be ordered at http://www.ajph.org by clicking the J Health Care Poor Underserved. 2008;19(1):135---148.
“Reprints” link.
12. Brown JP, Tracy JK. Lesbians and cancer: an over-
This article was accepted October 15, 2012.
looked health disparity. Cancer Causes Control. 2008;19
(10):1009---1020.
Contributors
13. Ramirez JE, Ramos DM, Clayton L, Kanowitz S, Moscicki
B. M. Charlton, H. L. Corliss, M. Rosario, and S. B. Austin
AB. Genital human papillomavirus infections: knowledge,
conceptualized and planned the analysis while H. L.
perception of risk, and actual risk in a nonclinic population of
Corliss, S. A. Missmer, A. L. Frazier, J. A. Kahn, and S. B.
young women. J Womens Health. 1997;6(1):113-- 121.
Austin oversaw data collection. B. M. Charlton analyzed
the data and wrote the article, and all authors critically 14. Guvenc G, Akyuz A, Acikel CH. Health belief model
reviewed the article and approved the final version. scale for cervical cancer and Pap smear test: psychomet-
ric testing. J Adv Nurs. 2011;67(2):428---437.
Acknowledgments 15. Rosenstock IM. Why people use health services.
This study was supported by the National Institutes of Milbank Mem Fund Q. 1966;44(3 suppl):94---127.
Health (research grant R01HD057368). H. L. Corliss 16. Becker MH, Maiman LA. Sociobehavioral determi-
and S. B. Austin are supported by the Leadership Edu- nants of compliance with health and medical care
cation in Adolescent Health Project (grant T71 MC recommendations. Med Care. 1975;13(1):10---24.
00009) from the Maternal and Child Health Bureau.
H. L. Corliss is also supported by the National Institute 17. Solarz AL; Institute of Medicine. Lesbian Health:
on Drug Abuse (grant K01 DA023610). B. M. Charlton Current Assessment and Directions for the Future. Wash-
is supported by the Training Program in Cancer ington, DC: National Academies Press; 1999.
Epidemiology (grant T32 CA 09001). 18. Field AE, Camargo CA Jr, Taylor CB, et al. Overweight,
This work was initially presented at the 33rd Annual weight concerns, and bulimic behaviors among girls and boys.
Meeting & Scientific Sessions of the Society of Behavioral J Am Acad Child Adolesc Psychiatry. 1999;38(6):754-- 760.
Medicine; April 11-14, 2012; New Orleans, LA.
19. Sawaya GF. Cervical-cancer screening---new guide-
lines and the balance between benefits and harms.
Human Participant Protection N Engl J Med. 2009;361(26):2503---2505.
This study was approved by the Brigham and Women’s
20. Remafedi G, Resnick M, Blum R, Harris L. De-
Hospital institutional review board.
mography of sexual orientation in adolescents. Pediatrics.
1992;89(4 Pt 2):714---721.
References 21. SAS Statistical Software, Release 9.2. In. Cary, NC:
1. Igra V, Millstein SG. Current status and approaches SAS Institue Inc.
to improving preventive services for adolescents. JAMA.
22. Kahn JA, Emans SJ, Goodman E. Measurement of
1993;269(11):1408---1412.
young women’s attitudes about communication with pro-
2. Bailey JV, Kavanagh J, Owen C, McLean KA, Skinner viders regarding Papanicolaou smears. J Adolesc Health.
CJ. Lesbians and cervical screening. Br J Gen Pract. 2000; 2001;29(5):344---351.
50(455):481---482.
23. Zeger SL, Liang KY. Longitudinal data analysis for
3. Ben-Natan M, Adir O. Screening for cervical cancer discrete and continuous outcomes. Biometrics. 1986;42
among Israeli lesbian women. Int Nurs Rev. 2009;56(4): (1):121---130.
433---441.
24. Hertzmark E, Pazaris M, Spiegelman D. The SAS
4. Carroll NM. Optimal gynecologic and obstetric care MEDIATE Macro. 2009. Available at: http://www.hsph.
for lesbians. Obstet Gynecol. 1999;93(4):611---613. harvard.edu/spiegelman/mediate/mediate.pdf. Accessed
5. Diamant AL, Schuster MA, Lever J. Receipt of pre- June 3, 2013.
ventive health care services by lesbians. Am J Prev Med. 25. O’Hanlan KA, Crum CP. Human papillomavirus-asso-
2000;19(3):141---148. ciated cervical intraepithelial neoplasia following lesbian
6. Marrazzo JM. Genital human papillomavirus in- sex. Obstet Gynecol. 1996;88(4 Pt 2):702---703.
fection in women who have sex with women: a concern 26. Forss A, Tishelman C, Widmark C, Lundgren E,
for patients and providers. AIDS Patient Care STDS. Sachs L, Tornberg S. ‘I got a letter. . .’ a qualitative study
2000;14(8):447---451. of women’s reasoning about attendance in a cervical

February 2014, Vol 104, No. 2 | American Journal of Public Health Charlton et al. | Peer Reviewed | Research and Practice | 325
Copyright of American Journal of Public Health is the property of American Public Health
Association and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.