Acceptance of the HPV vaccine for adolescent girls: analysis of state-added questions from the BRFSS.

Abstract

Previous research regarding human papillomavirus (HPV) awareness and vaccine acceptance has relied on convenience or other selected samples of the population. To assess the prevalence of HPV awareness and vaccine acceptance in Kentucky we added questions to the 2006 Kentucky Behavioral Risk Factor Survey System (BRFSS), a population-based survey of health behaviors.
Women who participated in the statewide BRFSS were asked two HPV-related questions: one assessed previous awareness of HPV, and another assessed vaccine acceptance for girls 10 to 15 years old. We used crosstabulations and multivariate logistic regression to determine which factors were associated with HPV awareness and vaccine acceptance. Because the HPV vaccine Gardasil was approved in June 2006, we conducted an analysis of pre- and postapproval HPV awareness and vaccine acceptance. We also compared results across Appalachian and non-Appalachian counties, two distinct regions of Kentucky.
Overall, 57.6% of women had heard of HPV, and 70.2% accepted vaccination for girls. HPV awareness increased after Gardasil's approval, but the increase was much smaller among Appalachian women. Prevalence of vaccine acceptance was unchanged in both regions. Awareness of HPV was not associated with vaccine acceptance, and factors significantly associated with vaccine acceptance in multivariate analysis differed by Appalachian status.
This population-based survey of Kentucky women found relatively high vaccine acceptance for girls. Also, many respondents reported not knowing whether they accept vaccination, and factors associated with vaccine acceptance varied by Appalachian status. These findings suggest that acceptance of the HPV vaccine for girls may improve with targeted interventions.

Full text

Page 1

Original article
Acceptance of the HPV Vaccine for Adolescent Girls: Analysis
of State-Added Questions from the BRFSS
W. Jay Christian, M.P.H.a, Amy Christian, M.S.P.H.a, and Claudia Hopenhayn, Ph.D.a,b,*
aMarkey Cancer Control Program, University of Kentucky, Lexington, Kentucky
bDepartment of Epidemiology, College of Public Health, University of Kentucky, Lexington, Kentucky
Manuscript received April 11, 2008; manuscript accepted September 2, 2008
AbstractPurpose: Previous research regarding human papillomavirus (HPV) awareness and vaccine accep-
tance has relied on convenience or other selected samples of the population. To assess the prevalence
of HPV awareness and vaccine acceptance in Kentucky we added questions to the 2006 Kentucky Be-
havioral Risk Factor Survey System (BRFSS), a population-based survey of health behaviors.
Methods: Women who participated in the statewide BRFSS were asked two HPV-related questions:
one assessed previous awareness of HPV, and another assessed vaccine acceptance for girls 10 to 15
years old. We used crosstabulations and multivariate logistic regression to determine which factors
were associated with HPV awareness and vaccine acceptance. Because the HPV vaccine Gardasil
was approved in June 2006, we conducted an analysis of pre- and postapproval HPV awareness
and vaccine acceptance. We also compared results across Appalachian and non-Appalachian counties,
two distinct regions of Kentucky.
Results: Overall,57.6%ofwomenhadheardofHPV,and70.2%acceptedvaccinationforgirls.HPV
awareness increased after Gardasil’s approval, but the increase was much smaller among Appalachian
women. Prevalenceof vaccine acceptance wasunchanged inboth regions. Awareness of HPV wasnot
associated with vaccine acceptance, and factors significantly associated with vaccine acceptance in
multivariate analysis differed by Appalachian status.
Conclusions: This population-based survey of Kentucky women found relatively high vaccine ac-
ceptance for girls. Also, many respondents reported not knowing whether they accept vaccination,
and factors associated with vaccine acceptance varied by Appalachian status. These findings suggest
that acceptance of the HPV vaccine for girls may improve with targeted interventions. ? 2009 Society
for Adolescent Medicine. All rights reserved.
Keywords:
Human papillomavirus; HPV; Vaccine; Behavioral Risk Factor Surveillance System; BRFSS; Appalachia; Kentucky
Particular high-risk strains of the human papillomavirus
(HPV) have been identified as a necessary factor in thedevel-
opment of cervical cancer [1]. In June 2006, the U.S. Food
and Drug Administration (FDA) approved a vaccine, Garda-
sil, which protects against high-risk HPV strains 16 and 18,
which are responsible forabout70% of cervical cancer cases,
and low-risk HPV strains 6 and 11, which are linked to gen-
ital warts [2]. The vaccine is recommended for girls 11 to 12
years old, but it is approved for young women 9 to 26 years
old [2]. To be most effective, the vaccine needs to be admin-
istered prior to initiation of sexual activity (and potential ex-
posure to HPV); thus, its recommendation for preteen girls
[2]. Although some parents might find it unnecessary, or
even unacceptable, to administer the HPV vaccine to girls
at such a young age, it is important to recognize that nation-
wide about 33% of ninth-grade students report having had
sexual intercourse, increasing to about 62% of students by
the time they are in 12th grade [3].
Although cervical cancer rates have decreased substan-
tially in the last several decades in the United States [4], rates
have been consistently higher among subgroups of the
FinancialsupportwasprovidedbytheNationalCancerInstitute’sCancer
Information Service—Mid South Region
*Address correspondence to: Claudia Hopenhayn, Ph.D., Markey Can-
cerControlProgram,UniversityofKentucky,2365HarrodsburgRoad,Suite
B100, Lexington, KY 40504-3381.
E-mail address: cmhope0@uky.edu
1054-139X/09/$ – see front matter ? 2009 Society for Adolescent Medicine. All rights reserved.
doi:10.1016/j.jadohealth.2008.09.001
Journal of Adolescent Health 44 (2009) 437–445

Page 2

population, including Appalachian women [5–8], and partic-
ularly in Appalachian Kentucky [6,9]. In addition, cervical
cancer rates tend to be higher in rural areas [10–12], again
putting women in Appalachian Kentucky at increased risk,
as 47 of the 51 Appalachian counties are considered rural
[13,14]. Thus, the acceptance and widespread use of the
HPV vaccine in Kentucky will be an important component
in reducing the number of women diagnosed with cervical
cancer.
Recent research in the United States regarding the accep-
tance of an HPV vaccine has been based mostly on samples
of convenience (e.g., medical clinics, classrooms, commu-
nity settings) [15–23] or focused on particular groups (e.g.,
minorities, lower income, physicians) [24–27], whereas an-
other study was a statewide survey of parents in California
[28]. A literature review, which included most of these stud-
ies, showed that,in general,acceptability foran HPV vaccine
was quite positive [29].
In the fall of 2005, we conducted a telephone survey of
adult women in two Appalachian Kentucky counties regard-
ing their knowledge of HPV and their attitudes toward a vac-
cine for themselves and for girls [30]. Some of the questions
on that survey related to HPV awareness and vaccine accep-
tanceweresubmitted totheKentucky BehavioralRiskFactor
Surveillance System (BRFSS), and were accepted as state-
added questions for 2006. To our knowledge, Kentucky
was the first state to use the BRFSS to survey a statewide
sample of women regarding their awareness of HPV and ac-
ceptance of an HPV vaccine. Here we report the findings
from this survey with respect to women’s awareness of
HPV and their acceptance of the HPV vaccine for girls.
Methods
The BRFSS is an annual telephone survey of noninstitu-
tionalized adults (aged 18 years or more) concerning health
conditionsandriskbehaviors[31].ItissupportedbytheCen-
ters for Disease Control and Prevention and conducted in all
50 U.S. states. In Kentucky, the BRFSS is administered
through the Kentucky Department for Public Health, Cabinet
for Health and Family Services. The BRFSS includes a set of
core questions required of all states; however, additional
questions may be added to the survey at each state’s discre-
tion. For the 2006 BRFSS, our request to include a few ques-
tions concerning HPV awareness and acceptance of an HPV
vaccine was approved. The analysis presented here focuses
on two of these questions:
1. Have you ever heard of HPV before? (the ‘‘awareness
question’’)
2. If an HPV vaccine was available, do you think girls
aged 10 to 15 should be vaccinated? (the ‘‘acceptance
question’’)
It is important to note that the questions were designed
prior to the approval of the vaccine, and thus formatted to as-
sess the acceptance of a hypothetical or potential vaccine.
Gardasil, an HPV vaccine manufactured by Merck, was ap-
proved for distribution about halfway though the survey pe-
riod, on June 8, 2006 [2].
Kentucky’s 2006 survey included 6174 respondents,
about67%(n¼4154)ofwhichwerewomen.TheHPVques-
tions were administered only to female respondents who in-
dicated they had not previously had a hysterectomy (n ¼
2221), and were preceded by a short introduction about
HPV and its link to cervical cancer. The awareness question
allowed the following responses: yes, no, don’t know. The
acceptance question allowed these options: definitely yes,
probably yes, probably no, definitely no, don’t know. For
this analysis, the first two responses to the latter question
wererecodedto‘‘yes’’andthefollowingtwoto‘‘no,’’topro-
duce a binary outcome variable for the assessment of respon-
dents’ acceptance of vaccination for girls. We included the
‘‘don’t know’’ responses in bivariate analysis, but eliminated
them from multivariate analysis.
We conducted bivariate analysis for the Appalachian and
non-Appalachian regions separately to (1) provide for a com-
parison between this study and Hopenhayn et al [30], which
was limited to two Appalachian counties, and (2) to discern
whether there were any differences between Appalachian
and non-Appalachian regions regarding factors associated
with acceptance of the vaccine. The analysis consisted of cal-
culating and comparing prevalence estimates of acceptance
of HPV vaccination for girls by key socioeconomic, demo-
graphic, and health behavior indicators also available in the
BRFSS.Theseincludedage(codedasacategoricalvariable),
marital status (categorical), income (categorical), educational
attainment (categorical), health insurance coverage (binary),
smoking status (binary), race (binary: white vs. black only),
recent Pap test (binary: last Pap test less than 3 years ago
vs.morethan3years agoornever),urbanversusrural county
residence (binary), presence of children in the home (binary),
and Appalachian county residence (binary). As noted, the
race categories include only white and black respondents
who comprised 92.4% and 3.7%, respectively, of those
who answered the HPV-related questions. The remaining
3.9%, made up of missing values and other race categories
rare in Kentucky, were excluded from the subsequent multi-
variate modeling. Urban and rural counties were so desig-
nated using rural–urban continuum codes, with codes 1–3
indicating urban counties and 4–9 indicating rural [14]. The
exact coding of other variables is largely equivalent to what
we had used in a previous analysis of similar questions
[30], and can be found in Table 1.
We conducted multivariate analysis using all the variables
from the bivariate analysis described above. This analysis
consisted of two survey-weighted logistic regression models,
one each for the Appalachian and non-Appalachian regions.
Finally, we also employed an ancillary bivariate analysis
to discern whether the FDA’s regulatory approval of the
HPV vaccine, as well as the associated news media coverage
and public debate regarding the vaccine, might have influ-
enced HPV awareness and/or acceptance of the vaccine for
W.J. Christian et al. / Journal of Adolescent Health 44 (2009) 437–445
438

Page 3

girls. Because the FDA approved the vaccine in June, 2006,
we compared rates of HPV awareness and vaccine accep-
tance between those who were surveyed in the first and last
6 months of the 2006 BRFSS survey. Using the original
BRFSS variable that records the month of each respondent’s
telephone interview, we created a new binary variable and
crosstabulated it with dichotomized results from both ques-
tions.
For this study weused Stata 10[32], which can accommo-
date the complex survey design of the BRFSS, for data man-
agement and statistical analysis. Sampling weights from the
BRFSS were used in all analyses, including summary statis-
tics in Table 1.
This study was approved by the University of Kentucky
Medical Institutional Review Board.
Results
There were 2169 respondents to both the awareness ques-
tion and the acceptance question. Their characteristics are
shown in Table 1, with those of all women in the 2006 Ken-
tucky BRFSS for comparison. There were very few missing
values for most variables (less than 1% in most cases), the
most notable exception being that 18.9% did not answer
the household income question. The vast majority of the
study population was white (88.9%), with blacks being the
next largest group (6.3%). The proportion of the population
from non-Appalachian Kentucky was just over two-thirds
(68.7%). Urban and rural populations were roughly equiva-
lent (51.1% and 48.9%, respectively). The BRFSS subgroup
of women in our analysis did not differ much from the entire
female population surveyed in the 2006 Kentucky BRFSS,
with the exception of age distribution. The larger proportion
of younger women in our subject population was likely be-
causeofourexclusionofwomenwhohadhadahysterectomy
(38.9% of BRFSS respondents in 2006, yielding a hysterec-
tomy prevalence estimate of 29.9% for women in Kentucky).
In turn, the difference in age distribution probably accounts
for the fact that we had a greater proportion of women who
had children at home (52.2% vs. 43.0%), and together with
the hysterectomy exclusion, also explains the difference in
Table 1
Characteristics of all women in the Kentucky Behavioral Risk Factor
Surveillance System (2006) and the women included in the study analysisa
Percent
Women in
this studyAll women
Age group
18–29
30–39
40–49
50–59
60–69
70þ
Missing
Marital status
Married
Divorced/separated
Widowed
Never married
Missing
Household income
<$15,000
$15,000–$25,000
$25,000–$50,000
$50,000þ
Missing
Education
Less than high school
High school graduate
Some college
College graduate
Missing
Health insurance
No
Yes
Missing
Smoking status
Nonsmoker
Smoker
Missing
Race
Black
White
Other
Missing
Pap test history
No recent Pap
Had a recent Pap (<3 years)
Missing
Appalachian status
Non-Appalachian
Appalachian
Urban–rural status
Urban
Rural
Children in household
Yes
No
Missing
Awareness of HPV
No
Yes
Don’t know
Missing
23.5
25.8
20.0
15.7
7.4
7.4
0.2
17.6
19.9
18.7
18.3
11.0
14.0
0.3
63.8
12.8
8.0
15.1
0.3
61.8
14.1
11.9
11.8
0.4
11.2
14.5
26.2
29.2
18.9
12.3
15.3
23.6
27.8
21.2
13.1
34.7
27.1
24.7
0.4
15.9
35.7
25.8
22.3
0.4
19.4
80.6
0.0
15.9
83.9
0.2
69.6
30.4
0.0
71.9
28.0
0.1
6.3
88.9
4.5
0.3
5.3
90.8
3.6
0.4
17.3
81.0
1.7
18.4
70.1
11.5
68.7
31.3
69.7
30.3
51.1
48.9
46.2
53.8
52.2
47.7
0.1
43.0
56.9
0.1
41.8
57.6
0.6
—
—
—
—
(Continued)
Table 1
Characteristics of all women in the Kentucky Behavioral Risk Factor
Surveillance System (2006) and the women included in the study analysisa
(Continued)
Percent
Women in
this study All women
Acceptance of HPV vaccine for girls
Yes
No
Don’t know
Missing
70.2
12.8
16.2
0.9
—
—
—
—
Note: percentages may not add to 100 due to rounding.
aFemales who have not had a hysterectomy and responded to the HPV
awareness question and HPV vaccine acceptance for girls question.
W.J. Christian et al. / Journal of Adolescent Health 44 (2009) 437–445
439

Page 4

the proportion who reported having had a recent Pap test
(81.0% vs. 71.1%). The overall rates of HPV awareness
and vaccine acceptance in our study population were
57.6% and 70.2%, respectively.
Table 2 displays the results of the bivariate analysis of
vaccine acceptance by Appalachian status of respondents.
Overall, 73.1% of Appalachian women and 69.7% of non-
Appalachian women signaled acceptance of vaccination for
girls. The percentages of women who were not accepting
ofthevaccine,aswellasthosewhodidn’tknow,weresimilar
in Appalachia and non-Appalachia. In both regions, a greater
percentage of women answered that they did not know
whether to accept vaccination than rejected it.
In Appalachian counties, women over 70 years of age
were much less likely to accept HPV vaccination for girls.
A similar trend was apparent in non-Appalachian Kentucky,
but was not statistically significant. Regarding marital status,
women in both regions who were divorced or never married
Table 2
Women’s acceptance of an HPV vaccine for girls, by selected factors (BRFSS 2006)
Appalachian PercentNon-Appalachian Percent
Yes NoDon’t know
p-valueYes NoDon’t know
p-value
Total
Age group
18–29
30–39
40–49
50–59
60–69
70þ
Marital status
Married
Divorced/separated
Widowed
Never married
Annual income
<$15,000
$15,000–$25,000
$25,000–$50,000
$50,000þ
Missing
Education
Less than high school
High school graduate
Some college
College graduate
Health insurance
No
Yes
Smoking status
Nonsmoker
Smoker
Urban/rural status
Rural
Urban
Awareness of HPV
No
Yes
Race
Black
White
Pap test history
No recent Pap
Recent Pap (<3 years)
Children in household
No
Yes
73.111.515.4—69.713.616.7—
79.7
82.5
74.0
63.1
67.9
42.7
11.3
8.4
9.9
15.1
13.2
17.2
9.1
9.1
16.1
21.8
18.9
40.1
<.01 75.1
72.8
64.8
72.2
66.1
55.0
12.3
13.8
16.1
12.2
12.0
14.0
12.6
13.4
19.0
15.6
21.9
31.0
.15
72.5
79.1
59.3
79.4
11.8
11.1
12.4
9.9
15.7
9.8
28.3
10.8
.1166.8
76.3
63.4
79.8
15.4
9.4
7.6
12.9
17.8
14.3
29.0
7.3
.01
79.8
84.6
70.8
72.3
59.7
7.3
8.4
15.8
11.0
12.6
12.9
7.0
13.4
16.7
27.7
<.0189.3
78.7
75.5
63.3
57.3
6.1
10.0
10.7
18.4
15.1
4.6
11.3
13.7
18.3
27.6
<.01
72.2
77.2
71.7
67.9
8.5
11.0
14.1
10.8
19.3
11.8
14.2
21.3
.29 80.2
72.4
74.7
58.0
8.8
12.6
11.0
18.9
11.0
15.0
14.3
23.1
<.01
83.5
69.5
7.5
12.9
9.0
17.5
.01 82.8
67.2
6.1
15.0
11.1
17.8
<.01
67.0
84.6
14.5
5.8
18.5
9.6
<.01 67.7
74.9
14.2
12.1
18.2
13.0
.19
72.4
75.8
11.5
11.5
16.1
12.8
.75 75.8
66.1
10.5
15.4
13.7
18.5
.01
70.1
76.3
12.7
10.7
17.2
13.0
.2970.8
69.4
12.1
14.7
17.1
15.9
.61
90.8
72.8
0.0
11.5
9.2
15.7
.55 79.4
68.8
14.2
13.6
6.4
17.6
.09
73.7
73.3
10.2
12.0
16.1
14.7
.82 68.9
70.0
11.9
13.8
19.2
16.2
.73
64.1
81.0
15.5
7.6
20.4
11.4
<.01 68.2
71.2
13.0
14.1
18.8
14.7
.38
Yes ¼ probably yes and definitely yes; No ¼ probably no and definitely no.
Children in household ¼ children of either gender, under age 18.
Percentages may not add to 100 due to rounding.
W.J. Christian et al. / Journal of Adolescent Health 44 (2009) 437–445
440

Page 5

were the most likely to accept vaccination for girls, and
widows were the least likely. Progressively higher annual in-
come was clearly inversely associated with acceptance in the
non-Appalachian area, but this trend was less clear in Appa-
lachia, although rates of acceptance differed significantly by
income group. Those with a college education or more were
the least likely to be accepting of vaccination, but this asso-
ciation was strong and significant in non-Appalachian Ken-
tucky, while slight and nonsignificant in Appalachia.
Health insurance coverage was significantly associated with
lower acceptance in both Appalachia and non-Appalachia,
with rates of acceptance being similar in both areas. We ob-
served significant differences between smokers’ and non-
smokers’ rates of acceptance, but this difference was
statistically significant in Appalachia only, and the rate of ac-
ceptance among Appalachian women was considerably
higher than that among women in non-Appalachia. Resi-
dence in a rural county was significantly associated with vac-
cine acceptance in non-Appalachia only. Acceptance of the
vaccine did not seem to be related to awareness of HPV in ei-
ther region. Race was marginally associated with vaccine ac-
ceptanceinnon-Appalachia.Inneitherregion dida history of
recent Pap testing appear to be related to vaccine acceptance.
Last, the presence of children in the household was signifi-
cantly associated with vaccine acceptance in Appalachia,
but this was not the case in non-Appalachia. The bivariate
analysis thussuggeststhat thefactorsassociatedwithvaccine
acceptance differ between Appalachian and non-Appala-
chian regions of Kentucky.
Our multivariate analysis results are shown in Table 3.
Broadly speaking, the results confirm those of the bivariate
analysis; they suggest that significant predictors of HPV vac-
cine acceptance differ between Appalachian and non-Appa-
lachian regions. In Appalachia, only smoking (odds ratio
[OR] ¼ 2.59) and the presence of children in the household
(OR ¼ 2.23) were significantly associated with vaccine ac-
ceptance, after adjusting for all other factors. Being married
was marginally significant in predicting nonacceptance of
the vaccine. These associations were not present in the
non-Appalachian group. There were, however, significant,
or near-significant, associations with age group (greatest ac-
ceptanceamongtheyoungestgroup),education(moreaccep-
tance among the less educated), health insurance coverage
(more acceptance among those with no coverage), and rural-
ity(moreacceptanceinruralareas)innon-Appalachia. These
associations were absent or nonsignificant in Appalachia.
Table 4 displays the results of the bivariate analysis con-
ducted to compare the responses to both study questions be-
fore and after FDA approval of the vaccine. Statewide,
awareness of HPV was substantially higher among those
surveyedinthelatterhalfof2006(47.5%vs.66.9%).Aware-
ness of HPV was increased in both Appalachian and non-
Appalachian areas, but this increase was large and significant
innon-Appalachian Kentucky (46.8% to70.9%), while mod-
est and nonsignificant in Appalachia (49.2% to 58.3%).
Additionally, awareness of HPV in non-Appalachian
counties was significantly greater than in Appalachian
counties in the final 6 months of the year, but not in the first
6. These results suggest awareness of HPV increased much
less in Appalachian Kentucky, if at all, compared to the
rest of the state.
Interestingly, the rates of vaccine acceptance for girls re-
mained virtually unchanged in both regions. Similarly, rates
of‘‘don’tknow’’responseswereverysimilarinbothregions,
and for both time periods.
Discussion
Because some HPV strains can be transmitted through
sexual contact, they should be viewed as sexually transmitted
infections. Also, because infection with a high-risk HPV
strain is a necessary condition for the development of inva-
sive cervical cancer, this disease can also be considered to
be sexually transmitted. Because of this, there are issues re-
garding administration ofan HPV vaccineto girls for the pre-
vention of cervical cancer based on the association of HPV
infection and sexual behavior. Some adults are concerned
that a vaccine for a sexually transmitted infection (STI) will
encourage sexual activity or condone unsafe or risky behav-
ior [17,19,33]. To our knowledge, this is the first large, pop-
ulation-based study addressing vaccine acceptance among
women of all ages.
The results of this study suggest that overall rates of vac-
cine acceptance were reasonably high at 70.2%, particularly
considering that the survey was initiated 6 months before the
HPV vaccine was approved, and before the widespread me-
diacoveragethathasensuedsincethen. TherateofHPVvac-
cine acceptance for girls was similar to results found in
a number of other recent studies in the United States, where
rates variedfromabout
[17,18,21,23,28,30]. Interestingly, there was no significant
shift toward greater acceptance in the second half of the
year after the vaccine entered the market and became more
visible through advertisement, even though this probably
had an impact on the increased awareness of HPV.
Despite similar rates of a positive response, the factors as-
sociated with acceptance of HPV vaccination for adolescent
girls among Appalachian and non-Appalachian women of
Kentucky differed. In the non-Appalachian region, younger
age, lack of a college degree, and residence in a rural county
were factors positively related to vaccination for girls. In Ap-
palachia, on the other hand, smoking and having children liv-
ing at home were the factors linked to vaccine acceptance,
while being marriedhad a negative impact.These differences
in the factors associated with vaccine acceptance by Appala-
chian status suggest that strategies to promote HPV vaccina-
tion need to be tailored somewhat differently depending on
regional or population characteristics.
Related to some extent is the association between vaccine
acceptance and rural residence only in the non-Appalachia
region. It is possible the lack of such associations in Appala-
chia was because of the fact that the three largest urban areas
66%to75% ofadults
W.J. Christian et al. / Journal of Adolescent Health 44 (2009) 437–445
441