2. P.M. Frew et al. / Vaccine 32 (2014) 1736–1744 1737
Fig. 1. Socioecological factors associated with influenza immunization among
pregnant minority women.
framing has been proposed as a potential method to promote pos-
itive health-seeking behavior such as immunization. According to
Prospect Theory, individuals tend to avoid risks when considering
gains and are willing to take on risks when considering losses [21].
Positively-oriented (“gain-frame”) messages communicate infor-
mation by emphasizing the benefits of the target health behavior,
while negatively-oriented (“loss-frame”) messages emphasize the
risks of not engaging in the behavior [16]. Furthermore, gain-
frame messages appear to be most persuasive when advocating
for preventive health behaviors (e.g., immunization). Loss-frame
messages are most persuasive when advocating for a behavior that
detects a health problem (e.g., getting a Pap smear) [16,22–26].
Prospect Theory therefore has been previously operationalized in
HIV and influenza vaccine studies to understand confluent factors
that shape immunization decision-making [26,27].
1.2. Socioecological influences
To examine intersecting factors that contribute to pregnant
women’s decision-making when faced with specific immunization
messages, we drew upon a socioecological framework for empiric
investigation of the dynamic interactions across individual, social
network/social influence microsystem, and community levels. Our
socioecological model (Fig. 1) posits that individual behavior is
influenced and defined by their surrounding ecology, environ-
ment, or systems [28]. From a conceptual standpoint, immunization
decision-making is informed by a “chain reaction” series of events
driven by an entire ecological system including direct and indirect
influencers.
This model specifies three key levels of behavioral dynamics.
The first level of the system – the individual level – refers to factors
that are individually exclusive such as perception of disease vulner-
ability [28]. The second level, the social network/social influence
microsystem, involves peer and family support for immunization
decision-making within a person’s immediate surroundings. Inter-
actions at this level are theorized to have a strong impact on
individual’s health decision-making due to strong interpersonal
dynamics in these relationships [28]. Finally, the community level
represents the healthcare utilization and provider communica-
tion on immunization, influences considered driven by structural
concerns such as culturally-competent providers and insurance
provision in geographic proximity to pregnant women [28].
Multilevel influences guide individual health decisions and
attitudes. As a result, message framing effects alone may not
be sufficient to promote immunization behaviors among at-risk
populations. Individual, social, and community-level factors may
significantly sway receptivity of these messages, whether gain- or
loss-framed. In addition to understanding the role of message fram-
ing in immunization decision-making behavior, we also sought to
understand these spheres of influence. By identifying socioeco-
logical influences on immunization behavior, we can craft more
comprehensive community-based messages to guide pregnant
minority women towards health-promoting behaviors [28].
To our knowledge, message framing within a socioecological
model of health behavior has not been evaluated on mater-
nal immunization outcomes among minority women. Our study
endeavored to test messages that would articulate maternal ben-
efits associated with vaccination (gain-frame) and, conversely,
illustrate negative consequences of foregoing immunization (loss-
frame) on their decision to obtain the seasonal influenza vaccine
[29–31]. Based on previous research, we hypothesized that gain-
framed messages would be more effective than loss-framed
messages in persuading women to obtain influenza immunization
[16,22–25]. We also sought to pinpoint socioecological effects on
immunization behavior. Incorporation of these influences in future
health campaigns and vaccine messages may promote preventive
health behavior of at-risk populations.
2. Materials and methods
2.1. Study design and sample
The study protocol was approved by the Emory University Insti-
tutional Review Board. We enrolled 276 women on the basis of
achieving at least 80% power to detect a 20% difference between
baseline and the two intervention groups. We assumed a baseline
immunization rate of approximately 15% with a significance level of
˛ ≤ 0.05 and 10% attrition at 30 days postpartum follow up. Through
data cleaning, we identified four duplicate enrollments. The first set
of survey data for each duplicate enrollee was retained for analy-
sis. The baseline survey assessed sociodemographic characteristics,
psychosocial factors associated with their intention to immunize
themselves and their infants, and message framing influences on
their immunization decision-making. The postpartum question-
naire assessed their health during pregnancy, infant outcomes, and
receipt of recommended immunizations.
Cohort recruitment began at the inception of influenza season
from September 2011 to May 2012. Follow-up began with women
due from October 2011 until we reached all women responsive
to our survey as they reached the 30-days postpartum milestone,
thereby concluding our follow-up by May 2013. Eligible partici-
pants included those who self-identified as Black/African American,
Hispanic/Latina, and/or Multiracial/multiethnic reflecting lineage
in these racial/ethnic categories. They also must have been 18–50
years old, able to read and write English and/or Spanish, and able
to provide written informed consent. Women who reported receipt
of influenza vaccine during the 2011–2012 influenza season were
excluded from the study.
Using venue-based sampling, we approached approximately six
hundred women to participate in the study. Those who expressed
interest were then asked a brief series of questions to assess their
eligibility. Nearly two-thirds of the women were ineligible for
enrollment based on preset criteria such as age, ability to read and
write English and/or Spanish, and previous receipt of the seasonal
3. 1738 P.M. Frew et al. / Vaccine 32 (2014) 1736–1744
influenza vaccine during their current pregnancy (occurring
between September 2011 and May 2012). Following informed
written consent procedures, women were immediately random-
ized to receive one of three types of vaccine messages—standard
vaccine information sheet (control), gain-frame, or loss-frame
messages. Baseline surveys were conducted in semi-private or
private locations.
Thirty days after their expected due dates, all participants were
re-contacted to complete postpartum follow-up questionnaires.
We attempted to reach all women via email and telephone (i.e.
calls and text messages). We also reached them through alternate
locator contacts if needed. Participants responsive to our requests
completed a 5–10 min telephone interview with project staff at a
time mutually convenient, and were subsequently compensated
with a mailed $5 gift card for their time and inconvenience.
2.2. Survey design and measurement
Study materials were developed in English and Spanish. Bilin-
gual community members reviewed these documents to ensure
their adequate readability and item comprehension prior to admin-
istration. We determined that the Flesch–Kincaid reading score of
5.9 of the final survey in either language met the acceptable criteria
of 6–8th grade reading level for our target population [32,33].
Initial survey questions assessed sociodemographic measures
(i.e., age, race/ethnicity, education, healthcare utilization, relation-
ship status). At the individual level, key behavioral assessments
such as past immunization history, perceived susceptibility to
become ill with influenza, and perceived vaccine efficacy were
completed. The questionnaire included scale items designed to
measure psychosocial indicators of attitudes and perceptions
surrounding immunization decision-making. Each scale item
was measured by a 5-point Likert scale (1-strongly disagree to
5-strongly agree), designed to assign meaningful values to an
underlying continuum of ratings [34]. Maternal uptake of the
influenza vaccine was assessed by asking women whether or
not they obtained influenza immunization during pregnancy.
Following a response of Yes or No, women were asked to explain
factors which influenced their decision.
To assess factors at the social level, participants were asked
to rate the degree to which friends and family supported immu-
nization during pregnancy, as well as questions ascertaining
immunization behavior of church members, fellow employees, and
other close members of their community. Finally, socioecological
factors at the broader community level were assessed with ques-
tionnaire items related to experience with the healthcare setting
and healthcare providers, including provider recommendation at
any point during the pregnancy (Table 4).
2.3. Statistical analyses
We evaluated characteristic differences between enrolled arms
using cross-tabulations, descriptive analyses, and two sample
t-tests; using the same methods, we also evaluated any dif-
ferences between enrolled and not enrolled participants. Using
seasonal influenza immunization as our primary outcome vari-
able, bivariate and multivariate correlations were conducted to
assess effect of demographic and social network/support indicators
(e.g. race/ethnicity, interpersonal support). Pearson’s chi-square
and Fisher’s exact tests were used to analyze associations between
sets of the outcome variable, socioecological, and message-framing
variables. Significance levels were set at ˛ = 0.05.
Regression analyses were performed to understand socioeco-
logical factors associated with influenza immunization among
members of this population. Variables were identified for the mul-
tivariate analysis based on bivariate outcomes. The multiple logistic
regression analysis evaluated the association between message
framing predictor variables (gain-frame/loss-frame/control mes-
sage), socioecological factors, and the outcome variable (maternal
uptake of influenza immunization during pregnancy). Interaction
was assessed by comparing full and reduced models.
3. Results
The majority (89.6%, n = 112) of the study population included
Black/African American women. We also retained Hispanic/Latina
Table 1
Participant sociodemographic characteristics–baseline vs. follow-up (n = 272).
Characteristic Total (n = 272) Baseline group (n = 146, 53.68%) Follow-up group (n = 126, 46.32%) p-Value
Age (mean = 26, missing = 7) 0.08
18–25 147 (55.47%) 86 (61.43%) 61 (48.80%)
26–35 99 (37.36%) 47 (33.57%) 52 (41.60%)
36–45 19 (7.17%) 7 (5.00%) 12 (9.60%)
Educational attainment (missing = 6) 0.04
Less than high school 45 (16.92%) 28 (20.00%) 17 (13.49%)
High school or GED 141 (53.01%) 79 (56.43%) 62 (49.21%)
More than high school 80 (30. 80%) 33 (23.57%) 47 (37.30%)
Racial/ethnic background (missing = 2) 0.65
Black/African American 239 (88.52%) 127 (87.59%) 112 (89.60%)
Hispanic/Latino/Chicano 19 (7.40%) 10 (6.90%) 9 (7.20%)
Other 12 (4.44%) 8 (5.52%) 4 (3.20%)
Employment status (missing = 2) 0.28
Employed 105 (38.89%) 57 (39.10%) 48 (38.10%)
Unemployed 150 (55.56%) 82 (53.97%) 68 (53.97%)
Other 15 (5.56%) 5 (3.47%) 10 (7.94%)
Annual household income (missing = 20) 0.72a
Less than $20,000 176 (69.84%) 98 (72.598%) 78 (66.67%)
$20,001–$40,000 41 (16.27%) 21 (15.56%) 20 (17.09%)
$40,001–$80,000 27 (10.71%) 12 (8.89%) 15 (12.82%)
More than $80,000 8 (3.17%) 4 (2.96%) 4 (3.42%)
Relationship status (missing = 1) 0.93a
Single/never married 198 (73.06%) 107 (73.79%) 91 (72.22%)
Married/domestic partner 52 (19.19%) 26 (17.93%) 26 (20.63%)
Divorced/separated 8 (2.95%) 5 (3.45%) 3 (2.38%)
Widowed 1 (0.37%) 1 (0.69%) 0 (0.00%)
Other 12 (4.43%) 6 (4.14%) 6 (4.76%)
a
Fisher’s exact test p-value.
4. P.M. Frew et al. / Vaccine 32 (2014) 1736–1744 1739
Table 2
Post-partum follow-up demographic characteristics (n = 126).
Characteristic Among all entrants
(n = 126)
Control group
(n = 39, 30.95%)
Gain frame group
(n = 45, 35.71%)
Loss frame group
(n = 42, 33.33%)
p-Value
Age (mean = 26,
missing = 1)
0.93
18–25 61 (48.80%) 18 (47.37%) 24 (53.33%) 19 (45.24%)
26–35 52 (41.60%) 17 (44.74%) 17 (37.78%) 18 (42.86%)
36–45 12 (9.60%) 3 (7.89%) 4 (8.89%) 5 (11.90%)
Educational
attainment
(missing = 1)
0.24
Less than high
school
17 (13.49%) 9 (23.08%) 5 (11.11%) 3 (7.14%)
High school or
GED
62 (49.21%) 18 (46.15%) 24 (53.33%) 20 (47.62%)
More than high
school
47 (37.30%) 12 (30.77%) 16 (35.56%) 19 (45.24%)
Race/ethnicity 0.21
Black/African
American
112 (89.60%) 36 (92.31%) 42 (95.45%) 34 (80.95%)
His-
panic/Latino/Chicano
9 (7.20%) 2 (5.13%) 2 (4.55%) 5 (11.90%)
Other 4 (3.20%) 1 (2.56%) 0 (0.00%) 3 (7.14%)
Employment status 0.95
Employed 48 (38.10%) 16 (41.03%) 18 (40.00%) 14 (33.33%)
Unemployed 68 (53.97%) 20 (51.28%) 24 (53.33%) 24 (57.14%)
Other 10 (7.94%) 3 (7.69%) 3 (6.67%) 4 (9.52%)
Annual household
income
(missing = 14)
0.63
Less than
$20,000
78 (66.67%) 28 (75.68%) 28 (68.29%) 22 (56.41%)
$20,001–$40,000 20 (17.09%) 5 (13.51%) 6 (14.63%) 9 (23.08%)
$40,001–$80,000 15 (12.82%) 4 (10.81%) 5 (12.20%) 6 (15.38%)
More than
$80,000
4 (3.42%) 0 (0.00%) 2 (4.88%) 2 (5.13%)
Relationship status 0.28
Single/never
married
91 (72.22%) 26 (66.67%) 35 (77.78%) 30 (71.43%)
Married/domestic
partner
26 (20.63%) 11 (28.21%) 7 (15.56%) 8 (19.05%)
Divorced/separated
3 (2.38%) 0 (0.00%) 0 (0.00%) 3 (7.14%)
Widowed 0 (0.00%) 0 (0.00%) 0 (0.00%) 0 (0.00%)
Other 6 (4.76%) 2 (5.13%) 3 (6.67%) 1 (2.38%)
(7.2%, n = 9) and Multiracial/multicultural women (3.2%, n = 4) in
the study (Table 1). The mean age of the participants was 26.5
years; most were under 35 years old. Additionally, the majority of
the 126 participants are from lower-income households with total
earnings among family members comprising ≤$20,000 per year
(66.7%, n = 78). Fifty-four percent (n = 68) of the women indicated
that they were unemployed and 86.5% (n = 109) achieved a high
school equivalent education or more. Most participants reported
their relationship status as single or never married (72.2%, n = 91).
There were no significant sociodemographic differences observed
at baseline among those assigned to the three study arms (con-
trol, gain- or loss-frame arms) (Table 1). Notably, we found that
despite a relatively low retention rate (46%, n = 126), women who
participated in the 30-days postpartum data collection did not dif-
fer significantly from those lost to follow-up on key demographic
variables such as household income and employment status with
the exception of educational attainment (p = 0.04) (Table 2).
Over half of the participants indicated at baseline that they
were unlikely to obtain influenza immunization during pregnancy
(n = 65, 56.0%). The proportion of women who indicated that they
were undecided about immunization during pregnancy (n = 26,
22.4%) did not differ significantly from that of women indicating
they were likely to obtain the seasonal influenza vaccine during
pregnancy (n = 25, 21.6%). In the follow-up survey, nearly one-third
of study participants indicated that they received the seasonal
influenza vaccine during pregnancy (n = 36, 28.6%) following
exposure to our presented messages.
3.1. Message framing outcomes
The results of the multivariate analyses between paired mes-
sage framing groups [Model 1: gain vs. loss (n = 87); Model 2:
gain vs. control (n = 84); and Model 3: loss vs. control (n = 81)] do
not indicate any differences between gain- and loss-framed mes-
sages on immunization (Model 1). Furthermore, compared to those
receiving the control message, neither those receiving the gain-
[OR = 0.5176, (95% CI: 0.203,1.322)] nor loss-framed [OR = 0.5000,
95% CI: (0.192,1.304)] messages had significant association with
increased likelihood of immunization during pregnancy (Models
2 and 3). Yet, we observed that mothers who received a recom-
mendation from a healthcare provider were more likely to obtain
influenza immunization as demonstrated in Model 2 [OR = 10.45,
95% CI: (2.26,48.36)]. Additionally, women who perceived greater
than 50% influenza vaccine effectiveness were also more likely
to obtain immunization during pregnancy as demonstrated in
Model 1 [OR = 7.03, 95% CI: (1.88,26.30)]. Finally, women who
obtained prenatal care at clinics within or connected to hospitals
had the strongest likelihood of obtaining influenza immunization
5. 1740 P.M. Frew et al. / Vaccine 32 (2014) 1736–1744
Table 3
Factors associated with seasonal influenza vaccine uptake (n = 126).
Odds ratio estimate 95% confidence interval p-Value
Logistic model (n = 123)
Vaccine recommendation by healthcare provider
Recommended 3.934 1.331, 11.627 0.013*
Not recommended Referent
Primary source of prenatal care during pregnancya
Hospital-based clinic 2.584 1.091, 6.122 0.031*
Other clinical site Referent
Normative support for influenza immunization during pregnancyb
Agree 3.405 1.412, 8.212 0.006**
Disagree or neutral Referent
Message framing models
Model 1 (n = 87)–gain vs. loss
Gain vs. loss 1.0353 0.387, 2.767 0.945
Influenza vaccine >50% effective vs. <50% effective 7.0345 1.881, 26.303 0.0019**
Prenatal care at hospital-based clinic vs. other clinical sites 2.8438 1.032, 7.835 0.0392*
Model 2 (n = 84)–gain vs. control
Gain vs. control 0.5176 0.203, 1.322 0.166
Provider vaccination recommendation vs. none 10.4516 2.259, 48.363 0.0004**
Prenatal care at hospital-based clinic vs. other clinical sites 3.2888 1.257, 8.604 0.0133*
Model 3 (n = 81)–loss vs. control
Loss vs. control 0.5000 0.192, 1.304 0.154
Other factors, univariate analysis
Perceived likelihood of being sick with influenza during pregnancyc
Likely 2.414 1.047, 5.568 0.037*
Not Likely or neutral Referent
Perceived influenza vaccine efficacy
≥50% 3.071 1.292, 7.302 0.010*
≤50% Referent
Relationship status
Single, divorced, widowed 3.100 1.094, 8.782 0.028*
Married/domestic partner Referent
Likelihood of obtaining influenza immunization during pregnancyc
Likely 8.760 3.381, 22.696 <0.001*
Not likely or neutral Referent
a
Other sources of prenatal care (e.g. community health clinic, OB/Gyn office, etc.) were not significant and thus not included.
b
All participants indicating ‘strongly agree’ or ‘agree’ are classified as ‘agree’. All participants indicating ‘neutral/not sure’, ‘disagree’, or ‘strongly disagree’ are classified as
‘disagree, or neutral’.
*
p < 0.05.
**
p < 0.01.
c
From the original Likert scale, those indicating a six or greater were classified as ‘likely’, those indicating a five or less were classified as ‘not likely, or neutral’ (wherein
five was presumed to indicated the participant was neutral).
during pregnancy as demonstrated in Model 1 [OR = 2.84, 95% CI:
(1.03,7.84)] and Model 2 [OR = 3.29, 95% CI: (1.26,8.60)].
3.2. Socioecological and psychosocial influences
The results of individual correlations reflect factors such as
individual attitudes and health-seeking behaviors that contributed
to influenza immunization (Table 3). These included intent to
obtain the influenza vaccine during pregnancy [OR = 8.76, 95% CI:
(3.38,22.67)] and receipt of the influenza immunizations in the five
preceding years [OR = 3.45, 95% CI: (1.54,7.70)]. Individuals’ percep-
tion of susceptibility to influenza during pregnancy significantly
increased likelihood of immunization during pregnancy [OR = 2.41,
95% CI: (1.05,5.57)] compared to those who did not perceive
themselves as susceptible to influenza-related illness during preg-
nancy. The results further indicate that women were more likely
to have received the vaccine if they were single, divorced, wid-
owed or never married [OR = 3.10, 95% CI: (1.09,8.78)], compared to
married women. Additional individual-level demographic factors
such as race and ethnicity, educational attainment, income, and
employment status were not significant factors affecting immu-
nization.
We also examined additional variables such as perceived sever-
ity of the influenza during pregnancy, number of household
members including children who could potentially transmit the
influenza virus, and women’s perceived vulnerability in the context
of a household of infected individuals. These additional variables
were not significant in the analyses.
Social support, healthcare provider recommendation, and
source of prenatal care were also examined in multivariate analy-
ses. We assessed potential interactions among these variables and
found that none of them were significant. We therefore excluded
them from the final model. We identified that social support
was strongly associated with influenza immunization during preg-
nancy. In analyses that controlled for provider recommendation
and source of prenatal care, we found that receipt of influenza vac-
cine was 3 times higher among those who indicated that family
and friends were supportive of receiving influenza immunization
[OR = 3.41, 95% CI: (1.412,8.212)].
We also assessed the effect of community-level factors such as
source of prenatal care and the recommendation of a healthcare
provider such as a doctor or nurse on decision-making. In analyses
that controlled for social support and source of prenatal care, we
found that receipt of influenza vaccine was approximately 4 times
higher among women who received healthcare provider recom-
mendation [OR = 3.93, 95% CI: (1.33,11.63)]. Finally, women were
more likely to obtain influenza immunization during pregnancy if
their primary source of prenatal care was a clinic based in or near
a hospital [OR = 2.54, 95% CI: (1.091,6.122)], controlling for social
support and provider recommendation.
6. P.M. Frew et al. / Vaccine 32 (2014) 1736–1744 1741
Table 4
Survey measures of maternal intent to obtain immunization.
Survey item Response choices
Main outcome
Did you get a flu shot during your pregnancy? Yes/no/don’t know
Baseline variables
Please indicate how protective you think the flu vaccine will be for a pregnant
woman?
0% (Not protective), 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%
(completely protective)
I would be less likely to get a flu vaccine if it had any minor side effects such as
fatigue or fever
Same as above
On a scale of 0 (definitely not) to 10 (definitely so), please rank your likelihood
of getting a flu shot during your pregnancy:
0 (Definitely not)–10 (definitely so) [dichotomized as 0–5 = “unlikely”,
6–10 = ”likely”]
My friends and family would support my decision to get a flu shot while
pregnant
1—Strongly agree
2—Agree
3—Neutral/not sure
4—Disagree
5—Strongly disagree
Follow-up variables
During your pregnancy, did a healthcare provider, such as a doctor or nurse,
recommend that you get the flu shot?
Yes/no
During your pregnancy, where did you usually go for your prenatal care? a. Primary-care doctor’s office
b. Ob/Gyn doctor’s office
c. Community/public health clinic
d. Storefront clinic (i.e. CVS, RiteAid, Walgreen’s clinic)
e. Hospital
f. Emergency room
g. Prison clinic
h. School health clinic
i. Worksite health clinic
j. Other:
Sociodemographic Indicators
What is your date of birth? MM/DD/YYYY
What is the highest level of school that you have completed? a. K-8 grade
b. 9-11 grade
c. High school graduation/GED
d. Technical/vocational or associates
e. Bachelor degree
f. Master’s degree
g. Doctorate
How would you describe your ethnic background? a. African American/black
b. Hispanic/Latino/Chicano
c. Caucasian/white
d. Other—please specify
What is your relationship status? a. Single/never married
b. Married/domestic partner
c. Divorced/separated
d. Widowed
e. Other—please specify
Which of the following best describes your employment status? a. Employed—full time
b. Employed—part-time
c. Unemployed
d. Other—please specify
What is your annual household income (i.e. combined income of all
members of your family)?
a. Less than $20,000
b. $20,001–$40,000
c. $40,001–$60,000
d. $60,001–$80,000
e. $80,001–$100,000
f. More than $100,001
Messages presented
Gain-frame “Getting a flu shot during pregnancy will protect your baby from getting the
flu, and will continue to protect your baby for up to 6 months after she is born.”
Loss-frame “Don’t risk the life of your unborn child by skipping a flu shot. Did you know. . .
Flu-related illness could jeopardize the lives of both mother and baby?”
Message resonance
“This ad is very appealing to me” 6—Strongly agree
7—Agree
8—Neutral/not sure
9—disagree
10—Strongly disagree
“This ad is easy to remember” 1—Strongly agree
2—Agree
3—Neutral/not sure
4—disagree
5—Strongly disagree
“This ad is new and fresh” 1—Strongly agree
2—Agree
3—Neutral/not sure
4—Disagree
5—Strongly disagree
7. 1742 P.M. Frew et al. / Vaccine 32 (2014) 1736–1744
4. Discussion
Reaching pregnant women with culturally-relevant and highly
salient messages in the context of prenatal care visits represents
an optimal time to promote influenza immunization [35,36]. In
our study, we demonstrate the critical importance of promot-
ing vaccine messaging within the context of healthcare delivery,
while demonstrating the simultaneous influences of socioecologi-
cal factors such as individual perceptions and social support. This
study contributes to a growing body of literature demonstrating
the importance of promoting immunization as a standard of care
during the prenatal care visit timeframe [37]. The results also sup-
port that message framing has a weak effect on vaccination uptake
as a prevention behavior [15,22,38,39]. Furthermore, our results
indicate the important role of multilevel influences on influenza
immunization among pregnant minority women.
4.1. Community effects: Healthcare facilities and the role of the
provider
We observed a four-fold increase in immunization with vac-
cine recommendation delivered by a healthcare provider such as
a doctor or nurse. Concurrently, pregnant women in our popula-
tion were nearly three times more likely to choose to influenza
immunization if receiving prenatal care in a hospital-based clinic
setting. Pregnant minority women who receive care in a hospital
or clinic-based setting may be exposed to more vaccine promotion
messages than those receiving care in non-hospital based settings,
or may find it more convenient to obtain the vaccine while in the
hospital for other medical care. Further research is necessary to elu-
cidate the qualitative differences between those receiving care in
a hospital-based clinic setting and those receiving care elsewhere.
Nevertheless, these preliminary results suggest that location and
context of prenatal care plays a critical role in the decision to immu-
nize.
Provider recommendations during pregnancy and use of
a hospital-based clinic setting for prenatal care may shape
pregnant minority women’s decision to vaccinate during preg-
nancy. Exposure to vaccine messages at hospitals or clinics and
recommendation by providers is therefore critical during the pre-
natal period. Health campaigns at hospitals have been shown to
contribute to positive health seeking behaviors [29,30,40–43], and
may similarly impact influenza immunization among pregnant
minority women.
In a previously conducted trial, women expressed a preference
for information given to them during prenatal visits [37]; however,
maternal vaccine education during the prenatal period is not part
of standard obstetric care [44]. We argue that vaccination mes-
sages should be incorporated in provider visits as a standard of care
as well as delivered through hospitals, which serve as significant
points of intervention.
4.2. Social effects: Family, friends and a woman’s
decision-making behavior
Social networks and social influences operate as significant
mediating factors in preventive health seeking behaviors, including
immunizations [13]. Pregnant minority women in this popula-
tion were three times more likely to obtain the seasonal influenza
vaccine during pregnancy if they indicated that they felt that
their family and friends would support their decision to receive
the vaccine during pregnancy. Friends and family who support a
woman’s decision to obtain immunization during pregnancy may
promote her sense of self-efficacy in protecting her and her unborn
child(ren) [45], or may be reflective of broader social norms which
support immunization during pregnancy [46–48].
Women who report the support of family and friends
in receiving the influenza immunization during pregnancy
may also experience greater overall support and enjoy higher social
capital than those who indicate that their friends and family would
not support their decision to obtain the flu shot during pregnancy.
As such, this item may operate as a broader indicator of social cap-
ital and support from close family and friends, which in turn has
been tied to broad health outcomes and health-seeking behaviors
[49]. Conversely, results of the logistic regression model indicate
that the odds a woman will obtain influenza immunization dur-
ing pregnancy are significantly increased if she identifies as single,
divorced, widowed, or never married. Women in this relationship
status may exert more control over their health and the health
of their unborn child, as they are likely the primary providers for
themselves and their unborn child.
4.3. Individual knowledge attitudes and practice: The decision to
vaccinate
The 2011–2012 seasonal influenza vaccine demonstrated an
approximate 52% effectiveness rate against influenza virus [50]. In
our sample, 43% (n = 52) of participants believed vaccine efficacy
to be less than 50%. Those who believed the influenza vaccine was
more than 50% effective were more likely to immunize themselves.
Similarly, participants who believed that they were likely to get the
influenza during pregnancy were nearly three times more likely
to immunize. These results therefore illustrate the significance of
individual knowledge and attitudes regarding vaccine efficacy and
individual susceptibility. Such attitudes may be affected not only
by message framing but also by healthcare encounters and social
experiences.
Family and friends’ support may increase a woman’s belief
that the vaccine is effective in protecting herself and her unborn
child against influenza. Conversations with healthcare providers
may lead pregnant women to consider their susceptibility to
influenza and the protection offered by the vaccine. Furthermore,
gain-framed messaging may improve a woman’s attitude towards
vaccine efficacy and positively impact her decision to obtain the
influenza vaccine during pregnancy.
4.4. Study limitations
We acknowledge that the sampling of minority women from
one southeastern city is not representative of other cities in the
United States. Additionally, we recognize the potential for partici-
patory bias as women who were agreeable to participating in the
study were included and therefore may not be representative of
the actual population of pregnant Hispanic or African American
women. We experienced challenges associated with retention of
the study population as we offered a $5 gift card for their time
and inconvenience associated with completion of the 30-day post-
partum questionnaire. This was likely insufficient for many thereby
resulting in our final retention of 46% of the recruited study popula-
tion. No significant differences other than educational attainment
were found between those retained and those lost to follow-up
in our sample. Additionally, we understand that new mothers
are subject to newborn and stress-related distraction during the
postpartum follow-up, which may have affected participants’
response and contributed to lower retention rates.
5. Conclusion
Maternal perceptions of vaccine efficacy and susceptibility
to influenza during pregnancy are significantly associated with
influenza immunization. Our study demonstrates the importance
of understanding the context in which presented messages are
8. P.M. Frew et al. / Vaccine 32 (2014) 1736–1744 1743
evaluated. In particular, women who receive healthcare provider
recommendations about vaccination and those who have obtained
prenatal care at hospital-based clinic practices were more likely
to obtain immunization, compared to those who did not have
these encounters. Vaccination messages at hospital-based settings
should also seek to educate non-pregnant individuals, particularly
family members and partners, as our evidence suggests they exert
significant influence over pregnant women in their lives, as well as
the way pregnant women are likely to receive and internalize such
messages. Finally, this study demonstrates that message framing
effects are subject to individual perceptions and attitudes that may
be mediated by other socioenvironmental factors.
Sources of support
This study was partially supported by a Kaiser Permanente
Georgia community benefits grant and a grant from the Centers
for Disease Control and Prevention (CDC), grant 5P01TP000300 to
the Emory Preparedness and Emergency Response Research Center,
Emory University (Atlanta, Georgia).
Conflict of interest statement
The authors report no conflict of interest.
Acknowledgments
This study was partially supported by a Kaiser Permanente
Georgia community benefits grant and a grant from the Centers
for Disease Control and Prevention (CDC), grant 5P01TP000300 to
the Emory Preparedness and Emergency Response Research Cen-
ter, Emory University (Atlanta, Georgia). Special thanks to Drs.
Ruth Berkelman and Robert Davis, along with Ms. Ellen Whitney
and Mr. Rick Kern. Its contents are solely the responsibility of the
authors and do not necessarily represent the official views of the
CDC. Finally, we express our deepest gratitude to all of our com-
munity participants for their study involvement and to the venue
owners who offered permission to conduct recruitment on their
premises.
References
[1] Beigi RH, et al. Economic value of seasonal and pandemic influenza vaccination
during pregnancy. Clin Infect Dis 2009;49(12):1784–92.
[2] Moro PL, et al. Safety of seasonal influenza and influenza A (H1N1) 2009 mono-
valent vaccines in pregnancy. Expert Rev Vaccines 2012;11(8):911–21.
[3] Centers for Disease Control and Prevention. Maternal and infant outcomes
among severely ill pregnant and postpartum women with 2009 pandemic
influenza A (H1N1)—United States April 2009–August 2010. MMWR Morb Mor-
tal Wkly Rep 2011;60(35):1193–6.
[4] Georgia Department of Health. Annual vaccine preventable diseases
surveillance report—Georgia, 2010; 2010 October. Available from:
http://health.state.ga.us/pdfs/epi/vpd/2010%20VPD%20Report.pdf.
[5] Louie JK, et al. Severe 2009 H1N1 influenza in pregnant and postpartum women
in California. N Engl J Med 2010;362(1):27–35.
[6] Ahluwalia IB, et al. Seasonal influenza vaccine coverage among pregnant
women: pregnancy risk assessment monitoring system. J Women’s Health
2011;20(5):649–51.
[7] Fiscella K, et al. Disparities in health care by race ethnicity, and language among
the insured: findings from a national sample. Med Care 2002;40(1):52–9.
[8] Neuzil KM, et al. Impact of influenza on acute cardiopulmonary hospitalizations
in pregnant women. Am J Epidemiol 1998;148(11):1094–102.
[9] Read JS, Riley L. Progress in overcoming barriers to influenza immunization of
pregnant women. Am J Obstet Gynecol 2012;207(Suppl 3):S1–2.
[10] Flowers L. Racial and ethnic disparities in influenza and pneumococcal immu-
nization rates among Medicare beneficiaries. Issue Brief (Public Policy Inst (Am
Assoc Retired Pers)) 2007;IB83:1–6.
[11] Logan JL. Disparities in influenza immunization among US adults. J Nat Med
Assoc 2009;101(2):161–6.
[12] Frew PM, et al. Acceptance of pandemic 2009 influenza A (H1N1) vaccine in a
minority population: determinants and potential points of intervention. Pedi-
atrics 2011;127(Suppl 1):S113–9.
[13] Langlie JK. Social networks, health beliefs, and preventive health behavior. J
Health Soc Behav 1977;18(3):244–60.
[14] Shui I, et al. Factors influencing African–American mothers’ concerns about
immunization safety: a summary of focus group findings. J Nat Med Assoc
2005;97(5):657–66.
[15] Ferguson E, Gallagher L. Message framing with respect to decisions about vac-
cination: the roles of frame valence, frame method and perceived risk. Br J
Psychol 2007;98(Pt 4):667–80.
[16] Bartels RD, Kelly KM, Rothman AJ. Moving beyond the function of the health
behaviour: the effect of message frame on behavioural decision-making.
Psychol Health 2010;25(7):821–38.
[17] Satia JA, et al. Qualitative study to explore Prospect Theory and message fram-
ing and diet and cancer prevention-related issues among African American
adolescents. Cancer Nurs 2010;33(2):102–9.
[18] Tversky A, Kahneman D, Kahneman D. Advances in prospect theory:
cumulative representation of uncertainty. J Risk Uncertainty 1992;5:
297–323.
[19] Tversky A, Kahneman D. The framing of decisions and the psychology of choice.
Science 1981;211:4481–538.
[20] Kahneman D, Tversky A. Prospect theory—analysis of decision under risk.
Econometrica 1979;47(2):263–91.
[21] Abhyankar P, O’Connor DB, Lawton R. The role of message framing in promoting
MMR vaccination: evidence of a loss-frame advantage. Psychol Health Med
2008;13(1):1–16.
[22] Gallagher KM, Updegraff JA. Health message framing effects on atti-
tudes, intentions, and behavior: a meta-analytic review. Ann Behav Med
2012;43(1):101–16.
[23] Jung ME, et al. Increasing calcium intake in young women through gain-
framed, targeted messages: a randomised controlled trial. Psychol Health
2011;26(5):531–47.
[24] Gallagher KM, et al. Perceived susceptibility to breast cancer moderates the
effect of gain- and loss-framed messages on use of screening mammography.
Health Psychol 2011;30(2):145–52.
[25] Hull SJ. Perceived risk as a moderator of the effectiveness of framed HIV-test
promotion messages among women: a randomized controlled trial. Health
Psychol 2012;31(1):114–21.
[26] Evangeli M, et al. Does message framing predict willingness to participate in
a hypothetical HIV vaccine trial: an application of Prospect Theory. AIDS Care
2013;25(7):910–4.
[27] McCaul KD, Johnson RJ, Rothman AJ. The effects of framing and action
instructions on whether older adults obtain flu shots. Health Psychol
2002;21(6):624–8.
[28] Bronfenbrenner U. Ecological models of human development. In: International
encyclopedia of education. Oxford, UK: Elsevier; 1994.
[29] Snyder LB. Health communication campaigns and their impact on behavior. J
Nutr Educ Behav 2007;39(2 Suppl):S32–40.
[30] Snyder LB, et al. A meta-analysis of the effect of mediated health communi-
cation campaigns on behavior change in the United States. J Health Commun
2004;9(Suppl 1):71–96.
[31] National Cancer Institute. Making health communication programs work.
National Cancer Institute; 1989. Available at: http://www.cancer.gov/
cancertopics/cancerlibrary/pinkbook/page1.
[32] Davis TC, et al. Reading ability of parents compared with reading level of pedi-
atric patient education materials. Pediatrics 1994;93(3):460–8.
[33] Miller B, Bodie M. Determination of reading comprehension level for effective
patient health-education materials. Nurs Res 1994;43(2):118–9.
[34] Meyers LS, Gamst G, Guarino AJ. Applied multivariate research: design and
interpretation. Thousand Oaks, CA: Sage Publications; 2006. p. 722.
[35] Steinhoff MC, et al. Influenza immunization in pregnancy—antibody responses
in mothers and infants. N Engl J Med 2010;362(17):1644–6.
[36] Tamma PD, et al. Safety of influenza vaccination during pregnancy. Am J Obstet
Gynecol 2009;201(6):547–52.
[37] Vannice KS, et al. Attitudes and beliefs of parents concerned about vaccines:
impact of timing of immunization information. Pediatrics 2011;127(Suppl
1):S120–6.
[38] Gallagher KM, Updegraff JA. Erratum to: health message framing effects on
attitudes, intentions, and behaviors: a meta-analytic review. Ann Behav Med
2013;46(1):p127.
[39] Gerend M, Shepherd M. Message framing it does a body good: effects of message
framing and motivational orientation on young women’s calcium consumption.
J Health Psychol 2012.
[40] Llupià A, Olivé MG, Quesada V, Aldea S, Sequera M, Ríos VG, García-Basteiro
J, Varela AL, Bayas P, Trilla JMA. Evaluating influenza vaccination campaigns
beyond coverage: a before-after study among health care workers. Am J Infect
Control 2013;41(8):674–8.
[41] Stockwell MS, Fiks A. Utilizing health information technology to improve
vaccine communication and coverage. Hum Vaccines Immunother
2013;9(8):1802–11, http://dx.doi.org/10.4161/hv.25031.
[42] Bartick M, et al. Closing the quality gap: promoting evidence-based breastfeed-
ing care in the hospital. Pediatrics 2009;124(4):pe793–802.
[43] Williams HC, Smith D, du Vivier AW. Evaluation of public education campaigns
in cutaneous melanoma: the King’s College Hospital experience. Br J Dermatol
1990;123(1):85–92.
[44] Link-Gelles R, et al. Missed opportunities: a national survey of obstetricians
about attitudes on maternal and infant immunization. Matern Child Health J
2012;16(9):1743–7.
9. 1744 P.M. Frew et al. / Vaccine 32 (2014) 1736–1744
[45] Bandura A. Health promotion by social cognitive means. Health Educ Behav
2004;31(2):143–64.
[46] Berkowitz AD. Applications of social norms theory to other health and social
justice issues. In: Perkins HW, editor. The social norms approach to preventing
school and college age substance abuse: a handbook for educators, counselors,
clinicians San Francisco, CA/Trumansburg, NY: Jossey-Bass; 2002.
[47] de Visser R, Parikh WL, Lawrie CA. The importance of social norms for uptake
of catch-up human papillomavirus vaccination in young women. Sex Health
2011;8(3):330–7.
[48] Berkowitz AD. The social norms approach: theory, research, and annotated
bibliography. NY: Trumansburg; 2004. p. 14886–5701. http://www.cdc.gov/
flu/pastseasons/1112season.htm
[49] I Kawachi, B.P.K., and R Glass, Social capital and self-rated health: a contextual
analysis. American Journal of Public Health. 89(8):p. 1187-1193.
[50] 2011–2012 Influenza Season: Questions and Answers. Centers for Disease
Control and Prevention, National Center for Immunization and Respiratory Dis-
eases. Last updated November 22, 2013. Accessed December 18, 2013. URL:
http://cdc.gov/flu/pastseasons/1112season.htm.