Multilevel factors influencing hepatitis B screening and vaccination among Vietnamese Americans in Atlanta, Georgia.

Chronic hepatitis B virus (HBV) infection may lead to liver cirrhosis, chronic liver disease, and liver cancer. Immunization rates are suboptimal among Asian Americans/Pacific Islanders (AAPIs), who remain disproportionately affected by these illnesses. We investigated socioecological factors affecting HBV prevention among 316 Vietnamese Americans in Atlanta, Georgia. Social and community support of HBV vaccination was associated with screening (OR=1.69, 95% CI [1.21,2.38]), vaccination (OR=1.89, [1.27,2.81]), and intent to vaccinate (OR=1.77, [1.13,2.78]). Misconceptions decreased screening likelihood (OR=0.67, [0.46,0.99]) and vaccination (OR=0.55, [0.35,0.86]). Those able to pay for medical treatment (OR=1.23, [1.01,1.50]) were also more likely immunized, and greater transportation access (OR=1.42, [1.07,1.87]) was associated with greater intention to vaccinate. Multi-level factors facilitated HBV vaccination in this population. Tailored, culturally appropriate communication strategies will positively influence immunization uptake.

Introduction

Chronic hepatitis B virus (HBV) infection is an international health problem affecting 350 million people globally [1-4]. In 2011, the Centers for Disease Control and Prevention (CDC) estimated that between 700,000 and 1.4 million individuals in the United States were living with chronic HBV [4]. Prior studies estimated chronic HBV prevalence in persons over the age of 6 years at approximately 0.27 percent (95% CI, 0.20% to 0.34%) among all ethnic groups in the United States between 1999 and 2008 [5]. Recent CDC estimates point to 18,000 cases of acute HBV as of 2011 (7,400 to 86,200 cases as of 2011) [4]. Immunization coverage is suboptimal in the United States, and in 2011, overall vaccination coverage among adults ages 19 to 49 years was 35.9 percent [6]. Chronic HBV infection, a leading cause of cirrhosis, chronic liver disease, and liver cancer (hepatocellular carcinoma), is known as a “silent killer” because carriers remain asymptomatic for many years [1,7-11]. Asian Americans/Pacific Islanders (AAPIs) have the highest rate of liver cancer in the United States (18.7 cases per 100,000 persons), and an estimated 1 in 12 AAPIs are living with HBV in this country [12]. While HBV infection can be prevented by vaccination, vaccination uptake is low among AAPI [13-20].

Of the many Asian groups represented by these statistics, the Vietnamese population is particularly affected by HBV infections. Vietnamese men are disproportionately burdened with an HBV infection rate 15 times that of Caucasian men (7.2 cases per 100,000 persons) [21-24]. A recent study found that 16 percent of Vietnamese seeking clinical care were infected with HBV, yet 33 percent of those chronically infected were unaware of their infection [25]. Furthermore, despite the provision of free HBV vaccinations in community settings, a minority (33 percent) of Vietnamese individuals surveyed reported knowledge of these immunization events [15,16,26,27].

According to U.S. Census data, the Vietnamese population increased substantially from 1990 to 2000, including in the southern United States [28,29]. During this period, many first-generation Vietnamese Americans, who immigrated to the United States and subsequently became naturalized citizens, were exposed to mass vaccination campaigns targeting HBV endemic in their country of origin [7]. By contrast, Vietnamese Americans who immigrated to the United States more recently may be less likely to immunize against HBV.

Studies examining knowledge of HBV among Vietnamese Americans found most can identify mother to child, sexual intercourse, sharing toothbrushes, and sharing needles with infected persons as potential HBV transmission routes [15,16,18,19,30,31]. However, at least 33 percent of those surveyed also held incorrect beliefs about HBV transmission, indicating that the virus could be transmitted through cigarette smoking, airborne methods, and communal foods [15,16,18,20,30,32]. Studies have found that the majority of Vietnamese Americans believe HBV causes cancer and is terminal, yet a majority also believe HBV could be treatable if detected early and can be prevented by vaccination [14,16,18,19,26,27,33]. Knowledge and attitudes surrounding HBV, including perceived severity, have been associated with vaccination [18,26,27].

Previous studies have demonstrated the importance of physician vaccine recommendation on HBV immunization acceptance among Vietnamese Americans [19,34,35]. Although this group experiences greater HBV-related health disparities compared to other racial and ethnic populations, the CDC’s Racial and Ethnic Approaches to Community Health (REACH) 2010 Risk Factor Survey found that only 38 percent of physicians treating AAPIs reported HBV-related discussions in clinical encounters with patients [36]. Similarly, less than one-third of primary care providers routinely screen AAPI patients for HBV [37]. Although most physicians (83%, N=393 sampled) perceive HBV to have a significant role in the health status of AAPIs, many (62 percent) report unfamiliarity with current HBV screening and treatment guidelines [37].

Many studies have assessed issues related to HBV vaccine acceptability, yet few have investigated the role of social and community factors in health care decisions [13-16,18-20,25-27,30,31,33-36,37]. Evidence indicates that peer and social support is important in promoting screening behaviors for cancer [16,26]. One study further assessed general health communication among Vietnamese and concluded that healthy Vietnamese men were less likely to get information from doctors or nurses, compared to those with poor health status [38]. Among Vietnamese men in Seattle, most received health information through Vietnamese news sources or friends and family [38].

Among metropolitan areas with large Vietnamese populations, Atlanta ranks 10th in the United States and third in the southern region, with an estimated Vietnamese population of almost 45,000 in the metropolitan area [29]. The purpose of this study was to investigate barriers to HBV vaccine uptake and to identify potential intervention approaches that may increase HBV immunization in this vulnerable population. Specifically, we explored the extent to which individual-, dyadic- (provider/patient), and community-level factors influence HBV screening and vaccination decision-making.

Methods

Study Design and Sample

During the fall of 2010, participants were recruited in venues during randomly selected blocks of time on varying dates, a proven method for obtaining representative populations in cross-sectional survey samples [39]. Recruitment venues were located throughout metropolitan Atlanta, Georgia. The sampling frame comprised 14 locations, including health fairs, community-based organizations, offices, churches, temples, festivals, and other community events.

The target population included English- and Vietnamese-speaking adults of Vietnamese descent. About 385 individuals were invited to participate, and 316 provided written informed consent, an 82 percent response rate. Adults at least 18 years of age who could read and speak English or Vietnamese were eligible. Participants were asked to complete a 157-item questionnaire that measured attitudes toward HBV disease, screening, and vaccination. Questionnaires were conducted in Vietnamese or English. Participants were offered a $10 gift card or health promotion incentive. Table 1 displays sociodemographic characteristics of the sample.

Table 1

Sociodemographic Characteristics of Respondent Population (N=316).

	Frequency (%)
Gender
Male	136 (43)
Female	170 (54)
Missing	10 (3.2)
Country of Birth
Vietnam	288 (91)
USA	16 (5.1)
Other	9 (2.8)
Missing	3 (0.9)
Primary Language
Vietnamese	259 (82)
English	36 (11)
Other	13 (4.1)
Missing	8 (2.5)
Educational Attainment
Some school	65 (21)
English	36 (11)
High School graduate	115 (36)
Associate/Vocational	57 (18
Bachelor	54 (17)
Masters/Professional	12 (3.8)
Doctorate	3 (0.9)
Missing	10 (3.2)
Employment Status
Employed full-time	138 (44)
Employed part-time	52 (16)
Unemployed	74 (23)
Other	39 (12)
Missing	13 (4.1)
Annual Household Income
Less than $20,000	120 (38
$20,001-$40,000	79 (25)
$40,001-$60,000	47 (15)
$60,001-$80,000	19 (6.0)
$80,001-$100,000	11 (3.5)
More than $100,000	13 (4.1)
Missing	27 (8.5)
Median Age (IQR)	41 (29-54 years)
Missing	29
Percent of Life Spent in U.S.
Less than 50%	189 (60)
More than 50%	73 (23)
Missing	54 (17)
Insurance Status
Insured	173 (55)
Uninsured	122 (39)
Missing	21 (6.6)
Historical Hepatitis B Status
Negative	205 (65)
Positive	31 (9.8)
Don't Know	48 (15)
Missing	32 (10)
Vaccination Status
None	145 (46)
At least one dose	106 (34)
One shot only	37 (12)
Two shots only	29 (9)
Three shots series	40 (13)
Do not recall	34 (11)
Missing	29 (16)
Screen Status
Yes	154 (49)
No	121 (38)
Don't Know	21 (6.6)
Missing	22 (7.0)

All 316 survey respondents were Vietnamese American, a majority of whom were born in Vietnam (91%, n=288). Most indicated that they primarily spoke Vietnamese at home (82%, n=259). Participants were evenly divided by gender and had a median age of 41 years (inter-quartile range 29 to 54), and 60 percent (n=189) had lived less than half of their life in the United States. The majority of the respondents possessed at least a high school education (n=241, 76%). The majority of participants had a household income less than $40,000 (63%, n=199), and 38 percent (n=120) reported earning less than $20,000 per year. Though a majority (55%, n=173) were insured, 39 percent (n=122) of the participants did not carry health insurance. Forty-nine percent (n=154) indicated that they had been screened for hepatitis B. Forty-six percent (n=145) had not received any HBV vaccine shots, 12 percent (n=37) had received one shot, 9 percent (n=29) had received two shots, and only 13 percent (n=40) had received all three doses in the series. About 10 percent (n=31) of the survey respondents reported HBV infection.

Measures

Initial survey questions were adapted from reliable measures created for similar vaccine acceptability studies [40-42]. Three outcome variables were selected to reflect past HBV screening, HBV vaccination history, and intent to obtain future HBV vaccinations. HBV screening record was assessed through the question “Have you ever had a screening for the following illnesses?” including a row indicating “Hepatitis B” with options “Yes,” “No,” and “Don't Know.” For the purpose of this analysis, answers of “Don't Know” were treated as missing. To determine HBV vaccination history, participants indicated the number of HBV vaccine shots they had received to date. In this analysis, we examine characteristics of participants who have received at least one vaccine shot, compared to those who have not received any shots. Intention to vaccinate was determined through the survey item “On a scale of 0 (definitely not) to 10 (definitely so), please rank your likelihood of getting a HBV vaccination shot within the next year.” This measure was subsequently dichotomized, with scores of 0 to 4 indicating participants who do not intend vaccinate for HBV in the near future and scores of 5 to 10 indicating those who intend to vaccinate.

We examined factors at the community level, among providers and social networks, and at the individual level. These factors evaluated the influence on HBV screening and vaccination across a range of factors including sociodemographic, psychosocial, provider, access to care, and community issues. Sociodemographic factors examined included age (measured in years), gender (female or male), educational attainment (dichotomized to “some degree beyond high school” vs. “high school degree or less”), yearly household income (measured in $20K categories up to $100K or greater), and length of residence in the United States (dichotomized at the median to “less than 17 years in the US” or “greater than or equal to 17 years in the US”).

Access to care factors included insurance coverage and availability of funds to pay for medical treatment, availability of transport to medical services, and price elasticity of HBV screening and vaccination (that is, the amount participants were willing to pay for those services). Insurance coverage was dichotomized into “insured” (including private and public insurance) and “uninsured.” Availability of funds to pay for medical treatment was measured through the question “How often do you have sufficient funds to pay for treatment of illnesses?” and availability of transportation to medical services was measured by asking “How often do you have access to transportation to get to a healthcare provider, clinic, or hospital?” with answers given on a five-point Likert scale ranging from “Never” to “Always.” Both variables entered the statistical models as continuous variables. Price elasticity of screening and vaccination was measured using the question “If you have to pay for your next hepatitis B screening blood test/vaccination shot, how much are you willing to pay?” with answer categories starting at “$0/free/will not pay,” then increasing in $10 increments to “$31 or more.” These variables were also used as continuous covariates in our statistical analysis.

Peer influence on HBV screening and vaccination decision making was assessed through two items asking, “If people I know had it first and then recommended it to me, I would take the hepatitis B (screening blood test/vaccination shot),” with five-point Likert responses ranging from “Strongly Disagree” to “Strongly Agree.” Participants were also asked to assess the seriousness of chronic hepatitis B on a six-point Likert scale from “Not Serious at All” to “Extremely Serious.” Peer influence and hepatitis B seriousness were also entered into statistical models as continuous variables.

Using our understanding of unique community-level factors influencing health behaviors among Vietnamese Americans, three large scales were developed for the survey to assess psychosocial vaccination issues with this population. A 66-item questionnaire assessed individual attitudes about hepatitis B infection, screening, and vaccination. We constructed a subset of items (n=24) to measure community attitudes toward those same issues. We also extensively measured health information sources and patterns inclusive of rankings for community-based immunization locations that would be convenient in the future (n=23 items). Scale items were measured on a five-point Likert scale (1: Strongly Disagree to 5: Strongly Agree).

The questionnaire items were analyzed using exploratory factor analyses with identified resultant subscales further subjected to internal consistency analyses. Exploratory factor analyses of the three psychosocial subscales utilized the principal component extraction methodology with varimax rotation. Factors with eigenvalues larger than one were retained. Subscales were determined by items with factor loading greater than 0.5. Internal reliability of resulting subscales was estimated using Cronbach’s alpha; Cronbach’s alpha greater than 0.70 was considered acceptable reliability for each subscale [43]. Finally, subscale construct mean factor scores were computed as the average of component answers loading on the subscale. Three of the resulting subscales were chosen for inclusion for assessment as predictors of HBV vaccination and screening: “Perceived Social Approval,” “Belief in Myths,” and “Perceived Stigma."

Perceived Social Approval

This scale comprises four items assessing perceived approval of doctors and work colleagues toward HBV screening and vaccination.

Belief in Myths

This subscale is composed of six items measuring HBV disease misconceptions. Three items measured incorrect perceptions of the severity of HBV infections. Two items assessed the incorrect beliefs that HBV screening is harmful. One item assessed incorrect knowledge of HBV transmission.

Perceived Stigma

This five-item subscale assessed perceived disapproval and discrimination of those living with HBV by spiritual leaders, family, friends, and the community. Three items measured perceived stigma experienced by those living with HBV. One item assessed fear to disclose HBV infection status. The final item assessed general discouragement of HBV-related topics by spiritual and religious leaders.

Individual scale items for these subscales are described in Table 2, along with Cronbach’s alpha estimates of internal consistency. Factor loadings measuring the strength of the association between the measured item and the underlying subscale factor are also reported.

Table 2

Descriptive Statistics for Factor Scales, Factor Loadings, Alpha Reliability Estimates, and Subscale Items (n = 316).

Factor	Mean	SD	Min	Max	Factor Loading
Perceived Social Approval (α = 0.89, n=4 items)
I think my doctor would approve of my getting a hepatitis B vaccine shot in the next 6 months	3.58	0.877	1.00	5.00	0.823
I think my doctor would approve of my getting a hepatitis B screening test in the next 6 months	3.62	0.867	1.00	5.00	0.880
I think my work colleagues would approve of my getting a hepatitis B screening in the next 6 months	3.43	0.922	1.00	5.00	0.894
I think my work colleagues would approve of my getting a hepatitis B vaccine in the next 6 months	3.42	0.945	1.00	5.00	0.817
Belief in Myths (α = 0.75, n=6 items)
Getting a hepatitis B screening test seems risky	2.66	1.034	1.00	5.00	0.627
Hepatitis B will not impair liver function	2.57	1.046	1.00	5.00	0.708
Hepatitis B is caused by smoking cigarettes	2.77	0.948	1.00	5.00	0.578
Acute hepatitis B will not lead to chronic hepatitis B	2.86	0.913	1.00	5.00	0.681
Hepatitis B cannot be treated	2.72	0.985	1.00	5.00	0.514
Hepatitis B blood test can deplete the body of energy	2.77	1.037	1.00	5.00	0.684
Perceived Stigma (α = 0.76, n=5 items)
People avoid those who have hepatitis B	2.73	1.005	1.00	5.00	0.606
My religious or spiritual leaders will discourage anything related to hepatitis B	2.86	1.051	1.00	5.00	0.599
My family will not talk to me if I have hepatitis B	2.33	1.042	1.00	5.00	0.794
My friends will avoid me if I have hepatitis B	2.36	0.995	1.00	5.00	0.763
If I have hepatitis B, I am afraid to tell anyone	2.58	1.055	1.00	5.00	0.602

Statistical Analyses

IBM SPSS Statistics for Windows (Version 22) was used for statistical analyses (IBM Corp., Armonk, NY). Descriptive statistics and cross-tabulations were generated for variables of interest. Bivariable correlations were also generated to explore key relationships.

The independent contributions of demographic, socioenvironmental, and psychosocial factors associated with each dichotomous outcome of interest (e.g., HBV screening, previous receipt of HBV vaccine doses, and future HBV vaccination) were assessed using bivariable and multivariable logistic regression models. Multivariable logistic regression models the odds of an individual’s outcome as the product of a baseline odds and the individual’s predictor variable terms. These predictor variable terms are composed of the odds ratio for that predictor, taken to power of the value of the predictor for that individual. Bivariable odds ratios were computed using separate simple logistic regression models with a single dichotomous outcome and a single predictor variable (either dichotomous or continuous, as indicated in the discussion of measures).

Adjusted odds ratios were estimated from multivariable logistic regression models, each estimating the simultaneous independent effects of all multilevel predictor variables on the odds of a single outcome variable. Models for HBV screening were fit using the full study population. HBV positive participants were excluded from the analysis of factors associated with prior HBV vaccination. HBV-positive participants and participants who had completed all three HBV immunization shots were excluded from the analysis of factors associated with intention to vaccinate for HBV within the next year. Independent predictors were considered significant if the associated p-value was less than 0.05. All multivariable logistic regression models were assessed for multicollinearity using variance inflation factors.

Though overall item completion was good, about 85 percent for each multivariable model, many respondents were missing at least one model item. In both the bivariable and multivariable logistic regression models, these missing items were accounted for by using pooled estimates from multiply imputed datasets. Multiple imputation has shown to be an effective technique for improving analytic validity when listwise deletion of incomplete cases would omit a large proportion of cases and missingness is likely related to observed variables [44]. Prior to imputation, missing data patterns were inspected for structure. Imputation datasets were constructed separately for each of the three outcomes, utilizing all variables in the respective multivariable models. Following the recommendations of White and colleagues [45], 100 imputed datasets were constructed for each outcome.

Results

Factors Associated with HBV Screening

We assessed factors associated with past hepatitis B screening using bivariable and multivariable logistic regression models fit with pooled estimates through multiple imputation. Of the 316 participants, 43 did not know or did not provide a response on whether they recall obtaining previous HBV screening (14 percent). These missing outcomes were estimated together with the missing independent variables during the multiple imputation. Overall, only 14 percent of outcome and independent variable items were missing, yet 73 percent of cases were missing at least one item. Inspection of patterns within the missing data revealed no problematic structures.

The bivariable analysis revealed that persons who recalled HBV screening also perceived the seriousness of chronic hepatitis B infection. Among these cases, each one-point increase on the perceived seriousness Likert scale was associated with 28 percent increase in the odds of screening (OR=1.28, 95% CI [1.06,1.55]) over the baseline odds of 0.55 (95% CI [0.28,1.07]) for participants with perceived seriousness score of 0 (“Not Serious at All”) (Table 3).

Table 3

Bivariable and multivariable associations with hepatitis B screening (n = 316). Pooled estimates using multiple imputation.

		Bivariable			Multivariable
Factors	% Miss.	OR	95% CI	p	OR	95% CI	p
Gender (ref = male)	3	.077	(0.48,1.25)	0.298	0.87	(0.49,1.55)	0.633
College education (ref = high school or less)	3	1.60	(0.98,2.61)	0.058	1.93	(0.99,3.75)	0.054
Age	9	0.99	(0.98,1.01)	0.436	1.01	(0.98,1.03)	0.528
≥ 17 Years in US	11	0.63	(0.39,1.04)	0.068	0.43	(0.23,0.83)	0.012
Household income	8	1.06	(0.87,1.28)	0.588	0.99	(0.77,1.27)	0.948
Insurance status (ref = no insurance)	7	1.42	(0.87,2.30)	0.162	0.88	(0.44,1.72)	0.700
Access to transportation to health care services	16	1.13	(0.94,1.36)	0.200	1.02	(0.80,1.30)	0.866
Ability to pay for medical treatment	27	1.19	(0.99,1.43)	0.068	1.22	(0.96,1.56)	0.102
Price elasticity of screening	35	1.22	(0.98,1.52)	0.076	1.24	(0.97,1.60)	0.085
HBV+ family member in household	21	1.93	(0.92,4.05)	0.081	2.26	(0.96,5.32)	0.061
Perceived seriousness of chronic hepatitis B	16	1.28	(1.06,1.55)	0.012	1.14	(0.91,1.44)	0.263
Influence of peer recommendation to screen	11	1.38	(1.10,1.73)	0.006	1.27	(0.95,1.69)	0.103
Perceived social approval of vaccination and screening	13	1.69	(1.21,2.38)	0.002	1.39	(0.94,2.06)	0.096
Belief in HBV myths	12	0.67	(0.46,0.99)	0.046	0.75	(0.44,1.30)	0.307
HBV stigma	11	0.84	(0.58,1.21)	0.351	1.07	(0.64,1.79)	0.808

Perceived social approval of vaccination and screening outcomes was also statistically associated with previous HBV screening (OR=1.69, 95% CI [1.21,2.38]). Among this group of individuals, each one-point increase in the average score across items on the “Perceived Social Approval” subscale was associated with a 69 percent increase in screening odds over the value of 0.94 at baseline (95% CI [0.71,1.25]), reflecting a “Neutral/No Opinion” stance (value of 3 of response options).

HBV screening history was also associated with an ascribed lack of belief in HBV myths (OR=0.67, 95% CI [0.46,0.99]). We found that each one-point decrease in the average score on the “Belief in HBV Myths” subscale was associated with a 33 percent reduction in the odds of prior screening, with baseline screening odds of 1.09 (95% CI [0.84,1.41]) for participants with a subscale score of 3 (“Neutral/No Opinion”).

Persons who screened for HBV infection also reported the influence of peers on the screening behavior (OR=1.38, 95% CI [1.10,1.73]). Among these persons, we found a 38 percent increase in odds of prior screening for each one-point increase in Likert score for peer influence. The baseline prior screening odds was 0.97 (95% CI [0.72,1.30]) for participants with score of 3 (“Neutral/No Opinion”) on the response options ranking the influence of peer recommendation to screen for HBV infection. Other variables did not have statistically significant bivariable associations with prior screening, including gender, education, age, time in the United States, household income, insurance status, access to transportation to health care, ability to pay for health care, price elasticity of screening, living with family members who had HBV illness, and perceived HBV stigma.

The multivariable logistic regression model for HBV screening was significant with Wald chi square p-value < 0.0001. Variance inflation factors for all independent variables were less than 2, indicating that multicollinearity was not adversely affecting the model. Seventeen years or more of residence in the United States corresponded to a statistically significant 57 percent decrease in adjusted odds of previous screening behavior compared to individuals who had spent less time in the United States (OR=0.43, 95% CI [0.22,0.85]) (Table 3). Perceived seriousness of chronic hepatitis B, influence of peer recommendation to screen, perceived social approval of vaccination and screening, and belief in HBV myths were not significant in the multivariable model, nor were the other incorporated cofactors.

We estimated the baseline odds of prior screening from the multivariable model as 0.27 (95% CI [0.09,0.81]). Together with the adjusted odds ratio estimates from the multivariable model, the baseline odds provides a basis for estimating the odds of prior screening for an individual. Baseline characteristics were defined as male, did not hold a high school degree, had annual household income less than $20K per year, were 41 years old, lived in the United States less than 17 years, and did not have insurance. We also included those who indicated “Never” as a challenge for access to transportation to health care facilities and had the ability to pay for health care. Other characteristic variables included those who indicated “$0/free/will not pay” for willingness to pay for a screening, those who did not live with a family member who had HBV, those who marked “Not Serious at All” when asked about the seriousness of chronic hepatitis B, those who indicated “Neutral/No Opinion” for potential influence of peer recommendation to screen, and those who had subscale scores corresponding to an average item score of “Neutral/No Opinion” for the three psychometric subscales.

Factors Associated with Prior Receipt of HBV Vaccine

In order to understand motivational correlates associated with immunization initiation, we examined factors associated with the receipt of at least one HBV dose (37%, n=106) among participants who were not living with HBV. Of the 285 participants who met this criteria, 63 did not know or did not report prior vaccination (22 percent). These missing values were estimated in the multiple imputation, along with missing independent subscale factor data. Though only 14 percent of the outcome and independent items were missing overall, 77 percent of cases were missing at least one item. Inspection of missingness structure revealed no problematic patterns within the data.

Bivariable analysis revealed that increased likelihood of previous HBV vaccination was associated with several factors (Table 4). The odds of HBV vaccination initiation decreased by 4 percent for each year of age (OR=0.96, 95% CI [0.94,0.98]). The odds ratio for comparing odds of vaccination for participants with an age difference of more than 1 year was computed by taking the odds ratio for 1 year to power of the desired age increase. Thus, a 20-year increase in age corresponds to a 56 percent decrease in the odds of prior vaccination (OR=0.42, 95% CI [0.27,0.64]). Estimated baseline odds of prior vaccination was 0.64 (95% CI [0.48,0.86]) for 41-year-old participants (the median age of our sample).

Table 4

Bivariable and multivariable associations with hepatitis B vaccination and intent to vaccinate. Pooled estimates using multiple imputation.

	Started Vaccination Series (n=285)							Intent to Vaccinate (n=2530)
	Bivariable				Multivariable			Bivariable				Multivariable
Factors	% Miss.	OR	95% CI	p	OR	95% CI	p	% Miss.	OR	95% CI	p	OR	95% CI	p
Gender (ref = male)	4	.074	(0.43,1.27)	0.276	0.66	(0.34,1.31)	0.233	4	1.29	(0.71,2.33)	0.408	1.24	(0.60,2.57)	0.553
College education (ref = high school or less)	4	1.01	(0.60,1.72)	0.969	0.85	(0.42,1.75)	0.664	4	0.98	(0.54,1.78)	0.958	1.00	(0.45,2.21)	0.989
Age	10	0.96	(0.94,0.98)	0.000	0.95	(0.93,0.98)	0.000	11	1.01	(0.99,1.04)	0.215	1.02	(0.99,1.05)	0.124
≥ 17 Years in US	11	0.54	(0.31,0.92)	0.025	0.65	(0.31,1.36)	0.254	12	0.87	(0.49,1.54)	0.624	0.63	(0.29,1.36)	0.238
Household income	10	1.32	(0.83,1.20)	0.972	0.95	(0.73,1.23)	0.688	10	0.88	(0.70,1.09)	0.228	0.88	(0.65,1.20)	0.420
Insurance status (ref = no insurance)	7	1.32	(0.77,2.27)	0.313	1.08	(0.50,1.34)	0.843	7	0.92	(0.52,1.62)	0.762	1.04	(0.45,2.40)	0.935
Access to transportation to health care services	15	1.10	(0.88,1.38)	0.393	1.05	(0.78,1.40)	0.760	16	1.42	(1.07,1.87)	0.015	1.59	(1.12,2.25)	0.009
Ability to pay for medical treatment	28	1.23	(1.01,1.50)	0.036	1.22	(0.92,1.62)	0.162	27	0.85	(0.68,1.06)	0.152	0.80	(0.59,1.09)	0.157
Price elasticity of vaccination	39	1.06	(0.84,1.33)	0.640	1.01	(0.75,1.35)	0.955	38	1.06	(0.82,1.37)	0.640	1.11	(0.82,1.52)	0.496
HBV+ family member in household	22	1.58	(0.68,3.63)	0.286	1.84	(0.67,5.06)	0.235	23	1.15	(0.48,2.74)	0.745	1.33	(0.48,3.73)	0.592
Perceived seriousness of chronic hepatitis B	17	1.19	(0.96,1.48)	0.113	1.00	(0.76,1.32)	0.979	18	1.35	(1.01,1.80)	0.040	1.18	(0.85,1.65)	0.324
Influence of peer rec. to vaccinate	11	1.45	(1.10,1.90)	0.007	1.30	(0.92,1.83)	0.131	13	1.64	(1.15,2.33)	0.006	1.51	(1.03,2.21)	0.034
Perceived social approval	13	1.89	(1.27,2.81)	0.002	1.82	(1.10,3.01)	0.020	13	1.77	(1.13,2.78)	0.013	1.55	(0.91,2.65)	0.108
Belief in HBV myths	12	0.55	(0.35,0.86)	0.010	0.58	(0.32,1.08)	0.085	13	0.91	(0.58,1.43)	0.672	1.46	(0.75,2.85)	0.271
HBV stigma	11	0.76	(0.51,1.13)	0.173	0.97	(0.54,1.73)	0.906	12	0.81	(0.52,1.27)	0.351	0.73	(0.39,1.36)	0.318

Individuals who had spent 17 years or more in the United States were also less likely to have had at least one HBV vaccination shot (OR=0.54, 95% CI [0.31,0.92]), with a 46 percent reduction in vaccination odds compared to individuals who had been in the United States for less than 17 years. Baseline odds for those who had been in the United States for less than 17 years was 0.87 (95% CI [0.61,1.24]).

Positive perception of social approval for vaccination was associated with increased likelihood of prior vaccination (OR=1.89, 95% CI [1.27,2.81]). Each point increase in the average response scores for the “Perceived Social Approval” subscale items increased the odds of having at least one HBV shot by 89 percent over a baseline odds value of 0.48 (95% CI [0.33,0.68]) for participants with a “Perceived Social Approval” score of 3 (“Neutral/No Opinion”). We also found that each one-point increase of measured influence of peer recommendation to vaccinate increased the odds of prior vaccination by 45 percent (OR=1.45, 95% CI [1.10,1.90]) over a baseline odds of 0.50 (95% CI [0.34,0.73]) for participants who selected a “Neutral/No Opinion” response (value of 3). Reduced belief in myths was also significantly associated with prior vaccination, with each point decrease in the average response option associated with a “Belief in Myths” corresponding to a 45 percent decrease in the prior vaccination odds (OR=0.55, 95% CI [0.35,0.86]), with baseline odds of 0.56 (95% CI [0.41,0.77]) among those with “Neutral/No Opinion” responses.

Finally, each point increase in the ability to pay for medical treatment scale increased the odds of vaccine receipt by 23 percent (OR=1.23, 95% CI [1.01,1.50]). The estimated baseline odds of prior vaccination was 0.42 (95% CI [0.25,0.70]) for participants marking their ability to pay for medical treatment as 0 (“Never”). Factors not significantly associated with vaccine receipt included gender, education, household income, insurance status, transportation access for health care service utilization, perceived severity of HBV, perceived stigma, presence of family members living with HBV, and price elasticity (how much respondents are willing to pay for vaccines).

The multivariable logistic regression model for prior HBV vaccination displayed overall statistical significance (all 100 imputed data sets displayed overall significance with p<0.0001). No multicollinearity issues were evident in the model: all independent variables had variance inflation factors less than 2. Prior HBV vaccination was significantly associated with reduced age (OR=0.95, 95% CI [0.93,0.98]). Among these cases, each additional year of age was associated with a 5 percent decrease in the odds of vaccine receipt, after adjustment for other variables in the model. Using a 20-year difference in age, a 60 percent decrease in the adjusted odds was observed on vaccine receipt.

We observed that “Perceived Social Approval” corresponded to an 82 percent increase in the adjusted odds of vaccine receipt with adjustment for other covariates (OR=1.82, 95% CI [1.10,3.01]). Length of U.S. residency, ability to pay for medical treatment and provider visits, belief in HBV myths, and influence of peer recommendation to vaccinate became non-significant in the multivariable model. Baseline odds of vaccine receipt were estimated from the multivariable logistic model as 0.27 (95% CI [0.07,1.02]), with baseline defined as in the multivariable logistic model for HBV screening.

Factors Associated with Intent to Receive HBV Vaccine

To understand facilitators and barriers to future vaccination, we also examined factors associated with intent to vaccinate for HBV within the next year among participants who have not completed all three doses of HBV vaccine and were not HBV positive (n=253). Of the 253 participants who were not HBV positive but had not completed all three doses of the HBV vaccine, only 75 (30 percent) indicated positive intention to seek vaccination in the near future (5 or greater on the 10-point scale). Of the rest, 125 (49 percent) did not intend to vaccinate (4 or lower out of 10) and 53 (21 percent) did not report their intention to vaccinate. Among those who did report their intention to vaccinate within the next year, the average intention score was 3.39 (standard deviation = 3.89) on a 10-point scale from “Definitely Not” at 0 to “Definitely So” at 10. Thus, on average, HBV-negative individuals who had not completed their vaccination schedules expressed low intention to start or continue vaccination.

Missing outcomes were imputed together with missing independent variables during multiple imputation. Overall, 74 percent of cases were missing at least one item, but only 15 percent of item responses were missing. Analysis of patterns within the missing data did not detect any problematic structures.

Intention to receive HBV vaccination in the next year among those participants who had not received all three vaccination shots was significantly associated in bivariable analysis with several assessed factors (Table 4). Increased reported influence of peer recommendation to vaccinate was associated with an increased intention to vaccinate for HBV (OR=1.64, 95% CI [1.15,2.33]). Each point increase in indicated peer recommendation corresponded to a 64 percent increase in the odds of intending to vaccinate in the future, utilizing a baseline odds of 0.41 (95% CI [0.26,0.66]) (“Neutral/No Opinion” values of 3).

Access to transportation for utilization of heath care services was also a significant predictor of intention to vaccinate (OR=1.42, 95% CI [1.07,1.87]). We observed that each point increase in the given response corresponded to a 42 percent increase in odds of intending to vaccinate. The estimated baseline odds of intention to vaccinate was 0.20 (95% CI [0.07,0.55]) for those indicating response option of 0 (“Never”) when asked how often they had access to transportation to health care services.

Intention to obtain HBV vaccination was associated with positive perceived social approval of screening and vaccination (OR=1.77, 95% CI [1.13,2.78]). Each point increase in the “Perceived Social Approval” score increased the odds of favorable vaccine receipt intent by 77 percent over the baseline odds of 0.45 (95% CI [0.30,0.67]) for those stating “Neutral/No Opinion” on the “Perceived Social Approval” subscale.

Participants who perceived chronic hepatitis B infection as serious were also more likely to intend to seek vaccination in the future (OR=1.35, 95% CI [1.01,1.80]). Each point increase in perceived HBV seriousness was associated with a 35 percent increase in the odds of intending to vaccinate in the near future, with baseline odds of 0.22 (95% CI [0.07,0.64]) for participants with perceived seriousness score of 0 (“Not Serious at All”). Other variables did not have statistically significant bivariable associations with intention to vaccinate for HBV.

The overall multivariable logistic regression model for vaccination intention was significant. All 100 imputed datasets presented significant Wald chi square test, with p-values less than 0.0001. Multicollinearity assessment found no variance inflation issues, as all variance inflation factors were less than 2. The model revealed that reported influence of peer recommendation to vaccinate (OR=1.51, 95% CI [1.03,2.20]) and access to transportation to health care services (OR=1.59, 95% CI [1.12,2.25]) were significantly associated with intent to vaccinate (Table 4). Each point increase on response options for reported influence of peer recommendation to vaccinate was associated with a 51 percent increase in the adjusted odds of reporting intention to vaccinate in the near future, while each point increase in the scale for access to transportation to health care services was associated with a 59 percent increase in those adjusted odds. Perceived seriousness of chronic hepatitis B and perceived social approval of screening and vaccination were not significant in the multivariable model. The baseline odds of intention to vaccinate in the near future was estimated for the multivariable logistic model as 0.07 (95% CI [0.01,0.39]), with baseline variable values defined as in the multivariable logistic model for prior screening.

Discussion

The CDC recently released an action plan for the prevention, care, and treatment of viral hepatitis [46]. Those with chronic HBV infections have higher risk of developing liver cancer, and approximately 65 percent remain asymptomatic and unaware of their infection status [46]. Despite the fact that Vietnamese Americans have the highest rate of liver cancer out of all racial and ethnic groups [3,22,35,47-49], our study found that less than 50 percent had received any HBV vaccine doses and only about 13 percent had received the full three-dose series. This is consistent with the literature, as reported vaccination rates among AAPI in the United States vary from low estimates of 10 percent to 18 percent to high estimates of 24 percent to 38 percent [15,25-27,33,36,37,50].

Previous studies have found HBV screening and vaccination to be influenced by several factors, including demographics (e.g., age, family history of HBV), consistency of care, knowledge, perceived health beliefs (e.g. “people die from HBV”), physician recommendation, and suggestion by friends or family [14,19,30,34,51]. Our study found that younger Vietnamese Americans and those who have been in the United States for less than 17 years are more likely to be vaccinated. Our study also indicated that Vietnamese Americans who have been in the United States less than 17 years are more likely to have been screened for HBV. This is a reversal of trends found in previous studies, which suggested that younger immigrants may not have been exposed to World Health Organization and local efforts in Vietnam to monitor hepatitis B infection and routine vaccination, which peaked in the late 1990s [18,52].

Studies have found that at least one-third of the Vietnamese-American population has incorrect knowledge and beliefs about HBV [15,16,18,20,30,32]. Our findings indicate that those who believe in HBV-related myths and misconceptions are less likely to be vaccinated. Misconceptions about HBV infection, screening, or vaccination may impede attitudes and behaviors related to HBV prevention. Common misconceptions include belief that HBV screening tests can be harmful or that hepatitis B is not treatable, inadequate knowledge of the detrimental health effects of hepatitis B (such as liver damage), and belief that hepatitis B is caused by smoking tobacco. Efforts are needed to correct misconceptions, raise awareness of the severity of HBV, and promote HBV screening and vaccination among the Vietnamese-American community. Our study suggests that HBV awareness can translate directly into screening and vaccination, as participants who perceived chronic hepatitis B as more serious were more likely to have been screened for the disease and more likely to report intention to vaccinate.

We found that participants with the ability to pay for medical treatment and health care visits were more likely to be vaccinated. In addition, those with transportation to health care services were more likely to intend to vaccinate in the near future. Given the low income level of most study participants (less than $20,000) and the relatively high price of the HBV vaccination series, health care access and cost are serious barriers to vaccination and screening among this population. In addition to awareness-raising efforts among Vietnamese-American communities, ensuring the affordability of the vaccine is critical to improving vaccination uptake. Effective promotion of such provisions are also needed; prior studies have found that many Vietnamese were unaware of free HBV vaccinations in their communities [7,27-29,35].

Qualitative studies of migrants’ cultural identity suggest that migrants from various ethnic backgrounds experience a greater sense of ethnic identity and therefore identify with the collective rather than the individual [53]. Furthermore, evidence indicates that the strong sense of collectivism among Vietnamese-American women is positively associated with higher levels of self-efficacy with regard to screening behavior [54]. Therefore, social support from community members may have a significant impact on individual health decision-making behavior. Accordingly, among our study population, the majority of participants reported that they would be swayed by peer recommendations to screen or vaccinate and reported influence of peer recommendation to vaccinate was a significant factor in both prior vaccination behavior and future vaccination intent. That is, participants who reported being positively influenced by peer recommendations were more likely to follow (and intend to follow) good vaccination practice. Additionally, we found that perceived social approval of HBV vaccination and screening by doctors and family was positively associated with individual health behaviors. Peer and community-level efforts among the Vietnamese community have been effective in addressing cancer prevention [14], and our study indicates that similar approaches may be effective for HBV prevention as well.

Limitations

This study utilized self-reported measures, which we acknowledge carry the risk of recall bias. Because of the wide range of responses and the relatively low number of “Don't Know” answers, we feel that the effect of recall bias on this analysis is likely to be relatively small. Our treatment of “Don't Know” answers to outcome questions as missing could also potentially bias results if the relationships between outcomes and predictors were very different for those who could not recall their outcome status. The relatively low number of missing outcomes suggests that this bias in not likely to be large, and multiple imputation helped preserve relationships between the outcomes and predictors.

Actual prevalence of HBV infection is likely higher than reported, as studies indicate that 33 percent of chronically infected individuals are unaware of their status [12]. Because this study evaluated Vietnamese-American populations in the greater Atlanta region, results may be related to local environmental factors, so care must be exercised in generalizing to Vietnamese-American populations in other locations. The high proportion of cases with at least one missing covariate could potentially lead to biased results, but because the overall proportion of missing items was relatively small for each analysis, the use of multiple imputation allowed us to mitigate this issue.

Conclusion

Our study showed that despite recent CDC recommendations for culturally relevant HBV vaccination programs and education among at risk populations, the rate of HBV testing and vaccination are inadequate among the Vietnamese Americans living in Atlanta, Georgia. Only half of those surveyed had a HBV screen, and among those who reported HBV immunization, only 27 percent had completed the three-dose series. We identified sociocultural factors that might impact immunization uptake among Vietnamese Americans, such as misinformation, lack of HBV awareness, and the need for social acceptance of HBV vaccine. These identified factors generally agree with results of previous studies, though we found increased vaccination uptake among younger Vietnamese Americans, contrary to previous findings. Our findings suggest that a culturally congruent, targeted approach to reach these populations will prompt greater intention to engage in HBV screening and vaccination. Community partner organizations serving AAPIs may be best positioned to leverage their influence as trusted organizations and as message deliverers in HBV health promotion efforts. By enabling health care providers to deliver care in and through community organizations, they will be able to shape social and cultural norms and health promoting beliefs to increase routine HBV screening among this population and enhance efforts for subsequent referral to care.