Exploring the Behavioral Determinants of COVID-19 Vaccine Acceptance among an Urban Population in Bangladesh: Implications for Behavior Change Interventions

Study site and context

This Barrier Analysis study was conducted in different urban areas of Dhaka – the capital city in Bangladesh – from January 09-15 2021. Dhaka is the residence of more than 10.3 million people – 6.29 percent of total population of the country (34). Between 8 March 2020 and 19 April 2021, there were 736,074 COVID-19 confirmed cases with 10,781 reported death (4). While Dhaka makes up about 6.29% of the total population of Bangladesh, almost 71% of the COVID-19 cases in Bangladesh reported as of 19 April in Dhaka (61% in Dhaka city alone) (4) and which found as one of the largest hotspots of spreading COVID-19 (35).

Study tool

A Barrier Analysis (BA) study was conducted to better understand the behavioral determinants of COVID-19 vaccine hesitancy in Dhaka. BA is a research tool that was developed in 1990 by Davis (36). Based on HBM and TRA, BA studies explore respondents’ beliefs about a behavior. Sometimes certain beliefs about a behavior (e.g., possible COVID-19 vaccination side effects) are common in a population, but are not necessarily associated with vaccine acceptance. BA is meant to, identify the most likely determinants of that behavior(37). A key feature of BA is that responses from those accepting a behavior (‘Doers’ or ‘Acceptors’) are compared with those who are not (the ‘Non-doers’ or ‘Non-acceptors’) to identify behavioral determinants associated with a particular behavior (e.g., handwashing with soap, getting a vaccine) so that practitioners can develop more effective behavior change messages and activities. BA has been used in more than 40% of low-to-middle-income countries and used extensively by World Vision and other organizations during both the Ebola (38) and COVID-19 pandemics (39). The beliefs and other responses regarding behavioral determinants assessed during BA (see Box 1), are identified with a focus on the most actionable findings. The other details of BA approach can be found elsewhere (40–45). There are BA studies in the peer-reviewed literature on exclusive breastfeeding (46) HWWS among internally displaced women in the Kurdistan region of Iraq (40) timely oral polio vaccination agricultural extension behaviors in India (47), dietary salt reduction in Nepal (48) transition from the lactational amenorrhea method to other modern family planning methods in Bangladesh (36) and cervical cancer screening in Senegal (49).

Questionnaire development

This study modified the standardized Barrier Analysis questionnaire from the Designing for Behavior Change (DBC) training manual (45).

The BA questionnaire is divided into three parts. The first part includes a set of screening questions to identify whether the participant as either an ‘Acceptor’ or a ‘Non-acceptor’. In the screening section of the questionnaire, self-reported intention questions were used. (Please see Supplementary file 1.) In the second part of the questionnaire, information on the respondents’ age, gender, level of education and profession are obtained. In order to assess 11 of the usual 12 determinants assessed with BA (see Box 1), the third section consists of a set of open- and closed-ended questions (37,44). Based on input and a literature review from NGO experts who worked on increasing vaccine acceptance in the Ebola Vaccine Deployment and Compliance Project, several questions were also added to the standard BA questionnaire to explore respondent’s beliefs on the safety and effectiveness of COVID-19 vaccines, trust in vaccine information provided by both community and religious leaders, exposure to misinformation, and level of education. Based on local social media listening, one question was added on perceptions around herd immunity. The questionnaire was developed in English and then was translated into Bengali and back-translated into English to check for accuracy.

Sampling

We interviewed adult men and women for this BA study and selected them randomly through a convenience sampling strategy from five different areas of Dhaka – the capital city of Bangladesh. The Barrier Analysis approach recommends a minimum sample size of 45 Doers (Acceptors) and 45 Non-doers (Non-acceptors) in order to detect statistically-significant Odds Ratios of 3.0 or higher, an alpha error of 5%, and a power of 80% (37). The data collection team approached adult men and women until they reached 90 respondents (45 Acceptors and 45 Non-acceptors).

Data collection, management and analysis

We collected data between 9 to 13 January 2021 through individual interviews with responses recorded on paper-based questionnaires by two teams composed of one female and one male member. Male enumerators interviewed male respondents, and female enumerators interviewed female respondents. A research supervisor assured the quality of data. Following data collection, the data collection team and the lead author coded the open-ended responses thematically, using both an inductive and deductive coding process. At the end of this process, the team quantified the responses in each category for Acceptors and Non-acceptors separately. These categories and the number of responses registered for each were then entered into a standardized BA tabulation sheet that revealed whether differences in the proportion of Acceptors and Non-acceptors providing each response were statistically significant and should be addressed through the behavior change strategy. For each question and category of responses, the BA tabulation calculates the percentage of responses for both Acceptors and Non-Acceptors; the Odds Ratio, the Standard Error, and its confidence interval; the Estimated Relative Risk (ERR); and p-values (see supplementary file 2). This allows identify those differences between Acceptors and Non-acceptors that are statistically significant (at p<0.01) and to see the strength of the associations between each response and the behavior (based on the ERR).

Ethical considerations

We performed all procedures in studies in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study protocol was approved by the institutional Ethics Committee of the Chattogram Veterinary and Animal Sciences University, Bangladesh (permit ref. no. CVASU/Dir (R and E) EC/2020/169). We informed respondents about the study objectives, and obtained their written consent before conducted interview. The data collection activities have performed following the COVID-19 safety protocols in Bangladesh that enacted by Directorate General of Health Services in Bangladesh (50).

3.1 Perceived Social Norms

Some of the strongest predictors of vaccine acceptance in this population are beliefs around both injunctive and descriptive social norms, who the respondent thinks approve or disapproves of COVID-19 vaccination, and the proportion of people that they think will go for a COVID-19 vaccine when it is available. The results are shown in Table 1. Specifically, Acceptors were 3.2 times more likely to say they would be “very likely” to get a COVID-19 vaccine if a doctor or nurse approved (p<0.001), while Non-acceptors were 2.6 more likely to say it would be “not likely” that they would get a vaccine if a doctor or nurse recommended it (p<0.001). Acceptors were twice as likely to say that “most people” they know will get a vaccine (p<0.001), and 1.3 times more likely to say that “most close family and friends” will get a vaccine (p=0.003), while conversely, Non-acceptors were 3.5 times more likely to say that “very few people” they knew would get a vaccine (p<0.001) and 1.3 times more likely to say that “most of their close family and friends” would not get a COVID-19 vaccine (p=0.003). In terms of respondents’ impressions concerning who disapproves of their getting a COVID-19 vaccine, Acceptors were 1.7 times more likely (than Non-acceptors) to say that “no one” would disapprove (p<0.001). Non-acceptors were 1.5 times more likely to say that “my mother” (p<0.004), 1.4 times more likely to say “elderly people” (p=0.009), and 1.7 times more likely to say “people who will not get the vaccine” (p=0.006) would disapprove of their getting a COVID-19 vaccine. In addition, Acceptors were and more likely to say their mother would approve of their getting a COVID-19 vaccine (p<0.001). With regards to community and religious leaders’ influence on the decision to get a COVID-19 vaccine, Acceptors were 1.3 times more likely to say that “most community and religious leaders” would want them to get a vaccine (p=0.007), while Non-acceptors were 1.5 times more likely to say that most community leaders and religious leaders would not want them to get a COVID-19 vaccine (p=0.002).

3.2 Perceived Safety of COVID-19 Vaccines

As in many places in the world, concerns about safety of COVID-19 vaccines are affecting intention to accept vaccination in this population. When asked how safe the COVID-19 vaccines are, Non-acceptors were 1.8 times more likely to say that COVID-19 vaccines are “not safe at all” (p<0.001) while Acceptors were 1.4 times more likely to say that COVID-19 vaccines are “mostly safe” (p<0.001) (Table 2).

3.3 Perceived Risk / Susceptibility (to COVID-19)

Perceived risk of getting COVID-19 – and the level of emotional concern about getting COVID-19 – also appears to be driving intended vaccine acceptance in Dhaka. Acceptors were 1.4 times more likely to say they it was “very likely” that someone in their household would get COVID-19 over the next 3 months (p<0.001) while Non-acceptors were 1.3 times more likely to say that was only “somewhat likely” (p=0.005). When respondents were asked how concerned they were about getting COVID-19 (or someone in their household getting COVID-19), Acceptors were 1.3 times more likely to say that they were “very concerned” (p=0.002) while Non-acceptors were 1.7 times more likely to say that they were only “a little concerned” (p<0.001). Additionally, Non-acceptors were 1.7 times more likely to say that “very few people” have had COVID-19 in their communities (p<0.001) (Table 2).

3.4 Perceived Self-efficacy

The respondents were asked two open-ended questions to understand what would make it easier or more difficult to get a COVID-19 vaccine. From the results (Table 3), factors related to how and where the vaccine would be given affects intention to vaccinate, including whether or not proper COVID-19 social distancing and prevention measures are maintained during vaccination.

Concerning what made it easier, Acceptors were 1.4 times more likely to say “if vaccination is provided by government health care facilities” (p=0.000), 1.2 times more likely to say “school-based vaccination” (p=0.003), 1.3 times more likely to say “if the vaccines are provided by establishing kiosks” (p=0.001) and 1.2 times more likely to say “if proper health & safety (COVID-19) protocols are maintained while giving the vaccine” (p=0.006) all would make it easier to get a COVID-19 vaccine. Conversely – and interestingly – Non-acceptors were 1.2 times more likely to say “when vaccines are given at home” (p=0.010), and 1.7 times more likely to say “when the vaccine has no side effects” (p=0.002) would make it easier for them to get a COVID-19 vaccine.

When asked what would make it more difficult to get a COVID-19 vaccine, Acceptors were 1.3 times more likely to say either “if there are no health measures in the vaccination center due to overcrowding” or “risk of getting infected with COVID-19 while vaccinating” (p=0.001), and 1.2 times more likely to say “if the vaccinator does not follow proper COVID-19 preventive measures” (p=0.005) would make it difficult to get a COVID-19 vaccine. Non-acceptors were 2.4 times more likely to say “when the vaccine has severe side-effects” (p<0.001) and 2 time more likely to say “if there is bribery” (p=0.004) would make it difficult to get a COVID-19 vaccine.

3.5 Perceived Positive Consequences and Perceived Negative Consequences

Respondents were also asked what the positive and negative consequences (e.g., advantages and disadvantages) would be of getting a COVID-19 vaccine. While reducing the reduced risk of infection was important, Acceptors were more likely to point out benefits related to livelihood and economic benefits and life getting back to normal. Specifically (Table 4), Acceptors were 1.3 times more likely to say that reducing the risk of COVID-19 infection (p=0.003), 1.3 times more likely to say “we can attend social and cultural activities” (p=0.003), 1.2 times more likely to say “children can start school again” (p=0.003), 1.3 times more likely to say “reduction in COVID-19 related costs” (e.g. masks, hand sanitizer, tests; p<0.001), 1.3 times more likely to say “employment and income opportunities will be increased” (p<0.001), and 1.3 times more likely to say “attending prayers in congregation” (p<0.002) were all reported advantages of getting a COVID-19 vaccine.

When asked about the negative consequences (disadvantages), Acceptors were 1.3 times more likely to mention mild side effects of vaccination such as “weakness” (p<0.002) and 1.3 times more likely to say “itching and skin problems” (p<0.001) were disadvantages of getting a COVID-19 vaccine. Meanwhile, Non-acceptors were 1.6 times more likely (than Acceptors) to mention life-threatening side effects (p<0.001), 1.5 times more likely to say “unknown/new diseases” (p<0.004), and 1.9 times more likely to say “infertility” (p<0.001) were all disadvantages of getting a COVID-19 vaccine.

3.6 Perceived Action Efficacy

Counter-intuitively, Acceptors were 1.3 times more likely to say they were “somewhat likely” to get COVID-19 once they were vaccinated against it (p<0.001) while Non-acceptors were 2.8 times more likely to say that they were “not likely at all” to get COVID-19 once one was vaccinated against it (p<0.001) (Table 5). Focus group discussions may be used at a later point in time to explore this finding. Related to perceived action efficacy, respondents were asked if they agreed or disagreed with the statement “Most people will eventually get infected with COVID-19, so getting a COVID-19 vaccine is unnecessary.” Non-acceptors were 1.7 times more likely to say that they “agree a little” or “agree a lot” with the statement (p<0.001) while Acceptors were 1.5 times more likely to “disagree a lot” with the statement (p<0.001).

3.7 Trust in COVID-19 vaccines

As expected, trust in COVID-19 vaccines is a key driver of intended vaccine acceptance in Dhaka. Acceptors were twice as likely to say that they trust the COVID-19 vaccines “a lot” or a “moderate amount” (p<0.001). Conversely, Non-acceptors were 1.7 times more likely to say that they “trust them a little” (p<0.001) and 2.3 times more likely to say that they “do not trust [COVID-19 vaccines] at all” (p=0.001) (Table 6).

3.8 Perceived Severity (of COVID-19)

The perceived severity of COVID-19 was also highly correlated with intended vaccine acceptance. Respondents were asked how serious it would be if they or someone else in their household got COVID-19. Acceptors were 1.3 times more likely (than Non-acceptors) to say that it would be “very serious” (p=0.001) (Table 6).

3.9 (Perceived) Access

Perceived difficultly in reaching clinics that normally provide vaccines is associated with intended COVID-19 vaccine acceptance in Dhaka. Non-acceptors were 1.6 times more likely to say that it was “very difficult” to get to the facility that normally provides vaccines (p=0.000), while Acceptors were 1.3 times more likely to say it was “somewhat difficult” to get to that facility (p=0.000) (Table 6).

3.10 Perceived Divine Will

Personal agency and religious beliefs often come into play with vaccine acceptance. In this study (Table 7), we asked for respondents’ degree of agreement or disagreement with the statement: “Whether I get COVID-19 or not is purely a matter of God’s will or chance -- The actions I take will have little bearing on whether or not I get COVID-19.” Personal agency was correlated with intended vaccine acceptance: Acceptors were 1.2 times more likely to say that they “disagree a lot” (p=0.005) that “whether I get COVID-19 or not is purely a matter of God’s will or chance” (p<0.001). Respondents were also asked whether they believed that Allah (or God or the gods) approves or disapproves of people getting COVID-19 vaccines. While 80% of Acceptors and 78% of Non-acceptors said that a deity approves, there were no statistically significant differences between Acceptors and Non-acceptors for this question.

3.11 Rumors/ Culture

Respondents were asked if there were any cultural or religious reasons that they would not get a COVID-19 vaccine (results are in Table 7). Acceptors were 1.2 times more likely to say that there were no cultural or religious reasons they would not get a COVID-19 vaccine (p=0.01), while Non-acceptors were 1.3 more likely to say that there were reasons (p=0.002). When asked what those reasons were, Non-Acceptors were 1.6 more likely to say that they had heard that ‘the vaccines were made with pork fat which is not allowed (haram) by Islam’ (p=0.001) and 1.4 times more likely to say that ‘vaccines were made with haram ingredients’ (p=0.003). Respondents were also asked if they had seen or heard of anything that would stop them or others from seeking to get a COVID-19 vaccine. If they said yes, they were asked a follow-up question on what would stop them or others from seeking the vaccine. Regarding that question, Acceptors were 1.3 more likely to say that producers’ “hide and seek activities” related to vaccine accuracy in the clinical testing would stop them or their peers from getting a COVID-19 vaccine (p<0.002).

Implications for Behavior Change Interventions

Aligned with WHO’s Technical Advisory Group on Behavioral Insights and Sciences for Health recommendation on social and behavioral drivers on COVID-19 vaccination (33), our study identified important beliefs and responses associated with different determinants of COVID-19 vaccine acceptance among urban population in Bangladesh which could be valuable to informing contextualized behavioral intervention and engagement strategies to support COVID-19 vaccination. For example, to increase perceived positive social norms, especially for those who found to be important influencers of this behavior (e.g., medical staff and mothers), videotaping individuals giving testimonials in each neighborhood on why they plan to get the vaccine, and distributing them over media is one possible approach. Other activities can be used to make acceptance more visible (e.g., stickers on households that say, “We plan to vaccinate!” or lapel pins with the same message). To increase the perception that COVID-19 is serious (to address perceived severity), testimonials by people who have lost or almost lost family members due to COVID-19 disease could be used. To address perceived divine will, religious leaders can be assisted in creating sermon outlines on maintaining one’s health (and linking that with COVID-19 vaccines), and supported in creating radio spots to promote COVID-19 vaccines. In addition to the COVID-19 prevention, other positive consequences of immunization mentioned by Acceptors should be disseminated. While not repeating any misinformation, it will be important to provide clear information on the known minor risks of COVID-19 vaccination as a way to combat misinformation on side effects that were mentioned more often by Non-acceptors (e.g., life-threatening conditions, new diseases, infertility). Information on how vaccines are made should be disseminated to counter misinformation (e.g., that vaccines are made with pork fat or other haram ingredients). Stakeholders should also take into account the findings on things that may make vaccination easier for people, such as providing the vaccine in schools and kiosks (in addition to government health facilities) and to assure that the population understands that proper COVID-19 health and safety protocols will be maintained in places where vaccines are given.