Ethics statement

The study was approved by ethical committees of local host institutes including the National Hospital for Tropical Diseases (Hanoi, Vietnam), the Hospital for Tropical Diseases (Ho Chi Minh City, Vietnam), CDC Daklak (Vietnam), Nepal Health Research Council (Kathmandu, Nepal), Patan Hospital (Lalitpur, Nepal), Faculty of Medicine, University of Indonesia (Jakarta, Indonesia), and the Oxford Tropical Research Ethics Committee (Oxford, UK).

Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included as supporting information (S1 Checklist).

Study settings

The study teams were located at Oxford University Clinical Research Unit (OUCRU) sites in Indonesia, Nepal and Vietnam. We included a total of 13 study locations across these three countries that were part of the SPEAR study. In Indonesia, sites included the urban settings of Jakarta and Bandung and West Sumba, a rural site. In Nepal, five distinct locations spanning the country’s diverse geographical regions were selected, representing both rural areas (Mustang, Sindhupalchwok districts) which are less densely populated, and urban areas (Bhaktapur, Morang, Patan) with high population density. In Vietnam, about half of the participants were selected from two major metropolis, Hanoi and Ho Chi Minh City as well as rural and mountainous regions, including Nam Dinh and Dak Lak provinces.

In each setting, the nature of the pandemic varied greatly over time, including the case load, the stringency of the public health responses, the nature of media sources, among others. For example, in Jakarta and Bandung, there were higher cases of COVID-19 throughout the pandemic compared with the island of Sumba in Nusa Tenggara Timur [19]. At the start of the pandemic, Vietnam initiated a zero COVID policy and closed borders rapidly, enacting a more stringent response at that time [20–22]. In Nepal, there was an extended nationwide lock down period that was longer than in other settings [23]. Each country also had different dates of vaccine rollout and access to different vaccines at different times, e.g., Indonesia started nation-wide vaccination in mid-January 2021 [24], Nepal started at the end of January 2021, and Vietnam started vaccination in early March of 2021 [11,25].

Sampling

We used purposive sampling, based on location, gender, profession, vaccine status, education and age to select participants for in-depth interviews. In Indonesia, we also selected participants based on responses from a survey component of SPEAR study. We contacted key informants at study sites prior to data collection to select a list of participants, based on location, gender, profession or work experiences, COVID-19 vaccination status, education and age. We aimed to recruit an estimated sample of 18–24 community participants per country. The researchers in each country compiled lists of potential participants, contacted them via phone, and arranged interviews either online or offline. We sent reminders to the candidates prior to the interview. If the participants did not join the interview, we replaced them with other participants having similar characteristics. For those who did not participate, we did not require them to report reasons for not participating. In parallel with the interview process, we conducted debrief sessions with the research team to discuss highlighted topics from the participants across sites and to refine the interview guide for further topics to explore or when we reached saturation of a topic. We also discussed and refined the sampling frame as needed during these sessions. Recruitment challenges included potential participants not responding to calls, some were no longer willing to participate or they provided invalid phone numbers in the survey. We reminded participants that there were no right and wrong answers during the interview process, interviewers as remained neutral as possible, and valued silent moments to help participants feel comfortable expressing their views. However, if participants directly discussed vaccine misinformation during the interview, we probed on the topic to gather more information and remained as neutral as possible during the interview. At the end of the interview, we addressed misinformation that the interviewee felt comfortable discussing and/or referred participants to discuss with HCWs.

Data collection

Study investigators developed the semi-structured interview guide, and the study team revised the topics and questions following routine debrief meetings and based on the evolution of the local situation. We explored the following topics: an open narrative of COVID-19 experiences; acceptance and accessibility of COVID-19 vaccines; perceptions, production of vaccines and other related topics (S1 File). The interviews were conducted between December 2021 and June 2022, when the first two doses of COVID-19 vaccine had been introduced across all the sites and the booster doses were underway at some sites. Due to mobility restrictions in these three countries, we conducted the majority of the interviews via phone or audio calls, but we also held face-to-face interviews when feasible.

The researchers provided the participants with details on the study and requested permission to record the interviews. We collected audio-recorded verbal informed consent for those who had a remote interview; those who had in-person interviews were asked to sign a paper consent form. The interviews were audio-recorded, and the interviewers made notes during the interviews.

Data analysis

The audio-recorded interviews were transcribed in the language spoken and translated into English by qualitative researchers of each country. Handwritten notes were typed and along with other fieldnotes, all files were uploaded into NVivo 12 to support further analysis. The bilingual researchers at each country checked and coded both original transcripts and English translations to ensure consistency.

We had at least one to two coders per country to conduct two cycles of coding on their country’s dataset initially [26]. First, we developed attribute coding (i.e., categorising the transcripts based on demographics data; gender, settings, age, profession, and COVID-19 vaccination status), and provisional coding using open or inductive coding. In the second coding cycle, each analyst compared themes from first cycle to the framework of vaccine hesitancy developed by the World Health Organisation [5]. The framework helped us map potential vaccine hesitancy across the three countries and ensured that we captured the main determinants related to vaccine hesitancy, including contextual influences, individual and group influences, and vaccine specific issues. At the end of this cycle, we had extensive discussions across the sites to organise comparable codes into categories and patterns. We merged datasets from three countries in NVivo 12 and then used a shared Excel sheet and visual thematic mapping to help us understanding the themes and patterns. We continued with more discussions and reflections about the agreements and disagreements across the sites and finally, refined the main themes.

Risk

Participants spoke about risk when considering whether to vaccinate. On the one hand, respondents spoke about the fear and risk of acquiring COVID-19 infection and the vaccine was viewed as a protection. Many participants who were fully vaccinated, including those who had received a booster shot, often described their motivation for getting vaccinated as predominantly driven by the fear of contracting COVID-19.

“I was excited and felt happy about the thing [vaccine] that I would get immunity from the vaccine to fight against the scariest disease.” (P48, 42, male, filmmaker, rural, Nepal)

Similarly, participants with children expressed fear about them being exposed to COVID-19 outside the home, which prompted them to view vaccination as a safeguard for their family members. The fear was fuelled by the media coverage of severe COVID-19 cases, which often led to feelings of worry, motivating them for vaccination.

“They [parents] still let their children get vaccinated. They were worried because the kids meet a lot of people at school, and they didn’t know what might happen. The kids might only meet their friends at school, but there might be other people around them after school who might be wandering around with COVID. Their parents were worried, so they let them get vaccinated.” (P35, 27, female, housewife, rural, Vietnam)

On the other hand, the narratives surrounding risk evolved into dialogues about the perceived risk of acquiring infection and, later, the severity of infection. While the first two doses were implemented under the combined policies of travel restrictions and mass vaccination requirements, there was less legislation around the booster doses, and participants’ views about booster doses better illustrated how disease threat and severity were integral to their decision to vaccinate. The perceived sources of risks (or lack of) varied among participants across the sites and were closely related to the country’s situation at the time of data collection.

“So those who are not forced… who don’t have any job, including me, are not vaccinated until today. My child who does not work in offices is not vaccinated. Except for those who work, it’s hard to be avoided […] because people think the vaccine is not a medication, but it only protects by immunity” (P3, 70, male, retired, urban, Indonesia)

Another facet of risk was highlighted in participants’ perceptions about the risk of spreading COVID-19 to their family members whom they perceived as vulnerable. For some, the propensity to vaccinate relied on how much risk they would pose to their family members if they became ill. For participants with underlying medical conditions, comorbidities, or pregnant women concerns revolved around severe side effects or long-term health problems from vaccination. In these instances, some decided to postpone their vaccinations, as advised by early vaccination programme instructions, doctors at vaccination centres, or their instincts.

“I was doubtful at that time because of my own health condition, I had pneumonia. Because I went through that, I was afraid of getting vaccinated.” (P5, 36, female, non-government organisation, rural, Indonesia)

Additionally, pregnant or breastfeeding women were concerned about the potential consequences for their infants and their health, such as high fever or reduction in their milk supply, which they heard about from others or on social media.

“At the time of vaccination, I was breastfeeding my son. Initially, we talked about whether I should get vaccinated or not … and deciding it shouldn’t be taken during that time [whilst breast feeding] so I didn’t take the vaccine.” (P52, 27, female, banker, urban, Nepal)

Trust

Another critical element that influenced the acceptance of COVID-19 vaccination across the three countries was trust. We found a variety of forms of trust, including trust related to national governments, COVID-19 vaccines, and interpersonal trust [12].

Logistics

During the COVID-19 vaccine rollout, we identified similar practical issues across the three countries, including inequitable access, issues regarding COVID-19 vaccine certificates, and administrative challenges. Here we present data demonstrating how these issues affected uptake of vaccines.

Social norms

Lastly, social norms and practices influenced COVID-19 vaccine uptake and/or hesitancy, including traditions to rely on alternative remedies when dealing with common infections and gender influences in decision-making. Some of the social norms that we identified were country-specific. For some participants, their religion influenced their social norms, such as those from Indonesia who referred to their religious faith as an alternative approach to their healthcare practice.

“Well, my wife was strong. We slept together. I don’t know what power she has, she didn’t get infected. Yes. But she is diligent in reciting the Quran. That’s probably the religious approach. Well, that’s my wife. My grandson also slept with the one with COVID, he was okay as well. Yes, only 1.5 years old, he was okay. If it’s time to die, then let it be.” (P23, 59, male, taxi driver, urban, Indonesia)

Additionally, it was important for vaccines to have halal status for some Indonesian participants, particularly when people were making decisions whether to have COVID-19 vaccine booster shots or not. The halal issues were predominant when more vaccines were available to the public rather than in the early pandemic stages since a fatwa of the Indonesian Ulema Council or in Indonesian called Majelis Ulema Indonesia (MUI – Indonesia’s highest religious body) had stated: “In an emergency, it is permitted to do what is normally prohibited” [30], thereby allowing the Muslim population to accept non-halal vaccines, such as the Sinovac vaccine which was the first vaccine available in the country. In the later stages of the pandemic, it was perceived as less of an emergency situation. The halal concern was a dominant driver identified in the urban study sites of Jakarta and Bandung where the population is predominantly Muslim, but not so in the rural Indonesian study site of Sumba, where the population is predominantly Christian.

“I’m currently considering whether to take it [the booster dose] immediately or not, I’m sorry… related to the halal status. That’s one of the things I consider, apart from the types. Even though to my knowledge, there are only a few vaccines that have been certified as halal. Even though it is currently being promoted, what I know is also for boosters, Pfizer, and AstraZeneca, to my knowledge it’s not halal yet.” (P11, 31, male, teacher, urban, Indonesia)

Similar religious concerns were not mentioned by participants in Nepal and Vietnam.

For some participants, their approach to COVID-19 was influenced by their usual healthcare practises, particularly if their experience of COVID-19 infection was mild. They likened it to common seasonal infections for which they normally self-treat, and therefore they did not seek vaccination for COVID-19. In addition, after the peak of the pandemic and with fewer severe cases, people returned to their social norms for treatment of illnesses.

“Two aunties have not got it [vaccines]. (I: Yes) Saying this is just a common cold so they don’t wear masks. There are people like this here who are not getting vaccinated. Well, a lot of (diseases) came and go. They have not got vaccinated saying they are healthy, they won’t wear mask and won’t take medicine for the simple common cold.” (P48, 42, male, film maker, rural, Nepal)

“She [participant’s wife] followed me. When she wants to go to the doctor [to get the COVID-19 vaccine], I said why go to the doctor, there is herbal medicine [laughs].” (P3, 70, male, retired, urban, Indonesia)

Embracing social norms

The communities in our study functioned within the existing social norms to respond to the outbreak, which influenced their motivations and decisions about COVID-19 vaccines. For example, some participants opposed the COVID-19 vaccines, yet at the same time, expressed a positive attitude towards immunisation programmes for children. This might be explained by the idea that positive attitudes toward vaccines do not always translate into practice [34]. However, adult immunisations are generally not included in national programmes, except for tetanus immunisation for pregnant women. This could contribute to the common paradigm that immunisation is given during childhood.

Another norm was that people were unlikely to accept something “new”. This highlights that vaccine mandates, either implicit or explicit, need to be carefully considered in vaccination policy. Although vaccination uptake might increase, it might also create scepticism [35]. Given that vaccination is also dynamic [36], strengthening the visibility of pro-vaccine actors beyond the government [34] may support collective action and build more proactive stories while allowing individuals time to process and accept vaccination. This is particularly relevant in both outbreak and non-outbreak settings, where the protective value of herd immunity may be perceived as an abstract concept or an individualised benefit, for instance, in the decline of booster uptake following the decrease in COVID-19 cases. In addition, some viewed immunisation and vaccination as distinct concepts, with vaccination perceived as a ‘new’ development emerging during the COVID-19 pandemic. We argue that improving communication and awareness about vaccines [32,33,37], and immunisations is essential to improve vaccine acceptability and uptake in the local context [38].

In Indonesia, religion is an important social norm, and religious considerations can have a prominent influence on vaccination programmes, such as the refusal of the measles rubella vaccine campaign in 2018 [39]. It was concluded that the Sinovac type was halal [40], while the AstraZeneca vaccine used trypsin derived from porcine in the manufacturing process, but it was still allowed to be used in the context of an emergency situation [40,41]. However, even within Indonesia, this norm is context-specific and it was mainly discussed by participants in urban areas with a Muslim majority and less mentioned in the rural site, Sumba, which has a Christian majority. Therefore, in Indonesia, religious leaders are an important group to be engaged with when introducing new vaccines.

Although we did not approach the topic through a gender lens, as it was beyond the scope of the initial study aims, the influence of gender on vaccine acceptability was a norm in certain settings— some women’s decisions depended on the approval of men (husbands) for taking COVID-19 vaccines. Gender can also influence other health issues in these countries—for instance, in reports of leprosy cases in Nepal and Indonesia cultural traditions often require women to depend on their husbands or in-laws for decisions [42]. Similarly, some women need their husband’s permission to use contraception, demonstrating a lack of freedom for women on health choices [43].

In the third phase of the pandemic, there was a decline in severe cases of COVID-19 across the three countries. Although the respective governments continued their vaccination programmes, there was a waning in public motivation to accept booster doses. Sustained engagement and understanding of the social norms in each context is essential to increase motivation towards vaccines and foster trust [38].

Fostering trust

There is substantial recent research examining how trust influences acceptance of public health control measures [44,45]. The literature on COVID-19 vaccine acceptance particularly identified distinct types of trust: trust in institutions, including governments and healthcare providers and also trust in the COVID-19 vaccines [12]. Understanding different drivers of trust are essential to design effective approaches to increase and maintain vaccine acceptance.

In terms of institutional trust, a global study reported that citizens in Vietnam have the highest levels of confidence in national health services [46], leading to strong compliance with the government policies, resulting in high rates of acceptance of the COVID-19 vaccine. This aligns strongly with our data. We conclude that trust is built by the public’s experience of a competent health service, as seen in the rapid response to the first COVID-19 cases in Vietnam, but it is also likely that the public remembers the Vietnamese government’s effective control of the Severe Acute Respiratory Syndrome (SARS) epidemic in 2003 [47,48]. Similarly, a study from Nepal found a significant association between perceptions of health system performance and local-level trust in Nepal, demonstrating that service delivery and performance are direct drivers of political trust [45]. In contrast, the eroding of trust in institutes and state services is compounded by low satisfaction in the health services, such as long waiting times, [49] uncertain schedules and unequal access [50]. We conclude that addressing logistical and administrative challenges would have improved uptake not only to the COVID-19 vaccines, but also to other life-course vaccinations provided for the communities. The second focus of trust was on the COVID-19 vaccines themselves. Although not all views from the participants were negative, it is important to note that geopolitical prejudices could hinder acceptance of certain vaccines.

Finally, trust in states and institutions can also be eroded when people are exposed to inconsistent information in the media [51]. In the same way, lack of transparency about vaccine development and research drove mistrust [17,52]. Mainstream media in Vietnam is state-controlled, and was seen as a reliable source of information during the pandemic [47]. On the other hand, analysis of media in Indonesia during this time revealed a lack of transparency in the dissemination of COVID-19 information and data in Indonesia driving mistrust in the government’s handling of the pandemic [53]. Although the abundance of online media enables the rapid circulation of misinformation and rumours, there is also a correlation with digital literacy and acceptance of disease control measures. A study on media literacy in Vietnam showed a positive correlation between young people’s media literacy and their trust in government, compliance with public health regulations, and vaccine acceptance [54]. A study in Indonesia reported that hesitancy towards booster doses in Indonesia was linked to lack of knowledge, including about side effects, halal status and benefits [17]. Our data similarly showed that lack of information and knowledge about the vaccines and the government control measures led to mistrust in the government and the vaccines, and hesitancy to accept vaccination.

These findings highlight the need for transparency from governments and health providers in times of public health emergencies, with more focused campaigns and communication that build the public’s knowledge about research and vaccines and sustain trust in health authorities and hinder spread of misinformation.