Study design

A mixed-method design was employed. A series of cross-sectional surveys and key informant interviews with was conducted between March and August 2021. Survey respondents were the general Thai population, while key informants from interviews were specific samples who live in geographical areas classified by the level of risks of outbreak.

Participants and data collection

For quantitative data, six rounds of surveys were conducted among the general public aged over 15 years old. The research team worked closely with the Department of Health Service Support, Ministry of Public Health (MOPH) to integrate this module with the ongoing survey "Stay Home, Save Lives", which aims to monitor people’s compliance on the disease control measures. Survey data were collected fortnightly through the network of village health volunteers (VHVs) in all 77 provinces using Google Form. Rounds 1, 2, 3, 4, 5 and 6 were conducted in 1–15 March, 16–30 April, 16–31 May, 16–30 June, 16–31 July, 16–31 August of 2021. Each round took 15 days before closing the survey for analysis.

In 2020, Thailand had more than 1.05 million VHVs nationwide. They are recruited by local communities and trained by MOPH, and each is responsible for 10–15 households in their community [21]. Since 2019, the monthly honorarium has been increased to 1000 Baht (US$ 33) to each VHV supported by the government and additional 500 Baht (US$ 16) for six months to support their contribution to COVID-19 containment during the pandemic [22].

The participants were conveniently selected based on their willingness to respond to the Google form disseminated by VHVs. We did not require VHVs to keep track of the same respondents to respond to the following survey rounds. Additionally, an online survey via Facebook or website was distributed to ensure additional coverage. Sample size was calculated from Yamane’s formula (n = N/1+Ne²). N stood for the range of population, 194,573 to 5,527,994 per province and e referred to 0.05; for which 400 samples per province are required [23]. To cover all 77 provinces, the total 30,700 samples per round (400 samples x 77 provinces), which is equivalent to total 184,800 samples in all six rounds (30,700 samples x 6 rounds) are required.

For the qualitative study, purposive sampling was conducted through online announcements and recommendations by the Provincial Islamic Committees, with the following inclusion criteria:

• The priority groups of vaccine rollout according to MOPH definition: health professionals, individuals with seven priority diseases (chronic respiratory disease, cardiovascular disease, end stage of chronic kidney disease, neurovascular disease, cancer, diabetes, obesity (weight > 100 kilograms or BMI > 35 kg/m² [24]
• People who have high-risk occupations such as employees working in public spaces or the tourism sector, as they are the priority for the next phase of vaccine rollout [24]
• Those who were more likely to have low vaccine acceptance, identified from previous literature [25], such as Muslims in deep southern provinces.

These three groups of participants were recruited from low and high outbreak provinces [12]. The research team offered 500-Baht (US$ 16) compensation for their participation in the study.

Measurement

The questionnaire comprised five sections: 1) demography (gender, age, education, occupation, place of residence, religion, health status, village health volunteer status (responses are multiple choices except age); 2) vaccine acceptance (multiple choices of yes/no/uncertainty); 3) have heard or seen information (multiple choices of yes/no) and ability to differentiate true and false statement (determining susceptibility of misinformation)(multiple choices of true/false/unsure); 4) trustworthy sources of information (Likert scale ranks from 1–5 from totally disagree, disagree, neutral, agree to totally agree); and 5) other factors related to vaccine acceptance such as self-perceived risk of infection, vaccine confidence, trust in the government, healthcare providers and vaccine manufacturers (Likert scale ranks from 1–5 from totally disagree, disagree, neutral, agree to totally agree) [8,20].

In each round of survey, there are five different true and false statements relating to the vaccine to assess participants’ ability to differentiate true and false information. These statements were generated from social listening by researchers and capture the dynamic of information flows on social media related to vaccines during vaccine rollout by the government. The process of selecting statements for each round of survey starts with monitoring headlines related to COVID-19 vaccine, then pick up the topics that were frequently mentioned or quoted by news agencies or widely discussed on social media platforms such as Facebook or Twitter. These findings were discussed among the research team and stakeholders, such as representatives from the local office of United Nations Children’s Fund (UNICEF) and WHO, and the Thai National Vaccine Institute and Department of Disease Control, MOPH to ensure validity and decide if the statement was true or false. It took 30–45 minutes to complete the self-administered questionnaire.

For qualitative data collection, the interview guide was adapted from the WHO Interim Guidance: for gathering and using data on the behavioural and social drivers of vaccination and the WHO European Region field guide to qualitative research for new vaccine introduction [26,27]. A semi-structured questionnaire comprising three main parts was developed, revised and finalised: 1) decision to vaccinate; 2) reasons to accept or refuse vaccine such as perceived risk of COVID-19 infection, vaccine confidence, trust in government and healthcare providers; and 3) what are the trustworthy sources of information? The interviews lasted approximately one hour via Zoom, Line, or by telephone, depending on the preferences of the interviewees. More than two researchers participated in each interview to avoid bias or domination; the conversation was electronically recorded after consent approval for detailed analysis.

Data analysis

The quantitative surveys were analyzed by using STATA 14.0; descriptive and inferential statistics were applied. The characteristics of respondents, vaccine acceptance and trustworthy sources of information were presented in percentages while impact of the vaccine information and other related factors on vaccine uptake applied multivariable binary logistic regression with 95% confidence level.

In multivariable binary logistic regression analysis, only those who were exposed to information were analyzed because it is not valid to assess the capacity among participants who were not exposed to such information. The independent variable was demographic data whereas vaccine acceptance was a dependent variable. The other independent factors such as ability to differentiate true and false information, confidence in the vaccine, trust in the systems, and perceived risk of infection were analyzed separately with the same independent and dependent variables mentioned above.

Only high frequency of exposure and general knowledge statements, which were not limited to the Thai context, were selected and separately analyzed in each true and false statement (See S3 Table). Those who have the ability to differentiate information are defined as giving correct answers to the selected statements. The Likert scale of trustworthy sources of information, vaccine safety, effectiveness and importance, and trust in government, healthcare providers and vaccine producers were defined as high level if the answer is ‘agree’ or ‘strongly agree’ or score 4–5 [20].

In-depth interviews were transcribed and coded by the research team in an Excel program. A deductive analysis was employed using three main themes: 1) factors and their trade-off for making decision to vaccinate; 2) appraisal and trustworthy to sources of information; and 3) other factors related to vaccine uptake in specific groups such as religious belief.

Ethical consideration

The research ethics of this study was approved by the Development of Human Research Protections (IHRP No. 059–2564). The participants and guardians/authorized representatives provided verbal informed consent before starting the surveys and interviews.

Theme 1: Trend of vaccine acceptance, hesitance and refusal

We defined vaccine acceptance as those who reported they want a vaccine; hesitance as those who were still indecisive to whether they will get a vaccine or not; and refusal as those who definitely rejected a vaccine.

The overall trend of vaccine acceptance increased over the six surveys. About 60% of all respondents accepted vaccines in March 2021 then it decreased to 44.0% in April 2021; it increased to 69.6% in May 2021 and peaked at 88.8% in August 2021. At the same time, 24.7% hesitancy in March 2021 decreased to 8.1% and 5.8% in July and August 2021. Similarly, the number of those who refused a vaccine decreased over the six-month period from 15.0% in March to 5.4% in August 2021. See Fig 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The trend of vaccine acceptance, hesitance and refusal.

https://doi.org/10.1371/journal.pone.0276238.g002

Theme 2: Trustworthy sources of information

From the survey, health professionals (74.9%), VHVs (73.2%), and academics (69.1%) were reported as the three most trustworthy sources of vaccine information. The government, family and friends, and the media were reported as less trustworthy accounting for 62.2%, 58.0% and 50.0% of respondents respectively. Public figures, such as celebrity or social media influencers, were seen as having the least trustworthy source of vaccine information (35.6%). See Fig 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The proportion of respondents who trusted different sources of information.

Note: n = 26,037.

https://doi.org/10.1371/journal.pone.0276238.g003

From in-depth interviews, the respondents had different judgments subject to their ability to understand, appraise information they received and search for additional evidence to verify information they received. Interviewees who were frontline health staff or had higher education proactively sought and verified with additional evidence to confirm the reliability of the information they received, usually through reliable academic journals, or international media published by WHO or UNICEF. Non-health professional interviewees commonly received information from healthcare workers and media through various channels such as online media (Facebook, Twitter, Clubhouse application) and television. Older people or people who were not able to search additional information on their own often got information from television and trusted individuals such as their younger relatives or close friends. A few key informants relied on their trusted sources and ignored sources that they considered unreliable, while some sought multiple sources of information before making their decisions.

“(I get my information) mainly from my children. […] My son told me not to go out if I am not vaccinated because there are several infected cases in our neighborhood.” (EPRN2)

Some key informants reported that information from official sources such as the Centre for COVID-19 Situation Administration (CCSA) were inaccessible, unclear and unreliable. They noted that there was no clear evidence on vaccine safety and efficacy, and they were not in line with international sources of information. They noted that CCSA promoted only information to support government-provided vaccines (in this case Sinovac and AstraZeneca). These key informants suggested the authorities should have proactively communicated by providing factual, understandable, accessible, and responsive information via appropriate communication channels for each population group.

“One thing is that the government sector information is very difficult to access. Only people who are really interested can reach it. […] When I tried googling government information, I found nothing. It’s really hard to access them. […] Moreover, the government websites are not user-friendly. It is difficult to search.” (IRY1)

Theme 3: Impact of vaccine information and other factors on vaccine uptake

The exposure to misinformation and ability to differentiate true and false information was assessed in all six rounds and only high exposure and general knowledge statements, which were not limited to the Thai context, were selected for analysis. Table 2 contains eight statements of which four are true and four are false. Details of other statements are shown in the Supplementary file, see S2 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. The number and proportion of respondents who were exposed to and provided correct and incorrect answers to eight selected true and false statements.

https://doi.org/10.1371/journal.pone.0276238.t002

In Table 2, more than half of respondents have seen these eight statements. Those exposed to true statements were more likely to give correct answers compared to those who saw false statements.

In the surveys, we maintained certain key parameters for trend monitoring which can inform policy in all six rounds. There was an upward trend of confidence in COVID-19 vaccines in terms of safety, efficacy and that vaccination was important to them; although trust in government, healthcare providers and vaccine producers saw a less significant increasing trend. In the first round of the survey (March 2021), less than half of respondents agreed that the vaccines are safe (40.3%), vaccines are important for them (42.9%) and vaccines are effective (44.4%). Participants’ confidence on safety and efficacy rose to more than 60% in May 2021. In round 6 (August 2021), confidence in vaccines reached its peak with respondents saying that the vaccine was important for them (78.0%), effective (73.0%) and safe (71.7%).

In the first round of the survey (March 2021), more than half of participants trusted healthcare providers (51.1%) while around two-fifths trusted vaccine producers and the government (42.2% and 37.3% respectively). See Fig 4. However, trust in healthcare providers and the government increased slightly then dropped significantly in July by 5 to 10 percentage points. Nevertheless, in the last survey round in August 2021, most respondents trusted the health system (68.1%) followed by vaccine producers (62.7%). The government had the least trust, at less than 50%. The high level of perceived risk of COVID-19 infection increased from 49.7% in the first round of survey to 71.8% in the last round.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Trend of key parameters related to COVID-19 vaccine in the population.

https://doi.org/10.1371/journal.pone.0276238.g004

From the parameters generated from quantitative surveys, we conducted multivariable binary logistic regression analysis on the association between demographic characteristics and other factors on vaccine acceptance, categorized by accept, or hesitant that combined refusal. Tables 3 and 4 analyzed participants who had been exposed to eight selected true and false statements.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The frequency of vaccine acceptance classified by various factors among respondents who were exposed to eight selected true and false statements.

https://doi.org/10.1371/journal.pone.0276238.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The multivariable binary logistic regression analysis of association between various factors and vaccine acceptance among respondents who were exposed to eight selected true and false statements.

https://doi.org/10.1371/journal.pone.0276238.t004

Analysis suggested that education, health status, area of residence, risk perception, perceived importance and safety of vaccine, trust in vaccine manufacturers, and ability to differentiate true and false statements significantly correlated with willingness to vaccinate.

In Table 3, respondents who had the ability to differentiate true and false information were 1.2–2.4 times more likely to accept vaccine (P-value<0.001–0.012). Respondents who had secondary or higher levels of education tended accept vaccination more than those with completed primary education (Adjusted odds ratio (AOR) = 1.6–4.1; P-value<0.001). Additionally, those who perceived a higher risk of infection (AOR = 2.6–4.7; P-value < 0.001), and perceived vaccine safety (AOR = 1.4–2.4; P-value<0.001–0.005) and the importance of vaccination (AOR = 2.3–5.1; P-value<0.001) were more likely to accept vaccination.

Respondents who had higher risks of infection such as chronic diseases (AOR = 0.7–0.9; P-value<0.001–0.024) were less likely to accept vaccination than healthy people, whereas those who lived in high-risk areas (high outbreak numbers) tended to have higher vaccine acceptance than those lived in low-risk areas (AOR = 1.4–3.0; P-value<0.001). Participants who trusted the vaccine producer were also more likely to get vaccinated (AOR = 1.9–3.2; P-value<0.001).

From interviews with key informants, there were various reasons for accepting, hesitating or refusing vaccination. Our multivariable binary logistic regression analysis clearly showed risk perception was one of the key determinants, followed by areas in which they lived, health status, or their work status or lifestyle, which exposed them to infection. This was also confirmation from in-depth interviews with, for example, interviewees who lived in areas with high outbreaks and those with jobs that exposed them to many people who have asymptomatic infections, were more likely to get vaccinated. Some interviewees with chronic diseases decided to vaccinate due to their personal risks but some were still concerned about the side effects and became hesitant. Those who were reluctant to be vaccinated usually perceived a low risk of infection and felt that they could protect themselves from infection.

“I really have not met anyone but the delivery people. And I have no reason to worry about risk, because I do not have to go out and meet people. So, I think, it is OK that I can wait (and see the vaccine).” (IRY2)

Interviewees also described different degrees of vaccine confidence. Some recommended that the risk of side effects (which is extremely low) should be a trade-off with the benefits of the vaccine (which meant minimizing risk of mortality once infected). The benefits of the vaccine were perceived differently across individuals, for instance, some decided to wait for vaccines that had higher efficacy to prevent infection, whereas some individuals thought that benefits gained from reduced morbidity and mortality was more than enough. They also considered the importance of being vaccinated in terms of their own protection and to protect others such as family members. This means that if they are not infected, they will not pass infections to others in the family. Although the vaccine acceptance rate was generally increasing, some interviewees still refused the vaccine because they wanted to wait for the most effective vaccine; however, some finally accepted the vaccines provided by the government as their preferred vaccines. At that time they were viral vector AstraZeneca and inactivated Sinovac, as mRNA such as Pfizer and Moderna were not available.

“I think all vaccines can prevent severity of infection and minimize mortality. Although, some people said that Sinovac is imperfect, I still believe that it can prevent morbidity so it is enough to avoid overwhelming the health system … Moreover, I am afraid that I can be a virus carrier to my parents and grandmother who are both at risk of severity if infected.” (ORY3)

Finally, religious belief seems to have limited effects on vaccine acceptance. Though Halal is one of the concerns among the Muslim community, the Chulalongkorn University Halal Science Center has proved and announced that the COVID-19 vaccine is Halal and this was further endorsed by The Central Islamic Council of Thailand. Misconceptions about vaccines and other preventive practices spread among some groups of Muslims. However, Muslim leaders reassured that having a COVID-19 vaccination does not conflict with Islamic principles. Many Muslim populations, especially in deep south provinces, also received information from the media and people in other Islamic countries owing to the similarity in religion and language. Muslim leaders and the believers suggested that the government should appoint an influential ambassador who people in the community respect and trust, such as imams, Muslim medical personnel, Sheikhul Islam Office and the Provincial Islamic Committees, to help disseminate accurate information in local languages and work with local health authorities.

“There was a message about the vaccine and the side effects they received from Malaysia. I saw it too. […] First of all, (it is popular because) we share the same religion, as most Malays, especially those who live near the border, are Muslims like us. And second, the language, which is very similar. It is an understandable language across the border.” (MSY1)