Study design and setting

This cross-sectional survey was part of a mixed-methods study on influenza vaccine hesitancy among healthcare workers in Nakhon Phanom province, which is located in the plateau region of northeastern Thailand, approximately 735 km from Bangkok. The province borders Lao PDR and has a total area of 5,528.88 sq. km. The province consists of 12 districts, with Muang district as the capital, along with 97 sub-districts and 1,123 villages [34]. There are 14 public hospitals in the province: one provincial general hospital, 11 community hospitals, a psychiatric hospital, and a military hospital. From 2003 to 2015, there was active surveillance of respiratory illnesses in the public hospitals of this province [34–36].

Study population

Thai HCWs of Nakhon Phanom province were the targeted population of the study. Although the WHO defines HCW as “all people engaged in actions whose primary intent is to enhance health” [37], we limited HCW’s definition to those who were currently working in public hospital settings in Nakhon Phanom for operational feasibility. As per a report of DMS, 1,863 registered HCWs were working in the province in 2018 (DMS, MOPH). The HCWs were broadly categorized into two groups according to the amount of patients’ contact/exposure and as grouped by WHO [38]: a) Health service providers who delivered health services and came in direct contact with patients and substances such as patients’ blood and other specimens, and b) Health management and support staff who assisted in the functioning of the health system without directly providing health services and as such may or may not have had direct contact with the patients. According to the nature of their work, health service providers were further sub-categorized into six groups—medical doctors, dentists, nurses, pharmacists, public health officers, and other service providers [38].

Participant selection criteria

Thai HCWs of both sexes (male and female) aged above 18 years, having a good command of the written and spoken Thai language, and currently working at a public hospital in Nakhon Phanom province were included in the study. HCWs who were not present in the hospital at the time of data collection and who were not willing to participate in the study voluntarily were excluded.

Sample size and sampling technique

As there was no prior estimate of influenza vaccine hesitancy among Thai HCWs, a conservative proportion of 50% was taken to calculate the sample size for the survey. A sample size of 350 was calculated using the single proportion formula and a finite population correction factor in OpenEpi (www.openepi.com), assuming a Type I error of 5%, absolute precision of 7.5 percentage points, and a design effect of 2 given the cluster sampling used for the population size of 1,863 HCW in Nakhon Phanom after inflating 10% for refusal and absences.

HCWs were recruited in the survey using a two-stage cluster sampling. In the first stage, hospitals were selected as clusters to recruit 50 HCWs in each cluster (i.e., seven clusters in total). The hospitals were chosen by a systematic random sampling using probability proportional to size (PPS) of the number of HCWs. As Nakhon Phanom Provincial hospital was a large hospital, it got selected twice. Therefore, 100 HCWs were recruited from that particular hospital, and 50 HCWs were selected from five other hospitals. In the second stage, HCWs were chosen using stratified random sampling according to their professional categories and proportional distribution in each hospital. Microsoft Excel sheets were used to generate the random numbers for the sampling in both stages, as described above.

Survey instrument and data collection

The survey questions were modified from the sample questions of SAGE’s vaccine hesitancy determinant matrix for the Thai context to measure influenza vaccine hesitancy and its three broad determinants: contextual or sociocultural, individual/group, and influenza vaccine-specific influences [16, 39]. Clear and concise language was used for ease of understanding and questions that were related were grouped to avoid jumping between unrelated topics. The participants were asked to indicate their level of agreement or disagreement with the items of the constructs on a 5-point Likert scale, with ‘1’ meaning strongly disagree and ‘5’ strongly agree. The questionnaire took around 30–35 minutes to complete. The study instrument was validated for use in the Thai context by another research project which used qualitative methods in the development of the questionnaire and had rigorous steps such as pilot testing (n = 30) for assessing feasibility, acceptability, and face validity, and a larger survey among a nationally representative sample of healthcare workers (n = 2352) for evaluating construct validity and reliability [40].

In this survey, the questionnaire was self-administered to the HCWs who were approached through designated focal points at the six hospitals in the province, identified by the sampling technique, from 1^st August to 30^th September 2020. A total of 350 questionnaires were distributed to the participants. The participants were assured that their responses would remain anonymous and confidential to reduce the pressure to conform to socially desirable responses. Clear instructions were provided on how to answer the questions, emphasizing the importance of honest and accurate responses.

Variables and measurement

The dependent variable of the study, influenza vaccine hesitancy, was defined as either delaying getting influenza vaccines, accepting the vaccines but being unsure, or refusing the vaccine despite seeing its value. It was measured as a binary variable (yes/no). “Yes” meant those HCWs who displayed some form of hesitancy, and “No” meant HCWs who accepted the influenza vaccine without any doubts.

The independent variables included demographic and work characteristics. Participant’s age and years of experience as HCW were first measured as continuous variables and then converted to appropriate categories. Categorical variables were sex (binary—male/female), religion, level of the hospital (Provincial/District), HCW type (Health service providers/Health management and support staff), daily exposure to patients (<1 meter/≥1 meter/no contact), education level, and self-reported presence of any pre-existing medical conditions (yes/no). Participants’ knowledge of influenza and vaccination was assessed using seven items that were adopted as a subset of a questionnaire from a previous study among Thai physicians [29], and then grouped into three categories using a modification of Bloom’s original cut-off points, such that six or more correct answers (i.e. ≥80%) represented “good knowledge”, four to five correct answers (i.e. 60–79%) meant “fair knowledge” and three or less correct answers (i.e. <60%) were considered as “needed improvement”.

The determinants of vaccine hesitancy were measured as constructs (i.e., composite variables). Contextual factors included the negative influence of media/social media (2 items), politics/policies (2 items), pharmaceutical industry influences (2 items), and the level of trust in the health system (2 items). Individual/group influences contained constructs of experience with past vaccination (3 items), beliefs and attitudes about influenza and vaccination (3 items), vaccination as a social norm (3 items), and influence of other healthcare professionals (2 items). Influenza vaccine/vaccination-specific issues included constructs of risk/benefit (perceived, heuristic) (4 items), risk of adverse events due to vaccination (3 items), fear of painful injections (2 items), and access to vaccines in the hospital (4 items) (S1 Table). Both positively and negatively worded statements were used, and the direction of negatively worded questions was reversed for data analysis. Items under each construct were averaged to create aggregated mean scores for the construct. The average scores were case-ranked and then used to divide the construct into three categories–high, moderate, and low.

Data analysis

Descriptive statistics were first employed to provide a summary of the selected variables and study samples. The mean scores of vaccine hesitancy determinants were compared between hesitant and non-hesitant HCWs using independent t-tests. The associated factors of influenza vaccine hesitancy were identified by a series of binary logistic regression models. As the dependent variable (influenza vaccine hesitancy) was categorical, the logistic regression analysis was performed using non-hesitant HCW as the reference category. All variables showing at least some evidence of association (Type III p values <0.1 in the binary logistic regression models) were entered into a multivariable logistic regression model, adjusting hospital locations to account for clustering. Variables with significant adjusted odds ratio (aOR) in the final model were recognized as associated factors (p-value <0.05). All data analysis was conducted using SPSS version 20 (IBM Corp., Armonk, NY).

Ethical considerations

Written informed consent was obtained from each participant before data collection. The study protocols and all related study documents were reviewed and approved by the Mahidol University—Central Institutional Review Board (MU-CIRB) (Protocol number MU-CIRB 2020/138.1606). Permission for data collection was also received from the Provincial Chief Medical Officer, Nakhon Phanom Provincial Public Health Office (ref no. 78.02/ 06115).

Response rate

Out of the total 350 survey questionnaires distributed, 338 filled questionnaires were received back from the participants, implying a response rate of 96.6% (Fig 1). The response rates varied at different hospitals, ranging from 80–100% (S2 Table).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Study flow and influenza vaccine hesitancy among healthcare workers in Nakhon Phanom Province, Thailand.

https://doi.org/10.1371/journal.pone.0310761.g001

Participant characteristics

Of those 338 participants, the mean age of the participants was 37.2±9.7 years, with a minimum of 20 years and a maximum of 60 years. Almost all participants were Buddhists (333; 98.5%). A majority of the participants were female (280; 82.8%), nurses (136; 40.2%), from district hospitals (238; 70.7%), with bachelor’s degree (223; 65.9%), and without any pre-existing chronic medical conditions (264; 78.1%) (Table 1). The participants had a mean work experience of 13.8±9.8 years as HCW, and around half of them reported having high daily exposure with patients in the proximity of less than 1 meter (172; 50.9%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Participant characteristics (n = 338).

https://doi.org/10.1371/journal.pone.0310761.t001

Knowledge about influenza and vaccination

A high majority of HCWs were aware of the MOPH recommendation of influenza vaccination for them (302; 89.3%). However, most participants were misinformed about the influenza vaccines as only 52% correctly identified that the influenza vaccine may not work if it contained the wrong virus strains than the circulating ones, and only 30% firmly believed that the influenza vaccine does not itself cause some people to get influenza (Table 2). Overall, only a quarter of the participants had a good knowledge regarding influenza and vaccination (85; 25.1%) (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Participants’ correct responses to each item of knowledge on influenza and vaccination (n = 338).

https://doi.org/10.1371/journal.pone.0310761.t002

Influenza vaccine hesitancy among the participants

A majority of the participants reported some degree of hesitancy towards influenza vaccines (197; 58.3%), and 7 HCWs (2.1%) were outright refusers of influenza vaccines who were not hesitant in their refusal (Fig 1). On closer inspection, most of the hesitant HCWs were “hesitant compliers” who had doubts about the influenza vaccines but still accepted them (172; 87.3%). Only 3 (1.5%) refused the vaccine while being considerate of its benefits, and 22 (11.2%) delayed receiving the vaccine for themselves (Fig 2). More than 60% of the health management staff, other medical professionals (paramedics), and nurses reported vaccine hesitancy, while other HCW types had less hesitancy than the sample prevalence of 58.3%; the least hesitancy was seen among pharmacists (31.6%) (Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Types of hesitant HCW among the participants.

https://doi.org/10.1371/journal.pone.0310761.g002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Influenza vaccine hesitancy levels among different HCW types.

https://doi.org/10.1371/journal.pone.0310761.g003

Vaccine hesitancy determinants

Removing the seven HCWs who refused influenza vaccines left with an analytical sample of 331 HCWs. The aggregated mean scores and standard deviations of the vaccine hesitancy determinants are presented in Table 3. The mean scores are on a scale of 1–5, where higher scores signify higher hesitancy or more negative sentiment towards the construct. Overall, the participants reported the highest negative perceptions towards pharmaceutical industry influences (mean score 2.8, SD 1.0) and influence of media/ social media (mean score 2.7, SD 0.8) (Table 3). When the mean scores were compared, HCW with influenza vaccine hesitancy were more likely to have a higher lack of trust in the health system (2.7 vs 2.3; p-value 0.008), negative past experience of vaccination (2.4 vs 2.1; p-value 0.017), negative perception of vaccination as a social norm (2.3 vs 2.0; p-value 0.011), negative influence of other HCW (2.2 vs 1.9; p-value 0.006), higher perceived risk and lack of benefit of influenza vaccine (2.4 vs 2.1; p-value 0.001), risk of adverse events due to vaccination (2.7 vs 2.3; p-value 0.001), fear of painful injections (2.5 vs 2.2; p-value 0.007), and lack of access to vaccines (2.5 vs 2.2; p-value <0.001) (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Influenza vaccine hesitancy determinants aggregated mean scores (n = 331).

https://doi.org/10.1371/journal.pone.0310761.t003

Factors associated with influenza vaccine hesitancy

In a series of univariable logistic regression analyses, different variables were found to be significantly associated with influenza vaccine hesitancy (Table 4). Compared to young HCWs (aged 30 years and below), HCWs aged above 50 years were nearly three times more likely to be vaccine-hesitant (OR 2.8; 95% CI 1.2–6.3). Similarly, HCWs with 20 years or more work experience were more than two times more likely to be vaccine-hesitant than HCWs with experience of 5 years or less (OR 2.2; 95% CI 1.2–4.2). In contrast, HCWs having fair knowledge of influenza and vaccines were less likely to be hesitant compared to HCWs with less knowledge (OR 0.5; 95% CI 0.3–0.8).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Factors associated with vaccine hesitancy among HCW (n = 331).

https://doi.org/10.1371/journal.pone.0310761.t004

Among the contextual factors, HCWs with negative perceptions of politics/policies and lack of trust in the health system had significant odds of being vaccine-hesitant compared to HCWs with positive perceptions (Table 4). Similarly, in the individual/group influences, significant determinants of hesitancy were negative past experience of vaccination (High—OR 2.1, 95% CI 1.2–3.7; Moderate–OR 1.7, 95% CI 1.0–2.9), moderate level of negative beliefs (OR 2.0, 95% CI 1.2–3.5), high negative perception of vaccination as a social norm (OR 2.2, 95% CI 1.3–3.9), and negative influence of other HCW (High–OR 3.3, 95% CI 1.9–5.7; Moderate—OR 2.7, 95% CI 1.5–4.9). Among the factors specific to influenza vaccines, low perceived risk (Low—OR 2.5, 95% CI 1.4–4.45; Moderate–OR 1.9, 95% CI 1.1–3.2), high risk of adverse events after vaccination (OR 2.8, 95% CI 1.5–5.3), fear of painful injections (High–OR 1.7, 95% CI 1.1–2.7; Moderate—OR 2.3, 95% CI 1.0–4.8), and reduced access to vaccines in the hospital (Low–OR 2.0, 95% CI 1.2–3.5; Moderate–OR 2.1, 95% CI 1.2–3.5) were significantly associated with the hesitancy.

However, in the multivariable model, only age (above 50 years–aOR 3.2, 95% CI 1.3–8.5), fair knowledge (aOR 0.4, 95% CI 0.2–0.8), and negative influence of other HCW (High–aOR 2.3, 95% CI 1.1–4.8; Moderate—aOR 2.1, 95% CI 1.1–4.4) remained significant factors of influenza vaccine hesitancy (Table 4). Work experience was not entered in the final model as it showed a high correlation with age.