Prevalence and associated factors of intention of COVID-19 vaccination among healthcare workers in China: application of the Health Belief Model

ABSTRACT

Healthcare workers (HCWs) are at an increased risk of coronavirus disease 2019 (COVID-19) and warrant COVID-19 vaccination to reduce nosocomial infections. This study investigated: (1) the prevalence of behavioral intention of COVID-19 vaccination (BICV) under eight scenarios combining vaccines’ effectiveness/safety/cost, plus two general scenarios of free/self-paid vaccination given governmental/hospital recommendations, (2) perceptions involving preferred timing of COVID-19 vaccination and impacts of various attributes on BICV, and (3) factors of BICV based on the Health Belief Model. An anonymous online cross-sectional survey was conducted among 2,254 full-time doctors/nurses in three Chinese provinces during 10/2020–11/2020. The prevalence of BICV was 75.1%/68.0% among nurses/doctors under the most optimum scenario of this study (free/80% effectiveness/rare mild side effects); it dropped to 64.6%/56.5% if it costed 600 Yuan (USD90). Similar prevalence was obtained (72.7%/71.2%) if the vaccination was recommended by the government/hospitals but dropped to <50% if effectiveness was 50% or mild side effects were common; 13.0% preferred to take up COVID-19 vaccination at the soonest (81.8% would wait and see). Scientific proof (completion of phase III clinical trials and approval from health authorities) was rated the highest in its impacts on vaccination decision, followed by vaccines’ performance, and then logistics. Multivariable logistic regression analyses showed that perceived severity, perceived barriers, cues to action, and self-efficacy (but neither perceived susceptibility nor perceived barriers) were significantly associated with the two BICV outcomes. The coverage of COVID-19 vaccination would be high only if the vaccines perform well. Health promotion may take the findings into account.

KEYWORDS:

• Vaccination
• intention
• the Health Belief Model
• healthcare workers
• China

Introduction

The ongoing Coronavirus Disease 2019 (COVID-19) pandemic has hit almost all countries. Since non-pharmaceutical measures commonly adopted in many countries were unable to control the pandemic,^1,2 COVID-19 vaccination is one of the few hopes left, as a high coverage of COVID-19 vaccination may result in herd immunity against COVID-19.³ This study was conducted from October to November 2020, when eleven COVID-19 vaccines had entered phase III clinical trials.⁴ As of February 24, 2021, 10 vaccines have been approved in various countries; their reported efficacies ranged from 50.4% to 95%;⁵ some countries have started rolling out COVID-19 vaccination.⁵ Since COVID-19 vaccines were not commercially available during the study period, this study can only focus on behavioral intention, which is a known predictor of the actual behavior.⁶ COVID-19 vaccination hesitancy has been widely reported.^7-12

Healthcare workers (HCWs) are important priority groups for COVID-19 vaccination for several reasons. First, they are vulnerable to COVID-19 infection. About 15.6% of the COVID-19 patients were asymptomatic according to a meta-analysis.¹³ For instance, 29.0% of the COVID-19 cases staying at a hospital in Wuhan of China during the initial COVID-19 outbreak were HCWs self-reporting possible nosocomial infections.¹⁴ In the U.S., 55% of HCWs infected by COVID-19 and having contacts with infected patients mentioned healthcare exposure as the sole route of transmission.¹⁵ Second, HCWs with asymptomatic COVID-19 infections may transmit the virus to patients, including older and/or chronic disease patients at risk of high case fatality of COVID-19.¹⁵ Third, HCWs may find COVID-19 infections among their colleagues stressful and demotivating¹⁶; COVID-19 vaccination may ease their mind.

A meta-analysis reported that 14 out of the 20 reviewed studies found that the coverage of influenza vaccination was <50% among HCWs.¹⁷ Vaccination hesitancy among HCWs hence raises a concern. In literature, only two studies have investigated the topic. They reported prevalence of COVID-19 vaccination intention of only 40% among the Hong Kong Chinese nurses¹⁸ and 78%/61% among the doctors/nurses in the U.K.,¹⁹ and associated factors including chronic disease status, contact with suspected/confirmed COVID-19 patients, and previous influenza vaccination.^18,19 Their questions about vaccination intention did not consider the vaccines’ effectiveness and safety, which were found to be positively associated with vaccination behaviors.^7,20 Health promotion of COVID-19 vaccination among HCWs is required.

This study used the Health Belief Model (HBM) as the conceptual framework. The theory postulates that determinants of health-related behaviors include: (1) perceived susceptibility of the health problem (e.g., perceived risk of developing COVID-19), (2) perceived severity of the health problem (e.g., perceived severity and potential harms of COVID-19), (3) perceived benefits of the behavior (e.g., protectiveness of COVID-19 vaccines), (4) perceived barriers against the behavior (e.g., obstacles against taking up COVID-19 vaccination), (5) cues to action (e.g., stimuli prompting COVID-19 vaccination), and (6) self-efficacy (e.g., confidence in taking up COVID-19 vaccination).²¹ The HBM has commonly been used in vaccination studies,^22,23 including those on influenza vaccination;^24-26 a study also reported significance between HBM constructs and COVID-19 vaccination intention in the general population.⁷ As factors related to HBM are modifiable, they can be used to guide the design of health promotion for COVID-19 vaccination among HCWs.

Stratified by the two occupations of doctors and nurses, (1) this study investigated the prevalence of behavioral intention of COVID-19 vaccination (BICV) during the first six months since its availability, under eight scenarios of specific effectiveness/safety/cost combinations and two scenarios (free and self-paid vaccination) of governmental/hospital recommendations; (2) it investigated perceptions related to COVID-19 vaccination descriptively (a) preferred timing of taking up COVID-19 vaccination, and (b) perceived levels of impacts of various attributes (evidence, performance, and others) on COVID-19 vaccination intention; (3) it investigated factors of BICV based on the HBM. It is hypothesized that the six HBM constructs would be significantly associated with BICV.

During the study period (October 19–November 26, 2020), the national average daily newly reported COVID-19 cases in China was 21; no death had been reported;²⁷ COVID-19 was put under control. During the initial phase of the COVID-19 outbreak in China, most of the cases were found in Hubei province. For instance, as of April 1, 2020, 67,802 out of the cumulative 81,589 COVID-19 cases were reported in Hubei (83.1%), while 1,501, 1,018, and 183 cases were reported in Guangdong, Hunan, and Yunnan provinces where this study was conducted, respectively. Although not ‘many’ cases were reported in some provinces, all hospitals in the entire country, disregarding the number of cumulatively/newly reported local COVID-19 cases, were then at the highest alert level, as there were strong worries about nation-wide outbreaks. Furthermore, the study was conducted prior to the roll-out period regarding COVID-19 vaccination which started after the registration of the first Chinese COVID-19 vaccine on December 31, 2020.²⁸ During the study period, a limited scheme of the experimental COVID-19 vaccines was offered to high-risk international travelers and a very small percentage of HCWs in China. COVID-19 vaccination was thus unavailable to almost all HCWs during the study period.

Materials and methods

Participants and data collection

An anonymous cross-sectional survey was conducted from October 19 to November 26, 2020. Data were collected from five hospitals in three provinces (two in Hunan, two in Guangdong, and one in Yunnan) in different regions of mainland China. The inclusion criteria were (a) full-time staff, (b) doctors or nurses, (c) being employed since January 2020 (since the outbreak of COVID-19 in China), and (d) access to mobile phones. Facilitated by the hospital administrators, all eligible doctors/nurses working in the selected hospitals’ major departments (e.g., internal medicine, surgery, gynecology and obstetrics, pediatrics, emergency, infectious diseases, and otolaryngology) received an invitation letter via the existing WeChat/QQ groups used for daily work-related communications, which included all doctors/nurses of the departments. (WeChat/QQ are the most commonly used social media applications in China that have over 1.2/0.7 billion users). The letter briefed them about the study’s background, anonymity, restriction to academic use, the right to quit at any time, and that return of the completed questionnaire implied informed consent. No incentives were given to the participants. Ethics approval was obtained from the Survey and Behavioral Research Ethics Committee of the corresponding author’s affiliated institution (Reference No. SBRE-20-094).

A total of 3,104 invitations were sent out; 2,287 completed questionnaires (73.7%) were returned to the research office via the online platform; six of which were invalid and being removed from data analysis. Among the rest 2,281 participants, data obtained from 17 (0.7%) participants who had taken up COVID-19 vaccination under the country’s scheme of limited use of experimental COVID-19 vaccines were excluded from data analysis. No participants had made appointments for COVID-19 vaccination. The effective analyzed sample size was 2,254 [doctors: 362 (16.0%); nurses: 1,902 (84.0%)].

Measures

Background information was collected (hospital, department, job seniority rank, marital status, whether living with elderly people, and whether having children <18 years old).

Behavioral intention of COVID-19 vaccination (BICV) was assessed by perceived chance of taking up COVID-19 vaccination during the first six months since the vaccines’ availability (1 = definitely not to 5 = definitely yes) under eight scenarios (S1–S8) combining vaccines’ effectiveness (80% versus 50%), safety (rare mild side effects versus common mild side effects), and cost (free versus 600 Yuan), and two scenarios of free or self-paid COVID-19 vaccination involving recommendations given by the government/hospitals (S9–S10).

Perceptions related to COVID-19 vaccination

1. Attitude toward the timing of taking up COVID-19 vaccination (one item: at the soonest/wait until obtaining comprehensive knowledge of the vaccines’ effectiveness and safety/as late as possible/avoid vaccination as much as possible/definitely not).

2. Perceived levels of impact of attributes of COVID-19 vaccines on COVID-19 vaccination decision (0 = no impact at all to 10 = extremely large impact). There were three categories: (a) scientific proof [i.e., completion of Phase III clinical trials and approval given by the State Food and Drug Administration (SFDA)], (b) performance (i.e., effectiveness, side effect, reported news about severe side effects, and duration of protectiveness), and (c) others (i.e., the convenience of vaccination, vaccination at the working hospital, and cost).

Factors of BICV based on the HBM

1. The 3-item perceived susceptibility of COVID-19 infection scale: Perceived chance of (a) contracting COVID-19, (b) contacting COVID-19 infected cases, and (c) infection among colleagues (1 = extremely low to 5 = extremely high; Cronbach’s alpha of 0.94).

2. The 3-item perceived severity of COVID-19 scale: Perceived levels of negative consequences of COVID-19 on (a) physical health, (b) mental health, and (c) life in general (1 = extremely low to 5 = extremely high; Cronbach’s alpha of 0.94).

3. The 6-item perceived benefits of COVID-19 vaccination scale: Perceived chances of gaining specific benefits: (a) self-protection, (b) protecting patients and colleagues, (c) less frequent use of facemasks, (d) restoration of normal social life, (e) psychological relief from worrying about COVID-19 infection, and (f) contribution to controlling COVID-19 in China (1 = extremely low to 5 = extremely high; Cronbach’s alpha of 0.89).

4. The 4-item perceived barriers of COVID-19 scale: Perceived chances of experiencing barriers against COVID-19 vaccination: (a) COVID-19 infection caused by vaccination, (b) severe side effects, (c) inconvenient logistics due to requirement of multiple doses, and (d) limited protectiveness (1 = extremely low to 5 = extremely high; Cronbach’s alpha of 0.83).

5. The 2-item self-efficacy scale: “You are confident in taking up COVID-19 vaccination if you want to” and “It is easy for you to take up COVID-19 vaccination if you want to” (1 = extremely disagree to 5 = extremely agree; Cronbach’s alpha of 0.79).

6. The four items on cues to action: (a) Having discussed with acquainted HCWs/non-HCWs and found them supportive toward COVID-19 vaccination (yes/no), and (b) frequencies of browsing COVID-19 vaccination information via governmental/other social media (1 = extremely rarely to 5 = extremely often). No scale was formed, as the question items involved different formats.

Data analysis

Chi-square/t-test was used to compare group differences. Two types of BICV were used as the binary dependent variables (1 = probably yes/definitely yes versus 0 = else): (a) free COVID-19 vaccination of 80% effectiveness plus rare mild side effects (S1), and (b) self-paid (600 Yuan or about USD90) COVID-19 vaccination of 80% effectiveness plus rare mild side effects (S2). Stratified by the occupation group, (1) univariable logistic regression analyses tested the associations between background factors and free/self-paid BICV; (2) multivariable logistic regression analyses tested the associations between the six HBM constructs and free/self-paid BICV, adjusted for background factors. Crude odds ratios (ORc) and adjusted odds ratios (ORa) and 95% confidence intervals (CIs) were derived. SPSS 21.0 was used for data analysis. Statistical significance level was defined as two-tailed p < .05.

Results

Descriptive statistics

Background factors (Table 1)

Of the participants, the mean (SD) age was 32.7 (7.4) years. The majority were female (89.0%); 49.7% had had a junior job rank; 70.3% were married; 68.2% lived with elderly people; 58.8% had children aged <18 years old. Doctors were more likely than nurses to be older, males, having senior job ranks, being married, and having children <18 years old (p < .05).

Table 1. Descriptive statistics of background characteristics and perceptions

	All (n = 2,254)	Doctors (n = 362)	Nurses (n = 1,902)	p^¶
	n (%)	n (%)	n (%)
Background characteristics
Sex				<.001
Female	2,015 (89.0)	175 (48.3)	1,840 (96.7)
Male	249 (11.0)	187 (51.7)	62 (3.3)
Age [Mean (SD)]	32.7 (7.4)	37.6 (8.7)	31.8 (6.8)	<.001
Hospitals				<.001
Hospital A in Yunnan	215 (9.5)	1 (0.3)	214 (11.3)
Hospital A in Guangdong	213 (9.4)	36 (9.9)	177 (9.3)
Hospital B in Guangdong	223 (9.8)	67 (18.5)	156 (8.2)
Hospital A in Hunan	1,271 (56.1)	219 (60.5)	1,052 (55.3)
Hospital B in Hunan	342 (15.1)	39 (10.8)	303 (15.9)
Department				<.001
Internal medicine	706 (31.2)	146 (40.3)	560 (29.4)
Surgery	561 (24.8)	46 (12.7)	515 (27.1)
Gynecology and obstetrics	85 (3.8)	30 (8.3)	55 (2.9)
Pediatrics	272 (12.0)	62 (17.1)	210 (11.0)
Infectious diseases	106 (4.7)	13 (3.6)	93 (4.9)
Emergency	166 (7.3)	12 (3.3)	154 (8.1)
Otolaryngology	46 (2.0)	11 (3.0)	35 (1.8)
Others	322 (14.2)	43 (11.6)	280 (14.7)
Job seniority rank				<.001
Junior	1,125 (49.7)	80 (22.1)	1,045 (54.9)
Middle	878 (38.8)	152 (42.0)	726 (38.2)
Vice-senior	180 (8.0)	94 (26.0)	86 (4.5)
Senior	36 (1.6)	31 (8.6)	5 (0.3)
Others (e.g., uncertain)	45 (2.0)	5 (1.4)	40 (2.1)
Marital status				<.001
Single	616 (27.2)	54 (14.9)	562 (29.5)
Married	1,591 (70.3)	299 (82.6)	1,292 (67.9)
Others (e.g., separated/divorced, widowed)	57 (2.5)	9 (2.5)	48 (2.5)
Living with elderly people				.174
No	719 (31.8)	126 (34.8)	593 (31.2)
Yes	1,545 (68.2)	236 (65.2)	1,309 (68.8)
Having children under 18 years old				.005
No	932 (41.2)	125 (34.5)	807 (42.4)
Yes	1,332 (58.8)	237 (65.5)	1,095 (57.6)
Attitude toward the preferred timing of taking up COVID-19 vaccination				.052
At the soonest	294 (13.0)	55 (15.2)	239 (12.6)
Wait-and-see	1,852 (81.8)	279 (77.1)	1,573 (82.7)
As late as possible	38 (1.7)	8 (2.2)	30 (1.6)
Avoid vaccination as much as possible	72 (3.2)	17 (4.7)	55 (2.9)
Definitely not	8 (0.4)	3 (0.8)	5 (0.3)

^¶Chi-square test or t-test when applicable.

Prevalence of BICV in various contexts (Figure 1)

First, the highest prevalence of BICV was >70% for the free vaccination involving 80% effectiveness plus rare mild side effects (S1: doctor/nurse: 75.1%/68.0%; p = .007), followed by that involving government/hospital recommendations (S9: doctor/nurse: 72.7%/71.2%; p = .572). Second, regarding self-paid vaccinations, the prevalence dropped to about 60% in the two similar contexts of 80% effectiveness/rare mild side effect (i.e., S2: doctor/nurse: 64.6%/56.5%; p = .004, and S10: doctor/nurse: 58.0%/57.4%; p = .833). Third, the other six scenarios involving either 50% effectiveness or common mild side effects exhibited a low prevalence of BICV of <50% (18.5–46.4%), of which no statistically significant differences between doctors and nurses were found; the intentions of self-paid vaccination were lower than those of free vaccination.

Figure 1. Prevalence of behavioral intention of COVID-19 vaccination (BICV) under different scenarios (p refers to chi-square test of the prevalence of BICV between doctors and nurses)

Rating on perceived impacts of various attributes on COVID-19 vaccination decision

The group of attributes about scientific proof (completion of Phase III clinical trial and approval of SFDA) was given the highest rating (mean: 7.6 to 8.4; range = 1–10), followed by the group of performance-related attributes (in descending order), namely, effectiveness, side effect, reported news about severe side effects, duration of protectiveness (mean: 7.0 to 8.2). The group of attributes that were rated lower in importance included vaccination at the working hospitals, convenience of vaccination, and cost (mean: 5.5 to 6.9). In general, the doctors gave higher ratings to the first two groups than the nurses (see Figure 2).

Figure 2. Mean values of the perceived impacts of attributes on COVID-19 vaccination decision among doctors and nurses (scale ranged from 0 to 10; p refers to t-test on score differences between doctors and nurses)

Preferred timing of COVID-19 vaccination

About 13% of the participants would like to take up COVID-19 vaccination at the soonest; the majority (81.8%) would wait and see until gaining comprehensive knowledge about the vaccines’ effectiveness and safety; only 5.3% were reluctant to take up COVID-19 vaccination (as late as possible/avoid vaccination as much as possible/definitely not). Doctors and nurses showed no statistically significant differences (see Table 1).

The HBM constructs

The ranked mean values (range = 1 to 5) of the HBM construct were 3.9 for perceived severity, 3.6 for perceived benefits, 3.4 for self-efficacy, 3.2 for perceived barriers, and 2.8 for perceived susceptibility. Regarding cues to action, the mean value was 3.3/3.2 for browsing COVID-19 vaccination information via governmental/other social media; about half of acquainted HCWs (49.1%) and non-HCWs (46.9%) showed supportive attitudes toward COVID-19 vaccination. Nurses showed higher levels of perceived susceptibility, perceived severity, perceived benefits, and perceived barriers than doctors; no statistical differences were found for other constructs (see Table 2).

Table 2. Descriptive statistics of constructs of the Health Belief Model^#

	All (n = 2,254)	Doctors (n = 362)	Nurses (n = 1,902)	p
	Mean (SD)	Mean (SD)	Mean (SD)
Perceived susceptibility	2.8 (0.9)	2.6 (0.9)	2.8 (0.9)	<.001
Perceived severity	3.9 (0.8)	3.8 (0.7)	3.9 (0.8)	.028
Perceived benefits	3.6 (0.7)	3.5 (0.8)	3.6 (0.7)	.015
Perceived barriers	3.2 (0.8)	3.0 (0.8)	3.2 (0.8)	<.001
Cues to action
Supportive attitudes toward COVID-19 vaccination from HCWs [n (%)]	1,111 (49.1)	181 (50.0)	930 (48.9)	.700
Supportive attitudes toward COVID-19 vaccination from non-HCWs [n (%)]	1,061 (46.9)	167 (46.1)	894 (47.0)	.761
Browsing information about COVID-19 vaccination via governmental social media	3.3 (0.9)	3.3 (0.9)	3.3 (0.9)	.626
Browsing information about COVID-19 vaccination via other social media	3.2 (0.9)	3.2 (0.9)	3.2 (0.9)	.968
Self-efficacy	3.4 (0.7)	3.5 (0.8)	3.4 (0.7)	.496

^#The score range was 1 to 5.

Factors of BICV (S1, S2)

In general, participants of specific departments (e.g., infectious diseases) were more likely than others to show two types of BICV, while the reverse trend was obtained among participants of specific hospitals and job ranks (e.g., vice-senior doctors). The other socio-demographics were not significantly associated with the two BICV variables (see Table 3).

Table 3. Associations between background characteristics and vaccination intention

	Doctors (n = 362)		Nurses (n = 1,902)
	Free + 80%+ rare mild side effects	600 RMB + 80%+ rare mild side effects	Free + 80%+ rare mild side effects	600 RMB + 80%+ rare mild side effects
	ORc (95% CI)	ORc (95% CI)	ORc (95% CI)	ORc (95% CI)
Sex
Female	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Male	1.03 (0.64–1.66)	1.10 (0.71–1.69)	0.87 (0.50–1.51)	1.40 (0.84–2.32)
Age	0.99 (0.96–1.02)	0.99 (0.97–1.02)	1.00 (0.99–1.01)	1.00 (0.99–1.01)
Hospitals
Hospital A in Yunnan	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Hospital A in Guangdong	NA	NA	0.94 (0.65–1.37)	0.88 (0.62–1.26)
Hospital B in Guangdong	0.90 (0.30–2.73)	0.79 (0.30–2.06)	0.93 (0.63–1.38)	0.65 (0.44–0.94)*
Hospital A in Hunan	0.43 (0.17–1.09)	0.55 (0.24–1.26)	0.67 (0.45–1.00)*	0.46(0.31–0.68)***
Hospital B in Hunan	0.90 (0.39–2.07)	0.89 (0.43–1.86)	1.13 (0.86–1.49)	0.88 (0.68–1.15)
Department
Internal medicine	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Surgery	0.75 (0.36–1.56)	0.97 (0.49–1.91)	1.09 (0.85–1.41)	0.94 (0.74–1.20)
Gynecology and obstetrics	0.90 (0.37–2.20)	1.08 (0.48–2.43)	1.62 (0.85–3.09)	2.00 (1.08–3.71)*
Pediatrics	1.12 (0.56–2.27)	1.31 (0.70–2.45)	0.94 (0.67–1.31)	0.76 (0.56–1.05)
Infectious diseases	1.09 (0.29–4.18)	1.00 (0.31–3.20)	1.72 (1.03–2.88)*	0.80 (0.52–1.24)
Emergency	1.64 (0.34–7.82)	3.11 (0.66–14.72)	0.90 (0.62–1.31)	0.95 (0.66–1.36)
Otolaryngology	0.39 (0.11–1.36)	0.75 (0.22–2.56)	1.45 (0.67–3.15)	0.89 (0.45–1.77)
Others	1.39 (0.59–3.28)	1.75 (0.82–3.77)	1.09 (0.80–1.49)	1.13 (0.84–1.51)
Job seniority rank
Junior	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Middle	0.58 (0.29–1.17)	0.75 (0.42–1.36)	1.08 (0.88–1.32)	1.00 (0.83–1.21)
Vice-senior	0.44 (0.21–0.91)*	0.55 (0.29–1.03)	1.12 (0.70–1.81)	1.20 (0.76–1.88)
Senior	0.67 (0.24–1.86)	0.99 (0.40–2.46)	0.73 (0.12–4.39)	3.13 (0.35–28.13)
Others (e.g., uncertain)	0.13 (0.02–0.85)*	0.27 (0.04–1.72)	1.14 (0.57–2.26)	1.31 (0.68–2.51)
Marital status
Single	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Married	1.17 (0.61–2.24)	1.57 (0.87–2.82)	0.96 (0.77–1.18)	0.98 (0.80–1.20)
Others (e.g., separated/divorced, widowed)	3.08 (0.35–26.75)	1.60 (0.36–7.07)	1.00 (0.53–1.89)	0.82 (0.46–1.48)
Living with elderly people
No	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Yes	0.98 (0.59–1.62)	0.92 (0.58–1.45)	0.82 (0.66–1.01)	0.88 (0.72–1.07)
Having children under 18 years old
No	Ref = 1.0	Ref = 1.0	Ref = 1.0	Ref = 1.0
Yes	1.29 (0.79–2.11)	1.16 (0.74–1.82)	1.07 (0.88–1.30)	1.02 (0.85–1.23)

*p < .05; **p < .01; ***p < .001.

Adjusted for the background variables and in general, the scales regarding perceived severity, perceived benefits, self-efficacy, and the individual cues to action items were significantly associated with the two BICV variables among doctors and nurses, while perceived susceptibility (except for S2 among nurses) and perceived barriers were not significantly associated with the two BICV variables among both doctors and nurses (see Table 4).

Table 4. Adjusted individual associations between constructs of the Health Belief Model and vaccination intention

	Doctors (n = 362)		Nurses (n = 1,902)
	Free + 80%+ Rare mild side effects	600 RMB + 80%+ Rare mild side effects	Free + 80%+ Rare mild side effects	600 RMB + 80%+ Rare mild side effects
	ORa (95% CI)	ORa (95% CI)	ORa (95% CI)	ORa (95% CI)
Perceived susceptibility	1.09 (0.82–1.45)	1.00 (0.77–1.30)	1.09 (0.98–1.22)	1.14 (1.02–1.26)*
Perceived severity	2.42(1.64–3.58)***	1.90(1.36–2.66)***	1.76(1.55–2.01)***	1.70(1.50–1.93)***
Perceived benefits	2.73(1.84–4.06)***	2.29(1.62–3.24)***	2.53(2.17–2.96)***	2.03(1.77–2.33)***
Perceived barriers	1.04 (0.75–1.45)	1.01 (0.75–1.35)	1.03 (0.91–1.16)	1.05 (0.94–1.18)
Cues to action
Supportive attitude toward vaccination from HCWs	4.16(2.33–7.43)***	3.71(2.23–6.16)***	3.49(2.83–4.31)***	3.06(2.53–3.72)***
Supportive attitude toward vaccination from non-HCWs	3.60(2.02–6.41)***	3.99(2.39–6.66)***	3.24(2.63–4.01)***	3.15(2.59–3.83)***
Browsing information about COVID-19 vaccination via governmental social media	1.89(1.38–2.58)***	1.87(1.41–2.48)***	1.30(1.16–1.45)***	1.41(1.26–1.56)***
Browsing information about COVID-19 vaccination via other social media	1.75(1.29–2.39)***	1.58(1.20–2.07)**	1.30(1.16–1.45)***	1.35(1.22–1.51)***
Self-efficacy	2.66(1.81–3.90)***	1.55 (1.15–2.11)**	2.39(2.03–2.81)***	2.10(1.81–2.43)***

*p < .05; **p < .01; ***p < .001. The adjusted models were adjusted for studied hospital, sex, age, department, job seniority rank, marital status, whether living with elderly people, and whether having children under 18 years old.

Discussion

One of the key findings is that about 80%/70% among doctors/nurses would take up COVID-19 vaccination having good effectiveness (80%) and safety (rare mild side effects), corroborating a U.K. study reporting a similar prevalence of vaccination intention of 78%/61% among doctors/nurses.¹⁹ Although some COVID-19 vaccines (e.g., Pfizer-BioNTech and Moderna) have reported efficacy of >90%,⁵ they are unlikely to be used by HCWs in China who are likely to take up the Chinese vaccines with efficacies of 50.3% and 79.3%.⁵ Thus, the COVID-19 vaccination coverage rates among Chinese nurses/doctors may approach 70–80% if the Chinese vaccine of higher efficacy of 79.3% (Sinopharm) were used. All in all, health promotion is still needed. Corroborating the aforementioned U.K. study¹⁹ and others focusing on influenza vaccination,^29,30 doctors sampled in this study were more motivated to take up COVID-19 vaccination than nurses. It might be due to stronger concerns toward vaccines’ safety among nurses than among doctors;¹⁹ the involved reasons are complicated and need further exploration. Health promotion targeting nurses may require a higher priority.

If the COVID-19 vaccination was recommended by the government/hospitals, the prevalence of intention of free/self-paid optional COVID-19 vaccination (S9 and S10) was similar to that of the most optimum scenario in this study (S1). It is plausible that HCWs in China tend to comply with official recommendations; they might consider COVID-19 vaccination as a job and/or social responsibility. Such recommendations may improve vaccination coverage among HCWs. Relatedly, the potential strong impacts of governmental recommendation may be based on the strong trust/satisfaction toward the governmental COVID-19 policies, which have successfully controlled COVID-19 in China.³¹ Trust toward the government was also positively associated with influenza vaccination behavior/intention.^32-34 Equally strong impacts on HCWs may not be found in countries where there is a lack of trust toward governments’ COVID-19 policies and performance. Building up or restoration of trust toward the government may be a dire driving force promoting BICV among HCWs.

It is noteworthy that there is in general a discrepancy between intention and behavior, which can be as large as 70%.³⁵ Thus, the actual COVID-19 vaccination coverage rate remains uncertain. A charge of 600 Yuan (about USD90) may reduce the prevalence of intention under the high prevalence scenarios (S1 and S9) by about 10%; either low effectiveness of 50% or frequent side effects (even mild side effects) might pull down the prevalence substantially; inferior vaccine performance might hence result in a low vaccination coverage. Thus, the prevalence is sensitive to cost, effectiveness, and safety. The consideration of such dimensions in the prevalence estimation is a strength of the present study. Interpretations of the other data may keep such considerations in mind.

The present study examined how much specific attributes of COVID-19 vaccines might influence BICV. Completion of Phase III clinical trials and SFDA approval were the two most influential attributes; another consistent finding is that the majority (>80%) would wait until they have gained a good understanding of the vaccines’ effectiveness. Besides these two attributes, effectiveness and safety mattered most and about equally; news about severe side effects, even rare, would be about equally impactful on BICV. In general, doctors paid stronger attention to the afore-mentioned attributes than nurses in this study; in contrast, the logistics (e.g., convenience and cost) were less important. Health promotion should thus present solid evidence of effectiveness and safety; recommendations given by famous professionals and celebrities in vaccine marketing campaigns³⁶ and clarification of exaggerated or false information about COVID-19 vaccines in social media³⁷ are potentially important.

The HBM factors of perceived severity/perceived benefits/self-efficacy/cues to action (via acquaintances and social media) were in general positively associated with intention of COVID-19 vaccination. Programs promoting BICV among HCWs may hence modify these constructs (e.g., impacts of COVID-19, protectiveness for oneself, patients and colleagues, psychological relief, contribution to controlling the COVID-19 pandemic, easiness to take up COVID-19 vaccination, and supportive news via social media). Under the most optimum scenario in this study (S1), some factors (e.g., perceived benefits and cues to action) were significant among nurses but not among doctors. It is uncertain whether the differential associations were due to the smaller sample size of the doctor group (362 versus 1,902). The findings need to be confirmed in future studies involving larger samples. Longitudinal studies testing such factors’ prediction over actual COVID-19 vaccination behavior are useful.

The observed level of perceived susceptibility was relatively low; the higher level among nurses than doctors is understandable as nurses might have closer contact with patients. The level of perceived barriers was modest and might not be of major concern. The levels of perceived severity, perceived benefits, and self-efficacy were moderate. The relatively higher level of perceived severity was possibly due to the severe stress experienced during the initial phase of the COVID-19 outbreak in China. Perceived short-term benefits might not be very high, as the pandemic seemed to have been put under control in China.³⁸ The moderate level of self-efficacy might be a mixture of the high priority of vaccination among HCWs as a national policy and experimental use that have started among HCWs in China,³⁹ amid uncertainties of unprecedented arrangements. There are substantial rooms for modifications. Since the levels of such perceptions and their relationships with intention of COVID-19 vaccination may vary across countries, further studies are needed to confirm our findings.

The study has the strengths of being one of the few that investigated factors of BICV among HCWs and involved a relatively large sample size. It has several limitations. First, the cross-sectional study design did not allow for causal inferences. Second, this study only involved five hospitals in urban areas of three Chinese provinces. Although the response rate was reasonable (73.7%), generalization to other Chinese provinces and other countries should be made cautiously. Third, since no HBM constructs in the context of COVID-19 existed, a panel of experienced researchers constructed such measurements of the HBM with references to previous vaccination studies.^40-42 Fourth, social desirability bias may exist as vaccination behaviors among HCWs seems socially desirable. Fifth, the sample sizes of the two occupational groups were uneven (a smaller sample of doctors) and may affect the significance of the factors examined in the two groups. Sixth, other potential determinants such as inter-personal factors and psychological factors were not included in this study. Moreover, factors’ associations with intention might not predict the actual behavior. A large proportion of participants in this study showed a wait-and-see attitude toward COVID-19 vaccination. However, we did not measure how long they plan to wait for a decision about vaccination or under what conditions they would take action to receive COVID-19 vaccination. Lastly, the highest specified efficacy was only 80% in this study, while some COVID-19 vaccines have reported efficacies of >90% (Pfizer-BioNTech and Moderna),⁵ as the study was designed prior to the announcement of any Phase III clinical trial results. However, the two Chinese vaccines approved in China (Sinovac and Sinopharm) have efficacies of 50.4% and 79.3%, respectively,⁵ which is within the efficacy range specified in this study. Both Chinese vaccines have also been approved in countries such as Brazil and Egypt.⁵ As possibly most of COVID-19 vaccination among HCWs in China would use the Chinese vaccines, our specified efficacy range (50–80%) is applicable to the Chinese context. Generalization to the vaccination intention among HCWs in other countries may, however, take into account more efficacious options.

In summary, the prevalence of BICV among HCWs in China varied according to the combinations of the vaccines’ effectiveness, safety, and cost. It might be reasonably high if the vaccines are free and have good effectiveness and safety, but health promotion is still needed as over 30% of the nurses indicated no vaccination intention and given the discrepancy between intention and behavior. Health promotion targeting HCWs may make use of solid scientific evidence, governmental/hospital recommendations, and modification of HBM factors (e.g., perceived severity, perceived benefits, self-efficacy, and cue to action). Randomized controlled trials are needed to create evidence-based intervention programs for increasing BICV. The HBM is partially supported and awaits confirmation from longitudinal studies predicting actual COVID-19 vaccination among HCWs. Such studies may involve comparisons across study populations and countries.

Authors’ contributions

Conceptualization: JTFL and YY; Methodology: JTFL and RS; Investigation: JTFL, RS, XC, LPL, LJL, and XJC; Software: YY; Formal analysis: YY; Data curation: YY and RS; Validation: JTFL; Resources: JTFL; Writing-original draft: YY and JTFL; Writing-review & editing: YY and JTFL; Supervision: JTFL; Funding acquisition: JTFL.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Ethics statement

The study was approved by the Survey and Behavioral Research Ethics Committee of the Chinese University of Hong Kong (No. SBRE-20-094).

Acknowledgments

We would like to thank, Lihui Zhu of Hunan Children’s Hospital, Huifang Tan of The First Affiliated Hospital of Nanhua University, Zepeng Huang of the Second Affiliated Hospital of Shantou University Medical College, Ling Guo of Yunnan Kungang Hospital, Lijun Zhu of Dali Bai Autonomous Prefecture People’s Hospital, and Huixia Lu of The First Affiliated Hospital of Dali University for their assistance in data collection.

Additional information

Funding

The study was supported by the Internal research funding of the Centre for Health Behaviour Research, the Chinese University of Hong Kong.