The influence of social network on COVID-19 vaccine hesitancy among healthcare workers: a cross-sectional survey in Chongqing, China

ABSTRACT

The factors that lead to coronavirus disease 2019 (COVID-19) vaccine hesitancy among health-care workers (HCWs) are unclear. We aimed to identify the factors that influence HCWs’ hesitancy, especially the influence of their social network. Using an online platform, we surveyed HCWs in Chongqing, China, in January 2021 to understand the factors that influence the COVID-19 vaccine hesitancy among HCWs. Proportional allocation stratified sampling method was used to recruit respondents. Multivariable logistic regression and social network analysis (SNA) were used to analyze the influence factors. A total of 5247 HCWs were included and 23.3% of them were vaccine-hesitant. Participants were more hesitant if they had chronic diseases (OR = 1.411, 95% CI: 1.146–1.738), worked in tertiary hospitals (OR = 1.546, 95% CI: 1.231–1.942), and reported a history of vaccine hesitancy (OR = 1.637, 95% CI: 1.395–1.920) and refusal toward other vaccines (OR = 2.433, 95% CI: 2.067–2.863). The participants with a social network to communicate COVID-19 immunization were less hesitant (OR = 0.850, 95% CI: 0.728–0.993). Several influential members with social networks were found in SNA. Most of these influential members in the networks were department leaders who were willing to get COVID-19 vaccines (P < .05). Hesitant subgroups among Chinese HCWs were linked to the lack of a social network to communicate COVID-19 immunization. Our findings may lead to tailored interventions to enhance COVID-19 vaccine uptake among HCWs by targeting key members in social network.

Graphical abstract

KEYWORDS:

• coronavirus
• COVID-19
• vaccine
• vaccine hesitancy
• healthcare worker (HCW)
• social network analysis (SNA)

Background

The number of coronavirus disease 2019 (COVID-19) cases was still increasing in December 2020 in China despite China’s strict pandemic measures.¹Vaccine, which provides immunity at individual and population-level,² promises hope to bring COVID-19 pandemic under control.³ It has become a critical tool in the battle against COVID-19.⁴ On the last day of December, 2020, an inactivated virus vaccine obtained conditional marketing authorization from the National Medical Products Administration (NMPA) of China.⁵ However, to successfully guard the population against COVID-19 does not only depend on the efficacy and availability of vaccines but also the vaccine acceptance level among the population.⁶ With more and more vaccines being approved and becoming readily available, the World Health Organization (WHO) expresses concerns on vaccine hesitancy which presents the next biggest challenge in the fight against the pandemic.⁷ Whether or not to get the COVID-19 vaccine has become a tough decision faced by the public.

As the most trusted source of vaccine-related information for the public, health-care workers (HCWs) have a strong influence on vaccination decisions.⁸ If HCWs recommend vaccines less frequently to their patients, it would undermine the confidence and lead to vaccine hesitancy among the general population.^6,9–11 However, HCWs are losing confidence in vaccination for themselves as well as for their children and patients.^9,12–14 Hesitancy against new vaccines is of particular concern.¹⁵ The decreasing influenza vaccination rate among HCWs has become an international health issue,¹⁶ and the COVID-19 vaccination among HCWs may also be a thorny issue. Mathematical models estimated that if the COVID-19 vaccine is 80% effective, its coverage in the population must be at least 55–82% in order to slow the transmission and decreasing the risk of infection.^17,18 Before COVID-19 vaccines were available, several studies showed that the willingness of HCWs to receive COVID-19 vaccines varied from 27.7% in the Democratic Republic of the Congo to 78.1% in Israel.^19,20 In China’s indigenous studies, the willingness rate among HCWs also ranged from 22.2% to 79.1%.^6,21–28 Although HCWs are more willing to be vaccinated against COVID-19 compared to the general population, such a vaccine acceptance level is still insufficient to protect HCWs against COVID-19 and ensure the smooth running of the health-care system.⁶ Vaccine hesitancy is defined as “a delay in acceptance or refusal of vaccination despite availability of vaccination services” by the Strategic Advisory Group of Experts on Immunization (SAGE) Working Group.²⁹ In China, COVID-19 vaccination is strongly encouraged but not mandatory.³⁰ Multiple survey studies showed that the COVID-19 vaccine hesitancy among the general population changed over time in many countries.¹⁸ In light of this, examining the hesitancy toward COVID-19 vaccines among HCWs after the vaccine receives conditional marketing authorization has broader policy implications in mass vaccination campaigns.

In the first several months after the issuance of vaccine emergency use authorization (EUAs) by NMPA of China, surveys among HCWs in China showed that the actual acceptance rate ranged from 66.5% to 86.2%, the rates of vaccine hesitancy ranged from 13.8% to 43.5%, and the willingness to vaccinate varied considerably among different regions.^31–39 However, those surveys did not study the important role of the social factors that may influence the hesitancy. Considering that the attitude toward vaccine is not only influenced by the characteristics of each individual but also by the people around them, we introduce social network analysis (SNA) to measure the influence of social relations on vaccine hesitancy in addition to the classical multivariate analysis in order to understand the factors that lead to the hesitancy toward COVID-19 vaccines among HCWs. SNA, which offers a way of exposing and mapping the communication and information flow among people from important groups within an organization, has been commonly used to improve the effectiveness and efficiency of decision-making processes in organizations.⁴⁰ This study could potentially establish a framework to understand vaccine hesitancy among HCWs in China and provide strategies to improve COVID-19 vaccination rate for a better pandemic control.

Methods

Study design

From January 14 to 20, 2021, a cross-sectional self-administered online survey was conducted among HCWs at all hospitals in Yuzhong district, Yubei district and Shapingba district of Chongqing municipal, China. Hospitals in China are designated as primary, secondary and tertiary institutions. Primary hospitals are typically a township hospital that contains fewer than 100 beds. Secondary hospitals contain more than 100 beds, but fewer than 500. Tertiary hospitals round up the list as general hospitals at the city with a bed capacity exceeding 500. These three districts include the majority of tertiary hospitals in Chongqing, hence the social network information of HCWs from hospitals of all three care levels can be collected in the three districts.

Based on the results from our pilot study and to ensure a maximum sample size, the rate of hesitancy (p) was estimated at 20%, and the required sample size of this cross-sectional survey in each of the three districts was calculated as 1600. This experiment design allowed us to investigate the actual vaccine hesitancy (the main outcome of interest) with an allowable error of 0.1p and a cutoff corresponding to 0.05 significance level. Assuming no more than 10% loss to follow-up in each district, the required total sample size was calculated as 5280.

Proportional allocation stratified sampling method was used, and the sample size required from each hospital was determined by the number of HCWs registered at the Department of Health Administration in these districts. Quick Response (QR) code of the questionnaire was distributed in their work groups. The participation was voluntary, and informed consent was obtained on the survey platform before the participant was allowed to proceed to survey items. If the HCWs agreed to participate, he/she would scan the QR code to complete an Internet-based questionnaire on a platform called Wenjuanxing with functions equivalent to Amazon Mechanical Turk. If the respondents refused, the manager would find another HCW in their work group.

The data of 5247 HCWs who completed the questionnaire were recorded in our study, while the other uncompleted questionnaires cannot be recorded by the platform. All the records were anonymous and without any personal identifiers.

Survey

The questionnaire was presented in Chinese. According to the literature on the vaccine acceptance and social networks in China and United States,^13–16 the contents of our questionnaire included items on demographic information (gender, age, levels of hospital care, the last 6 digits of phone numbers to construct the vaccination consulting network, occupation, health related characteristics, and medical behaviors) (Table 1), the attitude, experiences and behaviors toward vaccination, the hesitancy toward COVID-19 vaccines, self-perceived risk of contracting COVID-19, and the experience of treating COVID-19 patients (Table 2). Medical behaviors included the number of patients directly contacted per day by the HCW and whether the survey takers had recommended influenza or COVID-19 vaccines to their patients. If the survey takers mentioned they had at least one friend to discuss COVID-19 vaccination, they were grouped into the class with a vaccination consulting network. Attitude toward vaccination was measured by self-perceived importance of vaccination, the confidence in general vaccine safety and effectiveness, the trust in vaccination information provided by the government and vaccination doctors, and the history of vaccine hesitancy and refusal toward other vaccines not due to valid medical reasons such as preexisting conditions or allergies. Except for the questions on vaccination hesitancy and refusal, which were answered in yes or no, the answers to all other questions were selected from one out of four categories with 1 being the most and 4 being the least. During analysis, the first two categories (1 and 2) were grouped into one class, and the rest two were grouped into another. Behaviors toward vaccination were assessed based on the immunization record of the influenza vaccine and the 23-valent pneumococcal polysaccharide vaccine (PPSV 23) in prior years. Each variable was explained in detail, and the jump and logic controls were set up in the questionnaire to reduce the measurement bias. The questionnaire consisted of 16 items regarding attitude toward vaccination. The reliability of this questionnaire was assessed with Cronbach’ α coefficient, and the total Cronbach’ α coefficient was 0.723. In addition to the good content-related validity recognized by the experts, the validity of the questionnaire was also evaluated by factor analysis. The KMO coefficient was 0.855, and the significance of Bartley spherical test was <0.01. The exploratory factor analysis extracted 16 common factors to explain the cumulative variance contribution rate of 52.2%. Based on these analyses, it is concluded that the questionnaire has good reliability and validity.

Table 1. Demographic and health-related characteristics of participants (n = 5247). Demographic and health-related characteristics of participants are shown. Many variables are significantly different in hospitals of different care levels, especially the proportion of participants with a social network to communicate COVID-19 immunization

Variable	Primary hospital		Secondary hospital		Tertiary hospital		Total		x^2
	n	%	n	%	n	%	n	%
Demographic characteristics
Gender
Male	194	15.0	193	18.6	394	13.5	781	14.9	15.492**
Female	1103	85.0	845	81.4	2518	86.5	4466	85.1
Age (years)
18~	199	15.4	155	14.9	331	11.3	685	13.1	64.878**
26~	345	26.6	282	27.2	849	29.2	1476	28.1
31~	484	37.3	382	36.8	1261	43.3	2127	40.5
41~	213	16.4	180	17.3	325	11.2	718	13.7
51~	51	3.9	38	3.7	139	4.8	228	4.4
60~	5	0.4	1	0.1	7	0.2	13	0.2
Education
High school or lower	67	5.2	24	2.3	34	1.2	125	2.4	834.354**
Diploma	590	45.5	347	33.4	366	12.6	1303	24.8
Bachelor’s degree	628	48.4	608	58.6	1968	67.6	3204	61.1
Master’s degree	12	0.9	59	5.7	459	15.8	530	10.1
Doctorate degree	0	0.0	0	0.0	85	2.9	85	1.6
Living arrangement
With children only	387	29.8	277	26.7	787	27.0	1451	27.7	32.584**
With elderly(>60 years) only	147	11.3	101	9.7	255	8.8	503	9.6
With children and elderly	394	30.4	365	35.2	862	29.6	1621	30.9
No child nor elderly	369	28.5	295	28.4	1008	34.6	1672	31.9
Occupation and medical care setting
Medical care setting
Pediatrics	484	37.3	278	26.8	991	34.0	1753	33.4	229.676**
Adult departments	508	39.2	567	54.6	1666	57.2	2741	52.2
Non-direct contact with patients	305	23.5	193	18.6	255	8.8	753	14.4
Occupation
Doctor	359	27.7	310	29.9	642	22.1	1311	25.0	517.559**
Nurse	467	36.0	522	50.3	1879	64.5	2868	54.7
Medical and technical staff	65	5.0	35	3.4	175	6.0	275	5.2
Medical laboratory staff	46	3.6	19	1.8	35	1.2	100	1.9
Pharmacist	74	5.7	32	3.1	48	1.7	154	2.9
Administrative staff	286	22.1	120	11.6	133	4.6	539	10.3
Position
Head of department	189	14.6	93	9.0	86	3.0	368	7.0	225.475**
Deputy head of department	73	5.6	44	4.2	76	2.6	193	3.7
Ordinary staff	1035	79.8	901	86.8	2750	94.4	4686	89.3
Health status and health-related characteristics
Chronic diseases
Yes	162	12.5	126	12.1	436	15.0	724	13.8	7.645*
No	1135	87.5	912	87.9	2476	85.0	4523	86.2
Health insurance
Public insurance only	1087	83.8	827	79.7	2087	71.7	4001	76.3	81.387**
Commercial insurance	210	16.2	211	20.3	825	28.3	1246	23.7
Wearing masks all the time to comply with government COVID-19 guidelines
Yes	1232	95.00	997	96.0	2805	96.3	5034	95.9	4.159
No	65	5.0	41	4.0	107	3.7	213	4.1
Constantly washing hands to comply with government COVID-19 guidelines
Yes	928	71.6	757	72.9	2180	74.9	3865	73.7	5.434
No	369	28.4	281	27.1	732	25.1	1382	26.3
History of vaccine hesitancy
Yes	516	39.8	488	47.0	1433	49.2	2437	46.5	32.221**
No	781	60.2	550	53.0	1479	50.8	2810	53.5
History of vaccine refusal
Yes	285	22.0	272	26.2	877	30.1	1434	27.3	30.784**
No	1012	78.0	766	73.8	2035	69.9	3813	72.7
Treated AEFI before
Yes	275	21.2	195	18.8	369	12.7	839	16.0	56.147**
No	1022	78.8	843	81.2	2543	87.3	4408	84.0
History of influenza vaccination
Yes	754	58.1	386	37.2	1026	35.2	2166	41.3	203.089**
No	543	41.9	652	62.8	1886	64.8	3081	58.7
Medical behavior
Number of patients directly contacted per day
0-	615	47.4	412	39.7	903	31.0	1930	36.8	140.543**
30–99	587	45.3	493	47.5	1511	51.9	2591	49.4
100-	95	7.3	133	12.8	498	17.1	726	13.8
Recommended influenza or COVID-19 vaccination to their patients
Never	201	15.5	255	24.6	1285	44.1	1741	33.2	549.874**
Occasionally	595	45.9	517	49.8	1261	43.3	2373	45.2
Frequently	501	38.6	266	25.6	366	12.6	1133	21.6
Have the vaccination consulting network
Yes	492	37.9	320	30.8	845	29.0	1657	31.6	33.351**
No	805	62.1	718	69.2	2067	71.0	3590	68.4
Total	1297	100.0	1038	100.0	2912	100.0	5247	100.0

Primary hospitals are typically a township hospital that contains fewer than 100 beds. Secondary hospitals contain more than 100 beds, but fewer than 500. Tertiary hospitals round up the list as general hospitals at the city with a bed capacity exceeding 500. COVID-19 = coronavirus disease 2019;

*P < .05; **P < .001.

Table 2. Chi-square tests of COVID-19 vaccine hesitancy for categorical variables (n = 5247). Chi-square tests of COVID-19 vaccine hesitancy for categorical variables are shown. Several variables are correlated with the hesitancy toward COVID-19 vaccines in chi-square tests (P < .05). All variables with P < .20 in chi-square tests are input into an initial logistic multivariable regression model

	No. COVID-19 vaccine hesitancy/total no. in the category (%)	χ^2	P	OR (95% CI)
Demographic characteristics
Gender
Male	157/781 (20.1)	5.217	.022	ref.
Female	1065/4466 (23.9)			1.245 (1.031–1.502)
Hospital levels
Primary hospitals	190/1297 (14.7)	80.054	.000	ref.
Secondary hospitals	238/1038 (22.9)			1.733 (1.403–2.141)
Tertiary hospitals	794/2912 (27.3)			2.184 (1.835–2.600)
Occupation and medical care setting
Medical care setting
Pediatrics	374/1753 (21.3)	19.136	.000	ref.
Adult department	703/2741 (25.7)			1.272 (1.103–1.467)
Non-direct contact with patients	145/753 (19.3)			0.879 (0.710–1.089)
Occupation
Doctor	294/1311 (22.4)	26.786	.000	ref.
Nurse	706/2868 (24.6)			1.130 (0.967–1.319)
Medical and technical staff	66/275 (24.0)			1.092 (0.805–1.483)
Medical laboratory staff	17/100 (17.0)			0.709 (0.414–1.213)
Pharmacist	50/154 (32.5)			1.663 (1.159–2.387)
Administrative staff	89/539 (16.5)			0.684 (0.527–0.889)
Age (years)
18~	140/685 (20.4)	51.484	.000	ref.
26~	400/1476 (27.1)			1.447 (1.163–1.800)
31~	533/2127 (25.1)			1.302 (1.055–1.606)
41~	115/718 (16.0)			0.742 (0.565–0.975)
51~	32/228 (14.0)			0.636 (0.419–0.965)
60~	2/13 (15.4)			0.708 (0.155–3.230)
Education
High school or lower	16/125 (12.8)	44.186	.000	ref.
Diploma	242/1303 (18.6)			1.554 (0.903–2.674)
Bachelor degree	783/3204 (24.4)			2.203 (1.296–3.747)
Master’s degree	164/530 (30.9)			3.053 (1.750–5.323)
Doctorate degree	68/85 (20.0)			1.703 (0.807–3.594)
Position
Head of department	53/368 (14.4)	25.463	.000	ref.
Deputy head of department	30/193 (15.5)			1. 094 (0.673–1.779)
Ordinary staff	1139/4686 (24.3)			1.909 (1.416–2.572)
Living arrangement
No child nor elderly	456/1672 (27.3)	24.388	.000	ref.
With children only	315/1450 (21.7)			0.740 (0.628–0.873)
With elderly(>60 years) only	120/504 (23.8)			0.833 (0.661–1.050)
With children and elderly	331/1621 (20.4)			0.684 (0.582–0.804)
Health Status and Health-Related Characteristics
Chronic diseases
No	1034/4523 (22.9)	3.370	.066	ref.
Yes	188/724 (25.9)			1.184 (0.989–1.417)
Health insurance
Public insurance only	935/4001 (23.4)	0.060	.807	ref.
Commercial insurance	287/1246 (23.0)			0.981 (0.844–1.141)
Wearing masks all the time to comply with government COVID-19 guidelines
Yes	1171/5034 (23.3)	0.053	.818	ref.
No	51/213 (23.8)			1.039 (0.753–1.432)
Constantly washing hands to comply with government COVID-19 guidelines
Yes	911/3865 (23.6)	0.649	.421	ref.
No	311/1382 (22.5)			0.942 (0.813–1.090)
Medical behaviors Number of patients directly contacted per day
0-	468/1930 (24.3)	1.822	.402	ref.
30-	594/2591 (22.9)			0.929 (0.809–1.067)
100-	160/726 (22.0)			0.883 (0.720–1.083)
Attitude on vaccination History of vaccine hesitancy
No	423/2810 (15.1)	229.716	.000	ref.
Yes	799/2437 (32.8)			2.753 (2.408–3.146)
History of vaccination refusal
Never refused a vaccine	652/3813 (17.1)	299.233	.000	ref.
Refused a vaccine	570/1434 (39.8)			3.198 (2.794–3.662)
Recognition of the importance of all vaccines
Much/lots of confidence	1210/5227 (23.2)	-	.000*	ref.
Little/no confidence	12/20 (60.0)			4.980 (2.031–12.210)
Confidence in general vaccine safety
Much/lots of confidence	1206/5213 (23.1)	10.822	.001	ref.
Little/no confidence	16/34 (47.1)			2.953 (1.501–5.809)
Confidence in the effectiveness of all vaccines
Much/lots of confidence	1213/5233 (23.3)	-	.000*	ref.
Little/no confidence	9/14 (64.3)			5.965 (1.995–17.833)
Trust in vaccination information provided by vaccination doctors
Much/lots of confidence	1214/5230 (23.2)	5.394	.020	ref.
Little/no confidence	8/17 (47.1)			2.941 (1.132–7.638)
Trust in vaccination information provided by the government
Much/lots of confidence	1212/5232 (23.2)	15.843	.000	ref.
Little/no confidence	10/15 (66.7)			6.634 (2.263–19.445)
Have heard of vaccine scandals at Changchun Changsheng Biotech Co. or Shandong Province or about hepatitis B vaccine
All/most	1167/5012 (23.3)	0.002	.966	ref.
Little/no	55/235 (23.4)			1.007 (0.739–1.371)
Treated AEFI before
No	1030/4408 (23.4)	0.092	.762	ref.
Yes	192/839 (22.9)			0.973 (0.817–1.160)
Behaviors toward vaccination
Vaccinated with PPSV23 in prior years
Received	51/320 (15.9)	10.310	.001	ref.
Did not receive	1171/4927 (23.8)			1.644 (1.210–2.234)
Vaccinated with influenza vaccine in prior years
Received	308/2166 (14.2)	169.845	.000	ref.
Did not receive	914/3081 (29.7)			2.544 (2.205–2.936)
Have treated COVID-19 patients before
Yes	27/235 (11.5)	19.175	.000	ref.
No	1195/5012 (23.8)			2.412 (1.607–3.620)
Pay attention to the relevant knowledge of COVID-19 actively
No	303/909 (33.3)	62.083	.000	ref.
Yes	919/4338 (21.2)			0.538 (0.460–0.628)
Self-perceived risk of contracting COVID-19
Little/no	904/3475 (26.0)	42.765	.000	ref.
Much/lots	318/1772 (17.9)			0.622 (0.539–0.718)
Have the vaccination consulting network
No	893/3590(24.9)	15.989*	.000	ref.
Yes	329/1657(19.9)			0.748 (0.649–0.863)

Primary hospitals are typically a township hospital that contains fewer than 100 beds. Secondary hospitals contain more than 100 beds, but fewer than 500. Tertiary hospitals round up the list as general hospitals at the city with a bed capacity exceeding 500. COVID-19 = coronavirus disease 2019; OR = odds ratios; CI = confidence interval; ref. = reference group.*Fisher’s definite probability methods was used.

In order to conduct the SNA on vaccine hesitancy, the questionnaire also collected the last six digits of the mobile phone numbers of the participants’ friends with whom they were willing to discuss COVID-19 vaccination. According to the 11-digit coding rule of China’s mobile phone number, the first three digits represent the operator, the middle four represent the region, and the last four represent a personal code. Therefore, individuals with the same last six digits in the mobile phone number were considered the same person. With the help of these individuals, a social network to communicate COVID-19 vaccination was established among the respondents, and this social network was named vaccination consulting network.

Data analysis

In this study, the percentage of HCWs who had vaccine hesitancy toward COVID-19 vaccines was first estimated. The influencing factors, particularly social relations factors that could predict COVID-19 vaccine hesitancy, were identified. A comparative analysis was conducted to examine how the social network could minimize the hesitancy toward COVID-19 vaccines.

Descriptive statistics (e.g. frequencies) were calculated for all variables first. Then, chi-square tests were used to examine the difference between participants from hospitals of different care levels and investigate the association between these variables and COVID-19 vaccine hesitancy. All variables with P < .20 in chi-square tests were input into an initial logistic multivariable regression model followed by a backward selection procedure that retained variables with P < .05 to create a final multivariable model. The regression model produced odds ratios (ORs) and 95% confidence intervals (CIs). Lastly, the factors that would influence participants’ decision on receiving COVID-19 vaccines were determined. Data were analyzed with IBM SPSS version 19 (IBM Corp., Armonk, NY), and all statistical tests were two-tailed. A further analysis by stratification was used to examine the relationship between the lack of social network and vaccine hesitancy among HCWs from hospitals of different care levels.

In SNA, the vaccination consulting network of all participants was drawn first. The vaccination consulting network was plotted by connecting individuals who communicated information on COVID-19 vaccination. Each respondent and his/her friends with whom they discuss COVID-19 vaccination formed a sub-network. With the help of the unique individuals identified by their cell phone numbers, several sub-networks were linked together, which established the vaccination consulting network. In the network, individuals with more connecting ties had greater opportunities to influence others, and the extent of influence was measured by the degree. After calculating the degree of members in the network, influential individuals, ranked by their degree in the network, were identified. By further analyzing the vaccination willingness and social characteristics within the network, we attempted to propose a way of decreasing COVID-19 vaccine hesitancy. The SNA was analyzed by UCINET version 6 (Borgatti, S.P., Everett, M.G. and Freeman, L.C. 2002. Ucinet for Windows: Software for Social Network Analysis. Harvard, MA: Analytic Technologies).

Results

Participants’ characteristics and attitudes toward vaccines

5247 HCWs were included in our study. The mean time of survey completion was about 14.6 min, the 25% quartile of the completion time was 8.6 min, the 75% quartile was 15.3 min, and the median was 11 min. Most respondents were female (85.1%), from tertiary hospitals (55.5%), had a bachelor’s degree (61.1%), and worked as a nurse (54.7%) and as the ordinary staff (89.3%) at the adult departments (52.2%). The age distribution of the participants included 13.1% aged between 18–25, 29.2% aged between 26–30, 59.3% aged between 31–40, 11.2% aged between 41–50, 4.8% aged between 51–59, and 0.2% aged 60 and above. 31.9% of the participants did not live with any child or the elderly.

About 13.8% of the participants indicated chronic diseases, and only 23.7% of the participants bought commercial insurance. 95.9% of participants wore masks all the time and 73.7% washed hands on regular basis in compliance with government’s COVID-19 guidelines. 46.5% of the participants had a history of vaccine hesitancy and 27.3% refused to receive vaccines in the past. 41.3% of the participants took influenza vaccine before, and 31.6% had a vaccination consulting network. Only 16.0% treated adverse events following immunizations (AEFI) before. While 49.4% of the participants directly interacted with 30–99 patients per day, 33.2% had never recommended influenza nor COVID-19 vaccines to their patients. Many variables were significantly different in hospitals of different care levels (P < .05), including the proportion of participants with a social network to communicate COVID-19 immunization (Table 1).

Predictors of COVID-19 vaccine hesitancy

The vaccination rate was estimated at 43.8% (2300/5247) among HCWs within one month after the vaccine approval. 76.7% (4025/5247) of the participants were classified as willing to receive COVID-19 vaccines, which included the ones who had already been vaccinated and the ones who had been scheduled or were preparing for vaccination soon (32.9%). 23.3% (1222/5247) were classified as vaccine-hesitant (18.3% were uncertain, and 5.0% were certainly unwilling). If children were eligible for COVID-19 vaccination in the future, 65.8% (3451/5247) were willing to get their current children or future children vaccinated, 30.2% (1584/5247) were uncertain, and 4.0% (212/5247) were certainly unwilling.

Several variables were correlated with the hesitancy toward COVID-19 vaccines in chi-square tests (Table 2). In multivariable analyses (Table 3), participants were more hesitant toward COVID-19 vaccines if they were pharmacist (OR = 2.005, 95% CI: 1.299–3.097), aged 26–30 (OR = 1.355, 95% CI: 1.059–1.735) or 31–40 (OR = 1.517, 95% CI: 1.157–1.990), had a chronic disease (OR = 1.411, 95% CI: 1.146–1.738), and worked in secondary hospitals (OR = 1.467, 95% CI: 1.159–1.858) or tertiary hospitals (OR = 1.546, 95% CI: 1.231–1.942), especially at adult departments of these hospitals (OR = 1.276, 95% CI: 1.085–1.501). Participants were also more hesitant toward COVID-19 vaccines if they reported a personal history of vaccine hesitancy (OR = 1.637, 95% CI: 1.395–1.920) or refusal (OR = 2.433, 95% CI: 2.067–2.863), and did not receive the influenza vaccine in prior years (OR = 1.957, 95% CI: 1.673–2.290). Participants were less hesitant toward COVID-19 vaccines if they lived with children only (OR = 0.624, 95% CI: 0.514–0.757) or with both children and the elderly (OR = 0.574, 95% CI: 0.466–0.706), recognized the importance of vaccines to health (OR = 0.328, 95% CI: 0.113–0.948), had treated patients with COVID-19 before (OR = 0.409, 95% CI: 0.267–0.625), paid attention to the relevant information of COVID-19 actively (OR = 0.665, 95% CI: 0.560–0.791), perceived a higher risk of COVID-19 (OR = 0.646, 95% CI: 0.553–0.755), and had access to the vaccination consulting network (OR = 0.850, 95% CI: 0.728–0.993). A further analysis by stratification revealed that the lack of the vaccination consulting network had an important influence on the vaccine hesitancy among HCWs from secondary hospitals (Table 4).

Table 3. Multivariable correlates of COVID-19 vaccine hesitancy. Multivariable correlates of COVID-19 vaccine hesitancy are shown. Participants are more hesitant if they have chronic diseases, work in tertiary hospitals, and report a history of vaccine hesitancy and refusal toward other vaccines. The participants with a social network to communicate COVID-19 immunization are less hesitant

Variable	Bivariate OR (95% CI)
Hospital levels
Primary hospital	ref.
Secondary hospital	1.467 (1.159–1.858)**
Tertiary hospital	1.546 (1.231–1.942)**
Working area
Pediatrics	ref.
Adult department	1.276 (1.085–1.501)*
Non-direct contact with patients	1.047 (0.788–1.390)
Occupation
Doctor	ref.
Nurse	1.061 (0.837–1.344)
Medical and technical staff	1.113 (0.769–1.610)
Medical checkers	0.798 (0.423–1.507)
Pharmacist	2.005 (1.299–3.097)*
Administrative staff	1. 018 (0.712–1.456)
Age (years)
18~	ref.
26~	1.355 (1.059–1.735)*
31~	1.517 (1.157–1.990)*
41~	0.820 (0.566–1.187)
51~	0.522 (0.308–0.887)*
60~	0.955 (0.186–4.903)
Living arrangement
No child nor elderly	ref.
with children only	0.624 (0.514–0.757)**
With elderly (>60 years)	0.927 (0.716–1.201)
With children and the elderly	0.574 (0.466–0.706)**
Chronic disease
No	ref.
Yes	1.411 (1.146–1.738)*
History of vaccine hesitancy
No	ref.
Yes	1.637 (1.395–1.920)**
History of vaccination refusal
No	ref.
Yes	2.433 (2.067–2.863)**
Approve the importance of vaccines to health
Little/no confidence	ref.
Much/lots of confidence	0.328 (0.113–0.948)*
Get a influenza vaccine in prior years
Received a flu vaccine	ref.
Did not receive a flu vaccine	1.957 (1.673–2.290)**
Have treated patients in COVID-19
No	ref.
Yes	0.409 (0.267–0.625)**
Pay attention to the relevant information of COVID-19 actively
No	ref.
Yes	0.665 (0.560–0.791)**
Self-perceived risk of contracting COVID-19
Little/no	ref.
Much/lots	0.646 (0.553–0.755)**
Have the vaccination consulting network
No	ref.
Yes	0.850 (0.728–0.993)*

Final multivariable model followed by a backward selection procedure included only those variables presented in this table. Primary hospitals are typically a township hospital that contains fewer than 100 beds. Secondary hospitals contain more than 100 beds, but fewer than 500. Tertiary hospitals round up the list as general hospitals at the city with a bed capacity exceeding 500. COVID-19 = coronavirus disease 2019; OR = odds ratios; CI = confidence interval; ref. = reference group.

*P < .05; **P < .001.

Table 4. Multivariable correlates of COVID-19 vaccine hesitancy among HCWs in different hospital levels. Multivariable correlates of COVID-19 vaccine hesitancy among HCWs in different hospital levels are shown, and the lack of the vaccination consulting network have an important influence on the vaccine hesitancy among HCWs from secondary hospitals

	Bivariate OR (95% CI)
Variable	Primary hospital	Secondary hospital	Tertiary hospital
Working area
Pediatrics	ref.	ref.	ref.
Adult department	1.068 (0.700–1.630)	0.775 (0.526–1.142)	1.442 (1.182–1.758)**
Non-direct contact with patients	1.221 (0.702–2.121)	0.902 (0.455–1.789)	1.038 (0.702–1.535)
Occupation
Doctor	ref.	ref.	ref.
Nurse	0.959 (0.617–1.489)	1.066 (0.710–1.599)	0.949 (0.753–1.195)
Medical and technical staff	0.733 (0.298–1.802)	0.837 (0.291–2.404)	1.061 (0.697–1.617)
Medical checkers	1.014 (0.359–2.861)	0.625 (0.136–2.873)	0.499 (0.185–1.343)
Pharmacist	1.441 (0.700–2.967)	0.719 (0.236–2.191)	3.254 (1.665–6.359)**
Administrative staff	0.765 (0.407–1.438)	0.774 (0.344–1.743)	1.084 (0.662–1.774)
Age (years)
18~	ref.	ref.	ref.
26~	1.292 (0.758–2.203)	1.794 (1.067–3.018)*	1.359 (0.991–1.865)
31~	1.540 (0.874–2.713)	2.078 (1.169–3.693)*	1.536 (1.093–2.159)*
41~	0.800 (0.408–1.568)	0.712 (0.356–1.425)	0.797 (0.523–1.213)
51~	0.731 (0.248–2.158)	0.197 (0.042–0.939)*	0.492 (0.281–0.863)*
60~	7.276 (0.949–55.771)	0.000	0.000
Living arrangement
No child nor elderly	ref.	ref.	ref.
with children only	0.555 (0.353–0.874)*	0.612 (0.394–0.950)*	0.617 (0.482–0.790)**
With elderly(>60 years)	0.593 (0.322–1.091)	1.000 (0556–1.797)	0.997 (0.713–1.394)
With children and the elderly	0.587 (0.361–0.955)*	0.313 (0.191–0.511)**	0.656 (0.506–0.850)*
Chronic disease
No	ref.	ref.	ref.
Yes	1.321 (0.800–2.183)	1.431 (0.847–2.417)	1.364 (1.055–1.762)*
History of vaccine hesitancy
No	ref.	ref.	ref.
Yes	2.453 (1.672–3.598)**	2.000 (1.369–2.922)**	1.425 (1.164–1.743)**
History of vaccination refusal
No	ref.	ref.	ref.
Yes	1.654 (1.116–2.451)*	3.674 (1.823–3.921)**	2.604 (2.118–3.201)**
Get a influenza vaccine in prior years
Received a influenza vaccine	ref.	ref.	ref.
Did not receive a influenza vaccine	2.088 (1.479–2.950)**	1.979(1.350–2.901)**	1.963 (1.604–2.403)**
Have treated patients in COVID-19
No	ref.	ref.	ref.
Yes	0.151 (0.020–1.148)	0.351 (0.140–0.878)*	0.455 (0.276–0.751)*
Pay attention to the relevant information of COVID-19 actively
No	ref.	ref.	ref.
Yes	0.611 (0.405–0.921)*	0.558 (0.377–0.825)*	0.731 (0.586–0.910)*
Self-perceived risk of contracting COVID-19
Little/no	ref.	ref.	ref.
Much/lots	0.627 (0.428–0.917)**	0.503 (0.349–0.725)**	0.658 (0.541–0.800)**
Have the vaccination consulting network
No	ref.	ref.	ref.
Yes	0.884 (0.624–1.250)	0.680 (0.465–0.995)*	0.889 (0.729–1.084)

Final multivariable model included only those variables presented in this table. Primary hospitals are typically a township hospital that contains fewer than 100 beds. Secondary hospitals contain more than 100 beds, but fewer than 500. Tertiary hospitals round up the list as general hospitals at the city with a bed capacity exceeding 500.

COVID-19 = coronavirus disease 2019; OR = odds ratios; CI = confidence interval; ref. = reference group.

*P < .05; **P < .001.

Factors that influence vaccination decisions

All the participants answered the question regarding their source of relevant COVID-19 information. As shown in Figure 1, social media was the most common channel of acquiring information on COVID-19 (83.8%), and other channels, such as Center for Disease Control and prevention (CDC) staffs (35.9%) or family/relatives/friends (30.8%), also accounted for a significant proportion. 32.9% of the participants who were willing to get COVID-19 vaccines indicated that several factors shown in Figure 2 would influence their vaccination decisions. The most important factor was the official call for vaccination (non-mandatory) (66.1%). The recommendations by hospital management (30.7%), by friends and relatives (27.5%), or by colleagues (11.4%) also accounted for a significant proportion. 7.8% chose to vaccinate under the pressure from colleagues. 23.3% of the participants who were hesitant about COVID-19 vaccines indicated that several factors shown in Figure 3 would influence their vaccination decisions. Their biggest concerns were the safety (61.5%) and efficacy of COVID-19 vaccines (45.5%).

Figure 1. Ways of acquiring relevant knowledge of COVID-19 vaccines ranked from the least to the most popular. 11 ways ranked by the proportion of responses are shown, and the social media is the most popular one.

Figure 2. Factors that would promote participants’ willingness to take COVID-19 vaccination ranked by the proportion of responses. 10 factors ranked by the proportion of responses are shown, and the most important factor is in response to the official call for vaccination (non-mandatory). The missing words are shown in the answers to Question 44 in the questionnaire in the Supplementary S1.

Figure 3. Factors that would matter in participants’ vaccine hesitancy toward COVID-19 vaccination ranked by the proportion of responses. 18 factors are shown, and participants’ biggest concern is the safety of vaccine. The missing words are shown in the answer to Question 45 in the questionnaire in the Supplementary S1.

SNA

1657 HCWs mentioned they have a vaccination consulting network, with a total of 4213 friends with whom they were willing to discuss COVID-19 vaccination. After screening out duplicated individuals through phone numbers and deleting isolated individuals with no social connections with others, a vaccination consulting network with 1817 members is constructed (Figure 4). Although this network is loose with a centrality of 0.3%, many sub-networks are linked together, and the vaccination information is transmitted between these sub-networks through influential members in the network.

Figure 4. The vaccination consulting network of the respondents. 1657 HCWs mention they have the vaccination consulting network, with a total of 4213 friends with whom they are willing to discuss COVID-19 vaccination. After screening for duplicated individuals through the phone numbers and deleting isolated individuals with no social connections with others, the vaccination consulting network with 1817 members is constructed.

The degree of these members in the network, which indicates their power of influence, ranged from 0 to 7 with a mean degree of 1.68. Twenty-two most influential members of the network with degrees greater than 5 were identified, and their sub-networks are shown in Figure 5. The blue nodes in Figure 5 represent the individuals who were willing to get COVID-19 vaccines, the red nodes represent the individuals who were hesitant, and the yellow nodes represent the individuals who were mentioned by the participants but did not participate in this study. In general, most of the individuals in Figure 5 were willing to get COVID-19 vaccines. The social characteristics of these 22 individuals were further analyzed by Fisher’s definite probability methods. Most of them were department leaders (40.9%), with a bachelor’s degree (54.6%), and more likely to get COVID-19 vaccines for themselves (95.5%). These individuals showed social characteristics that were statistically different from the entire group of participants (P < .05) (Table 5).

Table 5. Correlation between the 22 influential members ranked by degree in the vaccination consulting network and the entire participants. The social characteristics of the influential members found in SNA which are statistically different from the entire participants are shown, and most of them are department leaders, with a bachelor’s degree, and more likely to get COVID-19 vaccines for themselves

Variable	n	%	n	%	P
Education					.025
High school or lower	4	18.2	121	2.32
Diploma	5	22.7	1298	24.84
Bachelor’s degree	12	54.6	3192	61.09
Master’s degree	1	4.6	529	10.12
Doctorate degree	0	0.0	85	1.63
Position
Head of department	7	31.8	361	6.91	.001
Deputy head of department	2	9.1	191	3.66
Ordinary staff	13	59.1	4673	89.44
Willing to get a COVID-19 vaccine
No	1	4.6	4004	76.63	.040
Yes	21	95.4	1221	23.37

Fisher’s definite probability methods was used. Only those variables with significant difference presented in this table.

Figure 5. The vaccination consulting network of 22 participants ranks by degree. Twenty-two most influential members in the network with degrees greater than 5 are identified, and their sub-networks are shown. The blue nodes represent the individuals who are willing to get COVID-19 vaccines, the red nodes represent the individuals who are hesitant, and the yellow nodes represent the individuals who are mentioned by the participants but do not participate in this study.

Discussion

Based on our findings, the COVID-19 vaccine hesitancy rate among HCWs in China is estimated. Different from many other studies among HCWs, our investigation explores the influence of social network on vaccine hesitancy.^{6,21–28,31–39} Compared to people in western countries who are independent selves and self-centered in their self-structure, oriental selves rely on interdependent social relations. Their self-structure consists of a series of intimate people, and their attitudes and views are more prone to the influence by the relationship.⁴¹ Therefore, it is sensible and meaningful to investigate the influence of social network on vaccine hesitancy in China. Although the vaccine hesitancy will change with time, the positive role of social network in the initial stage when a new vaccine first becomes available can be used to guide future mass vaccine campaigns in China.

Vaccine hesitancy is undermining the protection of vaccine-preventable diseases in many specific subgroups across the globe. Targeting specific groups, such as HCWs, has resulted in the largest observed increase (>25%) in vaccine uptake,^42,43 and is believed to be an effective way of promoting vaccination. In this study, we surveyed the COVID-19 vaccine hesitancy among HCWs within the first month after the inactivated COVID-19 vaccine was approved, and found that 23.3% of HCWs in Chongqing, China, showed hesitancy toward vaccination for themselves and 34.2% showed hesitancy toward vaccination for their children if the kids would be eligible for COVID-19 vaccines in the future. As studies suggest, an immunity rate of 55–82% among the entire population is required for COVID-19 vaccine to guard the population against the virus.¹⁷ Given that a substantial proportion of the population are ineligible for COVID-19 vaccination due to age, immunological status or other preexisting medical conditions, a higher immunity rate is entailed among the eligible population.⁴⁴ In consideration of the exemplary role of HCWs, actions should be taken immediately to reduce the vaccine hesitancy among HCWs.

It is believed that the severity of the COVID-19 pandemic is a determining factor of vaccine uptake.⁴⁵ But contrary to the expectation, this study finds that the fear of the disease alone is not enough to promote the acceptance of COVID-19 vaccines. Similar to previous studies in Malta,¹⁰ our results also show that the likelihood of refusing the COVID-19 vaccine is associated with the likelihood of refusing the influenza vaccine, which indicates the hesitancy toward both vaccines. As such, addressing vaccine hesitancy is not a simple task as a multitude of factors can potentially influence people’s decision on whether to accept vaccination for themselves or their children.^20,46

The hesitant subgroups may be linked by geography, culture, socioeconomic and/or other factors.⁴⁷ For example, hesitant subgroups are defined by their behavioral factors and determinants categorized by complacency, confidence and/or convenience in the European Region,^20,31 whereas the decrease in public confidence in vaccines and the vaccine delivery system leads to increased vaccine hesitancy and refusal in the USA.⁴⁸ In this study, hesitant subgroups among HCWs in China are linked to the age, health status, care level of hospitals, working area, living arrangement, history of vaccine hesitancy and vaccination refusal, perceived risk of infection, and the lack of a social network to communicate COVID-19 immunization. The two biggest concerns of the hesitant participants are the safety and efficacy of the COVID-19 vaccines.

It is encouraging to find a high degree of trust in the government in spite of the vaccine scandals in China in recent years. It is also gratifying to see a lower hesitancy rate among those participants who live with children only or with both children and the elderly, as the two age groups have a high burden of COVID-19.¹⁰ However, younger HCWs are less likely to receive COVID-19 vaccines, possibly because they consider their symptoms could be mild or they are preparing for pregnancy or breastfeeding. In addition, the younger group is more likely to be influenced by the hesitant information,⁴⁹ which agrees with the prior studies on the acceptance of influenza vaccines.^10,49 The vaccine hesitancy among HCWs with chronic diseases is of greater concern because of the higher mortality rate of COVID-19 in this group.^50,51 What is more worrying is the lower influenza vaccination rate and the lower number of vaccination recommendations made by HCWs with longer experience and more direct contact with patients, especially those HCWs in tertiary hospitals. The recommendation made by HCWs is a key determinant of vaccination behaviors,⁵¹ and HCWs in tertiary hospitals are exemplars in terms of healthcare behaviors as they are the most approached and trusted sources of information in many countries.^51,52 This finding confirms the previous research results which showed that only a limited number of healthcare providers would recommend and administer vaccines to patients.⁵³ In general, it is imperative to address the vaccine hesitancy among HCWs, especially the youths and the ones with chronic diseases in tertiary hospitals.

Given the complexity of vaccine hesitancy and the limited evidence available on how to address this issue,⁴⁸ it is exciting to see the importance of the participants’ social network on vaccine hesitancy in our survey. Individuals with a social network to communicate immunization are less hesitant to get the COVID-19 vaccine. A significant portion of HCWs obtain the information of COVID-19 vaccines from the staffs at CDC, and the opinions of colleagues and friends would also influence their vaccination decisions. The threshold degree of the influential members was set at 5 in SNA in order to maximize the sphere of influence and communication efficiency for future intervention while minimizing the number of targeted members due to limited resource. Twenty-two most influential members of the network with degrees greater than 5 are identified. It is further understood that most of the individuals in the vaccination consulting network are willing to get COVID-19 vaccines, and the most influential individuals in the network are department leaders with a bachelor’s degree. Combined with the findings that HCWs who live with children are less hesitant and those who report a history of vaccine hesitancy and refusal are more hesitant, the department leaders who live with children and get influenza vaccination before could be facilitated by CDC staffs to build and grow their consultation social networks actively. Through this strategy, COVID-19 vaccination rate can be improved.

A further analysis by stratification reveals that the lack of the vaccination consulting network has a strong influence on the vaccine hesitancy only among HCWs in secondary hospitals. This finding may be attributed to the fact that HCWs in primary hospitals are deeply engaged in the vaccination campaign and members in their social network are more willing to be vaccinated. HCWs in tertiary hospitals are unfamiliar with immunization work, and members in their social network are more hesitant toward vaccines. This result does not disapprove the influence of social network on vaccine hesitancy. Instead, it points out the importance of including HCWs from secondary hospitals in the social network, particular in a network with influential members from primary hospitals, as a way of reducing vaccine hesitancy.

The advantages of our study include a large sample size with participants from hospitals at all three care levels in China, as well as that a wide range of possible correlates are examined. This study also focuses on the communication among HCWs from hospitals of different care levels in a particular geographical area, which can be used to analyze the influence of socialization on vaccination willingness. The way of distributing QR codes by hospital management ensures that the participants have an equal opportunity to participate in this study.

A key limitation of this study is the self-administered survey design. In order to ensure the quality of the survey, we set up jump and logic controls for the questionnaire. An additional limitation is the lack of available data on non-respondents. Given that the data collection occurred during the early stages after the approval of the COVID-19 vaccine, the actual acceptance of COVID-19 vaccines may be under estimated as the vaccine acceptance rate would continue to improve (e.g. with more evidence on vaccine safety and efficacy becomes available. Next, we need to further study the influence of social network on vaccine hesitancy by the snowballing method. This survey method takes the influenced individuals in the consulting network as the starting point, and continuously collects new individuals mentioned by existing individuals in the network until no more new person is mentioned. Then a network with a directed and/or ranked relation can be mapped and a more detailed evaluation to understand the influence of social network can be done.

Conclusions

COVID-19 vaccination is a key public health strategy to fight the pandemic. Our study provides an early insight into the hesitancy toward COVID-19 vaccines among HCWs. While the results indicate that many HCWs in China are willing to get vaccinated, the acceptance level needs to be further increased as HCWs is a special group facing a high risk of COVID-19 infection, and their participation in COVID-19 vaccination program would set a good example for the general population. Our results highlight that hesitant subgroups among HCWs in China are linked to the age, health status, care level of hospitals, working area, living arrangement, history of vaccine hesitancy and refusal, perceived risk of infection, and especially the lack of a social network to communicate COVID-19 immunization. It is therefore crucial to commence an information campaign based on our findings to increase the uptake of COVID-19 vaccines among HCWs as soon as possible. Department leaders with a bachelor’s degree who are willing to receive vaccines should be regarded as key figures to enhance the vaccine uptake among HCWs. Particular focus should be put on the young HCWs and the HCWs with chronic diseases who are working in tertiary hospitals, because the former will set an example for other youths and the latter have a higher risk when infected.

Author contributions

Conceptualization, Qing Wang; study design, Binyue Xu and Qing Wang; data analysis, Binyue Xu, Yi Zhang; investigation, Lei Chen, Linling Yu, Lanxin Li; resources, Yi Zhang; data curation, Binyue Xu; writing—original draft preparation, Binyue Xu; writing—review and editing, Yi Zhang, and Qing Wang; visualization, Binyue Xu; supervision, Binyue Xu; project administration, Binyue Xu; funding acquisition, Binyue Xu. All authors involved in the discussion of results and have read and agreed to the published version of the manuscript.

Ethics approval and consent to participate

The research is an interview survey in nature and hence is exempted from the IRB review. Informed consent to participate was obtained from all subjects involved in the study.

Acknowledgments

The authors wish to express great appreciation to all the participants in the hospitals for their assistance in the data collection of the survey.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data presented in this study are available on request from the corresponding author. The data are not publicly available due to participant’s privacy concerns.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/21645515.2021.2004837

Additional information

Funding

This research and the APC were funded by Health Commission of Chongqing municipal, China with [Grant number 2019QNXM032].