Beyond intention: Predicting children’s COVID-19 vaccine uptake using the theory of planned behavior

ABSTRACT

This prospective study tested if parental factors from the theory of planned behavior (TPB) predicted children’s uptake of the COVID-19 vaccine and examined whether parents’ intention to vaccinate children against COVID-19 would mediate such associations. Participants were 852 Hong Kong parents of 1076 children aged 5–12. At Time 1, parents reported on items measuring the TPB predictors (i.e. attitudes, subjective norms, and perceived behavioral control) and intention. At Time 2 (approximately 4 months after Time 1), parents reported whether their children had received the COVID-19 vaccine. Attitudes, perceived behavioral control, and intention predicted children’s actual uptake of the COVID-19 vaccine. Intention mediated the relations between two TPB predictors, namely attitudes and subjective norms, and children’s COVID-19 vaccination uptake. The TPB is considered a useful framework in the development of future COVID-19 vaccine programs for children to promote parents’ intention and the subsequent uptake of the COVID-19 vaccine among children.

KEYWORDS:

• Theory of planned behavior
• uptake
• COVID-19 vaccine
• children
• parents’ intention
• Hong Kong

Among data available from 100 countries, 21% of confirmed COVID-19 cases occurred in children and adolescents under the age of 20.¹ The World Health Organization (WHO) stated the importance of vaccinating children against COVID-19 in order to decrease community transmission risks, given the substantial increase of infection cases among children during the Omicron variant surge.² Although the emergency phase of COVID-19 is over, COVID-19 remains a threat as the virus continues to spread and evolve. Currently, the WHO recommends getting COVID-19 vaccines and boosters as it helps decrease the likelihood of emergence of new variants.³ However, according to a recent systematic review and meta-analysis, the estimated prevalence of parental acceptance to vaccinate their children with the COVID-19 vaccine across countries was 57%.⁴ As such, understanding the factors that predict children’s COVID-19 vaccination uptake is vital to boosting the vaccination rate of children further and for providing insights to other countries where parental COVID-19 vaccination hesitancy is high.

Hong Kong has experienced five waves of the pandemic. During the last and largest wave of the pandemic, the confirmed COVID-19 cases among the population in Hong Kong surged from 12,650 to 2,863,475. In particular, the cases among young children under the age of 12 rose from 655 to 225,001.⁵ Following the implementation of the COVID-19 vaccination programme for all Hong Kong residents aged 12 or above in 2021, the Hong Kong government has authorized the use of the COVID-19 vaccines for children as young as five years old, three years old, and six months in January 2022, February 2022, and August 2022, respectively. Similar to studies found in other countries, Hong Kong parents were found to have high hesitancy in getting their children vaccinated against COVID-19 when the COVID-19 vaccination program was first launched.^6,7 However, there was a sharp increase in the vaccination rate among Hong Kong children after the highly transmissible Omicron variant was imported. At present, the vaccination rate of the second and third doses is 69.9% and 27.2% for children aged 0–11 and 92.5% and 82.5% for the population aged 12 and above cases in Hong Kong.⁸ The current study was conducted during the period during which the COVID-19 vaccines were first authorized to young children and children’s vaccination rate has drastically increased in the following months. The timing of this study provided a unique context for the examination of factors that predict actual vaccine uptake among children.

While vaccination intention was found to be a predictor of subsequent vaccination behavior, COVID-19 vaccination success among children depends upon parents’ decisions.⁹ Prior studies have also found that parents’ intention to vaccinate children against COVID-19 was influenced by various factors, including a perception of threat from the pandemic as well as attitudes toward and knowledge about the vaccines.^10,11 In particular, the theory of planned behavior (TPB) has been used successfully to predict vaccination behaviors in general and parents’ intention to vaccinate their children against COVID-19.^12,13 However, no research has examined factors that predict parents’ decision to actually vaccinate their children against COVID-19. The lack of such evidences may result in the lack of strategic public health policies to promote the uptake of the COVID-19 vaccine. Our study extends the existing literature by utilizing the TPB as the theoretical framework to investigate factors that correlate with parents’ intention to let children receive COVID-19 vaccine, which in turn predicts children’s actual uptake of the COVID-19 vaccine.

The TPB postulates that behavioral intention is the most direct predictor of the actual behavior in a given health context.¹⁴ Three factors are involved in the formation of an intention to have their children to receive the COVID-19 vaccine. Specifically, a parent would (1) evaluate the risks and benefits of the vaccine for their children (i.e., attitudes), (2) consider whether their significant others would support such behavior (i.e., subjective norms), and (3) appraise their perceived control over the arrangement of their children receiving the vaccine (i.e., perceived behavioral control). Previous studies have demonstrated that intention to receive the vaccine will predict subsequent vaccine uptake.^15,16 As such, this study examined (1) whether the TPB predictors (i.e., attitudes, subjective norms, and perceived behavioral control) and intention of parents would predict children’s COVID-19 vaccine uptake, and (2) whether intention would mediate the associations between the TPB predictors and children’s uptake of COVID-19 vaccination (Figure 1). We hypothesized that the TPB predictors and intention of parents would positively predict children’s COVID-19 vaccine uptake and that intention would mediate the relations between the TPB factors and children’s COVID-19 vaccination uptake.

Figure 1. Conceptual model of the associations between TPB predictors, intention to vaccinate their children, and children’s actual uptake of the COVID-19 vaccine.

Methods

Participants and procedure

Participants were a subsample from a large-scale longitudinal survey (involving 11,141 parents of young children in Hong Kong) that investigated family health behaviors and well-being during COVID-19 (Time 1: February 2022; Time 2: May 2022). This study utilized the full dataset involving two waves of assessments to test the predictive pathways of the TPB factors, which were unique to that of the previous studies. At Time 1, shortly after the authorization of the use of the COVID-19 vaccines for children as young as 3 years old, participants were recruited to assess their attitudes, subjective norms, and perceived behavioral control, and intention to vaccinate their children against COVID-19. At Time 2, which occurred three months after Time 1, participants completed a follow-up survey to determine whether their children had received the COVID-19 vaccine. While parental hesitancy was initially high at Time 1, the number of child vaccination has drastically increased as the number of COVID-19 cases rose in the following months. Therefore, the three-month time interval between the two time points allowed the research team to examine the factors that predicted the actual uptake of the COVID-19 vaccine among children. At the end of the online survey at Time 1, parents were given the opportunity to leave their contact information if they were interested in participating or knowing more about the follow-up study. Out of the 4193 parents who provided contact information, we randomly selected and invited 2500 to complete the Time 2 survey through WhatsApp. Parents were instructed to complete the online survey if they agreed to participate in the follow-up study. Completed online surveys were received from 852 parents of 1076 children aged 5–12 (87.6% mothers, 61.9% aged 31–40 years, 46.8% having a bachelor’s degree or above, and 55.3% having a family monthly income of HK$40,000 or above (US$1 = HK$7.8). Results of Little’s MCAR test suggested that the missing was not random. Parents received tokens of appreciation for their participation at Time 2. This study received approval from the Human Research Ethics Committee at the authors’ university.

Measures

Time 1 TPB predictors

Three TPB predictors, namely parents’ attitudes (six items; Cronbach’s α = .98), subjective norms (three items; Cronbach’s α = .88) and perceived behavioral control (two items; Cronbach’s α = .76), were measured using self-devised items and items used in Chu and Liu’s¹⁷ study on a 5-point scale. A higher score indicating more positive attitudes, stronger subjective norms, and higher levels of perceived behavioral control of parents.

Time 1 parents’ intention

Parents’ intention to vaccinate children against COVID-19 was measured using the self-devised item “Without any policy restrictions, I will arrange this child to take COVID-19 vaccines” on a 5-point scale. A higher score indicated a stronger intention.

Time 2 COVID-19 vaccination

Whether children received the COVID-19 vaccine was measured using a single self-devised item “Did this child receive any COVID-19 vaccines.” Participants received a score of “0” for zero doses and a “1” for at least one dose. Parents who had plans to let their children receive the COVID-19 vaccine were excluded from the analysis.

Data analysis

First, we performed descriptive statistics and bivariate correlations for the main study variables in SPSS. Second, we calculated the intraclass correlation (ICC) for the outcome in Mplus8.3 to determine whether multilevel modeling should be used, given the data is nested in nature (i.e., parents/children nested in families).¹⁸ As the outcome is binary, we calculated the ICC using the formula ICC = between variance/(π²/3 + between variance).¹⁹ The results showed that the between variance was 93.10 and thus the ICC of the outcome variable in this study was 0.97, suggesting that a large proportion of variance differed across families and that multilevel modeling should be used. Subsequently, following Preacher et al.’s¹⁹ guidelines and Mplus syntax, we fit a 2-1-1 mediation model with a multilevel structural equation modeling (MSEM) paradigm. In this model, T1 attitude, T1 subjective norms, and T1 perceived behavioral control are the three level-2 predictors, while T1 intention to vaccinate children and T2 parents’ actual behavior of having their children vaccinated are the level-1 mediator and level-1 outcome, respectively. In the MSEM framework, the level-1 mediator and outcome will be modeled at both the within-level (i.e., within-level b path) and the between-level (i.e., between-level b path), and the mediation effect will be calculated with the b path at the between-level in this study. We examined the model controlling for relevant within-level (e.g., child sex and age) and between-level (e.g., parent age and their educational level) covariates. Since the outcome is binary, the estimated effects are logits. Therefore, we used the exponential function to convert the logits into odd ratios (ORs) to facilitate interpretation. The “Bayes” estimator was used to facilitate model convergence. Given the missing was not random, missing data was handled with listwise approach.

Results

Descriptive statistics and bivariate correlations

The results indicated that at T1 the levels of parents’ attitude (M = 1.91, SD = 1.19) and social norms about children receiving COVID-19 vaccination (M = 2.03, SD = 1.30) were low while their levels of perceived behavioral control were low-to-medium (M = 2.54, SD = 1.35). Moreover, parents’ intention to vaccinate their children at T1 was also low (M = 1.78, SD = 1.19). At T2, about 681 (63.3%) children had received vaccination, suggesting that most parents changed their mind and had their children to receive vaccination later on.

Regarding the bivariate correlations, the results showed that the attitude, social norms, and perceived behavioral control were positively related to intention (r = .86, .67, and .42, respectively; ps < .001) and children’s actual uptake (r = .44, .36, and .33, respectively; ps < .001). Moreover, intention was also positively related to children’s actual uptake (r = .41, p < .001). All coefficients were at medium effect sizes or above.

The associations between TPB predictors, intention, and actual uptake

Given that the residual variance of the between-level variance was 14.87, the model explained (93.10–14.87)/93.10 = 84.03% between-level variance of the outcome. After controlling for a number of within-family and between-family covariates, the results showed that T1 attitude and subjective norms were positively related to T1 intention (Table 1). Moreover, after controlling for the covariates, T1 attitude, perceived behavioral control, and intention (between part) were all positively related to T2 actual uptake. Since the coefficients summarized in the table were logits, we transformed them into ORs for interpretation. The results showed that the ORs for the “T1 attitude → T2 actual uptake”, “T1 perceived behavioral control x T2 actual uptake”, and “T1 intention → T2 actual uptake” links were 4.48 (95% CI = [1.84, 9.30]), 1.57 (95% CI = [1.15, 2.29]), and 3.16 (95% CI = [1.54, 7.17]), respectively. In other words, a one-point increase in parents’ attitude, perceived behavioral control, and intention (between part) at T1 increased 4.48, 1.57, and 3.16 times for parents to have their children vaccinated at T2, respectively.

Table 1. Multilevel mediation model of T1 TPB predictors and intentions on T2 actual uptake.

Predictors	T1 Intention				T2 Actual Uptake
	B^#	Posterior S.D.	p^^	95% CI	B^#	Posterior S.D.	p^^	95% CI
Within-level
Child sex (1 = boy; 2 = girl)	−0.01	0.02	.323	[−0.06, 0.03]	0.15	0.23	.245	[−0.44, 0.49]
Child age	0.02	0.01	< .001	[0.01, 0.03]	0.05	0.09	.254	[−0.10, 0.30]
T1 Intention					−0.55	0.96	.262	[−2.52, 1.12]
Residual variances	0.06	0.01	< .001	[0.05, 0.08]
Between level
Parent sex (1 = mother; 2 = father)	0.03	0.05	.242	[−0.07, 0.13]	−0.95	0.58	.065	[−2.03, 0.28]
Parent age (30 years old or below as reference)
31–40 years old	0.07	0.08	.183	[−0.10, 0.23]	1.12	0.66	.040	[−0.11, 2.39]
41 years old or above	0.05	0.09	.265	[−0.12, 0.22]	1.40	0.72	.031	[−0.05, 2.69]
Parent educational level (middle school or below as reference)
Diploma/associate bachelor	0.01	0.94	.499	[−1.85, 1.89]	−0.29	1.57	.417	[−3.49, 2.77]
Bachelor or above	−0.02	0.94	.490	[−1.91, 1.85]	−0.32	1.59	.431	[−3.46, 2.93]
Family monthly income (HK$20,000 or below as reference)
HK$20,001 – HK$40,000	−0.14	0.07	.018	[−0.27, −0.01]	0.22	0.61	.363	[−0.94, 1.35]
HK$40,001 – HK$60,000	−0.14	0.07	.032	[−0.28, .001]	−0.35	0.67	.285	[−1.86, 0.87]
HK$60,001 – HK$80,000	−0.11	0.08	.096	[−0.25, 0.05]	−0.53	0.74	.248	[−2.02, 0.89]
HK$80,001 – HK$100,000	−0.08	0.09	.203	[−0.25, 0.11]	0.53	0.88	.262	[−1.23, 2.24]
> HK$100,000	−0.07	0.08	.216	[−0.24, 0.10]	0.03	0.72	.492	[−1.48, 1.36]
Parent history of COVID-19 vaccines (0 = no; 1 = yes)	0.03	0.05	.242	[−0.07 0.13]	2.50	0.56	< .001	[1.68, 3.88]
T1 Attitudes	0.80	0.03	< .001	[0.75, 0.85]	1.50	0.43	.001	[0.61, 2.33]
T1 Subjective norms	0.15	0.02	< .001	[0.11, 0.20]	0.01	0.23	.489	[−0.43, 0.51]
T1 Perceived behavioral control	−0.03	0.02	.037	[−0.07, 0.00]	0.45	0.18	.005	[0.14, 0.83]
T1 Intention					1.15	0.40	.001	[0.43, 1.97]
Residual variances	0.31	0.02	< .001	[0.28, 0.35]	14.87	4.32	< .001	[8.38, 24.29]

^#Coefficients are logits. ^{^}the p-value in the model with Bayes as the estimator refers to one-tail p-value. Significant results are bolded.

We then examined the extent to which intention at T1 would mediate the associations between T1 TPB predictors and T2 actual uptake (Table 2 and Figure 2). The results showed that the mediation effects of intention at T1 were significant in the association between T1 attitude and T2 actual uptake (B = 0.93, Posterior SD = 0.32, p = .001; 95% CI = [0.34, 1.58]) and the one between T1 subjective norms and T2 actual uptake (B = 0.17, Posterior SD = 0.07, p = .001; 95% CI = [0.06, 0.32]). But the mediation effect of intention at T1 in the association between T1 perceived behavioral control and T2 actual uptake was not significant (B = −0.03, Posterior SD = 0.03, p = .038; 95% CI = [−0.10, 0.00]). Again, we transformed these logits into ORs. The results showed that the mediation effect of T1 intention was 2.53 (95% CI = [1.40,4.85]) in the “T1 attitude → T2 actual uptake” link and was 1.19 (95% CI = [1.06,1.38]) in the “T1 subjective norms → T2 actual uptake” link. In other words, a one-point increase in attitude and subjective norms respectively increased 2.53 and 1.19 times for parents to actually have their children vaccinated at T2 through their intention at T1.

Table 2. Summary of the mediation effects.

Mediational paths	B^#	Posterior S.D.	p^^	95% CI	OR	95% CI
T1 attitudes → T1 parents’ intention to vaccinate their children →T2 children’s actual uptake of the COVID-19 vaccine	0.93	0.32	.001	[0.34, 1.58]	2.53	[1.40, 4.85]
T1 subjective norms → T1 parents’ intention to vaccinate their children → T2 children’s actual uptake of the COVID-19 vaccine	0.17	0.07	.001	[0.06, 0.32]	1.19	[1.06, 1.38]
T1 perceived behavioral control → T1 parents’ intention to vaccinate their children → T2 children’s actual uptake of the COVID-19 vaccine	−0.03	0.03	.038	[−0.10, 0.00]	0.97	[0.90, 1.00]

^#Coefficients are logits. ^{^}the p-value in the model with Bayes as the estimator refers to one-tail p-value. Significant results are bolded.

Figure 2. Associations between TPB predictors, intention to vaccinate their children, and children’s actual uptake of the COVID-19 Vaccine.

Non-significant paths.

Discussion

COVID-19 continues to pose a significant threat as the virus spreads and evolves, and it is crucial to address this ongoing concern. The World Health Organization (WHO) recommends COVID-19 vaccines and boosters as a means to decrease the likelihood of new variants emerging. By ensuring that individuals receive all the necessary vaccine doses, including booster doses, we can further reduce the risk of new variants arising. Therefore, the findings of the present study are vital for boosting the overall vaccination rate among children and can also serve as a valuable resource for other countries facing similar challenges with parental COVID-19 vaccination hesitancy.

Prior studies have shown that the TPB predictors were associated with parents’ intention to vaccinate children against COVID-19.^20,21 Going beyond and above, our findings provided the first evidence that the TPB predictors, namely attitude and perceived behavioral control, and intention of parents to vaccinate children, were notably predictors of children’s actual uptake of the COVID-19 vaccine. Consistent with findings that vaccination intention significantly predicted subsequent vaccination behavior,²² our findings also showed that parents’ intention to vaccinate children against COVID-19 mediated the relation between two TPB predictors, namely attitude and subjective norms, and children’s actual uptake of COVID-19 vaccination.

Together, our findings highlighted the need to promote positive attitudes, create a supportive subjective norm, and increase parents’ perceived behavioral control over the arrangement of their children’s uptake of COVID-19 vaccination. Specifically, concerns about vaccine safety and potential side effects can contribute to vaccine hesitancy among parents, leading them to feel reluctant to vaccinate their children. On the other hand, subjective norms, influenced by the people in their social networks and their vaccination behaviors, can impact parents’ decision-making regarding their children’s vaccination. If parents perceive that those around them, such as family members, friends, or community members, are hesitant or have chosen not to vaccinate their children, they may be influenced by these norms. Lastly, parents may consider the ease of accessing and receiving the vaccine as a determining factor in their decision. Challenges such as limited availability, inconvenient vaccination locations, or complex appointment scheduling procedures can create barriers and contribute to hesitancy.^7,13,23 In particular, to promote parents’ positive attitude toward the COVID-19 vaccines, transparency in communicating about the beneficial effect and the safety of vaccination for children, as well as the importance of herd immunization are essential. Such information can be disseminated through people (e.g., specialist in pediatrics and online parent influencers) and organizations (e.g., schools and community family centers) trusted by parents using various formats (e.g., parent talks, leaflets, and videos). To enhance parents’ subjective norms, health service providers, schools, and community organizations may facilitate the discussion of children’s vaccination among parents. For instance, sharing among parents on their decision to vaccinate their children can be encouraged in different parent groups in the community to make their action to vaccinate their children against COVID-19 more visible to other parents. Finally, to increase parents’ perceived behavioral control, the government should facilitate vaccination for children through multiple channels with simple logistic arrangements, such as government and private health centers and clinics, and arrange school outreach vaccination services, to make vaccines more accessible to families with young children.

Three limitations of this study should be acknowledged. First, participants were recruited in one Chinese city, so our findings might not be generalized to other societies. Future studies should be conducted in other social contexts with different health care systems or political environments. Second, the homogenous use of self-report survey for the TPB predictors, intention, and vaccination behavior could make our responses vulnerable to response bias (e.g., social desirability) and consistency tendency.¹¹ Third, selection bias may exist as only around 34% of invited participants who provided contact for our follow-up study eventually completed our survey at Time 2. Specifically, the communication and invitation process was conducted solely through WhatsApp, without any personal contact initiated. This approach may have inadvertently decreased parents’ motivation to participate in the follow-up study and could have influenced the level of engagement and willingness of parents to complete the Time 2 survey. Consequently, it is likely that parents who refused to join the study might have different characteristics (e.g., stronger interests or attitudes) compared to those who responded to our invitation. Similarly, the online mode of data collection might also restrict our participants to parents who are more attentive to the social media platforms and be familiar with online surveys. It would be promising for future studies to utilize more diverse recruitment methods to achieve a more representative sample.

Acknowledgments

The authors would like to thank Miss Patricia Ng for her assistance in carrying out the research. The authors would also like to extend their gratitude and acknowledgments to all study participants for their time spent on this study.

Disclosure statement

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

Additional information

Funding

This study was partially supported by the Research Impact Clusters (funded by the Department of Early Childhood Education, The Education University of Hong Kong) led respectively by the second author and last author.