Introduction

Health education and its promotion in schools is a critical way to improve health levels and equity among children and young people (Breil and Lillich 2023; Guo et al. 2023; Knisel et al. 2020). Among the school-based programmes, health literacy (HL) is an important outcome, as it describes individuals’ cognitive and social skills to acquire, understand, and use information to promote and maintain good health (Nutbeam 1998). Lower HL is associated with increased hospitalizations, more frequent use of medical services, and undesirable health behaviours (Berkman et al. 2011; Fleary et al. 2018). There has been an increased effort on implementing HL education programmes, but the population-level HL skills are still considered insufficient (Chari et al. 2014; Chu-Ko et al. 2021; Sorensen et al. 2015). A lack of specific and consistent intervention targets is a limitation of HL programmes, in which some interventions were not built on empirical evidence (Larsen et al. 2022). Identifying factors associated with HL for the target population (e.g., primary school students) and developing interventions based on the evidence are ways to address the challenges in HL interventions.

Functional HL refers to individuals’ ability to understand and apply health knowledge. Compared with HL, functional HL is focused on applying individuals’ knowledge in specific situations (e.g., understanding what amount of meat is good for health), which makes it a desirable outcome of HL intervention. A large body of research has used HL and functional HL interchangeably, ignoring that they are two distinct concepts (Rocha and Lemos 2016). The few studies on functional HL found that it is associated with better health-related outcomes (e.g., obesity) and quality of life (Bhagat et al. 2018; Chari et al. 2014; Sharif and Blank 2010).

There has been little research focused on HL or functional HL in Hong Kong (Huang et al. 2023). The current study is a part of the GoSmart project, an initiative to promote health among local primary and secondary school students. This study focused on the primary school sample to examine factors associated with children’s HL. In addition, the current study  examined the potential pathways to improve children’s functional HL, by exploring the underlying mechanism of functional HL. There were three research questions: (1) What factors are associated with primary school students’ health literacy and functional health literacy? (2) Is health literacy associated with functional health literacy? (3) Is health literacy a mediator in the underlying mechanism of functional health literacy?

Methods

Data collection

A letter of invitation was sent to 18 primary school principals requesting permission of participating the current study for the 2021/2022 school year. Once schools gave their consent, hard copies of the questionnaires were delivered to schools. The study used baseline data from primary school students. Parent consent was collected by schools. Participation in the study was voluntary, and students were permitted to leave items blank if they were not willing to answer.

The target population was students between grades 4 and 6. A total of 762 students in the 18 intake schools participated in the study. After removing missing data, there were 405 responses left for a complete case analysis. The final sample of children were aged from 8.74 to 12.16 years (M = 10.0, SD = 0.8), and 54.6% were female (n = 221).

Survey instruments

Participants were invited to complete a self-administered questionnaire at the beginning of the school year. The questionnaire had 49 questions including demographic characteristics, health literacy measures, self-rated health, source of health information, screen time, health behaviours, and cognitive ability. The estimated completion time was 20–30 min.

Measures

The primary outcomes were subjective HL and functional HL. Subjective HL was measured by the Health Literacy Survey for Children (HLS-Child-Q15) scale. The HLS-Child-Q15 scale (Hahnraths et al. 2021) consists of 15 items assessing the self-perceived ease or difficulty in finding, understanding, appraising, and applying health information. Responses were based on a four-point Likert scale ranging from 1 = “very difficult” to 4 = “very easy”, and the measurement score was the sum of all item scores. Higher scores represented higher health literacy. Responses were recorded in both linear and binary form. Children who scored above and below the sample average level were considered as having “high” and “low” health literacy, respectively. Functional HL was assessed by a reading comprehension test with three questions (nine sub-items), with each sub-item scored as 1. The questions required students to apply their health knowledge in real-life contexts. Responses were recorded in both linear and binary form. A score of 9 was counted as a high functional health literacy level and a score below 9 was a moderate level.

We measured a wide range of factors of interest: (1) children’s self-perceived physical health level, with responses recorded as moderate–poor, good, and very good; (2) the number of sources where children obtain health information, which was assessed by a question with 27 sub-items. Children were asked to indicate the sources they used, with the total number of sources as the measurement score; (3) screen time, in which students were asked to indicate the number of hours spent on TV, games, and social media during school days and working days, with responses recorded as ≤ or > 2 hours following a previous HL study among Hong Kong students (Huang et al. 2023); (4) healthy behaviours, including care in hand washing and dental cleaning, breakfast habits, physical activity per week, and sleeping hours (sleeping hours were recorded on a continuous scale, and the remaining measures were collapsed into always careful/not always careful, having breakfast every day/not every day, and sufficient/insufficient physical activity following the distribution of data and the World Health Organization (WHO) guideline (World Health Organization 2022); and (5) cognitive ability, in which students’ self-efficacy in general problem-solving and motivation to learn health knowledge were measured by individual single items, with responses coded as disagree or agree.

In terms of demographic characteristics, age, gender, students’ grades at school, and the Family Affluence Scale (FAS) were measured. The FAS is a well-validated measure of the socio-economic status of children (Andersen et al. 2008; Liu et al. 2012). The FAS score is calculated by summing individual item scores, with scores of 0–3 representing low FAS, and 4–7 representing medium-high FAS.

Statistical analysis

Data were analysed using R (version 4.3.1). Logistic regression was used to explore factors associated with HL and functional HL, with high and low HL and functional HL as the outcomes. Bivariate linear regression was used to examine the linear association between HL and functional HL. Mediation analysis was used to examine whether health literacy is involved in the underlying mechanism of functional HL, with the factors associated with functional HL as predictors, scores of HL as the mediator, and scores of functional HL as the outcome.

Results

Descriptive statistics

Table 1 illustrates the socio-demographic characteristics and the descriptive statistics of the primary school students (N = 405). The overall mean age of children was 10.0 years, and 54.6% were female. The majority of children (75.1%) had relatively low family socio-economic status. The average health literacy score was 47.6 (SD = 5.8) and functional health literacy score was 8.1 (SD = 1.4).

Table 1 Descriptive statistics of children overall, with high and moderate health literacy, and with high and moderate functional health literacy
Full size table

Factors associated with health literacy

Children’s self-rated physical health level (good: OR = 2.23, p = 0.004; very good: OR = 4.09, p < 0.001), number of health information sources (OR = 1.10, p < 0.001), care in hand washing (OR = 1.83, p = 0.003) and dental cleaning (OR = 1.76, p = 0.005), and self-rated cognitive ability (efficacy: OR = 4.45, p < 0.001; motivation: OR = 2.61, p < 0.001) showed significant positive associations with having a high HL level (Table 2). A dose–response relationship was found between self-rated physical health level and high health literacy—the association was stronger when students felt they had very good physical health than when they rated their physical health level as good. Having breakfast every day was associated with a higher likelihood of having high health literacy levels only after adjusting for confounding variables (adjusted OR = 1.70, p = 0.035). Physical activity (OR = 1.30, p = 0.185) and sleeping hours (OR = 1.14, p = 0.176) showed positive associations with high HL levels, but the associations were non-significant.

Table 2 Odds ratios of logistic regression with having high health literacy level as the outcome
Full size table

Spending more time on TV (school days: OR = 0.60, p = 0.013; holidays: OR = 0.47, p < 0.001) was significantly associated with less chance of having a high health literacy level. Spending time on game (school days: OR = 1.04, p = 0.870; holidays: OR = 0.71, p = 0.097) and social media (school days: OR = 0.76, p = 0.342; holidays: OR = 0.89, p = 0.647) showed the same negative association patterns with high HL levels, although the associations were non-significant. Socio-demographic characteristics age (OR = 1.22, p = 0.117), gender (OR = 1.32, p = 0.166), FAS (OR = 1.28, p = 0.289), and grade 5 (OR = 1.53, p = 0.059) showed trends of positive associations, while grade 6 (OR = 0.95, p = 0.864) showed a trend of negative association. However, the associations for socio-demographic characteristics and high HL level were all non-significant. Patterns remained the same before and after controlling for confounding variables, except for the variable of having breakfast every day.

Factors associated with functional health literacy

Having breakfast every day (OR = 2.08; p = 0.003) was significantly associated with a higher likelihood of having high functional health literacy (Table 3). More sleeping hours (OR = 1.07, p = 0.507), better self-efficacy (OR = 1.19, p = 0.457), and higher motivation of learning health knowledge (OR = 1.28, p = 0.297) showed trends of positive associations with high functional HL levels, but the associations were non-significant. Number of health information sources students used (OR = 0.94, p = 0.008), care in hand washing (OR = 0.57, p = 0.006), having sufficient physical activity (OR = 0.66, p = 0.042), and spending time on games (school days: OR = 0.52, p = 0.003) and social media (school days: OR = 0.30, p < 0.001; holidays: OR = 0.42, p = 0.001) showed negative associations with having a high functional health literacy. Self-rated physical health (good: OR = 0.97, p = 0.989; very good: OR = 0.78, p = 0.316), care in dental cleaning (OR = 0.79, p = 0.254), and screen time on TV (school days: OR = 0.88, p = 0.531; holidays: OR = 0.98, p = 0.929) and games during holidays (OR = 0.87, p = 0.482) showed non-significant negative associations with high functional HL levels. In terms of socio-demographics, being in grade 5 showed a significant positive association with high functional HL (OR = 1.76, p = 0.016). Age (OR = 1.06, p = 0.673) and gender (OR = 1.16, p = 0.478) showed non-significant positive associations with high functional HL, and FAS (OR = 0.89, p = 0.623) and grade 6 (OR = 0.93, p = 0.801) showed non-significant negative associations. All patterns remained the same before and after controlling the confounding variables.

Table 3 Odds ratios of logistic regression with having high functional health literacy level as the outcome
Full size table

The mechanism of functional health literacy

A significant positive linear relationship was found between the scores of health literacy and functional health literacy, but the strength of the relationship was weak (b = 0.034, t(404) = 2.69, p = 0.007). Mediation analysis showed that health literacy partially mediated the associations between functional health literacy and the number of health information sources (indirect effect = 0.011, p = 0.013; direct effect = −0.075, p < 0.001) and with care in hand washing (indirect effect = 0.078, p = 0.021; direct effect = −0.470, p = 0.001) (Table 4). As the direct effects remained significant, a portion of these relationships were explained by the mediating mechanism of health literacy. Having breakfast everyday (total = −0.743, p < 0.001; direct = 0.696, p < 0.001; indirect = 0.047, p = 0.101), screen time on games (total = −0.719, p < 0.001; direct = −0.679, p < 0.001; indirect = −0.039, p = 0.117) and social media during school days (total = −1.354, p < 0.001; direct = −1.319, p < 0.001; indirect = −0.035, p = 0.200), and screen time on social media during holidays (total = −0.539, p = 0.005; direct = −0.532, p = 0.006; indirect = −0.008, p = 0.762) showed significant total effects on functional HL when HL was the mediator. However, the significant direct effect can explain the significant total effects. HL did not show any mediation role in the effects of these variables on functional HL.

Table 4 Results of mediation analysis with health literacy score as the mediator in the association between functional health literacy scores and factors associated with high functional health literacy
Full size table

Discussion

Early health literacy promotion is a critical way to improve health level and equity (Breil and Lillich 2023; Guo et al. 2023; Knisel et al. 2020). Among the HL promotion programmes, functional HL is a desirable outcome that has received little attention in research, as it emphasizes the ability to apply health knowledge in real-life situations. The current study aimed to contribute to the development of HL intervention programmes in Hong Kong by exploring factors associated with HL and functional HL among primary school students, while also examining the underlying mechanism of functional HL to identify factors that have direct effects. The summary of results includes the following: (1) Children’s HL levels were associated with higher self-rated health levels, more health information sources, healthy behaviours, better cognitive ability, and less screen time on TV. (2) Children’s functional HL showed positive associations with being a student from grade 5 and regular breakfast eating. Functional HL also showed negative associations with more health information sources, better personal hygiene, sufficient physical activity, and screen time on games and social media. (3) HL showed a weak and positive linear association with functional HL. (4) Health literacy showed partial and full mediation effects on functional HL.

Health literacy

HL was found to be associated with a wide range of factors among primary school students in Hong Kong. Self-perceived health level is a quick screening of populational health status (Toci et al. 2015). A positive dose–response relationship between self-rated health levels and HL suggested that high HL is associated with better childhood health, consistent with previous research evidence (Berkman et al. 2011; Fleary et al. 2018). An earlier health literacy study among secondary school students in Hong Kong found that children with higher HL adopted better healthy behaviours, especially personal hygiene (Huang et al. 2023). The current study demonstrated the same results. Personal hygiene has always been an important component of school-based health promotion programmes worldwide (Pradhan et al. 2020; Watson et al. 2021). It is also one of the few behaviours children themselves can control, compared with their diets, which are still managed by caregivers. The current study highlighted the critical role of personal hygiene in children’s HL development. Our results also suggested that learning resources and opportunities, as indicated by the number of sources to obtain health information, are important for children’s HL levels. Easily accessible technology-based techniques have been found to be effective for HL promotion among secondary school students (Huang et al. 2023). The same techniques might be also effective for primary school students, especially since online learning showed its flexibility during the COVID-19 lockdown when children had limited access to schools (Yan et al. 2021; Zuo et al. 2021). In addition to building interventions based on the current findings, the practical implementation of intervention needs to account for children’s motivation and competence—whether they are motivated to learn more and whether they are capable of understanding the knowledge. The two factors showed an association with children’s HL levels, which can be confounding factors for children’s HL development.

In terms of screen time, children who spend less time watching TV have higher HL. This finding was expected, as screen time on TV is associated with poor health outcomes (e.g., childhood overweight, unhealthy dietary behaviours, insufficient physical activity) (de Jong et al. 2013; Fletcher et al. 2018; Pearson et al. 2014). Children with sufficient health knowledge should understand that they need less screen time on TV to stay healthy. Moreover, TV can be an unreliable source of health information—previous studies identified negative associations between HL and the use of television as a health information source (Chen et al. 2018). Even though children might not consciously use TV as a medium to acquire health knowledge, they could passively acquire information spreading through TV shows (Chen et al. 2018; Fagnano et al. 2012). The unreliability of TV information might be the other explanation for the negative association between HL and TV time. Since children are not able to judge the reliability of information from TV, and they spend a great amount of leisure time watching it (Owen et al. 2010), screen time on TV should be a careful focus of HL research and interventions.

Functional health literacy

The current study, in line with previous research, suggested that HL and functional HL are distinct concepts and should be studied independently (Rocha and Lemos 2016). There were different factors associated with functional HL compared with HL. The interchangeable use of functional HL and HL ignored the fact that functional HL represents the ability to apply health knowledge in addition to assimilating the knowledge (Nutbeam 1998).

Our results demonstrated that learning opportunities/sources might be harmful to children’s functional HL development. This finding was inconsistent with previous evidence, which demonstrated that individuals with higher functional HL tend to have a higher number of information sources (Suka et al. 2015; Yoshida et al. 2014). Participants in previous studies were adults, whereas the current study recruited young children in grades 4–6. Children at this age have less mature cognitive skills and ability to process and memorize information compared with adults (Baddeley et al. 2015; Richland et al. 2006). The rich sources of information might actually confuse children, as they cannot properly judge the reliability, which might explain the inconsistency in findings.

In terms of healthy behaviours, there was a trend wherein children who attempted more healthy behaviours were less likely to have higher functional HL, despite the positive association shown with regular breakfast eating. This finding was unexpected, as early HL studies suggested that children who have a better ability to apply health knowledge would maintain better personal hygiene and exercise more (Buja et al. 2020; Huang et al. 2023). A study of coronavirus infection demonstrated that moderate hand washing can reduce the risk of infection, whereas higher frequency of hand washing does not (Beale et al. 2021). Therefore, careful hand hygiene might not show a dose–response effect on health outcomes—moderate care might be enough to benefit health outcomes. Alternatively, the current study dichotomized children’s responses, which might lose the variability of data and result in unexpected findings. However, as the study had a wide range of variables, dichotomized variables were necessary for ease of interpretation. Future studies might benefit from exploring the linear characteristics of hand washing and physical exercise and re-examining this finding.

Our results also demonstrated that children who know how to apply health knowledge in daily life understand that they should have less screen time playing games and on social media. This was expected and consistent with early evidence (Huang et al. 2023). There was an absence of a significant negative association between time spent watching TV and functional HL. Playing video games and using social media are active information-seeking behaviours, where children actively pay attention to elements in games and social media (Wilson 2000). This active information-seeking is similar to functional HL, in which children are asked to actively apply their knowledge in specific contexts. In comparison, watching TV is a passive information-seeking behaviour, where children receive messages from the TV without any explicit attention. Such differences in information transmission might explain the patterns regarding screen time.

Underlying mechanism of functional health literacy

If HL education programmes set functional HL as the intervention outcome, it is necessary to understand the pathway from potential intervention target to functional HL. The current study explored the underlying mechanism of functional HL, to examine whether the factors we identified have a direct impact on this outcome. Results of mediation analysis showed that, while health information sources and care in hand washing might influence functional HL by first influencing HL (i.e., indirect association), screen time on games and social media showed direct associations with functional HL. To maximize the effectiveness of the HL education programme, it might be more beneficial to focus on factors that have direct impacts on functional HL. There have been discussions around using video games and social media as tools for HL education (Tse et al. 2015; Tuijnman et al. 2022). Proper use of games and social media might compensate their negative effects on children’s functional HL.

Socio-demographic characteristics

In the current study, socio-demographic characteristics demonstrated non-significant effects on HL and functional HL in general, except for grades, in which students in grade 5 showed a significant positive association with functional HL. Previous studies suggested that socio-demographic characteristics in terms of gender, ethnicity, and educational attainment were significantly associated with functional HL (Rocha and Lemos 2016). Given that the current study had a younger sample than previous studies, it might be evidence that socio-demographic factors begin to interact with HL and functional HL at later stages of development. Therefore, if an effective HL promotion programme can be implemented in the early school years, children from different backgrounds may have the potential to benefit from the programmes equally.

Strengths and limitations

The current study explored health literacy not just as a subjective measurement, but also in a functional context. A wide range of factors were included to explore the associations with HL, and the current study was also the first to examine the underlying mechanism of functional HL. However, there were also limitations. First, the sample size might not be sufficient to detect significant effects. Second, the current study only used cross-sectional data, so we cannot draw any causal conclusion from our results. This study is a baseline analysis of the GoSmart project, and our team plan to examine the longitudinal data in future studies. Third, the questionnaire items were self-rated, so students might have chosen the more socially acceptable answers. Fourth, the current measurement of functional HL was not sufficiently detailed. However, there were 49 items in the questionnaire. Adding more detailed measurements would increase the completion time, which might discourage children from participating in the studies.

Implications

The current study highlighted the difference between HL and functional HL, suggesting that future research should not use these two terms interchangeably. The factors identified from the study are potential intervention targets for HL and functional HL for primary school students in Hong Kong. It is worth noting that the key factors of HL and functional HL for primary school students may be distinct from those of secondary school students and adults. Thus, evidence from other populations might not be applicable to young children. Future studies should explore more HL among the younger population, as early education has greater effects on health promotion and equity. The current study also revealed that early HL was not associated with children’s background, which further highlights the importance of early education. To develop an effective intervention, the study identified a few factors that have direct impacts on the outcome functional HL. Continued research on the underlying mechanism is necessary, preferably with longitudinal data which can provide causal evidence.