How different health literacy dimensions influences health and well‐being among men and women: The mediating role of health behaviours

Abstract Background Health literacy, the ability to access, understand, evaluate and apply health information, was found to contribute to positive health outcomes, possibly via promoting healthy behaviours. However, the specific pathways linking different health literacy skills to health and well‐being have remained unclear. Methods A cross‐sectional survey with structural questionnaires was administered among 2236 adults in Hong Kong (mean age = 46.10 ± 19.05). Health literacy was measured by HLS‐Asian‐47. Participants' physical conditions and subjective well‐being were predicted by health literacy and health behaviours with structural modelling path analysis. Results Health literacy in finding and understanding information showed a direct effect on enhancing physical health, while applying information capacity had an indirect positive effect via promoting health behaviours, which was moderated by sex. Only among women, this indirect effect predicting fewer physical symptoms and better well‐being was significant. Conclusions Different health literacy dimensions showed distinct direct and indirect pathways in influencing health for men and women. Based on the findings, skill trainings should be developed to enhance both gender's abilities of finding and understanding health information, while the ability of applying health information should also be improved for modifying lifestyle and promoting health, particularly for women. Patient or Public Contribution Two thousand and two hundred thirty‐six adults from different districts of Hong Kong participated in the study, and responded to questions on health literacy, behaviours and health status.


| BACKG ROU N D
Health literacy, referring to a set of abilities to access, comprehend, appraise and apply information to effectively promote and maintain health in different contexts, 1 was found to play a key role in individual's health behaviours and health status. 2,3 However, the prevalence of inadequate health literacy was considerably high, particularly among older adults. For example, the Agency for Healthcare Research and Quality (AHRQ) has found that approximately one third of Americans only has limited health literacy, and this rate went up to 70% among those aged 75 and above. 4 A recent systematic review on the prevalence of limited health literacy in Southeast Asia has reported that with a large variation across five countries (ie, Laos, Malaysia, Myanmar, Singapore and Thailand), on average, over 50% of the population showed limited health literacy, 5 and the rate was even higher in healthcare settings (67.5%). Similar results were reported in Hong Kong population by a recent study, 6 while the prevalence of limited health literacy was even higher in mainland China. In a sample of 1360 participants (aged  in Shanghai, the prevalence of limited health literacy was approximately 85%. In consistent, when looking at certain type of health literacy , over 70% of people showed limited health literacy about chronic disease 7 ; and about 80% did not have adequate health literacy about infectious disease. 8 With such a high prevalence of limited health literacy in the Chinese society, it is possible that the general public has remained unaware about the impact of health literacy. Previous literature, mostly with western samples, has showed that greater health literacy was consistently associated with various benefits for individual's health, including more healthcare actions, 2 better health status 3 and greater subjective well-being. 9 A systematic review reported that limited health literacy was associated with poorer physical health and higher all-cause mortality rate even after controlling for cognitive functioning. 10 Inadequate health literacy could also lead to lower medical adherence among patients with cardiovascular disease, 11 poorer glycaemic control in type-2 diabetes 12 and higher hospital admission. 13 In addition, health literacy was also found to affect individual's mental health, 14,15 although the existing evidence has been relatively thin. In the review by Berkman et al, 10 only one study showed low health literacy was related to more depressive symptoms after controlling for the confounders. 14 By investigating the relationship between health literacy and happiness, Angner et al, 9 found inadequate health literacy, in addition to poverty and poor health, was associated with lower level of happiness. However, despite this finding being widely cited, the single-item measurement for health literacy (ie, 'how confident are you in filling out medical forms by yourself') may not accurately capture individual's ability to process health-related information, and the question about happiness is not sufficient to indicate one's well-being. Therefore, the current study would address the effects of health literacy on both subjective wellbeing and physical health.
Another objective of the current study is to explore the under-  17 In the studies of Gebeors (2014, 2016), a three-item questionnaire was adopted to evaluate people's perceived capacity in understanding hospital or medical instructions. 20,22 However, how the impact of health literacy may vary across different domains of health literacy capacities were rarely covered. As an exception, by addressing the capacities of accessing to, comprehending, applying and evaluating health-related information, Panahi et al,(2017) found the first three were more important for smoking cessation among college students. 23 Therefore, instead of general health literacy, the current study examines the relationships between different dimensions of health literacy and health outcomes.
In addition, we also aim to address the potential sex difference in the effects of health literacy. Past research has found women's life expectancy is usually 4-5 years longer than men, although they are less healthy than men at any age. 24 When including the sex differences in health behaviours (eg, consumptions of tobacco, alcohol or drug) into the economic model of health deficit accumulation, Schünemann et al, 25 found an additional 89% of the gender gap in life expectancy was explained. However, where this sex differences in health behaviour arises from has remained unclear, and psychosocial factors such as health literacy may play a major role. 2,26 Existing literature showed that the level of health literacy was usually lower among men than women (for a review, see 27 ), despite their tendency to over-report when answering health literacy questions. 28 Probably because women watch more health-related television programmes and have greater social engagement, resulting in higher health literacy and healthier lifestyle. 29 However, no study has examined the link between sex difference in health literacy and health behaviour in predicting health status.
With a sample of 2236 community-dwelling individuals, the current study has a threefold research purpose: (a) to explore how the effects of different health literacy skills would be mediated by health behaviours; (b) to test whether the mediation pathways of health behaviour differ between predicting physical conditions and subjective well-being; (c) to clarify whether the indirect effects of health literacy are moderated by sex. Smoking, drinking and physical exercise were selected to indicate people's health behaviour, which are common behaviours in Hong Kong society. Also, these behaviours could reflect gender-specific preferences, that is the rate of habituated smokers and alcohol users are higher among older male than female, and men tend to have more frequent physical exercise than women. [30][31][32] In addition to providing an updated profile of health literacy in Hong Kong, the current study also aimed to obtain valuable insights for tailoring educational programmes to promote public health.

| Design
A cross-sectional study was conducted with structured questionnaires among individuals from different districts of Hong Kong.
For younger and mid-age adults, the questionnaires were selfadministered. For those aged 65 or above, the questionnaires were administered by a trained research assistant, in case they may have difficulties in understanding the questions due to the relatively low level of education. Written consent was obtained at the beginning of the study.

| Sample
A random sampling was used to recruit two thousand six hundred and thirteen adults from different districts of Hong Kong, with a multi-age stratified clustered sampling method (ie, including similar number of participants aged 18-29, 30-64 and 65 or above). The number of participants from different gender was also balanced.
Health literacy (HL) was measured by the Chinese version of HLS-EU (HLS-Asian-Q . 29 HLS-Asian-Q includes 47 items assessing the information-processing abilities across three domains of health, that is health care, disease prevention and health promotion. Four types of information-processing abilities were evalu- Vietnam. 29 The Cronbach alpha in our sample was 0.98, suggesting good internal consistency.

| Statistical analyses
The Lavaan package in R was used to conduct the structural equation modelling, 38  which only omitted specific variables with missing data on an analysis by analysis basis, to maximize the available data. The scores of four HL capacities were entered as predictors, with the presence of physical symptoms and subjective well-being index as outcome variables. Since we are interested in investigating how health literacy may differ between men and women, and whether this could further lead to sex difference in health behaviours, sex was included as a moderator in the pathway between health literacy to health behaviour. The hypothesized model was displayed in Figure 1.

| RE SULTS
Two thousand and two hundred thirty-six adults (aged from 18 to 93, mean = 45.07 ± 19.05) participated in the survey. 53.8% of the participants were female and 55.1% were married, with the majority having secondary education or above (79.7%, for details, see Table 1). The average number of reported physical symptoms  Table 1). Since sex was proposed as a moderator, we have tested the sex difference in having limited HL (inadequate and problematic) and adequate HL The correlations among predictors, mediator, moderator and health outcomes were displayed in Table 2 To probe the moderating effect of sex, the indirect effects of health literacy in applying information were compared between men and women. The results showed that among men, the capacity of applying information has marginally significant indirect effects on physical health and well-being (physical health: β= -0.023, P = .088; well-being: β = 0.144, P = .067). However, among women, the indirect effects on physical symptoms and well-being were both significant, such that via reducing unhealthy behaviours, the capacity of applying information was associated with fewer physical symptoms (β= −0.063, P = .001) and greater well-being (β = 0.387, P < .001).  The details of the moderated mediation models were presented in Tables 3 and 4.

| D ISCUSS I ON
Health literacy was found to be a key contributor to individual's health. Although the related changes in health behaviours were proposed to be an underlying mechanism of the health literacy, limited evidence has been found regarding the mediating role of health behaviour. By conducting a large-scale survey across different age, the current study has tested a moderated mediation model of health literacy predicting physical health and subjective well-being through influencing health behaviours. Furthermore, we have looked at the effects of specific health literacy skills, that is finding, understanding, evaluating and applying health-related information.
In a sample of 2236 adults, we found the prevalence of limited HL was 55%, which was close to the average levels in Malaysia and Singapore. 5 The prevalence of limited HL was higher among women than men, which was inconsistent with the previous findings that men actually have lower level of HL. 26,39 This sex difference might be driven by that men's education level was higher in our sample (χ 2 = 24.99, P < .001), which contribute to higher health literacy. In fact, with logistic regression, it showed that although female are 1.19 times more likely to have limited HL than male (P = .04), when education entering the model, the sex difference became insignificant, and people with lower education are 2.47 times more likely to report limited HL (P < .001). It is also possible that men may use over-report when answering health literacy questions, thus leading to a higher HL score. 28  on health via behaviours. Interestingly, the capacity of evaluating information didn't show any effect on physical symptoms or wellbeing, which is consistent with previous findings that smoking cessation was associated with various HL capacities except evaluating information. 23 It is possible that evaluating information was usually perceived as the most difficult (eg,, 6

| CON CLUS ION
To conclude, the current study provided evidence suggesting that promotions for health literacy are still urgent for Hong Kong population, such that over 50% of the lay public showed a limited level of health literacy, and this prevalence was a little higher among women.
In addition, by addressing specific direct and indirect effects of four dimensions of health literacy on physical health and subjective well-being, the findings showed that better perceived capacities in finding and understanding health information could lead to better physical health, while greater capacity in applying health information is associated with healthier lifestyle. Based on the findings, more educational programmes to advocate health literacy and increase the awareness of healthy lifestyle should be developed for the lay public. In particular, skill trainings for finding and understanding health information should be provided to promote both men and women's physical health; while for women, it is important to enhance the ability to apply health information to behaviour modifications, which may further benefit their health and well-being.

| E THI C S APPROVAL AND CON S ENT TO PARTI CIPATE
The ethical approval for the study was obtained from the Human Research Ethics Committee of The University, and participants have provided written consent before taking part in the study.

| CON S ENT TO PUB LIS H
Not applicable.

| COMPE TING INTERE S TS
There was no competing interest that needs to be declared.

ACK N OWLED G EM ENTS
The author would like to thank Prof. Cheng Sheung-Tak for his constructive suggestions and help for the manuscript.

CO N FLI C T O F I NTE R E S T
The manuscript is not under simultaneous consideration elsewhere, and all authors have no conflict of interest to declare.

AUTH O R S ' CO NTR I B UTI O N S
ZF has analysed and interpreted the data, as well as written and edited the manuscript. PO was a major contributor in generating the research idea, collecting data and editing the manuscript. JC was also a contributor in generating the research idea and data collection. All authors have read and approved the final version of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Derived data supporting the findings of the study are available from the corresponding author on request.