Digital Health Literacy and Sport-Physical Activities: The Moderating Role of Age

Background : The present study extends understanding of digital health literacy by exploring issues that clarify how and when sport-physical activities enhance digital health literacy measured by four types of health-related information searched on the Internet i.e., general health information, information on a speci�c injury or condition, speci�c information on a medical treatment or procedure and information regarding second opinion after visiting a doctor. Methods: We employ Tobit model on the large European data (N=5,318). Results: We �nd support for direct relation between sport-physical activities and digital health literacy. In addition, we �nd that the effect of sport-physical activities on digital health literacy is reversed when interacted with age. Conclusion: This evidence is consistent with the view that sport-physical activities can be considered as an important tool for digital health literacy improvement but only under certain socio-demographic characteristics.


Background
Internet technologies are considered as major developments in the last twenty years that affect all sectors around the globe as well as healthcare sector.Actually, as concluded in [1], Internet is recognised as primary mechanism for health-related information dissemination.Therefore, Internet has become an imperative in every attempt to improve healthcare [2].Scholars [3], [4] identi ed the bene ts of Internet for healthcare sector such as instant access to a vast array of information, improved communication between patients and doctors, and support for interpersonal interactions and social support.However, there are a number of constraints to be ful lled, in order to contribute to better utilisation of Internet for healthcare purpose.
One of the main factors that determine the bene ts of Internet usage for healthcare outcomes is associated to digital health literacy [5].
Digital health literacy is de ned as degree to which individuals have the capability to obtain, process, and understand basic health information and services needed to make appropriate health decisions [5][6][7].Furthermore, poor health literacy could drive to several adverse health outcomes, such as delayed diagnoses, poor adherence to treatment regimens, mortality, hospitalization and emergency care visits [8][9].In the same vein, in [10], it is argued that comparing to individuals who have high health literacy, those with low health literacy have poorer health outcomes.
It is considered that poor health literacy is linked with unhealthy lifestyle behaviours such as poor diet, smoking and a lack of physical activity [11][12][13][14].Actually, it is argued that sports activities establish a platform for learning various personal health-related skills that are not only related to motor skills [15][16][17].In addition, previous scholars underline that health literacy is dependent on health behaviours such as sport-physical activity [7], [18][19][20].However, in [21], it is indicated that there is limited data on the relation between health literacy and health promoting behaviours associated to sport-physical activity.Therefore, considering importance of sport-physical activities for digital health literacy, our rst objective is to empirically examine this relationship.What more since little is known about the link of health literacy with health information-seeking type [22][23] this paper distinguishes between four digital health literacy indicators associated with different types of health-related information searched on the Internet i.e., general health information, information on a speci c injury or condition, speci c information on a medical treatment or procedure and information regarding second opinion after visiting doctor.
Moreover, one stream of literature speci cally suggests that the digital health literacy is contingent upon socio-demographic characteristics [24][25].As suggested in [7], investigating contingent factors associated to digital health literacy would also be of great importance because these might be objectives for further improvements.Technology acceptance suggests that individual demographics affect the embracement of new technologies [26][27][28].Age is associated to demographic characteristics that represents experience (or lack of experience) with certain health issues [25], but also to the Internet utilisation.For instance, rates of poor health literacy are associated with older age, low income and low educational level [29][30].Consequently, we build and test a model of the impact of sport-physical activity on digital health literacy by introducing a moderator factor, namely, age.
The main intended contribution of this study is therefore to develop a more detailed understanding of how and to what extent sportphysical activities enhance digital health literacy.Does age moderate relationship between sport-physical activities and digital health literacy?Answering these questions contributes to emerging research concerning the drivers that could help digital health literacy improvement.Actually, comparing to previous studies on the subject our analysis will include individuals from various age groups and countries since in order to enhance health literacy at the population level, the analysis needs to cover sample across various age groups and settings [17].
To address these questions empirically, we rely on the European citizens' digital health literacy data which is conducted by TNS Political & Social at the request of the European Commission, Directorate-General Communication Networks, Content and Technology (DG CONNECT).Respondents from different social and demographic groups were interviewed via telephone in their mother tongue on behalf of the European Commission.Noteworthy, respondents form 28 EU countries are present in the sample, what makes important factors when considering generalisability of the ndings.In addition, as argued in [19], [31], [32] research is still lacking regarding the analysis on overall populations, due employing European citizens' digital health literacy data will give us possibility to obtain more general understanding concerning the link between digital health literacy and sport-physical activities.
The rest of this article is organized into four sections.In the next section, we present previous literature leading up to our hypotheses for direct and moderating effects.In the two subsequent sections, we detail our method and obtained results.In the nal section, we discuss the obtained ndings and practical implications.
The relationship between health literacy and sport-physical activities Previous scholars [17], [20], [33], [34] have recognised the link between health literacy and health behaviours.More precisely, health behaviour is related to physical activity, fruit and vegetable consumption, smoking behaviour, breakfast consumption, alcohol consumption, and body mass index (BMI), all of which are known to be linked to health status and health outcomes [17][18][19][20], [35], [36].
Among above mentioned indicators that are associated with health behaviours, in [7] and [20] is identi ed physical activity as the important pathway between health literacy and health outcomes.Accordingly, in [17] is stressed out that sport-physical activities are associated with better coping skills needed in daily life what includes also health literacy empowerment.Moreover, in [37] is proposed that health promotion action, such as physical activities, creates an intermediate set of health promotion outcomes, including health literacy, which in turn affects health outcomes.
Empirical work investigating the link between health literacy and sport-physical activities has mainly con rmed the positive relationship.For instance, conducting a cross-sectional study using data from the Rush Memory and Aging Project, which includes 556 older persons, in [21] is found that higher health literacy scores were associated with more frequent participation in health promoting behaviours, including cognitive, physical and social activities.In the same sense, using data from The Netherlands on older people, in [7] is found a negative association between inadequate health literacy and physical activity.In addition, in [20] is con rmed the ndings on the sample of 3,241 participants, showing that low health literacy was associated with insu cient physical activity, insu cient fruit and vegetable consumption, lack of regular breakfast consumption, obesity and low alcohol use, but not with smoking.Furthermore, working on the population-based survey among colorectal cancer (CRC) survivors, in [38] is demonstrated that CRC survivors with low health literacy did not meet the prescribed physical activity guidelines compared with survivors with medium or high health literacy.Moreover, the authors indicate that CRC survivors with low health literacy reported statistically signi cantly lower levels of mental and physical health-related quality of life and higher distress levels compared with survivors with medium and high health literacy.Working on the date from Finland among school children, in [17] is shown that perceived health literacy was higher among adolescents who participated in sports club activities.
Based on previous discussion, we posit that: H1.Sport-physical activities have a positive impact on digital health literacy.

The moderating effect
Both health literacy and sport-physical activities are considered to be contingent upon socio-demographic characteristics.Moreover, socio-economic and demographic factors such as age, income, education and health status determine people's probability to use the internet to seek health information [39][40].Accordingly, in [40] is found that socio-demographic characteristics particularly age, education, income, perceived health and social isolation are associated with having access to the internet con rming that a 'digital divide' exists among various groups.
Considering that ageing populations presents fundamental European challenges, achieving enhancing health literacy level among different age groups is crucial.However, even that Internet offers the elderly population signi cant potential for remaining independent longer [41], the elderly lag behind in using and bene ting from information technology (IT) [42].Accordingly, older adults are at a risk of low health literacy [32].Moreover, rates of inadequate health literacy have been shown to be higher in adults with older age [14], [20], [43], [44].The Canadian Council on Learning [45] even reported that health literacy decline begins in early adulthood and continues to fall into old age.The literature identi ed that the reason for not using IT among older adults are cognitive decline associated with aging processes and attitudes such as anxiety about computer use and the perception that the technology was not useful for them [46][47].On other hand, younger population is considered to be more frequent health information seekers [25], [46], [48], [49].
Therefore, we present the following hypothesis: H2.Age moderates negatively the relationship between sport-physical activities and digital health literacy.

Data and sample
This study relies on the Flash Eurobarometer report No. 404 "European citizens' digital health literacy" which aims to assess the extent to which Europeans already use the Internet and online resources to help manage their own health.The data provide information on the type of health-related information people most commonly look for, where they look for it and why they look for it, how likely they would be to use the Internet to look for health-related information in the future, level of Internet usage among Europeans, and also of the health of European citizens and other health-related issues (e.g.doctor visits, frequency of physical exercise).
The data used was carried out by TNS Political & Social network in the 28 Member States of the European Union during September 2014.Respondents from different social and demographic groups were interviewed.After eliminating observations with missing values that are relevant for this study, our nal sample counts for 5,318 observations (the sample changes is related to the dependent variable utilisation).

Dependent Variable
As it was indicated previously, we distinguish between four types of health-related information searched on the Internet i.e., general health information, information on a speci c injury or condition, speci c information on a medical treatment or procedure and information regarding second opinion after visiting doctor.More precisely, the questions were de ned as following: On average, within the last 12 months, how often have you used the Internet to (1) access general information on health-related topics or ways to improve health; (2) access on a speci c injury, disease, illness or condition; (3) get a second opinion after having visited doctor; (4) nd speci c information on a medical treatment or procedure.The questions are designed on 5-point scale anchored by 1 = "few times within the past year", 2 = "approximately once every two months", 3 = "approximately once a month", 4 = "Several times a month", 5 = "once a week or more often".

Independent variable
Our main independent variable is related to sport-physical activities and it is measured using following question: How often do you exercise or play sports?By 'exercise', I mean any form of physical activity which you do in a sports context or sports-related setting, such as swimming, training in a tness centre or a sports club, running in the park etc.The variable is also designed using 6-point scale with value 1= "never", 2 = "less often", 3 = "1 to 3 times a month", 4= "1 to 2 times a week", 5= "3 to 4 times a week", 6= "5 times a week or more".

Control Variable
Furthermore, we control for respondent speci c characteristics that are found to determine health literacy by including additional control variables.Actually, following previous scholars [7], [20], [39], [40] we include following control variables: (1)  Noteworthy, we interact our independent variable sport-physical activities with age to see if it moderates the sport-physical activitieshealth literacy link.In order to avoid multicollinearity between the interaction terms and their components, we follow the approach suggest in [51].More precisely, we mean-center the direct terms by subtracting the mean of each variable from the values of each observation.
In Table 1 we present descriptive statistics of variables used.

Results
Tables 2 and 3 report the estimation results of our empirical models, direct and moderating effect, respectively.
Model 1 presents the direct model where dependent variable represents general information on health-related topics or ways to improve health.We can observe that sport-physical activities have signi cant and positive effect on digital health literacy (t = 0.12, p < 0.01).
The results are going in the same direction when looking Model 2 where dependent variable refers to the information on a speci c injury, disease, illness or condition.More precisely, we can conclude that sport-physical activities improve responds' digital health literacy (t = 0.11, p < 0.01).Furthermore, Model 3, reveals that sport-physical activities improve digital health literacy measured by information about opinion after having visited doctor (t = 0.24, p < 0.01).Finally, the results are also con rmed in the Model 4 where dependent variable is related to the speci c information on a medical treatment or procedure (t = 0.11, p < 0.01).The obtained ndings using all four indictors of digital health literacy support our H1.In addition, our results lend support to previous studies that show positive link between sport-physical activities and digital health literacy [7], [17], [20].Accordingly, we may suggest that participation in sport-physical activities creates skills as well as environment that support digital literacy empowerment, including also health literacy.
Turning to the control variables, we can see that age, gender, profession, general health knowledge and internet search engines in uence negatively digital health literacy.On the other hand, general health condition, frequency of doctor visits, types of information searched, speci c internet source such as online newspapers or magazines, speci c and dedicated websites, blogs or forums and online social networks could be considered as important drivers of digital health literacy improvement.Noteworthy, the ndings also indicate that being self or surrogate seeker (who seeks health information for others) [52], also improve digital health literacy.However, the results are not con rmed when searching is oriented to children.]The moderator analysis is presented in Table 3. Model 1a indicates that age moderates negatively the relationship between sportphysical activities and digital health literacy (t=-0.003,p < 0.01).Furthermore, we can observe that positive effect of sport-physical activities on health literacy disappears in Model 2a when sport-physical activities are interacted with age.In other words, we obtain non-signi cant results.The similar situation is when looking Models 3a and 4a.Therefore, we can only partly support H2. Overall, the results from Table 3 are quite disappointing.Actually, despite all potential advantages of digital health literacy, the elderly is less likely to exploit them.What more, even that sport-physical activities could be considered as important mechanism for digital health literacy enhancement (as we can see from Table 2), it is not the case for older people.Actually, the positive effect of sport-physical activities is vanished by age.The obtained ndings could be explained by two facts.Firstly, even that older people have more time to engage in healthy behaviours, they are less likely to be su ciently physically active than younger adults [19].Secondly, as discussed previously, older population shows signi cant resistance towards IT utilisation [46], [47].

Conclusion
In order to achieve objectives de ned in "Health 2020″ regarding enhancement of population health outcomes, promoting health literacy is considered as one of the crucial mechanism [53].Moreover, in [54] is underlined that addressing the challenges related to low health literacy should be considered as an urgent priority.Previous scholars tried to identify mechanism that could help improvement of digital health literacy, and one of the proposed mechanisms is sport-physical activities [17], [20].However, based on the existing literature, in [31] is concluded that health literacy and sport-physical activities are both known to be related to population health, however further investigation is needed to understand better their relation.
Therefore, in our study, we analysed the relation between sport-physical activities and digital health literacy measured by four indicators associated with types of health-related information searched on the Internet.In addition, we examine how the relation between sport-physical activities and digital health literacy may be moderated by age.Accordingly, a comprehensive model is presented here, aiming to shed some light on a complex real phenomenon where sport-physical activities, digital health literacy and age are interlinked.The results based on the large European data, show that sport-physical activities are positively correlated to digital health literacy.The results are con rmed for all four indicators of health literacy namely search on general health information, information on a speci c injury or condition, speci c information on a medical treatment or procedure and information regarding second opinion after visiting doctor.
Regarding the moderating effect, the signi cantly negative association between sport-physical activities and health literacy (associated to search on general health information) when moderated by age con rms our belief that older population shows important resistance regarding utilisation of IT, therefore, the positive effect of sport-physical activities could not hold when interact with age.In addition, the positive effect of sport-physical activities also disappears when interacted with age for three other indicators of digital health literacy.Therefore, our study highlights the importance of socio-demographic characteristics when analysing digital health literacy.Accordingly, in order to improve health literacy among elderly population speci c tailored health literacy program should be established that targets only this category of population [32].In addition, online health providers have to create different interfaces that are suitable for different demographic groups [55].
Overall, our study also implies that in order to identify the mechanisms that could improve digital health literacy, we need to focus on the moderating effects.In this vein, our ndings add to the recent literature which focuses on better identifying the mechanisms through which digital health literacy could be improved.
This study focuses on one socio-demographic characteristic, that is, age.Thus, we acknowledge the limitation of excluding potential impact from other socio-demographic characteristics as potential mediators.Furthermore, future analysis should identify other mechanisms that could serve as tool for improving digital health literacy a part of sport-physical activities.In addition, the date used is cross-sectional, due it does not allow for the assessment of de nitive causal relationships.Longitudinal data would prove useful in establishing cause-and-effect relationships.

Table 1
Since in this paper the dependent variables are censored with an upper limit of 1, the Tobit regression model is an adequate econometric technique for determining the relationship between digital health literacy and sport-physical activities.The Tobit model expresses the dependent variable in relation to latent variables:

Table 2
Results of Tobit model-direct effect

Table 3
Results of Tobit model-moderating effect