Promotion International, 2019;34:e1–e17 doi: 10.1093/heapro/day062 Advance Access Publication Date: 9 August 2018 Article

Does health mediate the relationship between socioeconomic status and health disparities? Integrative review Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 Coraline Stormacq1,2,*, Stephan Van den Broucke3, and Jacqueline Wosinski2

1Faculty of , Catholic University of Louvain, Clos Chapelle-aux-Champs 30, B-1200 Bruxelles, Belgium, 2Institut et Haute Ecole de Sante´ La Source, HES-SO, University of Applied Sciences and Arts Western Switzerland, Avenue Vinet 30, CH-1004 Lausanne, Switzerland and 3Faculty of Psychology and Educational Sciences, Psychological Sciences Research Institute (IPSY), Catholic University of Louvain, Place du Cardinal Mercier 10, B-1348 Louvain-la-Neuve, Belgium

*Corresponding author. E-mail: [email protected]

Summary While socioeconomic disparities are among the most fundamental causes of health disparities, socio- economic status (SES) does not impact health directly. One of the potential mediating factors that link SES and health is health literacy (HL). Yet although HL can be considered a modifiable risk factor of socioeconomic disparities in health, the relationship between SES, HL and health disparities is not well understood. This study reviewed the evidence regarding the mediating role of HL in the relation- ship between socioeconomic and health disparities. Medline, Cinahl, Embase, PsychInfo, Eric, Web of Science, Google, Google Scholar, Mednar, Doaj and Worldcat were used to retrieve studies that spe- cifically addressed socioeconomic and socio-demographic factors related to low HL levels, as well as the mediating role of HL in the relationship between SES and disparities in health outcomes. Selected studies were assessed for methodological quality. Sixteen published studies were retained for inclusion and content analyzed using the constant comparison method. The review indicates that dis- advantaged social and socioeconomic conditions contribute to low HL levels, whereby low SES, and particularly educational attainment, is the most important determinant of HL, and that HL mediates the relationship between SES and health status, quality of life, specific health-related outcomes, health behaviors and use of preventive services. HL can be considered as a modifiable risk factor of socioeconomic disparities in health. Enhancing the level of HL in the population or making health services more accessible to people with low HL may be a means to reach a greater equity in health. Key words: health literacy, socioeconomic status, mediating role, social determinant of health, health disparities

INTRODUCTION (Adler and Newman, 2002). As such, socioeconomic Socioeconomic status (SES) is linked to a wide range of disparities, traditionally measured through levels of edu- health related lifestyles and to poor health outcomes cation, income and/or occupation, are considered as the

VC The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please email: [email protected] e2 C. Stormacq et al. most fundamental cause of health disparities (Adler and include accessing and understanding information on the Newman, 2002). Health disparities are differences determinants that impact the populations’ health, under- in health status and health outcomes (Whitehead, 1991; standing public health messages, participating in civic Braveman and Gruskin, 2003) between social action, resolving problems, and taking decisions that groups that are ‘not only unnecessary and avoidable, but benefit the whole community (Gazmararian et al., 2005; in addition, are considered unfair and unjust’ Freedman et al., 2009). [(Whitehead, 1991), p. 2019], and which systematically In both the clinical and population approaches, a and adversely impact socially and economically disad- low SES is considered as a potential risk factor for low vantaged groups (Braveman and Gruskin, 2003; HL (Lee et al., 2010; Sun et al., 2013; Van Der Heide Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 Braveman, 2014). et al., 2013a), which in turn is associated with poor Although SES indicators are powerful determinants health outcomes such as higher rates of chronic diseases of health, they do not impact health directly, but instead (Baker et al., 2007; Kickbusch and Maag, 2008; serve as ‘proxies’ for other determinants (Adler and Bostock and Steptoe, 2012), higher rates of mortality Newman, 2002). Indeed, many intermediate factors (Baker et al., 2007; Bostock and Steptoe, 2012), and ad- have been highlighted in the relationship between low verse health-related behaviors (Adams et al., 2009; SES and poor health outcomes, such as poor environ- Bostock and Steptoe, 2012). Population surveys have in- mental conditions (exposure to toxic wastes, air and wa- dicated that nearly half of the American population ter etc.), poor housing quality and living presents difficulties to act on health information (Kutner conditions or poor work environments and working et al., 2006), and that approximately half of the popula- conditions (physically arduous tasks, occupational haz- tion in Europe has limited HL skills (Sorensen et al., ards, repetitive strain etc.) (Evans and Kantrowitz, 2015). Given the large numbers of people who are af- 2002). More specifically regarding health behaviors, one fected by low HL and its numerous adverse health of the factors that have been considered as linking SES effects, HL is regarded as an important public health and health is health literacy (HL). HL has been defined concern (Van den Broucke, 2014). as ‘a person’s knowledge, motivation and competences On the other hand, HL can also be regarded as a le- to access, understand, appraise, and apply health infor- verage to improve health, both at the individual and the mation in order to make judgments and take decisions population level. As a potential mediator of the relation- in everyday life concerning healthcare, disease preven- ship between SES and health status, HL can be consid- tion and to maintain or improve qual- ered as a modifiable risk factor of socioeconomic ity of life during the life course’ [(Sorensen et al., 2012), disparities in health. Consequently, enhancing the level p. 3]. In this definition, ‘access’ refers to the ability to of HL in the population or making health services more seek, find and obtain health information; ‘understand’ accessible to people with low HL may be a means to to the ability to comprehend the health information that reach a greater equity in health. However, the relation- is accessed; ‘appraise’ to the ability to interpret, filter, ship between SES, HL and health disparities is not well judge and evaluate the health information that has been understood (Bennett et al., 2009), and research investi- accessed; and ‘apply’ to the ability to communicate and gating the mechanisms that allow HL to decrease socio- use the information to make a decision to maintain and economic and health disparities remains scarce. Only improve health (Sorensen et al., 2012). two systematic reviews have thus far been published In the literature, two approaches to HL exist which examined the relationship between socioeco- (Pleasant and Kuruvilla, 2008; Van Der Heide et al., nomic disparities and HL. The first one (Paasche-Orlow 2013a). The ‘clinical approach’ encompasses the skills et al., 2005) reviewed the socioeconomic and socio- that are required of an individual or patient to function demographic characteristics that are associated with in a clinical setting (e.g. understanding and applying poor HL, but did not consider the mediating role of HL medical forms or prescriptions, communicating with specifically. Moreover, this study used a limited concep- healthcare professionals (Kickbusch, 2001; Garcia et al., tualization of HL focusing on and numerical 2010). According to this approach, poor HL skills are skills only. The second review (Mantwill et al., 2015) considered a risk factor for poor health outcomes explored the association between HL and educational, (Nutbeam, 2008). In contrast, the ‘public health ap- racial and gender-related health disparities. While this proach’ considers (public) HL as an asset for health pro- review provides theoretical insights into the contribution motion (Nutbeam, 2008). It focuses on the skills that of HL to health disparities, it only mentioned the medi- are required to promote health and prevent diseases at ating role of HL as a possibility, and did not cover it in the population level (Freedman et al., 2009), which the review itself. The mediating role of HL between SES and health disparities e3

To our knowledge, no published literature review statistically tested mediation analysis such as, but not has so far specifically focused on the mediating role of limited to, path analysis or structural equation model- HL in the association between SES and health dispar- ing. Any study design was considered, including ran- ities. This study addresses this gap by reviewing studies domized controlled trials, longitudinal studies, cohort that consider the association between social or SES and studies, cross-sectional studies, qualitative studies and health disparities, with a view to clarify the role of HL other designs if relevant to our topic. Studies published in contributing to health disparities. in French, English, Spanish or Dutch and published up to the moment of the review were considered. To define disadvantaged populations depending on the context Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 METHODS in which primary studies took place, and to guide To investigate whether HL mediates the association be- data extraction, the PROGRESS factors (Place of resi- tween SES and health disparities, an integrative litera- dence, Race/ethnicity, Occupation, Gender, Religion, ture review was performed. An integrative review is ‘a Education, Socioeconomic status and Social capital) specific review method that summarizes past empirical (O’Neill et al., 2014) were used. These are ‘socially or theoretical literature to provide a more comprehen- stratifying factors that drive variations in health out- sive understanding of a particular phenomenon or comes’ [(O’Neill et al., 2014), p. 58] that are systemic healthcare problem’ [(Whittemore and Knafl, 2005), both within and between countries. For example, people p. 546]. For our review, studies were considered that living in rural areas, or belonging to an ethnic minority, addressed HL, assessed by any described measurement or having low educational levels or low income, are con- tool, and offered relevant content with regard to the re- sidered as disadvantaged and at risk of health dispar- lationship between social status or SES and HL, ities. PROGRESS factors are widely used in systematic health outcomes, or health disparities. To trace pub- reviews and recognized as indicators of health inequities lished and unpublished studies investigating these topics, by the Cochrane collaboration, the Campbell collabora- an initial search was undertaken on Medline/Pubmed tion, and the World Health Organization (WHO) to identify index terms and keywords that are used to (Cochrane Field and Campbell Equity describe relevant articles. Using these index terms and Methods Group, 2008; Tugwell et al., 2010). Studies keywords, and using the Medline/Pubmed initial that were selected using this retrieval procedure were strategy as a basis, a full search was performed in assessed for methodological quality, using standardized Medline, Cinahl, Embase, PsychInfo, Eric and Web of critical appraisal instruments from the Joanna Briggs Science. In addition, an internet search was performed Institute (JBI) (The Joanna Briggs Institute, 2014) prior on Google, Google Scholar, Mednar, Doaj and to analysis. Worldcat to find additional studies and gray literature For the data analysis use was made of the constant (technical and official rapports, conference papers, comparison method (Whittemore and Knafl, 2005). proceedings etc.). For each database, an individual First, references were classified according to the type of search strategy was developed combining 12 keywords publication (primary study, official report) and type of (‘Social Determinants of Health’[Mesh], ‘Health Status design. Next, data were extracted based on the research Disparities’[Mesh], ‘Socioeconomic Factors’[Mesh], question and inclusion criteria. The results of primary ‘Social Class’[Mesh], ‘Healthcare Disparities’[Mesh], studies were organized in two main categories: socioeco- ‘Ethnic Groups’[Mesh], ‘Minority Health’[Mesh], nomic and socio-demographic characteristics influenc- ‘Vulnerable Populations’[Mesh], ‘socioeconomic status’, ing HL levels, and the mediating role played by HL. ‘health disparities’, ‘health inequalities’ and ‘equity’) Results concerning the same variable were grouped to- with the search terms ‘health literacy’ and ‘public health gether and systematically compared, allowing the identi- literacy’ using the Boolean operator OR. The reference fication of trends and patterns, as well as similarities lists of all identified references were searched for addi- and differences. tional studies. Titles and abstracts of studies retrieved in the search were assessed for relevance against the inclu- sion/exclusion criteria. To be included in the review, studies needed to spe- RESULTS cifically address socioeconomic and socio-demographic Out of a total of 5656 identified references, 334 publica- factors related to low HL levels, as well as the mediating tions were considered for detailed examination after ex- role of HL in the relationship between social status or cluding duplicates and exclusion based on titles. Of SES and disparities in health outcomes, using a these, 318 did not meet the inclusion criteria, leaving a e4 C. Stormacq et al. total of 16 references. A flow diagram summarizing the reading, and skills regarding health in- study selection process is given in Figure 1. formation, while only two studies (Friis et al., 2016; Zou et al., 2016) considered HL from a more compre- Characteristics of included studies hensive perspective. To assess HL, seven different mea- Table 1 shows the characteristics of the studies that surement tools were used. For this reason, and because were included in the review. Of the 16 original studies, different outcomes were considered in the studies, a 12 were cross-sectional studies, two were secondary meta-analysis was not possible. Therefore, results are analyses of a randomized controlled trial (Gwynn et al., presented in a narrative form.

2014; Gupta et al., 2016), one was a prospective cohort Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 study (Curtis et al., 2012), and one was a retrospective Socio-demographic and socioeconomic factors cohort study (Howard et al., 2006). Twelve studies were influencing HL conducted in the USA, two in Europe (Van der Heide The socio-demographic and socioeconomic factors re- et al., 2013b; Friis et al., 2016), one in Australia (Adams lated to HL in the studies included in this review com- et al., 2013), and one in China (Zou et al., 2016). Five prised SES, race/ethnicity, and gender. studies were population-based (Bennett et al., 2009; Adams et al., 2013; Van der Heide et al., 2013b; Hovick SES: education, income, occupation and perceived social et al., 2014; Friis et al., 2016), and 11 community-based status (Howard et al., 2006; Schillinger et al., 2006; Osborn Socioeconomic factors, including educational attain- et al., 2007, 2009, 2011; Ayotte et al., 2009; Pandit ment, income, occupation and perceived social status, et al., 2009; Curtis et al., 2012; Gwynn et al., 2014; were addressed in seven studies (Howard et al., 2006; Gupta et al., 2016; Zou et al., 2016). Fourteen studies Ayotte et al., 2009; Bennett et al., 2009; Pandit et al., considered functional HL (FHL), which refers to basic 2009; Adams et al., 2013; van der Heide et al., 2013b;

Fig. 1: Flow diagram of selection process for the review. Table 1: Characteristics of studies included in the review disparities health and SES between HL of role mediating The

Authors Aim and sample Variables and analysis Main results

Adams et al. Mediation role of FHL between SES, Independent variables: sex; education; income; Lower education and lower income significantly associated (Adams et al. 2013) and perception of the risk of occupation; area of residence with lower HL skills Australia lifestyle behaviors for cancer Dependent variables (outcomes): cancer risk percep- Cancer risks more likely to be perceived as no important (or do N¼2824: representative Australian tion related to lifestyle behaviors (smoking, not know) among those with inadequate HL skills sample, population-based study alcohol, diet, weight, exercise); health behaviors Statistically significant mediation effect of FHL in the pathway (fruit and vegetable intake, level of physical from SES to perceptions of risk and lifestyle behaviors activity, alcohol consumption, smoking status) Mediation effect: 29.4% of the total effect Mediators: FHL (NVS) Analysis: Structural equation modeling Ayotte et al. Mediation role of FHL between Independent variables: gender; race (Whites, Female sex, White race, and higher levels of education signifi- (Ayotte et al. 2009) demographic factors and health African Americans); financial status; education cantly associated with better health information recall USA information recall Dependent variables (outcomes): health information Higher level of HL associated with higher financial status, N¼1190 hypertensive adults recall White race, higher levels of education, and female sex Mediators: FHL (REALM) Higher HL skills associated with better health information re- Analysis: path analysis call Test for mediation: indirect paths from financial status (b¼0.02, p<0.01), race (b¼–0.05, p<0.01), education (b¼0.11, p<0.01), and gender (b¼0.02, p<0.01) to health information recall through HL significant Bennett et al. Mediation role of FHL in racial/eth- Independent variables: education; race/ethnicity Higher HL scores among Whites, followed by Blacks and (Bennett et al. nic and education-related dispar- (White, Black, Latino/Hispanic) Latinos 2009) ities in self-rated health status and Dependent variables (outcomes): self-rated health Lower educational attainment associated with lower HL scores USA preventive health behaviors status; preventive health behaviors (influenza vac- Poorer self-rated health status among Blacks than Whites, and among older adults cination, dental care, mammography) for those with less than high school degree N¼2668 adults aged 65 years and Mediators: FHL (NAAL) Lower influenza rates among Blacks. No less use of older, population-based study Analysis: mediation analysis using criteria of Barron mammography or dental care services by Black and Latina and Kenny (Baron and Kenny, 1986) women Lower influenza vaccination use, mammogram and dental care use for those with less than high school attainment Increased HL independently associated with better health sta- tus, influenza vaccination, mammography, or dental care use Significant mediation effect of FHL: association of race/ethnic- ity with health status and influenza vaccination reduced, as- sociation of education and all the outcomes reduced

(continued)

e5 Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded Table 1: Continued e6 Authors Aim and sample Variables and analysis Main results

Curtis et al. (Curtis Mediation role of FHL in explaining Independent variables: race/ethnicity (Latino, Lower HL skills among Blacks (38.3%), followed by Latinos et al. 2012) racial differences in asthma out- African American, White) (31.3%) and Whites (13.2%) (p¼0.002) USA comes Dependent variables (outcomes): asthma quality of For all outcomes: significantly worse asthma outcomes for N¼353 adults aged 18–40 with per- life (AQOL); asthma-related emergency depart- African Americans than Whites. Latinos had lower quality sistent asthma ment visits; asthma-related hospitalizations; of life (b¼0.47; 95% CI [0.79, 0.14]; p¼0.01) and asthma control worse asthma control (RR¼0.63; 95% CI [0.41, 0.98]; Mediators: FHL (REALM) p¼0.04) than Whites Analysis: multivariate models Limited HL skills associated with lower AQOL (b¼0.56; 95% CI [0.79, 0.33]; p<0.001), more emergency depart- ment visits (RR¼1.67; 95% CI [1.27, 2.18]; p<0.001), more hospitalizations (RR¼2.10; 95% CI [1.16, 3.82]; p¼0.01), and lower asthma control (RR¼0.49; 95% CI [0.34, 0.71]; p<0.001) Asthma disparities between African Americans and Whites re- duced by HL by 13–17% Asthma disparities in AQOL between Latinos and Whites re- duced by HL by 21.2% (b¼0.33; 95% CI [0.62, 0.05]; p¼0.02 to b¼0.26; 95% CI [0.53, 0.00]; p¼0.05) Friis et al. (Friis et al. Mediation role of HL between edu- Independent variables: education Low educational levels significantly associated with smoking, 2016) cational level and obesity and Dependent variables (outcomes): health behaviors physical inactivity, unhealthy diet and obesity Denmark health behaviors (smoking, physical inactivity, poor diet), obesity Significant partial mediation effect of HL (‘Understanding N¼29473 adults aged 25 years or Mediators: two sub-dimensions of HL as measured health information’ sub-dimension) between low educational more, population-based study with the HLQ: Understanding health informa- levels and daily smoking (6.6% of the variance), physical in- tion, Actively engage with providers activity (20.1%), unhealthy diet (13.3%), and obesity Analysis: multiple linear regressions, multiple medi- (16.2%) ation analyses Significant partial mediation effect of HL (‘Actively engage with health care providers’ sub-dimension) between low edu- cational levels and daily smoking (4.5% of the variance), physical inactivity (5.4%), and obesity (4.2%) Gupta et al. (Gupta Mediation role of FHL in racial dis- Independent variables: race (White, African Race significantly associated with HL (p<0.001), with higher et al., 2016) parities in cognitive decline American) odds of low HL levels among African-American compared USA N¼198 elderly aged 65 years or Dependent variables (outcomes): cognitive decline to Whites (OR¼34.45, 95% CI [8.78, 135.11]) older (Trail-Making Test TMT) Higher HL levels associated with less cognitive decline Stormacq C. Mediators: FHL (S-TOFHLA) (p¼0.01) Analysis: multiple linear regressions according to Significant association between race and cognitive decline Baron and Kenny (Baron and Kenny 1986) (p¼0.001)

Mediation analysis: HL partly mediated the relationship be- al et tween race and cognitive decline (25.3% of the variance) .

(continued) Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded Table 1: Continued disparities health and SES between HL of role mediating The Authors Aim and sample Variables and analysis Main results

Gwynn et al. (Gwynn Mediation role of FHL in racial dis- Independent variables: race (White, Black) Indirect path between black race and patient activation et al., 2016) parities in patient activation Dependent variables (outcomes): patient activation through HL significant (p¼0.008) USA N¼225 urban minority elderly (Patient Activation Measure PAM) Mediators: FHL (S-TOFHLA) Analysis: path analysis Hovick et al. (Hovick Factors that contribute to disparities Independent variables: SES (education, income); Lower HL levels among Blacks and Hispanics than Whites et al., 2014) in cancer risk knowledge and in- race/ethnicity (Black, Hispanic, White) (p<0.05) USA formation seeking Dependent variables (outcomes): cancer risk knowl- HL and cancer risk knowledge positively associated (b¼0.33, N¼1007, population-based study. edge; cancer risk information seeking p<0.001), no association with information seeking (b¼–.01, Mediators: FHL (NVS); social capital; perceived p¼0.703) seeking control; attitude toward seeking; subjec- Cancer risk knowledge positively associated with income and tive norms education, but not with race/ethnicity (p<0.05) Analysis: multiple linear regressions according to Indirect effects of SES and race on cancer risk knowledge Baron and Kenny (Baron and Kenny, 1986) through HL acting as a mediator. HL did not mediate the ef- fect of SES and race on cancer risk information seeking Howard et al. Impact of HL on differences in Independent variables: education; race (White, Inadequate HL more prevalent among those without high (Howard et al., health status and vaccination by Black, Hispanic); income school degrees (45%) than those with a high school degree 2006) educational attainment and race Dependent variables (outcomes): health status (13%) (p<0.001), and among Blacks (52%) compared to USA N¼3260 elderly persons enrolling in (physical health Short Form-12 [SF-12], mental Whites (19%) (p<0.001) a Medicare managed care plan health SF-12, self-reported health status); receipt Inadequate HL skills significantly associated with worse health of influenza vaccination; receipt of pneumococcal outcomes and less influenza vaccine use vaccination Significant positive association between higher education levels Mediators: FHL (S-TOFHLA) and physical and status (SF-12) (p¼0.013 and Analysis: regression models 0.004, respectively), and higher self-reported health (p<0.001). No significant association between education and receipt of influenza and pneumococcal vaccines (p¼0.117 and 0.206, respectively) Significant association between Black race and worse health sta- tus compared to whites (p¼0.012), and lower receipt of in- fluenza (p<0.001) and pneumococcal vaccines (p<0.001). Insignificant differences in SF-12 scores between Blacks and Whites Educational differences in physical and mental health reduced by HL by 25–41% respectively Racial differences in self-reported health reduced by HL by 25%. HL did not reduced racial differences in vaccination rates e7

(continued) Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded e8

Table 1: Continued Authors Aim and sample Variables and analysis Main results

Osborn et al. (Osborn Mediating effect of limited FHL on Independent variables: race (White, African Significant association between African-American race and low et al., 2007) the relationship between race and American) HL skills (AOR¼7.4, 95% CI [1.49, 10.9]) USA HIV-medication Dependent variables (outcomes): HIV medication African Americans race associated with nonadherence to medi- N¼204 patients infected with HIV adherence (Patient Medication Adherence cation, compared to Whites (AOR¼2.40 (95% CI [1.14, Questionnaire PMAQ) 5.08]) Mediators: FHL (REALM) Low HL skills associated with nonadherence to medication Analysis: multivariate regression models compared with adequate HL skills (p¼0.01) HL reduced the effect of African-American race on medication adherence by 25% to a point of nonsignificance (AOR¼1.80, 95% CI [0.51, 5.85]) Low HL: significant independent predictor of nonadherence (AOR¼2.12, 95% CI [1.93, 2.32]) Osborn et al. (Osborn Mediating effect of FHL, general nu- Independent variables: race/ethnicity (White, Association African-American race and glycemic control signif- et al., 2009) meracy, and diabetes-related nu- African American) icant USA meracy on the relationship Dependent variables (outcomes): Glycemic control Significant associations between African American race and between race and poor glycemic (A1C value) limited HL skills (r¼0.39, p<0.001), limited general nu- control in patients with diabetes Mediators: FHL (REALM); diabetes-related numer- meracy skills (r¼0.43, p<0.001), and limited diabetes-re- N¼383 adults with type 2 diabetes acy; general numeracy lated numeracy skills (r¼0.46, p<0.001) Analysis: Structural equation models Only diabetes-related numeracy significantly predicted glycemic control (r¼0.15, p<0.01), not HL or general numeracy African American race differences in glycemic control reduced by diabetes-related numeracy to nonsignificance (r¼0.10, NS) Osborn et al. (Osborn Mediating role of FHL and/or nu- Independent variables: race (White, African Low HL, low diabetes-related numeracy, and low general nu- et al., 2011) meracy in racial diabetes medica- American) meracy more prevalent among African Americans than USA tion adherence differences Dependent variables (outcomes): Diabetes medica- Whites N¼383 adults with type 2 diabetes tion adherence (medication adherence subscale of African American race significantly associated with less medica- the Summary of Diabetes Self-Care Activities tion adherence (r¼0.10, p<0.05) questionnaire SDSCA) Direct effect of HL on medication adherence (r¼0.12, p<0.02). Mediators: FHL (REALM); diabetes-related numer- Diabetes-related numeracy and general numeracy not related acy; general numeracy to diabetes medication adherence Stormacq C. Analysis: path analysis model Effect of race on adherence reduced by HL to nonsignificance (r¼0.09, p¼0.14)

(continued) tal et

. Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded h eitn oeo LbtenSSadhat disparities health and SES between HL of role mediating The

Table 1: Continued Authors Aim and sample Variables and analysis Main results

Pandit et al. (Pandit Mediating role of FHL in the associ- Independent variables: education attainment Lower education significantly associated with limited HL skills et al., 2009) ation between education, and hy- Dependent variables (outcomes): Hypertension (r2¼0.52, p<0.001) USA pertension knowledge and control knowledge (characteristics and symptoms of high Poorer hypertension knowledge significantly predicted by lower N¼330 hypertensive patients blood pressure); blood pressure control educational attainment and lower HL skills (unadjusted mul- Mediators: FHL (S-TOFHLA) tivariate analysis) Analysis: meditational analysis according to Baron Lower educational attainment and lower HL skills significantly and Kenny (Baron and Kenny, 1986) associated with uncontrolled hypertension (unadjusted mul- tivariate analysis) Relationship between education and hypertension knowledge no longer significant when considering HL as a mediator, with lower HL acting as a significant independent predictor of poor hypertension knowledge. Relationship between edu- cation and blood pressure control only slightly reduced Schillinger et al. Mediating role of FHL the relation- Independent variables: educational attainment Significant relationships between higher educational attainment (Schillinger et al., ship between education and glyce- Dependent variables (outcomes): Glycemic control and better glycemic control 2006) mic control among diabetes (HbA1c value) Higher HL skills associated with better glycemic control USA patients Mediators: FHL (S-TOFHLA) When HL added to the model, direct relationships between ed- N¼395 diabetes patients Analysis: path analysis using structural equation ucation and glycemic control reduced. Partial mediation modeling effect Van der Heide et al. FHL as a pathway between educa- Independent variables: education Lower education levels associated with lower HL skills (Van der Heide tion and health Dependent variables (outcomes): self-reported gen- Higher HL skills associated with better self-reported general et al., 2013b) N¼5136, population-based study eral health; self-reported physical health; self- health (b¼0.002, SE¼0.0006, p<0.001), physical health The Netherland reported mental health (Medical Outcomes Study (b¼0.017, SE¼0.006, p¼0.005), and mental health Short Form 12) (b¼0.012, SE¼0.004, p¼0.007) Mediators: FHL (HALS) Lower education associated with poorer self-reported general Analysis: mediation models, linear regression health, physical health, and mental health analyses Associations between education and self-reported general health, self-reported physical health, and self-reported men- tal health partially mediated by HL

(continued)

e9 Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded e10

Table 1: Continued Authors Aim and sample Variables and analysis Main results

Zou et al. (Zou et al., Factors that mediate the relationship Independent variables: SES (education, employ- Education, employment and income not associated with de- 2016) between SES and depressive ment, income); subjective social status pressive symptoms China Symptoms Dependent variables (outcomes): depressive symp- Significant association between lower subjective social status N¼321 patients with heart failure toms (Chinese version of Depression Subscale of and depressive symptoms (OR¼1.321; 95% CI [1.067, Hospital Anxiety and Depression 1.642]; p¼0.012) Scale HADSD) No significant associations between subjective social status and Mediators: access to health care; social support; HL access to healthcare and social support (Chinese version of HL Scale for Patients with Significant association between subjective social status and HL Chronic Disease) (b¼0.168; p¼0.001) Analysis: meditational analysis according to Baron Low HL levels significantly associated with depressive symp- and Kenny (Baron and Kenny, 1986) toms (OR¼1.065; 95% CI [1.036, 1.095]; p<0.001) Mediation analysis: relationship between subjective social sta- tus and depressive symptom fully mediated by HL (p<0.001)

AOR, adjusted odd ratio; CI, confidence interval; FHL, functional health literacy; HALS, Health Activities and Literacy Scale; HL, health literacy; HLQ, Health Literacy Questionnaire; NAAL, National Assessment of Adult Health Literacy; NS, nonsignificance; NVS, Newest Vital Sign; REALM, The Rapid Estimate of Adult Literacy in Medicine; RR, risk ratio; SES, socioeconomic status; S-TOFHLA, Test Of Functional Health Literacy in Adults, short version. .Stormacq C. tal et

. Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 September 27 on guest by https://academic.oup.com/heapro/article/34/5/e1/5068634 from Downloaded The mediating role of HL between SES and health disparities e11

Zou et al., 2016). With regard to educational attain- health-related outcomes and quality of life; five focused ment, a significant independent association between ed- on the mediating role of HL in the relationship between ucation level and HL was found, in that people with low SES and determinants of health-related behaviors; five educational attainment are more likely to have low HL studies focused on the mediating role of HL in dispar- levels. For example, Howard et al. (Howard et al., ities in health-related behaviors; and two on its mediat- 2006) found that inadequate HL was more prevalent ing role in the access to and use of healthcare. among people without a high school degree than among people with a high school degree (45% vs. 13% respec- tively, p <0.001). Moreover, HL shows a gradient Health-related outcomes and quality of life Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 depending on the level of education (Bennett et al., In a study using mediation analysis to examine the path- 2009): the lower the level of education, the more the ways between education and health, Van der Heide risk of low HL increases (Van der Heide et al., 2013b). et al. (Van der Heide et al., 2013b) found that partici- Two studies found that low HL is related to lower in- pants with a lower education level had lower HL and come (Ayotte et al., 2009; Adams et al., 2013). The rela- reported a worse self-rated general, physical and mental tionship of employment status with HL was only health status than those with adequate HL. While the addressed in one study (Pandit et al., 2009), where un- authors concluded that HL partly mediated the relation- employed or retired individuals tend to have lower HL ship between education and these three self-rated health skills than those employed. One study (Zou et al., 2016) indicators, they also found that the mediating role of HL examined the association of perceived social status with differed according to the indicator used, with a greater HL, showing that a lower subjective social status was as- effect of HL on the relationship between education and sociated with low HL levels (p<0.001). self-rated mental health. Similar results were found in the study by Bennett et al. (Bennett et al., 2009) among Race/ethnicity older adults, where low HL levels partially explained ra- Race and ethnicity are considered as markers of SES in cial and educational disparities in self-reported health the USA. Eleven studies reported an influence of race/ status. This result was also found in the study by ethnicity on HL (Howard et al., 2006; Osborn et al., Howard et al. (Howard et al., 2006), also among an 2007, 2009, 2011; Ayotte et al., 2009; Pandit et al., older population: HL partly explained racial and ethnic 2009; Bennett et al., 2009; Curtis et al., 2012; Gwynn disparities in health status, reducing educational and ra- et al., 2014; Hovick et al., 2014; Gupta et al., 2016), cial differences by 25% in physical health, and by 41% showing that being African American/Black or Latino/ in mental health measured with the SF-12 questionnaire. Hispanic is associated with higher odds of limited HL. Focusing on specific health-related outcomes, Friis et al. In a longitudinal cohort study exploring racial differen- (Friis et al., 2016) explored the mediating role of HL on ces in asthma outcomes, Curtis et al. (Curtis et al., the relationship between differences in educational levels 2012) found that 38.3% of African Americans had low and obesity, finding that HL partly mediates this rela- HL skills, compared with 31.3% for Latinos and 13.2% tionship. The ability to understand health information for Whites (p ¼0.002). A similar result was found in the accounted for 16.2% of the variance; and the ability to study by Gupta et al. (Gupta et al., 2016), in which actively engage with health care providers for 4.2%. 53.3% of African Americans were found to have low Curtis et al. (Curtis et al., 2012) used mediation analysis HL levels, compared to 4.8% for Whites. to explore the role of HL in racial disparities in asthma- related outcomes (asthma quality of life, emergency vis- its and hospitalization, and asthma control), revealing Gender that HL is a partial mediator between Latino race and The influence of gender on HL was only addressed in disparities for all these outcomes, and between African one of the studies in this review (Ayotte et al., 2009), American race and asthma quality of life, control and showing that women had significantly higher levels of emergency visits disparities. HL reduced asthma dispar- HL than men. ities between African American and White adults by 13–17%, and asthma quality of life disparities be- HL as a mediator of the relationship between tween Latinos and Whites by 21.2%. In a study explor- social and health-related disparities ing racial disparities in cognitive decline (Gupta et al., Seven of the studies included in the review examined the 2016), a mediation analysis showed that differences in mediating role of HL in the relationship between socio- HL levels partly explained racial disparities in cognitive economic and socio-demographic characteristics and decline, accounting for 25.3% of the variance in this e12 C. Stormacq et al. relationship. Finally, in a study exploring the association A mediation analysis by Pandit et al. (Pandit et al., between SES and depressive symptoms among patients 2009) showed that differences in HL levels could fully with heart failure, Zou et al. (Zou et al., 2016) found a explain educational disparities in hypertension knowl- full mediation effect of HL in the relationship between edge, with low HL skills acting as a significant predictor social status and depressive symptoms. of poor hypertension knowledge. Although HL was sig- nificantly associated with blood pressure control, its me- diating role between education and hypertension control Determinants of health-related behaviors was limited. Five studies explored the mediating role of HL in the re- Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 lationship between social disparities and determinants of health-related behaviors such as cancer risk percep- Health-related behaviors tions, information recall, health-related knowledge and Five studies explored the mediating role of HL in the re- patient activation. Adams et al. (Adams et al., 2013) ex- lationship between social disparities and health-related plored the mediating role of HL in the association be- behaviors. A study using path analysis to explore the as- tween SES and cancer risk perceptions related to lifestyle sociation between educational differences and glycemic behaviors, using structural equation modeling. In this control among diabetic patients showed that HL par- study, particular health-related behaviors (lack of physi- tially mediated this relationship (Schillinger et al., cal activity, smoking, lack of fresh fruit and vegetables 2006). In contrast, Osborn et al. (Osborn et al., 2009), consumption) were significantly linked to disadvantaged using structural equation modeling to explore various socioeconomic conditions such as low income and a potential mediators of racial disparities in glycemic con- poor educational level, yet people with a low HL level trol including HL, general numeracy, and diabetes- did not perceive these behaviors as potential risk factors related numeracy, did not find HL to mediate the for cancer. Structural equation modeling showed that relationship between race and glycemic control as did HL was a statistically significant mediator between SES diabetes-related numeracy, although being African and the perception of cancer risk related to lifestyle American was significantly associated with having lim- behaviors, with a partial mediation amounting to 29, ited HL skills. However, when exploring racial dispar- 4% of the total effect. In a similar vein, Ayotte et al. ities in diabetes medication adherence using path (Ayotte et al., 2009) used path analysis to show that HL analytic modeling on the same dataset, the same authors partially mediated the association between socioeco- found that differences in HL levels between African nomic characteristics (poor financial status, low educa- Americans and Whites partly explained the racial dis- tion, male gender, Black race) and health information parities in adherence to diabetes medication, to the point recall among 1190 patients with hypertension, conclud- where ‘HL reduced the effect of race on adherence to ing that ‘one of the mechanisms by which low HL might non-significance’ [(Osborn et al., 2011), p. 4]. One study influence health outcomes is by its relationship with the also explored the mediating role of HL in the relation- ability to recall important health information’ [(Ayotte ship between race and adherence to HIV treatment et al., 2009), p. 429]. This hypothesis seems to be con- (Osborn et al., 2007), showing that while African firmed by a number of other studies. In a study explor- American participants had lower HL levels than Whites ing socioeconomic and ethnic disparities in specific (OR ¼7.40, 95% CI [1.49, 10.9]) and were less adher- health-related knowledge and information seeking using ent to HIV medication (OR¼2.40, 95% CI [1.14, multiple linear regressions with the Baron and Kenny 5.08]), considering HL reduced the effect of race on (Baron and Kenny 1986) method, Hovick et al. (Hovick medication adherence by 25% to nonsignificance. et al., 2014) found HL to partially mediate the relation- Finally, Friis et al. (Friis et al., 2016), when investigating ship between race and SES (educational level and the mediating role of two dimensions of HL (‘under- income) on the one hand, and cancer risk knowledge on standing health information’ and ‘actively engaging with the other hand. In this study, HL did not mediate the re- health care providers’) in the association between educa- lationship between SES and cancer risk information tion disparities and various unhealthy behaviors such as seeking. Another study, which used path analysis to ex- smoking, physical inactivity and poor diet, found both plore the mediating role of HL in the association be- components of HL to only partly mediate the relation- tween racial disparities and patient activation (Gwynn ship between low educational attainment and smoking, et al., 2014), showed that HL fully mediated racial dis- with ‘understanding health information’ and ‘actively parities in patient activation, with Blacks showing signif- engaging with health care providers’ accounting for icantly lower levels of HL than Whites (p<0.001). 6.6% and 4.5% of the variance, respectively. The same The mediating role of HL between SES and health disparities e13 was found for the associations between low education seems a logical choice, as education impacts on both em- levels and physical inactivity, where ‘understanding ployment status and income (O’Neill et al., 2014), health information’ accounted for 20.1% of the vari- which in turn allows access to various resources such as ance, and ‘actively engaging with health care providers’ better living conditions, healthy food and health insur- for 5.4%. In the relationship between low educational ance (O’Neill et al., 2014). As such, education can be attainment and poor diet, only ‘understanding health in- regarded as a major determinant of HL. However, the formation’ was found to be a partial mediator (13, 3% use of this indicator is often disputed. Van der Heide of the variance). et al. (Van der Heide et al., 2013b) showed that, al- though education and HL levels are strongly associated, Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 within each group defined by educational level there is a Access and use of healthcare services percentage of people with limited HL. This result is con- Two studies addressed the mediating role of HL in dis- firmed by other studies (Gazmararian et al., 1999; parities in the access to and use of healthcare. Bennett Adams et al., 2009; Kaphingst et al., 2012; Van der et al. (Bennett et al., 2009) explored differences in pre- Heide et al., 2013a,b). This presence of low HL across ventive healthcare use using Baron and Kenny’s ap- groups defined by educational level suggests that educa- proach (Baron and Kenny, 1986) to analyze the tion, measured by the number of school years completed mediation effect, finding that educational disparities in or the highest degree obtained, is an ‘inaccurate indica- flu vaccine, mammography and dental care use were sig- tor of someone’s true educational attainment’ nificantly mediated by HL. HL also significantly medi- [(Gazmararian et al., 1999), p. 549], which does not ated the relationship between ethnic disparities (black vs take into account the differences in cognitive skills white) and flu vaccine rates. Conversely, Howard et al. across individuals (Matsuyama et al., 2011). Individuals (Howard et al., 2006) examined the extent to which HL with the same educational level may indeed have differ- mediated the relationship between education and race ent levels of cognitive abilities (Raudenbush and Kasim, and vaccination disparities (influenza and pneumococ- 1998). This is partly due to different educational oppor- cal) among Medicare enrollees, but found only a weak tunities outside the school, leading to differences in effect of HL in explaining differences in vaccination knowledge and skills acquisition, as well as to differen- rates. ces in the quality and type of education and educational systems across regions and countries (Matsuyama et al., 2011; Yamashita et al., 2013). Thus, to fully capture the DISCUSSION extent to which people can access, understand, appraise Although socio-economic disparities in health are one of and apply health information, it is necessary to measure the biggest challenges to public health, and despite the actual cognitive skills and specific knowledge, rather potential role of HL as a factor that may explain the re- than simply asking about the level of education com- lationship between social disparities and health out- pleted (Kaphingst et al., 2012). Race/ethnicity with re- comes, research investigating the relationship between gard to its role in HL was also very frequently social disparities, HL and health disparities remains addressed. This is not surprising given that most of the scarce. This paper provides an integrative review of included studies have been conducted in the USA, where those studies that explored socioeconomic and socio- race and ethnicity are often considered as indicators of demographic factors associated with low HL levels, and the SES. However, our conclusion about the relationship shed light on the mediating role of HL in the relation- between race/ethnicity and HL must be nuanced. ship between SES and health disparities, synthesizing the Indeed, and although recent European studies tend to findings from 16 published studies. show that people from some ethnic minority groups The results confirm, first of all, that a disadvantaged have lower HL levels than the indigenous population social or SES contributes to low HL. The main socioeco- (van der Gaag et al., 2017), the role of race/ethnicity is nomic and socio-demographic factors that seem to influ- not yet sufficiently studied and described in an ence HL are educational attainment, income, European context to draw the same conclusions as for occupation and race/ethnicity. These results are congru- the US studies. Therefore, to generalize this finding, ent with the systematic review of Paasche-Orlow et al. more European research on the relationship between (Paasche-Orlow et al., 2005), who also found that edu- race/ethnicity and HL are needed. cation, income and ethnicity are associated with low HL Our review not only confirmed the contributing role levels. Among the indicators of SES, educational level is of HL in health disparities as shown in the systematic re- the most frequently addressed in HL research. This view of Mantwill et al. (Mantwill et al., 2015), but also e14 C. Stormacq et al. offered a deeper understanding by highlighting the par- advanced review methods such as a meta-analysis. Even tial mediating role of HL in the relationship between so- so, this review allowed to present the current status of cial and socioeconomic factors and health disparities. the research into this topic and to suggest directions for Specifically, HL was shown to mediate the association further research. between socioeconomic characteristics on one hand, and Secondly, across the publications included in this re- health outcomes, health-related behaviors, and the ac- view, a large variety of measurement tools were used to cess to and use of healthcare on the other hand. In other assess HL. As a result, the thresholds to define low or words, as a result of disadvantaged social conditions, a limited HL may also vary across studies. This possibility low or inadequate level of HL in turn can strongly con- is illustrated in a study by Barber et al. (Barber et al., Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 tribute to health disparities. As such, the persistence of 2009) showing that in the same sample, the rates of low health disparities can partly be explained by the fact that HL differ according to the scale that is used, ranging people with a more favorable SES have a better access to from 6.8% of inadequate or marginal HL with the (new) health information, make more effective use of TOFHLA, to 10.6% with the REALM, and 26.0% with that information, and have a better access to resources the NVS. In this regard, it is important to note that to act upon this information (Phelan et al., 2010; Diez twelve of the 16 studies included in this review focused Roux, 2012; Mackenbach, 2012). The mediating role of on functional HL, which mainly refers to reading skills HL that is highlighted by this review not only leads to a and reading comprehension, and does not consider more better understanding of health disparities, but also offers advanced HL skills such as decision-making, problem- possibilities to counter these disparities in the sense that solving, or critical thinking. As such, these studies do HL becomes a potential leverage for action on the social not evaluate the fullest extent of the HL concept, and do determinants contributing to health inequalities and dis- not evaluate how people access, understand, appraise parities. Given the difficulty of acting directly on social and act upon health information, as is the case with conditions, strategies to reduce health inequalities can broader conceptualizations of HL (Sorensen et al., focus on intermediate factors such as HL. Interventions 2012). This finding represents a limitation of the current that aim to increase HL or that take people’s low HL literature. While the predominance of studies focusing level into account will not lift people from disadvan- only on functional HL can be explained by the fact that taged socioeconomic conditions but can be considered many studies were conducted in the USA, where the as a ‘midstream’ strategy to reduce the impact of ‘up- functional perspective of HL is predominant, and by the stream’ socioeconomic determinants on ‘downstream’ fact that comprehensive operational definition of HL are disparities in health. Adopting a proportionate univer- relatively recent, further research is needed that consid- salist approach (Marmot Review, 2010), health services ers not only the functional aspect of HL, but also the and health promotion interventions should therefore be communicative and critical aspects of the concept as de- developed and implemented with an ‘equity lens’ scribed by Nutbeam (Nutbeam, 2000) and Sorensen (O’Neill et al., 2014), and be as accessible, clear, under- et al. (Sorensen et al., 2012). HL need to be considered standable and usable as possible for everyone in the pop- in a multidimensional way ‘as a distinct concept, rather ulation, but with a particular focus on targeting, than a derivative concept from literacy and numeracy reaching and engaging the most vulnerable groups dis- skills’ [(Adams et al., 2009), p. 534]. proportionately affected by low HL (Yamashita, 2011). It should also be noted that most of the studies in- However, although HL appears as a promising strat- cluded in this review regard (low) HL as a risk factor, egy, addressing low HL levels should not be considered rather than an asset and a potential leverage for the de- as an ‘easy option’ to alleviate health disparities. It does velopment of skills and capacities that enable people to not preclude the need to tackle the root causes of health exert greater control over their health and the factors disparities and to implement profound changes at the that shape health (Nutbeam, 2008). Future research into political, economical and organizational levels to ad- the mediating role of HL in the relationship between so- dress these disparities. cial disparities and health outcomes should therefore re- While the above recommendations derive from the gard HL as a multidimensional, enabling concept and findings of this review, it is important to point out three explore its role as a ‘resource for daily living’ (Adams limitations of the existing literature on HL and social et al., 2009). disparities. Firstly, the number of published studies that A third limitation of the existing literature is that the directly investigated the mediating role of HL in the rela- majority of studies used cross sectional or observational tionship between socioeconomic inequalities and health designs, and therefore provided a description of associa- inequalities is limited, which prevented the use of more tions between variables rather than causal relationships. The mediating role of HL between SES and health disparities e15

More prospective or longitudinal studies are needed to strategic, and statistical comparisons. Journal of Personality improve the understanding of the causal mechanisms and , 51, 1173–1182. linking social condition, HL and health outcomes. Bennett, I. M., Chen, J., Soroui, J. S. and White, S. (2009) The contribution of health literacy to disparities in self-rated health status and preventive health behaviors in older adults. CONCLUSION Annals of Family Medicine, 7, 204–211. Bostock, S. and Steptoe, A. (2012) Association between low This integrative review showed that disadvantaged so- functional health literacy and mortality in older adults: lon- cial and socioeconomic conditions contribute to low HL gitudinal cohort study. BMJ, 344, e1602.

levels, whereby low SES, and particularly educational Braveman, P. (2014) What are health disparities and health eq- Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 attainment, is the most important factor influencing HL. uity? We need to be clear. Public Health Reports, 129, 5–8. It was also shown that HL impacts on a variety of Braveman, P. and Gruskin, S. (2003) Defining equity in health. health-related outcomes, and that HL mediates the rela- Journal of and , 57, 254–258. tionship between SES and health disparities. As such, Cochrane Health Equity Field and Campbell Equity Methods HL can be considered as a modifiable risk factor for Group. Kavanagh, J., Oliver, S and Lorenc, T. (2008) health disparities on which it is possible to act in order Reflections on developing and using PROGRESS-Plus. to reduce health disparities and reach greater equity in Equity Update, 2, 1–3. health. Curtis, L. M., Wolf, M. S., Weiss, K. B. and Grammer, L. C. (2012) The impact of health literacy and socioeconomic sta- tus on asthma disparities. Journal of Asthma, 49, 178–183. FUNDING Diez Roux, A. V. (2012) Conceptual approaches to the study of health disparities. Annual Review of Public Health, 33, This research did not receive any specific grant 41–58. from funding agencies in the public, commercial, or not- Evans, G. W. and Kantrowitz, E. (2002) Socioeconomic status for-profit sectors. and health: the potential role of environmental risk expo- sure. Annual Review of Public Health, 23, 303–331. REFERENCES Freedman, D. A., Bess, K. D., Tucker, H. A., Boyd, D. L., Tuchman, A. M. and Wallston, K. A. (2009) Public health Adams, R. J., Appleton, S. L., Hill, C. L., Dodd, M., Findlay, C. literacy defined. American Journal of Preventive Medicine, and Wilson, D. H. (2009) Risks associated with low func- 36, 446–451. tional health literacy in an Australian population. Medical Friis, K., Lasgaard, M., Rowlands, G., Osborne, R. H. and Journal of Australia, 191, 530–534. Maindal, H. T. (2016) Health literacy mediates the relation- Adams, R. J., Piantadosi, C., Ettridge, K., Miller, C., Wilson, C., ship between educational attainment and health behavior: a Tucker, G. et al. (2013) Functional health literacy mediates Danish population-based study. Journal of Health the relationship between socio-economic status, perceptions , 21, 54–60. and lifestyle behaviors related to cancer risk in an Australian Garcia, S. F., Hahn, E. A. and Jacobs, E. A. (2010) Addressing population. Patient Education and Counselling, 91, low literacy and health literacy in clinical oncology practice. 206–212. Journal of Supportive Oncology, 8, 64–69. Adler, N. E. and Newman, K. (2002) Socioeconomic disparities Gazmararian, J. A., Baker, D. W., Williams, M. V., Parker, R. in health: pathways and policies. Health Affairs, 21, 60–76. M., Scott, T. L., Green, D. C. et al. (1999) Health literacy Ayotte, B. J., Allaire, J. C. and Bosworth, H. (2009) The associa- among Medicare enrollees in a managed care organization. tions of patient demographic characteristics and health in- JAMA, 281, 545–551. formation recall: the mediating role of health literacy. Gazmararian, J. A., Curran, J. W., Parker, R. M., Bernhardt, J. Aging, Neuropsychology, and Cognition, 16, 419–432. M. and DeBuono, B. A. (2005) Public health literacy in Baker, D. W., Wolf, M. S., Feinglass, J., Thompson, J. A., America: an ethical imperative. American Journal of Gazmararian, J. A. and Huang, J. (2007) Health literacy and Preventive Medicine, 28, 317–322. mortality among elderly persons. Archives of Internal Gupta, V. K., Winter, M., Cabral, H., Henault, L., Waite, K., Medicine, 167, 1503–1509. Hanchate, A. et al. (2016) Disparities in age-associated cog- Barber, M. N., Staples, M., Osborne, R. H., Clerehan, R., Elder, nitive decline between African-American and Caucasian C. and Buchbinder, R. (2009) Up to a quarter of the populations: the roles of health literacy and education. Australian population may have suboptimal health literacy Journal of the American Geriatrics Society, 64, 1716–1723. depending upon the measurement tool: results from a Gwynn, K. B., Winter, M. R., Cabral, H. J., Wolf, M. S., population-based survey. Health Promotion International, Hanchate, A. D., Henault, L. et al. (2014) Predicting cancer 24, 252–261. risk knowledge and information seeking: the role of social Baron, R. and Kenny, D. (1986) The moderator-mediator vari- and cognitive factors. , 29, able distinction in social psychological research: conceptual, 656–668. e16 C. Stormacq et al.

Hovick, S. R., Liang, M. C. and Kahlor, L. (2014) Predicting explains racial disparities in diabetes medication adherence. cancer risk knowledge and information seeking: the role of Journal of Health Communication, 16, 268–278. social and cognitive factors. Health Communication, 29, Osborn, C. Y., Cavanaugh, K., Wallston, K. A., White, R. O. 656–668. and Rothman, R. L. (2009) Diabetes numeracy: an over- Howard, D. H., Sentell, T. and Gazmararian, J. A. (2006) looked factor in understanding racial disparities in glycemic Impact of health literacy on socioeconomic and racial differ- control. Diabetes Care, 32, 1614–1619. ences in health in an elderly population. Journal of General Osborn, C. Y., Paasche-Orlow, M. K., Davis, T. C. and Wolf, Internal Medicine, 21, 857–861. M. S. (2007) Health literacy: an overlooked factor in under- Kaphingst, K. A., Goodman, M., Pyke, O., Stafford, J. and standing HIV health disparities. American Journal of

Lachance, C. (2012) Relationship between self-reported ra- Preventive Medicine, 33, 374–378. Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 cial composition of high school and health literacy among Paasche-Orlow, M. K., Parker, R. M., Gazmararian, J. A., community health center patients. and Nielsen-Bohlman, L. T. and Rudd, R. R. (2005) The preva- Behavior, 39, 35–44. lence of limited health literacy. Journal of General Internal Kickbusch, I. and Maag D. (2008) Health Literacy. In Medicine, 20, 175–184. Heggenhougen, K. and Qah, S. (eds), International Pandit, A. U., Tang, J. W., Bailey, S. C., Davis, T. C., Bocchini, Encyclopedia of Public Health, 1st edn. Academic Press, San M. V., Persell, S. D. et al. (2009) Education, literacy, Diego, pp. 204–211. and health: mediating effects on hypertension knowledge Kickbusch, I. S. (2001) Health Literacy: addressing the health and control. Patient Education and Counselling, 75, and education divide. Health Promotion International, 16, 381–385. 289–297. Phelan, J. C., Link, B. G. and Tehranifar, P. (2010) Social condi- Kutner, M., Greenberg, E., Jin, Y. and Paulsen, C. (2006) tions as fundamental causes of health inequalities: theory, The Health Literacy of America’s Adults: Results from the evidence, and policy implications. Journal of Health and 2003 National Assessment of Adult Literacy (NCES Social Behavior, 51, S28–S40. 2006-483). National Center for Education Statistics, Pleasant, A. and Kuruvilla, S. (2008) A tale of two health litera- Washington. cies: public health and clinical approaches to health literacy. Lee, S. Y. D., Tsai, T. I., Tsai, Y. W. and Kuo, K. N. (2010) Health Promotion International, 23, 152–159. Health Literacy, health status, and healthcare utilization of Raudenbush, S. W. and Kasim, R. M. (1998) Cognitive skill and Taiwanese adults: results from a national survey. BMC economic inequality: findings from the national adult liter- Public Health, 10, 614. acy survey. Harvard Educational Review, 68, 33–79. Mackenbach, J. P. (2012) The persistence of health inequalities Schillinger, D., Barton, L. R., Karter, A. J., Wang, F. and Adler, in modern welfare states: the explanation of a paradox. N. (2006) Does literacy mediate the relationship between Social Science and Medicine, 75, 761–769. education and health outcomes? A study of a low-income Mantwill, S., Monestel-Umana,~ S. and Schulz, P. J. (2015) The population with diabetes. Public Health Reports, 121, relationship between health literacy and health disparities: a 245–254. systematic review. PLoS One, 10, e0145455. Sorensen, K., Van den Broucke, S., Fullam, J., Doyle, G., Marmot Review. (2010) The Marmot Review Final Report: Fair Pelikan, J., Slonska, Z. et al. (2012) Health literacy and pub- Society, Healthy Lives: Strategic Review of Health lic health: a systematic review and integration of definitions Inequalities in England Post 2010. Marmot Review, and models. BMC Public Health, 12, 80. London. Sorensen, K., Pelikan, J. M., Ro¨ thlin, F., Ganahl, K., Slonska, Matsuyama, R. K., Wilson-Genderson, M., Kuhn, L., Z., Doyle, G. et al. (2015) Health literacy in Europe: com- Moghanaki, D., Vachhani, H. and Paasche-Orlow, M. parative results of the European health literacy survey (2011) Education level, not health literacy, associated with (HLS-EU). European Journal of Public Health, 25, information needs for patients with cancer. Patient 1053–1058. Education and Counselling, 85, e229–e236. Sun, X., Shi, Y., Zeng, Q., Wang, Y., Du, W., Wei, N. et al. Nutbeam, D. (2000) Health literacy as a public health goal: a (2013) Determinants of Health Literacy and health behavior challenge for contemporary health education and communi- regarding infectious respiratory diseases: a pathway model. cation strategies into the 21st century. Health Promotion BMC Public Health, 13, 261. International, 15, 259–267. The Joanna Briggs Institute. (2014) Joanna Briggs Institute Nutbeam, D. (2008) The evolving concept of health literacy. Reviewers’ Manual: 2014 Edition. Adelaide, The Joanna Social Science and Medicine, 67, 2072–2078. Briggs Institute. O’Neill, J., Tabish, H., Welch, V., Petticrew, M., Pottie, K., Tugwell, P., Petticrew, M., Kristjansson, E., Welch, V., Clarke, M. et al. (2014) Applying an equity lens to interven- Ueffing, E., Waters, E. et al. (2010) Assessing equity in tions: using PROGRESS ensures consideration of socially systematic reviews: realising the recommendations of the stratifying factors to illuminate inequities in health. Journal Commission on Social Determinants of Health. BMJ, 341, of Clinical Epidemiology, 67, 56–64. c4739. Osborn, C. Y., Cavanaugh, K., Wallston, K. A., Kripalani, S., Van den Broucke, S. (2014) Health literacy: a critical concept Elasy, T. A., Rothman, R. L. et al. (2011) Health literacy for public health. Archives of Public Health, 72, 10. The mediating role of HL between SES and health disparities e17 vanderGaag,M.,vanderHeide,I.,Spreeuwenberg,P.M.M., Whittemore, R. and Knafl, K. (2005) The integrative review: Brabers, A. E. M. and Rademakers, J. J. D. J. M. (2017) updated methodology. Journal of Advanced , 52, Health literacy and primary health care use of ethnic minorities 546–553. in the Netherlands. BMC Health Services Research 17,350. Yamashita T. (2011) Health Literacy and Health Outcomes: Van der Heide, I., Rademakers, J., Schipper, M., Droomers, M., Implications for Social Determinants of Health, Health Sorensen, K., and Uiters, E. (2013a) Health literacy of Disparity and Learning for Health over the Life Course. Dutch adults: a cross sectional survey. BMC Public Health, (Doctor of Philosophy). Miami University, Oxford, OH. 13, 179. Yamashita, T., John Bailer, A. and Noe, D. A. (2013) Van der Heide, I., Wang, J., Droomers, M., Spreeuwenberg, P., Identifying at-risk subpopulations of Canadians with limited

Rademakers, J. and Uiters, E. (2013b) The relationship be- health literacy. Epidemiology Research International, 2013, Downloaded from https://academic.oup.com/heapro/article/34/5/e1/5068634 by guest on 27 September 2021 tween health, education, and health literacy: results from the 110 (Article ID 130263). Dutch Adult Literacy and Life Skills Survey. Journal of Zou, H., Chen, Y., Fang, W., Zhang, Y. and Fan, X. (2016) The Health Communication, 18, 172–184. mediation effect of health literacy between subjective social Whitehead, M. (1991) The concept and principles of equity and status and depressive symptoms in patients with heart fail- health. Health Promotion International, 6, 217–228. ure. Journal of Psychosomatic Research, 91, 33–39.