| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Dan Morrow, PhD, Beckman Institute of Advanced Science & Technology, University of Illinois at Urbana–Champaign, Urbana, IL 61801. E-mail: dgm{at}uiuc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Health literacy • Cognitive ability • Health communication • Heart failure
More specifically, we focused on relationships between health literacy and cognitive abilities to better understand which skills and abilities are impaired in low-literacy patients. This knowledge, in turn, could lead to a more comprehensive model of the antecedents of health literacy and consequences for health care. Such a model would help explain associations between health literacy and demographic variables such as age and race, which researchers have investigated because of their interest in health literacy as a mediator of differences in health care outcomes related to socioeconomic status (Ad Hoc Committee, 1999; DeWalt et al., 2004). The model would also guide development of strategies for mitigating effects of low literacy on health behaviors and outcomes.
We adapted an initial model, presented in Figure 1, from Nielson-Bohlman and colleagues (2004, p. 33) and elaborated on it in terms of theories of comprehension (Kintsch, 1998) and cognitive determinants of self-care (Park & Jones, 1997). It organizes what is already known about correlates of health literacy and frames the issues motivating our study. Ratzan and Parker (2000, pp. ix) has defined health literacy as the " ... capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions." This definition links health literacy, like the broader concept of literacy, to a wide range of skills such as comprehension of linguistic, graphic, and numeric information (Nielsen-Bohlman et al.). Although comprehension is central to the concept of health literacy, investigators rarely use theories of comprehension to analyze this critical concept.
|
Demographic and Education Variables
Researchers have found age-related decreases for health literacy as well as general literacy measures that are not fully explained by differences in health status (Gazmararian et al., 1999; Schillinger et al., 2002; Williams, Davis, Parker, & Weiss, 2002). African American and Hispanic middle-aged and older adults tend to score lower than non-Hispanic White or Asian adults on standard health literacy measures (Arnold et al., 2001; Bennett et al., 1998; Gazmararian et al.; Schillinger et al.). These differences partly reflect health differences, because African American adults are more likely than non-Hispanic White adults to have chronic conditions (Lafata, Pladevall, Divine, Heinen, & Philbin, 2004) that compromise cognition (Izquiero-Porrea & Waldstein, 2002).
Number of years of education often explains large amounts of variance in health literacy as well as general literacy measures (Baker, Gazmararian, Sudano, & Patterson, 2000; Gazmararian et al., 1999; Schillinger et al., 2002). Education may enhance literacy by systematically imparting comprehension strategies (e.g., drawing inferences) and accelerating reading experience by promoting successful engagement with text, which automatizes word recognition processes and increases the knowledge that supports situation models (Stanovich, West, & Harrison, 1995).
However, number of years of education does not fully explain differences in general literacy (e.g., Barnes, Tager, Satariano, & Yaffe, 2004; Manly, Jacobs, Touradji, Small, & Stern, 2002) or health literacy (Andrus & Roth, 2002). This may reflect the fact that the number of school years a person has is a poor measure of educational attainment (Baker, Parker, Williams, Clark, & Nurss, 1997; Manly et al.) or that the many years of post-education reading experience is more important than formal education for determining literacy (Stanovich, West, & Harrison, 1995). Similarly, controlling for age differences in education often attenuates but does not eliminate age differences in literacy (Baker et al., 2000).
Cognitive Abilities and Health Literacy
Education, cognitive ability, and literacy are intertwined, although it is difficult to unravel the direction of causality. Education often improves cognitive function as well as literacy, but the more cognitively adept are also likely to advance further in education and create opportunities for cognitive growth (Gottfredson & Dreary, 2004; Perfetti, 1994). Still, there is evidence that cognition (e.g., working memory capacity) relates more to literacy than to years of education among older adults (Barnes et al., 2004; Manly et al., 2002). Therefore, it is likely that cognitive differences help explain the effects of education and demographic variables such as age on literacy. Differences in sensory abilities may also mediate the effects of age on literacy because sensory and cognitive abilities are increasingly coupled in advanced age (Lindenberger & Baltes, 1994). This conclusion is consistent with evidence that age-related differences in cognitive abilities partly mediates relationships between age and self-care (Park & Jones, 1997). Although Baker and colleagues (2000) found that a global measure of cognitive impairment (the Mini-Mental State Examination) was not associated with age-related differences in health literacy, they did not measure cognitive abilities more directly related to comprehension. For example, successful comprehension requires integrating current with previously read information, and integration is more successful to the extent previous information is easy to access from working memory (Kintsch, 1998). Thus, individual differences in working memory capacity, processing speed, or other cognitive abilities may explain differences in health literacy.
Figure 1 also suggests that education influences literacy via paths other than cognition by serving as a marker for broader cultural factors. Highly educated adults more likely than less educated adults to share cultural norms and beliefs with health care professionals, which enables them to more easily access and understand health-related information or predisposes them to accept the information as relevant (Nielsen-Bohlman et al., 2004). We did not investigate these factors in the present study.
Study Objectives
Previous studies have found that the association of health literacy with demographic variables is partly mediated by education. However, the association between literacy and cognitive abilities is less well established. In the current study, we examined whether health literacy in a diverse sample of adults with CHF was related to sensory and cognitive abilities (independent of education) and whether these relationships mediated effects of demographic variables on literacy.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
A self-report questionnaire administered at study recruitment measured demographic variables (age, gender, race) and education. Participants self-identified their racial group by choosing from a set of options. For the regression analyses, we coded racial group as African American or Other. We measured health status with the New York Health Association classification of CHF functional status (Critical Committee, 1964) and by the Charlson Comorbidity Index (Charlson, Pompei, Alex, & MacKensie, 1987).
We measured health-related literacy with the reading component of the Short Test of Functional Health Literacy in Adults (STOFHLA; Baker, Williams, Parker, Gazmararian, & Nurss, 1999). This test requires about 7 min to administer and measures the ability to read and understand actual health-related passages with readability levels of 4.3 and 10.4 grade level (Gunning–Fog Index). Scores on this version of the test range from 0 to 36 (0–16 = inadequate health literacy, 17–22 = marginal literacy, and 23–36 = adequate literacy) and tend to decline with age (Baker et al., 2000).
We measured visual function with the Snellen Chart, a standardized test of visual acuity consisting of a chart with a series of letters in gradually decreasing sizes placed at a standard distance from the subject. We assessed auditory function with the Speech Discrimination Screening Task, part of the Arizona Battery for Communication Disorders of Dementia (Bayles & Tomoeda, 1993), which tests the ability to discriminate word pairs that differ only in the first phoneme (e.g., "bare" and "dare"). We assessed speech comprehension with the Revised Token Test (McNeil & Prescott, 1978), which requires listeners to manipulate objects of varying size, shape, and color in response to spoken instructions. This test has discriminated children with various neurological or linguistic deficits from normally developing children, in part because of the working memory demands of comprehension (Aram & Ekelman, 1987). We measured verbal working memory, or the ability to simultaneously store and manipulate verbal information, directly with the Listening Span task. In this task, participants answer questions about progressively larger sets of simple spoken sentences (1–6 sentences) and then recall the final word of each sentence in the set (for details on materials and scoring, see Salthouse & Babcock, 1990). The working memory measure was the size of the sentence set for which the participant could reliably recall the sentences' final words. Age-related differences on this task account for part of the age-related variance in reasoning and memory tasks (Salthouse & Babcock). We measured processing speed with the Letter Comparison and Pattern Comparison tasks (Salthouse, 1991). In these timed paper-and-pencil tasks, participants decide as rapidly as possible whether pairs of letter sets or line patterns are the same or different. Age differences on these tasks account for age-related variance in verbal reasoning (Salthouse) and memory (Park et al., 1996) tasks.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Second, we assessed which of the variables that had been significant in the simple linear analyses predicted health literacy when the influence of the other variables was controlled. Following Figure 1, we first examined the association of the demographic and health variables (age, race, gender, and health as indexed by the comorbid measure entered together) with the STOFHLA measure (Step 1). We then examined whether the effects of these variables remained significant when we entered education into the equation (Step 2). Rather than present all higher-order regression models (e.g., effects of demographics and education with cognitive but not sensory variables controlled), we present here a parsimonious model that accounts for the most variability in health literacy with the fewest predictor variables (Step 3). For comparison, we also present the model with all variables entered (Step 4).
Participants
The majority (67%) of participants (age range, 47–89 years old) were female. With respect to race, 48% of participants were African American, 49% were non-Hispanic White, 2% were Asian, and 1% were Native American. Forty percent of participants scored at the two most severe levels of the New York Heart Association measure for CHF function (Level 1 [least severe] = 19%, Level 2 = 41%, Level 3 = 35%, and Level 4 [most severe] = 5%). In addition, participants reported on average 3.2 co-morbid conditions and took a mean of 9.5 medications (SD = 3.6) at the time of the study. Almost one third (28%) of participants had marginal or inadequate health-related literacy (STOFHLA score < 16). This was consistent with previous studies with older adult samples (e.g., Gazmararian et al., 1999) and indicated an adequate range of scores to investigate correlates of health literacy.
Simple Linear Regression Analyses of Health Literacy
We examined associations between the STOFHLA measure and the demographic, health, and cognitive/sensory measures. Table 1 presents mean and standard deviation values for the continuous demographic and individual difference variables. This table also presents the demographic and individual difference scores separately for the African American and Other groups. Typical of health literacy and cognitive aging studies, age was negatively associated with the health literacy measure (r = –.11, p <.05) and with the cognitive and sensory ability measures (rs –.14 to –.34, ps <.05). Table 2 presents results of these regression analyses with STOFHLA scores separately regressed on each predictor. The scores were lower for participants who were male, African American, older, less educated, had more comorbidities, or scored lower on all cognitive ability measures as well as on the speech discrimination test. Literacy was not significantly associated with the Snellen test or the New York Health Association measure. A comparison of the R2 values directly obtained from the regression model fitting procedure (which is interpreted as the percentage of the variation in the criterion variable explained by the regression model) showed that education (R2 = 21.83) and the cognitive measures (Comparison Test R2 = 29.4; Revised Token Test R2 = 26.64) accounted for the largest amount of variability in health literacy.
|
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In other studies, education was associated with the STOFHLA measure but did not eliminate significant associations of age and health literacy when controlled (e.g., Baker et al., 2000). This may reflect differences in samples across studies. For example, because mean participant age was higher in Baker and colleagues' (2000) study than in our study (73 vs 63 years, respectively), the association of age and literacy may have been greater in the former study.
Health Literacy and Cognitive Function
Perhaps the most interesting finding from our study was that cognitive ability explained so much variance in health literacy with education controlled. The finding that processing speed was more predictive than the working memory measure of health literacy (Step 3) may reflect limitations on the working memory measure because performance on the listening span task was generally low, with limited variability. Indeed, in an earlier study we found that differences in patients' memory for CHF medication instructions were related to health literacy, which could be partly explained by differences on a working memory measure (Morrow et al., 2005).
The relationship between processing speed and health literacy is unlikely due (only) to the fact that the STOFHLA is a timed test. Processing speed measures have also been found to predict performance on untimed cognitive tasks (involving comprehension, memory, and reasoning; e.g., Salthouse, 1991). Adults with reduced speed of processing are likely to more slowly access word meanings and integrate them into a coherent text representation, which may impair comprehension. Thus, individual differences in health literacy among middle-aged and older adults partly reflect differences in mental processing speed and other fundamental cognitive mechanisms, as predicted by theories that identify such mechanisms as constraints on comprehension efficiency (Kintsch, 1998; Wingfield & Stine-Morrow, 2000). Baker and colleagues (2000) found that age-related declines in literacy were reduced but not eliminated by a global measure of cognitive function (the Mini-Mental State Examination). As suggested by these authors, this may reflect the lack of direct measures of specific cognitive abilities in that study.
Sensory measures did not predict health literacy in the present study (also see Baker et al., 2000). Studies with more complex visual function measures (which likely share variance with cognitive measures) have found associations of health literacy and sensory functioning (Echt & Schuchard, 2004).
Relationships of health literacy to race were not explained by differences in education and cognition. Other studies have found that the impact of race on cognitive performance is attenuated when controlling for education or other socioeconomic variables (Aiken, Marsiske, & Whitfield, 2004). Perhaps differences in the quality rather than amount of education accounted for differences in health literacy between racial groups in our study (Gazmararian et al., 1999). Indeed, Manly and colleagues (2002) found that differences between African American and White older adults on standard cognitive ability measures were reduced when controlling for differences on a general literacy measure, although this latter measure may more directly reflect educational attainment than quality of education. It is also possible that the impact of race on health literacy could have been mediated by unmeasured variables in our study. For example, differences in community resources such as level of civic engagement (social capital) may have played a role.
Finally, men had lower STOFHLA scores than did women in our study, whereas other studies have tended to find no gender differences in STOFHLA scores (Gazmararian et al., 1999) or have found lower scores for women (e.g., Schillinger et al., 2002). These different patterns likely reflect differences across samples related to socioeconomic status, health, or other variables.
Study Limitations
In the present study, we measured health literacy with the STOFHLA, but health literacy is a multifaceted concept that encompasses abilities in addition to comprehension skill—abilities such as decision making and others needed to navigate the health care system (e.g., Nielsen-Bohlman et al., 2004). Therefore, any single measure may underestimate poor health literacy. For example, we likely underestimated effects of cultural differences. This may help explain why education was associated with health literacy even after we controlled for other factors, because cultural factors were likely correlated with education (also see Beers et al., 2003). A related limitation is that we did not measure constructs related to social and cultural context in our general model (Figure 1) in the present study.
Nonetheless, the association of the STOHFLA measure with general cognitive abilities has several implications for the importance of cognitive function in daily health care. First, the comprehension abilities tapped by the STOFHLA are a core set of skills related to the broader concept of health literacy, as shown by evidence that performance on the STOFHLA is related to health outcomes (for a review, see Nielsen-Bohlman et al., 2004). Second, cognitive abilities are likely to influence more abilities and behaviors than just comprehension. For example, working memory measures explain age-related variance in medication adherence (Park et al., 1999) and medical decision making (Zwahr, Park, & Shifren, 1999).
A second limitation of this study relates to our sample of chronically ill adults. Because chronic illness such as CHF compromises cognitive function (e.g., Elias, D'Agostino, Elias, & Wolf, 1995; Izquiero-Porrea & Waldstein, 2002), it is difficult to generalize our findings about the relationships between cognitive ability and health literacy. An important next step will be to investigate cognitive ability and health literacy in samples of healthy elders.
A final caution relates to the correlational design of our study. We identified associations between health literacy and cognitive measures but did not establish causal relationships. Although the model in Figure 1 suggests that cognitive deficits contribute to poor health literacy, it is also possible that poor literacy skills contribute to cognitive declines, perhaps by reducing health. Prospective longitudinal studies that also include measures of the sociocultural context may disentangle relationships between these important skill domains and determine how they change over time.
Implications for Improving Health Literacy
The importance of cognitive abilities such as mental processing speed for health literacy suggests the value of training in basic cognitive skills for improving literacy. Such training improves older adults' performance on a variety of tasks (for a review, see Kramer & Willis, 2002). The relationship between cognitive abilities and literacy (Barnes et al., 2004; Manly et al., 2002) suggests this would also be the case for literacy. However, it is unclear to what extent training in basic cognitive abilities "scales up" to performance of complex everyday tasks (Kramer & Willis, 2002). Therefore, basic cognitive training may not be the most effective strategy for improving the ability of older adults with inadequate literacy skills to navigate the health care system.
The relationship between cognitive abilities and health literacy suggests another approach to mitigating the impact of low health literacy on health care. Simplifying tasks that tax health literacy skills would reduce constraints on performance imposed by declining cognitive abilities and limited comprehension skills. Researchers frequently suggest that low-literacy participants will benefit from communication with simple language, pictorials, and other characteristics that simplify comprehension (Ad Hoc Committee, 1999; Doak, Doak, & Root, 1996). However, the relationship between general cognitive abilities and health literacy identified in the present study suggests that such strategies will only help low-literacy patients if they reduce demands on general as well as health-specific cognitive abilities. For example, a sequence of pictorials may not help low-literacy patients if these pictures must be mentally integrated in order to be understood. Effective health communication should reduce demands on cognitive abilities (e.g., working memory) as well as on language-specific abilities (e.g., word knowledge). We are developing multimedia (text with pictures) instructions for CHF medications that may improve older adults' comprehension by addressing both types of abilities (e.g., Morrow et al., 2005).
We plan to elaborate the model in Figure 1 to encompass relationships between health literacy, adherence, and health outcomes. A preliminary analysis suggests that lower health literacy is associated with lower adherence to CHF medications in our intervention study (Murray, Wu, et al., 2004). We plan to investigate whether relationships between health literacy and adherence are mediated by differences in cognitive abilities and whether these relationships also mediate the impact of demographic variables on adherence and health outcomes. Such findings would be consistent with functional definitions that view health literacy in terms of how individuals' skills match the demands of specific health care tasks or contexts (DeWalt et al., 2004; Nielsen-Bohlman et al., 2004).
|
Footnotes |
|---|
1 Beckman Institute of Advanced Science and Technology and the Institute of Aviation, University of Illinois at Urbana-Champaign.
2 Indiana University School of Medicine, Indianapolis; Regenstrief Institute, Inc., Indianapolis, IN; and Indiana University Center for Aging Research, Indianapolis.
3 University School of Medicine, Indianapolis, IN.
4 Department of Psychological Sciences, University of Missouri–Columbia.
5 Pharmaceutical Policy and Evaluative Sciences Division, Center for Excellence in Pharmaceutical Outcomes Research, University of North Carolina at Chapel Hill.
Decision Editor: Linda S. Noelker, PhD
Received for publication November 28, 2005. Accepted for publication May 9, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. O. Clark, R. M. Frankel, D. L. Morgan, G. Ricketts, M. J. Bair, K. A. Nyland, and C. M. Callahan The Meaning and Significance of Self-Management Among Socioeconomically Vulnerable Older Adults J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2008; 63(5): S312 - S319. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
