Health literacy has been the subject of multiple reports, and the U.S. Department of Health and Human Services, Institute of Medicine, and World Health Organization promote and exhort improving health literacy as a public health goal.1,51,52
Proposals have even been made recently to recognize low health literacy as a risk factor warranting clinical screening.53–55
Despite this, controversy remains as to the definition of health literacy - whether it is an individual risk factor or asset, a reflection on healthcare providers’ skills and health systems’ accessibility, or all of these.56
Clearly, the term has sparked unprecedented interest around simplifying healthcare and helping individuals manage their health.
The intent of the LitCog study was to revisit the measures previously used, almost exclusively, in health literacy research and better understand the latent psychological traits being evaluated. Our findings strongly suggest that the problem of limited health literacy mostly reflects individual differences across a broad set of cognitive skills that include but are not limited to reading and numeracy. Associations between health literacy and performance on common health tasks were substantially explained by 1) fluid abilities necessary to actively learn and apply new information, and 2) crystallized abilities such as background knowledge.
The TOFHLA and NVS were more strongly correlated with one another and aligned with fluid abilities. REALM associations with the various health tasks were explained more by crystallized abilities. In multivariable models, assessments of fluid and crystallized abilities together with a health literacy test best explained performance on everyday health tasks. There is likely not a limited set of skills that can be isolated as most important in managing one’s health. The roles individuals assume in healthcare require reading and numeracy, but also health-related knowledge, speed and efficiency of thought, critical thinking, multi-tasking, and memory among other abilities. It is therefore not surprising cognitive traits explained such a large degree of the relationship between health literacy and task performance, or that health literacy measures also provided an independent contribution.
Our findings are limited in that our sample was English-speaking only and predominantly female. We also included more assessments of fluid abilities than crystallized tests. Since fluid and crystallized abilities were comparable in explaining health literacy associations, our findings might under-estimate the importance of background knowledge. In addition, performance on everyday health tasks was measured using hypothetical scenarios. Participants might have applied greater effort to tasks if they had been more salient to their current personal health. However, when comparing scores between those with and without experience with the task or condition in each scenario, differences were not found. LitCog participants are now being followed as a cohort, allowing for opportunities to prospectively study relationships between health literacy, cognitive abilities, and outcomes including risk of hospitalization and mortality.
A general critique of our findings might be that the assessment of task performance is similar to health literacy measures. However, we required individuals to demonstrate functional skills across a wide array of health scenarios beyond solely reviewing print materials - the basis of existing health literacy tests. This criticism can also be directed at many seminal health literacy studies that have examined associations with the ability to perform common self-management tasks.4,57–59
In fact, assessments of health literacy closely resemble cognitive tests, supporting the primary assertion of the LitCog study. The most notable similarity can be seen between the REALM and the American-National Adult Reading Test (AM-NART); both require individuals to correctly pronounce lists of words (r
0.73). The strength of correlations among health literacy, cognitive tests, and performance on health tasks should be understandable and expected.
These crude literacy assessments have proven to be useful research tools, and it is possible they may eventually demonstrate equivalent clinical utility. All are highly predictive of an individual’s ability to perform routine healthcare tasks; the choice to use one versus another should depend more on test attributes (i.e. availability in Spanish, time to administer) and less so out of concerns for misclassification. In this case, the NVS might be a logical choice as it is as brief as the REALM, but available in Spanish where the REALM is not.
Brief measures that assess global cognitive function, such as the mini-mental status exam (MMSE), might also serve as proxies for health literacy. Strong associations between health literacy measures, such as the TOFHLA, and the MMSE have previously been established.25,26,60
The advantages to this approach are that many of these cognitive screeners are already administered in clinical settings. In addition, their face validity makes it less likely a researcher or clinician would fall victim to superficial interpretations of the problem and its solution. Yet these tools would likely need revised scoring thresholds, as individuals could be free of a clinically defined cognitive impairment yet still have limited health literacy.
The proposition that the most common measures of health literacy are actually crude assessments of general cognitive abilities should not distract attention from broader efforts to redesign health materials, improve clinician communication skills, enhance the navigability of health systems, or engage communities to assume public health roles.27
Our findings affirm the need to help patients build appropriate background knowledge and skills, but to also reduce the cognitive demands of health systems through unnecessarily complex health tasks. As a start, health literacy interventions should move beyond plain language approaches and deconstruct the tasks required of patients within a particular healthcare context. Depending on the task, steps could be taken that follow cognitive and human factors principles to improve performance. This might include giving individuals more time to accurately process information, limiting and layering new information to reduce cognitive burden, use of increasingly available technologies or ‘external aids’ (i.e. pill box organizers) to enhance recall and prompt health behaviors, or eliminate tasks all together if the health system could assume responsibility instead.
Future evaluations of interventions should always collect sufficient data to determine if a strategy mitigates the impact of ‘low health literacy’ on outcomes.61
A modified perspective of health literacy that includes an expansive view of cognitive skills necessary to manage health could also inform the development of more precise clinical assessments to identify those at risk.12
Despite calls for clinicians to follow universal precautions and assume all patients may have health literacy concerns, remediating inadequate cognitive skills for self-care might require clinical screening. This would then allow a greater allocation of resources, in terms of education and follow-up, to those struggling to learn and apply health information and instructions.