Background
Health literacy has been defined as the degree to which people are able to access, understand, appraise, and communicate information to engage with the demands of different health contexts in order to promote and maintain health across the life-course [
1]. Health literacy is related to general literacy, but it more specifically encompasses a person’s understanding of health information, both in spoken and in written form [
2]. Also, in contrast with general literacy, health literacy is considered a more dynamic and context-dependent ability [
3]. Multiple studies have shown that older adults are an especially vulnerable group with regard to health literacy, with rates of low health literacy ranging from 30 to 68 % [
4‐
7], with some studies already finding lower health literacy in adults above the age of 50 [
4,
5].
Low health literacy is strongly associated with undesirable health outcomes, such as poor physical fitness [
8], higher rates of arthritis and hypertension [
9], and higher mortality [
10]. It has been suggested that difficulty in adhering to medical advice may partly explain why low health literacy leads to poor health outcomes [
11‐
13]. Adherence can be defined as the extent to which a person’s behavior such as following a diet, taking medication, and/or executing lifestyle changes, are in agreement with recommendations from a health professional [
14]. This includes any behavior to prevent, cure, or care for health problems. It also includes many behaviors that are commonly considered to be part of self-management [
15]. Rates of poor adherence can be as high as 47 % [
16,
17]. Poor adherence has been shown to be associated with several factors, such as poor cognitive abilities [
18,
19], a higher number of prescribed medications [
18,
19], and the presence of depressive symptoms [
20]. Adherence may also be an important factor through which health literacy impacts health outcomes.
Previous reviews have assessed whether an association between health literacy and adherence exists, with mixed results [
12,
21,
22]. For example, the reviews of Jin et al. [
22] and Witte [
12] suggest that high levels of health literacy contribute to successful adherence to therapy. However, Loke et al. [
21] found no association between health literacy and adherence in hospitalized patients with diabetes or cardiovascular disease. Also, reviews do not often focus on older adults, which makes it hard to draw conclusions about this specific population. If low health literacy plays an important role in poor adherence among older adults, interventions to improve adherence may be effective to improve the relatively poor health outcomes of older adults with low levels of health literacy. It is, however, unclear whether adherence interventions are effective among older adults with low levels of health literacy.
We conducted a meta-review as a means to adequately assess and summarize a large number of existing reviews and meta-analyses. The meta-review methodology, also called ‘review of reviews’ or ‘overview of reviews’, is a review that only includes systematic reviews and meta-analyses. This methodology is also used by the Cochrane collaboration [
23]. It is considered a suitable methodology to summarize existing evidence on topics on which multiple reviews have already been published [
23,
24]. Systematic reviews and meta-analyses are considered to be the highest level of evidence. Policy makers and healthcare professionals should make decisions based on systematic reviews, but the vast increase in number of systematic reviews may cause people to become overwhelmed. This is particularly true for topics which are clinically important (e.g. adherence) and a need exists to summarize all findings. Meta-reviews have the potential to identify consistent patterns of results on a large level by taking into account an even larger body of evidence than regular systematic reviews. In a meta-review, differences in the objectives and the quality of the systematic reviews can also be explored. In fact, Smith et al. [
25] point out that “A logical and appropriate next step is to conduct a systematic review of reviews of the topic under consideration, allowing the findings of separate reviews to be compared and contrasted, thereby providing clinical decision makers with the evidence they need.” The meta-review methodology has been increasingly used over recent years [
26‐
31].
A broad definition of adherence was adopted for this meta-review, including any behavior that was recommended by health professionals. In modern healthcare, adherence goes beyond medication adherence, and health professionals often advise their patients to perform various other health behaviors, including, for example, doing regular blood glucose checks, increasing physical activity, or decreasing salt intake. Earlier research has also shown that health behaviors are often associated and have shared determinants [
32,
33]. This makes it plausible that interventions that aim to improve any kind of adherence may impact on various health behaviors.
In this study, our aims are firstly to evaluate the association between health literacy and adherence in older adults above the age of 50 by performing a meta-review of existing systematic reviews. Secondly, we assess whether interventions to improve adherence are effective among this population.
Methods
Search strategy
Systematic searches were conducted for systematic reviews. The searches were conducted in eight electronic databases: MEDLINE, Education Resources Information Center (ERIC), EMBASE, PsycINFO, Cumulative Index to Nursing and Allied Health Literature (CINAHL), DARE, The Cochrane Library, and Web of Knowledge. All databases were searched through September 2014. Combinations of keywords were used, including health literacy, numeracy, adherence, compliance, and self-management. When possible, built-in filters for reviews were used. The full search strategy is presented in Additional file
1. In addition, reference lists of included systematic reviews were manually searched for further reviews that could add to our meta-review.
Selection of reviews
After the completion of the search and removal of duplicates, two independent reviewers (BG and YKL or JB) screened the titles and abstracts of all articles for potential eligibility for inclusion in our meta-review. Any article selected by at least one of the reviewers was included for full-text review. In the title/abstract review, the inter-rater agreement was around 95 %. Two independent reviewers (BG and JB or YKL) then read the selected articles in full. The reviewers were not blinded to authorship of the reviews. Disagreements in the full-text review were resolved by discussion (BG and JB or YKL).
Reviews were included if they provided information on at least one of our objectives, based on the following criteria:
(1)
The article was a systematic review (we defined this as a literature review involving a systematic search with application of selection criteria and a description of the number and nature of included studies), either with or without a meta-analysis (i.e. statistical pooling of the results).
(2)
The review either assessed the association between health literacy and adherence or evaluated the effectiveness of interventions to improve adherence in adults with low health literacy.
(3)
The review focused on behaviors that need to be maintained for an extended period of time. Reviews that focused on behaviors that are only performed once, such as diagnostic tests and participation in screening, were excluded.
(4)
At least part of the results of the studies included in the review were specific for the objectives of our meta-review. To confirm this, we verified that the included primary studies considered at least a subset of older adults (mean or median age of at least 50 years) and assessed health literacy with a validated measure, such as the S-TOFHLA [
34] or REALM [
35]. As an additional criterion, we checked whether the studies were performed in westernized developed countries (USA, Canada, Europe, New Zealand, or Australia).
There were no restrictions regarding type of publication (e.g. report, journal article) or the type of primary studies that were evaluated in the systematic reviews. Also, no restrictions were imposed regarding language. Reviews in non-English languages were translated using online translation services. Native speakers could be contacted in case a non-English review was selected for data-extraction. However, this did not occur.
Data extraction and quality assessment
Data were extracted from the reviews that met all criteria, using a coding form that captured bibliographic information, the main research question, methodological data, characteristics of the studied population, data about the content and procedures of the included studies, the results, and conclusions as reported by the authors.
The AMSTAR tool was used to check the quality of the included reviews [
36]. AMSTAR consists of 11 questions and assesses, among other things, whether a comprehensive literature search is performed, whether duplicate study selection and data-extraction were performed, and whether a full list of included studies is provided. Data-extraction and quality assessment were independently performed by at least two reviewers (BG and JSB or YKL) with disagreements resolved by discussion or by consulting the third reviewer.
Analyses and reporting
As meta-reviews report on the level of systematic reviews, detailed reporting or pooling of statistics is only possible when at least some of the included systematic reviews conducted a meta-analysis. As we only identified one systematic review with a meta-analysis, a narrative synthesis was used to report our results. First, we summarize the quality of the included reviews. Then, we report on the conclusions of the reviews regarding the association between health literacy and adherence in older adults. Finally, we discuss conclusions of the reviews regarding the effectiveness of adherence interventions in the population of older adults with low levels of health literacy.
In accordance with the meta-review methodology, only information from the systematic reviews is reported in this study. However, to ensure the validity of the results of our meta-review, we also performed data verification by checking whether the reported general conclusions of the reviews were supported by the results of the primary studies that were specifically relevant for our meta-review. A complete overview of these primary studies is presented in Additional file
2.
Discussion
Our meta-review provides only weak evidence in support of an association between health literacy and adherence in older adults. Our results show some evidence that interventions on adherence are effective among older adults with low health literacy, but this evidence is limited.
The results of our systematic meta-review cast doubt on the existence of a strong association between health literacy and adherence among older adults, as the identified systematic reviews only support the existence of a weak association. While non-systematic reviews and health literacy frameworks have suggested that adherence in an important factor through which health literacy impacts health outcomes [
11‐
13], our results do not strongly support this notion.
However, studies on health literacy and adherence in older adults may also have missed a genuine association. First, one of the reviews suggests the possibility of a nonlinear association between health literacy and adherence, in which adherence rates are lowest among those with moderate health literacy [
45]. This idea is supported by a survey study that shows that people with low health literacy mostly fail to adhere as a result of a lack of understanding of the given instructions, while people with high health literacy more often non-adhere as a result of deliberately choosing to disregard recommendations [
62]. If the association between health literacy and adherence is nonlinear, studies that treat health literacy as a categorical variable with only two or three categories may fail to observe any such associations.
Second, the way in which health literacy and adherence are measured may limit the possibility to draw strong conclusions on the association between the two concepts. The two most commonly used measurement tools for health literacy are the Rapid Estimate of Adult Literacy in Medicine (REALM) and the Short Test of Functional Health Literacy in Adults (S-TOFHLA). Both tools have been criticized for not adequately covering the range of competencies required for adequate health literacy [
3,
21]. It has also been suggested that both tools measure limited different elements of health literacy [
63]. Also, many tools to measure adherence are based on self-report, but it has been shown that self-reporting is not always an accurate measure of adherence due to people overestimating their adherence, especially when their actual adherence is poor [
64].
Third, the inconsistent results on the association between health literacy and adherence may be the result of a confounding effect of age in some studies. Whereas older adults tend to have poorer health literacy [
4‐
7], other research has shown that older age is positively related to adherence [
65,
66], which may complicate any association between health literacy and adherence in older adults.
The results of our meta-review on interventions suggest that interventions on adherence are at least as effective for people with low health literacy as for those with high health literacy. The interventions on adherence described in our reviews seem to focus mainly on education and on lowering the health literacy demands of adherence instructions. However, as the reviews provided only limited information on the effectiveness of adherence interventions among older adults with low health literacy, we were not able to draw conclusions regarding which type of intervention could be most beneficial for this population. We identified a clear gap in the available literature, as none of the included systematic reviews drew specific conclusions on the topic of adherence interventions among older adults with low health literacy. Additionally, the input of four of the reviews was based on the results of the same study [
53]. In total, the reviews provided information on only seven unique intervention studies.
The reviews included in our meta-review were of varying quality. However, similar conclusions were found among both reviews of higher and lower quality. Our quality assessment indicates that the reviews were mostly based on thorough searches, that the selection procedures and data-extraction were mostly well conducted, and that most reviews gave sufficient information about the included primary studies. However, many of the included reviews did not conduct a quality assessments of the included studies. This is problematic and certainly requires improvement, as it makes it impossible to assess whether conclusions are based on high quality evidence.
Strengths and limitations
The strengths of our meta-review included the use of a broad definition of adherence, which also includes behaviors outside the cure and care setting, and our extensive search strategy.
However, our meta-review had some limitations. First, we cannot rule out the possibility of selective analysis and outcome reporting in both the primary studies and the reviews. Some research may not have been published if deemed insufficiently novel, positive or significant. Second, as none of our included reviews reported specifically on the effectiveness of adherence interventions among older adults with low health literacy, we could only report on their limited conclusions about individual studies. Third, in many reviews, only part of the studies on the association between health literacy and adherence focused on older adults, leaving unclear to what degree the conclusions of the review are generalizable to this group. However, data verification confirmed that the patterns of results found in the reviews did not change substantially when only considering the primary studies that met the criteria for our meta-review. Finally, in a systematic meta-review, a review based on including systematic reviews, the most recent primary studies may not be covered.
Implications for public health and future research
Our results suggest that health literacy and adherence exert partially independent effects on the health outcomes of older adults. Public health practitioners should be aware that initiatives that aim to mitigate the negative impacts of low health literacy on health outcomes among older adults should not focus solely on adherence. Initiatives that aim to improve adherence rates among older adults could focus on education and on lowering the health literacy demands of adherence instructions, as evidence on these strategies is the strongest.
Although we adopted a broad definition of adherence in this meta-review, most of the included reviews focused specifically on medication adherence or disease management. None of the reviews focused on adherence with guidelines for general health behavior, such as healthy nutritional behavior and physical activity. Future reviews on the impact of health literacy on the health outcomes of older adults could consider these behaviors as well. Some studies have focused on the association between health literacy and general health behaviors among older adults [
67,
68].
Also, despite our extensive search strategy, we found no reviews that focused on improving adherence specifically among older adults with low health literacy. To close this gap in the available literature, future research could focus on reviewing primary studies on specifically this topic, as this could further advance our understanding of the role of health literacy in adherence interventions among older adults. Many of the reviews included in our meta-review stressed the need for more high quality intervention research among people with low health literacy. Further intervention research could indeed help identify which types of interventions are most beneficial for older adults with low health literacy, which could be valuable for clinical practitioners.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BG developed the protocol for the study, conducted the literature searches, selected studies for inclusion in the review, conducted the quality assessment, interpreted the data and wrote the manuscript. JSB and YKL assisted in designing the search strategy, selecting studies for inclusion in the review, conducting the quality assessment, and interpreting the data. CJMJ and CS critically reviewed the manuscript for content. SAR was involved in drafting the manuscript and had an advisory role throughout the study. AFW was involved in the development of the protocol, designing the search strategy, interpretation of the data, and had an advisory role. All authors contributed substantially to the writing of the manuscript and have read and approved the final manuscript.