Associations of health literacy with self-management behaviours and health outcomes in chronic kidney disease: a systematic review

Introduction Low health literacy is widely reported in people with chronic kidney disease (CKD) and has been associated with reduced disease self-management, poor health outcomes, increased mortality and poorer quality of life. However, these associations are still not well understood. Methods Electronic-based systematic searches were performed to identify studies examining associations between health literacy and self-management behaviours and/or health outcomes in patients with CKD. A tabular and narrative synthesis of the data was performed. Meta-analysis was not appropriate due to the heterogeneity of study designs and methods. Results Searches identified 48 studies that met the inclusion criteria. A total of 41 published articles, six conference abstracts, and one thesis were included. Of the 48 studies, 11 were cohort and 37 were cross-sectional. In total there were 25,671 patients; 16,952 from cohort studies. Median study sample size was 159 (IQR 92–275). Study quality was high (5), moderate (24) and poor (19). Thirteen measures of health literacy were used. Despite the limitations of the available evidence, there appear to be consistent relationships between higher health literacy and favourable self-management behaviours for patients with CKD. Definitive relationships between health literacy and patient outcomes are far less clear and remain incompletely understood. Discussion Conclusive evidence describing a causal link between health literacy and patient outcomes remains limited, but for many outcomes, a consistent association is described. In addition to associations with mortality, hospitalisation and clinical events, there were consistent associations between health literacy and favourable self-management behaviours which could support the development of patient education aimed at improving health literacy. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s40620-022-01537-0.


Introduction
The term health literacy refers to the cognitive and social skills, which establish the motivation and ability of an individual to gain access to, and to understand and use information in ways that promote good health, prevention of disease and improve quality of life (QoL) [1]. Early conceptualisations of health literacy focused on functional aspects, such as basic skills in reading and writing. In recent years, definitions and measures have evolved to include a more multidimensional approach encompassing behavioural and cognitive skills. Communicative health literacy represents the more advanced cognitive skills that allow information to be extracted and meanings derived [2]. Critical health literacy is the analysis of information and its application to achieve greater control over life events [3]. It is established that the health literacy of patients with chronic kidney disease (CKD) is often low and may limit the extent to which patients can effectively self-manage aspects of their health as well as directly or indirectly affect health outcomes [4]. The factors which contribute to health literacy are complex and interrelated, but link to education, income and socioeconomic background [5].
Self-management behaviours refer to the ability to learn and practice skills and behaviours which allow people to better manage their day-to-day health, and include behaviours related to medication adherence, physical activity, managing symptoms, information seeking and interaction with healthcare providers [1]. Health literacy and self-management behaviours are inherently linked, but account for and explain different aspects of patient experience. Gaining a greater understanding of the relationships between health literacy and self-management behaviours in patients with chronic illnesses will improve the way complex interventions are designed to account for the relative influence of these different factors on patient outcomes. This is especially true of patients with CKD, for whom the personal burdens of care are often high [6] and which leave many feeling overburdened and unable to be actively involved in decisions about their care or to appropriately engage with their treatment [4].
A previous systematic review evaluated the relationship between health literacy, mortality and patient outcomes in individuals with CKD [4]. Whilst there was no robust causal evidence linking health literacy and patient outcomes, associations between low health literacy and adverse health outcomes were evident.
This systematic review aimed to update and expand the previous review by examining the relationships between health literacy and self-management behaviours and health outcomes. Better defining these relationships will allow better planning of prospective studies and the design of complex interventions that seek to improve health literacy and self-management behaviours to improve patient experience and outcomes for people with CKD.

Protocol registration and eligibility criteria
The PRISMA flow diagram is shown in Fig. 1. This review was prospectively registered (PROSPERO 2020 reference: CRD42020201602).
Studies were considered eligible if they: (i) Were retrospective, prospective, cross-sectional, longitudinal, case-control, cohort studies or randomised controlled trials (RCTs). Conference abstracts were included where sufficient data was present and data were not published elsewhere. Review articles, drug intervention studies, trial protocols, qualitative studies, and case reports were excluded, along with studies that were not in the English language. (ii) Included adults, children, and adolescents with CKD stages three to five (estimated glomerular filtration rate (eGFR) less than 60 mL/min/1.73 m 2 ) including those requiring renal replacement therapy (haemodialysis, peritoneal dialysis or kidney transplantation). Studies containing participants with CKD stages 1 and 2 exclusively were excluded. (iii) Used an accepted tool for the assessment of health literacy (functional, communicative and/or critical health literacy). (iv) Reported associations between health literacy and self-management behaviours and/or health outcomes as part of study outcomes.

Search strategy
Electronic-based searches from the date of inception to December 2021 were performed to identify studies examining associations between health literacy and self-management behaviours and/or health outcomes in patients with CKD.

Data extraction, synthesis and quality appraisal
Search results were managed using Endnote (Clarivate Analytics, Philadelphia, PA, USA) referencing software. Full texts were obtained and reviewed for articles meeting the inclusion criteria. Two reviewers independently reviewed two articles using the data extraction tool (an adapted version of the Cochrane Data Extraction Template) to ensure suitability. From therein, reviewers extracted data from each article including study design, sample-size, recruitment method, inclusion/exclusion criteria, primary aims, reported associations between health literacy and outcomes, and statistical testing. Crosschecking was performed by a second reviewer (R.E.B). Titles and abstracts were screened by two reviewers (A.T and S.F.A) based on the criteria below. Agreement between the two reviewers was 86.28% (k = 0.354; 95% CI 0.292-0.416). Discrepancies were resolved with the inclusion of a third reviewer (R.E.B).
Study quality was independently assessed by three reviewers (D.S.M, S.F.A, and R.E.B) using a modified Newcastle Ottawa Scale, and final grading was decided following discussion. Quality scoring is presented within the supplementary material (appendix S2). Statistical significance was accepted as p < 0.05 unless individual studies stated otherwise.
Data are presented in tabular form including study characteristics, sample sizes, demographics of participants, health literacy measurement tools, outcome measures, associations tested between health literacy and self-management behaviours and health outcomes in univariate and multivariate analyses, covariates used, and any significant associations that were found. A narrative synthesis of findings, detailing the association between health literacy and self-management behaviours and/or health outcomes was also performed. Meta-analysis was not appropriate due to the heterogeneity of study designs and methods.

Results
Searches identified 48 studies that met the inclusion criteria ( Fig. 1 Table 1 and an overview of the associations between health literacy and outcomes is provided in Table 2. Study quality was graded low for 19 studies, moderate for 24 studies, and high for 5 studies.  Jain et al. [33] Patzer et al. [13] UM U Schrauben et al. [10] Balhara et al. [34] Kazley et al. [43] U Kazley et al. [46] M M

Self-management behaviours
Three cohort and 24 cross-sectional studies including a total of 5367 patients, explored elements of self-management. Study qualities were: 2 high, 15 moderate, and 10 low. Several studies reported on self-management behaviours as a whole, and others reported individual behaviours.
In 410 patients with CKD stages 1-5, health literacy was independently associated with self-management behaviours [7]. However, social support was the strongest determinant of behaviour. Total health literacy score was associated with diabetes self-management in 63 patients on haemodialysis [8]. In regression analysis, critical and communicative health literacy were associated with diabetes self-management and functional health literacy was not. Health literacy was not associated with self-management knowledge [9] or self-care behaviour scores [10] in 150 and 401 patients with nondialysis CKD, respectively. After adjustment for demographics, CKD knowledge, CKD awareness, and education, lower health literacy was associated with lower kidney disease knowledge in 401 patients with non-dialysis CKD [11]. Health literacy had a moderate positive direct effect on self-management behaviours in 245 patients with CKD stages 2-3 [12].

Medical management
In a cohort study of 99 KTRs, limited health literacy was independently associated with higher odds of medication non-adherence [13]. Health literacy was not associated with medication adherence in 208 patients on haemodialysis, 252 KTRs, and 137 patients with non-dialysis CKD [14][15][16], but was associated with medication management capacity in cross-sectional analyses [16]. In three studies utilising the HLQ, health literacy was associated with medication adherence in 96 KTRs and 187 patients with non-dialysis CKD [17,18] and with treatment adherence in 30 patients with ESKD [19]. In two of these studies, the domain 'actively managing health' was positively associated with medication adherence [17,18].

Lifestyle
Diet One study of 452 patients on haemodialysis reported that lower health literacy was associated with lower adherence to dietary and fluid intake recommendations [20]. Higher health literacy was associated with increased adherence to diet, fluid, and medication instruction in 42 patients on haemodialysis [21]. Health literacy, after adjustment for socioeconomic factors, was the only determinant of dietary adherence in 84 patients on haemodialysis and correlated with self-management skills [22]. In 155 patients with nondialysis CKD, health literacy did not associate with dietary sodium knowledge [23]. Lower health literacy associated with lower intake of sugary drinks and lower fast food consumption in 137 patients with non-dialysis CKD after adjusting for demographics, education, and diabetes status [14].
Exercise No association between health literacy and physical activity was shown in 137 patients with CKD stages 3-5 [14]. However, there was an association between higher health literacy and greater exercise habits in 200 patients with CKD stages 3-5 and an association between lower health literacy and lower total metabolic equivalent for physical activity (unadjusted analysis only in 2340 patients with CKD stages 2-4) [24].

Functional Communicative
Critical The scores of each item are summed to determine the total score Health literacy was categorised into the following four levels based on total score: "inadequate", 0-25 points; "problematic" 25-33 points; "sufficient", 33-42 points; and "excellent", 42 points or more. "Inadequate" and "problematic" with 33 points or less were defined as "limited" [63]

Mortality and hospitalisation
Seven cohort studies with a median sample size of 480 (IQR 260-2274) and follow-up times between 12 and 35 months examined the association between health literacy and mortality. One study was high quality and six studies were moderate quality. Six studies were of patients on haemodialysis and one study was of kidney transplant candidates. Three studies showed no association between health literacy and mortality in 2820 patients on haemodialysis [26][27][28]. The HLQ domain 'actively managing my health' was independently associated with mortality after adjusting for sociodemographic and clinical factors in 221 patients on haemodialysis [29]. One study reported an independent association between limited health literacy and increased mortality in patients on dialysis (HR 1.54; 95% CI 1.01-2.36), after adjustment for age, sex, race, and diabetes but not socio-economic status [30]. Similarly in 11,476 patients on dialysis, low health literacy was associated with increased mortality (HR = 1.65; 95% CI 1.28-2.12 Low vs. High) [31]. In 1578 kidney transplant candidates, there was an increased risk of waitlist mortality in those with limited health literacy [32]. This was robust to adjustment for income, comorbidity, and insurance type in a sensitivity analysis. Eight studies (six cohort and two cross-sectional, 1216 patients) explored the association between health literacy and hospitalisation (median sample size 122, IQR 80-250). Study quality was: three low and five moderate. Lower hospitalisation rates were associated with the HLQ domain 'actively managing my health' in 221 patients on haemodialysis after adjustment for age, comorbidity, and education [29]. Lower health literacy was associated with ESKD-related hospitalisation but not total hospitalisation in an adjusted analysis of 286 patients on haemodialysis [26]. In a cross-sectional analysis of 102 patients with ESKD, higher scores on the HLQ domain 'understanding health info well enough to know what to do' were associated with lower emergency department admissions. One study showed no association between health literacy and hospitalisation in a cross-sectional analysis of 32 patients on peritoneal dialysis [33]. Another cross-sectional study showed no association between health literacy and missed dialysis sessions resulting in hospitalisation in 49 patients on haemodialysis [34]. Three cohort studies showed no association between health literacy and hospitalisation in patients with CKD stages 2-5 [35] (142 patients), 3-4 [36] (74 patients), and KTR [13] (99 patients).

Kidney function, comorbidities and clinical parameters
Six studies (one cohort and five cross-sectional) totalling 3096 patients (median 175, IQR 112-741) explored the  [43][44][45][46][47][48][49][50] association between health literacy and kidney function. One study was low quality and five studies were moderate quality. In a cohort study of 74 patients with CKD stages 3-4, patients with inadequate or marginal health literacy had lower eGFR than those with adequate health literacy [36]. Limited health literacy was associated with lower eGFR in a cross-sectional analysis of 2340 patients with CKD stages 2-4 [24] and in 150 patients with CKD stages 1-4 [9]. This did not remain significant in the latter study after adjustment for age. In 124 KTRs, cross-sectional analysis did not describe any significant relationships between health literacy and eGFR [37]. However, health literacy was negatively correlated to serum creatinine after adjustment for time after transplant.
Cross-sectional analysis of 2340 patients with CKD stages 2-4 showed those with limited health literacy were more likely to report a history of cardiovascular disease after adjusting for age, gender, race, BMI, and education [24]. In unadjusted analysis, limited health literacy was associated with a reduced likelihood of having blood pressure below 130/80 mmHg and an increased prevalence of diabetes [24]. Adequate health literacy was associated with lower blood pressure and lower mean arterial pressure but not blood lipids after adjusting for both demographic and socioeconomic variables in 72 patients on dialysis [25]. In a crosssectional analysis of 225 patients on dialysis, limited health literacy was associated with poor blood pressure control in women with diabetes but not in any other subgroup [38]. No health literacy domains were associated with the presence of comorbidities (< 3 or ≥ 3) in 102 patients with CKD stage 5 [39] or with the presence of diabetes in 223 patients with CKD [40].

Quality of life and cognition
Ten cross-sectional studies explored QOL and/or outcomes relating to cognition with a total of 1643 patients (median 173, IQR 96-236). Four studies were moderate quality and six were low quality. In 187 patients with CKD stages 3-5, significantly better QOL was found in patients with higher health literacy [18]. However, in 260 patients on haemodialysis, QOL was not significantly different between those with adequate and those with limited health literacy [41]. This study did suggest depression may be worse in patients with limited health literacy, but this did not reach statistical significance. In 100 patients on dialysis, cross-sectional analysis showed that patients with higher health literacy reported fewer depressive symptoms [42]. Similarly, in 223 patients on dialysis and patients with non-dialysis CKD, depression was associated with worse health literacy in 8/9 domains on the HLQ [40]. Cognitive impairment was not associated with health literacy in this study. In unadjusted analyses, lower health literacy was associated with lower decision-making capacity in patients with advanced CKD [43]. Three cross-sectional studies showed an association between higher health literacy and high self-efficacy in KTR [44], patients on haemodialysis [22], and patients with CKD stages 2-3 [12].

Modality-specific outcomes
Four studies (three cohort and one cross-sectional) evaluated the association between health literacy and aspects of the transplantation process. Study quality was: one high, two moderate, and one low. In a cohort study of 2274 patients receiving dialysis, limited health literacy was associated with a reduced likelihood of receiving a kidney transplant of any type but was not associated with pre-emptive transplant listing [27]. In a different study of 1578 kidney transplant candidates, limited health literacy was independently associated with a decreased likelihood of listing for transplantation [32]. In a cohort study of 62 patients on haemodialysis, lower health literacy was associated with a lower likelihood of being referred for transplant evaluation after controlling for race, gender, comorbidities, age, and support [45]. Higher health literacy was associated with transplant listing and receiving a transplant in a cross-sectional analysis of 92 patients with advanced CKD [46].
Seven studies (three cohort and four cross-sectional) explored the association of health literacy and vascular access. Study quality was: two high, three moderate, and two low. Patients with lower health literacy were more likely to choose a catheter over arteriovenous fistula (AVF) in a cross-sectional analysis of 56 patients on haemodialysis [47]. Similarly, catheter use was more common (versus AVF) in patients with limited health literacy after adjustment for age, gender, race, and dialysis vintage in a crosssectional analysis [48]. In a cross-sectional analysis of 542 patients on haemodialysis, patients with a greater ability to engage with healthcare providers, those with a better ability to navigate the healthcare system, those more able to find good health information, and those who understood it well enough to know what to do were more likely to dialyse via an AVF [49]. Conversely, in a cohort study of 2274 patients on haemodialysis, limited health literacy was not associated with dialysis catheter use [27].

Discussion
This systematic review synthesised data from 48 studies and 25,671 patients. There were more cross-sectional studies (n = 37) than cohort studies (n = 11), but cohort studies tended to be larger accounting for 16,952 of the patients included. Study quality was highly variable with only five studies classified as high quality. This variability, the heterogeneity of study designs and the limitations of statistical analyses mean it is not possible to define causal links between measures of health literacy and outcome measures assessed in this systematic review. The heterogeneity of the study designs and the wide variety of tools used to assess health literacy and outcomes meant it was also not possible to meta-analyse the available data. Fewer than one-quarter of studies assessed the relationships between health literacy and 'hard' outcomes such as mortality or hospitalisation, and there were limited numbers of studies that included KTRs or patients on peritoneal dialysis. No studies included paediatric patients. Whilst 27 studies explored the relationship between health literacy and self-management behaviours, only three of these were cohort studies. In a previous review of the relationships between health literacy and patient outcomes in patients with CKD [4], the majority of studies were only available as conference abstracts (13 articles; 16 conference abstracts). In the present review, there were 41 published articles (plus six abstracts; one thesis).
Within the limitations of the data available, there are consistent associations between health literacy, discrete selfmanagement behaviours, and knowledge of CKD for patients with CKD, patients on haemodialysis and KTRs [7,11,13,[17][18][19][20][21][22]. For KTRs health literacy appears to associate most closely with health behaviours related to medication adherence [13]. This also seems to be the case for patients on haemodialysis [19] and those with CKD [17,18]. These findings have several important implications for studies that seek to improve health-related behaviours such as medication adherence. Health literacy and self-management behaviours are inherently linked but describe different aspects of the patient experience. Even though the direction of causality between health literacy and health behaviours cannot be definitively defined from the available data, intuitively interventions that seek to improve self-management behaviours in patients with CKD will require carefully developed components that improve relevant aspects of health literacy. Defining the components of low health literacy for different populations of patients with kidney disease and between sub-groups of patients must be a priority for the kidney community to be able to support the development and implementation of appropriate self-management interventions. Our incomplete understanding of these issues and the relationships between health literacy and self-management behaviours may explain why there is only weak evidence that health literacy interventions in patients with CKD lead to meaningful improvements in knowledge, decision-making and selfcare behaviours [50]. It is also clear that whilst appropriately designed, stand-alone interventions addressing aspects of health-related behaviours are needed, they do not have to be tested in isolation. Self-management interventions with appropriate components of health literacy could (and should) be developed to be tested alongside lifestyle interventions, pharmacological interventions and even interventions that involve medical devices. The inequities in access to participation in clinical trials for patients with kidney disease are well described at both an individual [51][52][53] and institutional level [54]. Embedding relevant health literacy and self-management interventions into clinical trials may go some way to improving recruitment and retention to trials for patients with low health literacy [55]. Appropriate designs would also allow testing of the effects of contemporaneous health literacy/self-management interventions alongside lifestyle, pharmacological or device-related interventions.
Whilst some studies suggest there is no clear link between health literacy and mortality for patients on haemodialysis [26][27][28], larger, prospective studies suggest that there is a link for these patients [30,31], and for potential kidney transplant candidates [32]. Links between health literacy and mortality in patients with CKD and KTRs have not been well studied. Whilst smaller studies did not show clear links between health literacy measures and hospitalisations [13,[33][34][35][36], larger studies described a clear and consistent relationship for patients on haemodialysis [26,29]. There were no data describing a consistent relationship between health literacy and kidney function, and it is not possible to draw any firm conclusions from data assessing possible relationships between health literacy and medical co-morbidities as the nature of the study designs and analyses do not account for confounding variables. There were no consistent associations described between health literacy and quality of life for patients on dialysis, KTRs or patients with CKD. There were clear and consistent associations between lower health literacy and a lower likelihood of being listed for a kidney transplant [27,32,45,46], although whether this is independent of social or demographic factors is not clear. For patients on haemodialysis, whilst cross-sectional studies appear to describe a relationship between health literacy and likelihood of dialysing with an AVF [47,48,56], this finding was not corroborated by a larger cohort study [27], suggesting there is no clear link between health literacy and vascular access type.
Thirteen different health literacy questionnaires were used, with the REALM and the HLQ being the most common. The HLQ provides scores across nine independent health literacy domains, rather than a summative score, and therefore was not included in the previous systematic review [4]. However, its popularity has grown over recent years with the majority of the literature using this tool being published in 2019/20. The concept of health literacy has developed to become far more multidimensional, including social, critical, and interactive dimensions rather than just individual reading and numeracy skills [1,57].
The strengths of this systematic review lie in the broad search strategy and the inclusion of paediatric studies. That there were no studies evaluating links between health literacy and self-management behaviours in paediatric patients with kidney disease is telling and identifies an immediate area for research. This review offers a significant update to the previous review [4], with data from over 7000 patients. Despite this, the findings of this review are broadly similar in that there remains significant heterogeneity in study design, in selection of outcome measures and in the way health literacy (and self-management behaviours) are assessed, thereby making a direct comparison between studies difficult and meta-analyses impossible. There are inherent limitations in the use of screening tools to assess health literacy, but they have been validated against comprehensive assessments [58]. Whilst there were fewer cohort studies than cross-sectional studies, there were more patients in cohort studies. Investigators should be encouraged to work collaboratively on large, multi-centre prospective studies. Consensus should be reached on the assessment of health literacy and of core outcomes to allow synthesis of studies and future meta-analyses and meta-regressions. Without consensus across these areas, the evidence base will remain piecemeal, inconclusive and limit our ability to inform the design of interventional studies of complex interventions.

Conclusions
Conclusive evidence describing a causal link between health literacy and patient outcomes remains limited, but for many outcomes, a consistent association is described. In addition to associations with mortality, hospitalisation and clinical events, in this systematic review, we were also able to describe consistent associations between health literacy and self-management behaviours. Gaining a better understanding of the nature of these relationships is a priority for the kidney community to support the development of evidencebased interventions to support health literacy.