The role of health literacy in cancer care: A mixed studies systematic review

Background Patients diagnosed with cancer face many challenges and need a good understanding of their diagnosis and proposed treatments to make informed decisions about their care. Health literacy plays an important role in this and low health literacy has been associated with poorer outcomes. The aims of this review are to identify which outcomes relate to health literacy in patients with cancer, and to combine this through a mixed studies approach with the patient experience as described through qualitative studies. Methods Four electronic databases were searched in January 2021 to identify records relating to health literacy and patients with cancer. Records were independently screened then assessed for inclusion by two reviewers according to the following criteria: patients aged ≥18 years with cancer, English language publication AND health literacy measured with validated tool and measured outcome associated with health literacy OR qualitative study exploring the role of health literacy as patients make decisions about health. Quality was independently assessed by two reviewers. A narrative synthesis was performed, and findings integrated through concept mapping. This systematic review was registered with PROSPERO, entry CRD42020166454. Results 4441 records were retrieved. Following de-duplication, 2496 titles and abstracts were screened and full texts of 405 papers were reviewed for eligibility. 66 papers relating to 60 studies met the eligibility criteria. Lower health literacy was associated with greater difficulties understanding and processing cancer related information, poorer quality of life and poorer experience of care. Personal and situational influences contributed to how participants processed information and reached decisions about their care. Conclusion This review highlights the important role of health literacy for patients with cancer. Outcomes are poorer for those who experience difficulties with health literacy. Further efforts should be made to facilitate understanding, develop health literacy and support patients to become more involved in their care.


Introduction
Patients with cancer are expected to understand complicated information about their diagnosis and management. They must learn a new language of health terminology, provide consent for treatments and procedures, turn up at the right time and place for their appointments and seek help in an appropriate way and in a timely manner. Health literacy is integral to this.
Health literacy has been defined as "the combination of personal competencies and situational resources needed for people to access, understand, appraise and use information and services to make decisions about health. It includes the capacity to communicate, assert and act upon these decisions" [1]. Two distinct views of health literacy, as either a 'risk' or an 'asset', have been proposed [2]. The 'risk' approach is largely associated with work in the clinical domain, exploring the impact of health literacy on individual and health system outcomes. Health literacy in this context is used to describe an individual's literacy skills, and low health literacy is seen as a risk factor that must be compensated for. In the 'asset' approach however, which has developed from work in public health, health literacy is seen as an asset to be built, comprising more than just functional skills and including the development of more advanced social and communication skills as a means of increasing patient empowerment [2].
As a 'risk', health literacy is associated with hospitalisation, use of emergency care, uptake of preventative services, ability to understand health information and take medications appropriately, and, in older people, with health status and mortality [3]. Crucially, it is modifiable [4,5] and improving health literacy is increasingly recognised as a way of improving outcomes, including in Europe's Beating Cancer Plan [6].
In addition to relationships with health outcomes, health literacy is a prerequisite for shared and informed decision making [7,8] and has close ties with person-centred care, which aims to support patients to develop their knowledge, skills and confidence to participate in a partnership with their healthcare provider [9]. Edwards et al.'s Health Literacy Pathway Model considers health literacy from the 'asset' perspective, and portrays the development of health literacy as a process over time, influenced by personal, emotional and facilitating factors, leading towards active involvement in consultations and shared decision making [8]. Such involvement is particularly important in cancer care, where patients are often faced with preferencesensitive decisions, and these closely related concepts are therefore very relevant to this setting. This paper aims to provide an up to date overview of the literature, enabling us to understand the clinical relevance of health literacy in cancer care more broadly than existing reviews focussing on self-management behaviours [10], limited to studies from the USA [11] and exploring interventions to improve health literacy [12] have allowed. Given the complexity of the concept and the personal preference-specific nature of decisions made in the oncology setting, a mixed studies approach was chosen. This was to ensure that the patient voice was heard alongside the quantitative findings and to provide further insight into the patient experience than might have been possible through analysis of measured patient reported outcomes. To our knowledge, no prior systematic reviews have sought to bring together such a comprehensive outline of the field in this way. The objectives were: 1) to identify which outcomes relate to health literacy in patients with cancer and 2) to explore the role and consequences of health literacy, reported by qualitative studies, as patients with cancer access, understand, appraise and use information and services to make decisions about health.

Methods
The review protocol was prospectively registered with the International Prospective Register for Systematic Reviews (PROSPERO), entry number CRD42020166454. Wording of the qualitative objective has been refined since registration.

Search strategy
Searches were carried out on four electronic databases (MEDLINE, EMBASE, PsycINFO and CINAHL) in January 2021. Publications addressing cancer and health literacy were sought using search terms identified through review of the existing literature, including MeSH terms (neoplasms, health literacy) and keywords (cancer, malignancy, neoplasm, tumour, carcinoma, health literacy and health competence). Specific outcomes were not stipulated due to the anticipated varied nature of the studies. The search strategy was reviewed by an experienced librarian and is shown in S1 File. Visual scanning of reference lists from included studies was undertaken. Citations were managed through Endnote X9 and Microsoft Excel.
Screening of titles and abstracts was undertaken by two independent reviewers, with one screening all papers (CH) and three reviewers screening a third of papers each (AH, RW, SW), with a preference for inclusion if there were disagreements. Following the initial screening process, full texts of the remaining studies were obtained and independently reviewed for eligibility by two authors (CH and AH, RW or SW) according to the following criteria:

Inclusion
• Patients aged �18 years with malignancy of any site (if mixed group, data able to be separated) • English language AND • Quantitative papers: • Health literacy assessed with validated tool (concerning general or cancer health literacy, used in its validated form in its entirety) • Measured outcomes associated with health literacy OR • Qualitative papers: • Studies exploring the role of health literacy as patients access, understand, appraise and use information and services to make decisions about health

Exclusion
• Use of the term 'health literacy' but referring to disease specific knowledge only • Case reports, review papers, conference proceedings, opinion pieces, editorials, letters to the editor, dissertations/theses, book chapters, protocols At all stages, disagreements were resolved through discussion.

Data extraction
One reviewer (CH) extracted data from all papers, with independent extraction from eight papers by a second reviewer (RW, SW) to check for accuracy. Data on study characteristics (author, year, country study undertaken, setting, design, aims/objectives, inclusion/exclusion, recruitment procedure, health literacy measure used and how limited health literacy defined), sample (age range, sex ratio, cancer site, stage, number of participants and number of eligible participants if mixed group, proportion limited health literacy according to measure used and by tumour site), outcomes (as reported in individual studies, measures used and effect of health literacy on these) and qualitative methods used, data analysis procedure, key themes and findings and participant quotes were collected.

Quality appraisal
Quality was assessed using the Mixed Methods Appraisal Tool (MMAT) [13], allowing all study types to be appraised using a single tool for consistency. MMAT scores are given out of a total of 5, with a point scored for each 'Yes' answer, and no points awarded for 'No' or 'Can't tell' responses. Studies were considered higher quality if they scored 4/5 or 5/5. Quality assessment was carried out by two independent reviewers, with one assessing all papers (CH), and three assessing a third of the papers each (AH, RW, SW). Disagreements were resolved through discussion.

Data synthesis
Statistical pooling of data was not performed due to the varied study designs, outcomes, health literacy assessment tools and thresholds used to identify participants with lower health literacy. Drawing on guidance developed by Popay et al. [14], a narrative synthesis was undertaken. After extraction of the data, studies were grouped and tabulated based on the two review objectives. To address the second, qualitative objective, a thematic analysis was performed [14]. The key themes, authors' descriptions and interpretations, and supporting quotations were extracted from the results sections of the original qualitative papers, alongside relevant contextual data. Using an iterative process, similar themes were then grouped and used to develop meta-themes, drawing on existing definitions and theoretical frameworks [1,8,15]. An initial grouping and development of meta-themes was made by CH before being further refined by RW, after which all authors reviewed the primary texts and met to discuss each meta-theme and contributory theme. Any disagreements were resolved through discussion until consensus on the final grouping was reached.
Finally, relationships between studies across both objectives were explored through concept mapping, again drawing on existing models as appropriate [1,8], and findings integrated.

Results
4440 records were retrieved from the searches (Fig 1). After removal of duplicates, 2496 titles and abstracts were screened. Full texts of 405 papers were reviewed for eligibility, and 66 papers relating to 60 studies were ultimately selected for inclusion. One additional eligible study was identified through reference list scanning.

Objective 1: Outcomes relating to health literacy in patients with cancer
Fifty-eight papers relating to 52 studies addressed this objective, of which 49 studies were of quantitative non-randomised design. The majority were conducted in the USA (31/52), and the most common health literacy assessment tools were variations on the Chew screening questions (16/52) and Rapid Estimate of Adult Literacy in Medicine (REALM) (12/52). Breast (N = 12) and prostate cancer (N = 8) were the most studied individual tumour sites, and a further 21 studies included participants with a variety of cancer diagnoses. Thirty-two papers were considered higher quality (MMAT score 4/5 or 5/5). See Table 1 for details of included studies and Table 2 for a summary of the reported associations between health literacy and outcomes. Additional study details can be found in S1 Table. When referring to the significance of associations, the threshold for statistical significance is taken to be p<0.05.
Information processing. Five higher and seven lower quality studies considered outcomes relating to information processing. Lower health literacy was associated with lower ease of understanding, as well as higher and more variable estimates of risk relating to breast cancer recurrence in women with early stage disease [16,25]. Participants with lower health literacy had significantly higher unmet information needs in another large (N = 1060) study of patients with newly diagnosed breast cancer, although confounding variables were not controlled for [69]. A smaller but higher quality study of patients with mixed tumour sites, which did consider confounders, reported a significant association between health literacy and information needs in bivariate analysis only [37]. For radiotherapy outpatients with lower health literacy, the need for information about treatment at a single centre decreased significantly from pre-initial consultation to 3-5 weeks after the initial visit [50], though again, confounders, including time between consultations and treatment course length, were not accounted for.  Higher health literacy was associated with higher information recall in patients with breast cancer and in older patients with mixed tumour sites [34,49]. It was also associated with greater disease specific knowledge about human papilloma virus (HPV) among patients with head and neck cancer [20], greater prostate cancer knowledge in patients with the disease [60], trials knowledge [64] and, in a small single centre study, with chemotherapy knowledge [62].

Category Association Outcomes
Information processing Lower health literacy: Lower understanding [16,25] Poorer estimation of recurrence risk [16] Greater information needs and greater decrease in needs over time [37,50,69] More time spent on prognostic information and infographic (eye tracking) [73] Higher health literacy: Higher recall [34,49] Greater knowledge [20,60,62,64] Decision making Higher health literacy: Preference for more active participation [34] Higher perceived involvement [28] No association: Preference for more active participation [44] Observed shared decision making [67] Quality of life Lower health literacy: Poorer quality of life [19,23,29,30,33,39,42,47,48,52,65,68] No association: Quality of life [18] Treatment and health service use Lower health literacy: Increased number and length of hospital admissions [17] Increased likelihood of GP follow up for cancer [41] Increased use of post-operative discharge services [59] Increased likelihood of treatment complications [54,63] Higher health literacy: Increased odds of receiving chemotherapy [26] Increased likelihood of breast reconstruction [24] Lower likelihood of receiving unproven treatment [36] Greater treatment continuity [72] No association: Hospital admissions and emergency department visits [63] Adherence Lower likelihood of patient participation in multidisciplinary tumour conferences [57] Other Lower health literacy: Lower levels of physical activity [19,43] Higher cancer related and unexpressed needs [53,56] Greater need for psychological support [61] Increased financial hardship [38] Increased fear of progression or recurrence [30,55] Greater treatment regret [58] Lower self-care management [32] Greater distress [19,22,41] Increased upper extremity disability after breast cancer [29] Higher health literacy: Greater self-efficacy [46,53] Preference for patient centred care [27] Greater patient engagement [45] Fewer practical concerns [46] Increased willingness to participate in a clinical trial if offered [44] No association: Self-efficacy [42] Mortality [26] Distress [22] Perception of doctors' communication of recurrence risk [70] Perceived changes to longevity with treatment or observation [40] Presentation with early stage disease [26] Self-care behaviours [33] https://doi.org/10.1371/journal.pone.0259815.t002 A small study of patterns of eye tracking reported a difference between time spent on aspects of a web based prostate cancer decision aid by those with higher and lower health literacy [73]. Those with lower health literacy appeared to spend longer on the prognostic text and infographic, but this was based on very limited data from 12 participants. Decision making. Four studies exploring health literacy and decision making (three of higher quality) found mixed results. Using self-report measures, an association between higher health literacy and preference for more active participation in decision making was reported in one study of women making decisions about breast cancer recurrence risk testing [34], and with higher perceived involvement in shared decision making in another cross-sectional study of cancer patients at a single centre [28]. Yet no association was found when assessing preference for involvement in decision making regarding participation in clinical trials [44]. A secondary analysis of data from a randomised controlled trial evaluating decision aids for breast cancer, the only study to measure observed shared decision making, did not find a difference according to health literacy [67]. The chosen cut point for the health literacy screening question was higher than is recommended [74,75], with a higher sensitivity but lower specificity for detecting lower health literacy, which may account for the lack of difference seen.
Quality of life. Twelve studies, of which eight were higher quality and five had over 500 participants, reported an association between lower health literacy and poorer quality of life. Studies included patients with colorectal [19,39], breast [29], prostate [52,65], lung [33], head and neck [23,30] and mixed tumour sites [42,47,48,68], and used a variety of health literacy and quality of life assessment tools. Only a single survey did not find a significant difference in quality of life between patients with low and higher health literacy [18], which may be due to its relatively small sample size compared with other large higher quality studies [19,29,47]. The study included a convenience sample of 97 patients with mixed tumour sites recruited from the waiting rooms of two clinics, and assessed health literacy using three different tools. Lower health literacy ranged from 5%-46% using the different measures, though the authors note there was no association between any measure and quality of life.
Treatment and health service use. Eleven studies considered treatment and health service use, six of which were higher quality. Higher health literacy was significantly associated with increased odds of receiving adjuvant chemotherapy for stage III/IV colorectal cancer [26]. It was also associated with an increased likelihood of reconstruction after mastectomy in a cross sectional study of 336 women with breast cancer attending a single centre [24], though this was significant in univariate analysis only. A further study of men with prostate cancer identified a trend for those with higher health literacy having a lower likelihood of undergoing unproven salvage androgen deprivation therapy for prostate specific antigen (PSA) recurrence [36], but this was again significant in univariable analysis only. In a single centre study of patients receiving chemotherapy, those with lower health literacy experienced more grade 3 and above adverse drug reactions [54].
Although no association was found in one retrospective study [63], lower health literacy was significantly associated with increased number and length of hospital admissions in a cohort study of patients with mixed tumour sites (N = 752) [17] after controlling for diagnosis, receipt of chemotherapy, comorbidities and other variables. In a national survey of 4045 French cancer survivors 5 years post diagnosis, those with lower health literacy were more likely to see their general practitioner for follow up of their cancer, which may suggest increased health service use, though data on frequency, reasons for visits, and contact with a specialist was not collected [41].
Patients requiring post-cystectomy discharge services in one centre had lower health literacy scores; significant on bivariate analysis [59], however, a change in practice during the study period led to an increase in the number of patients receiving discharge services regardless of risk factors is likely to have affected outcomes. In the same centre, those with lower health literacy were significantly more likely to experience a minor post-operative complication [63]. Treatment continuity for patients with non-muscle invasive bladder cancer was significantly higher in those with adequate health literacy in another study [72], but it is not clear how this was assessed, and confounders were not controlled for in the analysis. In another study, self-reported adherence to follow up after bowel cancer was not associated with health literacy [38].
Medication adherence. Four studies, of which one was higher quality [21], explored the association between health literacy and oral medication adherence. Adherence to general medications [71], specific oral chemotherapy (capecitabine) [21], and to various anti-cancer medications, including hormonal and targeted treatments [51,66] was assessed. Higher health literacy was associated with higher levels of adherence in three studies of up to 100 participants [21,51,71]. One study of patients with gynaecological cancers (N = 100) did not report a significant association, though it was not powered to detect predictors of non-adherence [66]. All but one study [21] relied on self-report.
Care coordination. Three studies considered aspects of care coordination, of which two were small but considered higher quality [31, 35]. One survey of Chinese migrants with cancer in Australia found a positive correlation between higher health literacy and better experience of care coordination [35]. Another, of patients with mixed tumour sites [31], found an association between lower health literacy and higher input required from a patient navigator, although this did not remain significant in multivariate analysis. The third study involving 863 women with breast cancer found that those with 'inadequate' health literacy, as determined by the HLS-EU-Q16, were significantly less likely to participate in multidisciplinary tumour conferences than those with 'sufficient' health literacy [57].
Other outcomes. A range of other outcomes were also explored. Lower health literacy was associated with lower levels of physical activity, significant on bivariate analysis in a large study of patients with colorectal cancer [19], and in stepwise regression analysis of patients with breast cancer [43], and with significantly increased upper extremity disability after breast cancer in bivariable analysis [29]. It was also associated with higher cancer-related [53] and unexpressed needs [56], increased likelihood of need for psychological support [61], increased fear of progression in a study of older patients with breast cancer [55], higher fear of recurrence in patients with head and neck cancers [30], and greater treatment regret in a small study of men with prostate cancer [58]. Lower health literacy was associated with greater distress in three studies [19,22,41], though the same association was not found when one of the studies used a different measure of health literacy [22]. Lower health literacy was significantly and independently associated with increased cancer treatment related financial hardship [38], and self-care management scores were lower for patients with lower health literacy in another single centre study [32]. Higher health literacy was associated with greater preference for patient centred care [27], patient engagement [45], and self-efficacy in two studies [46,53], though no association was found in a third single-centre study [42]. Those with higher health literacy were significantly more likely to report willingness to participate in a clinical trial if one was offered [44], and men with early prostate cancer and higher health literacy reported significantly fewer practical concerns [46].
Mortality [26], presentation with early stage disease [26], self-care behaviours [33], perception of doctors' communication of recurrence risk [70] and perceived changes to longevity with treatment or observation [40] were not associated with health literacy.

Objective 2: Qualitative studies exploring the role and consequences of health literacy as patients with cancer access, understand, appraise and use information and services to make decisions about health
Eight qualitative studies were identified and add the patient voice to the findings of this review (Table 3). Studies included patients with prostate cancer [76][77][78], breast cancer [79,80] and haematological malignancies [81]. One included patients with different primary tumours [82] and one study did not report on tumour site [83]. Six studies were of higher quality according to the MMAT, and one mixed methods study scored highly for the qualitative component but achieved a lower score overall. Table 4 demonstrates how the original themes reported by the individual studies were grouped to form meta-themes. The meta-themes identified included situational influences (networks and system), personal influences, information processing, and consequences of health literacy. Situational influences refer to the factors external to the person which influence their ability to process information. They include network influences, incorporating sources of information and support outside of the healthcare environment, as well as system influences, relating to professionals within the healthcare system and structural factors involved in care delivery. Personal influences refer to more internal factors that might contribute to health literacy, such as prior experience, cultural values and emotions. Information processing encompasses the strategies described by patients to help them deal with and process the information they face. Consequences refer to the outcomes of these influences and processing, and include negative aspects, such as fear or uncertainty, as well as more positive outcomes, such as empowerment and better understanding.
Situational influences. All eight papers described themes relevant to the role of external or situational influences on health literacy. Two key areas were identified: the importance of networks, which were largely supportive and facilitated understanding; and the system, which often acted as a barrier and inhibitor to the development of health literacy.
Situational influences-Networks. Social and informational networks played important roles as facilitators of health literacy and were among the situational resources available to patients enabling them to access, understand, appraise and use information and services. Although some participants expressed a preference to deal with their diagnosis by themselves [76], many relied on friends and family as sources of information and support [76-78, 80, 81]: "I was a little reluctant because I really didn't know that much about the IORT at first. But then I talked, actually after talking to a friend of mine who had, you know, the traditional radiation, she said, "Man, I can't imagine how much better it would be just to do it once, just to have one dose of radiation.". . . So after talking to my friend who had a very bad experience, she got burned. . . I just decided I didn't want to do the traditional" (female phase 2 clinical trial participant, adequate health literacy) [80]. Learning from other patients about their experiences, often through support groups or organisations, allowed participants to develop a greater understanding of their diagnosis and treatment [77,78,81]. Support groups also offered participants the opportunity to hear from and talk to 'experts' outside the consultation setting, helping to build confidence to ask questions: "You find confidence and get encouraged to talk to health professionals, ask questions, and that will only come through building confidence. If you have any problem, try to seek the answer for it" (73 year old attendee at prostate cancer support group for 14 years) [77]. Only occasionally, these social connections acted as barriers, such as when the knowledge imparted was inaccurate or led to increased fear [83]. The internet was a valuable resource for many participants, who were able to use it to find further information and additional support [76,77,80,81]: "I have done research through the Internet. The Leukemia Society, I called them, and they got me, they hooked me up with another patient that had   Table 4. Meta-themes and the contributory themes extracted from original papers.

Meta-themes Themes from original papers
Situational influences Relying on others [81] • Networks Involvement of their spouses and the people around them [76] Group information processing [77] Recognising opportunities to learn from each other [83] Social and informational networks [78] Their use of the internet for information retrieval [76] Pragmatic decision making [80] Situational influences Overcoming professional medical language [78] • System Self-management support [82] Delivery system design [82] Support from and interaction with healthcare professionals [76] Opportunities for practical services and programmes for health literacy in relation to cancer [83] Inequality of access to health information [79] They did not tell me [81] Decision support [82] Confidence in provider recommendation [80] Personal influences Recognising susceptibility to cancer [83] Community characteristics [82] Spiritual and emotional influences [78] Literacy levels [78] Silence among men [78] Information processing Numbers and measures as the foundation of prostate cancer literacy [77] Deductive and hypothetical reasoning [78] Weight of risks and benefits [80] Consequences Shopping around [77] Decision dilemmas [81] Fears of dying [81] Tough symptoms and side-effects [81] Medication usage and adherence [79] Acquisition of medication information [79] https  [81]. Situational influences-System. Health literacy was also influenced by 'system' factors that are outside participants' control. Professionals within the healthcare system played a vital role in imparting information, and, when done well, participants' confidence in their clinicians made them feel more comfortable in their decision making [80]: "I was concerned and I was very open, and they were open with me in explaining what the procedure would be. . . I almost made it right there on the spot because I just felt so secure that my problem would be taken care of" (female phase 2 clinical trial participant, marginal health literacy) [80]. Yet this information giving was not always done in a way that participants could understand [78,82]: "They used too many big words. . . It is a complicated procedure. They explained everything, but you still don't get it." (35 year old postal carrier undergoing stem cell transplant) [81]. The healthcare system itself placed high demands on participants' health literacy, with over-complicated forms which some participants signed without fully understanding: "I have signed a lot of papers without reading. I figure they ain't gonna give me nothing to sign if it's bad" (rural cancer patient) [82], and through inconsistent access to resources and opportunities to further understanding [79]. Participants in one study described cancer care as a "foreign" experience and didn't know what to ask about their treatment options [82], providing support more generally for the recommendations by Treloar et al. [83] for improved community education to raise awareness and help prepare people for such a diagnosis.
Personal influences. Participants' health literacy was also affected by personal influences. These included cultural and community values such as stoicism, which led to patients 'suffering in silence' rather than 'bothering' healthcare professionals [82] and a "silence among men" impeding open discussion and thus understanding [78]. Silence was exacerbated by limited experience of cancer prior to diagnosis: "Cancer has never sort of crossed my life till now. . . I used to be a health worker, an educator, but cancer was never part of my life, I never knew anyone with cancer, I never seen anyone with cancer, maybe on TV but not in the here and now, cos I was always busy with Aboriginal health and teaching Aboriginal health, but cancer was never part of our programme, which was a shame" (Aboriginal patient who had previously worked in health sector) [83]. Participants' general literacy was influenced by social and cultural exposures over time, typically encouragement at school or at home, and fed into their approach to learning about their condition [78].
Information processing. Some participants used strategies to help them process information, highlighted by two studies of male attendees at prostate cancer support groups. Focussing on numbers relating to pathological grading or biomarkers and the relationship of these to treatment options facilitated understanding of prostate cancer and allowed men to assess their options: "Researcher: In what ways did it [the prostate cancer support group] help you steer your treatment? Participant: By giving me information about how each of the approaches is and how it works, the long-term and short-term effects of each, the certainties and uncertainties around each one, and certainly the cure rate" (59 year old attendee at prostate cancer support group for three years) [77]. In another study, patients used a process of deductive and hypothetical reasoning, comparing information from different sources, or comparing themselves with others, to further their understanding, monitor their response to treatment, and verify information given to them by healthcare professionals [78]. A different study, assessing perceptions of the risks and benefits of participation in a trial of a novel radiotherapy technique for breast cancer, found that many participants did not believe there were any risks, and most focussed instead on the positives, such as convenience of the treatment, which were influential in their decision to take part [80].
Consequences. The result of some of these influences and processes can be seen in the wider consequences of health literacy. Where there was conflicting advice or poor understanding, this led to decision dilemmas, and participants were prepared to accept a treatment without full comprehension as a way of moving on and progressing their care [81]. When the information patients needed was not given, or not in a way they could understand, they experienced greater fear: "Many of the things you fear are those you don't understand" (42 year old industrial worker undergoing haematopoietic stem cell transplant) [81]. Poorer prior understanding also led to more unanticipated side effects [81], which in turn influenced decisions about medication adherence: FGA: "I don't take everything they give me." "If it has too many side effects, I don't take it." FGB: "I didn't take anything because I was afraid of the side effects". FGB: "They tell you some of the side effects but they don't tell you all the side effects" (quotes from female African American participants with breast cancer from two focus groups (FGA and FGB)) [79].
Conversely, those who had developed a good understanding and the confidence to do so were able to effectively navigate the healthcare system and exercise their rights to ensure that they received 'good care': "I felt this urologist was pushing me for surgery and I thought geez, I don't have enough knowledge, I want data. So, I kept pushing him, to tell me where the groups [PCSGs] were and he was reluctant to tell me, but finally he agreed, and I went, and I never went back to this guy and I started my search and the prostate cancer groups were very instrumental in helping me to make my decision. They gave me knowledge." (63 year old attendee at prostate cancer support groups for 10 years) [77]. Patients accessed information in different ways, influenced by personal factors including the desire not to be a bother, and situational influences, such as time pressures on healthcare staff. When these influences were removed, if patients knew their pharmacists well or they appeared to have time to talk, for example, or if patients had the confidence to overcome these influences, it was possible for them to seek and obtain the information they needed [79].

Combined synthesis and conceptual map
The concept map below (Fig 2), shows the relationships between the outcomes associated with health literacy as identified by the quantitative studies (Objective 1) and the meta-themes identified from the qualitative synthesis (Objective 2). It draws on the existing framework developed by Edwards et al. [8] and the definition of health literacy proposed by the International Union for Health Promotion and Education [1].
Patients with lower health literacy may have more difficulty understanding and recalling the information they have been given, demonstrate lower knowledge and have higher unmet information needs (' Processes', Fig 2). The qualitative data suggest that situational influences, including the way that information is delivered, the complexity of the system, and the lack of resources available to patients make these tasks more challenging ('Influences'). As such, patients are prepared to consent to treatments without fully understanding what they may entail, or what the potential risks and benefits are ('Outcomes'). Fear and unanticipated side effects may arise as a consequence of lower health literacy through lack of understanding ('Outcomes') yet fear itself may influence and hinder comprehension ('Influences').
Those with higher health literacy are better able to process information ('Processes'), engage more with health promoting activities such as exercise, and experience a better quality of life ('Outcomes'). They may be more likely to seek out the additional information they need, perhaps learning to interpret numbers relating to their condition or finding opportunities to learn from others ('Processes'). With greater understanding and knowledge of their disease and their rights, they may take a more active role in making decisions about their care and have greater confidence in navigating the system ('Outcomes').

Discussion
The findings from this mixed studies systematic review demonstrate the role and consequences of health literacy in the oncology setting. The outcomes associated with health literacy are varied, with some having clear implications for care delivery, and others demonstrating the negative impact of health literacy difficulties on the experience of care as reported by patients themselves. While the quantitative data gives evidence for the measurable outcomes associated with health literacy, the qualitative findings complement this by adding the patient voice, identifying some of the influences of health literacy, and offering an insight into some of the associations seen. Findings relating to information processing and decision-making highlight some of the 'Processes' affected by these 'Influences' and demonstrate how health literacy may link to the described 'Outcomes'. Ensuring that the system is considerate of the burden it places on patients, taking steps to simplify information and processes, providing patients with the confidence and opportunities to speak up, and making support available is therefore essential.
Although further empirical work is needed to determine the nature of these associations, the causal links between health literacy and health outcomes have been hypothesised [84,85]. These models consider the range of mediating factors that may influence the pathway, including patient and system factors affecting access and utilisation of health care, provider-patient interaction and self-care [84], as well as health status, attitudes, emotions, motivation, self-efficacy and ecological resources [85], some of which are included as associated outcomes in this review. In addition, health behaviours and outcomes may in turn influence these mediators and health literacy skills [85]. Poorer quality of life, for example, which was consistently associated with lower health literacy, may be linked with other outcomes identified in this review, such as increased fear, greater financial hardship or a worse experience of treatment, as such outcomes may influence or indeed act as mediators in the pathway. Whether improving health literacy itself leads to better quality of life is as yet unknown [86], and this is an important outcome for further study.
Our review supports the Health Literacy Pathway Model presented by Edwards et al. [8], which draws on Nutbeam's conceptualisation of health literacy as an asset that can be developed over time [2]. The model incorporates internal and external influences that may positively or negatively affect a person's health literacy [8], factors we have also found to be important in patients with a cancer diagnosis. Such patients face many new challenges at a highly emotional time. It is therefore crucial that the systems and networks are in place to support patients, making it easier for them to access, understand, appraise and use the information they want and need by removing as many additional barriers as possible. In doing this, patients are afforded the best chance of being able to develop and use their health literacy to take an active role in their health and make informed decisions based on what is important to them.
The decision-making preferences and degree to which patients with lower health literacy feel able to take on a more active role in the oncology setting require further study. But whether a patient wishes to be actively involved in decision making or prefers to be guided by their clinician, an understanding of the aims and potential risks of treatment are key to informed consent [87]. The General Medical Council (GMC) guidance for doctors in the United Kingdom highlights the importance of taking steps to facilitate understanding, acknowledging that patients have different information needs and may prefer to receive information in different formats [87]. Our findings suggest that this is not always achieved.
One limitation of this review is the exclusion of studies using measures relating to health literacy but referring only to literacy. This was to ensure that health literacy remained the subject of interest, but other studies may have been missed. Secondly, to achieve consistency in a field with a range of measures, it was agreed that only those health literacy assessment tools used in their validated form would be included. Although this excluded some studies using non-validated adaptations, it was deemed important in order to be able to draw any comparisons between studies. As found elsewhere, the range of health literacy measures and identification of participants with lower health literacy makes such comparisons difficult. Over half of the included studies were conducted in the USA, with none carried out in the UK, which may limit the relevance of some results to other healthcare settings.
A major strength of the review is the use of the mixed methods approach, bringing together a more comprehensive picture of health literacy in the oncology setting, incorporating the patient voice and allowing us to better understand the experience from the patients' perspective. The broad inclusion criteria allowed us to identify the association between health literacy and a wider range of outcomes than has previously been addressed [10,11]. Additionally, involvement of a multidisciplinary team of experienced researchers and clinicians at all stages of the review ensured consistency and rigour throughout the process.

Conclusion
Health literacy plays a key role in cancer care, with important implications for patient experience and outcomes. Those with lower health literacy face greater difficulties processing information, report poorer psychological outcomes and experience a poorer quality of life, whilst those with higher health literacy appear better informed and able to take on a more active role in managing their health. Future interventions aimed at supporting person centred care in this setting should therefore take account of health literacy and consider the factors influencing its development. Further research is required to better understand the decision making processes and preferences of those with lower health literacy receiving care for cancer.
Supporting information S1