The influence of health literacy on the timely diagnosis of symptomatic cancer: A systematic review

Abstract Low health literacy has been associated with poor cancer screening uptake, difficulty in making treatment choices and reduced quality of life following a cancer diagnosis, yet it is unclear whether and how health literacy influences the pathway to diagnosis for patients with cancer symptoms. This systematic review aimed to evaluate the influence of health literacy on the timely diagnosis of symptomatic cancer. Literature was searched between January 1990 and May 2017 using MEDLINE, Embase, Scopus, ASSIA, CINAHL and PsycINFO. Only three papers met the inclusion criteria. These reported two qualitative studies and one quantitative, with adult patients diagnosed with gastrointestinal (colon, rectum and pancreas), cervical and breast cancer. The definition and assessment of health literacy varied between the studies, as did the descriptions of the pathway to diagnosis. Due to the methodological weaknesses identified, the conclusions are limited; however, the studies did highlight important considerations in the definition and measurement of health literacy. Further research is required that clearly defines health literacy and follows the principles of the Aarhus Statement to assess the influence of health literacy on the pathway to cancer diagnosis. The protocol for this review was registered with PROSPERO (CRD42016048917).

studies of cancer screening (colorectal, breast, cervical and prostate) and concluded that there was a trend towards low screening rates with low health literacy. This could be influenced by the association of low health literacy with low knowledge of cervical cancer screening (Lindau et al., 2002), or the burden on patients with low health literacy to read and understand written information in relation to colorectal cancer screening (von Wagner et al., 2009). Understanding information, both written and oral, can also be difficult for those with low health literacy faced with complex treatment options, thereby limiting engagement and participation in shared decision-making (Amalraj et al., 2009). Difficulties in communication may also act as a barrier for accessing support services, possibly contributing to worse mental health outcomes as seen for low health literate men newly diagnosed with prostate cancer (Song et al., 2012). While this suggests that low health literacy may affect cancer screening and treatment pathways, it remains unclear how health literacy may influence the timely diagnosis of cancer in symptomatic patients.
Promoting timely diagnosis has been a priority in improving outcomes for cancer patients since the launch of the National Awareness and Early Diagnosis Initiative (NAEDI, 2008) in the UK. Key findings have identified a significant variation in the time to diagnosis across cancers, from symptom onset to first presentation in primary care, subsequent referral and cancer diagnosis Lyratzopoulos et al., 2015). Frameworks such as the Categorisation of Delay model (Olesen, Hansen, & Vedsted, 2009) and the Model of Pathways to Treatment (Scott, Walter, Webster, Sutton, & Emery, 2013;Walter, Webster, Scott, & Emery, 2012) are useful for describing the intervals along the pathway to diagnosis, while the Aarhus Statement suggested ways to improve the design and reporting of studies (Weller et al., 2012). Outcomes could be improved by decreasing the time between a patient first noticing a potential cancer symptom and seeking help (appraisal and help-seeking, or patient interval), and reducing the time between initial consultation with a healthcare professional, referral to secondary care and diagnosis (diagnostic, or primary and secondary care interval). In considering the pathway to diagnosis, it is possible that an individual's health literacy may influence timely diagnosis through a person's ability to access and understand cancer symptom information, appraise the information in relation to their own bodily changes and navigate the healthcare system; presenting to a healthcare practitioner and accessing the specialist care required to receive a diagnosis.
In this systematic review, we therefore aimed to evaluate the impact of health literacy on the timely diagnosis of symptomatic cancer.

| Search strategy
We searched peer-reviewed literature published worldwide from 1 January 1990 to 19 May 2017. The search was limited to 1990 onwards as health literacy is a relatively new field, with the number of studies expanding following the publication in 1991 of the first widely used health literacy instrument, the Rapid Estimate of Adult Literacy in Medicine (REALM) (Davis et al., 1991). Six bibliographic databases were searched, with the strategy developed and run in MEDLINE (Table 1)

| Inclusion and exclusion criteria
Studies published in any language were included where they focused on adult patients (aged 18 years and older) with a primary diagnosis of any cancer and explored the influence of health literacy (or literacy/numeracy related to health yet not termed as "health literacy") in relation to the time to diagnosis of symptomatic cancer. This included studies evaluating the total time to diagnosis, from symptom onset to diagnosis, or focusing on one or more intervals along the pathway: appraisal, help-seeking or

| Study selection
Following removal of duplicate references, EH screened the titles and abstracts against the inclusion and exclusion criteria, with a random sample (10% of the total) assessed by FMW and JB to confirm agreement. The full text was obtained for all studies identified as potentially relevant to the review. Three reviewers (EH, FMW and JB) screened all the full-text articles to identify studies for inclusion in the review.

| Data extraction
Data were extracted by EH from each of the included studies: study type, recruitment setting, data collection details, participant characteristics, patient pathway/interval data as defined within the study, health literacy data including the definition and health literacy instrument used (if any), and the findings in relation to time to diagnosis. The extracted data were reviewed by FMW and JB to confirm completeness.

| Quality assessment
Quality of the studies was assessed by EH and reviewed by FMW and JB, using the Joanna Briggs Institute Critical Appraisal Tools (Joanna Briggs Institute, 2016). The aim of the quality assessment was to determine the validity of the results based on the design, methods, analysis and conclusions of each study and to assess the relative contribution of each study to the review.

| Protocol registration and reporting
Prior to starting the review, the protocol was registered with PROSPERO (CRD42016048917), an international prospective register of systematic reviews.

| Study selection
The search identified 5,188 citations, and after removing duplicates, 2,304 titles and abstracts were screened against the inclusion and exclusion criteria to identify 26 potentially includable studies ( Figure 1). Following full-text assessment, three studies qualified for

| Study characteristics
The three included studies were set in Japan, the USA and Egypt, and included patients diagnosed with gastrointestinal (colon, rectum and pancreas), cervical or breast cancer (Table 2). Two studies used qualitative methods, while the third reported a survey. The number of participants ranged from 10 to 37 and the overall sample was predominately female (92%). The studies set in the USA and Egypt were published in English, while the study set in Japan was published in Japanese and was therefore professionally translated. Due to the heterogeneity of studies, it was not possible to synthesise the findings; therefore, we chose to use a descriptive approach to analyse and report the findings.

| Quality of included studies
The studies were assessed based on methodological quality and conceptual clarity in relation to definitions of "time to diagnosis" and "health literacy." Table 3 summarises the methodological quality of the studies, which ranged from poor to adequate. Study details such as setting, sampling strategy and exclusion criteria were poorly described, with only one study fully describing the participant inclusion criteria (Nakagami & Akashi, 2010). The small sample sizes impacted the analysis, with Nakagami and Akashi unable to reach theoretical saturation based on the grounded theory approach used (Strauss & Corbin, 1998). From the reporting of the qualitative studies, it was also unclear whether the data accurately reflected the voices of most participants or a minority (McEwan, Underwood, & Corbex, 2014;Nakagami & Akashi, 2010). While acknowledging the limitations in the study design and conclusions, studies were not excluded based on the quality assessment alone.

Time to diagnosis
All the studies aimed to explore the diagnostic pathway from the patients' perspective, from symptom onset to diagnosis or start of treatment. Table 4 summarises the definition(s) used by the authors to describe the intervals along the pathway, and the corresponding interval defined by the Model of Pathways to Treatment (Scott et al., 2013;Walter et al., 2012) and the Categorisation of Delay model (Olesen et al., 2009 Age as defined within the study. The precise age of the participants was not stated. b Setting defined within the previous associated quantitative study (Corbex, 2010

Health literacy
The definition and assessment of health literacy varied between the three included studies (  People, 2000), and Nakagami and Akashi referring to seven definitions, with a focus on three (Mancuso, 2008;Murata, Arakita, & Shirai, 2006;Nutbeam, 2000). McEwan et al. used their own definition, describing health literacy in relation to symptom interpretation and knowledge networks. However, they also used a conceptual framework, the "social ecological model" (Scheidner, 2006), to analyse and present their data, which extended the definition of health literacy to include patient beliefs around risk factors.
To assess health literacy, Tecu and Potter used the short-form of the REALM (REALM-SF) (Davis et al., 1993), a validated instrument designed to assess pronunciation of medical words. It has been widely used in health literacy research but is primarily an assessment of comprehension rather than an assessment of ability to obtain and use health information. From the analysis, the assessment of the REALM-SF score in relation to time to presentation was limited (Tecu & Potter, 2012). The two qualitative studies explored health literacy via interviews with patients and in relation to the definitions of health literacy as proposed within each study.

TA B L E 3 Methodological quality of included papers
Author (

| Influence of health literacy on time to diagnosis
The two qualitative studies took different approaches to exploring Focusing on the diagnostic interval, Nakagami and Akashi found that it was lengthened where an alternative diagnosis was suggested or the patient was monitored or given medication for their  (2010) Nutbeam a (Nutbeam, 2000) Speros b (Speros, 2005) Mancuso c (Mancuso, 2008) Abbreviated definition as stated in the paper: "'Ability to acquire information relating to illness, medicine and health' (information about medicine and health), 'ability to compute', 'ability to read medical and health information', 'ability to understand medical and health information', 'ability to take the role of the patient', and 'ability to take appropriate decisions and to evaluate'"

Explored via interviews
Murata (Murata et al., 2006) "'Reading and writing and computation', 'information acquisition', 'perception, cognition and understanding'. 'analysis, selection and evaluation', 'action', response' and 'provision to others'" World Health Organisation (Nutbeam, 1998 The cognitive and social skills which determine the motivation and ability of individuals to gain access to understand and use information in ways which promote and maintain good health.  (Scheidner, 2006) Explored via interviews a Original definition from Nutbeam (2000): "The personal, cognitive and social skills which determine the ability of individuals to gain access to, understand, and use information to promote and maintain good health." b Speros 2005 does not provide a unique definition of health literacy and instead references the definitions provided by the World Health Organisation, American Medical Association, and US Department of Health and Human Services as above. c Original definition from Mancuso (2008): "A process that evolves over one's lifetime and encompasses the attributes of capacity, comprehension, and communication. The attributes of health literacy are integrated within and preceded by the skills, strategies, and abilities embedded within the competencies needed to attain health literacy." symptoms. In these instances, patients had to re-start the process of understanding, evaluating and acting on their symptoms, which relied on their health literacy ability.
In contrast to the qualitative studies, the survey study conducted by Tecu and Potter quantified the time to diagnosis and explored health literacy using the REALM-SF. The mean REALM-SF score was 60.08 (SD 12.63, range 0-66), with eight (22%) women scoring 54 or less and demonstrating low health literacy.
The authors stated that no statistically significant correlations were found between the REALM-SF scores and the patient interval (time from symptom onset to first presentation) across the whole cohort; however, they did not provide data to substantiate this. A sub-group analysis of the eight women with low health literacy found that four had a patient interval of 6-12 months; although, patient intervals were not reported for the 29 women who scored >54 on the REALM-SF. The survey also explored symptom experience, knowledge and help-seeking behaviours, yet presented these as descriptive statistics without any analysis of how these factors may correlate with health literacy or the patient interval.

| D ISCUSS I ON
This systematic review sought to explore how health literacy can influence the patient's pathway to diagnosis with cancer, as health literacy may affect a patient's ability to access and understand cancer symptom information, appraise the information in relation to bodily changes and navigate the healthcare system to access the special- This provides a flexible approach suitable for use with multiple methods to explore health literacy along the pathway to diagnosis for symptomatic cancer.

| Health literacy and stage at diagnosis
Two of the 23 studies excluded from the review following full-text assessment used validated measurement tools to assess health literacy in relation to stage at diagnosis (Bennett et al., 1998;Busch, Martin, DeWalt, & Sandler, 2015). Advanced stage disease is a major contributory factor to the poor survival outcomes across many cancers in the UK in comparison with Europe (Robb et al., 2009), and there are major efforts being made internationally to detect cancer at an earlier stage (Lyratzopoulos et al., 2012). A recent systematic review demonstrated the association between time to diagnosis and disease stage for some common cancers with equivocal findings for less common cancers . As it is unclear whether stage is an accurate indicator of time to diagnosis, the studies exploring health literacy and stage at diagnosis were excluded from the current systematic review. A further review investigating qualitative time to diagnosis research and evaluating the results in respect to a recent multi-dimensional definition of health literacy may now be needed.

| CON CLUS ION
Due to the few studies identified from the systematic search, and their methodological weakness and relatively poor quality, it was not possible to fully evaluate the influence of health literacy on the timely diagnosis of symptomatic cancer. However, the studies provide a starting point for research within this area and identify important aspects that need to be addressed in future research. When exploring diagnostic routes for cancer, researchers should be guided by the Aarhus Statement and underpin their research with a conceptual framework and clear definitions. Where health literacy is explored, researchers should also be aware of the range of health literacy definitions and assessment tools currently in use, and how these could relate to their research. Again, they would be advised to choose a definition best suited to their research area and to reflect on this and the aim of the research when choosing an appropriate instrument or method for exploring health literacy.
Reducing the patient interval is important for earlier diagnosis of cancer and it is possible that health literacy may influence the pathway, which could have important implications for developing targeted awareness campaigns for recognition of cancer symptoms and to prompt timely help-seeking, as well as informing GP-patient communication strategies. Research exploring the time to diagnosis should also consider the relation of health literacy to other factors affecting the pathway. In conclusion, further research is required that clearly defines health literacy and adheres to the principles of the Aarhus Statement to assess the influence of health literacy on the timely diagnosis of symptomatic cancer.

CO N FLI C T O F I NTE R E S T
None.

AUTH O R CO NTR I B UTI O N S
All authors contributed to the design of the study and analysis of data, and were involved in drafting and revising the manuscript.