A total of 24 relevant papers were included in the review. The initial stage of the search strategy identified four reviews that focused on health outcomes among our age groups of interest [17–20]. Of the 57 primary studies included within these four reviews, 11 met our inclusion criteria.
The second stage of the search strategy identified 2400 citations. Figure 1 shows the number of documents excluded at each stage. Additional reviews identified at this stage were also searched for primary papers not appearing in the citations from the database searches. Exclusions at full text appraisal stage were largely due to studies not meeting our criteria for age or ability to speak the dominant language.
The 11 studies identified from the initial reviews were also identified by the second stage search. The 2400 citations from the second stage search yielded an additional 13 studies that met our inclusion criteria.
The twenty-four papers that were included reported mainly on studies conducted in the USA (two were from the UK and one from Canada) (Additional File 2, Table S2). Studies used diverse methods to investigate the implications of functional or health literacy for various health-related issues in a range of health care contexts. The health related issues that the twenty-four papers considered were grouped into one or more of five outcome categories reflecting the five areas of interest in the review questions. An additional category of 'emotional responses' that it seemed inappropriate to treat simply as examples of (end state) health status emerged during the process and has been included (Additional file 2, Table S2).
Three measures of health literacy (REALM, TOFHLA and s-TOFHLA) and two measures of functional literacy (NART and the Test of Basic Adult Education) were used across the 24 papers included in the review (Additional file 2, Table S2). Implementation of the measures and cut-off points to determine low health literacy differed even when the same measure was used. (Additional file 2, Table S2).
All but three papers [21–23] considered potentially confounding demographic factors in their analysis. Only two papers [24, 25] clearly reported that the person who assessed health data was blinded to study participants' health literacy status. We now summarise the key findings relating to each of the five health-related outcome domains.
Relationship between functional or health literacy and health status
There is some evidence from 3 cross-sectional studies that lower functional or health literacy is associated with poorer health status, assessed by self-report or more objectively [26–28].
Studies of 1892 emergency department walk-in patients and of 339 people living with HIV-AIDS both found that those with lower health literacy were significantly more likely to self-report their health as poor [26, 27]. In the study of people with HIV-AIDS, recorded CD4 cell counts and undetectable viral loads in the medical notes confirmed the poorer health status of those with lower health literacy [27]. Both studies used the TOFHLA to measure health literacy but compared different cut-off points: Baker et al. [26] compared the highest and lowest of three health literacy levels (0-59 vs. 75-100) while Kalichman et al. [27] compared those above and below 80% correct.
A third study measured reading level using the Test of Basic Adult Education in 193 adult learners. Those with very low reading levels (at or below 4th grade), had significantly lower scores on the physical and psychosocial domains of the Sickness Impact Profile than those with higher (5th grade+) reading levels [28]. This study used an objective measure of health but focused on a group of people who were motivated to address their literacy difficulties and so were not necessarily representative of the general population with low literacy. People who have sought help with literacy may be more likely to let health professionals know they have difficulty with reading and writing. Psychosocial health impairment may be more prevalent in those who do not seek help with literacy education and so may be underestimated by this study.
Relationship between functional or health literacy and health promoting or health risk behaviours
Five studies were found to have investigated the relationships between health literacy levels and preventive health or health risk behaviours [22, 29–32]. All used the REALM to measure health literacy but no two used the same levels for comparison. Findings from these studies were complex and mixed.
Two studies found some higher health risk behaviours in those with lower health literacy but also some potentially conflicting evidence [22, 29]. In a US study of 130 women referred for colposcopy after abnormal pap smear, those with higher health literacy reported a greater number of risk factors for cervical cancer. Differences for individual risk factors varied; those with higher health literacy were more likely to report oral contraceptive use and having had 5 or more sexual partners in total while those with lower health literacy had higher parity. Health literacy was not associated with intercourse aged ≤18 years or with history of sexually transmitted disease other than HPV [29]. One UK study of 505 family planning clinic users, found that women with lower health literacy were: more likely to have been aged under 16 at first sexual intercourse; less likely to have used contraception at that time; and more likely to have had two or more partners in the previous 6 months [22]. The UK study of family planning clinic users found no significant difference across health literacy levels in planned or unplanned pregnancies, previous use of emergency hormonal contraception or number of sexual partners in the previous four weeks [22].
Comparison of these two studies is difficult because they categorised health literacy levels differently. Although both used the REALM, in the US study, participants fell into a broad range of health literacy levels and those scoring below 9th grade were compared with those at 9th grade or above [29]. The UK study converted the scores to UK reading ages; all participants had a reading age of 12 and above and comparisons were made between those with a reading age of 12-14 and 15+ [22]. This may explain some of the variance in the evidence; however, the studies also differed in the age by which first sexual intercourse was reported and the time period over which previous sexual partners were reported. Sexual health behaviours may also have been subject to different cultural influences in the two study settings. Multivariate analysis was not carried out in the UK study but the authors of the US study considered years of education, knowing someone with cervical cancer and having previous colposcopy as potential confounding factors.
Multivariate analysis from two further US studies found no association between health literacy and health risk behaviour. In a study of 600 pregnant women, no correlation was found between reading level and smoking prevalence [30]. In this group, race was significantly associated with smoking practice, African American women being significantly less likely to smoke and being significantly more likely to have lower health literacy. In a study of 423 female prison inmates, many of whom had dropped out of school, HIV risk behaviour was associated with educational attainment but not with health literacy [31]. One of the studies [31] compared 3 and the other [30], 4 health literacy levels across the REALM score.
In a further US study, lower health promoting behaviour in those with low literacy did not reach statistical significance. A cross-sectional study of 61 new mothers found that those with lower health literacy were less likely to initiate and sustain breastfeeding for the first two months of their infant's life. Breast feeding for at least two months was associated, but not significantly, with higher (12th grade+) health literacy [32]. This study had an insufficient number of participants and only two literacy categories, 7th-8th grade and 12th grade+ and this may have contributed to the lack of statistical significance.
This small group of 5 studies did not produce convincing evidence of a clear association between functional or health literacy and preventive health or health risk behaviours.
Relationship between functional or health literacy and access to and use of health services
One small qualitative study of 8 adults who participated in a community college literacy program and had been hospitalised met our inclusion criteria [33]. We assessed this study to establish what issues were identified in relation to the research questions and to ascertain whether any of these had been investigated in the quantitative studies. Participants had experienced impaired decision-making and given uninformed consent to interventions [33]. They reported having been unsure of what was expected of them as patients because they had been unable to read instructions, for example, on menus and notices [33]. They experienced fear; worry; powerlessness; stigma; vulnerability; diminished self-efficacy in accessing health services, and they balanced the risks of exposure of their literacy difficulties (stigma, decreased self-esteem) with the risks of non-disclosure (lack of knowledge gathering). Some, however, felt that the hospital was a special place where vulnerability could be shared and staff would keep information confidential. The frequency and distribution of these issues were not explored in the quantitative studies.
A further 6 quantitative studies focused on access to and use of health services. Two studies found no evidence [21, 25] and one found some evidence [26] of associations between health literacy and uptake of services. One study found an association between low health literacy and poorer access to treatment [27] in some cases. Evidence of association between health literacy and relationships with healthcare staff was mixed and unclear [34, 35].
A study of 543 parents found, as a secondary outcome measure, no correlation between total REALM score and accessing of preventive services for their children [25]. This could also have been considered as health promoting behaviour. A study of 202 African American women's use of prenatal care found no difference between high and low health literacy groups in the proportions of women beginning prenatal care. This study was underpowered, however, and had sought to exclude women who had no prenatal care notes [21].
Evidence from three studies suggested that low health literacy - as measured by the TOFHLA - may be associated with less appropriate use of health services or access to optimum treatment. A cross-sectional study of 1892 people attending an emergency walk-in department found that those with inadequate health literacy were more likely to have been hospitalised in the previous year than those with adequate health literacy [26] although their more frequent use of health services in general was non-significant after adjustment [26]. A study of 339 people with HIV-AIDS found those with lower health literacy were less likely to have been prescribed antiretroviral medication [27]. A further study reported an association between health literacy and relationships with healthcare staff, which may have implications for ensuring access to optimum treatment. Among 294 people living with HIV/AIDS, those with lower health literacy were no less likely to say that the doctor answered all their questions, but they were significantly less likely to say their doctors asked their opinion about treatment, or that they explained things so they could understand [35].
Collectively these 6 studies suggest that in relation to health service use, the differences between people with higher and lower levels of literacy are to be found less in terms of initial gaining of access to services and more in terms of the appropriateness of patterns of use and the securing of appropriate treatment.
Relationships with healthcare staff featured in an additional study of 157 parents of children aged one to four who had visited a well-child clinic. Those with a REALM score below 9th grade reported higher quality patient-provider relationships compared to those with scores of 9th grade or higher, through better family-centred care, helpfulness and confidence building [34].
Relationship between functional or health literacy and self-management of health problems
Eight studies examined associations between functional or health literacy and aspects of self-management of manifest health problems.
Four studies used quantitative techniques to assess adherence to medication and reported a relationship between lower functional literacy or health literacy and poorer adherence, [36–39] two of them in relation to parents administering medication to their children [38, 39]. A further two studies focused on parental ability to administer medication to their children [25, 40]. One study investigated women's compliance with follow up treatment [24].
One qualitative study of 25 people infected with HIV investigated the perceived clarity and level of difficulty of self-report HIV medication adherence measurement tools. Patients found it difficult to define adherence, had difficulty identifying medication and in recalling missed doses [41]. These difficulties have implications both for patients' adherence and for research that seeks to investigate this.
Studies of people living with HIV/AIDS found that in a sample of 381 people, those with lower health literacy were significantly more likely to miss at least one dose of medication over a 2 day period [36] and in another study of 87 HIV+ patients, that higher health literacy was associated with 95% or greater adherence over 3 months [37]. The two studies used different health literacy measures, Kalichman et al considering low health literacy to be less than 86% correct on the TOFHLA [36] and Graham et al comparing those with a REALM score below 9th grade level with those 9th grade or above [37]. Both studies considered relatively small variations in adherence but findings were consistent over the two widely different timescales.
In a retrospective cohort study of 150 parents of children with asthma, those with low health literacy had used rescue medication for their children more frequently and in greater amounts. They also had a significantly greater incidence of hospitalisation and days missed from school as well as an increase in emergency department visits which approached significance [38]. In a cross-sectional study of 78 children with type 1 diabetes, glycemic control was correlated with mothers' functional literacy as measured by NART scores [39].
In one cross-sectional study of 181 parents and caregivers, those with lower health literacy, measured by the TOFHLA, reported greater use of nonstandardised dosing instruments to give their children medication and this may impact on their adherence to the medication [40]. Another study of 543 parents found no association between parents' total scores on the REALM and their ability to administer their child's medication [25].
Compliance with recommended follow-up interventions was the focus of one study of 68 women who had had an abnormal pap smear. This study considered both physicians' subjective assessments of women's health literacy and more objective measurement using the REALM. Although there was a high level of agreement between the two, only subjective physician assessment of patient health literacy was a significant predictor of failure to follow up [24].
Knowledge of particular health risk or health conditions as a mediator between functional health literacy and health behaviours
Twelve studies focused on or included associations between functional or health literacy levels and knowledge about health conditions or treatment [25, 27, 35, 37, 38, 40–42] or health risks [22, 23, 30, 43]. Most, but not all of the studies, demonstrated lower knowledge of the various topics of interest in those with lower health literacy; two studies found that knowledge did not necessarily mediate behaviour [30] or adherence [37]. One study found that lack of knowledge was associated with behaviour likely to impact on adherence but adherence itself was not assessed [40].
Knowledge of HIV/AIDS in 372 patients offered HIV testing was poorer in those with lower health literacy [42]. One paper reported that in a sample of 294 people living with HIV/AIDS, those with lower health literacy were significantly more likely to believe that HIV transmission was less likely if anti-HIV medication was taken or if viral load was undetectable [35]. Poorer knowledge of their health status; [27, 35] poorer knowledge of medication [41] and more mistaken beliefs about their treatment [37] were also reported among patients with HIV/AIDS and lower health literacy.
One study of 181 parents and caregivers found that those with lower health literacy lacked knowledge about weight-based dosing and this was associated with the use of nonstandardised medication dosing instruments [40]. Another study of 150 parents reported that low health literacy was associated with less parental asthma related knowledge, characterised by a two point difference in a 20 point questionnaire [38].
A study of 600 pregnant women reported that those with lower reading levels had lower knowledge and less concern about the health effects of smoking on their unborn babies [30]. Other studies of 406 women in the community [43] and 505 female family planning clinic patients [22] found women with low health literacy were more likely to want to know more about birth control, [43] had lower knowledge of sexually transmitted infections [22] and were less likely than those with adequate health literacy to know about fertile times within their menstrual cycle [22, 43].
The 10 studies which found associations between knowledge of specific health issues and health literacy used 4 different measures and 8 different cut-off points for comparison, so although results suggest that knowledge is related to health literacy, (as would be expected, given the definition of health literacy), as with other relationships with health outcomes, it is unclear what aspects or levels of health literacy are most important.
Two studies found no association between knowledge of health issues and health literacy score. One study found that among 543 parents, knowledge of their child's diagnosis, medication name, purpose and instructions for use was not associated with health literacy score [25]. In this particular study, parents with lower health literacy considered their child more sick for the same degree of illness compared with those with higher literacy and this may have had an impact on parental management of their child's medication [25]. Another paper reported that among 400 women attending a family planning clinic, knowledge of contraception was generally poor, and although it tended to be better in those with higher health literacy, understanding of side effects of oral contraception and what to do about multiple missed pills was not associated with health literacy [23]. It is unclear why this particular study differs from the others with a similar focus examined here.
Two studies compared knowledge with related behaviour. One found that knowledge did not mediate smoking behaviour among pregnant women. Those with higher health literacy had greater knowledge but the trend was towards higher smoking in this group although the relationship was not significant [30]. Another study reported that some beliefs about medication did not mediate the relationship between health literacy and adherence and although beliefs about adherence norms were associated with adherence itself, this was independent of health literacy [37].
Emotional responses
A further two studies focused on emotional responses of patients, either to their actual condition [29] or to scenarios related to their condition [44]. Although emotional wellbeing can be considered as a contributory indicator of health status, we have reported these studies separately because it seems important not to obscure the possibility that the 'outcomes' they report might mediate other health status changes. Among 130 women at risk of developing cervical cancer, those with lower health literacy were more likely to have excessive levels of distress [29]. In a sample of 294 people living with HIV/AIDS, those with low health literacy had greater symptoms of affective depression but less evidence of negativistic thinking; they were more likely to endorse feelings of emotional distress, lower optimism and maladaptive coping when presented with a scenario of increased viral load [44].