Parent/caregiver health literacy among children with special health care needs: a systematic review of the literature

PubMed, OVID Medline and CINAHL were searched in October 2014 for peer-reviewed original research published prior to that date. The search strategy included: a combination of MeSH terms and key word searches including the terms: pediatric, child, parent, caregiver, health literacy, children with special health care needs, chronic disease. To ensure all relevant articles were captured, bibliographies of included articles were also reviewed for inclusion of any additional articles.

The PRISMA flow diagram detailing study selection and inclusion is found in Fig. 1. All studies were independently assessed for eligibility by two reviewers (JKM and LL). Once articles were identified from the database searches, duplicates were removed, then all remaining abstracts were reviewed for eligibility. Full-text articles were assessed for eligibility with only two articles excluded at this stage (because there was no assessment of parent/caregiver health literacy). Fifteen studies were included for thematic and contextual purposes, and of these, thirteen studies were included in the final analysis and systematic review. Two studies were eliminated because they were reviews and not empirical studies.

A data extraction instrument was developed based on elements important to both health literacy and outcomes pertinent to CSHCN. The following criteria was collected by two reviewers (JKM and LL): author/year, study design, disease of focus, sample ascertainment, assessment of health literacy (measure), percentage of population who had limited/low/marginal health literacy, outcome measures, notable findings (that were health-literacy specific) and the study quality score. To aid in interpretation of included studies, a quality review score was used to assign a study design level and ranking criteria based on the consistency and generalizability of reported elements (quality review criteria found in Table 1). This quality review criteria was used because it does not penalize studies for being observational, and a study can still yield high, good, or low quality based on the consistency, generalizability, and potential impact of the individual study on the systematic review. The review criteria utilizes a number (based on study design) and A, B, or C (high, good, low study quality). A low study quality ranking (“C”) does not mean a poorly conducted study, it only means that there is limited ability for the authors to use the findings of that study for overall interpretation in this systematic review. The reviewers assigned quality scores independently and if there was a discrepancy, met to discuss the study to agree on a consensus score between the two reviewers.

Thirteen studies meeting inclusion criteria were included in this review and study elements are described in Table 2 [15‐27]. All of the studies were conducted in various settings within the United States, with the majority occurring in pediatric clinics associated with academic medical settings. The majority of studies (n = 7) assessed parents with much of their samples comprised of mothers, but several studies also included the term ‘caregiver’ to be inclusive of non-parental primary caregivers caring for the child (including parents, guardians, grandparents, etc.). In terms of the health literacy assessment, six studies used the Rapid Estimate of Adult Literacy in Medicine (REALM) [28], six studies used either the full or shortened Test of Functional Health Literacy in Adults (TOFHLA or S-TOFHLA) [29], two studies used the Newest Vital Sign (NVS) [30], one study used the Parental Diabetes Numeracy Test (PDNT) [31] to assess numeracy. Several studies used a combination of measures. Two studies did not provide percentages of those in their sample with limited or marginal health literacy [17, 24] and in another study all caregivers were screened as having adequate health literacy [21], which limited the interpretation of the findings in these studies for specific CSHCN populations. For the remainder of the studies, the percentage of the sample with low, limited, or marginal health literacy ranged from 5.6 % [23] to 49 % [16]. Farber and colleagues [16] who cited 49 % of their population had limited health literacy, recruited caregivers of children with asthma from an inner-city ED. Wood and colleagues [27] cited 44.4 % with a possibility of limited health literacy, specifically recruited African-American caregivers of children with asthma. In both of these cases, the confluence of race, geography, and income could explain higher sample estimates of limited health literacy than the general population, but are extremely important for the context of specific sub-populations who are at a greater risk for exhibiting limited health literacy.

Sample size and study design were two of the components that were considered during the quality assessment period. The context of the current state of the science of health literacy and outcomes of CSHCN were also considered, as most studies (69 %) reported cross-sectional designs (n = 9) [16, 18‐21, 24‐27]. There were three cross-sectional studies that had adequate sample sizes and utilized appropriate statistical methods to control for confounding variables, that scored a IIIA on the quality scale due to their ability to generalize the findings to populations of CSHCN [19, 20, 25]. These three studies also described estimates of limited health literacy that were similar to rates described of the general population of adults. There was one study that utilized a retrospective cohort approach and it was also scored high on the quality scale (IIIA) because of the adequate sample size and sound statistical approach used to make conclusions on the relationship between health literacy and health service utilization [15]. There was one prospective observational study that had a low-moderate sample size where the authors’ were able to demonstrate an important relationship between health literacy and medication adherence, but they did not provide the percent of parents that had limited or marginal health literacy, therefore it scored lower as a IIIB [17].

There were also two randomized control trials (RCTs) that were a part of this analysis. Macy and colleagues [22] assessed health literacy of all enrollees, and randomized equally (not stratified based on health literacy) to either a video (intervention) or written materials (control). This study scored as a IB because the primary outcome was asthma knowledge (not a child health outcome related to asthma). Among parents with limited health literacy, asthma knowledge increased in both groups (p < 0.01) [22]. Porter and colleagues [23] assessed health literacy of all enrollees, and then randomized parents of children with attention deficit hyperactivity disorder (ADHD) into either a paper-based or computer-based data collection. Quality for this study was rated as IB primarily because of its lack of generalizability for CSHCN due to the outcome and primary findings (those with adequate health literacy report higher levels of sufficient and accurate clinical data) [23]. All of the included studies were from single-site recruitment, most often occurring from a pediatric clinic in academic health systems.

Of the 13 studies in this review, eight assessed parents or caregivers of children with asthma [15, 16, 18, 19, 22, 25‐27]. DeWalt and colleagues [15] focused on health service utilization among a parent/child with asthma dyad and demonstrated that children of parents with limited health literacy were more likely to use the ED (IRR 1.4, CI 0.97-2.0) and exhibited a statistically significant amount of hospitalizations (IRR 4.6, CI 1.8-12) and days missed from school (IRR 2.8, CI 2.3-3.4) after adjusting for asthma-related knowledge, medication use, and various socio-demographic factors (quality rating IIIA). Harrington and colleagues [19] focused on how provider perception of health literacy impacts asthma treatment recommendations (p = 0.001) and the way the information is delivered (p = 0.001) (quality rating IIIA). They also demonstrated that providers interpretation of health literacy had a low concordance with actual parental health literacy assessed through a validated measure [19]. Shone and colleagues [25] demonstrated that low health literacy was associated with perception of child’s health as fair/poor (OR 3.96, CI 2.4-6.4), greater parental worry about the condition (OR 1.85, CI 1.2-2.8), needing help to read the forms (OR 2.03, CI 1.3-3.1) and lower health related quality of life (β = −0.097; p = 0.047). They did not find any independent associations between health literacy and health care utilization (quality rating IIIA) [25].

There were three studies that focused on a population of parents/caregivers among children with Type 1 Diabetes Mellitus [20, 21, 24]. Hassan and colleagues [20] demonstrated a statistically significant relationship between caregiver health literacy and glycemic control as measured by HbA1c (p, 0.004; R² 0.23) (quality rating IIIA). Alternatively, Janisse and colleagues [21] were not able to demonstrate the same results and did not find a relationship between health literacy and HbA1c. This interpretation is limited by the fact that their entire caregiver population screened as having adequate health literacy at baseline. For a sub-group of adolescents who were on intensive insulin regimens, higher levels of caregiver health literacy were associated with greater adherence to the diabetes management scale (p < 0.01) (quality rating IIIC) [21]. Finally, Pulgaron and colleagues [24] demonstrated no relationship between health literacy and HbA1c but did find that those parents with lower levels of numeracy had worse HbA1c outcomes (r = −0.52, p <0.01,) (quality rating IIIB).

Two additional studies reported on relationships between health literacy and other clinical diagnoses. Freedman and colleagues [17] conducted a prospective observational study with a small sample size that reported decreased parental health literacy was associated with decreased medication adherence among children with glaucoma, but this interpretation was limited in that they did not report overall percentages of limited health literacy (quality rating IIIB). Porter and colleagues [23] initiated the RCT among parents of children with ADHD, but it was not focused on ADHD-specific outcomes (quality ranking IB).

Correlates of the impact of limited parental/caregiver health literacy on child health outcomes were synthesized across all studies (Table 3). The impact of limited health literacy was assessed for utilization, disease management, communication/knowledge transfer, and information appraisal outcomes. The impact of parental/caregiver health literacy on three of the four factors (utilization, communication and knowledge transfer and information appraisal) were deemed inconclusive because of the limited number of studies that were able to establish associations. Limited health literacy was determined to have an overall mediocre association on disease management because five of the included studies established correlations between limited health literacy and poorer disease management outcomes, while two studies described non-significant findings.

Most of the studies included in this review were observational studies, which leaves room for bias in interpretation of the results. To attempt to offset this bias, we utilized review criteria to help the authors’ interpret the strength of the study in terms of the generalizability and implications for CSHCN. Additionally, because we wanted to understand the impact of parental/caregiver health literacy on a wide range of child health outcomes we included all diagnoses that would classify the children as CSHCN. The heterogeneity in disease samples may limit the ability to interpret our results. Finally, we did not include child health outcomes related to healthy children, health knowledge or health promotion strategies. The vast majority of parent/caregiver health literacy research to-date includes healthy children so there may be important relationships we were not able to fully describe due to this exclusion.