Parental Health Literacy and Acute Care Utilization in Children With Medical Complexity

We conducted a cross-sectional study of 236 parents of CMC receiving primary care at an ∼400 bed free-standing midwestern United States children’s hospital. Parents of CMC receiving primary care at the study site participated in a research repository. The CMC included in the repository received primary care either through the study site’s primary care clinic dedicated to CMC or the site’s general pediatric primary care clinic. Eligibility criteria for the repository included child age <18, ≥1 CCCs,¹⁸  and an English-speaking parent because of the lack of survey instruments validated in other languages. Of 363 repository participants, 236 (65.0%) parents completed the Single Item Screener (SILS) and were included in the current study.¹⁹  The hospital institutional review board approved this study.

We measured parental health literacy using the validated SILS, chosen for its brevity and ability to be administered in written format.¹⁹  The SILS asks: “How often do you need to have someone help you when you read instructions, pamphlets, or other written material from your doctor or pharmacy?” Parents respond using a 5-point Likert scale from never to always. In our primary analysis, we used the traditional SILS categorization, wherein responses of “sometimes,” “often,” or “always” indicate inadequate health literacy and “never” or “rarely” indicate adequate health literacy.¹⁹  In post hoc sensitivity analyses, we restricted adequate health literacy to only responses of “never” (indicating the highest level of health literacy).

The main outcomes were annual rates and costs of ED visits and hospitalizations, collected from the Pediatric Health Information System (PHIS) database. Surveys and PHIS were linked by medical record number. The duration for measuring the outcomes varied based upon the child’s duration of care at the study site. The median duration (interquartile range [IQR]) for measuring the outcomes was 7.4 years (IQR: 5.1–11.3).

We summarized data as medians and IQR and compared groups using the χ² test and the Wilcoxon Rank Sum or Kruskal-Wallis test for continuous variables. We used generalized linear models with log link and time from first to last visit as offset, adjusting for study covariates. We included covariates with known utilization associations. We collected covariates from PHIS (insurance type, number of CCCs, technology dependence), the electronic health record (enrollment in CMC clinic), and the survey (marital status, language spoken at home, parental education, and household income as a percentage of the federal poverty limit). We calculated adjusted rate ratios (aRR) with 95% confidence intervals (95% CI). In a post hoc analysis, we stratified by whether the patient received care in the CMC clinic. We conducted analyses using SAS Enterprise Guide 7.1 (Cary, NC).

We performed 2 post hoc sensitivity analyses. For sensitivity analysis #1, we restricted adequate health literacy to only responses of “never” (indicating the highest level of health literacy). For sensitivity analysis #2, we stratified our analyses by whether the patient received care in the CMC clinic. The stratified analyses were performed for both our main adjusted regressions and for the regressions in sensitivity analysis #1, which restricts adequate health literacy to only responses of “never.”

Of the 236 CMC parents, ∼95% had adequate health literacy (Table 1). The study population had a median of 1.9 annual hospitalizations (IQR 1.1–3.4) and 2.9 annual ED visits (IQR 1.7–4.3) (Table 2). Parental health literacy was not associated with acute care utilization in either the bivariate analyses (Table 2) or the adjusted regressions (Fig 1).

Sensitivity analysis #1 – definition of adequate health literacy: Adequate health literacy was defined as only responses of “never” needing assistance. Health literacy was not associated with annual acute care utilization in bivariate analyses (Supplemental Table 3). In adjusted analyses, ED costs were 188% higher (aRR 2.88 [95% CI: 1.63–5.07]) and hospitalization costs were 126% higher (aRR 2.26 [95% CI: 1.49–3.44]) among CMC whose parents need any literacy assistance (Fig 1).

Sensitivity analysis #2 – stratification by clinic: When the main adjusted regressions were stratified by whether the patient received care in the CMC clinic, hospitalization costs of CMC seen in the general pediatric clinic were 86% lower among CMC whose parents had inadequate health literacy (aRR 0.14 [95% CI: 0.02–0.85], P = .03). No associations were found in the stratified analyses when the definition of adequate health literacy was restricted to “never” needing assistance.

Most parents of CMC had adequate health literacy. We found no statistically significant differences in acute utilization associated with parental health literacy. Inadequate health literacy was associated with increased ED and hospitalization costs after restricting “adequate health literacy” to those parents with the highest degree of health literacy.

We found no differences in acute care utilization associated with parental health literacy using the traditional SILS classification. Our findings are consistent with Morrison et al who found no association between inadequate parental health literacy and nonurgent ED visits in children with special health care needs.⁸  This contrasts with the majority of prior research that demonstrated the association of inadequate health literacy with increased acute care utilization.^2,5–9  Most prior research focused on the general pediatrics population or single or an unspecified number of chronic conditions.¹⁷  The lack of association in our study may reveal the importance of other health literacy-related predictors of acute care utilization in CMC. We measured parental health literacy; however, CMC often receive care from multiple caregivers (eg, private duty nurses, other family). Consequently, acute care utilization in CMC may reflect not only parental health literacy but also of additional caregivers (“collective heath literacy”).²⁰ 

The lack of association may also be related to limited power because of the unexpectedly low percentage of parents with inadequate health literacy. Our study population’s high educational attainment (75% attended college) and English fluency may account for this finding. Education and English proficiency are the strongest predictors of parental health literacy.¹  Few studies describe health literacy levels among parents of CMC.¹²  In studies of general pediatric patients, 30% of parents in the ED⁶  and 75% of parents with hospitalized children¹⁰  had low health literacy. In studies of children with special health care needs, inadequate parental health literacy ranged from 5.6% to 49%, but most focused on single chronic conditions such as asthma.^15,17  The use of SILS in the current study may contribute to the different levels of inadequate health literacy found in studies using different instruments. In studies using the SILS, inadequate health literacy ranged from 24.1%²¹  to 33.3%²²  among adults and 28.6% among parents.²³  The SILS has a sensitivity of 54%, specificity of 83%, and area under the receiver operating curve of 0.73.¹⁹  Therefore, the SILS potentially underidentified parents with inadequate health literacy. Our sensitivity analysis restricting the criteria for adequate health literacy may have partially offset this weakness. Health literacy in parents of CMC may also differ from other parents because of the inherent necessities of providing sophisticated care to their child since health literacy can change over time and with acquisition of disease-specific knowledge.^4,14 

In sensitivity analyses, acute utilization costs were significantly higher for CMC of parents who reported ever needing assistance reading health materials, despite a similar number of ED visits and hospitalizations. There was no association with ED and hospitalization rates. Further study is needed to investigate these differences, particularly since we used a nontraditional SILS categorization. If confirmed, it could indicate that parents of CMC with lower health literacy underestimate illness severity and seek care later in their child’s illness trajectory when they may require more costly care. The causal mechanisms between health literacy and increased costs in this population would need further investigation.

This study has several limitations. First, the small number of parents with inadequate health literacy reduced our power to detect associations with acute care utilization. Second, generalizability is limited because of the study population being limited to parents recruited from primary care clinics and who were English-speaking because of the lack of validated survey instruments in other languages. Third, although the SILS has been validated as a self-administered written instrument, the format may have self-selected participants with higher literacy skills, thereby introducing selection bias. Additionally, this instrument does not account for all health literacy skills (eg, listening, speaking, numeracy)^24,25  and is subject to social desirability bias. Finally, our data do not include the duration of each CCC, how long the guardian has been caring for the CMC, or other caregivers’ health literacy levels.

Most CMC parents had adequate health literacy. We found no association between parental health literacy and acute care utilization. We only found differences in associated costs in a sensitivity analysis. A better understanding of health literacy among parents of CMC could result in potential interventions to improve organizational health literacy, particularly among CMC. Further multicenter study in a varied population including non-English speaking parents of CMC is needed to comprehensively measure health literacy and its impact on CMC health.

We wish to acknowledge the parents who participated in the study and Dr Jacqueline Walker and Dr Timothy Ryan Smith who contributed to the interpretation of the data.