This was a hospital-based, prospective, cross-sectional study. Health literacy levels were measured by using the short form of Test of Functional Health Literacy in Adults (S-TOFHLA) (16). Patient education was provided on the stroke unit at University of Florida Health–Jacksonville (UF Health–Jacksonville) by using standardized protocols, in compliance with Joint Commission specifications. The education sessions occurred as part of routine patient care. The subjects’ level of retention of poststroke care education was assessed on day of discharge by asking a series of 5 questions that were derived from the stroke performance measures on stroke education as required by the Joint Commission for certified stroke centers.

From August 2009 through June 2010, 101 eligible patients with acute ischemic stroke (AIS) who were consecutively admitted to the inpatient stroke unit at UF Health–Jacksonville provided written consent to participate in the study. UF Health–Jacksonville is a 750-bed tertiary-care hospital located north of downtown Jacksonville, Florida, with an AIS catchment area that includes the city of Jacksonville, 10 northeast Florida rural counties, and 2 southeast Georgia rural counties. This hospital has been certified as a primary stroke center by the Joint Commission since 2004.

Study subjects were patients older than 18 who were admitted to the hospital stroke unit with an AIS diagnosis. Subjects were identified and recruited by either the treating neurologist or a stroke unit nurse manager. Diagnosis of AIS was confirmed by a neurologist, and a computerized tomography (CT) or magnetic resonance imaging (MRI) scan confirmed the location and size of the brain infarction. Subjects who were unable to provide informed consent were excluded. Prisoners and subjects with diagnosed severe cognitive dysfunction due to dementia, global aphasia, or receptive aphasia were excluded. Once subjects were identified and their consent was obtained, subjects underwent literacy assessments. Of 189 subjects screened, 70 were excluded due to severe cognitive impairment. Eighteen eligible subjects refused to participate, 1 subject withdrew consent before completing the assessment, and 100 (84%) completed the study that constituted our study sample.

Demographic information, including age, race/ethnicity, annual household income, and education level, was collected through either patient interview or chart review. Residences of patients were coded and analyzed using zip codes, which have been aggregated into the city of Jacksonville’s 6 health zones. These zones, designated by the county health department and that reflect different racial/ethnic and other demographic characteristics of the residents, were based on regional boundaries used by city organizational entities, including the Duval County Public Schools, Jacksonville Sheriff’s Office, and the Community Planning Action Councils, as well as physical geographic boundaries and barriers (river, ocean). Growing recognition of the effect of community of residence (place matters) on health outcomes, particularly in the Jacksonville community (17), makes this demographic factor an important confounding variable.

The magnitude of neurological deficit at the time of hospital admission for each subject was delineated by the National Institutes of Health Stroke Scale (NIHSS) score (18). When not documented, scores were assigned by one of the study neurologists via retrospective review of the admission physical examination. Additionally, time between stroke onset and discharge from hospital and hospital length of stay in days was noted. Each study subject was administered a Mini-Mental Status Exam (MMSE), a test of cognitive impairment, after obtaining consent (19).

A trained research staff member administered the S-TOFHLA, which takes 8 minutes on average to administer (16). Briefly, it is a reliable, validated instrument used to assess a patient’s health literacy level (Cronbach’s α = 0.97; correlation with REALM [Rapid Estimate of Adult Literacy in Medicine]: r ² = 0.81) (16). The S-TOFHLA contains 36 reading comprehension items. The health-related passages on the S-TOFHLA use a modified cloze procedure, where every fifth to seventh word is omitted and subjects select the correct word from among a set of 4 options. Each selection was scored a 1 for correct or a 0 for incorrect. Scores were summed to create a total score. The passages contain information about an upper gastrointestinal tract X-ray procedure and the “rights and responsibilities” section from a Medicaid application; the passages have Gunning-Fox readability indices of fourth and tenth grade, respectively. The 36-point scale of the S-TOFHLA is divided into 3 categories of functional literacy: inadequate (0–16), marginal (17–22), and adequate (23–36).

Nurses on the stroke unit are trained in stroke patient education via completion of a self-learning module with yearly skill updates. Upon admission to the stroke unit, the registered nurse (RN) assessed the patient’s readiness for learning and developed an individualized stroke education plan. Stroke education was initiated as soon as possible during patients’ stay in the stroke unit. Educational sessions were designed to cover all of the topics that the Joint Commission mandates for a certified primary stroke center: personal risk factors for stroke, warning signs for stroke, activation of emergency medical services, need for follow-up after discharge, and medications prescribed for stroke prevention (20). The RN educator provided both verbal and written stroke education by using standardized American Stroke Association–approved handouts as learning tools during the patient’s hospital stay. Patients usually received 30 to 60 minutes of individualized stroke education that included their personal risk factors and lifestyle modification.

After the education session and before discharge a research staff member assessed the patients’ comprehension and retention of the information provided in the educational session by asking the 5 questions of the Stroke Patient Education Retention (SPER) score (Table 1). All retention assessments occurred on day of discharge. The SPER scores range from 0 to 10. Questions 1 and 2 were scored 0 = no correct content in response to the question, 1 = the patient provided some but not all of the correct content in response to the question, and 2 = patient provided all key content in response to the question. Questions 3–5 responses were scored 0 = no correct content in response to the question, and 2 = the patient provided the correct response to the question. A score of 10 indicates all correct responses for key content in the education sessions.

Statistical analyses and modeling were conducted using SAS version 9.3 software (SAS Institute, Inc, Cary, North Carolina). Continuous variables were described by using means, standard deviations (SDs), and medians. Categorical variables were presented as counts and percentages and analyzed using Pearson χ² or Fisher Exact tests, when appropriate. Univariate and multivariable analyses were used to identify factors independently associated with SPER. For univariate analysis, the differences in the SPER scores among the 3 S-TOFHLA groups were assessed using the nonparametric Kruskal–Wallis test and with Tukey’s adjustment for pairwise comparisons. For multivariate analysis, analysis of covariance (ANCOVA) was used to model the data and estimate the least squared means. A backward elimination variable procedure was conducted to eliminate weak predictors that contributed marginally in explaining variation in the SPER scores. Variables considered included S-TOFHLA score, sex, age (<60 years or ≥60 years), income (<$25,000 per year or ≥$25,000 per year), education (less than high school graduate or high school graduate or more), race (African American or other than African American), employment status (currently working, retired, unemployed, or other), NIHSS score, and MMSE score.

Variations in health literacy across the various patient demographic characteristics were not significant except for education level (P = .02) and MMSE score (P = .003) (Table 2). The mean education SPER score was 6.7 (SD, 2.0). Subjects with adequate health literacy outperformed the other 2 categories of health literacy on all 5 cognitive retention items (Table 3). Subjects who provided all key content in response to each of the SPER questions with marginal or adequate health literacy were compared with those with inadequate health literacy. Differences between having adequate health literacy and inadequate health literacy and between having marginal health literacy and inadequate literacy on the SPER mean scores were significant, but differences between adequate and marginal health literacy were not significant (Table 4).

Using multivariate analysis to examine the relationship of health literacy to education outcomes while considering other patient characteristics, a backward selection procedure was conducted to eliminate weak predictors that contributed marginally in explaining variation in the SPER scores using an ANCOVA regression modeling approach. The full model had STOFHLA score (P = .02), sex (P = .03), age (P = .95), income (P = .11), education (P = .43), race (P = .84), employment (P = .54), NIHSS score (P = .87), and MMSE score (P = .22) as predictors and the variables with the lower contribution (higher P value) to the model were excluded one by one. The significant predictors identified by the model were STOFHLA (P = .004), sex (P = .03), income (P = .007), and MMSE score (.04). The model explained 35% of the variation in SPER scores. Adjusting for all the other variables in the model and for multiple comparisons, subjects with marginal health literacy (adjusted mean SPER score = 7.59) and those with adequate health literacy (adjusted mean SPER score = 7.50) had better recall than subjects with inadequate health literacy (adjusted mean SPER score = 6.04) (P = .02 and P = .006 respectively). There was no significant difference estimated between subjects with adequate health literacy and subjects with marginal health literacy (P = .98) (Table 3).