Health Literacy Associated With Blood Pressure but not Other Cardiovascular Disease Risk Factors Among Dialysis Patients

Abstract

BACKGROUND

Limited health literacy is prevalent and has been linked to adverse patient outcomes. We examined the relationship between health literacy and cardiovascular disease (CVD) risk factors, including blood pressure (BP) parameters, lipids, waist-to-hip ratio (WHR), body mass index (BMI), and tobacco utilization among dialysis patients.

METHODS

We conducted a cross-sectional study of 72 participants in a prospective cohort study of vascular calcification in newly initiated dialysis patients. Health literacy was assessed using the Short Test of Functional Health Literacy (S-TOFHLA) in Adults. The study population was dichotomized into those with and without adequate literacy. Linear and logistic regression analyses were used to predict continuous and dichotomous cardiovascular risk factor variables, respectively.

RESULTS

Twenty-one percent had limited health literacy. Compared to limited health literacy, adequate health literacy was associated with lower BP parameters in multivariable analyses (systolic blood pressure (SBP): β −16.8, s.e. 6.7, P = 0.01; diastolic blood pressure (DBP): β −13.8, s.e. 4.1, P = 0.001; mean arterial pressure (MAP): β −14.8, s.e. 4.6, P = 0.002). Health literacy was not a statistically significant predictor of low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides, WHR, BMI, or tobacco utilization.

CONCLUSIONS

Limited health literacy is common in individuals on dialysis. Individuals with adequate health literacy skills had DBP readings that were on average 13.8 mm Hg lower and MAP that was 14.8 mm Hg lower than those with inadequate health literacy. Prospective studies to elucidate if improvements in health literacy skills will lead to improvement in BP control are needed.

End-stage renal disease (ESRD) is associated with diminished quality of life and reduced life expectancy.¹ The reasons for suboptimal outcomes in ESRD are multifactorial and may be attributable to the high comorbid disease burden, as well as the inherent complexity of ESRD management. Addressing nutrition, mineral–bone disease, acid–base homeostasis, electrolyte homeostasis, fluid balance, anemia, cardiovascular risk mitigation, renal replacement therapy, and other domains of ESRD care involves intricate dietary, medication, and fluid prescriptions. Frequent patient interactions with and navigation through a complicated healthcare system are an added barrier to optimal ESRD management.²

At the individual level, patient self-management and partnership with healthcare providers is a critical component of ESRD care, but may be hampered by limited health literacy. The Institute of Medicine defines health literacy as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.”³ Limited health literacy is exceedingly common in the general population, with estimates that 90 million adults in the Unites States have insufficient literacy to process and comprehend health information.⁴ Studies in dialysis patients have confirmed limited health literacy and numeracy in this subset of the population.^5,6

Health literacy significantly impacts healthcare utilization and patient outcomes. Higher health literacy is associated with improved glycemic control, blood pressure (BP) control, and satisfaction with healthcare services.^7,8 Limited health literacy, on the other hand, is associated with diminished disease-related knowledge and medication knowledge in non-chronic kidney disease (CKD) populations.^9–11 Individuals with limited health literacy are at increased risk of suboptimal chronic disease management and increased hospitalizations.^12,13 Emerging data regarding health literacy and CKD suggests that hemodialysis patients with limited health literacy are less likely to be referred for renal transplantation and experience higher mortality rates.^6,14

Little is known about the mechanisms by which health literacy may contribute to increased mortality among CKD patients. Traditional cardiovascular risk factors are common in individuals with CKD and may be affected by health literacy status. High systolic blood pressure (SBP) values among younger individuals are associated with increased mortality, regardless of race or diabetes status.¹⁵ Waist circumference as well as body mass index (BMI) are associated with increased mortality risk in adults with CKD.¹⁶ Lowering low-density lipoprotein (LDL) is associated with improved survival in individuals with CKD or on dialysis.¹⁷ Smoking is a lifestyle factor associated with increased mortality in new to dialysis individuals.¹⁸

We hypothesized that newly initiated dialysis patients with limited health literacy would have an adverse cardiovascular risk factor profile. We examined the association between health literacy and risk factors for cardiovascular disease (CVD), including BP, lipid profile, waist-to-hip ratio (WHR), BMI, and tobacco utilization.

METHODS

We conducted a cross-sectional study of newly initiated dialysis patients participating in the Dialysis Heart and Bone Study, a prospective cohort study of risk factors associated with vascular calcification. Participants were excluded if they had a history of coronary revascularization or cardiac devices (pace-makers, defibrillators) as well as if they weighed >350 lbs due to technical reasons. Participants attended a baseline research visit within 4 months of dialysis initiation and a follow-up visit at least 12 months following the initial baseline visit. During the baseline visit, a detailed medical history and demographic data were collected, including self-reported race, age, and lifestyle behaviors. A limited physical examination was performed and standard anthropometric measures were obtained. Blood was collected during each research visit. The study protocol was approved by the institutional review board of the University of Pennsylvania, and the study was conducted in accordance with accepted ethical principles of human subject research.

Newly initiated dialysis patients were recruited from the renal transplant evaluation clinics of the University of Pennsylvania Health System (UPHS), UPHS-affiliated dialysis units and non-UPHS-affiliated dialysis units in the Philadelphia metropolitan area from 2008 until 2010. Seventy-one participants completed the health literacy questionnaire and were included in the analysis.

Health literacy was measured using the Short Test of Functional Health Literacy (S-TOFHLA), a validated measure of health literacy that includes an assessment of reading comprehension, as well as numeracy. S-TOFHLA is scored on a scale of 0–100. A score ≥67 corresponds to adequate health literacy, a score of 54–66 delineates marginal health literacy, and a score ≤53 represents inadequate health literacy.^19,20 The number of participants with marginal (n = 5) and inadequate literacy (n = 10) were small, and in subsequent analyses, we combine marginal and inadequate health literacy categories into a single group referred to as limited health literacy. S-TOFHLA was administered by trained study personnel during the first research visit beginning in 2009. Participants received the S-TOFHLA during their baseline (n = 28) or follow-up visit (n = 44). Some participants were unable to return for follow-up visit due to organ transplantation (n = 7), death (n = 8), or coronary revascularization (n = 3). Educational attainment was categorized into three strata: less than high school diploma, high school diploma or trade school certification, and beyond high school. Income was categorized as follows: <$20,000, $20,000 to <$40,000, $40,000 to <$60,000, and ≥$60,000 per annum. CVD was defined as prior history of cerebrovascular disease, peripheral vascular disease, or history of coronary artery disease without revascularization procedures.

We evaluated the following cardiovascular risk factors: SBP, diastolic blood pressure (DBP), mean arterial BP (MAP), LDL, high-density lipoprotein (HDL), triglycerides, WHR, BMI, and active tobacco use. BP readings were measured three times during the research visit corresponding to S-TOFHLA administration, and the average of all three measurements were used in analyses. A lipid profile, including LDL, HDL, and triglycerides, were measured during the initial research visit. Lipid analyses were conducted in the clinical laboratories of the University of Pennsylvania using the Roche Hitachi 912 Automated Clinically analyzer and reagents from Roche Diagnostics (Indianapolis, IN).

Continuous and categorical variables were summarized using means and proportions, respectively. The distribution of continuous and categorical variables across health literacy categories was analyzed using t- test and χ², respectively. Predictors of health literacy were assessed using the covariates age, gender, race, income, and education in a multivariable linear regression model. To identify whether health literacy associates with the cardiovascular risk factors of interest, health literacy was included in multivariable linear regression models for continuous cardiovascular risk factors, and logistic regression models for smoking. The covariates age, gender, race, income, and educational attainment were selected a priori, based on known associations in the literature with health literacy.^4,21 Linear regression models for LDL, HDL, and triglycerides also include adjustment for statin use and the research visit number, to take into account the temporal separation between health literacy and lipid assessments that occurred only for this parameter in some of the participants. HDL additionally was adjusted for alcohol use. The statistical package STATA version 10.1 was used for all analyses (Stata, College Station, TX). Given the multiple testing, we adjusted the level of significance to a P value ≤ 0.008 using the Holmes correction.

We also evaluated the impact of adequate health literacy on reaching BP suggested goals (SBP > or ≤140; DBP > or ≤90, and MAP > or ≤92) using logistic regression.

RESULTS

Approximately 68% of the study population was comprised of men with an average age of 51.6 years. Fifty-eight percent of study participants self-identified as African American (AA), 37% as white, 1% as Asian, and 4% did not identify as belonging to any of the aforementioned groups. Hypertension was exceedingly common in this cohort, affecting 94% of individuals. Diabetes and CVD were present in 46 and 29.2% of participants, respectively. The vast majority of the population (87%) received in-center hemodialysis, 10% were on peritoneal dialysis, and 3% were on home hemodialysis.

The income of the population was distributed as follows: 27% reported an income of <$20,000, 24% reported an income between $20,000 and <$40,000, 23% reported an income between $40,000 and <$60,000, and 27% reported an income ≥$60,000. The majority of the cohort (57%) was well-educated with some college or other advanced degree while 30% obtained either a high school diploma or trade school certification, and 13% had less than a high school diploma. There were no baseline differences in demographics, lifestyle factors, lipid profile, BP, and anthropometric measures in participants who completed the S-TOFHLA and those who did not.

The majority of participants had adequate health literacy while 21% had limited health literacy. Age, gender, and the presence of comorbidities were comparable in the adequate and limited health literacy categories (Table 1). Race, education level, and income were statistically different across health literacy categories in unadjusted analyses. After adjusting for demographic variables, age, race, and income were statistically significant predictors of health literacy (Table 2). There was a significant difference between the distribution of education achieved by AA compared to non-AA (elementary/high school/beyond college 17.1, 39, 43.9% for AA vs. 6.7, 20, 73.3% for non-AA, P = 0.046). There was no significant statistical difference between the distribution of income achieved by AA compared to non-AA (<20k, 20–40k, 40–60k, >60k, 36.6, 24.4, 17.1, 22% for AA vs. 13.3, 23.3, 30, 33.3 for non-AA, P = 0.13). When we excluded race (AA vs. non-AA) and income from the model, education did not reach statistical significance as a categorical variable (elementary-reference, high school P value 0.45, and college P value 0.32) or as an ordinal variable (P value 0.07). When we stratified by race and included both education and income in the model, education reached statistical significance in non-AA as a categorical variable (elementary-reference, high school P value 0.13, and college P value 0.003) and as an ordinal variable (P value 0.001). Diabetes was not a predictor for health literacy.

Table 1 |.

Baseline characteristics of the study population stratified by literacy status

Variable	Limited health literacy, N = 15	Adequate health literacy, N = 57	P value
Age (years)	52.9 (10.3)	50.1 (13.2)	0.44
Gender
Male	80.0	63.2	0.22
Female	20.0	36.8
Race
African American	100.0	49.1	<0.001
Other	0	50.9
Systolic blood pressure (mm Hg)	144.9 (20.5)	137.3 (17.5)	0.15
Diastolic blood pressure (mm Hg)	84.1 (15.4)	76.6 (10.2)	0.03
Mean arterial pressure (mm Hg)	104.4 (15.2)	96.8 (11.5)	0.04
Comorbidities
Diabetes	53.3	44.6	0.55
Hypertension	100	92.7	0.30
Coronary artery disease	13.3	12.5	0.93
Any CVD	40	26.3	0.30
Income categories
<$20,000	53.3	16.7	0.02
$20,000 to <$40,000	20.0	25.9
$40,000 to <$60,000	20.0	24.1
≥$60,000	6.7	33.3
Education
< High school	13.3	13.0	0.01
High school	60.0	20.4
Beyond high school	26.7	66.7
Body mass index (kg/m2)	28.2 (5.7)	29.2 (5.4)	0.53
Waist-to-hip ratio	0.94 (0.09)	0.97 (0.16)	0.47
HDL (mg/dl)	54.3 (25.4)	42.7 (15.2)	0.03
LDL (mg/dl)	145.7 (41.0)	160.5 (39.5)	0.21
Triglycerides (mg/dl)	117.6 (58.3)	160.3 (78.7)	0.06
Current tobacco use	26.7	10.7	0.12
Any tobacco use	80	44.6	0.02
No alcohol use	35.7	28.6	0.75
Alcohol use >1–2 drinks/week	14.3	8.9	0.62
Dialysis modality upon initiation
In-center hemodialysis	86.7	86.8	0.04
Peritoneal dialysis	13.3	9.4
Home hemodialysis	0	3.8

CVD, cardiovascular disease; HDL, high-density lipoprotein; LDL, low-density lipoprotein.

Table 2 |.

Predictors of health literacy score

Outcome variable	Linear regression model
	Coefficient	P value
Age	−0.46 (−0.79 to −0.12)	0.01
Female	3.67 (−5.07 to 12.4)	0.41
African American	−12.5 (−21.3 to −3.8)	0.006
Income	5.77 (1.9 to 9.6)	0.004
Education	1.1 (−5.1 to 7.2)	0.73

In unadjusted analyses, there was a statistically significant association between health literacy and both DBP and MAP (DBP: β −12.4, s.e. 3.4, P < 0.001; MAP: β −12.3, s.e. 3.8, P = 0.002) but not SBP (β −11.9, s.e. 5.6, P = 0.04) (Table 3). After adjusting for both demographic and socioeconomic variables, the relationship between health literacy and diastolic and MAP remained statistically significant and increased in magnitude (β −13.8, s.e. 4.1, P = 0.001 and β −14.8, s.e. 4.6, P = 0.002, respectively) while the relationship with SBP reached borderline statistical significance (β −16.8, s.e. 6.7, P = 0.01). There was no statistically significant association between health literacy and either LDL or triglycerides. In contrast, adequate literacy, compared to limited health literacy, was associated with lower HDL levels (β −13.1, s.e. 6.2, P = 0.04) but it did not reach statistical significance. WHR and BMI were comparable across health literacy groups, and health literacy was not associated with WHR, BMI, or tobacco utilization in adjusted analyses (Table 3). In a sensitivity analysis, diabetes was also added to the models. However, diabetes did not modify the association of the cardiovascular risk factors with health literacy.

Table 3 |.

Regression coefficients of the association between health literacy and several cardiovascular disease risk factors

Outcome variable	Univariable model, β (s.e.)	P value	Model 1a, β (s.e.)	P value	Model 2b, β (s.e.)	P value
Systolic blood pressure	−11.9 (5.6)	0.04	−12.8 (6.5)	0.06	−16.8 (6.7)	0.01
Diastolic blood pressure	−12.4 (3.4)	<0.001	−12.9 (3.9)	0.002	−13.8 (4.1)	0.001
Mean arterial pressure	−12.3 (3.8)	0.002	−12.8 (4.4)	0.005	−14.8 (4.6)	0.002
Triglyceridesc	41.6 (21.5)	0.06	39.4 (24.2)	0.11	20.3 (24.2)	0.41
Low-density lipoproteinc	13.4 (11.6)	0.25	14.8 (12.7)	0.25	9.7 (13.1)	0.46
High-density lipoproteinc	−12.0 (5.2)	0.02	−12.7 (6.0)	0.04	−13.1 (6.2)	0.04
Body mass index	0.56 (1.6)	0.73	1.4 (1.9)	0.47	0.75 (2.0)	0.71
Waist-to-hip ratio	0.02 (0.03)	0.57	0.04 (0.03)	0.20	0.03 (0.03)	0.33
	OR (95% CI)	P value	OR (95% CI)	P value	OR (95% CI)	P value
Current tobacco use	0.20 (0.05–0.79)	0.02	0.40 (0.08–2.1)	0.28	0.40 (0.07–2.22)	0.29

HDL was also adjusted for alcohol use (<1–2 drinks/week vs. >1–2 drinks/week).

β, Regression coefficient describing the association between adequate health literacy compared to limited health literacy for corresponding outcome variable; CI, confidence interval;

HDL, high-density lipoprotein; OR, odds ratio.

^aModel 1-adjusted for age, gender, and race.

^bModel 2-adjusted for age, gender, race, income, and educational attainment.

^cIn addition to the covariates delineated by models 1 and 2, model also adjusted for visit number and statin use.

Participants with adequate health literacy were more likely to reach all suggested BP parameters goal, but it only reached statistical significance for DBP (unadjusted odds ratio (OR) 6.8 (95% confidence interval (CI) 1.7–27.1), P = 0.007 and after full adjustment OR 8.75 (95% CI 1.25–61.1), P = 0.03). For reaching SBP goals, the unadjusted OR for individuals with adequate health literacy compared to those without was 3.13 (0.73–13.4), P = 0.12 and was attenuated to 2.5 (0.78–8.02), P = 0.12 after full adjustment. Similarly, for achieving MAP goal, the unadjusted OR for individuals with adequate health literacy compared to those without was 3.49 (0.57–21.2), P = 0.18 and was attenuated to 3.61 (0.74–17.6), P = 0.11 after full adjustment.

DISCUSSION

We found that limited health literacy is common in individuals on dialysis and is associated with age, race, and income. There was a statistically significant relationship between health literacy and BP parameters among newly initiated dialysis patients. Individuals with adequate health literacy had DBP values that were ~13.8 mm Hg lower and MAP values that were 14.8 mm Hg lower than individuals with limited health literacy, respectively. However, health literacy was not associated with LDL, HDL, triglycerides, WHR, BMI, or history of tobacco utilization.

The prevalence of health literacy in our study population was within the range of health literacy rates reported in other studies. Grubbs and colleagues evaluated health literacy using the S-TOFHLA in a cohort of 62 hemodialysis patients and reported inadequate literacy in one-third of the study population.⁶ Green et al. used an alternate health literacy instrument, the rapid estimate of adult literacy in medicine, and identified limited health literacy in 16% of their population.²² Similar to the findings of Green and colleagues, race was a statistically significant predictor of health literacy.²² Age and income were statistically significant predictors of health literacy in our study which differs from the findings of Green et al.

Health literacy was associated with BP parameters. Studies regarding health literacy and BP are limited, but a cross-sectional study by Winkleby and colleagues found that higher educational attainment, which is highly associated with health literacy, predicted a lower prevalence of hypertension, after adjusting for age, gender, income, and occupation in a non-ESRD cohort.²³ Our findings further characterize the relationship between health literacy and BP in an ESRD cohort. In non-CKD cohorts, higher BP is associated with increased cardiovascular mortality and adverse outcomes. In dialysis populations, a U-shaped relationship exists between BP and mortality such that both low and high BPs are associated with increased mortality.^24,25 Evidence suggests that low SBP may confer increased mortality risk as the result of factors such as malnutrition, while elevated SBP remains a consistent risk factor for adverse cardiovascular events.²⁶ Our findings suggest that the presence of limited literacy, which is associated with higher BP, may confer increased cardiovascular risk in dialysis patients.

One possible explanation of the observed association between health literacy and BP may relate to the critical role of patient self-management in the achievement of BP control. Medication adherence is an essential component of BP management, and limited health literacy has been linked to poor cardiovascular medication adherence.²⁷ Individuals with adequate health literacy may be better poised to participate in volume management, mediated by differences in disease-specific knowledge. Knowledge has been linked to accurate medication reconciliation and permanent vascular access use in hemodialysis patients.^9,12,28,29 Interventions to achieve adequate BP control among ESRD patients, which has been reported to be as low as 30%, may be more successful if health literacy is considered.³⁰

Components of the lipid profile and body habitus were not associated with health literacy. The link between lipids and atherosclerosis in ESRD is further influenced by nutritional status.³¹ WHR correlates with cardiovascular events in the general population and may be a better predictor of cardiovascular events than BMI, a finding that was replicated in a study of peritoneal dialysis patients.^32–34 WHR was not associated with health literacy in our study.³⁵ We did not find an association of health literacy with tobacco utilization or BMI which confirms findings reported by Wolf and colleagues.³⁶

Health literacy may have important implications with regard to the elimination of racial and socioeconomic disparities in health. Health literacy differed across racial and ethnic groups such that blacks had lower health literacy, a finding that has been demonstrated in other studies.³⁷ Racial disparities in BP control among individuals with late CKD have been documented.³⁸ In our study, race was not a statistically significant predictor of various BP measures. To the extent that health literacy impacts determinants of BP, health literacy may impact racial disparities in BP control.

Interpretation of these findings is subject to limitations. This study is cross-sectional and therefore causality cannot be determined. Our cohort is small and there is concern for multiple testing. Therefore, our results should be interpreted with caution. Health literacy, measured at a single point in time, may fluctuate and is influenced by other factors such as cognitive ability, which was not assessed in this study. In addition to fluctuations in health literacy over time, health literacy may also have a variable effect on cardiovascular risk factors over time. Another important consideration is selection bias. Our study population was limited to a single geographical area and may not be generalizable to other populations. Our participants had a slightly lower prevalence of CVD than is expected for the US population. In addition, the death of the 8 participants who did not have S-TOFHLA administered could be informative. Participants of this study also agreed to participate in a larger epidemiological study that required multiple visits and therefore represent a distinct cohort. We did not measure medication adherence in this study or dietary measures such as sodium intake. There was no home BP monitoring.

Cardiovascular mortality remains a leading cause of mortality in the ESRD population. Individuals with ESRD are vulnerable to atherosclerotic CVD, heart failure, and sudden cardiac death. In conclusion, this report suggests that the adverse outcomes found in individuals with limited health literacy may, in part, be mediated by suboptimal BP management. These findings advance our knowledge regarding the crucial role health literacy may play in the management of hypertension. Additional research is needed to explore if interventions to improve health literacy translate into changes in BP levels.