Intern Emerg Med

DOI 10.1007/s11739-017-1651-7

IM - ORIGINAL

Association between health literacy and hypertension

management in a Chinese community: a retrospective cohort
study
Di Shi1 • Jiangbo Li1 • Yong Wang1 • Si Wang1 • Kai Liu1 • Rufeng Shi1 •

Qiang Zhang1 • Xiaoping Chen1

Received: 12 January 2017 / Accepted: 8 March 2017

Ó SIMI 2017

Abstract Low health literacy is associated with poor life in hypertensive patients. Low health literacy increases
clinical outcomes. The relationship between literacy and the 10-year risk of ICVD and incidence of artery stiffness
blood pressure (BP) has been inconsistent. We investigated in hypertensive patients. Improving health literacy should
the determinants of health literacy and the potential rela- be considered an important part of the management of
tionship between health literacy and hypertension man- hypertension.
agement. We conducted a retrospective cohort trial of 360
hypertensive patients. Scale measurements, physical Keywords Health literacy
Hypertension
Home blood
examination, and laboratory tests were performed based on pressure
Blood pressure control
a standard protocol. To determine factors associated with
health literacy, multiple logistic regression analysis was
performed and the discriminatory power of the scale score Introduction
for hypertension control was assessed by the area under the
receiver operating curve. After adjusting for potential Health literacy is currently defined as ‘‘the degree to which
confounders, our findings show that the level of education, people are able to access, understand, appraise and com-
home blood pressure measurement, regular medication, municate information to engage with the demands of dif-
and systolic blood pressure are significantly associated with ferent health contexts in order to promote and maintain
health literacy. Moreover, patients with high health literacy good health across the life-course [1].’’ Low health literacy
have better hypertension control, a lower risk of ischemic is a significant problem in patients with chronic diseases.
cardiovascular disease (ICVD), lower brachial an- An important feature of low health literacy is that people
kle pulse wave velocity values, and better health-related find it difficult to read the drug labels, and to understand
quality of life. In addition, our study also demonstrates that the registration card of physical examinations and health
we can identify the health literacy level of hypertensive education information [1, 2]. Moreover, a higher degree of
patients using the Chinese Health Literacy Scale for education does not necessarily indicate a higher level of
Hypertension. At a cut-off value of 13.5, we predict that health literacy.
patients will achieve long-term hypertension control. A number of studies demonstrate that the health literacy
Adequate health literacy is a contributing factor to better level independently predicts patients’ health-related
blood pressure (BP) control and better perceived quality of knowledge, behaviors (such as medication compliance),
and self-reported health status [3–6]. These factors can
make great contributions to important clinical outcomes;
D. Shi and J. Li contributed to this work equally. however, the relationship between literacy and blood
pressure (BP) has been inconsistent. Several studies show
& Xiaoping Chen that higher health literacy is associated with better BP
[email protected]
control [5, 7], but two studies report that higher health
1
Department of Cardiology, West China Hospital, Sichuan literacy is associated with worse BP in veterans with
University, Chengdu 610041, China hypertension [8, 9]. The inconsistency of these results may

123
 Intern Emerg Med

be attributed to different models of healthcare delivery alcohol consumption level, exercise habits, history of dia-
systems. betes mellitus, history of hypertension, and use of anti-
To better understand determinants of health literacy and hypertensive drugs). The presence of co-morbidities was
the potential relationship between health literacy and based on self-reported records acknowledged by the study
hypertension management in the healthcare system of participants. The body mass index (BMI) was calculated
China, we adapted and created a Chinese health literacy using the following formula: weight/height2 (kg/m2).
scale for hypertension, which had no cultural differences Fasting plasma glucose (FPG), fasting serum total choles-
and language barriers for a Chinese-speaking population. terol (TC), low-density lipoprotein cholesterol (LDL-C),
We hypothesize that high health literacy is associated with high-density lipoprotein cholesterol (HDL-C), and
better BP control in hypertensive patients. triglycerides (TG) were included. Blood was drawn from
the antecubital vein in the morning after a 12-h fast and
subsequently analyzed using an automatic biochemical
Methods analyzer (Glamour 1800; MD, Muskegon, MI, USA).

Related definitions
Measures
Hypertension was defined as a systolic blood pressure
(SBP) of at least 140 mmHg, or a diastolic blood pressure Home blood pressure measurement (HBPM)
(DBP) of at least 90 mmHg, or currently taking anti-hy-
pertensive medications. Diabetes mellitus was defined as During the visit, all participants were shown how to self-
follows: (1) fasting plasma glucose C7.0 mmol/L; (2) a measure BP. All devices for use at home were evaluated
positive response to the question, ‘‘Has a doctor ever told and validated according to international standardized pro-
you that you have diabetes;?’’, or (3) current use of insulin tocols [10]. Four BP measures were obtained each day
or oral hypoglycemic agents. Smoking was defined as an [twice in the morning (\1 h after awakening and empty-
average cigarette consumption of at least 1 cigarette per ing the bladder) and twice in the evening (near bedtime)],
day. Alcohol intake was defined as an average intake of for at least 3–4 days, and preferably for seven consecutive
alcohol of at least 50 g/day. Physical activity was defined days as proposed by the European Society of Hypertension
as performing moderate intensity of 3–6 METs aerobic [11]. Instructions were given to have the two measures
physical activity (e.g., Walking briskly, ballroom dancing, performed after at least 5 min of rest in a sitting position
and slow cycling) 3 or more times per week (at least and separated by 2-min intervals. Participants had to record
30 min each time). Participants who previously reported a all BP readings in a logbook that consisted of three pages
diagnosis of hypertension or were taking drugs to lower (one for each day) with two parts: ‘‘morning measure 1–2’’
their BP were classified as ‘patients on target hypertension’ and ‘‘evening measure 1–2.’’ As in previous studies,
if the SBP was \140 mmHg or the DBP was \90 mmHg HBPM was considered successful when at least 12 mea-
with respect to treatment goals. Long-term hypertension sures of 18 were properly performed [12]. We collected
control was defined as reaching treatment goals at least patients’ logbooks containing the BP values during the past
9 months during the year. year.

Study sample Chinese health literacy scale for hypertension (CHLSH)

A retrospective study was conducted involving residents We designed the CHLSH to measure the health literacy of
living in the Jinyang community health center in Chengdu, patients with hypertension, which was based on four cog-
China. Patients with hypertension were recruited from nitive levels, according to the revised Bloom’s Taxonomy:
consecutive participants undergoing health check-ups in remembering; understanding; applying; and analyzing [13].
the community between September 2014 and May 2015 The ‘remembering’ subscale measures an individual’s
(n = 523). Three hundred sixty patients who understood ability to read aloud commonly used terms frequently seen
the correct measurement of home blood pressure were in printed educational leaflets or information sheets in
enrolled in the study. hypertensive management. First, we reviewed some pre-
Personal characteristics included physical examinations vious studies, the 2014 Commentary on Chinese Guideli-
(involved assessments of height, weight, and waist cir- nes for the Management of Hypertension in the
cumference), socio-demographic characteristics (e.g., age, Community and the 2010 Chinese Guidelines for the
gender, marital status, education level, work status, and Management of Hypertension, and collected questionnaires
monthly income), and clinical data (e.g., smoking status, from five hypertensive patients [14–18]. Then, we selected

123
Intern Emerg Med

80 items as initial items of ‘remembering.’ Next, the low corrected item-total correlation coefficients were
research team invited two professors of cardiology from screened out. We used Cronbach’s alpha to assess the
the West China Hospital, three general practitioners from internal consistency of each subscale and the CHLSH.
the community health center, two community nurses, one Items that all subjects answered correctly were dropped.
community pharmacist, and one clinical laboratory physi- Then, we calculated the Cronbach’s alpha for each
cian to score each item using the Delphi technique [19]. subscale again. We compared the alpha of the deleted
The scoring criteria were as follows: 5, the most relevant to item with the Cronbach’s alpha of the subscale. The
hypertension care and 1, the least relevant to hypertension. alpha of the deleted item represented the overall relia-
Items scoring 4 or 5 were included in the second round of bility coefficient for internal consistency of the subscale
rating. The experts then rated the items a second and third if an individual item was removed from the subscale. If
time. Finally, ten items that best represent the usual prac- the alpha was greater than the overall alpha of the cor-
tice in communication with patients were selected. responding subscale, we dropped the item and recalcu-
The other three subscales (understanding, applying, and lated the Cronbach’s alpha. We repeated the procedure
analyzing) were developed based on situations in daily until no item could be dropped. Confirmatory factor
hypertension management in which patients need to make analysis was conducted to assess the number of sub-
decisions. These situations included prescription, patient scales in the CHLSH. Pearson correlations were calcu-
education information, and instructions on drug use. These lated to test the relationships between the CHLSH and
three subscales included a total of 14 items, as follows: 7 the other two scales. Test–retest reliability was calcu-
questions for understanding; 3 questions for applying; and lated by Pearson correlation to test the reliability of the
4 questions for analyzing. The experts reviewed and rated scale during a period of 2 weeks.
the draft questions. The experts considered whether or not
the chosen labels and questions represented actual situa- Demographics of the Respondents
tions encountered by patients with hypertension in a Chi-
nese community, and whether or not the labels and One hundred forty participants completed the CHLSH and
questions were clearly stated. These items subsequently other scales. Then, we received all information again,
became the subscales of understanding, applying, and including the demographic information, from 132 respon-
analyzing. dents. The age range of participants was 31–88 years. Of
the participants, 52.3% were female. Approximately 16.7%
Power calculation and participants of respondents had only completed primary school, 32.6%
had completed secondary school (middle school, 32.6%;
In general, a reliable scale requires a Cronbach’s alpha high school, 21.2%), and 29.5% had completed higher
C0.70. The effect size was estimated to be 0.26 if the education. The course of hypertension and work status of
Cronbach’s alpha of our scale was 0.80 [20]. With this respondents was also collected, and is shown in Table 1.
effect size, an alpha level = 0.05, and a power = 0.8, the
calculated sample size was 109 [21]. General practitioners Construct validity assessment
who work in Jinyang Community Health Center of
Chengdu assisted with subject recruitment. We recruited There were 10 items in the remembering subscale, 7 in the
140 participants from Jinyang Community Health Center in understanding subscale, 3 in the applying subscale, and 4 in
Chengdu between March and June 2014. the analyzing subscale (24 items in total) in the originally
designed CHLSH (model 1 in Table 2). After the pilot
Pilot study and analysis study, we eliminated the six items that were answered
correctly by all respondents in the remembering subscale;
Sixty-eight subjects from a total of 140 participants vol- the Cronbach’s alpha was 0.882 (model 2 in Table 2). In
unteered for the pilot study, which aimed to refine the the other three subscales (understanding, applying, and
questionnaire, assess the time needed to complete the analyzing), no item was dropped and the Cronbach’s alpha
CHLSH, and assess the clarity of the draft questionnaire. values were 0.686, 0.771, and 0.691, respectively
According to the results of the pilot study, we made (Table 2). The Cronbach’s alpha of the final CHLSH scale
appropriate modifications about the items which were not with 18 items was 0.878. We carried out confirmatory
clearly stated. factor analysis after item elimination and derived a model
We assessed several psychometric properties of the with RMSEA = 0.009 and CFI = 0.982, indicating that
CHLSH, including item analysis, construct validity, the presence of the four subscales in the CHLSH. The
discriminative ability, test–retest reliability, and corre- factor loading of the four subscales of CHLSH is shown in
lation with the other two relevant scales. Items that had Table 3.

123
 Intern Emerg Med

Table 1 Demographics of study participants (n = 132) Table 3 Factor loading of the CHLSH subscales
Items n % Items Remembering Understanding Applying Analyzing

Sex Male 63 47.7 1 0.605

Female 69 52.3 2 0.853
Age(years) B50 32 24.2 3 0.364
51–60 29 21.9 4 0.703
61–70 37 28.0 5 0.430
C71 34 25.7 6 0.680
Marital status Married 120 90.9 7 0.804
Divorced 11 8.3 8 0.838
Unmarried 1 0.8 9 0.786
Education level Primary school 22 16.7 10 0.767
Middle school 43 32.6 11 0.651
High school 28 21.2 12 0.807
Higher education 39 29.5 13 0.801
Working state On job 47 35.6 14 0.653
Retiring 65 49.2 15 0.782
Jobless 20 15.2 16 0.892
Course of hypertension (years) B1 27 20.4 17 0.835
1–5 32 24.2 18 0.877
6–10 27 20.4 CHLSH Chinese Health Literacy Scale for Hypertension
11–20 31 23.5
C20 15 11.4

Understanding was significantly correlated with applying

(r = 0.77, p \ 0.001) and analyzing (r = 0.56, p \ 0.001),
Table 2 Item analysis of the CHLSH and its subscales
whereas applying was significantly correlated with analyz-
ing (r = 0.48, p \ 0.001). The CHLSH had high internal
Scale Model 1 Model 2 consistency (Cronbach’s alpha = 0.94), and test–retest
Items (n) a Items (n) a reliability during a 3-week period was good (r = 0.82;
p \ 0.001; Table 4).
CHLSH 24 18 0.878
Remembering 10 0.851 4 0.882
Correlation with two relevant scales
Understanding 7 0.686 7 0.686
Applying 3 0.771 3 0.771
We tested the correlations between CHLSH and two other
Analyzing 4 0.691 4 0.691
scales [the Health Literacy Scale for Chronic Care
CHLSH Chinese Health Literacy Scale for Hypertension (HLSCC) and the High Blood Pressure-Health Literacy
Scale (HBP-HLS)] [22, 23]. The first measure is a vali-
dated Chinese scale that assesses knowledge of chronic
We reviewed the CHLSH relationship with age and edu- diseases and ability and self-confidence in chronic disease
cational level to assess discriminative ability. management. The latter measure is used for evaluating
There was a negative correlation between the CHLSH and hypertension management interventions in the community
age (r = -0.28, p \ 0.001). The CHLSH scores were setting. The HLSCC had good internal consistency
compared among participants with different educational (Cronbach’s alpha = 0.91) and good test–retest reliability
levels, as follows: primary school (M = 34.3, SD = 9.7); (r = 0.77, p \ 0.01) during a period of 3 weeks in the
middle school (M = 37.5, SD = 8.3); high school Chinese population of Hong Kong. The HBP-HLS was
(M = 41.6, SD = 4.6); and higher education (M = 44.5, developed and validated among hypertensive Korean
SD = 3.2; F = 20.18, p \ 0.001). We also calculated the Americans, and shown to have good internal consistency
relationship between each of the four subscales. The (Kuder-Richardson-20 coefficient = 0.98).
remembering subscale was significantly correlated with The CHLSH had a significant positive correlation with
understanding (r = 0.82, p \ 0.001), applying (r = -0.53, the HBP-HLS (r = 0.67, p \ 0.001) and the HLSCD
p \ 0.05), and analyzing (r = 0.48, p \ 0.001). (r = 0.45, p \ 0.001). The remembering subscale had a

123
Intern Emerg Med

Table 4 Results of validity and reliability of the CHLSH C80%: 18 9 80% = 14.4) and a low health literacy
Discriminant validity M (SD) Coefficient
group [(LLG); the correct answer rate was B80%].
This grouping method was based on the China’s
Age r = -0.28** Ministry of Health used in a previous investigation
Education level F = 20.18** [24].
Primary school 34.3 (9.7)
Middle school 37.5 (8.3) Arterial stiffness measurements
High school 41.6 (4.6)
Higher education 44.5 (3.2) Brachial ankle pulse wave velocity (baPWV) was mea-
Reliability Cronbach’s a sured by an automatic device (VP1000; Colin Co., Ltd.,
Overall scale 0.94 Komaki, Japan) with an appropriate size cuff. For at least
Subscale 5 min before the test, each subject rested in the supine
Remembering 0.90 position in a room at 25 °C. Participants were instructed
Understanding 0.67 to refrain from consuming food, tea, and caffeine or
Applying 0.56 smoking for 3 h before the measurements, and also to
Analyzing 0.51 refrain from consuming alcohol for 24 h before the
Test–retest reliability r measurements. The device simultaneously recorded PWV,
Overall scale 0.82** BP, electrocardiogram records, and heart rate (HR).
Subscale Detailed information for baPWV measurements has been
Remembering 0.92** reported elsewhere [25].
Understanding 0.60**
Applying 0.43* Other tools
Analyzing 0.49**
We used the 10-year risk for ischemic cardiovascular
Pearson’s correlation was used to assess the relationship between
disease (ICVD) to evaluate the risk of cardiovascular
CHLSH and age. Analysis of variance was used to assess the rela-
tionship between CHLSH and education level events. ICVD is a cardiovascular risk prediction model
CHLSH Chinese Health Literacy Scale for Hypertension appropriate for the Chinese population [26]. In addition,
* p \ 0.05, ** p \ 0.001 we used the Medical Outcome Study Short-Form 36-Item
Health Survey (SF-36 scale) to evaluate the association
between health literacy and health-related quality of life
Table 5 Pearson correlations of the CHLSH and its subscales with (HRQOL) [27]. The SF-36 evaluates the following eight
other measures physical and mental health areas: physical functioning
HLSCC HBP-HLS (PF); physical role functioning (RP); bodily pain (BP);
general health (GH); vitality (VT); social role functioning
CHLSH 0.45** 0.67** (SF); emotional role functioning (RE); and mental health
Remembering 0.88** 0.76** (MH). Each of the eight areas was scored on a scale of
Understanding 0.63** 0.58** 0–100, in which a higher score indicated a better per-
Applying 0.32** 0.34** ceived quality of life.
Analyzing 0.27* 0.36**
HLSCC Health Literacy Scale for Chronic Care, HBP-HLS High Statistical analyses
Blood Pressure-Health Literacy Scale
* p \ 0.05, ** p \ 0.001 Continuous variables are expressed as the mean ± stan-
dard deviation (SD). Differences in characteristics
between patients with high and low health literacy were
correlation with HLSCD (r = 0.88, p \ 0.001) and HBP- tested by independent t tests for normally distributed
HLS (r = 0.76, p \ 0.001). The other three subscales variables and by the non-parametric Mann–Whitney or
(understanding, applying, and analyzing) also had a cor- Wilcoxon test for skewed variables. Categorical variables
relation with HLSCD and HBP-HLS (Table 5). are expressed as frequencies and percentages. Differences
Finally, a total of 18 items comprised the CHLSH between two groups were tested by a Chi-square test. To
questionnaire. The scale was self-reported records determine factors associated with health literacy, a mul-
acknowledged by study participants. According to the tiple logistic regression analysis was performed to deter-
CHLSH score, we divided patients into a high health mine the OR and 95% CI values, and the discriminatory
literacy group ([HLG]; the correct answer rate was power of the CHLSH score for hypertension control was

123
 Intern Emerg Med

assessed by the area under the receiver operating curve and DBP of the HLG had a U-shape with seasonal
(ROC). Covariates, including age, waist circumference changes (Fig. 1a, b), while we only observed a U-shape
(WC), BMI, FPG, TG, TC, HDL-C, and LDL-C, were with seasonal changes for the DBP of the LLG (Fig. 1b).
fitted as continuous variables, and gender, alcohol intake, A comparison of BP control was made across the four
smoking, regular physical exercise, work status, monthly quartiles of health literacy (Fig. 2). With an increased
income, diabetes, HBPM, hypertension, and anti-hyper- CHLSH score, the rate of hypertension control became
tensive drug use, were fitted as categorical variables. The higher (p \ 0.001).
point representing the largest sum of sensitivity and
specificity on the ROC was calculated. The score of ROC curves analyses
health literacy was split into four quartiles. A comparison
of BP control was conducted across the four quartiles. The areas under the ROC curves were 0.717 (95% CI
The least significant difference (LSD) test for pairwise 0.649–0.786; p \ 0.001) in predicting the target blood
comparisons was used when needed. SPSS 19.0 and pressure achieved for the entire year, and 0.703 (95% CI
MedCalc 11.0 software were used. Statistical significance 0.633–0.772; p \ 0.001) in predicting the target blood
was defined as a p \ 0.05. pressure achieved in at least three quarters of the year
(Table 9). At a cut-off value of 13.5, the highest sensitiv-
ities and specificities were achieved in both of the afore-
Results mentioned situations (Fig. 3a, b).

Factors associated with health literacy Association between health literacy and ICVD

According to the CHLSH score, we divided patients into Compared with the HLG, the LLG had a much higher risk
HLG and LLG. Compared with those in the LLG, HLG for ICVD (9.7 vs. 6.6%, p \ 0.001; Fig. 4).
subjects had more males, a higher education level, fewer
unemployed, higher income, lower SBP, lower TC, and Association between health literacy and baPWV
lower TG, and were more compliant with regular medica-
tion, HBPM, and regular exercise (all p \ 0.05, Table 6). The average baPWV values of the HLG were lower than
the LLG (1605 ± 337 vs. 1946 ± 471 cm/s; p \ 0.05;
Multiple logistic regression analysis Fig. 5).

After adjustment for confounders (age, WC, BMI, HR, Association between health literacy and HRQOL
alcohol intake, smoking, regular exercise, diabetes, mar-
ital status, FPG, TG, TC, HDL-C, and LDL-C), education The total SF-36 score in the HLG was higher than the LLG
level, work status, HBPM, regular medication, and SBP (593.6 ± 121.8 vs. 523.5 ± 156.7; p \ 0.05).
were still significantly associated with health literacy
(Table 7). Only a middle school education and retired
from work were significantly associated with health lit- Discussion
eracy. Thus, subjects in whom their education level was
middle school, whose work status was retired, who did The goal of our study is to identify factors associated with
HBPM, and who took regular medication had better health literacy and to investigate the association between
health literacy. A higher SBP indicated lower health health literacy and hypertension management. After
literacy. adjusting for potential confounders, we show that educa-
tion level, HBPM, work status, regular medication use, and
Association between health literacy and long-term SBP are significantly associated with health literacy.
hypertension control Moreover, patients with high health literacy have better BP
control, a lower risk of ICVD, lower baPWV values, and
Fifty-two patients who could not provide complete data better HRQOL. Our study also demonstrates that we can
were excluded from this analysis. The average BP values identify the health literacy level of hypertensive patients
are shown in Table 8. The SBP of the two groups had using the CHLSH. At a cut-off value of 13.5, we can
significant differences at 1–3, 4–6, 7–9 months, and the predict that patients will do well in long-term BP control.
entire year. The DBP of the two groups had no signifi- A higher degree of education does not necessarily mean a
cant differences in any quarters of the year. The SBP higher level of health literacy. Our results show that education

123
Intern Emerg Med

Table 6 Analysis of factors

HLG (n = 161) LLG (n = 199) p
associated with health literacy
Male (n, %) 82 (51.0) 77 (38.7) \0.05
Age (years) 66.6 ± 6.2 66.6 ± 8.3 NS
Marital status (n, %)
Married 160 (99.4) 196 (98.5) NS
Divorce 1 (0.6) 3 (1.5) NS
Unmarried 0 0
Education level (n, %)
Primary school 16 (9.9) 68 (34.2) \0.001
Middle school 42 (26.1) 66 (33.2) \0.001
High school 42 (26.1) 44 (22.1) \0.001
Higher education 61 (37.9) 21 (10.6) \0.001
Work status (n, %)
Employed 29 (18.0) 28 (14.1) \0.001
Retired 109 (67.7) 120 (60.3) \0.05
Unemployed 25 (15.5) 50 (25.1) \0.001
Monthly income (n, %)
\1000 38 (23.6) 50 (25.1) NS
1000-3000 96 (59.6) 90 (47.1) \0.001
[3000 52 (32.3) 34 (17.8) \0.001
Course of hypertension (years) 8.7 ± 7.3 9.8 ± 10.3 NS
Regular medication (n, %)
Yes 138 (85.7) 94 (47.2) \0.001
No 10 (6.2) 41 (20.6) \0.001
No medication 13 (8.1) 64 (32.2) \0.001
HBPM (n, %) 109 (67.7) 55 (27.6) \0.001
Smoking (n, %) 25 (15.5) 27 (13.6) NS
Alcohol intake (n, %) 19 (11.8) 29 (14.6) NS
Regular exercise (n, %) 89 (55.3) 60 (30.2) \0.001
BMI (kg/m2) 24.3 ± 3.0 24.3 ± 3.2 NS
WC (cm) 84.7 ± 9.9 84.3 ± 10.1 NS
Diabetes (n, %) 37 (23.0) 65 (32.7) NS
SBP (mm Hg) 131.0 ± 12.8 144.6 ± 14.4 \0.001
DBP (mm Hg) 75.0 ± 5.8 75.3 ± 6.2 NS
Heart rate (beats/min) 73.1 ± 9.6 73.7 ± 10.2 NS
TC (mmol/L) 4.9 ± 1.1 5.2 ± 1.0 \0.05
HDL-C (mmol/L) 1.5 ± 3.4 1.5 ± 3.4 NS
LDL-C (mmol/L) 2.5 ± 0.7 2.5 ± 0.7 NS
TG (mmol/L) 1.5 ± 0.6 1.7 ± 0.8 \0.05
FPG (mmol/L) 6.1 ± 2.4 6.6 ± 2.9 NS
HLG high health literacy group, LLG low health literacy group, WC Waist circumference, BMI body mass
index, HBPM home blood pressure measurement, FPG fasting plasma glucose, TG triglyceride, TC total
cholesterol, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, NS no
statistical significance

level, work status, HBPM, regular medication use, and SBP are possible that we could not demonstrate a statistically signifi-
significantly associated with health literacy, which is in cant difference in the relatively narrow range of age. In addi-
agreement with another study [16]; however, we do not find a tion, only retirement is significantly associated with health
relationship between age and health literacy. The possi- literacy, which suggests that the retired have more time to care
ble reason is that most patients we recruited are older. It is about their own health.

123
 Intern Emerg Med

Table 7 Multiple logistic

B SE Wald P OR
regression analysis of factors
associated with health literacy Sex 0.133 0.482 0.082 NS 1.225
Age -0.008 0.047 0.363 NS 0.966
Education level (compared with primary school) 9.705 \0.05
Middle school -1.891 0.687 9.623 \0.05 0.194
High school -1.371 0.457 4.782 NS 0.442
Higher education -0.908 0.544 2.786 NS 0.403
Work status (compared with unemployed) 8.376 \0.05
Employed 1.752 0.562 3.644 NS 0.769
Retired 1.964 0.535 5.773 \0.05 1.064
Monthly income (compared with \1000) 2.696 NS
1000–3000 -1.273 0.813 5.273 NS 1.543
[3000 -0.879 0.668 3.273 NS 1.033
Course of hypertension 0.083 0.231 0.080 NS 1.023
HBPM (yes or no) 1.301 0.379 11.535 \0.05 3.769
Medication (yes or no) 7.473 \0.05
Regular medication 1.497 622 5.443 \0.05 3.618
Irregular medication 0.085 0.956 0.042 NS 1.116
SBP (mm Hg) -0.094 0.053 12.571 \0.05 0.977
Constant 6.877 2.845 6.583 \0.05 1072
B partial regression coefficients, SE standard error, Wald Chi-square value, OR odds ratios, NS no statistical
significance

Table 8 Average BP values of

HLG (n = 170) LLG (n = 138)
the two groups in the past year
SBP (mm Hg) DBP (mm Hg) SBP (mm Hg) DBP (mm Hg)

1– 3 month 136.7 ± 14.5 75.0 ± 7.9 154.7 ± 16.1* 75.8 ± 9.1

4–6 month 127.5 ± 12.3§ 74.5 ± 9.2 140.0 ± 16.8*§ 73.1 ± 9.1
7–9 month 123.1 ± 14.1§ 74.9 ± 10.1 144.0 ± 19.1*§ 75.3 ± 9.6
10–12 month 136.3 ± 18.5 75.7 ± 9.0 139.8 ± 18.4§ 76.6 ± 10.4
The entire year 131.0 ± 12.8 75.0 ± 5.8 144.6 ± 14.4* 75.3 ± 6.2
HLG high health literacy group, LLG Low health literacy group
* Comparison between two groups (p \ 0.05)
§
Compared with 1–3 month in one group (p \ 0.05)

There are no studies that investigate the association HLG is closer to normal physiologic conditions. Patients
between long-term BP control and health literacy. Our with low health literacy are lacking hypertension knowl-
results demonstrate that patients in the HLG have better edge and have poor compliance with medication [28, 29],
control of SBP than patients in the LLG during three which might explain the relationship between health liter-
quarters of the year. In addition, the SBP and DBP of the acy and hypertension control. In addition, the cut-off val-
HLG have a U-shape with seasonal changes, while only the ues for ROC curves are basically in agreement with the
DBP of the LLG has a U-shape with seasonal changes. This score that we initially set, which indicates that the CHLSH
phenomenon indicates that blood pressure variability of the is effective for identifying the level of health literacy.

123
Intern Emerg Med

We also find that patients with low health literacy have a

higher risk for ICVD and higher values of baPWV. Martin
et al. [30] assessed the 10-year risk of cardiovascular dis-
ease using the Framingham risk score, and reports similar
results. Because of a lack of hypertension knowledge and
poor compliance with medication, patients with low health
literacy may be accompanied by an unhealthy lifestyle and
more cardiovascular risk factors. Although carotid-
femoral pulse wave velocity (cfPWV) is one of the most
reliable methods for measuring arterial elasticity, baPWV
is closely correlated with cfPWV; the aortic PWV assessed
invasively and the presence of CV risk factors is linked
with an elevated baPWV value [31–33]. Moreover, the
recent AHA recommendation states that baPWV is useful
in cardiovascular outcome predictions in Asian populations
[34]. According to our results, health literacy is a new risk
factor of hypertension for ICVD and arterial stiffness
prediction.
HRQOL evaluations include comprehensive health-re-
lated and disease-specific assessments. Comprehensive
HRQOL evaluations include the SF-36, which was devel-
oped in the United States and is used worldwide. The SF-36
allows patients to quantitatively self-evaluate their physical
and emotional quality of life with relative ease. We exam-
ined the role of health literacy as a predictor of HRQOL that
has not been previously studied in hypertensive patients. In
the current study, the higher levels of health literacy posi-
tively influence HRQOL scores in hypertensive patients. In
addition, given that general health perception is among the
important independent determinants of health service uti-
lization [35], a higher HRQOL is likely to increase individual
health behaviors and self-care management for hypertensive
patients. Thus, high health literacy reflects a better ability for
health service utilization. Ultimately, health literacy might
Fig. 1 Bar Chart of SBP (a) and DBP (b) of two groups in the past
lead to improved HRQOL among hypertensive patients.
year (*p \ 0.001)
Some studies involving patients with various diseases report
that limited health literacy influences health outcomes, such
as medication cost, knowledge, self-care management,
adherence to medication, and HRQOL [36, 37]. The finding
is consistent with prior studies regarding the impact of health
literacy on HRQOL [37, 38].

Limitations

Our study has several limitations. First, the small sample

size might impact the study’s findings, and the results of a
single-site study may demonstrate atypical outputs. The
findings of this study might not be generalized to other
populations. Second, we could not provide patients with
Fig. 2 Hypertension control of health literacy in four quartiles The the same device for use at home. We evaluated and vali-
BP control rate increased with the improvement of health literacy. dated all devices according to international standardized
(*p \ 0.001 compared with the former group)

123
 Intern Emerg Med

Table 9 Areas under the ROC

AUC (95% CI) p Cut-off Sensitivity Specificity
curve for different periods of a
year used to predict BP control The whole year 0.717 (0.649–0.786) \0.001 13.5 0.680 0.755
using CHLSH
At least three quarters of the year 0.703 (0.633–0.772) \0.001 13.5 0.700 0.683

Fig. 4 Association between health literacy and ICVD

Fig. 5 Association between health literacy and baPWV

Fig. 3 Receiver operating characteristic curve analysis for predicting

target blood pressure achieved in the entire year (a) and in at least
three quarters of the year (b). Asterisk the cut-off point risk of ICVD and incidence of artery stiffness in hyper-
tensive patients. Improving health literacy should be con-
protocols. Third, because the average age of our sample is sidered as an important part of the management of
older, we cannot extend our findings to a general popula- hypertension.
tion. Finally, there is a potential selection bias to this ret-
rospective study. Acknowledgements The authors thank Dr. Xiao’s team at Jinyang
Community Health Center in Chengdu, China for their contributions
to this study.

Conclusion Compliance with ethical standards

Conflict of interest The authors declare that they have no competing

Adequate health literacy is a contributing factor to better interests (financial, political, personal, religious, ideological, aca-
BP control and better perceived quality of life in hyper- demic, intellectual, or any other) to declare in relation to this
tensive patients. Low health literacy increases the 10-year manuscript.

123
Intern Emerg Med

Statement of human and animal rights The research protocol was pressure-focused health literacy scale. Patient Educ Couns
reviewed by the Medical Ethics committee of West China Hospital, 87(2):165–170
Sichuan University. The study protocol conforms with the ethical 15. Leung AY, Lou VW, Cheung MK, Chan SS, Chi I (2013)
guidelines of Declaration of Helsinki. Development and validation of Chinese health literacy scale for
diabetes. J Clin Nurs 22(15–16):2090–2099
Informed consent Informed written consent was obtained from 16. Leung AYM, Cheung MKT, Lou VWQ, Chan FHW, Ho CKY,
patients before enrollment. The Medical Ethics Committee of West Do TL, Chi I (2013) Development and validation of the Chinese
China Hospital Sichuan University approved all procedures. Health Literacy Scale for chronic care. J Health Commun
18(sup1):205–222
17. Writing group of Commentary on Chinese guidelines for the
management of hypertension in the Community (2015) Com-
References mentary on Chinese guidelines for the management of hyper-
tension in the Community (2014 revised edition). Chin J
1. Canadian Public Health Association (2008) (Accessed September Hyperten. doi:10.3760/cma.j.issn.1674-0815.2015.01.004
30, 2015) A Vision for a Health Literate Canada: Report of the 18. Writing group of (2010) Chinese guidelines for the management
Expert Panel on Health Literacy. http://physicalactivitystrategy. of hypertension (2011) 2010 Chinese guidelines for the man-
ca/pdfs/Health_ Literacy.pdf agement of hypertension. Chin J Hyperten. doi:10.16439/j.cnki.
2. Williams MV, Baker DW, Parker RM, Nurss JR (1998) Rela- 1673-7245.2011.08.009
tionship of functional health literacy to patients’ knowledge of 19. Sinha IP, Smyth RL, Williamson PR (2011) Using the Delphi
their chronic disease: a study of patients with hypertension and technique to determine which outcomes to measure in clinical
diabetes. Arch Intern Med 158(2):166–172 trials: recommendations for the future based on a systematic
3. Gazmararian JA, Williams MV, Peel J, Baker DW (2003) Health review of existing studies. PLoS Med 8(1):e1000393
literacy and knowledge of chronic disease. Patient Educ Couns 20. van Zyl JM, Neudecker H, Nel DG (2000) On the distribution of
51(3):267–275 the maximum likelihood estimator of Cronbach’s alpha. Psy-
4. Williams MV, Baker DW, Parker RM, Nurss JR (1998) Rela- chometrika 65(3):271–280
tionship of functional health literacy to patients’ knowledge of 21. Length R (2003) Java applets for power and sample size. Divi-
their chronic disease: a study of patients with hypertension and sion of Mathematical Sciences, the College of Liberal Arts at The
diabetes. Arch Intern Med 158(2):166–172 University of Iowa. http://homepage.divms.uiowa.edu/*rlenth/
5. Pandit AU, Tang JW, Bailey SC, Davis TC, Bocchini MV, Per- Power/. Retrieved 7 Oct
sell SD, Wolf MS (2009) Education, literacy, and health: Medi- 22. Kim MT, Song HJ, Han HR, Song Y, Nam S, Nguyen TH, Kim
ating effects on hypertension knowledge and control. Patient KB (2012) Development and validation of the high blood pres-
Educ Couns 75(3):381–385 sure-focused health literacy scale. Patient Educ Couns
6. Gazmararian JA, Kripalani S, Miller MJ, Echt KV, Ren J, Rask K 87(2):165–170
(2006) Factors associated with medication refill adherence in 23. Leung AYM, Cheung MKT, Lou VWQ, Chan FHW, Ho CKY,
cardiovascular-related diseases: a focus on health literacy. J Gen Do TL, Chi I (2013) Development and validation of the Chinese
Intern Med 21(12):1215–1221 Health Literacy Scale for chronic care. J Health Commun
7. Rothman RL, DeWalt DA, Malone R, Bryant B, Shintani A, 18(sup1):205–222
Crigler B, Pignone M (2004) Influence of patient literacy on the 24. China’s Ministry of Health (2009) The first survey report of
effectiveness of a primary care–based diabetes disease manage- Chinese residents’ health literacy. 2011
ment program. JAMA 292(14):1711–1716 25. Luo F, Wang Y, Wang X, Sun K, Zhou X, Hui R (2009) A
8. Willens DE, Kripalani S, Schildcrout JS, Cawthon C, Wallston K, functional variant of NEDD4L is associated with hypertension,
Mion LC, Roumie CL (2013) Association of brief health literacy antihypertensive response, and orthostatic hypotension. Hyper-
screening and blood pressure in primary care. Journal of health tension 54(4):796–801
communication 18(sup1):129–142 26. Wu Y, Liu X, Li X, Li Y, Zhao L, Chen Z, Liu K (2006) Esti-
9. Powers BJ, Olsen MK, Oddone EZ, Thorpe CT, Bosworth HB mation of 10-year risk of fatal and nonfatal ischemic cardiovas-
(2008) Literacy and blood pressure–do healthcare systems influ- cular diseases in Chinese adults. Circulation 114(21):2217–2225
ence this relationship? A cross-sectional study. BMC Health Serv 27. Ware JE, Kosinski M, Dewey JE, Gandek B (2000) SF-36 health
Res 8(1):1 survey: manual and interpretation guide. Quality Metric Inc
10. O’brien E, Atkins N, Stergiou G, Karpettas N, Parati G, Asmar R, 28. Pandit AU, Tang JW, Bailey SC, Davis TC, Bocchini MV, Per-
Working Group on Blood Pressure Monitoring of the European sell SD, Wolf MS (2009) Education, literacy, and health: medi-
Society of Hypertension (2010) European Society of Hyperten- ating effects on hypertension knowledge and control. Patient
sion International Protocol revision 2010 for the validation of Educ Couns 75(3):381–385
blood pressure measuring devices in adults. Blood pressure 29. Gazmararian JA, Williams MV, Peel J, Baker DW (2003) Health
monitoring 15(1):23–38 literacy and knowledge of chronic disease. Patient Educ Couns
11. Council ES, Redon J, Narkiewicz K, Nilsson PM, Burnier M, 51(3):267–275
Viigimaa M, Tsioufis C (2013) 2013 ESH/ESC Guidelines for the 30. Martin LT, Schonlau M, Haas A, Derose KP, Rudd R, Loucks
management of arterial hypertension. Eur Heart J 34:2159–2219 EB, Buka SL (2011) Literacy skills and calculated 10-year risk of
12. Stergiou GS, Skeva II, Zourbaki AS, Mountokalakis TD (1998) coronary heart disease. J Gen Intern Med 26(1):45–50
Self-monitoring of blood pressure at home: how many measure- 31. Yamashina A, Tomiyama H, Takeda K, Tsuda H, Tomio ARAI,
ments are needed? J Hypertens 16(6):725–731 Hirose K, Yamamoto Y (2002) Validity, reproducibility, and
13. Anderson LW, Krathwohl DR, Airasian PW et al (2001) A tax- clinical significance of noninvasive brachial-ankle pulse wave
onomy for learning, teaching, and assessing: a revision of velocity measurement. Hypertens Res 25(3):359–364
bloom’s taxonomy of educational objectives. Addison Wesley 32. Yamashina A, Tomiyama H, Arai T, Hirose KI, Koji Y, Hir-
Longman, New York ayama Y, Hori S (2003) Brachial-ankle pulse wave velocity as a
14. Kim MT, Song HJ, Han HR, Song Y, Nam S, Nguyen TH, Kim marker of atherosclerotic vascular damage and cardiovascular
KB (2012) Development and validation of the high blood risk. Hypertens Res 26(8):615–622

123
 Intern Emerg Med

33. Tomiyama H, Yamashina A (2010) Non-invasive vascular literacy and heart failure knowledge, self-efficacy, and self-care
function tests. Circ J 74(1):24–33 adherence. Res Soc Adm Pharm 10(2):378–386
34. Townsend RR, Wilkinson IB, Schiffrin EL, Avolio AP, Chirinos 37. Halverson JL, Martinez-Donate AP, Palta M, Leal T, Lubner S,
JA, Cockcroft JR, Najjar S (2015) Recommendations for Walsh MC, Trentham-Dietz A (2015) Health literacy and health-
improving and standardizing vascular research on arterial stiff- related quality of life among a population-based sample of cancer
ness. Hypertension 66(3):698–722 patients. J Health Commun 20(11):1320–1329
35. Lam CL, Fong DY, Lauder IJ, Lam TPD (2002) The effect of 38. Macabasco-O’Connell A, DeWalt DA, Broucksou KA, Hawk V,
health-related quality of life (HRQOL) on health service utilisa- Baker DW, Schillinger D, Weinberger M (2011) Relationship
tion of a Chinese population. Soc Sci Med 55(9):1635–1646 between literacy, knowledge, self-care behaviors, and heart fail-
36. Chen AM, Yehle KS, Albert NM, Ferraro KF, Mason HL, ure-related quality of life among patients with heart failure. J Gen
Murawski MM, Plake KS (2014) Relationships between health Intern Med 26(9):979–986

123