Study on the Status of Health Literacy and its Relationship with Chronic Diseases in Poor Areas

We conducted an epidemiological survey of health literacy and chronic diseases among people in poverty-stricken areas in China to understand the current status of health literacy among poor individuals and the epidemic characteristics of chronic diseases. In June 2018, multistage stratied random sampling was used to conduct a face-to-face questionnaire survey of 1,700 residents (response rate: 97.71%) in 7 national-level poverty-stricken counties in Henan Province, China. The questionnaire assessed social demographics, health literacy, health-related behaviours, and the status of chronic diseases (such as diabetes, hypertension, and heart disease). Structural equation models were used to examine the relationships among health literacy, health-related behaviours, and chronic diseases. Statistical analysis was performed using IBM SPSS Statistics 21.0 and Mplus 7.0.

1 Background "Health literacy", proposed in 1974, is playing an increasingly important role in the eld of public health [1] .
The European Health Literacy Survey Alliance (HLS-EU) de nes health literacy (HL) as a skillset that requires people to have the knowledge, motivation and su cient ability and health promotion to improve their quality of life when they make decisions about health care and disease prevention in their daily life and when they obtain, understand, evaluate and implement health information [2] . Additionally, the Institute of Medicine (IOM) de nes health literacy as "An individual's ability to access, process, and understand basic health information and services needed to make sound health-related decisions and follow treatment guidelines." This study used the de nition given by the World Health Organization, i.e., the extent to which individuals have the cognitive and social skills necessary to obtain, understand and apply basic health information and services to promote and maintain good health [3][4] . Today, health literacy remains inadequate in many countries and regions, especially among the most vulnerable socioeconomic groups [5][6][7] . More than half of the German population has reported di culty processing health-related information, and with a limited awareness of health, the prevalence of limited health literacy was particularly high among individuals with low self-assessed social status [6] . Approximately 80 million American adults have limited health literacy, which puts them at risk of poor health, with the rate of limited health literacy being higher among poor individuals [7] .
Most young people from poor areas in China go to work in economically developed areas as oating populations, and the health literacy of the elderly individuals and left behind children in rural areas is generally low [8][9] . Low health literacy results in excessive risky behaviour and poor health outcomes, including poor health awareness, an increased incidence of non-communicable diseases (NCDs), poor labelling of intermediate diseases and the inadequate use of preventive measures in health services [1,10−13] . As such, improving health literacy has evolved as a public health goal, and many countries have taken measures to improve the level of health literacy of their citizens [14,15] .
In the Outline of the "Healthy China 2030 Blueprint", the Chinese government has set a goal of increasing health literacy to 30% by 2030. Henan Province has a large population but has poor medical and health conditions [16] . In 2018, there were 53 poverty-stricken counties in this province, including 31 national-level poverty-stricken counties, and more than 1 million residents in poverty-stricken areas suffered from diseases, accounting for 50.69% of the province's population of poor individuals. In this survey, seven state-level poverty-stricken counties in Henan Province were surveyed to assess the health literacy level, health-related behaviours and prevalence of NCDs among local residents and to explore the relationship among the these three factors to improve the health literacy of residents in poor areas, reduce the incidence of chronic diseases and improve the quality of life of low-income groups in poor areas through increasing health knowledge.

Participants and sampling
This cross-sectional study was conducted in June 2018. The survey was administered in 7 national-level poverty-stricken counties in Henan Province. All non-community residents who were aged 15 ~ 69 years and had been living continuously in these countries for more than 6 months were eligible to participate.
According to the sample size calculation formula: , urban and rural areas are divided into two layers. Here, deff = 2, the health literacy level of Henan Province in 2016 (weighted rate 7.8%, p = 7.8%). Based on this formula, the minimum effective sample size was calculated as 1380. Considering the possible allowable error and missing data, the sample size was increased by 20%, so we determined that the number of people surveyed should be 1,700.
We used multistage strati ed random sampling to recruit the study population (Table 1). A total of 1700 residents were surveyed face-to-face by trained investigators; of these, we obtained and analysed 1661 (97.71%) valid responses.

Measure
The questionnaire assessed social demographics, health literacy, health-related behaviours and NCDs.

Health literacy
This survey used the "National Resident Health Literacy Questionnaire" issued by the National Health Commission of the People's Republic of China. The content includes three aspects of literacy health, i.e., knowledge (23 items, 29 points), health behaviour (15 items, 20 points), and health skills (12 items, 16 points), and 6 types of health problems, i.e., scienti c and healthy ideas (7 items, 9 points), prevention of infectious diseases (6 items, 7 points), chronic disease prevention (10 items, 14 points), safety and rst aid (10 items, 14 points), basic medical information (12 items, 14 points), and health information (5 items, 7 points). A total of 56 items were included, and the maximum total score was 65 points. The scale had a Cronbach's alpha coe cient of 0.875.
Judgement questions and single-choice questions are awarded one point for correct answers, and wrong answers are not scored; for multiple-choice questions, two points are given for correct answers. Wrong choices, multiple choices, or fewer choices are not scored. According to the scoring standard of the Chinese health literacy monitoring survey, when the correct score of the health literacy assessment reaches more than 80% of the full score, it is determined that the survey subject has su cient health literacy. This standard applies to the determination of all three aspects of health literacy and all six types of health problems.

Health-related behaviours
According to research by Alden Yuanhong Lai [17] , Machi Suka [18] , Yong-Bing Liu [19] and other scholars, smoking and drinking were included in health-related behaviours herein. We assessed the frequency of smoking behaviour on a scale of 1 (never) to 4 (frequently). We assessed the frequency of drinking behaviour on a scale of 1 (never) to 5 (frequently). The sum of these two items (2 ~ 9 points) was used as an indicator of health-related behaviours: higher scores indicated a greater behavioural risk.

Health status
The assessment of health status included whether an individual was suffering from NCDs, how many NCDs the individual had, self-reported of health status, etc. Whether an individual had a chronic disease was represented by a binary classi cation variable (0 = no, 1 = yes), and the self-assessed health status was expressed by the item "How do you think your health is?" The response options were 3 = good, 2 = moderate, and 1 = poor.

Research hypothesis
Based on the related models combined with other health literacy studies [1-2,17−19] , it is assumed that the level of health literacy directly affects health-related behaviours and the chronic disease prevalence of residents in poor areas, and the health-related behaviours of residents in poor areas directly affect the prevalence of NCDs. A model of the relationships among health literacy, health-related behaviours and NCDs mediated by healthy behaviours was established (Fig. 1).

Statistical analysis
First, we used Epi Data 3.1 to enter the questionnaire data(double-entry method is adopted, and check after entry is completed) and then used SPSS Statistics 21.0 to perform descriptive statistical analysis (frequency, percentages, mean and standard deviation) and single-factor statistical analysis (chi-square test). A test level of α = 0.05 and P < 0.05 were considered statistically signi cant. Finally, Mplus 7.0 was used to build the structural equation model (SEM), the maximum likelihood estimation method was used to test the hypothesis, and the total scores of health literacy and health-related behaviours were used as potential variables. To test whether the estimated model was suitable for the data, we 4 t indices and their criteria [20] : root mean square approximation error (RMSEA) < 0.05; Tucker-Lewis index (TLI) and comparative t index (CFI) values > 0.90; standard root mean square residual (SRMR) < 0.05; χ 2 /df < 3. If all index values are close to or higher than these critical values, the model is considered to be a good t for the data.

Participants and health literacy status
In this survey, the mean age of the participants was 50.90 ± 12.97 years, with 56.53% women, 50.09% rural, and 97.17% Han respondents. The majority of participants had a primary school education level and below (49.91%), followed by junior high school (37.87%), senior high school (8.49%) and college or above (3.73%). Most participants were married (90.43%), and most were farmers and workers (89.22%).  Table 2).
The average scores of key variables in the whole sample: The mean scores for health literacy and healthrelated behaviours were 31.95 ± 11.81 and 3.06 ± 1.99, respectively. The mean value for the number of NCDs was 0.04 ± 0.20 (Table 3).
The results of this survey showed that 574 (34.56%) participants had one or more NCDs, of which 397 (69.16%) had 1 type of chronic disease, 129 (22.47%) had 2 types of chronic disease, and 48 (8.36%) had  Hypothesis test results show that the best-tting model demonstrated that health-related behaviour was a partial mediator of the relationship between health literacy and NCDs (root mean square error of approximation = 0.042 and comparative t index = 0.940). Speci cally, this model showed that less smoking and drinking behaviours were associated with higher health literacy (β = -0.002, 95% con dence interval (CI): -0.003~-0.001). The health literacy score is directly proportional to the number of NCDs (β = 0.041, 95% CI: 0.033 ~ 0.049). An analysis of the mediating pathways shows that health literacy and the number of NCDs are mediated by health-related behaviours, which supports our previous assumption (Fig. 3, Table 4).

Discussion
Having su cient HL might be more important than ever before because people are expected to participate in health decisions and to take responsibility for their own health despite more complicated health problems and the need to navigate a more complex health system [21][22][23] . In this survey, only 4.15% (31.95 ± 11.81 points on average) of the residents in poor counties achieved su cient health literacy. However, in the same period, the adequate health literacy of Chinese residents was 17.06%, and that of Henan residents was 14.38% [24] . Hence, the level of health literacy of poor county residents in Henan Province is far lower than that of the whole country and other regions in the same province. With respect to health knowledge, health behaviour and health skills, su cient health knowledge was most common (10.36%), followed by health behaviour (5.30%) and health skills (5.54%). This nding indicates that residents in poor areas do not engage in su ciently healthy behaviours, and more health guidance is needed. Studies in Europe and other countries have also found that the proportion of certain groups with limited health literacy is higher than that of the general population, especially individuals with less money [25] . The single-factor χ 2 test indicated that the effects of 8 factors, including gender, age, education, occupation, monthly income, locality census register, chronic disease, and self-assessment of health status, on health literacy are statistically signi cant. Among the participants, those with lower health literacy levels were more likely to be male, older, poorer, less educated, and suffering from chronic diseases. This is consistent with the research results of Liu L [1] , Verney SP [9] , S rensen K [25] and other scholars.   [19] . As such, the two unhealthy behaviours (smoking and drinking) were not higher in poorer areas than in the general population.
NCDs (such as cardiovascular disease, cancer, and diabetes) are the major source of the global burden of disease and mortality, and their share of the total global disease burden has steadily increased from 44% in 1990 to 61% in 2016 [11,26] . Multiple chronic conditions (MCC) further reduce the quality of life of patients and cause a huge economic burden [27] . In this survey, 34.56% of the participants had chronic diseases, and 30.84% had two or more diseases. The prevalence of chronic diseases in poor areas is high, and more than one-third of residents suffer from multiple chronic diseases, which is higher than that in non-poor areas of China [28][29] .
The structural equation model showed that the health-related factors were mediating variables with weak effects. There was a negative correlation between health-related behaviours and health literacy and a positive correlation between health literacy and the types of chronic diseases. The higher the score of health literacy is, the lower the score of health-related behaviours, which means that residents with a high level of health literacy engage in fewer unhealthy behaviours [2,30−31] . Improving the health literacy of residents in poor areas can reduce unhealthy behaviours, and people's lifestyle will be healthier [32][33][34] . Thus, the results of this study show that health literacy scores are positively correlated with the types of NCDs, which is contrary to expectations. However, patients with NCDs tend to pay more attention to healthy living habits and avoid unhealthy behaviours because they have acquired health knowledge from doctors or are already ill, so their health literacy scores are higher, which is consistent with the research results of Liu L,Qin L and Villaire M [1,35−36] .
To the best of our knowledge,this study is the rst one to investigate on the health literacy status of residents in impoverished counties in China, and this may kick off in the future. This survey was organized by the Henan Provincial Health Commission, Henan Provincial Center for Disease Control and Prevention and other government departments. The sample selection was standardized, the data quality was high, and health risk behaviors were used as intermediary variables to further study the relationship between health literacy and chronic diseases. At the same time, the survey results should be interpreted cautiously due to the following limitations: First, we are a cross-sectional survey, and there is insu cient comparison with the previous ones. Therefore, we should carefully check the causal inferences; The health literacy measurement tool we used was announced by the Chinese government in 2012. With the continuous development of the concept of health literacy, the measurement tool needs to be updated.

Conclusion
In summary, individuals living in poverty-stricken areas have a very low level of health literacy, engage in more unhealthy behaviours such as smoking and drinking, and have a high prevalence of NCDs. Health literacy has an impact on both health-related behaviour and NCDs prevalence. The relevant government departments should increase health education efforts to improve residents' health literacy, reduce the prevalence of chronic diseases and improve their quality of life. Figure 1 Theoretical model and hypotheses. X1=Scienti c and healthy ideas, X2=Prevention of infectious diseases, X3=Chronic disease prevention, X4=Safety and rst aid, X5=Basic medical information,

Figure 2
Types and proportion of NCDs of the surveyed subjects Figure 3