Study design
This was a cross-sectional study of health literacy and its geographic heterogeneity in China. Health literacy was measured using the National Resident Health Literacy Monitoring Questionnaire. The study subjects were permanent residents aged 15-69 years who had continuously lived in the survey areas for more than 6 months. Residents with cognitive impairment or hearing loss were excluded from the study.
The research protocol was reviewed and approved by the Medical Ethics Committee of Central South University. All participants aged 16 and older who agreed to participate in the study signed an informed consent form at the beginning of the survey. Written informed consent was obtained from a parent or guardian for participants under 16 years old.
Sampling methods
There are 34 provincial administrative regions in China. After factors such as geography, economic level, population, accessibility for investigation, funding, etc., were comprehensively considered, 25 provinces were non-randomly selected for inclusion. Multi-stage stratification and cluster sampling were used to randomly select a total of 70 survey points nationwide among the 25 provinces or municipalities (6 in the eastern region, 8 in the central region, and 11 in the western region) in China, with 1-2 urban and rural survey points selected in each province according to regional and population factors, followed by the random, step-by-step selection of the sample areas (counties), streets (townships) and neighbourhood committees (villages). Furthermore, 50 households were randomly selected from each community according to the community's resident roster, and then one eligible respondent was randomly selected from each selected household. The sample size was calculated according to the cluster sample method based on the following parameters: rate of adequate health literacy π=20%, significance level α=0.05, allowable error σ=0.02, intracluster correlation coefficient (ICC) ρ=0.01, average cluster size c=135. Then the sample size was estimated to be N=[Zα/22×π×(1–π)/σ2]×[1+(c-1)×ρ] = 3596 ≈ 3600.
Study measures
Demographic characteristics
The socio-demographic characteristics collected in this study included gender (male or, female), age (15~29, 30~49, or 50~69 years), place of residence (eastern, central or western region), community type (urban or rural community), marital status (single or married), education level (elementary school and below, junior high school, senior high school, or college and above), and economic status (poor, medium, or good). The surveyed residences were divided into the eastern, central, and western regions according to the region classification in the China Health Statistics Yearbook. Economic status was divided into the poor, medium, and good categories, with the cutoff points being 75% and 125% of the median annual household income per capita.
Health literacy
The National Resident Health Literacy Monitoring Questionnaire (NRHLMQ), prepared by the China Health Education Center, was used to measure health literacy. The overall Cronbach’s alpha of the NRHLMQ was 0.95, and the Spearman-Brown coefficient was 0.94 [12]. Confirmatory factor analysis showed that the questionnaire measured a unidimensional construct with three highly correlated factors [12]: (1) basic knowledge and attitudes (BKA), (2) healthy lifestyles and behaviours (HLB), and (3) health-related skills (HRS). The NRHLMQ covers six domains: scientific views of health (SVH), prevention and treatment of infectious diseases (PTID), prevention and treatment of chronic diseases (PTCD), safety and first aid (SFA), basic medical care (BMC), and health information (HI).
There are three types of questions on the NRHLMQ: true or false (with 1 point given for each correct response), single answer (a multiple-choice question with only one correct answer, where 1 point is given for each correct response), and multiple answer (a multiple-choice question with more than one correct answer, where 2 points are for each correct response). For the multiple-answer questions, a correct response was defined as one that contained all of the correct answers and none of the incorrect ones.
The maximum total score of the NRHLMQ is 66 points, with the maximum total scores of the three dimensions being 28 (BKA), 22 (HLB), and 16 (HRS) points. The maximum total scores for SVH, PTID, PTCD, SFA, BMC, and HI are 11, 7, 12, 14, 14, and 8 points, respectively.
A total score of 53 (80% of 66) points or above was considered to indicate adequate health literacy (adequate HL). A score of 0 to 52 was considered to indicate limited health literacy (limited HL). The health literacy level was defined as the proportion of participants who had adequate health literacy out of the total number of participants. The judgment criterion for adequate HL in each dimension or domain was ≥ 80% of the total score for the dimension or domain [12-13].
Health status
The self-evaluated health status was used as the evaluation index and was divided into good, fair, and poor levels. The original question was, "What do you think of your health status in the past year?"
Community health education
We used the number of health lectures given by the primary care practitioners as a proxy measure of community health education, determined by a question, "How many health lectures did you attend in your community during the past three years?" The self-reported frequency of participation in community health education was divided into three categories (0 times, 1-9 times, and ≥10 times).
Survey method
In the pre-investigation stage, a certain number of respondents were randomly selected from the sample locations for pre-surveys, focusing on whether the questionnaire items were unambiguous and clearly understood. The results showed that the respondents could understand the contents of the questionnaires. In the formal investigation phase, individual interviews were used to collect information. In the re-testing phase, which was two weeks after the formal investigation, 155 respondents were randomly selected from the overall sample using a computer-based simple random sampling technique, and the investigators re-tested those subjects by phone. All stages of the investigation were completed by undergraduate students majoring in preventive medicine. Prior to the investigation, all investigators were given uniform training for this survey. The investigation was conducted from January to April 2017.
Statistical analyses
Statistical analysis was conducted with SPSS version 19.0 (IBM Corp., Armonk, NY, USA) and MapInfo Professional version 7.0 (Pitney Bowes MapInfo Corp., Stamford, USA). An integrity check was performed before submitting the questionnaire, and questionnaires with missing values were not included in the analysis (there were 3 questionnaires with missing values). Prior to the analysis, data were screened for outliers and out-of-range values. No outliers or out-of-range values were found. The general conditions and health literacy of the sample were statistically described as the mean ± standard deviation, composition ratio, median, and frequency distribution table. In order to evaluate the factors of health literacy, the health literacy scores were dichotomized into two categories: adequate and limited. The chi-squared (χ2) test was used to compare the health literacy levels among different characteristic groups. A series of multiple logistic regressions was used to adjust for the relevant factors associated with the health literacy level in the total and regional samples. The logistic regression analyses were performed with gender, age group, marital status, community type, education level, economic status, self-rated health status, and frequency of participation in community health education as the independent variables; adequate health literacy served as the dependent variable in the overall and regional samples. An adequate HL equation was established using a multiple logistic regression model with stepwise forward selection. In all hypothesis tests, two-sided P-values of <0.05 were taken to indicate statistical significance.