The impact of predisposing, enabling, and need factors in utilization of health services among rural residents in Guangxi, China

China was geographically grouped into urban (i.e., cities) and rural (i.e., town or villages) areas. China’s per capita gross domestic product (GDP) was RMB 38354 (approximately US$6100) in 2012, and the average net income for farmers was RMB 7919 (approximately US$1250) [15].

Guangxi is an economically developing agricultural province in western China. The population was estimated at 52,820,000 inhabitants in 2012. The per capita GDP in Guangxi was RMB 30,588 (approximately US$4829), and the average net income of farmers was RMB 6213 (approximately US$981) [16]. Guangxi's income level was lower than the national average, especially in rural areas. This study focuses on Guangxi, one of typical developing provinces, to investigate the need of health service in rural China.

Guangxi province has more public health problems compared to developed provinces in China. Low government health expenditures and inadequate health resources is a common problem in these developing regions [17]. Most poor people in Guangxi are villagers in remote and mountainous regions with low levels of education and weak awareness of health protection, all of which are susceptible risks of the spread of infectious diseases [18].

The eligible participants in this study were the Guangxi peasants living in the sampling rural area for not less than 6 months. The participant was excluded if three investigation failures had happened or unable to communicate (verbally or written). The recall bias of self-reported prevalence is inevitable in a retrospective investigation. To reduce the bias, the survey required participants to recall illness during the last two weeks of the survey in accordance with international practice.

A cross-sectional study, using multi-stage stratified cluster random sampling for household survey (county and township cluster, and simple random sampling of households), was carried out from June 30 to July 20, 2012. The single proportion formula of the sample size was 1.96²pq (DEFF)/d ² (p = two-week prevalence of 17 % in rural Guangxi [19], q = 1 − p and d = margin of error of 1.5 %). The adjusted sample size was 4818 by multiplying a factor of 2 due to the design of multi stage sampling [20]. Considering a possible 10 % non-response rate, the final sample size was 5300.

Counties were stratified into two groups (above or equal to average, below the average), based on per capita GDP of counties in Guangxi [21]. Two counties, Rongxian County (above the average per capita GDP) and Luchuan County (below the average per capita GDP), were chosen. Three townships were selected in each county. Three villages were selected in each township, based on population size. Households were randomly selected from sample villages and all family members in a sampled household were individually interviewed.

The survey interviewed each household member in the sample families. The interviewers were trained students from the public health school of Guangxi medical university. The parents were asked to provide relevant information in cases of children younger than 15 years old.

In the household health interview, a questionnaire was used for data collection (Additional file 1). The main contents of the questionnaire were health service utilization in household members. It included characteristics of socio-economic status, demographics, and insurance for the residents; self-reported illness and 2-week prevalence prior to the survey; consultation, hospitalization, and expenditure of both inpatient and outpatient.

The outcome of healthcare utilization is measured in two perspectives: the utilization of outpatient visit (e.g., physician visit) and the utilization of inpatient visit (e.g., hospitalization). The utilization of outpatient visit refers to the questionnaire item “have you been to a clinic in the last two weeks?” The utilization of inpatient visit is based on the questionnaire item “Have you been hospitalized in the past half year?” These two variables were dichotomous (yes or no).

The database was created in Epidata 3.0. Statistical power was calculated with PASS version 11.0. Other analyses were performed using SPSS version 17.0. Descriptive statistics were used to describe the social, economic, and demographic characteristics of rural residents. To test the statistical significance of the difference between the various subgroups in the percentage of reported problems, χ² tests and z tests were used. Subsequently, multivariate logistic regressions were used to determine the association of health service utilization with the predisposing, enabling, and need characteristics. Multivariate analyses were performed in three different steps (i.e., 1. considering the need factor; 2. inserting the predisposing characteristics; and 3. considering the enabling resources), which allowed the assessment of the influence of need, controlling for possible confounding associations on health service utilization. Statistical significance was considered when two-sided P ≤ 0.05, and the results were expressed as odds ratios (OR) with 95 % confidence intervals (CI).

The questionnaire survey was conducted in 5300 potentially eligible individuals, out of which non-eligible people included 98 non-reachable for three attempts and 3 unable to communicate in mandarin. The participant rate was 98.09 % (5199/5300). Among the 5199 eligibility, 2 refused to cooperate with the interview due to privacy. The rest of 5197 participants completed the questionnaire with the correct participant rate of 98.06 % (5197/5300). Missing answers to the key questions, such as inpatient utilization and chronic diseases, were found in 563 participants, the rest 4634 valid subjects were analyzed in this study. The correct response rate was 87.43 % (4634/5300). The statistical power was 0.987 indicating reliable results.

In the 4634 subjects, 686 (14.80 %) subjects had used outpatient service in the last two weeks, and 280 (6.04 %) subjects had used inpatient service in the last six months. Descriptive analysis by socio-economic status is shown in Table 1. The sample consisted of 4634 respondents with an average age of 32.72 ± 21.76. More than half (52.42 %) were male, and 53.69 % were married. The average family size was 6.02 people. An educational level below junior middle school was reported by 48.70 % of the respondents. The average travel time to the nearest health facility was 10.52 min. The mean expenditure on healthcare per year was RMB 4242.16, 98.2 % covered by national health insurance with a cost of RMB 172.92 per year. There were 11.57 % of the subjects suffered from chronic diseases.

As one of the predisposing factors, age was positively associated with the utilization of physician visits (Table 2) and hospitalization (Table 3). Females and married people showed health seeking behaviors similar to the age factor. However, family size was negatively associated with the utilization of outpatient. As enabling factors, the travel time to health facility and expenditure on NCMS were associated with the utilization of physician visits, but not hospitalization. Conversely, expenditure on healthcare for inpatient but not outpatient was significant. Educational level was negative associated with the two utilizations. For the need factor, subjects with chronic diseases tended to use physician visits and hospitalization more than those without chronic diseases.

Table 4 summarizes a series of multiple logistic regression analyses to estimate the relationship between the health services utilization and the predisposing, enabling, and need factors. For the physician visit, there was a strong positive association between self-reported health problems and the utilization. In the unadjusted model, the need factor entered the initial model making a significant contribution. Those who had chronic diseases were more willing to use outpatient and inpatient health service. After adjusting all predisposing and enabling factors, the influences of chronic diseases were still observed (physician visits: OR = 5.87, 95 % CI = 4.71–7.32; hospitalization: OR = 4.04, 95 % CI = 2.90–5.61, respectively). In addition, the age factor was significantly associated with hospitalization but not physician visits. Married people had greater OR compared to singles. People with high education levels were less likely to use health services.

For the 536 subjects with chronic diseases, only 246 (45.90 %) reported physician visits, and only 96 (17.91 %) reported being hospitalized. The commonest reported reason for not using a health service even with chronic diseases was financial difficulty.

Table 5 shows medical expenditures, payments, and proportions in rural areas and the entire province of Guangxi. The medical expenditure per physician visit in rural areas of Guangxi was higher than that for the entire province of Guangxi, while the mean hospitalization expenditure in rural areas (RMB 2507) was much lower than that of the entire province (RMB 3648, of which RMB 976 could be reimbursed) (Table 5). The reimbursement proportions were 26.8 % (976/3648) in the entire province and 30.4 % (762/2507) in the rural areas.

We observed that the predisposing factors, including age, gender, family size and marital statue contributed significantly to the variance in health service utilization. Numerous studies showed a positive association between socio-economic status and health in Europe [22, 23], as well as in China’s previous studies [24, 25]. Among all predisposing factors, marital status had a stronger effect than age, gender and family size, which is consistent with previous studies [26, 27]. As marital status affects health seeking behavior, a possible explanation is that married people get help and advice from their spouse promoting a doctor visit. Younger individuals were found more likely to use inpatient health services, possibly due to parental attention to young children. In this study, the OR of gender was 1.2 for physician visits and 1.4 for hospitalization, indicating that females have more health needs and awareness than males. Gender equality is an important issue in health service utilization. Many studies showed the differences in health seeking behavior between males and females [28, 29]. In addition, the family size was negatively associated with the utilization of health services, in another word, smaller family size linked to more health service utilization. These data suggested that married females with smaller family sizes were more likely to endorse willingness to use health services. These people should be a key target group for health interventions.

Educational level was negatively associated with the health service utilization, which is consistent with several previous studies [17]. Higher education may lead to more knowledge of preventive health care and awareness of health services. For this population, it is important to improve their personal ability of health control by influencing their lifestyles and to increase the use of health services [30]. Travel time to the health facility, consistent with the previous findings [31, 32], was negatively associated with the utilization of physician visit. Therefore, more village clinics should be strategically set up to guarantee convenience in access to primary health services. The reason for significant personal expenditures associated with the hospitalization utilization may be the result of the present health policies, hospitalization means more out-of-pocket expenditures compared to outpatient services. Association between the expenditure of NCMS and the utilization of health services was not found, reflecting the benefits from the popularized China rural basic medical insurance. According to China health statistics yearbook, the scheme covered 832 million rural residents, or 97.5 % of Chinese farmers, and the government contribution to insurance premium increased to 246 RMB (US$38.51) per person in 2011 [33].

Chronic conditions represent potential needs for healthcare use. Our findings are comparable to other studies showing patients with chronic illness were more likely to use outpatient and inpatient services [34‐36]. These suggest that chronic diseases play a critical role in increasing health service utilization by promoting a more positive personal perception.

As mentioned above, many socio-economic characteristics play an important role in the utilization of health services in the process of seeking medical attention of rural residents. These factors affect their decisions and thereby affect the choice of medical service expenses. Primary health care institutions are of great importance in the rural health service system, village clinics serve as basic health maintenance, and county-level hospitals are the main choice for the treatment of serious disease. In this study, most outpatient service happened in village clinics and over a half of inpatient service happened in county-level hospitals. Thus, enhancing village clinics and county-level hospitals, including improving financial investment, infrastructure and staff training, is necessary for better access to the rural primary health care.

In rural areas of China, personal or family economic difficulty is a major problem, particularly in case of hospitalization. The reimbursement proportions in both this study and Guangxi province are low. Residents in rural areas must pay for the rest of medical expenditure, which still leads to a heavy economic burden. Since 1990, the increasing use of high-level medical technologies and expensive drugs has raised up expenditures on health service in China. Studies also showed that the health expenditure per capita has grown since the 1980s [37]. Therefore, many people with low income have experienced increased financial difficulty in seeking health services [38]. It is necessary to cut down medical expenditures in rural areas to relieve the economic burden. NCMS sets deductibles and a ceiling, poor residents might be consequently out of the coverage due to financial difficulties. Therefore, NCMS is a basic guarantee for health service and other supplemental insurances are needed.

This study has some limitations. Self-reported responses used in this investigation is a common way to evaluate the use of health services [39, 40]. Reporting bias is a major disadvantage in the assessment of health care because of self-reported data [41]. The recall period in health surveys, 2 weeks for outpatient and 6 months for inpatient, is accepted for eliminating the error. Clark et al. [42] found that the recall accuracy is positive associated with hospitalization time. Clarke et al. [39] reported recall period should be determined in accordance with survey technique, the outcome, and policy. The estimated sample size was 5300, while the subjects of analysis were 4634. Non-response is a source of missing data. There are different non-response bias effects, and non-response on disease contributes most to the bias [43]. The dependent variable in the regression was a simple categorical variable. Further studies need to conduct continuous variables and items to expand the basic knowledge and improve the attitudes in rural adults toward health services. Despite these limitations, the present study collected important information on health service in Chinese rural adults. Furthermore, our findings may provide baseline information for educational interventions to promote healthcare utilization awareness in rural populations.