Skip to main content
  • Research article
  • Open access
  • Published:

Does inpatient health services utilization vary by remoteness in the medical financial assistance population? Evidence from Shaanxi province, China



Medical Financial Assistance (MFA) provides health insurance and financial support for millions of low income and disabled Chinese people, yet there has been little systematic analysis focused on this vulnerable population. This study aims to advance our understanding of MFA recipients’ access to health care and whether their inpatient care use varies by remoteness.


Data were collected from the Surveillance System of Civil Affairs of Shaanxi province in 2016. To better proxy remoteness (geographic access), drive time from the respondent’s village to the nearest county-level or city-level hospital was obtained by a web crawler. Multilevel models were used to explore the impacts of remoteness on inpatient services utilization by MFA recipients. Furthermore, the potential moderating role of hospital grade (i.e. the grade of medical institution where recipient’s latest inpatient care services were taken in the previous year) on the relationship between geographic access and inpatient care use was explored.


The analytical sample consisted of 9516 inpatient claims within 73 counties of Shaanxi province in 2016. We find that drive time to the nearest hospital and hospital grade are salient predictors of inpatient care use and there is a significant moderation effect of hospital grade. Compared to those with shortest drive time to the nearest hospital, longer drive time is associated with a longer inpatient stay but fewer admissions and lower annual total and out-of-pocket (OOP) inpatient costs. In addition, these associations are lower when recipients are admitted to a tertiary hospital, for annual total and OOP inpatient expenditures, but higher for length of the most recent inpatient stay no matter what medical treatments are taken in secondary or tertiary hospitals for the most remote recipients.


Our results suggest that remoteness has a significant and negative association with the frequency of inpatient care use. These findings advance our understanding of inpatient care use of the extremely poor and provide meaningful insights for further MFA program development as well as pro-poor health strategies.

Peer Review reports


Health is a prerequisite for an individual’s all-round development. The current health reform in China aims to enable residents to have equal access to quality and affordable essential health care. Substantial progress has been achieved in ameliorating health care access including the establishment of basic medical insurance schemes (e.g. the New Cooperative Medical Scheme in rural areas and the Urban Resident Basic Medical Insurance in urban areas) that cover all residents, yet there is very limited evidence for the most financially disadvantaged population [1, 2]. According to the statistics from the National Poverty Alleviation and Development Information System, illness remained the top cause of poverty in rural poor households. Among more than 70 million poor families in China, the proportion of poverty caused by illness increased from 42% in 2013 to 44% in 2016 [3]. According to the Poverty Monitoring Report of rural China, in rural areas, the percentage of residents unable to receive timely health care due to financial handicaps dropped by 7.8%; however, the number increased by 10.1% due to long and essential drive times and transportation costs [4]. These vulnerable populations normally have poor health and live in rural and remote areas with limited access to health care [5,6,7,8,9]; they are also more likely to delay or deter essential medical treatments [10,11,12].

To protect the health of the vulnerable population and to reduce poverty, the Medical Financial Assistance (MFA) scheme was introduced in China as a purely pro-poor strategy. The MFA scheme targets extremely vulnerable individuals and families, and subsidizes basic medical insurance or provides cash assistance for medical care. In 2018, approximately 130 million individuals benefited from MFA [13]. A summary of eligibility criteria, basic information, and benefit packages of MFA is shown in Table 1. Although the MFA aims to expand basic medical insurance coverage, increase health care access, and to reduce financial risk for the poor, recent studies have found it has limited impacts on health promotion or financial risk protection, particularly for those who live in rural or remote areas [12, 14]. Furthermore, indirect costs, such as transportation costs, are not covered. Significant differences in health care access and use between rural and urban residents (geographic disparity) has made the limited access and delivery of health services for rural residents a widespread concern in China [5, 6, 15, 16].

Table 1 Summary of Medical Financial Assistance Scheme (MFA) at the national and provincial level

Andersen’s Behavioral Model of Health Services Use (consisting of predisposing factors, enabling factors, and needs) has been widely used to study health care access [17,18,19,20]. Studies have also suggested that access to health care comprises financial access (medical costs and people’s capacity to pay) and geographic access [21]. Geographic access denotes essential distance or time to medical institutions; the distance that must be traveled (or travel time) to the health facility to receive health care limits access to health services. In China, studies of financial access to health care are well established but only a few focused on geographic access [8, 22]. They found that geographic access is one of the dominant factors hindering health services utilization in China, especially in rural areas of the western region, due to the long distance to medical institutions and the undersupply of transport services [23].

There is no evidence on whether geographic access influences the MFA population in China. This study aims to fill the gap by investigating whether inpatient services utilization varies by remoteness in MFA recipients. Moreover, given the important role played by the hospital grade in disparities of health services access and utilization in China [22, 24, 25], this study explores whether hospital grade acts as a moderator in the associations between geographic access and inpatient services utilization. The findings will advance our understanding of inpatient care use by the most vulnerable subgroups in China, as well as provide unique insights for future MFA program design and implementation.


Data source

The study site is Shaanxi Province, which is located in the northwest of China, with an annual per capita net income of a rural resident of 9396 RMB (Chinese Renminbi Yuan, 1 RMB = 0.14 US dollars) and ranked 26 among 31 provinces in China in 2016. In this region, 80% are mountainous or plateau areas, particularly in rural areas. In this study, community-level indicators were drawn from the National Bureau of Statistics 2017 and Shaanxi Health and Family Planning Statistical Yearbook 2017, while data on individual-level characteristics were obtained from the Surveillance System of Civil Affairs of Shaanxi Province in 2016. The Surveillance System was established in 2012, and includes the vulnerable populations who are eligible for the MFA in all 73 counties in Shaanxi Province. The Civil Affairs Departments are responsible for data collection and data, which is submitted annually.

The MFA data comprises three independent groups according to how assistance is received: in the first group, recipients receive subsidies for enrolling in basic medical insurance schemes; recipients from the second group receive immediate assistance during health service utilization, and the third group receive assistance only after medical treatment. The MFA data contains individual information, including personal demographics (e.g. name, gender, age, marital status, health status), household information (e.g. household income, family size, and residential address) and assistance details (e.g. reimbursement method; outpatient/inpatient medical costs: total cost, self-paid payment and reimbursement cost; and inpatient admission and discharge dates).

Study sample

This study focuses on rural residents as they are the most vulnerable population. For the first two MFA groups introduced above, there was not sufficient personal information (e.g. economic status and complete hospitalization details) available to the researchers. Consequently, we focus on the third group, i.e. those who receive assistant after medical treatments. From the total of 11,346 recipients’ records in this group, we firstly exclude around 7% (n = 794) of records in which the inpatient care services were taken in township health centers which are normally located close to home and treat minor disease. Next, 1036 records with key information missing (e.g. inpatient medical costs) were dropped, leaving a final study sample of 9516 recipients within 73 counties.

Empirical strategy

Geographic access

Travel barriers are represented by either straight-line distance, driving distance, or drive time to the nearest health institution and are frequently used to measure geographic access to health care [26]. Straight-line distance is intuitive and easily calculated, while actual driving distance and time are regarded as better alternatives as they factor in the actual road network likely to be used [27]. These indicators used to be difficult to measure but have now become much easier due to the advancement of web crawler techniques [22]. A web crawler is a search engine that can systematically visit web sites and collects the relevant information with high accuracy [22, 27, 28]. Here we use this technology to obtain the actual driving distances and time from a resident’s home to the nearest secondary/tertiary hospital in Shaanxi Province; all requests to the API (Application Program Interface) of Amap (China’s leading solution provider of digital map content, navigation and location services) were completed by the web crawler with a self-compiled Python 3.6 program.

The names of village clinics/neighborhood committees, county-level and city-level hospitals in Shaanxi province were obtained through the Shaanxi Provincial Health Statistical Annual Report and Shaanxi Rural Health Statistical Yearbook. Coordinates of the village clinics/neighborhood committees were taken as the starting points while the coordinates of the county/city hospitals were regarded as the terminal points and the path planning interface was used to collect navigation data. The data strategy chose the fastest route without taking highways to obtain the time and distance from the village clinics/neighborhood committees to the county-level/city-level hospitals. Taking account of traffic conditions at different times, four crawler requests for Amap were conducted on Friday, November 23, 2018 (from 10:00 am to 11:00 am) and Tuesday, November 27, 2018 (from 14:00 to 15:00 PM), and the average of four times was used.

Eventually, the data for 10,350 clinics/neighborhood committees were obtained from 13,074 clinics/neighborhood committees in 73 counties of Shaanxi province. Given the importance of drive time in medical treatment, particularly for emergency treatment and severe diseases, we used drive time from residential districts to the nearest secondary/tertiary hospital to measure remoteness [25, 27]. Drive time (in minutes) was categorized in five groups: the shortest (time < 30) = 1; shorter (30 < = time < 60) = 2; medium (60 < = time < 90) = 3; longer (90 < = time < 120) = 4; longest (time > 120) = 5. Without a unified and standard measurement for changes in driving time, we chose a 30-min increase as the cutoff point based on the data distribution and previous literature which suggests 30 min is a significant increase in driving time [29].

Outcome indicators

The primary outcome variables in this analysis are: (1) length of the latest inpatient stay last year, (2) number of admissions last year, (3) total inpatient expenditure, and (4) out-of-pocket (OOP) inpatient expenditure. Length of the latest inpatient stay is calculated as the number of days for latest hospitalization in the prior year, number of admissions last year is measured by the number of admissions during the previous year, while total inpatient expenditure and OOP inpatient expenditure are gauged using all the inpatient care costs and self-paid inpatient care costs in the past year, respectively.


Our covariates were selected according to the framework of Anderson’s Behavioral Model: socio-demographic characteristics, e.g. gender, age, marital status, and education level were specified as predisposing factors; the presence of any chronic diseases and self-reported health status were selected as need factors. In terms of enabling factors that may facilitate health behaviors, economic status, driving time to the nearest hospital, and the hospital grade were grouped into individual-level enabling factors. Over the past decades, tremendous attention has been given to individual-level factors in examining their effects on health service utilization, while there is a growing appreciation that factors beyond individual characteristics also play important roles in the disparities of health care use [20]. Based on the literature, population density, per capita GCP, number of beds per 10,000 people, number of doctors per 10,000 people and number of nurses per 10,000 people in the county where the recipient lived were included as community-level enabling factors, which signify economic developments and medical resources in the counties that influence which health services are delivered [19]. Medical insurance schemes and Hukou status were not included in this study as more than 98% of the residents in these 73 counties were covered exclusively by the New Cooperative Medical Scheme and 95% of them were agricultural Hukou. More details about dependent and independent variables are presented in Table 2.

Table 2 Definitions of variables

Multilevel model

A multilevel model was adopted to allow for the clustered data structure [30]. Here, recipients (individual-level) were nested within counties (community-level), thus a two-level linear mixed model was adopted to simultaneously estimate individual-level and community-level effects with the estimating equation:

$$ {y}_{ij}={\alpha}_0+{\alpha}_1{\mathcal{x}}_{ij}+{\alpha}_2{\mathcal{w}}_j+{\mathcal{u}}_j+{\varepsilon}_{ij} $$

In the specification, yij represents the inpatient care utilization of recipient i in county j and α0 means the intercept. \( {\mathcal{x}}_{ij} \) and \( {\mathcal{w}}_j \) are the individual-level and community-level variables, with corresponding coefficients of α1 and α2, respectively. \( {\mathcal{u}}_j \) is the individual level error term while εij denotes the community level error. In using a multilevel model, the intra-class correlation coefficient (ICC) of between-county heterogeneity in the community level needs to be statistically significant.

Moderation effect

Moderation effects, if significant, change the direction or magnitude of the association between independent and dependent variables [31]. In this study, guided by previous literature [22, 24, 25], we hypothesize that hospital grade would function as a moderator in the relationship between geographic access and inpatient care utilization in the MFA population. In order to test this moderating effect, hierarchical multiple regression was conducted. In this model, the occurrence of moderation can be observed by the significance of predictor and moderating variables as well as the general model R2 in Block 1 (without interaction term), and a significant interaction term and a significant R2 change in Block 2 (interaction term added). Adjusted predictions (marginal effect) of the interaction term were calculated for ease of interpretation of the interaction effect.

Descriptive statistics for the total sample by hospital grade summarize the characteristics of the MFA recipients. Mean (SD) and T-test were used for continuous variables, N (%) and Chi-square test were used for categorical variables. All statistical analyses were carried out using Stata, version 15.0.


Table 3 provides descriptive statistics for our study sample. Of the 9516 MFA recipients within 73 counties, more than half of the recipients were males (53.8%), aged above 45 years (75.8%), married (78.7%), and had an education level of primary school or below (86.6%). Regarding health status, 21.4% of recipients had chronic diseases and 32.3% reported having a disability or serious illness. The average driving time was 88.95 min from home to the closest hospital. In terms of inpatient care use, the average length of stay for the latest inpatient care was 21.1 days, and on average individuals sought inpatient treatment 1.38 times in the previous year, with a total annual inpatient expenditure of 20,828.01 RMB and an average OOP expense of 4954.13 RMB.

Table 3 Basic characteristics of variables for the total sample and comparisons between hospital grade

Among the study sample, 37.8% (3592) of recipients had their latest inpatient care in secondary hospitals whilst 62.3% (5924) were in tertiary hospitals. For those hospitalized in secondary hospitals, more than 80% of them required up to 60 min to drive to the nearest hospital, whereas only 13% were within a 1 h drive to the nearest tertiary hospital. When comparing those who were hospitalized in secondary hospitals and tertiary hospitals, significant differences were found in a range of characteristics, such as age, whether they had a chronic disease, health status, driving time to the hospital, as well as population density, per capita GCP, number of beds per 10,000 people, number of doctors per 10,000 people and number of nurses per 10,000 people for the counties where the recipient lived.

The ICC showed that 11.6% of the total variance in length of the latest inpatient stay (19.1% of the total variation in total inpatient expenditures and 10.4% of the total variance in self-paid inpatient expenditures, respectively) could be explained by community-level differences (for details see Appendix Table 1). On the other hand, for the number of admissions, only 4.6% of the total variance can be attributed to between-county differences, and therefore the OLS regression was used.

Table 4 shows the detailed regression results. After controlling for a wide range of confounding factors, we find that drive time to the latest hospital is significantly associated with inpatient care use. More specifically, compared to those with shortest drive time, all other drive time groups had significantly longer inpatient stays for the latest inpatient care, but fewer admissions (not significant for the medium and the longer drive time groups), and less annual total and OOP costs (not significant for the longest drive time group). Furthermore, compared to those admitted in secondary hospitals, recipients hospitalized in tertiary hospitals had a significantly shorter length of inpatient stay (β = − 3.971, P < 0.001), but higher total hospitalization expenses (β = 12,705.810, P < 0.001) and OOP hospitalization expenses (β = 4174.214, P < 0.001). These results indicate that for recipients who had inpatient care in tertiary hospitals, on average their inpatient stay was nearly 4 days shorter but total inpatient costs and OOP inpatient costs were 12,705.810 RMB and 4174.214 RMB higher, respectively, compared to their counterparts in secondary hospitals. Recipients’ characteristics, such as age, chronic disease, and population density of counties where recipients lived also had significant effects on all outcome variables of interest.

Table 4 Multilevel model analysis on influencing factors for inpatient health services utilization among the MFA recipients

We tested the moderation effect of hospital grade in the association between driving time to the hospital and inpatient care use by creating interaction terms between driving time to the hospital and the hospital grade. Significant R2 changes and interaction terms were observed in predicting length of the latest inpatient stay (Δ R2 = 0.012, P < 0.001), total inpatient cost (Δ R2 = 0.039, P < 0.001) and self-paid inpatient cost (Δ R2 = 0.032, P < 0.001). Detailed results for the hierarchical multiple regression examining the moderation effect are reported in Appendix Table 2, and the adjusted predictions of the interaction terms on inpatient care use are shown in Fig. 1.

Fig. 1
figure 1

Adjusted predictions of the interaction terms on inpatient care use. a length of the latest inpatient stay, b total inpatient expenditure, and c OOP inpatient expenditure. Secondary means secondary hospital; Tertiary means tertiary hospital

Part (a) shows that no matter what time intervals the recipients were in, the length of inpatient stay was longer in secondary hospital than in a tertiary hospital: holding the other covariates at the mean level, the longest inpatient stay is predicted to be 36.1 days (P < 0.001) in a secondary hospital and 29.3 days (P < 0.001) in a tertiary hospital for the longest group. Part (b) shows that on average total inpatient expenditures in tertiary hospitals were higher than those in secondary hospitals and the maximum value of the marginal effect was 36,076.7 RMB (P < 0.001) in a tertiary hospital, which was observed in the shortest group. The above trend was also observed in predicating OOP inpatient expenses, with the maximum adjusted prediction of 8890.1 RMB (P < 0.001) in the tertiary hospital (Part (c)). These findings imply that regardless of the drive time, the total and OOP inpatient expenditures were higher in a tertiary hospital compared with a secondary hospital, and recipients who incurred the highest inpatient costs were those with the shortest time to the hospital.

The adjusted predictions of time to the hospital and hospital grade on inpatient services utilization in terms of remoteness are shown in Table 5. Compared to those with the shortest time to the hospital, when admitted in a secondary hospital, on average the length of inpatient stay in a tertiary hospital was significantly longer in the shorter (β = 9.035, P < 0.01), the medium (β = 11.769, P < 0.001), the longer (β = 11.581, P < 0.001) and the longest groups (β = 13.510, P < 0.001), respectively; whilst when admitted in a tertiary hospital, it was only significant in the shorter and the longest groups. After including the moderator variable, the impact of driving time to a hospital on length of the latest inpatient care was increased when admitted to a secondary hospital for all groups and in a tertiary hospital for the longest group. The above results suggest that the hospital grade enhanced the association between time to hospital and length of inpatient stay. Annual total and OOP inpatient expenditure were lower for recipients in the shorter, the medium, the longer and the longest than in the shortest group regardless of whether their treatments were obtained in a secondary hospital or a tertiary hospital. Moreover, the relationships between drive time to the hospital and both annual total and OOP inpatient costs were smaller when recipients were admitted to a tertiary hospital.

Table 5 Adjusted predictions of time to the hospital and hospital grade on inpatient services utilization


This study investigated whether inpatient health services utilization varied by the remoteness, which was proxied by the drive time to the nearest hospital according to the actual driving time from Amap. The findings suggest that remoteness had a significantly negative effect on the frequency of inpatient care use in MFA recipients, which was further moderated by hospital grade.

Compared to those with the shortest drive time to the nearest hospital, the shorter, the medium, the longer and the longest had a significantly longer inpatient stay in the latest inpatient care but fewer admissions as well as lower total and OOP inpatient costs over the past year. Within the moderation of hospital grade, these links were increased in the relation between time to the hospital and length of the latest inpatient stay, while decreased when recipients were admitted to a tertiary hospital for both total and OOP inpatient expenditures. The reduction of inpatient admissions and annual total and OOP inpatient costs along with the increase in drive time reflects that the demand for health care reduced with the increase in travel distance. This geographic access barrier has played a significant role for whether to seek medical care and where to receive medical care when needed among MFA recipients. The unmet need may be an issue with the MFA population, however, further examination of this issue is beyond the capacity of our data [25, 32]. With the moderation of hospital grade, the relation between time to the hospital and inpatient costs were decreased, suggesting that even though there was a relatively lower frequency for recipients with longer driving time to the hospital, once they were admitted in a tertiary hospital, the total and OOP inpatient care costs increased regardless of drive time.

Longer drive time was associated with a longer (latest) inpatient stay. This finding is similar to a previous study for rural China, which reported that driving longer to the nearest clinic predicted a higher level of health service utilization in mountainous areas [10]. It might be that MFA recipients who live in remote mountainous and plateau areas generally have poor health status (eg. high-altitude areas expose humans to sustained hypoxia, which may lead to severe health problems) [10, 15]. Although the frequency of inpatient care was lower on average for residents living in far-off areas with poor roads and transportation, when their health status was worse (i.e. suffering severe illnesses or undergoing particular procedures), drive time tends to matter less. Therefore, when these residents decide to visit a doctor, they drive farther in pursuit of optimal medical treatment in faraway secondary or tertiary hospitals, given that high-quality care is concentrated in large counties or cities [15, 16, 32, 33]. This could be the reason that on average a single hospital stay is longer but the total number of hospitalizations and costs within a year are lower for recipients living in the low accessibility areas.

The results also indicated that hospital grade was significantly associated with recipients’ inpatient care use: the higher the level of hospital, the higher total and OOP inpatient expenses whilst the shorter length of the last inpatient stay. Higher total and OOP costs in tertiary hospitals are generally as expected [15, 25, 34, 35]. A plausible explanation for the shorter length of inpatient stay in a tertiary hospital is that tertiary hospitals have strict controls over the length of inpatient stays with high turnover rates of beds as more granular costing can be realized by amortizing their costs over a large volume of new patients, while secondary hospitals predominantly provide rehabilitation and nursing services and thus have relatively low turnover rates (according to China Health and Family Planning Statistical Yearbook 2017, the turnover rate of bed was 98.85 in tertiary hospitals and 84.1% in secondary hospitals in 2016) [36,37,38].

Considering other factors that influenced inpatients health utilization, as expected, compared to those subjectively perceived in good or very good health, recipients who reported disability or were seriously ill had longer inpatient stays as well as higher total and OOP inpatient expenses. Apart from individual factors, the community level characteristics also played a vital role in predicting recipients’ inpatient care use. Our study revealed that the increase in total beds per 10,000 people, population density and per capita GCP were related to a lower inpatient cost. It could be that residents in these counties with better health resources are generally wealthier and have better health status compared with those in less affluent areas.

Results are subject to a number of limitations. Firstly, only “Dibao” recipients that received ‘after medical treatment’ assistance were included from the MFA data. Ideally, we would like such data for all MFA recipients, however, hospitalization details of certain subgroups were incomplete in the available data and can’t be collected ex-post, which leads to other problems in assessing inpatient care utilization. For our purpose, the “Dibao” recipient is a more effective choice. Secondly, it is very common in the geographic access literature to use the village location as a proxy for the individual’s residence in rural areas as we have done. However, the drive time might be under-estimated. Further research would be benefit from taking each respondent’s residence as the minimum unit with more precise estimation. Thirdly, although the “diagnosis group for hospital admission” was not collected in the MFA data, other health need factors such as chronic disease and health status were adjusted in this study, enabling us to obtain a fine-tuned understanding of the inpatient health services utilization of MFA recipients. Finally, since only cross-sectional data for MFA recipients was obtained in our analysis, the associations we observed should not be interpreted as causal.

Despite these limitations, to the best of our knowledge, this study is one of only a few efforts to assess geographic access based on digital maps with real-time transportation information, thus offering higher precision of geographic access in MFA recipients in a developing country; it is also the first investigation of this topic concentrating on the MFA recipients in China. Our findings suggest further development of MFA program to offer more assistance to those in remote areas would avoid potential barriers to timely and adequate care. The benefit package should also cover the indirect medical costs such as transportation costs for the most vulnerable recipients.


Our results suggest that time to the hospital and hospital grade are salient predictors of inpatient care use, and there is a significant moderation effect of hospital grade. Compared to those with the shortest drive time to the nearest hospital, recipients with longer drive time had significantly lower frequency of inpatient care use with the most distant residents most affected. These findings provide novel evidence of the weak access for the extremely poor and unique insights for pro-poor health strategies as well as further development of MFA scheme.

Availability of data and materials

Data used in this study belongs to the Surveillance System of Civil Affairs of Shaanxi province and contains personal information (e.g., name, ID, etc.) of recipients. Due to the sensitive nature of these data and restrictions imposed by the institution, the authors cannot make these data publicly available. Other researchers who want to use the data may contact the author for data requests.



Medical Financial Assistance


Catastrophic Health Expenditure


Ordinary Least Squares


Gross County Product




Intra-class Correlation Coefficient


  1. Yip W, Hsiao W, Chen W, et al. Early appraisal of China’s huge and complex health-care reforms. Lancet. 2012;379:833–42.

    PubMed  Google Scholar 

  2. Liu G, Vortherms S, Hong X. China’s health reform update. Annu Rev Public Health. 2017;38:431–48.

    PubMed  Google Scholar 

  3. Li J. Achievements and challenges of health poverty alleviation in China. Seeker. 2019;05:95–103 (in Chinese).

    Google Scholar 

  4. Hu R, Dong S, Zhao Y, et al. Assessing potential spatial accessibility of health services in rural China: a case study of Donghai County. Int J Equity Health. 2013;12(1):35.

    PubMed  PubMed Central  Google Scholar 

  5. Liu M, Zhang Q, Lu M, et al. Rural and urban disparity in health services utilization in China. Med Care. 2007;1:767–74.

    Google Scholar 

  6. Zimmer Z, Kwong J. Socioeconomic status and health among older adults in rural and urban China. J Aging Health. 2004;16(1):44–70.

    PubMed  Google Scholar 

  7. Akin J, Dow W, Lance P, et al. Changes in access to health care in China, 1989–1997. Health Policy Plan. 2005;20(2):80–9.

    PubMed  Google Scholar 

  8. Gu X, Zhang L, Tao S, et al. Spatial accessibility to healthcare services in metropolitan suburbs: the case of Qingpu, Shanghai. Int J Environ Res Public Health. 2019;16(2):225.

    PubMed Central  Google Scholar 

  9. Shi W, Chongsuvivatwong V, Geater A, et al. Effect of household and village characteristics on financial catastrophe and impoverishment due to health care spending in Western and central rural China: a multilevel analysis. Health Res Policy Syst. 2011;9(1):16.

    PubMed  PubMed Central  Google Scholar 

  10. Fang P, Han S, Zhao L, et al. What limits the utilization of health services among the rural population in the Dabie Mountains-evidence from Hubei province, China? BMC Health Serv Res. 2014;14(1):1–7.

    Google Scholar 

  11. Fang P, Su M. A discussion on the key problems and system construction of health poverty alleviation in China. Chin J Health Policy. 2017;10(06):60–3.

    Google Scholar 

  12. Liu K, Yang J, Lu C. Is the medical financial assistance program an effective supplement to social health insurance for low-income households in China? A cross-sectional study. Int J Equity Health. 2017;16(1):138.

    PubMed  PubMed Central  Google Scholar 

  13. Hai F. International health care system profiles. China; 2020. Accessed 5 June 2020.

  14. Ma X, Zhang J, Meessen B, et al. Social health assistance schemes: the case of medical financial assistance for the rural poor in four counties of China. Int J Equity Health. 2011;10(1):44.

    CAS  PubMed  PubMed Central  Google Scholar 

  15. Qian D, Pong R, Yin A, et al. Determinants of health care demand in poor, rural China: the case of Gansu Province. Health Policy Plan. 2009;24(5):324–34.

    PubMed  Google Scholar 

  16. Han Y, Wei J, Song X, et al. Accessibility of primary health care workforce in rural China. Asia Pac J Public Health. 2012;24(5):833–47.

    PubMed  Google Scholar 

  17. Penchansky R, Thomas W. The concept of access: definition and relationship to consumer satisfaction. Med Care. 1981;19:127–40.

    CAS  PubMed  Google Scholar 

  18. Derose K, Gresenz C, Ringel J. Understanding disparities in health care access-and reducing them-through a focus on public health. Health Aff. 2011;30(10):1844–51.

    Google Scholar 

  19. Levesque J, Harris M, Russell G. Patient-centred access to health care: conceptualising access at the interface of health systems and populations. Int J Equity Health. 2013;12(1):18.

    PubMed  PubMed Central  Google Scholar 

  20. Aday L, Andersen R. Equity of access to medical care: A conceptual and empirical overview. Med Care. 1981;XIX:4–27.

    Google Scholar 

  21. Center for Health Statistics and Information. An Analysis Report of National Health Services Survey in China. Beijing: Beijing Union Medical University Press; 2008. (in Chinese).

  22. Cheng G, Zeng X, Duan L, et al. Spatial difference analysis for accessibility to high level hospitals based on travel time in Shenzhen, China. Habitat Int. 2016;53:485–94.

    Google Scholar 

  23. Wu C, Fang P. Analysis of inequity of health Service for Rural Residents in West China and potential accessibility and countermeasure. Chin Health Serv Manag. 2007;08:560–2.

    Google Scholar 

  24. Li X, Huang J, Zhang H. An analysis of hospital preparedness capacity for public health emergency in four regions of China: Beijing, Shandong, Guangxi, and Hainan. BMC Public Health. 2008;8(1):319.

    PubMed  PubMed Central  Google Scholar 

  25. Yu W, Li M, Ye F, et al. Patient preference and choice of healthcare providers in Shanghai, China: a cross-sectional study. BMJ Open. 2017;7(10):2–11.

  26. Al-Taiar A, Clark A, Longenecker J, et al. Physical accessibility and utilization of health services in Yemen. Int J Health Geogr. 2010;9(1):38.

    PubMed  PubMed Central  Google Scholar 

  27. Boscoe F, Henry K, Zdeb M. A nationwide comparison of driving distance versus straight-line distance to hospitals. Prof Geogr. 2012;64(2):188–96.

    Google Scholar 

  28. Quan B, Gang X, Yong, et al. Analysis and detection of bogus behavior in web crawler measurement. Procedia Comput Sci. 2014;31:1084–91.

    Google Scholar 

  29. Hsia R, Shen Y. Changes in geographical access to trauma centers for vulnerable populations in the United States. Health Affairs (Project Hope). 2011;30(10):1912.

  30. Israel S. How social policies can improve financial accessibility of healthcare: a multi-level analysis of unmet medical need in European countries. Int J Equity Health. 2016;15(1):41.

    PubMed  PubMed Central  Google Scholar 

  31. Bolin J, Hayes A. Introduction to mediation, moderation, and conditional process analysis: a regression-based approach. New York: The Guilford Press. J Educ Meas. 2014;51(3):201–45.

  32. Liu Y, Zhong L, Yuan S, et al. Why patients prefer high-level healthcare facilities: a qualitative study using focus groups in rural and urban China. BMJ Global Health. 2018;3(5):2–7.

  33. Chen R, Du X, Yang Z, et al. Analysis of choice of healthcare services and the influencing factors among outpatients in Chengdu city. Mod Prev Med. 2016;43:4217–20.

    Google Scholar 

  34. Meng Q, Mills A, Wang L, et al. What can we learn from China’s health system reform? BMJ. 2019;365:l2349.

    PubMed  PubMed Central  Google Scholar 

  35. Yip W, Wang H, Liu Y. Determinants of patient choice of medical provider: a case study in rural China. Health Policy Plan. 1998;13(3):311–22.

    CAS  PubMed  Google Scholar 

  36. Ma W. Research on the checks and balances mechanism of medical insurance cost control. Tianjin: Tianjin University; 2010. (in Chinese).

  37. Taheri P, Butz D, Greenfield L. Length of stay has minimal impact on the cost of hospital admission. J Am Coll Surg. 2000;191(2):123–30.

    CAS  PubMed  Google Scholar 

  38. Li F, Sang P, Zhu B, et al. Analysis of hospitalization behavior of death patients with malignant tumors in Shanghai. China Health Econ. 2019;38(05):65–70.

    CAS  Google Scholar 

Download references


Not applicable.


This study was funded by National Natural Science Foundation of China (71874137), China Medical Board (15–277 and 16–262), Xi’an Social Science Planning Fund Project (19S78), Shaanxi Social Science Foundation (2017S024), Shaanxi Provincial Youth Star of Science and Technology in 2016.

Author information

Authors and Affiliations



YR contributed to the overall conceptualisation and analysis plan, synthesised the findings, and contributed to the drafting of this manuscript. ZZ, CS, MS, YS and DZ gathered data and information. GL, DC, TX, RN and TG contributed to the overall conceptualisation and data analysis. GC and JMF contributed to the revision of each section and made substantial contributions to revise the English of this manuscript. All authors have seen and approved the final manuscript before submission.

Corresponding author

Correspondence to Zhongliang Zhou.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests. We declare that Prof. Zhongliang Zhou is a member of the editorial board (Associate Editor) of this journal.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1: Appendix Table 1.

Two level null model on inpatient services utilization of MFA recipients. Appendix Table 2. Test results of hierarchical multiple regression in examining moderation effect.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ren, Y., Zhou, Z., Liu, G. et al. Does inpatient health services utilization vary by remoteness in the medical financial assistance population? Evidence from Shaanxi province, China. BMC Health Serv Res 20, 1051 (2020).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: