Choice of and Equity in First-Contact Care in China’s Zhejiang and Qinghai Province: A Comparative Study

The imbalance between health needs and healthcare resources has become more serious in China. The hierarchical diagnosis and treatment system has played an important role in solving the problem, one goal of which was to increase the use of basic-level medical facilities. This study aimed to compare the choice of and equity in rst-contact care among residents in Zhejiang and Qinghai provinces, China. First-hand data were obtained through a cross-sectional survey of residents of Zhejiang and Qinghai. We rst compared the proportion of residents who select primary healthcare institutions as the setting for rst-contact care. Then we compared horizontal equity using the concentration index decomposition method. An urban-rural subgroup analysis was carried out for both provinces.


Background
Primary healthcare institutions (PHIs) play an extremely important role in China's National Medical and Health Care System (NMHCS), a situation that is considered the foundation of an effective health system [1,2]. It is the responsibility of PHIs to tackle frequently encountered medical conditions, common minor diseases, and chronic conditions [3]. However, PHIs are still underutilized in China [3]. With the development of society and the improvement of living standards, people nd it more convenient to seek health care services by walking into one of the hospitals. Both PHIs and hospitals can be directly accessed by walk-in patients [3]. Meanwhile, given that there are more professional personnel and specialized resources in large hospitals, more patients would select large hospitals as their rst-contact care provider regardless of their health issue. The freedom of patients to select institutions has caused a great dilemma for large hospitals and PHIs. Large hospitals are crowded, whereas PHIs have few patients. This situation is referred to as the "inverted triangle" phenomenon of health resource allocation [1].
To alleviate this dilemma, the National Hierarchical Medical System (NHMS) was formed as an important part of the New Medical Reform in China in in 2009 [4]. The management, personnel, and health services of PHIs were entirely funded by the government [5]. It was proposed that improvements should be made in the structure of the health care system in order to promote the utilization of PHIs [6-8]. In the NHMS, patients must rst seek care at a PHI. Then with a referral from PHI, they can seek care at a hospital.
Patients who seek rst-contact care at a hospital without having been referred by a PHI, receive less reimbursement from their medical insurance provider. Meanwhile, a dual referral system was also launched, which enhanced the accessibility and equity of health care between PHIs and hospitals [3]. The implementation of the NHMS and other relevant measures have facilitated the utilization of PHIs.
However, there are signi cant differences in the proportion of residents selecting PHIs as rst-contact care among the different regions. For example, previous studies have found that the proportion of individuals who choose to seek rst-contact care from a PHI was 63% for Xiamen, 74.0% for Jilin, 60.2% for Shandon, 52.3% for Chengdu, and 82.1% for Xuzhou [9][10][11][12]. There is a gap between the proportion in several regions in China and the World Health Organization recommended proportion of above 80% [13].
China is a vast land with great differences in socioeconomic development among its regions; therefore, we were interested in the differences in choice of rst-contact care in regions with different levels of development. We chose two study sites with entirely different levels of development to compare the proportions of residents who selected primary health care institutions as their rst-contact care provider.
Zhejiang was chosen as the developed province, and Qinghai was chosen as the underdeveloped province.
Equitable access to healthcare services that meet health needs regardless of socioeconomic status (SES) is an import goal of the health care service system [14]. Because medical insurance reimbursement plays an important role in the NHMS, people with different levels of socioeconomic status can be in uenced to different degrees. Prior studies, primarily on associated factors, have revealed that the choice of rstcontact care is signi cantly affected by such factors as region type (urban or rural), income level, and distance from a PHI [1,5,9,13]. However, there is a lack of research on the relationship between health needs and the choice of rst-contact care. Equity with regard to the choice of rst-contact care for residents with similar health needs but different levels of SES and differences in other variables have yet to be examined. Equity is the core of primary health care; horizontal equity refers to the situation where people with the same health needs have the same access to health care. This concept has been applied in the delivery of health care in Europe and the United States [15,16]. The horizontal inequity (HI) index was rst proposed by Wagstaff et al.; the HI index can be used to determine whether, with the in uence of socioeconomic status and other variables, the utilization of health care is equitable for residents with similar health needs [15].
The objective of this study was to compare the choice of and equity in rst-contact care in Zhejiang (a developed province) and Qinghai (underdeveloped province). First, the proportions of residents choosing PHIs as their rst contact provider were compared. Next, the levels of equity were compared using HI.
Because China has an urban-rural dual economic structure, for each province, sub-comparisons of the choice of rst-contact care and level of equity in choice of rst-contact care between rural and urban regions were also carried out for each province.

Data source and sampling
A cross-sectional study was conducted in Zhejiang and Qinghai during a 1-year period between August 2016 and August 2017. Two counties in Zhejiang and three districts in Qinghai were selected using a multi-stage strati ed cluster random sampling method. The selection of study sites was based on two criteria: 1) the counties (districts) represented different levels of socioeconomic development; 2) the governments of the counties (districts) were able to support our research. In each county (districts), 10 neighborhood committees of one street were randomly selected (towns), or 10 administrative villages were randomly selected in each township (rural areas) [17].
Our questionnaire was developed based on the high-quality questionnaire of the National Services Survey, which contained several items on residents' sociodemographic information, health status (self-reported health, chronic diseases), health care utilization, and whether a PHI was their choice of rst-contact care [18]. Prior to conducting the eld surveys, all investigators were trained on the research protocol and the administration of the questionnaires. This study was conducted in accordance with the Declaration of Helsinki.

Data collection
The collection of data was conducted during the period of the eld investigation. We contacted with the governments of Zhejiang and Qinghai provinces to obtain the support of local health departments. Prior to conducting the eld surveys, all investigators were trained on the research protocol and the administration of the questionnaires. Any family not responding after three attempts was replaced by another randomly chosen family in the same neighborhood. If the residents were younger than 16 years old, their survey would be completed by their parents or other family members at least 16 years of age (For accuracy).

Indicator and variables
Choice of rst-contact care was applied as an indicator in this study, and was a dichotomous observed dependent variable (1 = selected PHI; 0 = did not select PHI).
In this study, variables were determined and classi ed based on previous studies and Andersen's behavioral model, to ensure that all possible in uencing variables would be fully considered [9, 11-13, 17, 19-21]. Age, sex, chronic disease status, and self-reported health score were classi ed as health needs variables. Income was used to assess SES. Because insurance status, education level, employment status, marital status, and distance from the nearest PHI had been previously reported to be possible in uencing factors of choices of rst-contact care, they were included in this study as other variables.

Data analyses
Descriptive analysis and χ 2 -tests were applied to compare the proportions of choices of rst-contact care by related factors for both Zhejiang and Qinghai.
There are some misconceptions with regard to equality and equity. Equality in health refers to a state in which each person has the same level of access to health care resources; this can be measured with the concertation index (CI) [17]. Two concepts related to equity in health are horizontal and vertical equity. Horizontal equity refers to a state in which people with the same health needs have access to the same health care resources, and vertical equity refers to a state in which people with more health needs are provided with more health care resources [15,16]. Horizontal equity is widely measured using the methods developed by Wagstaff and Doorslaer.
Horizontal inequity can be calculated by subtracting the contributions of the health needs variables to CI. Because the indicator was dichotomous, a probit model was adopted to calculate marginal effects for each factor. The formula is as follows [22]. Where y i is the indicator of choices of rst-contact care, β m j and γ m k are the partial effects, dy/dx j and dy/dz k of each variable treated as xed parameters and evaluated at sample means; x i denotes health needs variables, z i denotes other variables, and the μ i is the implied error term, which includes approximation errors [22,23].
Because the Eq. (1) is linearly additive and the decomposition result can be applied, the CI for y is as follows [22]: Where CI is nonstandardized, C j is the CI forx j , C k is the CI for γ k , and C μ is the CI for error; x j denotes the mean of x j , z k denotes the mean of z k , and μ denotes the error [23].
HI index can be calculated by subtracting the contributions of health needs variables from the nonstandardized concentration index. Based on the Eq. (2), the formula is as follows [22]: 3. Results

Household characteristics and their choices of rstcontact care
A total of 4899 residents were enrolled in this study (2872 from Zhejiang and 2027 from Qinghai). A signi cantly higher percentage of residents from Qinghai selected PHIs as their choice for rst-contact care compared with Zhejiang (82.5% vs 54.4% P < 0.001). There were signi cant differences between the two provinces in choice of rst-contact care by income level, education level, employment status, distance from the nearest PHI, and insurance status. There was no signi cant difference between the two provinces in choice of rst-contact care by marital status. Region type (urban or rural) was an in uencing factor among residents of Qinghai; however, it was not found to be an in uencing factor among residents of Zhejiang (Table 1).

Decomposition of the CI index and the value of HI index
When extracting the contributions of health needs variables to CI, the HI index was calculated to assess whether there was horizontal equity in the choice of rst-contact care (i.e., to determine whether residents with similar health needs had similar options with regard to rst-contact care). It was found that the contribution of health needs variables to CI was relatively low (from − 1.64-2.84%). In Zhejiang, income contributed mostly to the CI, with a proportion of contribution of 52.74%. Region type (urban or rural) followed with a proportion of contribution of 34.83%. Income contributed mostly to the CI both in urban and rural Zhejiang, with proportions of contribution of 49.03% and 96.49%, respectively. In Qinghai, the CI was mostly affected by region type (urban or rural; 59.48%), and for both urban and rural Qinghai, the CI was mostly affected by the distance from the nearest PHI (32.52%) and insurance status (36.17%) ( Table 2).
The HI indexes of Zhejiang and Qinghai were − 0.180 and 0.007, respectively. The inequity of Zhejiang was pro-poor; among residents with similar health needs, poorer residents were more likely to select a PHI for rst-contact care. The inequity of Qinghai was mild pro-rich (which was almost equitable, due to the fact that the HI index was very close to 0); with similar health needs, richer residents preferred to select a PHI for rst-contact care. The HI indexes were − 0.088, -0.160, 0.069, and − 0.058 in rural Zhejiang, urban Zhejiang, rural Qinghai, and urban Qinghai, respectively ( Table 2).

Discussion
In this study, a comparison of choices of rst-contact care in Zhejiang and Qinghai was carried out. It was found that the proportion of residents who chose PHIs as their rst-contact care medical institutions of Zhejiang was lower than that of Qinghai. The HI index was used to assess horizontal equity when selecting rst-contact care among residents with similar health needs. Pro-poor inequity in Zhejiang and mild pro-rich inequity in Qinghai were found. Inequity was greater among the poor in rural regions in the two provinces.
In total, the proportion of residents selecting PHI as their rst-contact care in Zhejiang was lower than that in Qinghai. One possible reason for the difference is the gaps in medical insurance between different institution levels. Prior studies reported that medical insurance plays an important role in the NHMS [3,9]. The gaps in medical package bene ts between different institution levels for Qinghai was larger than that for Zhejiang [24,25]. In Qinghai, for outpatients with urban and rural medical insurance (URRMI), the reimbursement ratio was 50% with a ceiling of 120 per capita per year; for outpatients with URRMI, the deductible was 100, 600, and 1500, and the reimbursement ratio was 90%, 80%, and 70% for PHI, secondary hospitals, and tertiary hospitals, respectively [24]. For Zhejiang, using Jiaxing as an example, for outpatients with URRMI, the reimbursement ratio were 40%, 20%, and 10% for PHI, secondary hospitals, and tertiary hospitals, respectively; for inpatients with URRMI, the deductibles were 300, 500, 1000, and the reimbursement ratio were 80%, 75% and 65% for PHI, secondary hospitals, and tertiary hospitals, respectively [26]. The larger gaps in deductibles and reimbursement in Qinghai may encourage resident to use PHI more. In addition to the difference in medical insurance package, Qinghai was at the forefront of implementation of the NHMS in China in 2013, while Zhejiang developed several pilot programs in 2014 [25]. Compared with Zhejiang, residents had more time to get used to NHMS in Qinghai, which might help to increase the utilization of PHI.
With similar health needs, the degree of pro-poor inequity of Zhejiang was signi cant; poorer residents preferred to select PHI as their rst-contact care. In Qinghai, a mild pro-rich inequity can be found, and richer residents preferred to select PHIs. Zhejiang is a relatively developed province in China, with 56.56 million inhabitants; its annual gross domestic product was 7793.69 billion dollars in 2017 [27]. Compared with Zhejiang, Qinghai is underdeveloped, with 5.98 million inhabitants, and its annual gross domestic product was 39.14 billion dollars in 2017 [28,29]. Also, as demonstrated in this study, the income per capita per year in Zhejiang was signi cantly higher than that in Qinghai. Rich residents in Zhejiang may not be as sensitive to the differences in reimbursement ratio for different institution levels, which may be because the differences for them are not as signi cant. For Qinghai, the choices of rstcontact care were almost equitable. The overall income level in Qinghai was moderate. Considering the larger gaps in reimbursement between different institution levels in Qinghai, residents with different SES status may be more likely to select PHI as their rst-contact care equally.
A subgroup analysis of region type (urban or rural) was also carried out. Results indicated that the propoor inequity was more serious in urban regions in the two provinces. In Zhejiang, a pro-poor inequity can be found both in rural and urban regions; however, the degree of inequity was more serious among urban residents. In Qinghai, in rural regions, rich residents tended to select PHI, whereas in urban regions, more poor residents selected PHI. One possible reason for this nding may be the urban-rural income gap, which is about 2.02:1 [30]. Compared with rural residents, urban residents were richer; therefore, they may be less sensitive to the gaps in medical reimbursement between different levels of institutions when they select rst-contact care providers. This situation could result in richer residents preferring not to select PHIs for rst-contact care in urban regions.
These ndings showed that for a country with vast land and a tremendous amount of regional socioeconomic differences, medical reform should be implemented according to local conditions. The focus of the reform should be based on socioeconomic development. For Zhejiang, residents were relatively richer, and they would be less sensitive to the gaps in medical insurance reimbursement, larger gap in the medical reimbursement ratio or other solutions can be employed to increase the utilization of PHI, especially in urban regions. For Qinghai, with similar health needs, the choices of rst-contact care were almost equitable, which may serve as an example for other regions. However, the equitable utilization of PHI in rural regions of Qinghai should be further promoted.
This study had several limitations. First, possible reasons based on prior research and information from government websites were considered to explain the ndings; however, not all possible factors were discussed in the study. Second, residents may have forgotten actions taken in the past regarding their health care, causing recall bias. Because the questionnaires were administered to each household, it is likely that family members discussed these past actions for better recall. Third, there were many proper nouns in the questionnaires, which may cause misunderstanding of residents. We added the explanation of these nouns, and face-to-face questionnaires were carried out to decrease misunderstandings.
There are three major policy implications from this study. First, medical reform policies should take socioeconomic status into consideration, developing different policies for different regions. Second, more focus should be placed on urban regions, in order to increase the equitable utilization of PHIs. Third, larger gaps in reimbursement at different institution levels may help increase the equitable utilization of PHIs.

Conclusions
In conclusion, the present study compared the choice of and equity in rst-contact care in Zhejiang and Qinghai. The differences between rural and urban regions within the two provinces were also explored. In Zhejiang, the proportion was relatively lower with a pro-poor inequity; larger gaps in the medical reimbursement ratio at different institutional levels can be considered to encourage the utilization of PHI. In Qinghai, the proportion was higher than the World Health Organization recommended proportion, and the choices were almost equitable. More effort should be made to promote the equitable utilization of PHI in urban regions of the two provinces.

Declarations
Ethics approval and consent to participate This study was approved by the institutional review board of Zhejiang University School of Public health. Written consent was obtained from all participants, for residents who were younger than 16 years old, their consent would be written by their parents or other family members at least 16 years of age (For accuracy) .

Consent to publish
Not applicable Availability of data and materials Please contact the corresponding author for data requests.

Competing interest
All authors declare that they have no competing interests. Author Contributions XS participated in the eld investigation, was responsible for quality control, training of investigators, data collection and analysis, and the initial version of the manuscript and the revisions; HZ participated in the design of the research study, design of questionnaires, and manuscript revisions; HZ, SG, and XZ participated in the design of the research study, the design of the questionnaires, and the eld investigation, and shared responsibility for quality control, investigator training, data collection, and manuscript revisions; YG, MH, and JW participated in the eld investigation, and shared responsibility for quality control, investigator training, and data collection; HD participated in the design of the research plan, designed the questionnaires, participated in the eld investigation, shared responsibility for quality control and investigator training, and reviewed the manuscript before nal submission.