Understanding the Impact of Community Family Physician Contracting (CFPC) on Community Medical Resources Consumption: A Case Study from Beijing in China

,


Introduction
Te universal health coverage, accessibility, and afordability of health services global health objective has been established by the World Health Organization (WHO).Health needs have become more complicated and diversifed as a result of population aging and changing illness profles, providing new problems for the medical and health care system.As a reaction, the WHO put out the conceptual framework for an integrated healthcare system that is based on primary healthcare principles [1] and people-centered integrated healthcare (PCIC) [2].Te WHO views primary care and two-way referrals between various levels of health care institutions as the key elements of an efective health service system, and these frameworks place emphasis on the fundamental role of primary health services in the whole medical and health service system.Family physician contracted to service is the key to the comprehensive establishment of a hierarchical diagnosis and treatment system and the construction of a high-quality, integrated, and effective healthcare service system, according to China's Healthy China 2030 planning outline, which was released in 2016 in response to the WHO's PCIC global strategy [3].
Te whole-population and whole-cycle health services provided by the Chinese government place a signifcant emphasis on the implementation of CFPC.Family doctors serve as the community's health gatekeeper, providing whole-life health management services to people in order to address the two fundamental goals of equitable access and systematic interconnection of national health services.Te implementation of graded diagnosis and treatment has also become a key component of CFPC promotion at the same time.China's family physician contracted service system was ofcially implemented in 2016 with the release of the guidance on promoting family physician contracted services, and as of now, the 1 + 1 + 1 model in Shanghai, the integrated medical and nursing care model in Hangzhou, the three divisions under the common management model in Xiamen, and the contractual general practitioner team-based service model in Wuhan and Shenzhen are the typical service forms (National Department of Health).Te widespread implementation of CFPC in China is still relatively new compared to international practice, the supporting primary care system has not yet been implemented nationwide, and there is still a dearth of empirical evidence on the role of family physicians in the prudent allocation of medical resources despite the fact that the success of these typical models serves as an example of the full-scale rollout of CFPC in China.
Family doctor services are an essential part of basic healthcare and are a widely accepted policy by nations throughout the globe to safeguard citizens' rights to life and health and enhance population quality.Tey also help patients have a more solid and ongoing relationship with their physicians.More than 60 nations and areas throughout the globe have family doctor contracting services in place as of now.According to empirical data from these nations, people who have signed up for family physician services tend to use healthcare services more frequently and also tend to receive primary care that is of higher quality, especially for patients with chronic illnesses.Te contracted group also has signifcantly higher levels of health and daily care quality than the uncontracted group [4].In addition, some research studies contend that universal family physician contracting might lessen health disparities in old age brought on by economic diferences and assist close the health gap across various socioeconomic groups [5].
Tere are two major bodies of work that focus on the family physician contractual system and residents' primary care visits in China.Te frst branch of the research, which is often based on interviews and surveys [6,7] focuses on the efect of CFPC on residents' willingness and satisfaction with primary care.Te second body of research focuses on how the CFPC afects residents' overall medical costs.Te majority of studies reveal that contracted residents' medical costs are lower than those of noncontracted residents.Researchers claim that increasing the proportion of primary care visits will reduce contracted residents' medical costs but they do not ofer empirical evidence to support these hypotheses.Instead, they rely on data on medical costs to draw their conclusions.Te Anderson model, which has been utilized extensively in the study of health care use accessibility [8][9][10][11], is a traditional framework for assessing individual health service consumption behavior.Based on the monthly settlement data provided by community hospitals in Beijing, this study uses three dimensions of propensity characteristics, enabling resources, and medical need to visually analyze the impact of CFPC on the medical expenses of residents in the community hospitals in order to examine the role of CFPC in promoting hierarchical triage and optimizing the allocation of medical resources.

Study Design and Data Collection.
Using quantitative research techniques and data on real medical payment data for residents given by a community hospital in Beijing, this study aims to evaluate the efect of family doctor contracting on residents' medical costs.Beijing made the following proposal in 2010: "strengthen the service function of primary health care institutions and promote family physician service."With CFPC's ongoing exploration, Beijing's community resident contracting rate has been rising.According to the assessment index of family doctor contracting services in Beijing in 2022, the contracted rate of the city's resident population is anticipated to reach more than 45% at the conclusion of the 14th Five-Year Plan, up from the contracted rate of 35% for residents in the city in 2018.Terefore, it is appropriate to research how the CFPC afects Beijing residents' use of primary healthcare resources.Te information used in this research came from a community hospital in Beijing, and it included information on 41 months of resident payments from January 2018 to May 2021 as well as records of every resident's consultations.5816 patients were eventually included in the sample after removing the consultation records, patients with missing fundamental information such as age and gender, and patients without medical data in one year of the statistical year, and a total of 238,456 valid data were acquired.

Dependent Variable.
Te medical expenditure (ME) of patients at a community hospital in Beijing served as the study's dependent variable.ME was determined by adding up all of the patients' monthly expenses at the specifc community hospital from January 2018 to May 2021 who had visit data.Patients lacking medical records for one continuous year were eliminated to further verify data continuity.

Independent Variable.
According to the fundamental data provided by community hospitals, the patient's contracted status is the independent variable.Whether the patient is contracted to a family physician is used as the indicator, and the contracted patient is given a value of 1 and the uncontracted patient is given a value of 0.

Control Variables.
In this study, Anderson's model is used as the basic framework, and in terms of propensity characteristics, the information of patient's age and gender is mainly referred to; in terms of enabling resources, the sample data mainly involve fve types of urban employees' medical insurance, public medical insurance, urban 2 Health & Social Care in the Community residents' medical insurance, new rural cooperative, and foreign medical insurance, each of which applies to diferent reimbursement ratios; patients' need factors are mainly referred to whether they sufer from chronic diseases; meanwhile, this study considers the frequency of visits and the frequency of consultation and the number of departments can be used to measure patients' recognition of the service quality of community hospitals, which is also the environmental factor in Anderson's model, so the frequency of consultation and the number of departments are included in the need dimension.Te specifc sample selection and assignment are shown in Table 1.

Statistical Analysis.
Medical spending data are a common example of zero-infated data, which also goes by the name semicontinuous data since there are a lot of zero values and the remaining values follow a continuous distribution.
According to Huang and Gan [12], the zero values in the medical cost data may be the result of actual medical costs (i.e., no medical expenditures were spent in the current time) or they may be the consequence of patients' ostensible discontinuation of medical treatment.Te data cannot accurately refect the genuine ME of patients if the zero value occurrence is due to patient self-selection.It also violates the OLS model's premise of the normality of random errors, leading to biased estimators [13].
Te two traditional methods for processing semicontinuous data are the two-part model and the Heckman two-stage model.Te two traditional methods vary in how null values are handled and the underlying premise assumptions [14,15].Te two models used in this study's frst phase look at whether residents opt to join up for family doctor services, and the second phase looks at the medical expenses recorded after signing up.Te two models difer in that they contend that there is no connection between the two stages of the decision-making process and that various mechanisms afect residents' decisions regarding whether to enroll and receive medical care at the community hospital.Te Heckman sample selection model, on the other hand, contends that the second component of the medical care expenditure at the community hospital afects whether or not residents choose to sign up for family doctor services and that whether people sign up or not and medical care expenditure are related.Te access choice model may address the self-selection bias issue with the medical cost model since there is a strong association between residents' enrollment decisions and medical expenditures.At the same time, the two models interpret value 0 diferently.Te Heckman twostage model contends that value 0 medical costs may be the result of selection bias, whereas the two-part model views value 0 medical costs as the genuine value.Te fact that the data in this paper are drawn from a community hospital in Beijing's system billing records helps to some extent ensure the sample's contracting status is distributed randomly (we think this is the case for patients whose visits to the hospital have visit data).To assure the reliability of the regression fndings, estimate in this study uses both the two-part model and the Heckman two-stage model.Te two-part model below utilizes a generalized linear model (GLM) in the second part to assess individual healthcare costs while the frst part uses a probit model to simulate the likelihood that residents would use community hospital health services.In our study, the Heckman two-stage model is applied; frst, the Heckman selection model is used to assess whether patients decide to sign a contract or not, and second, the Heckman outcome model is applied to assess the ME sufered by patients who decide to sign a contract and seek treatment in community hospitals.In our research, the statistical analysis was done using Stata16.0.

Two-Part Model.
Te two-part model in this paper divides patients' medical expenses in community hospitals into two stages.In the frst stage, the probit model is used to estimate the probability of patients signing up, i.e., to determine whether the explanatory variable is zero, while the generalized linear model (GLM) is used in the second part to analyze the efect of each characteristic factor on the expenses.
Te frst part is called the selection model: where the random perturbation term ε it o obeys the standard normal distribution if the sample has signed I i � 1; otherwise, I i � 0.
Te second part is the expenditure model: where the random perturbation term ε it ∼N(0, σ 2 ), Cov(ε it , μ it ) � 0. In the two-part model, the zero-valued medical cost data and nonzero are separated and it is assumed that the behavior of patients consuming medical services, whether they consume or not, is independent of the amount of consumption that occurs.Te dependent variable is nonzero medical costs, the explanatory variables are the same as those in the sample selection model, and the main coefcient of interest in this paper is β 1 , i.e., the efect of contracting on patients' health care expenditures in community hospitals.

Heckman Sample Selection Model.
Te Heckman twostage model used in this paper, the frst stage of which is a choice model, examines the factors infuencing patients' choice to contract family physician services.
where the explanatory variable contract it is binary variables that take the value of 1 if patient i is contracted in period t and 0 if not contracted.Explanatory variables include the patient's age (age), gender (gender), and other propensity characteristics, as well as indicators in terms of enabling resources health insurance type (type), the number of monthly visits (times), number of monthly visits to departments (numbers), and whether the patient is chronically ill (chornic), and other demand factors.v t and ε it represent the explanatory variable parameters and error terms, respectively.Te second stage, the scale efect, examines the factors that infuence the healthcare expenditures of contracted patients in community hospitals.fee it � β 0 + β 1 age it + β 2 chornic it + β 3 male it + β 4 times it Of these, fee it is the patient's health care expenditure in the community hospital, and Mills it is the inverse Mills ratio obtained from the frst-stage model estimation, thus controlling the sample selection bias problem.Te reason for the substitution of Mills it is that this indicator contains unobservable information from the frst-stage model and helps to correct for possible sample selection bias in the second stage.If the inverse Mills ratio is statistically signifcant, it indicates the existence of sample selection bias, and the results of the reference Heckman two-stage model are more reasonable than the results of the traditional OLS regression.

Basic Characteristics of Residences.
Tis study used sample data from patients who had consultation records in community hospitals and had at least one visit per year during the observation period (2018-2021).A total of 5816 patients were eligible for inclusion in the sample, and a total of 238,456 monthly medical resource consumption data for all samples were included.Te data source for this paper is a community hospital in Beijing, which covers 13 communities with a total resident population of 33,000.3,686 of them have contracted, which is a 63.4 percent rate.Table 2 displays the fundamental details of the sample's patients.Te sample's patients ranged in age from 56.53 to 19.09 years, with 43.7% being men and a generally equal distribution of the other genders.42.9% of the sample's patients had chronic conditions, which is consistent with a previous research that found patients who were in poor health and had a regular source of treatment seen more often [16].Te lowest and maximum values of ME, specifcally for data on patients' healthcare resource consumption, varied widely, with a standard deviation as high as 817.5, suggesting a wide range in cost per person each visit.Te greatest frequency of visits per month was 34, with the number of visits per month varying widely as well.
Table 3 displays the between-group mean test for each variable using the noncontracted patients in the sample as the control group in order to more accurately compare the willingness to use medical resources in community hospitals between patients who have signed up for family physician services and those who have not.Te data demonstrate that ME in community hospitals was signifcantly higher for patients who had signed up than for patients who had not, comparing the basic conditions of the two groups.Tis fnding ofers preliminary evidence that the family physician Health & Social Care in the Community contract service can increase residents' willingness to visit primary health services like community hospitals.Te average age of both groups was over 50, indicating a higher tendency for the elderly group with the relatively higher frequency of medical visits to seek treatment at the primary level.In contrast, the average age of the contracted group was 59.97, close to 60, indicating a higher willingness to contract among the elderly group.Te proportion of chronic disease patients among the contracted residents was 52.7%, which was signifcantly higher than that of the noncontracted group (25.8%).Tis fnding suggested that family physician contracting increased this group's willingness to use community hospitals' medical resources while also suggesting that chronic disease groups that required regular medical consultation were more responsive to CFPC.Te frequency of consultation and the number of departments visited by contracted patients were much greater than those of the noncontracted group, according to statistics on the demand for community medical services of both groups.Te frequency of consultation and the number of departments of patients who have signed a contract are much greater than those of patients who have not signed a contract, according to demand statistics for the two categories of community medical services.Te diference in community medical service demand between the two groups demonstrates how family physician contract service encourages locals to use more primary healthcare resources.Between the two groups, there were statistically signifcant variations in the fundamental patient and ME circumstances.

Te Impact of CFPC on the Medical Consumption of
Residents.Te coefcient of family physician contracting on the cost of patients' visits to community hospitals was signifcantly positive after controlling for basic patient characteristics such as age, gender, and the presence of chronic diseases.At the same time, contracting was also positively and statistically related to the frequency of patients' visits to community hospitals and the number of departments visited.Tis means that CFPC may successfully enhance patient care continuity by contracting with community residents and general practitioners, which not only dramatically increases medical treatment behavior in primary health service institutions but also solidifes the contractual link between community residents and general practitioners.From Table 4, the regression results are in line with other empirical studies that used data from primary health service visits by community members in Beijing.In those studies, family physician services that were contracted had a big impact on how likely residents were to visit community hospitals and how willing they were to make frst visits [17,18].

Te Factors Infuencing Residents' ME.
Te outcomes of the two-part model are used in the second portion of this article to show how much CFPC contributes to residents' ME at community hospitals.By removing the mutual interference between the two-choice processes of family physician contracting and community hospital visits, we will utilize the Heckman two-stage model in this part to further evaluate the variables afecting residents' ME in community hospitals.
Table 5 displays the fndings from our frst analysis of the variables afecting patients' ME in community hospitals throughout the whole sample.In this research, an OLS regression model and a random efect model regulating the time efect are utilized to examine the explanatory potential and logic of the Heckman two-stage model.Te OLS regression results are shown in column (1) of Table 5, the RE regression results controlling the time efect are presented in column (2), and the regression results of the Heckman twostage model are presented in columns (3) and (4), respectively.Te sign and signifcance of the variables for the OLS regression and random efects model are somewhat diferent from those of the Heckman two-stage model as can be seen from a comparison of the regression results in columns (1), (2), and (4).Te high signifcance of the inverse Mills ratio further indicates that there is self-selection bias in the sample, and the results with reference to the Heckman model are also somewhat diferent.Te fndings of the Heckman two-stage model in columns (3) and (4) provide the major foundation for the discussion that follows.
Te Heckman selection model's estimate fndings reveal that inhabitants in the sample have diferent propensity traits, enabling resources, and requirements.Age had a signifcant positive coefcient, indicating that residents' willingness to participate in CFPC increases with age.Tis is consistent with the special needs of older adults for continuity of care and continuity of relationship with the doctor as noted in studies [19,20].Age and gender both had a signifcant efect on the choice of CFPC from the perspective of propensity characteristics.In terms of enabling resources, the type of health insurance showed a signifcantly  Health & Social Care in the Community positive relationship, showing that patients with higher reimbursement ratios are more likely to choose CFPC and are also more likely to make ME after signing a contract.According to the theory, people's desire for health drives their need for medical care, and they tend to base their decision on their chosen degree of health stock on the cost of such services [21].In addition, the standard medical choice model and the RAND experiment, which contend that patients' willingness to use medical services would rise when the price declines substantially [22], are both consistent with the large impact of out-of-pocket costs on patients' ME.Te larger willingness of patients with chronic illnesses to sign up is due to the policies connected to CFPC focusing on patients with chronic diseases, which has a bigger infuence on the demand side of the equation [23].In addition, patients with chronic diseases who need accessibility and continuity of treatment for their own healthcare requirements have shown a propensity to contractualize their ties with doctors in order to strengthen such connections [24].Te fndings of current empirical testing are also compatible with the considerably positive chronic illness indicator [25,26].Overall, all three indicators of the demand dimension show a signifcant positive efect on patients' contracting decisions, indicating that for patients in community hospitals, individual demand for medical services was the main factor examined to measure contracting decisions.Te frequency of visits and the number of departments visited are also signifcantly positive indicators.
In community hospitals, contracted residents' ME per visit tends to rise with age, according to the Heckman scale model, which demonstrates that age has a strong positive efect on this statistic.On the one hand, such older persons have a high frequency of trips to community hospitals and a strong demand for their services, which results in a high ME per visit.On the other hand, it shows that the hardware and medical conditions of neighborhood hospitals can accommodate the everyday healthcare requirements of the elderly, which supports the growth of neighborhood-aged healthcare.Te reimbursement ratio of contracted patients plays a signifcant role in the ME of community hospitals, and the reimbursement ratio of various medical insurance types becomes a crucial factor for benefciaries seeking  ( Table 6 reports the results of the two-part model regression using the quarterly means of the number of visits and the number of departments visited as proxy variables.Table 7 presents the regression results for the random efects model and the Heckman two-stage model.Te results presented in the table indicate that after changing the main variables (times and numbers) to control for basic characteristics such as patients' age and gender, CFPC remains signifcantly and positively associated with residents' ME in community hospitals at the 1% signifcance level.Meanwhile, age, chronic, and type (type of health insurance) remained the main factors afecting patients' willingness to contract and interfering with patients' choice of MRC after contracting.In other words, the regression results did not change substantially after the substitution of key variables, indicating that the previous study fndings remain robust.

Endogeneity Test.
In addition, this research runs a sensitivity test on the response rate of CFPC in order to take into account the potential for reciprocal causation between the explanatory factors and the explained variables.Contracting behavior in the current era has an efect on the ME in community hospitals in the next period because it infuences residents' MRC decisions in the present and their capacity to use primary care resources in the future.As a result, in this section, we examine the relationship between CFPC and ME by substituting the delayed one-period contracting response rate for the current period's contracting status.Based on the OLS model, random efects model, and Heckman two-stage model, Table 8 shows the link between contracting behavior in the delayed era and ME in the present period.Te similarity of the regression results to the benchmark regression structure previously indicated further validates the accuracy of the fndings.

Disscussion
Te CFPC realizes the continuity of medical and health services through the long-term contractual relationship between doctors and patients [27], and this efcient two-way information transfer mechanism is conducive to improving the trust of contracted residents in family physicians, thereby promoting primary care and hierarchical triage.Te majority of research on CFPC and residents' consultation behavior has been on residents' overall ME and desire to attend primary care.Does hiring family doctors, nevertheless, encourage resident consultation at local hospitals?In addition, in its guidance on family doctor contracting services, the China Health and Wellness Commission identifed special populations as key populations, including the elderly and the chronically ill, and emphasized the need to concentrate on standardized management and health services for patients with major chronic diseases (guidance on promoting the high-quality development of family doctor contracting services, 2022).Consequently, this essay will further investigate whether the development of a fxed contract relationship with family physicians improves the daily healthcare and medication collection of important groups such as the elderly (≥65 years old) and patients with Health & Social Care in the Community chronic diseases for these groups in community hospitals.As early as 2010, Beijing, one of the locations in China that had already adopted CFPC, put into practice the family physician contract model that had been tested in Dongcheng, Xicheng, and Fengtai districts [28].Tis research conducted an empirical analysis of the efect of contract signing on residents' ME at community hospitals using billing data from community hospitals in Beijing.Te fndings indicate that compared to the noncontractual population, contracted residents in community hospitals had MEs that are considerably higher.Eforts should be undertaken to enhance CFPC coverage for important groups and raise the percentage of primary care visits for key populations, according to the guidance on promoting family physician contracted services published in 2016.Is there a diference in ME between the important population in this instance and other groups at community hospitals after the contract was signed?

Medical Expenses of Community Hospitals for Contracted
Residents.Te majority of domestic studies support the benefcial impact of contracting on boosting the willingness of key populations to attend primary hospitals with regards to the impact of CFPC on the usage of key populations, such as those with chronic conditions, in primary health care [25,29].Te following random efects model, which controls for time efects, is used to break down the ME and their infuencing factors for various subgroups in the contracted group in order to assess the impact of contracting on the utilization of community hospital services for key populations from the perspective of ME.Table 9's column (1) displays the relationship between ME for all contracted patients in community hospitals and the three-dimensional variables of forward-leaning factor, enabling resources, and demand; column (2) displays the relationship between ME for the contracted group of the elderly population (65 years and above); and column (3) displays the relationship between ME of the group and the three-dimensional variables of the forward-leaning factor, enabling resources, and demand.Te information in column (3) demonstrates the association between the ME of the group of people under 65 and the variables of demand, enabling resources, and forward-leaning factor.Te information in column (1) demonstrates that both parameters are strongly and favorably related to ME for the contractual group.Te comparison of the data in columns ( 2) and ( 3) also reveals that the contracted patients" frequency and number of department visits are signifcantly higher than those who are younger than 65; thus, the elderly group, which has more health care needs after contracting, is more willing to visit community hospitals than the contracted patients of other age groups.Te regression results are in line with empirical fndings from prior research using stepwise regression, according to which older patients are much more likely to pick community hospitals for routine medical treatment after enrolling in CFPC [30].Meanwhile, Liu et al.'s fndings that there is a substantial association between ME and the medical insurance reimbursement ratio among the Chinese senior population are in line with the considerably positive coefcient of the type of medical insurance among the people who signed the petition.
In Table 9, column (1) displays the relationship between the three-dimensional variables of the forward-leaning factor, enabling resources, and demand and the ME of all contracted patients in community hospitals; column (4) displays the relationship between the three-dimensional variables of the forward-leaning factor, enabling resources, and demand and the ME of contracted chronic patients; and column (5) displays the relationship between the three-dimensional variables of the forward-leaning factor, enabling resources, and demand.Te link between the ME of patients with newly acquired nonchronic diseases and the three dimensions of prospective variables, enabling resources, and demand is shown in column (5).Te comparison of the data in columns ( 4) and ( 5) further reveals that among contracted residents, patients with chronic diseases attend departments more often and on average more frequently than patients without such conditions.Te empirical fndings in Table 9 support Le et al.'s assertion that chronic illnesses would sharply increase the number of visits to the residents.Table 9's fndings show that, on the one hand, elderly and chronically ill patients are more likely to sign up because of their greater reliance on healthcare services and need for continuity of care, and that, on the other hand, the connections made between patients and doctors by CFPC further increase these two groups' willingness to choose community hospitals for consultation, proving that CFPC can successfully promote patient triage, prevention, and treatment.Te aforementioned fndings demonstrate that the family doctor contract service can assist in achieving "Healthy China 2030"'s primary goals of increasing life expectancy per capita and reducing mortality.It also has a positive impact on solving the health problems of the elderly, chronic disease patients, and other key populations.

Impact of COVID-19 on the Medical Expenses of Contracted
Residents.Family doctors have been crucial in grassroots epidemic prevention and management since the COVID-19 pandemic broke out in 2020.CFPC has been essential in protecting residents' life and health safety and satisfying the population's rising health needs, whether during or after epidemics.In the postepidemic period, regions have successfully pushed the coverage of family physician services and the frst consultation at community hospitals in order to guarantee that residents' demands for medical and health services are satisfed and to lower the risk of infections at hospitals.
Te visits of contracted residents to community hospitals between January 2018 and December 2019 and January 2020 and May 2021 are shown in Table 10.Te visits to community hospitals before and after the outbreak for all contracted residents are shown in columns (1) and (2) of Table 10, whereas the visits for two important populations-the elderly (65 years and older) and patients with chronic diseases-are shown in columns (3)-( 4) and ( 5)- (6).We focused on the frequency () and the number of departments () that residents visited in community hospitals before and during the pandemic to investigate whether CFPC had a role in boosting primary care during epidemic prevention and control.Te fndings in Table 10 demonstrate that after the epidemic, visits to community hospitals by all contracted residents increased signifcantly as did the number of departments visited.For the key populations, contracted elderly and chronic patients made more visits to community hospitals after the epidemic, and both the frequency and number of departments visited increased in comparison to the number before the epidemic.According to research on the family physician services ofered and residents' readiness to enroll in the postepidemic period in Wenzhou, contracted residents were more likely to seek medical advice from community hospitals after the epidemic [31].Te regression analysis's fndings further imply that CFPC helps promote primary care for important groups.

Strength and Limitation.
Te willingness of residents to get into contracts and the efects of contracting on residents' consultation behavior and ME have been the main topics of discussion on CFPC.Tere is also disagreement about whether the Chinese family physician system may act as a medical cost gatekeeper by directing graded healthcare despite international research, suggesting that family doctors function as both health gatekeepers and medical cost gatekeepers [32].Tere is still a relative lack of the pertinent empirical evidence despite the fact that the existing literature generally agrees that the CFPC currently implemented in China reduces overall ME by encouraging contracted populations to seek treatment for minor illnesses and routine care in primary health facilities like community hospitals [33,34].Trough the use of billing information and patient records provided by community hospitals in Beijing, we examine the relationship between contracting and patients' ME in this study.We then provide empirical support for the role that family physician contracting has played in the development of a new medical care pattern that places minor illnesses in the community, serious illnesses in the hospital, and recovery in the community.Te research also analyzes the contracted group's trips to community hospitals before and after the pandemic and fnds that family doctors have a bigger role in encouraging primary care visits during the postepidemic era.
Te study shows that the family doctor contracting service system in China signifcantly contributes to the development of an efective medical service utilization system of "primary care for the frst diagnosis" and "two-way referral," and it ofers empirical support for continued use of family doctor contracting service as a key strategy to realize "Healthy China 2030."Te utilization of family doctor contracted services as a crucial strategy to realize a "Healthy Health & Social Care in the Community China 2030" is also supported empirically.Achieving the goal of "Health for All" as outlined in the Alma-Ata Declaration from 1978 will depend on the population's health quality improving and health disparities between various economic and social groups decreasing as a result of the rise in family doctor contracting rates.Tis research has a number of drawbacks as well.Not all of the fundamental features of inhabitants, such as the marital status, education, and self-rated health status, were taken into account in the model due to data availability limitations.Due to the lack of data, it was also unable to examine in this study scientifcally whether enrolling in community care would result in fewer visits to other major hospitals and a consequent decrease in the cost of treatment over the course of a patient's lifetime.Only the monthly visit frequency and the department visited were used in the model to estimate the individual demand for community hospitals since the precise medication usage and treatment plan heavily rely on the residents' real sickness state and treatment outcomes.To better understand the role, shortcomings in development, and potential future directions of family doctor contracted services in China, we will conduct a more indepth analysis of the health performance of family doctor contracted services in conjunction with data on medical cost settlement in the upcoming study.We will collect as many demographic indicators of the sample as possible through questionnaires and interviews.

Conclusion
Our analysis shows that the CFPC in China has successfully improved the frequency and variety of departments that contracted residents visit in community hospitals, hence raising the residents' ME in these facilities.For some groups (elderly and patients with chronic illnesses), developing a solid connection with a family doctor enables this group of people to develop a habit of initially seeking treatment in the neighborhood, supporting the sensible distribution of medical resources.It is benefcial to encourage residents to join a community hospital primary care model because it helps to establish an efective and organized pattern of healthcare.On the other hand, it also helps to regulate the overall ME and guarantee that residents get timely medical attention.In order to further boost the acceptance and effcient usage of CFPC, we should implement a comprehensive spectrum of policy support measures and improve publicity and education.

Table 1 :
Description of relevant variables.

Table 3 :
Between-group mean diference test by sign.

Table 5 :
Regression results of the Heckman two-stage model.

Table 4 :
Regression results of the two-part model.represents the resident number i and t represents the t th quarter; the number of visits and the numbers of departments visited by patients in the three months of the t th quarter after the variable substitution was performed were replaced by the mean value.

Table 7 :
Regression results of the Heckman two-stage model after substituting the variables.

Table 8 :
Regression results of the Heckman two-stage model for lagged period data.

Table 10 :
Association between COVID-19 and medical expenses among the contracted groups.

Table 9 :
Infuencing factors of medical expenses among the contracted groups.