Sage Journals: Discover world-class research

Abstract

This study explored and compared the associations between pharmacy competition and drug expenditures by individuals with influenza. This study used a dataset consisting of 6 694 534 individuals who purchased drugs for influenza at pharmacies from 2015 to 2019 in China. Patients’ annual average influenza-specific drug expenditures per visit at pharmacies was the outcome variable of interest. Pharmacy competition was measured using the Herfindahl-Hirschman index. A 3-way fixed-effects model combined with a lagged identification strategy was constructed to estimate the association between pharmacy competition and drug expenditures. When the radius of the market was set to 1, 5, and 10 km, for each 10% increase in the degree of total competition in the market, an individual’s annual average influenza-specific drug expenditures per visit fell by 0.65%, 2.21%, and 5.20%, respectively. With a more detailed understanding of the underlying mechanism through which pharmacy competition affects the behaviors of health care providers, competition can be considered as a potential tool to assist decision makers in the design of policies to curtail the growth in drug expenditures.

Keywords

economic competition managed competition pharmacies health expenditures China

What do we already know about this topic?

Most of the extant literature concerning competition in the pharmacy sector has dealt with the impacts of pro-competition reforms and service-level competition.

How does your research contribute to the field?

To the best of our knowledge, we were not aware of any prior literature that has assessed the association between pharmacy-level competition and drug expenditures in China. This project filled this knowledge gap by revealing that pharmacy-level competition reduced drug expenditures by individuals with influenza in China. Particularly, this research found that pharmacies competed for any patient rather than for patients on the basis of a specific line of services or diseases when the market radius was over 3 km and found that competition among pharmacies was more strongly associated with reductions in drug expenditures compared to that between pharmacies and other non-pharmacy facilities.

What are your research’s implications toward theory, practice, or policy?

This study demonstrated that pharmacy competition reduced drug expenditures; competition is thus suggested to be promoted in the retail pharmacy sector. We highlight the need to promote competition between pharmacies and to increase competition between pharmacies and non-pharmacy healthcare facilities (eg, clinics). Specifically, policy makers are recommended to reduce the barriers for entry to the pharmacy sector by simplifying administrative and regulatory procedures. We also recommend that policy makers incentivize qualified incumbent and retired physicians to open clinics and allow such clinics to be covered by the public health insurance programs.

Introduction

As the WHO reported in 2017, drug expenditures accounted for 20% to 60% of total health expenditures in low- and middle-income countries, compared with just under 20% in countries belonging to the Organization for Economic Co-operation and Development.¹ In China, the number of wholesale pharmaceutical companies and retail stores reached 13 000 and 341 000, respectively, in 2007.² Spending on drugs accounted for a staggering 41.9% of total health expenditures (including both private and general government health expenditures) or 2.1% of China’s GDP in 2010.³ Annual drug expenditures reached 1850 CNY (260 USD) per capita in 2019, accounting for about 6% of per capita disposable income of Chinese citizens.⁴ Although the accessibility of drugs has improved significantly over the last 2 decades, high drug prices triggered mounting complaints from the public in China.^5
-7 Since 2009, the Chinese government has outlined a range of market-oriented drug-price reforms. For instance, the “Healthy China 2030” report published in 2016 recommends drug pricing to be based on the confluence of market and regulatory control.⁸ In the same year, the “Thirteenth Five-Year Plan for Deepening the Reform of the Medical and Health System” stresses the need to encourage patients to purchase prescription medications at retail pharmacies.⁹ The goal of this policy is to make retail pharmacies an important channel for the sale and provision of drugs.⁹ The establishment of a market-oriented drug pricing mechanism has been further advanced by a series of very recent policy documents.^10
-12 Although extensive efforts have been expended by the government to promote greater pharmacy competition in China, their impacts on the drug expenditures by individuals remain unknown.

This study is related to 3 strands of literature. The first strand of research has focused on the policy effects of pro-competition reforms carried out in countries across the world. For instance, a German study showed that the regulatory reforms for pharmacies resulted in lower entry cost and a more efficient spatial distribution of retail pharmacies.¹³ A Swedish study estimated the impacts of a deregulation reform that allowed for the entry of several private agents into the pharmacy market, and concluded that the reform led to higher availability for consumers.¹⁴ On the contrary, a study investigated the price reactions of pharmacies to changes made to over-the-counter drug regulations in Germany, showed that price competition between pharmacies was only taking place on a very small scale.¹⁵ Likewise, a study compared community pharmacy systems in 5 European countries, revealing that deregulation in the pharmacy sector was not associated with a lower drug price.¹⁶ The second strand of research has focused on product-level competition and analyzed the impacts of competition between generic and therapeutic drugs.^17
-20 A study conducted in China documented that both generic and therapeutic competition drove the price down.²⁰ This finding was confirmed by a subsequent study in 2014 which demonstrated that competition from generic and therapeutic competitors decreased drug prices in China.⁷ Another 2 studies, however, found that only competition among generics was linked with price reductions, while the impacts of competition among therapeutics were debated.^17,18 The third strand of research has focused on assessing the impacts of institutional-level competition in the pharmacy sector.²¹ Research from developed countries showed some signs of decreased drug prices in connection with more intense pharmacy-level competition.²² Nevertheless, we are not aware of any evidence in support of the beneficial effects of pharmacy-level competition in reducing drug expenditures in China. Retail pharmacies have emerged as a primary source of health care in China following economic and political reforms that dramatically altered the healthcare delivery system.²³ By the end of 2021, the number of retail pharmacies covered by public health insurance (PHI), including the Urban Employee Basic Medical Insurance (UEBMI) and Urban and Rural Residents Basic Medical Insurance (URRBMI), reached 423 000, accounting for over 96% of all healthcare institutions that opted into the PHI arrangements.²⁴ The total drug expenditures reimbursed for individuals under the UEBMI exceeded 221.19 billion CNY (31.07 billion USD) in 2022.²⁵ As such, it would be useful to investigate whether pharmacy competition plays a positive role in cutting rising drug expenditures in China.

In a summary, most of the extant literature concerning competition in the pharmacy sector has dealt with the impacts of pro-competition reforms and service-level competition. In contrast, little is known about the role of institutional-level competition in the pharmacy sector in reducing drug expenditures by individuals. Particularly, there has been a lack of research on the marketplace for the management for influenza disease wherein demand is anticipated to be highly elastic. Hence, the primary objective of this study is to estimate the association between pharmacy-level competition and drug expenditures by individuals with influenza. In China, various types of healthcare institutions (such as pharmacies, hospitals, primary care facilities, and clinics) all serve as the providers of drugs and therefore, all of these facilities were thus considered as potential competitors of a given pharmacy in our study. This indicates that pharmacy competition in our research not only includes competition between pharmacies but also competition between pharmacies and non-pharmacy institutions. As such, the secondary objective of this study is to further compare the association between different types of pharmacy-level competition and drug expenditures by individuals with influenza.

Methods

Design

The theoretical model used to guide this study was reported in Supplemental Appendix 1. This repeated cross-sectional analysis of a retrospective cohort study was conducted in Changde City, which is located in Hunan province, China and has a population of over 5 million in 2019.²⁶ Individual-level data were obtained from the administrative health and electronic medical record databases managed by the Changde Municipal Human Resource and Social Security Bureau, which routinely collect information on all patients with the PHI coverage who visited healthcare facilities that opted into the PHI arrangements. The dataset contains demographic data on individuals covered by public health insurance programs from 2015 to 2019 in Changde city and visit-level data on diagnosis of diseases and drug expenditures. Among various diseases, influenza has been our focus as influenza epidemics occur virtually every year and result in substantial disease, death, and expense (influenza-specific visits also have the largest market share across all disease categories in our dataset).²⁷ The inclusion criteria for our study sample were set as follows: individuals with the PHI coverage who purchased drugs for influenza (identified based on the diagnosis in their medical records) at pharmacies participating in the PHI program from 2015 to 2019. Accordingly, we excluded individuals without a diagnosis of influenza and those who did not purchase drugs for influenza at a participating pharmacy (more details could be found in Supplemental Appendix 2). Consequently, an unbalanced panel sample consisting of 6 694 534 individuals who purchased drugs for influenza at pharmacies from 2015 to 2019 in Changde city was obtained.

In our study, non-pharmacy institutions include primary care facilities (including community healthcare centers, community health stations, and township hospitals), hospitals, clinics (including village and private clinics), and others (such as the disease prevention and control center). As a first step in defining the market catchment area for a given pharmacy, we obtained the location of all places where drugs were dispensed, including 1198 pharmacies, 159 hospitals, 303 primary care facilities, 540 clinics, and 216 other healthcare facilities. We collected the addresses of these healthcare facilities from 3 online platforms that deliver basic registration information on companies in China (https://www.qc.com; https://www.qixin.com; https://www.qcc.com). We then used the “Baidu” map and manually collected the coordinates of all 2416 facilities using the “Baidu Coordinates Extraction System” (https://api.map.baidu.com/lbsapi/getpoint/index.html). To calculate the distance between each facility, we extracted data on the 1:250 000 scale base map of Hunan province and road networks from the China National Earth System Science Data Center.²⁸

Study Variables

Individuals’ annual average influenza-specific drug expenditures (per visit) at pharmacies was the outcome variable of interest. Since the distribution of the drug expenditures is right skewed,²⁹ the dependent variable was log transformed. The distribution of the dependent variable before and after log transforming can be seen in Supplemental Appendix 3. The Herfindahl-Hirschman index (HHI),²¹ the most commonly used measure of market competition, was used to measure the degree of competition faced by each individual pharmacy. The calculation of the HHI for pharmacy j was described in Equation 1: D_j,m denotes the demand experienced by pharmacy j (in our study j ∈ [1, 1198]) in market m; D_m represents the total demand of N number of healthcare facilities in market m. Pharmacy-level competition was calculated based on the number of visits that occurred at healthcare facilities in a given pharmacy’s catchment area. As a wide range of healthcare facilities could dispense drugs (such as pharmacies, hospitals, primary care facilities, and clinics), all of these facilities were taken into account in the calculation of the HHI in the catchment area of a given pharmacy.

Most of the extant literature measured the degree of competition in the healthcare market based on disease- and service-specific (inpatient, outpatient, or drug purchases) utilization.²⁹ However, when a pharmacy enters into a market, it does not specialize in a certain disease category; instead, pharmacies in China are pretty homogenous and sell a broad range of drugs. With the goal of offering a complete picture, 3 types of pharmacy competition were included by this study, including total competition, disease-specific competition, and disease-specific drug competition where demand was measured by total visits to healthcare facilities, influenza-specific visits to healthcare facilities, and influenza-specific visits for purchasing drugs to healthcare facilities, respectively. Specifically, we used the diagnosis of diseases (in our study, influenza) to split up the whole market to derive the degree of disease-specific competition. This specification considers a scenario when only healthcare facilities offering treatment in the respective diagnosis were considered as competitors by pharmacy j. We then used the diagnosis of diseases (in our study, influenza) and type of healthcare services (in our study, drug purchases) to further split up the whole market to derive the degree of disease-specific drug competition. This specification considers a scenario when only healthcare facilities offering treatment in the respective diagnosis and type of services were regarded by pharmacy j as competitors.

H H I_{j, m} = \sum_{j = 1}^{N} {(\frac{D_{j, m}}{D_{m}})}^{2}

(1)

Although a 5 km radius has been adopted by prior research to define the catchment area for hospitals,³⁰ no consensus has been reached as to the most appropriate radius to define the market of pharmacies. We used the fixed radius approach that assigns each pharmacy a unique market area, which is the region enclosed by a circle centered on the pharmacy and is defined by a fixed radius³¹ of 1 km to 5 km, and 10 km, to test whether our research findings were sensitive to different market radii. We found the sampled pharmacies, on average, lose about 11% to 15% of their market share for each 1 km increase in the market radius defined. When the market radius increased from 1 to 10 km, pharmacies included in this study lose approximately 70% of their initial market share. This result indicates that our selection of market radii was able to provide a relatively complete picture of the impacts of varying market radii. The change in the average market share of all the sampled pharmacies with different market radii can be seen in Supplemental Appendix 4. Our study also included a set of variables that have been shown to affect drug expenditures, including patient age (years), sex (female or male), type of health insurance program used to pay medical costs (UEBMI or URRBMI), and whether the patient had comorbidities (whether the patient had other diseases besides influenza).²¹

Statistical Analysis

One major challenge that needs to be addressed is that we lacked data on some determinants of drug expenditures,²⁹ which may result in omitted variable bias. This can be more problematic if individuals’ choices of healthcare institutions could be partly explained by unobserved variables contained in the error term. To address this statistical challenge, we adopted a 3-way fixed-effects model to capture the effects of unobserved heterogeneity across individuals, pharmacies, and years.

In this study, in the absence of an exogenous policy shock, we adopted the HHI index to measure competition. Since price is one of the key factors determining individuals’ choices of pharmacies, we must recognize that a demand-based measure of competition suffers from an endogeneity problem. As such, another challenge that needs to be addressed is that there may exist reverse causality between the HHI and drug expenditures.^18,22 To address the endogeneity issue arising from reverse causality, the instrumental variable (IV) approach, which has been widely applied by previous related studies,^29,32
-35 was adopted. Since we included 3 types of competition, it was hard to find external variables to instrument the HHI; hence, we used the lag of endogenous variables as an IV.^33,34 In the presence of serial correlation, the lagged IV is no longer exogenous and may be correlated with unobserved confounders. Inspired by prior research that instrumented current-period HHI with the value of the HHI preceding the introduction of competition,^36,37 we restricted our sample to individuals who had not visited the same pharmacy in the previous year. Specifically, we used the $H H I_{i^{-}, j, t - 1}$ as an IV for the $H H I_{i, j, t - 1}$ , which may help reduce the potential association between the lagged IV and unobserved individual-level variables (such as income). Through applying this approach, we aim to present a viable alternative identification strategy for comparison.

Four sensitivity analyses were carried out to test the robustness of our findings. First, we instead estimated the association between pharmacy competition and individuals’ annual average influenza-specific drug expenditures (per visit) at all healthcare institutions. Second, we further estimated the models using a HHI that is based only on the visits to pharmacies alone. This means that non-pharmacy healthcare facilities were no longer considered as potential competitors by pharmacy j described in Equation 1. Such analysis enabled us to compare the effects of within-type pharmacy competition (that refers to competition between pharmacies) with those of between-type pharmacy competition (that refers to competition between pharmacies and non-pharmacy institutions). Third, we checked the stability of results after removing influential observations from the dataset based on an interquartile method. Fourth, we aggregated individual-level drug expenditures as an indicator of institutional-level annual drug revenue in order to explore the association between pharmacy competition on pharmacies’ annual average influenza-specific revenue per visit.

A set of diagnostic tests were carried out: (1) the Wald-test was conducted to test the joint nullity of regression coefficients; (2) the weak instrument test was performed to check the validity of the lagged IV; (3) the Wu-Hausman endogeneity test was conducted to check for the presence of endogeneity; and (4) the R² was reported to measure the goodness-of-fit of all the regression models. We adopted Geographical Information System-based techniques to compute the distance matrix. The Origins-Destinations (OD) cost matrix analysis³⁸ was employed to calculate the distance matrix in a road network using the ArcGIS Desktop 10.³⁹ Data on distances were then loaded into the R Statistical Environment for all the econometric analyses.⁴⁰

Ethics

Ethics approval for conducting the study was obtained while written informed consent from all participants was waived as this study does require personally identifiable information and we only received and use de-identified data.

Results

The Descriptive Results of the Sampled Individuals

Figure 1 presents the distribution of all the healthcare facilities (including pharmacies, hospitals, primary care clinics, etc.) where drugs were dispensed in 2019 in Changde city. Pharmacies were found to be spatially clustered with other healthcare facilities (including primary care facilities, hospitals, clinics, etc.). Supplemental Appendix 5 reports annual growth in the number of healthcare facilities over the study period. The number of pharmacies increased from 773 in 2015 to 1156 in 2019. Supplemental Appendix 6 reports healthcare facilities’ average market share measured based on total visits, disease-specific visits, and influenza-specific visits for purchasing drugs. Pharmacies were found to be the major providers of drugs for individuals with influenza, taking up more than 90% of the market from 2015 to 2019. By the end of 2019, pharmacies accounted for about 85% and 55% of the market measured based on disease-specific visits and total visits respectively.

Figure 1.

The distribution of healthcare facilities in 2019. Yellow cross: Other facilities. Purple bolt: Clinics. Blue circle: Hospitals. Green pin: Primary care facilities. Red triangle: Pharmacies.

Table 1 reports the descriptive characteristics of the sampled individuals. A total of 6 694 534 individuals were included, among which the majority were male (56.21%), used UEBMI to pay their drug expenditures (99.93%), and did not have comorbidities (60.71%). The age of the sampled individuals was 52 years with a standard deviation (SD) of 15. Patients’ annual average influenza-specific drug expenditures were 94.39 CNY (SD = 119.02), where 6.91 CNY = $1 US in 2019. Among the 3 types of pharmacy competition, disease-specific drug competition reached the highest mean value, increasing from 0.19 (with a market radius of 1 km) to 0.05 (with a market radius of 10 km). There was a reduction in the mean value of the 3 types of pharmacy competition during our study period (Supplemental Appendix 7), implying the emergence of a more competitive market.

Table 1.

The Descriptive Statistics of the Sampled Patients.

Study variables	Minimum	Maximum	Median	Mean	Variance	SD
Dependent variable
Annual average influenza-specific drug expenditures per visit	0.003	66 031.71	71.763	94.39	14 166.130	119.022
Key independent variables
Disease-specific drug competition (1 km)	0.038	1.000	0.136	0.193	0.037	0.192
Disease-specific drug competition (2 km)	0.029	1.000	0.061	0.113	0.023	0.151
Disease-specific drug competition (3 km)	0.019	1.000	0.044	0.096	0.023	0.151
Disease-specific drug competition (4 km)	0.014	1.000	0.035	0.083	0.020	0.142
Disease-specific drug competition (5 km)	0.011	1.000	0.030	0.075	0.018	0.133
Disease-specific drug competition (10 km)	0.009	1.000	0.029	0.054	0.007	0.084
Disease-specific competition (1 km)	0.038	1.000	0.125	0.178	0.030	0.173
Disease-specific competition (2 km)	0.027	1.000	0.059	0.103	0.018	0.134
Disease-specific competition (3 km)	0.017	1.000	0.043	0.086	0.018	0.133
Disease-specific competition (4 km)	0.013	1.000	0.036	0.074	0.016	0.127
Disease-specific competition (5 km)	0.010	1.000	0.029	0.067	0.013	0.116
Disease-specific competition (10 km)	0.008	1.000	0.027	0.048	0.005	0.070
Total competition (1 km)	0.038	1.000	0.124	0.178	0.030	0.174
Total competition (2 km)	0.027	1.000	0.066	0.106	0.018	0.132
Total competition (3 km)	0.020	1.000	0.045	0.088	0.017	0.131
Total competition (4 km)	0.013	1.000	0.037	0.076	0.016	0.125
Total competition (5 km)	0.013	1.000	0.035	0.068	0.013	0.112
Total competition (10 km)	0.011	1.000	0.032	0.048	0.004	0.066
Control variables
Age	1	112	51	51.535	225.438	15.015
Sex	Female = 2 931 751 (43.793%)
Sex	Male = 3 762 783 (56.207%)
Insurance	Working residents with URRBMI = 4012 (0.060%)
	Non-working residents with URRBMI = 960 (0.014%)
	Working employees with UEBMI = 3 836 540 (57.309%)
	Non-working employees with UEBMI = 2 853 022 (42.617%)
Comorbidity	Without comorbidities = 4 064 028 (60.707%)
	With comorbidities = 2 630 506 (39.293%)

The Association Between Pharmacy Competition and Drug Expenditures (OLS Model)

The results of the OLS estimation of the association between pharmacy competition and drug expenditures are reported in Table 2. The estimated coefficient of total competition with a market radius of 1, 5, and 10 km was 0.06 (P > .1), 0.11 (P < .05), 0.17 (P < .05), respectively. This finding implies that for every 10% increase in the degree of total competition with a market radius of 1, 5, and 10 km, an individual’s annual average influenza-specific drug expenditures decreased by 0.55%, 1.06%, and 1.74%, respectively. In comparison with total competition, disease-specific drug competition had a larger negative association with individuals’ annual average influenza-specific drug expenditures when the market radius was set to 1 and 2 km. However, total competition had a larger negative association with individuals’ annual average influenza-specific drug expenditures than disease-specific drug competition when the market radius exceeded 3 km. The P-value of the Wald test for all the models was less than .05, rejecting the nullity of regression coefficients in these models. The R² for all the models was larger than .62, implying that these models could explain more than 62% of the variation in the dependent variable.

Table 2.

The Association Between Pharmacy Competition and Drug Expenditures (OLS Model).

	1 km	2 km	3 km	4 km	5 km	10 km
	The association between disease-specific drug competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Disease-specific drug competition	0.085** [0.005; 0.164]	0.083* [−0.007; 0.174]	0.077* [−0.013; 0.167]	0.071* [−0.016; 0.158]	0.085* [−0.003; 0.173]	0.140* [−0.006; 0.286]
Insurance_working residents	0.850 [−1916.6; 1918.3]	0.859 [−1911.7; 1913.4]	0.862 [−1911.3; 1913.0]	0.829 [−1911.1; 1912.8]	0.858 [−1910.3; 1912.0]	0.946 [−1903.2; 1905.1]
Insurance_non-working residents	−0.708 [−2465.1; 2463.7]	−0.654 [−2459.7; 2458.4]	−0.672 [−2457.8; 2456.5]	−0.682 [−2456.9; 2455.6]	−0.681 [−2456.3; 2454.9]	−0.632 [−2448.6; 2447.4]
Insurance_non-working employees	0.223 [−1985.8; 1986.3]	0.186 [−1983.7; 1984.0]	0.190 [−1983.8; 1984.2]	0.155 [−1985.4; 1985.7]	0.161 [−1984.4; 1984.7]	0.239 [−1980.1; 1980.6]
Comorbidity_yes	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]
	The association between disease-specific competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Disease-specific competition	0.064 [−0.029; 0.158]	0.074* [−0.021; 0.169]	0.077* [−0.013; 0.168]	0.069* [−0.020; 0.158]	0.081* [−0.001; 0.163]	0.118 [−0.070; 0.306]
Insurance_working residents	0.823 [−1907.8; 1909.5]	0.832 [−1902.5; 1904.2]	0.852 [−1901.6; 1903.3]	0.835 [−1902.8; 1904.5]	0.868 [−1902.0; 1903.7]	0.945 [−1895.8; 1897.7]
Insurance_non-working residents	−0.700 [−2453.0; 2451.6]	−0.672 [−2447.3; 2446.0]	−0.672 [−2445.9; 2444.6]	−0.676 [−2446.7; 2445.3]	−0.672 [−2446.8; 2445.5]	−0.645 [−2438.9; 2437.7]
Insurance_non-working employees	0.238 [−1982.5; 1983.0]	0.218 [−1979.6; 1980.0]	0.235 [−1979.1; 1979.6]	0.188 [−1981.5; 1981.9]	0.208 [−1981.2; 1981.6]	0.286 [−1977.7; 1978.3]
Comorbidity_yes	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]
	The association between total competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Total competition	0.055 [−0.019; 0.128]	0.076* [−0.013; 0.165]	0.089** [0.002; 0.175]	0.088** [0.005; 0.171]	0.106** [0.032; 0.180]	0.174** [0.027; 0.322]
Insurance_working residents	0.823 [−1912.7; 1914.4]	0.864 [−1907.8; 1909.5]	0.856 [−1903.1; 1904.8]	0.874 [−1899.9; 1901.6]	0.875 [−1896.9; 1898.6]	0.902 [−1879.1; 1880.9]
Insurance_non-working residents	−0.643 [−2459.1; 2457.8]	−0.634 [−2454.0; 2452.7]	−0.650 [−2449.4; 2448.1]	−0.636 [−2445.7; 2444.4]	−0.654 [−2444.4; 2443.1]	−0.660 [−2425.0; 2423.7]
Insurance_non-working employees	0.209 [−1981.9; 1982.3]	0.187 [−1977.3; 1977.7]	0.180 [−1973.7; 1974.1]	0.175 [−1972.8; 1973.2]	0.178 [−1969.4; 1969.8]	0.191 [−1955.2; 1955.6]
Comorbidity_yes	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.042**** [−0.054; −0.031]	−0.043**** [−0.054; −0.031]

Note. The significance codes: <.001****, <.01***, <.05**, <.10*; The R² for all these models was larger than .62; The P-value of the Wald test for all the models rejected the joint nullity hypothesis.

The Association Between Pharmacy Competition and Drug Expenditures (IV Model)

In the first-stage regression, the 3 types of the HHI were found to be positively associated with their lagged values. The value of the F-statistics for all the regression models in the first stage analysis was larger than 10, rejecting the presence of weak instruments. The P-value of the Wald test for all these models was less than .05, rejecting the presence of joint nullity of regression coefficients. These models had a high goodness-of-fit, with their R² ranging from 97.0% and 99.0%. The P-value of the (Wu-) Hausman test rejects the null hypothesis that the HHI was uncorrelated with the error term, which indicates the presence of endogeneity. These results demonstrated the appropriateness of using the lag of the HHI as an IV in our study.

The second-stage regression results of the association between pharmacy competition and drug expenditures are reported in Table 3. The negative association between the 3 types of pharmacy competition and drug expenditures were found to be more substantial using the lagged IV approach. The estimated coefficient of total competition with a market radius of 1, 5, and 10 km was 0.07 (P > .1), 0.22 (P < .1), and 0.52 (P < .05), respectively. This implies that each 10% increase in the degree of total competition with a market radius of 1, 5, and 10 km was associated with a decline of 0.65%, 2.21%, and 5.20% in annual average influenza-specific drug expenditures, respectively. We still found that total competition had a larger negative association with individuals’ annual average influenza-specific drug expenditures when the market radius was larger than 3 km. The R² for all the models using the IV approach was larger than .70, implying that all the independent variables explained more than 70% of the variation in the dependent variable. The P-value of the Wald-test for all the IV models was less than .05, implying the joint significance of regression coefficients.

Table 3.

The Association Between Pharmacy Competition and Drug Expenditures (IV Model).

	1 km	2 km	3 km	4 km	5 km	10 km
	The association between disease-specific drug competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Disease-specific drug competition	0.138* [−0.043; 0.319]	0.135 [−0.091; 0.362]	0.160 [−0.066; 0.386]	0.175* [−0.052; 0.402]	0.209* [−0.015; 0.432]	0.470** [0.196; 0.745]
Insurance_working residents	0.675 [−2173.8; 2175.1]	0.603 [−2175.6; 2176.8]	0.514 [−2174.4; 2175.4]	0.477 [−2173.7; 2174.7]	0.499 [−2173.7; 2174.7]	0.819 [−2171.1; 2172.8]
Insurance_non-working residents	−0.151 [−2979.7; 2979.4]	−0.020 [−2978.3; 2978.2]	0.163 [−2974.8; 2975.2]	0.196 [−2972.2; 2972.6]	0.115 [−2973.1; 2973.3]	−0.091 [−2969.5; 2969.3]
Insurance_non-working employees	−0.109 [−2036.5; 2036.3]	−0.130 [−2036.8; 2036.6]	−0.213 [−2037.2; 2036.7]	−0.272 [−2037.6; 2037.1]	−0.275 [−2037.8; 2037.2]	0.036 [−2039.3; 2039.4]
Comorbidity_yes	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]
	The association between disease-specific competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Disease-specific competition	0.107 [−0.145; 0.358]	0.117 [−0.123; 0.357]	0.158 [−0.058; 0.374]	0.175* [−0.037; 0.386]	0.215** [0.002; 0.429]	0.581** [0.212; 0.951]
Insurance_working residents	0.746 [−2174.0; 2175.5]	0.645 [−2175.1; 2176.4]	0.606 [−2174.4; 2175.6]	0.589 [−2174.2; 2175.4]	0.640 [−2174.2; 2175.5]	0.925 [−2171.4; 2173.3]
Insurance_non-working residents	−0.249 [−2978.4; 2977.9]	−0.206 [−2978.8; 2978.4]	−0.179 [−2977.9; 2977.5]	−0.184 [−2977.4; 2977.0]	−0.153 [−2978.7; 2978.4]	0.153 [−2976.5; 2976.8]
Insurance_non-working employees	−0.021 [−2035.9; 2035.9]	−0.059 [−2036.1; 2036.0]	−0.015 [−2036.6; 2036.6]	−0.132 [−2037.1; 2036.8]	−0.080 [−2037.0; 2036.8]	0.046 [−2039.3; 2039.4]
Comorbidity_yes	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.041*** [−0.054; −0.029]	−0.041*** [−0.054; −0.029]	−0.041*** [−0.054; −0.029]
	The association between total competition and drug expenditures
Variables	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]	Estimates [95% CI]
Total competition	0.065 [−0.109; 0.240]	0.108 [−0.099; 0.316]	0.155* [−0.050; 0.359]	0.182* [−0.024; 0.387]	0.221** [0.015; 0.427]	0.520** [0.217; 0.824]
Insurance_working residents	0.753 [−2175.9; 2177.4]	0.668 [−2175.6; 2176.9]	0.548 [−2174.6; 2175.6]	0.523 [−2174.4; 2175.5]	0.659 [−2174.3; 2175.6]	0.920 [−2173.8; 2175.6]
Insurance_non-working residents	−0.284 [−2979.3; 2978.8]	−0.167 [−2978.5; 2978.2]	−0.082 [−2976.6; 2976.4]	−0.060 [−2976.5; 2976.3]	−0.093 [−2977.5; 2977.3]	−0.176 [−2984.3; 2984.0]
Insurance_non-working employees	−0.062 [−2035.6; 2035.4]	−0.145 [−2034.7; 2034.4]	−0.263 [−2034.6; 2034.1]	−0.388 [−2034.5; 2033.7]	−0.259 [−2033.5; 2033.0]	−0.019 [−2031.8; 2031.8]
Comorbidity_yes	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]	−0.042*** [−0.054; −0.029]

Note. The significance codes: <.001****, <.01***, <.05**, <.10*.

Results of Sensitivity Analyses

The association between pharmacy competition and drug expenditures at all healthcare facilities are reported in Supplemental Appendix 8. The negative association between the 3 types of pharmacy competition and individuals’ annual average influenza-specific drug expenditures at all healthcare facilities were still found. Supplemental Appendix 9 reports the association between the HHI measured based only on the demand of pharmacies on drug expenditures. For each 10% increase in the degree of pharmacy competition measured based only on the demand of pharmacies, an individual’s annual average influenza-specific drug expenditures fell by 0.89%, 0.90%, and 1.37%, when the radius of market was set to 1, 5, and 10 km, respectively.

As depicted by Supplemental Appendix 10, the distribution of the dependent variable was more concentrated after removing outliers. Supplemental Appendix 11 reports the results of the IV model using a dataset without influential observations. Various forms of competition were still found to reduce individuals’ annual average influenza-specific drug expenditures. Supplemental Appendix 12 reports the association between pharmacy competition and pharmacies’ annual average influenza-specific drug revenue per visit. In line with findings from our primary analysis, competition was found to reduce pharmacies’ annual average influenza-specific revenue per visit. The consistency between the results of our primacy analysis and the sensitivity analyses strengthened the credibility of our conclusions regarding the negative association between pharmacy competition and drug expenditures.

Discussion

This study constructed a 3-way FE model and employed a lagged IV approach to estimate the association between pharmacy competition and individuals’ drug expenditures spent at pharmacies. We demonstrated that after adjusting for various covariates, pharmacy competition (evidenced by a lower HHI score) was associated with a reduction in drug expenditures by individuals with influenza in China.

Principal Findings

Our study demonstrated that competition contributed to a decline in drug expenditures, which is consistent with some previous work in China.^7,17,18 More precisely, we demonstrated that within a market radius of 3 km, disease-specific drug competition had a larger negative association with patients’ drug expenditures than total competition. But the negative association between disease-specific drug competition and patients’ drug expenditures were overtaken by that between total competition and patients’ drug expenditures when the market radius was beyond 3 km. This may be attributed to that with increased market size, consumers tended to consider institutions as substitutes for pharmacies when they want to purchase drugs, in which case pharmacies would compete for any patient rather than for patients seeking a specific type of service or to address a specific disease. Unexpectedly, we found that patients with reported comorbidities had lower drug expenditures than those without. This finding may be rationalized as our study focused on influenza-specific drug expenditures and variances in the estimated health effects of pharmacy competition for the 2 groups were extremely small across model specifications.

Our study provided some preliminary evidence about the varying associations between between- and within-type pharmacy competition and individuals’ drug expenditures. In our primary analysis, all the healthcare facilities competing in the market were regarded by pharmacies as competitors. One of our sensitivity analyses, however, considered a specific scenario when only pharmacies competing in the market were regarded as competitors by our sampled pharmacies. Competition between pharmacies had the same positive but a larger negative association with drug expenditures than competition between pharmacies and all the healthcare facilities. In China, one of the major providers of drugs and pharmaceutical services are community healthcare centers/stations (of which the majority had public ownership⁴¹) and hospitals. The public community healthcare centers/stations and hospitals are required to remove the profit margins on sales of all the national essential drugs and all drugs respectively. In comparison, pharmacies can profit from drug sales, though a price ceiling for drugs is introduced when they had PHI coverage and/or joined the bulk drug procurement program. These facts suggest that it would be difficult for pharmacies to initiate price competition (ie, attracting customers through lowering prices for the products they sell) with community healthcare centers/stations and hospitals. Instead, pharmacies were more likely to consider other pharmacies as competitors and tended to differentiate themselves from other pharmacies by promoting price competition, which may explain why competition between pharmacies led to a greater reduction in drug expenditures by individuals.

Our research findings offered some important policy implications for future drug pricing reform in China. This study revealed the essential role retail pharmacies play in the pharmacy industry, demonstrating that pharmacy competition contributed to a reduction in drug expenditures. The evidence offered herein suggests that policy makers may wish to further encourage competition among pharmacies to control the rising drug expenditures. We also suggest that promoting competition between retail pharmacies and other types of healthcare facilities more broadly may serve as a useful strategy to control drug expenditures. In recent years, some initial attempts to bolster the development of other healthcare facilities have been made by the Chinese government. For instance, the “Thirteenth Five-Year Health Plan” issued in 2016⁹ and the “Notice on Promoting the Development of Clinics” released in 2019⁴² support the development of clinics. These policies allow clinics to determine the prices of drugs and to be reimbursed by PHI programs. Under the guidance of these policies, specific action plans to promote inter-type pharmacy competition are suggested to be proposed.

Strengths and Limitations

This study has several strengths. To the best of our knowledge, the literature has not examined how pharmacy-level competition might affect drug expenditures by individuals in China. Additionally, most examinations were unique to a single type of competition, while our study contributed to the extant literature through comparing the associations between various types of pharmacy competition and drug expenditures. What is more, our study demonstrated the benefit of enhanced competition in the pharmacy industry was a reduction in drug expenditures by individuals. Our findings may shed light on the health benefits of pro-competition reforms that have been carried out in China since 2009. Our findings were also of significance for retail pharmacies that want to configure an optimal network of store locations and policy makers who must reform the pricing mechanism of drugs.

Certain limitations of this study should be recognized. First, data should be remedied before solid conclusions can be drawn regarding the association between pharmacy competition and drug expenditures by individuals without PHI coverage in Changde city. Specifically, as our study samples were restricted to those with the PHI coverage who purchased drugs for influenza at pharmacies that opted-in to the PHI arrangements from 2015 to 2019, we were unable to investigate how pharmacy competition might be associated with drug expenditures for all the patients regardless of whether they were covered by the PHI programs. However, since there was a high participation rate of individuals in PHI programs (97% in Changde city), our findings were generally applicable to most residents living in Changde city concerning the association between pharmacy competition and their drug expenditures. Another significant limitation was the inability to control for the effects of unobserved serially correlated characteristics, which may threaten the credibility of our IV estimation. In our study, the ordinary least squares estimates were smaller than the IV estimates. This result may be attributed to biases from measurement error (which leads to attenuation) and omitted variables (such as patients’ health status and income level) that correlate with both patients’ choices of pharmacies and their drug expenditures. However, these biases would likely have been more pronounced if we had not constructed the 3-way fixed effects models and employed the lagged IV identification strategy to address concerns related to unobserved individual-level variables. Third, since we lacked data on patients’ place of residence, we did not estimate the patient choice model and only performed a reduced-form analysis. Patients’ residential distance to healthcare facilities is a critical variable since it predicts patients’ choices of healthcare facilities and can satisfy the exclusion restriction of using the IV approach.⁴³ We highly recommend future research to apply a structural approach to estimate patients’ demand and then simulate the health effects of pharmacy competition. It would be challenging but meaningful to include consumers, insurance companies, and healthcare providers⁴⁴ into a holistic model. This would allow for an investigation of how the interaction between and among these players may affect patients’ health outcomes. Fourth, we did not undertake a detailed analyses of the underlying mechanisms through which the positive health effects may have resulted from the reduction in individual drug expenditures that resulted from increased pharmacy competition. We suggest future studies with data on patient-level characteristics to take steps toward exploring the association between pharmacy competition on patient welfare. Fifth, since we lacked information on pharmaceutical products consumed, we only estimated the association between institutional-level competition and drug expenditures. A prior Chinese study has offered some valuable insights on the differential impacts of competition in the generic and therapeutic markets on price for both branded and generic drugs.¹⁷ Researchers are suggested to seek answers to questions about the specific health consequences of various forms of product-level competition in the pharmacy sector.

Conclusion

This study demonstrated that pharmacy competition led to a decline in drug expenditures by individuals with influenza in China. This study stressed that the association between pharmacy competition and drug expenditures varied by the type of pharmacy competition considered. Particularly, this research found that pharmacies competed for any patient rather than for patients on the basis of a specific line of services or diseases when the market radius was over 3 km. The evidence offered herein suggests that competition may be a useful tool to alleviate the financial burdens on individuals by drug expenditures.

Supplemental Material

sj-docx-1-inq-10.1177_00469580241307446 – Supplemental material for Competition Among Pharmacies as a Determinant of Drug Expenditures

Supplemental material, sj-docx-1-inq-10.1177_00469580241307446 for Competition Among Pharmacies as a Determinant of Drug Expenditures by Zixuan Peng, Audrey Laporte, Xiaolin Wei and Peter C. Coyte in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Footnotes

Acknowledgements

Not applicable.

Authors’ Contributions

P.Z. conceived the original research idea, designed the model, analyze and interpret the data, and drafted the manuscript. A.P. and X.W. made critical revisions of the manuscript for important intellectual content. P.C. helped supervise the project and revise the manuscript. The authors read and approved the final manuscript.

Data Availability

The datasets generated and/or analyzed during the current study are not publicly available due to privacy concerns but are available from the corresponding author on reasonable request.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Start-up Research Fund of Southeast University (#4025002402).

Ethics Considerations

This research has been approved by the Research Ethics Board of University of Toronto (Protocol #:38551,approved On:4-Apr-23). All methods were carried out in this study were accordance with relevant guidelines and regulations issued by the Research Ethics Board of University of Toronto. We did not seek ethical approval from a local Institutional Review Board/Ethics Committee in the country where the data were sourced, in accordance with the data confidentiality agreements signed with the institutions that provided the dataset.

Consent

All data is de-identified and consent from the individual patient level is not applicable.

ORCID iD

Zixuan Peng

Supplemental Material

Supplemental material for this article is available online.

References

World Health Organization. Access to medicines: Making market forces serve the poor. 2017. Accessed November 10, 2024. https://cdn.who.int/media/docs/default-source/essential-medicines/fair-price/chapter-medicines.pdf?sfvrsn=adcffc8f_4&download=true

PRC

SC.

The white paper on status quo of drug supervision. 2008. Accessed November 10, 2024. http://www.gov.cn/jrzg/2008-07/18/content_1048857.htm

Institute CNHE. National Health Accounts Report 2012. Ministry of Health, People’s Republic of China; 2012.

PRC NHC. China health statistical yearbook 2020. 2020. Accessed November 10, 2024. http://www.nhc.gov.cn/mohwsbwstjxxzx/tjtjnj/202112/dcd39654d66c4e6abf4d7b1389becd01/files/7325fee02bf84676a15ebcfa827c9cd5.pdf

Shi

Pharmaceutical supply chain in China: current issues and implications for health system reform. Health Policy. 2010;97(1):8-15. doi:10.1016/j.healthpol.2010.02.010

Dong

Bogg

Rehnberg

Diwan

Drug policy in China: Pharmaceutical distribution in rural areas. Soc Sci Med. 1999;48(6):777-786. doi:10.1016/S0277-9536(98)00414-6

Liu

Pharmaceutical pricing: an empirical study of market competition in Chinese hospitals. Pharmacoeconomics. 2014;32(3):293-303. doi:10.1007/s40273-013-0099-5

PRC

SC.

Notice of the state council on printing the “outline of the healthy china 2030 plan.”

2016. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2016/10/25/art_104_6428.html

PRC

SC.

Notice of the state council on printing and distributing the thirteenth five-year plan for deepening the reform of the medical and health system. 2016. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2016/12/27/art_104_6429.html

10.

PRC NMSA. Notice of the national medical security administration on printing and distributing the “opinions on drug price management”. 2019. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2019/12/6/art_104_6468.html

11.

PRC

SC.

Opinions of the state council of the communist party of china on deepening the reform of the medical security system. 2020. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2020/3/5/art_104_6480.html

12.

PRC

NMIA

PRC

NHC.

Guiding opinions of the national medical insurance administration and the national health commission on establishing and improving the dual channel supply chain management for drugs included in national health insurance negotiation program. 2021. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2021/5/10/art_104_6527.html

13.

Aue

Spatial effects of price regulations and competition: A dynamic approach to the German retail pharmacy market. 2020. Accessed November 10, 2024. https://www.crctr224.de/en/research-output/discussion-papers/archive/2020/DP195

14.

Mia

Competition in the Swedish Pharmacy. University of Gothenburg; 2015. Accessed November 10, 2024. https://gupea.ub.gu.se/bitstream/handle/2077/39957/gupea_2077_39957_1.pdf;jsessionid=6484356125FAF5746A9CCF702F7F489A?sequence=1

15.

Stargardt

Schreyögg

Busse

Pricing behaviour of pharmacies after market deregulation for OTC drugs: the case of Germany. Health Policy. 2007;84(1):30-38. doi:10.1016/j.healthpol.2007.04.006

16.

Vogler

Habimana

Arts

Does deregulation in community pharmacy impact accessibility of medicines, quality of pharmacy services and costs? Evidence from nine European countries. Health Policy. 2014;117(3):311-327. doi:10.1016/j.healthpol.2014.06.001

17.

Liu

Yang

Hsieh

CR.

Regulation and competition in the Taiwanese pharmaceutical market under national health insurance. J Health Econ. 2012;31(3):471-483. doi:10.1016/j.jhealeco.2012.03.003

18.

Zhao

Impacts of regulated competition on pricing in Chinese pharmaceutical market under urban employee basic medical insurance. Expert Rev Pharmacoecon Outcomes Res. 2017;17(3):311-320. doi:10.1080/14737167.2017.1251318

19.

Castanheira

Ornaghi

Siotis

Frutos

M-A

. The unexpected consequences of asymmetric competition. An application to the pharmaceutical industry. 2017. Accessed November 10, 2024. http://i3health.eu/wp-content/uploads/2018/01/pharma.2017-10-19.pdf

20.

Wang

YR.

Price competition in the Chinese pharmaceutical market. Int J Health Care Finance Econ. 2006;6(2):119-129. doi:10.1007/s10754-006-5558-5

21.

Chen

The effects of competition on prescription payments in retail pharmacy markets. South Econ J. 2019;85(3):865-898. doi:10.1002/soej.12325

22.

Fura

Rantzien

The Swedish Pharmaceutical Market: Competition and Pricing Post-Deregulation. 2016. Accessed November 10, 2024. https://www.konkurrensverket.se/globalassets/dokument/kunskap-och-forskning/uppsatstavling/uppsatser/uppsats-2017_karin-fura_elsa-horn-rantzien.pdf

23.

Beach

Harbin role of pharmacies in Chinese world of health care. Lancet. 1999;354(9177):493-493. doi:10.1016/S0140-6736(05)75528-8

24.

PRC NHSA. China Health Security Statistical Yearbook 2022. China Statistics Press; 2022.

25.

PRC NHSA. China Health Security Statistical Yearbook 2023. China Statistics Press; 2023.

26.

PRC HPBoS. Hunan statistical yearbook 2020. 2020. Accessed November 10, 2024. https://www.hunan.gov.cn/zfsj/tjnj/tygl.html

27.

Conesa

López-Quílez

Martínez-Beneito

Miralles

Verdejo

FluDetWeb: an interactive web-based system for the early detection of the onset of influenza epidemics. BMC Med Inform Decis Mak. 2009;9:36. doi:10.1186/1472-6947-9-36

28.

PRC

NESSDC

PRC

NSTI.

Basic geographic data of Hunan province at a scale of 1:250,000. 2015. Accessed November 10, 2024. http://www.geodata.cn/data/datadetails.html?dataguid=85166584456169&docid=2653

29.

Pan

The association of hospital competition with inpatient costs of stroke: Evidence from China. Soc Sci Med. 2019;230:234-245. doi:10.1016/j.socscimed.2019.04.017

30.

T-H

Tung

Y-C

Wei

C-J.

Can hospital competition really affect hospital behavior or not? An empirical study of different competition measures comparison in Taiwan. Inquiry. 2017;54:46958017690289. doi:10.1177/0046958017690289

31.

Wong

Zhan

Mutter

Do different measures of hospital competition matter in empirical investigations of hospital behavior. Rev Ind Organ. 2005;26(1):27-60. doi:10.1007/s11151-004-6067-7

32.

Colla

Bynum

Austin

Skinner

Hospital competition, quality, and expenditures in the U.S. Medicare population. 2016. Accessed November 10, 2024. http://www.nber.org/papers/w22826

33.

Pan

Does hospital competition lead to medical equipment expansion? Evidence on the medical arms race. Health Care Manag Sci. 2021;24(3):582-596. doi:10.1007/s10729-020-09529-x

34.

Ding

Peng

Heim

Jordan

VS.

Service mix, market competition, and cost efficiency: a longitudinal study of U.S. ho spitals. J Oper Manag. 2020;66(1-2):176-198. doi:10.1002/joom.1050

35.

Pan

Qin

Messina

Delamater

PL.

Does hospital competition improve health care delivery in China?

China Econ Rev. 2015;33:179-199. doi:10.1016/j.chieco.2015.02.002

36.

Moscelli

Gravelle

Siciliani

Santos

Heterogeneous effects of patient choice and hospital competition on mortality. Soc Sci Med. 2018;216:50-58. doi:10.1016/j.socscimed.2018.09.009

37.

Moscelli

Gravelle

Siciliani

Hospital competition and quality for non-emergency patients in the English NHS. RAND J Econ. 2021;52(2):382-414. doi:10.1111/1756-2171.12373

38.

ESRI. OD cost matrix analysis. 2024. Accessed November 10, 2024. https://desktop.arcgis.com/en/arcmap/latest/extensions/network-analyst/od-cost-matrix.htm

39.

ESRI. ArcGIS Desktop 10. ESRI. 2024. Accessed November 10, 2024. https://www.esri.com/en-us/home

40.

R Core Team. R: A Language and Environment for Statistical Computing, Vol. 3. R Foundation for Statistical Computing; 2013:201.

41.

Pan

Pong

Miao

Qian

Different models of hospital–community health centre collaboration in selected cities in China: a cross-sectional comparative study. Int J Integr Care. 2016;16(1):1-12. doi:10.5334/ijic.2456

42.

PRC NHC, PRC NDaRC, PRC MoF, PRC MoHRaSS, PRC NMIA. Notice on printing and distributing the opinions on promoting the development of clinics. 2019. Accessed November 10, 2024. http://www.nhsa.gov.cn/art/2019/6/6/art_104_6452.html

43.

Skellern

The hospital as a multi-product firm: The effect of hospital competition on value-added indicators of clinical quality. 2017. Accessed November 10, 2024. http://eprints.lse.ac.uk/id/eprint/83618

44.

Krabbe-Alkemade

Groot

Boter

The impact of hospital competition and insurer concentration on health care volume and cost in Dutch hospitals. Maandblad Voor Accountancy en Bedrijfseconomie. 2019;93(7/8):203-213. doi:10.5117/mab.93.34164

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.67 MB