Regional price differences of medical services: evidence from China

Abstract

Background
Price levels of medical services may vary across regions with different income levels, which would raise concerns about the equal access to medical services. This study aimed to estimate the spatial price index of medical services to measure price levels across 31 provincial regions in China.

Methods
Price data were collected from medical service price schedule in each region. Two methods based on the Purchasing Power Parities were used to estimate the spatial price index and measure price differences across regions. The two-way fixed effects models were used to examine the association between medical service price levels and income levels, and further investigate the impacts of price differences on utilization of medical services and medical expenditure.

Results
The consistent estimation results were given by two methods. Medical service price level in the highest-price region was found to be 74% higher than the lowest. There was a significant negative correlation between price levels and income levels, as well as price levels and the utilization of outpatient services. Moreover, we also found a 1% increase in medical service price level was significantly associated with a 0.34% and 0.24% increase in the medical service expense per outpatient visit and per inpatient respectively.

Conclusions
Regions in China had significant gaps in medical service price levels. Policymakers should pay more attention to regional price differences and take great measures such as enhancing financial protection to ensure the equal access to medical services and better achieve the universal health coverage.

Full text

RESEARCH Open Access
© The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0
International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you
give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modied the
licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation
or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://
creativecommons.org/licenses/by-nc-nd/4.0/.
Li and Liu BMC Public Health (2024) 24:2353
https://doi.org/10.1186/s12889-024-19719-9
and the income, and ensuring the balanced development
across regions [1, 2]. Several national institutes, such as
the U.S. Bureau of Economic Analysis and the Oce for
National Statistics of the UK, have ocially released the
spatial price index to measure the price dierences across
states and metropolitan areas [3, 4]. With the develop-
ment of the International Comparison Program (ICP),
the methodology of spatial price index (SPI) based on the
Purchasing Power Parities (PPPs) has been widely used
in measuring regional price dierences [2, 5]. A number
of studies have measured the general price levels across
regions using the spatial price index [6, 7], and found a
positive correlation between price levels and income
Background
e price of goods or services diers not only across
countries, but also across regions within a country. It’s
of great importance to take regional price dierences
into account especially for assessing regional inequal-
ity, comparing the real Gross Domestic Product (GDP)
BMC Public Health
*Correspondence:
Bao Liu
liub@fudan.edu.cn
1Department of Health Economics, School of Public Health, Fudan
University, 130 Dong’an Road, Shanghai 200032, China
2Key Laboratory of Health Technology Assessment (Fudan University),
National Health Commission, 130 Dong’an Road, Shanghai 200032, China
Abstract
Background Price levels of medical services may vary across regions with dierent income levels, which would raise
concerns about the equal access to medical services. This study aimed to estimate the spatial price index of medical
services to measure price levels across 31 provincial regions in China.
Methods Price data were collected from medical service price schedule in each region. Two methods based on the
Purchasing Power Parities were used to estimate the spatial price index and measure price dierences across regions.
The two-way xed eects models were used to examine the association between medical service price levels and
income levels, and further investigate the impacts of price dierences on utilization of medical services and medical
expenditure.
Results The consistent estimation results were given by two methods. Medical service price level in the highest-
price region was found to be 74% higher than the lowest. There was a signicant negative correlation between price
levels and income levels, as well as price levels and the utilization of outpatient services. Moreover, we also found a
1% increase in medical service price level was signicantly associated with a 0.34% and 0.24% increase in the medical
service expense per outpatient visit and per inpatient respectively.
Conclusions Regions in China had signicant gaps in medical service price levels. Policymakers should pay more
attention to regional price dierences and take great measures such as enhancing nancial protection to ensure the
equal access to medical services and better achieve the universal health coverage.
Keywords Medical services, Spatial price index, Regional price dierence, China
Regional price dierences of medical services:
evidence from China
LuoLi1,2 and BaoLiu1,2*
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 2 of 10
Li and Liu BMC Public Health (2024) 24:2353
levels [8–10]. is correlation is also known as the Penn
Eect, where high-income regions tend to have higher
price levels [11].
In the healthcare sector, the price of medical services
can ensure the costs of services are covered and provide
incentives for providers [12]. It’s also the critical factor
accounting for growth in health expenditure [13]. Mea-
suring regional price dierences of medical services may
help explain regional variations in health expenditure and
help policymakers take actions to improve health sys-
tem performance and equity [14]. eoretically, the price
level of medical services should also have a positive asso-
ciation with the income level. Providers in high-income
regions tend to have higher costs for delivering medi-
cal services including employee wages and oce rent,
so they need to charge higher prices to oset the varia-
tion of costs due to geographic location [15]. Meanwhile,
patients in low-income regions could be more likely to
experience nancial hardship when faced with higher
prices, especially for those not covered by health insur-
ance. is would raise concerns about the access to medi-
cal services and the health equity. Some countries have
taken into account the price dierences of medical ser-
vices across regions. For example, the Centers for Medi-
care & Medicaid Services in the U.S. has employed the
Geographic Practice Cost Index to adjust fee-for-service
payments in the Medicare Physician Fee Schedule [15],
and the National Health Service in England has calcu-
lated the market forces factor presented in the payment
index to vary the prices to reect dierences in unavoid-
able costs between providers [16]. However, studies on
price comparisons across regions in the eld of health-
care mainly focused on drug and consumable prices [17–
19], whereas medical service prices were often compared
in time series to reect the price change over time [20].
Few studies have measured regional price dierences
of medical services and it’s still unclear the association
between medical service price levels and income levels.
In China, medical services are usually provided by
public medical institutions. e specication of medi-
cal service price schedule was formulated at the national
level, while the administration of medical service price
was decentralized to the provincial level. e provin-
cial department of medical service price administration
in each region has established its own medical service
price schedule with a unique code for each service and
set the price ceiling for each service in public medical
institutions within the region. Some provinces have fur-
ther devolved the price administration authority to the
prefecture-level. is directly led to the dierences in
codes and prices of medical services across regions. From
2012, China has implemented the nationwide reform of
public hospitals and gradually cancelled the drug markup
with price adjustments for medical services, which made
medical service revenue become the major source of
hospital revenue [21]. With the increasing health expen-
diture and the budget constraints of medical insurance
funds, medical service price has gradually become the
focus of healthcare reforms. To further establish the
dynamic price adjustment mechanism, the National
Healthcare Security Administration (NHSA) of China
has launched a pilot program for deepening medical ser-
vice price reform in 2021 and proposed that regions with
similar socioeconomic status should establish reasonable
price linkage for medical services. erefore, it’s impor-
tant to measure price dierences of medical services
across regions in China, especially when considering the
regional disparities of medical resources [22]. However,
the dierences of price schedules due to the decentral-
ization of administration made it dicult to develop a
basket of medical services that were comparable across
regions, which also limited related research. No study has
measured the price dierences of medical services across
31 provincial regions in China.
With the rapid development of information technology,
the NHSA of China has established the national unied
healthcare security information system in 2022 and con-
structed a national coding system for medical services.
Each provincial region has also mapped its own medical
service codes with the national codes so that we could
match the comparable services by the national codes,
which makes price comparisons across regions possible.
Take these needs into account, the present study aimed
to estimate the spatial price index of medical services to
measure price dierences across 31 provincial regions
in China. On this basis, this study examined the correla-
tion between price levels of medical services and income
levels, and further investigated the impact of price dif-
ferences on utilization of medical services and medical
expenditure. Findings from this study may have implica-
tions for further price adjustment in China, and may also
provide lessons for other countries.
Methods
Data sources
e medical service price schedules by September 2023
were obtained from the Health Commission and the
Healthcare Security Administration of each provincial
region. Furthermore, price schedules were updated to
December 2023 based on relevant price adjustment poli-
cies. e highest prices in price schedules of each region
were selected to ensure comparability. e annual aver-
age price for 2023, estimated as the simple arithmetic
average of prices at these two time points, was used as
price data in the calculation of the spatial price index
[23].
After matching using the national codes, 1116 medical
services were identied and formed the basket for price
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 3 of 10
Li and Liu BMC Public Health (2024) 24:2353
comparison. Some provinces didn’t set the specic price
for certain services in price schedules and therefore only
34,158 price data of 1116 services were obtained. e
national specication divided medical services into four
categories, followed as general medical services (includ-
ing registration services, nursing services, etc.), medi-
cal diagnosis services (including examinations, testing
services), clinical treatment services (including therapy
services, surgery services), and traditional treatment
services (including traditional therapy services). ese
1116 services could be divided into above four categories
based on the national specication. However, consider-
ing the availability of weights data, services in Category
“Clinical Treatment” and Category “Traditional Treat-
ment” were merged into one category (Category “Clini-
cal and Traditional Treatment”). erefore, 1116 services
were divided into three categories followed as general
medical services, medical diagnosis services, and clini-
cal and traditional treatment services. e three catego-
ries served as the basic heading in the calculation of the
spatial price index. Table1 showed the structure of the
basket compared with the national specication. e
Cramer’s V coecient was 0.09, which indicated the sim-
ilarity and the representativeness in structure.
To ensure the comparability as much as possible, we
used the Resistant Fences method to detect the outlier
price data [24]. For the log transform price of each service
in the basket, let
q25 =
the rst quartile,
q75 =
the third
quartile, and
H=q75 −q25
, the interquartile range. e
incomparable price was dened as less than
q25 −k∗H
or greater than
q75 +k∗H
, where
k
is a constant and
equal to 2 in this study. After that, 772 (2.26%) price data
were excluded, and 33,386 price data were used nally to
estimate the spatial price index.
e expenditure weight of each basic heading in each
region were also needed to estimate the spatial price
index, which could show the relative importance of each
category. e weights were computed using the revenue
data of public hospitals from the China Health Yearbook,
which provided data on medical services revenue for dif-
ferent categories including registration, nursing, exami-
nation, therapy, and surgery services in 2021. Specically,
we directly obtained the amounts of examinations and
surgery services in each region. For other categories, we
used the national average proportions in total outpatient
and inpatient revenue for each category, and the outpa-
tient and inpatient revenue in each region to estimate the
amounts of each category for each region. After that, we
were able to calculate expenditure weights of basic head-
ings in each region.
Methodology of the spatial price index
Due to 31 regions were involved in the price compari-
son, the spatial price index was constructed as a multi-
lateral price index rather than a bilateral price index for
comparison only between two regions. Compared with
the bilateral price index, the multilateral price index
could satisfy several fundamental properties such as
transitivity and base country invariance [2, 25]. Transi-
tivity means that the direct price index between any two
regions yields the same result as an indirect comparison
via any other region, and base country invariance means
that the price index between any two regions should be
the same regardless of the choice of base region [26]. e
PPPs, developed by the ICP, has been a widely used mul-
tilateral index for comparing price levels across regions
[27], which showed the ratio of prices for the same bas-
ket of goods and services in dierent regions. is study
used PPPs to estimate the spatial price index of medical
services to measure price dierences across 31 provincial
regions in China. e detailed steps are as follows.
The basic heading index
e rst step was to estimate the basic heading index
between every two regions. e Country Product
Dummy (CPD) method, which was rst introduced by
Summers (1973) [28], was used to estimate the basic
heading index. is method could deal with the fact
that not all prices of medical services were available
in all regions and was the standard method used in the
ICP. e CPD method can be described as the following
regression model:
ln
pij =

M
j=1γiDj+

N
i=1πiD∗
i+µij (1)
where
ln p
ij is the price of medical service
i
in region
j
,
Dj
(j=
1,2
,... ,M) and
D∗
i(i=1,2,... ,N)
repre-
sent, respectively, the dummy variables for
M
regions in
the comparisons and the dummy variables for
N
medi-
cal services in basic heading. e basic heading index
between region
j
and
k
can be derived as:
Table 1 The components of the basket compared with the
national specication
Category n (%) Cramer’s
V coef-
cient
(95% CI*)
The basket The national
specication
Total 1116
(100%)
4170
(100%)
0.09
(0.07,
0.12)
General medical
services
23
(2.06%)
90
(2.16%)
Medical diagnosis
services
185
(16.58%)
1097
(26.31%)
Clinical and traditional
treatment services
908
(81.36%)
2983
(71.53%)
* 95% Condence interval
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 4 of 10
Li and Liu BMC Public Health (2024) 24:2353
PPP
jk =
exp
γj

exp
(

γ
k
) (2)
Aggregation above basic headings
e second step was to aggregate basic heading index
into a bilateral price index between every two regions
using the expenditure weights data in each region. e
Fisher index, which was the geometric mean of Laspeyres
index and Paasche index, was used and shown as follow-
ing formulas:
PPPLaspeyres
jk =
N
ipij·qik

N
i
pik
·
qik
=N
iPPPi
jk ·Wik (3)
PPP
Paasche
jk =
N
ipij·qij

N
ipik
·
qij=1

N
i
1
PPPi
jk
·
Wij (4)
PPPFisher
jk =

PPPLaspeyres
jk
·
PPPPaasche
jk
1
2 (5)
where
Wik
represents the expenditure weight of basic
heading
i
in region
k
and
W
ij represents the expendi-
ture weight of basic heading
i
in region
j
.
The multilateral index
e nal step was to adjust the bilateral index between
every two regions to the multilateral index, which could
measure price dierences across regions. ere have been
various methods proposed to compute the multilateral
index with the development of the ICP, and two methods
were used in this study. e rst and main method was
the Gini-Éltetö-Köves-Szulc (GEKS) method, which orig-
inated with Gini (1930), and was independently redis-
covered by Éltetö and Köves (1964) and Szulc (1964). It’s
the standard method in ICP, which has the advantage of
that each region is treated in a symmetric way and is fully
consistent with the economic approach to index number
theory [23]. e multilateral index between region j and k
based on GEKS method can be calculated by the follow-
ing formula:
GE
KSjk =
M
l
[Fjl
•
Flk]1/M (6)
where
F
jl represents the Fisher index between region
j
and
l
,
Flk
represents the Fisher index between region
l
and
k
, and
M
represents the number of regions for
comparison.
Another one was the minimum spanning tree (MST)
method, which was introduced by R. J. Hill (1999). It’s
an alternative to the GEKS method in the ICP, which has
the advantages that it uses a superlative index number
formula for forming bilateral links and takes into account
substitution eects [23]. e multilateral index based on
MST method was calculated through a spanning tree
from the Paasche-Laspeyres Spread, which showed the
similarity between the Paasche index and the Laspeyres
index [29], and was given by the following formula:
PLSjk =





logPPP
Laspeyres
jk
PPPPaasche
jk





(7)
e intraclass correlation coecient (ICC) was used to
assess the agreement between two methods and ensure
the robustness of the estimation [30].
Statistical analysis
To further estimate the long-term spatial price index for
statistical analysis, this study rst used the Medical Ser-
vices Consumer Price Index (MSCPI) for each region to
extrapolate the GEKS results of 2023 by the following
formula [10]:
GE
KSjk,t−1=
GEKS
jk,t
×MSCPI
k,t
MSCPI
j,t
(8)
where
GEKS
jk,t and
GEKS
jk,t
−1
represent GEKS
index between region
j
and
k
in period
t
and
t−1
,
MSCPI
j,t and
MSCPI
k,t are MSCPI of region
j
and
k
, respectively. e MSCPI in some regions were
unavailable and the Healthcare Price Index was used
instead.
Based on the extrapolation, we constructed a panel
data on medical service price levels of 31 regions from
2015 to 2023. To examine the correlation between price
levels and income levels, we estimated a two-way xed
eects regression model (Model 1), which could be writ-
ten as follows:
lnPriceit =β0+β1lnIncomeit +µi+γt+it
(9)
where
lnPriceit
represents the log-transformed GEKS
index estimated by the formula 8 of region
i
in year
t
,
lnIncomeit
is the log-transformed disposable income per
capita of region
i
in year
t
,
µi
and
γt
are region xed
eect and time xed eect respectively,
it
is the error
term. We also added a set of time-varying control vari-
ables to the basic model (Model 2). According to relevant
policies, the capacity of medical insurance fund balance
was the key factor when the departments of price admin-
istration considered price adjustments. To avoid endo-
geneity problems, we used the aordable months by the
fund at the beginning of the year as the control variable,
which was dened as the accumulated fund balance at
last year-end divided by the average monthly expenditure
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 5 of 10
Li and Liu BMC Public Health (2024) 24:2353
over the last year. Besides, the population characteristics
including the proportion of population aged 65 years and
above, the health system characteristics including the
basic medical insurance coverage, the density of physi-
cians and hospital beds, were also included as the con-
trol variables. In addition, we selected the time interval
before the COVID-19 pandemic (from 2015 to 2019)
to repeat the above regression as the robustness check
(Model 3).
We also investigated the impact of price dierences
using the two-way xed eects regression model, which
could be written as follows:
ln
Y
it
=β
0
+β
1
lnPrice
it
+X

it
τ+µ
i
+γ
t
+
it
(10)
where
ln Yit
represent the log-transformed variables
of interest, including the average number of annual vis-
its per capita (Model 4), the hospitalization rate (Model
5), the average medical expense per visit (Model 6), the
average medical service expense per visit (Model 7), the
average medical expense per inpatient (Model 8), and the
average medical service expense per inpatient (Model 9).
lnPriceit
is the log-transformed GEKS index estimated
by the formula 8 of region
i
in year
t
. X

it
is a set of time-
varying control variables including the GDP per capita,
the proportion of population aged 65 years and above,
the basic medical insurance coverage, the density of phy-
sicians and hospital beds, the average length of stay in
hospitals, and the out-of-pocket health expenditure as a
share of the total health expenditure (THE).
µi
and
γt
are region xed eect and time xed eect respectively,
it
is the error term.
e average medical service expense was estimated by
deducting the average medication expense from the aver-
age medical expense. All variables above except for the
proportion variable were log transformed. Two-way clus-
ter-robust standard errors (at region and time level) were
used in all models to account for the potential cross-sec-
tionally and serially correlation [31]. All variables’ data
were collected from the China Statistical Yearbook and
the China Health Yearbook. All calculation process and
statistical analysis were done using R (version 4.3.2).
Results
Regional price dierences of medical services
Figure 1 shows the geographic distribution of medical
service price levels across 31 provincial regions in China.
Based on the GEKS method, Liaoning province had the
highest price level of medical services with an SPI of
118.7, which means the price level of Liaoning was 18.7%
higher than that in Shanghai, whereas in Guangxi prov-
ince the price level was the lowest with an SPI of 68.22
and medical services were priced at 68.22% of that in
Shanghai. e max/min ratio of the SPI was 1.74, which
means the price level of medical services in Liaoning
province was 74% higher than that in Guangxi province.
e detailed estimates of the SPI are shown in Table2.
e descriptive statistics of the spatial price index
under dierent estimation methods are shown in Table3.
e estimation results given by the two methods were
similar in terms of descriptive statistics. e ICC was
close to 1, which showed the consistency of results from
dierent methods.
Table4 presents the descriptive statistics of the extrap-
olated spatial price index. e standard deviation of the
SPI was 15.05 in 2015 and 12.77 in 2023. Similarly, the
max/min ratio also dropped from 1.95 in 2015 to 1.74 in
2023. e descending trend of these descriptive statis-
tics shows that the price dierences of medical services
across regions has narrowed over this period.
Correlation between price levels and income levels
e regression estimates from Model 1, Model 2 and
Model 3 are presented in Table5. In Model 1, there was
a signicant negative correlation between medical ser-
vice price levels and income levels. After controlling for
the covariates (Model 2), the sign and signicance of the
correlation are consistent with Model 1, which is also the
case in the robustness check (Model 3). Besides, there
was a positive association between medical service price
levels and the aordable months by the fund in both
Model 2 and Model 3, which further indicated the capac-
ity of medical insurance fund balance may be the critical
factor considered by the departments of price adminis-
tration when implementing price adjustment.
Impacts of price dierences
Table6 shows the impacts of medical services price dif-
ferences on variables of interest. A signicant negative
association between medical service price levels and
annual visits per capita was found in Model 4. By con-
trast, there was no signicant association between medi-
cal service price levels and hospitalization rates according
to Model 5. In terms of the impacts on medical expen-
diture, we found that a 1% increase in medical service
price levels was signicantly associated with a 0.34% and
0.24% increase in the medical service expense per outpa-
tient visit and per inpatient respectively, while there was
no signicant association between price levels and total
medical expense.
Discussion
To our best knowledge, this is the rst study to estimate
the spatial price index of medical services and measure
price dierences across 31 provincial regions in China.
e sample price data involved in this study covered
over 1000 medical services and had similar structure
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 6 of 10
Li and Liu BMC Public Health (2024) 24:2353
compared with the national specication, which highly
demonstrated the representativeness. In order to ensure
the robustness of the estimation, this study used two
dierent methods based on PPPs to estimate the spatial
price index and obtained consistent estimation results.
According to the results, this study found that there were
signicant price dierences of medical services across
regions in China. e price level in the highest-price
region was found to be 74% higher than the lowest. is
was understandable due to the decentralization of price
administration. is study also found a descending trend
in the descriptive statistics for the extrapolated results
of spatial price index, which shows the spatial price con-
vergence of medical services and is consistent with gen-
eral price converging in China found by previous studies
[32]. is study presented the methodology for estimat-
ing the spatial price index of medical services and mea-
sured current price dierences across regions, which
could provide implications for future price adjustments.
With the advancement of the deepening medical service
price reform in China, the departments of medical ser-
vice price administration should pay more attention to
the price dierences across regions and the spatial price
index of medical services could be used as a decision tool
to monitor regional price levels and provide references
for price adjustments.
To examine the association between price levels and
income levels, this study constructed a two-way xed
eects model based on the panel data of 31 provincial
regions. A signicant negative correlation between price
levels and income levels was found, which is in contrast
to the Penn eect shown in the general price level [10].
is is not surprising considering the health system
and the medical service price administration in China.
e commonweal of public medical institutions was
emphasized when the departments of price administra-
tion set the price ceiling for public medical institutions
to ensure the access to medical services. Such price may
not fully cover the cost of delivering medical services,
so China has implemented medical service price reform
and established the dynamic price adjustment mecha-
nism in recent years. However, the regional price levels
were given less consideration due to the decentralization,
while the capacity of medical insurance fund balance was
Fig. 1 The price levels of medical services across 31 provincial regions in China. Notes The price level of Shanghai was 100, based on the GEKS method
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 7 of 10
Li and Liu BMC Public Health (2024) 24:2353
the critical factor during the price adjustment process.
is was further conrmed by our empirical analysis. A
positive association between price levels and aordable
months by medical insurance fund was found, which
indicates the capacity of fund balance could facilitate the
price adjustment and thus drive the price level of medical
services higher. is may also explain to a certain extent
Table 2 The estimated spatial price index of medical services for
31 provincial regions in China
Region SPI
(GEKS Method)
SPI
(MST Method)
Liaoning 118.70 118.45
Inner Mongolia 114.01 113.77
Shandong 111.68 111.60
Guangdong 109.73 109.67
Tianjin 104.04 103.94
Qinghai 102.38 102.18
Hubei 102.38 102.46
Shaanxi 101.82 101.63
Beijing 100.91 101.05
Xinjiang 100.27 100.16
Shanghai 100.00 100.00
Henan 99.63 99.45
Hainan 96.33 96.26
Gansu 94.14 94.05
Jilin 93.32 93.11
Fujian 93.04 92.96
Sichuan 90.49 90.36
Heilongjiang 89.92 90.18
Guizhou 89.61 89.52
Shanxi 89.13 89.05
Jiangxi 89.10 89.10
Hunan 84.65 84.55
Chongqing 83.61 83.41
Anhui 82.81 82.81
Jiangsu 82.42 82.35
Zhejiang 79.41 79.37
Tibet 79.26 79.11
Hebei 77.67 77.60
Ningxia 75.25 75.32
Yunnan 68.55 68.55
Guangxi 68.22 68.21
Notes Shanghai was the base (= 100). SPI: Spatial price index. GEKS: Gini-Éltetö-
Köves-Szulc. MST: Minimum spanning tree
Table 3 Descriptive statistics of the spatial price index for
medical services
Variables nMean
(SD)
Median
(IQR)
Min, Max
(Max/Min)
ICC
(95% CI)
SPI
(GEKS Method)
31 92.66
(12.77)
93.04
(83.21, 101.37)
68.22, 118.70
(1.74)
0.98
(0.95,0.99)
SPI
(MST Method)
31 92.59
(12.73)
92.96
(83.11, 101.34)
68.21, 118.45
(1.74)
Notes SPI: Spatial price index. GEKS: Gini-Éltetö-Köves-Szulc. MST: Minimum
spanning tree. SD: Standard deviation. IQR: Interquartile range. ICC: Intraclass
correlatio n coecient. 95% CI: 95% Conden ce interval
Table 4 Descriptive statistics of the spatial price index from 2015
to 2023
Variable Mean
(SD)
Median
(IQR)
Min, Max
(Max/Min)
SPI for 2015 103.19
(15.05)
102.20
(92.78, 115.13)
68.27, 133.30
(1.95)
SPI for 2016 99.08
(14.31)
99.72
(91.55, 108.26)
66.23, 131.93
(1.99)
SPI for 2017 95.52
(14.39)
96.00
(86.07, 106.74)
63.80, 124.60
(1.95)
SPI for 2018 96.09
(14.80)
97.59
(84.30, 107.23)
64.23, 123.34
(1.92)
SPI for 2019 94.26
(13.97)
94.86
(83.52, 103.94)
62.24, 119.65
(1.92)
SPI for 2020 96.69
(14.20)
96.94
(86.83, 106.36)
67.60, 125.87
(1.86)
SPI for 2021 95.78
(13.68)
95.41
(86.10, 104.79)
69.39, 124.14
(1.79)
SPI for 2022 91.90
(12.86)
91.14
(82.78, 100.55)
67.34, 118.82
(1.76)
SPI for 2023 92.66
(12.77)
93.04
(83.21, 101.37)
68.22, 118.70
(1.74)
Notes SPI: Spatial price index. SD: Standard deviation. IQR: Interquartile range.
Results of SPI from 2015 to 2022 were extrapolated based on the GEKS method
results of 2023
Table 5 The two-way xed eects model results of correlation
between medical service price levels and income levels
Variables Medical service price level
(log transformed)
(1) (2) (3)
Disposable income per capita
(log transformed)
-0.83*** -1.06*** -
1.70***
(0.24) (0.29) (0.26)
Aordable months by the fund
(log transformed)
0.12*0.10*
(0.06) (0.05)
Proportion of population above
65
-1.23 -2.04*
(0.86) (0.96)
Basic medical insurance coverage -0.02 -0.02
(0.01) (0.01)
Density of physicians (log
transformed)
0.22 0.12
(0.13) (0.17)
Density of hospital beds (log
transformed)
-0.14 0.03
(0.08) (0.16)
Observations 279 217 155
R20.21 0.34 0.35
Notes***, ** and * denote the signi cance at the 0.1%, 1% and 5% level, respec tively.
Two-way cluster-robust standard errors are reported in parentheses. Medical
service price level is the spatial price index estimated using the GEKS method.
Aordable m onths by the fund is de ned as the accumulate d medical insurance
fund balan ce at last year-end divid ed by the average monthly e xpenditure over
the last year. Basic medical insurance coverage is dened as the percentage
of the population covered by Chinese basic medical insurance. Density of
physicians and hospital beds are de ned as the total number of p hysicians and
hospital beds per 10,000 population, respec tively
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 8 of 10
Li and Liu BMC Public Health (2024) 24:2353
why price levels of medical services was negatively asso-
ciated with income levels. Previous studies have found
that regions with higher socioeconomic levels tend to
have less medical insurance fund balance [33], which may
further restrain the rise in price level. is study aimed to
examine the correlation between price levels of medical
services and income levels, the determinants and inu-
encing mechanisms of medical service price levels need
to be further explored in future works. What’s more,
further studies should focus on the reasonable price dif-
ferences across regions. Considering the specic nature
of medical services, more indicators should be included
in addition to the cost and the income level, such as the
insurance fund balance and the economic burden for
patients. e pilot program for deepening medical ser-
vice price reform has proposed to establish the dynamic
price adjustment mechanism by medical service catego-
ries, so regional price levels could be further measured
for dierent categories.
As highlighted before, price dierences of medical
services across regions may have substantial impacts on
the utilization of medical services and regional medi-
cal expenditure variations, which were also indicated by
the results of our empirical analysis. is study found a
signicant negative association between medical ser-
vice price levels and average number of annual visits per
capita, which presents the price elasticity of demand for
medical services and is consistent with previous studies
[34]. In contrast, no signicant association was found
between medical service price levels and hospitalization
rates, which may be due to higher reimbursement rates
for inpatient services than outpatients in China, mak-
ing inpatient services less sensitive to price levels [35]. In
terms of the impacts on medical expenditure, no signi-
cant association between medical service price levels and
total medical expense was found for outpatients or inpa-
tients. is is mainly because the total medical expense
could also be determined by pharmaceutical prices and
other nonprice factors. Nevertheless, we found a posi-
tive association between medical service price levels and
medical service expense for both outpatients and inpa-
tients. More specically, a 1% increase in medical ser-
vice price levels may result in a 0.34% increase in medical
service expense per outpatient visit, while the elasticity
for inpatients was lower at 0.24%. However, the medical
service expense in this study was estimated by deduct-
ing the average medication expense from the average
medical expense, and such estimations for inpatients may
still contain non-service expense such as consumable
expense, thereby underestimating the impact of medi-
cal service price levels for inpatients. Combined with the
correlation between price levels and income levels, price
Table 6 The two-way xed eects model results for the impact of medical service price dierences
Outpatient Inpatient
Variables Annual visit Hospital-
ization rate
Medical
expense
Medical ser-
vice expense
Medical
expense
Medical
service
expense
(4) (5) (6) (7) (8) (9)
Medical service price level -0.19** 0.01 -0.04 0.34*** 0.08 0.24*
(0.07) (0.02) (0.04) (0.07) (0.06) (0.09)
GDP per capita 0.43*0.06*-0.06 0.17 -0.15 -0.22
(0.17) (0.02) (0.10) (0.16) (0.12) (0.16)
Proportion of population above 65 0.58 0.19 -1.97** -1.20 -1.03 -0.71
(0.58) (0.13) (0.68) (1.00) (0.75) (0.97)
Basic medical insurance coverage -0.02 -0.00 -0.00 -0.03 0.05*0.06*
(0.02) (0.00) (0.01) (0.03) (0.03) (0.03)
Density of physicians 0.34** 0.00 0.11 0.30** 0.12 0.14
(0.11) (0.03) (0.13) (0.12) (0.18) (0.23)
Density of hospital beds -0.02 0.16*** -0.25*-0.57*** -0.23 -0.19
(0.12) (0.02) (0.11) (0.09) (0.16) (0.23)
Average length of stay 0.51*** 0.53***
(0.08) (0.12)
Out-of-pocket expenditure (% of THE) -0.48 -0.02 0.11 0.60 0.35 0.33
(0.26) (0.06) (0.32) (0.35) (0.35) (0.35)
Observations 155 155 155 155 155 155
R20.40 0.37 0.18 0.40 0.26 0.23
Notes***, ** and * denote the signicance at the 0.1%, 1% and 5% level, respectively. Two-way cluster-robust standard errors are reported in parentheses. Annual
visit is the average number of annual visits per capita. Medical expense is the average medical expense per visit/per inpatient. Medical service expense is the
average medical service expense per visit/per inpatient. Average length of stay refers to the average length of stay in hospitals. Out-of-pocket expenditure (% of
THE) is dened as the out-of-pocket health expenditure as a share of the total health expenditure. All variables above except for the proportion variable were log
transformed
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 9 of 10
Li and Liu BMC Public Health (2024) 24:2353
dierences of medical services across regions may raise
concerns about the access to medical services and the
equity. In future price adjustments, China should narrow
and keep reasonable price dierences of medical services
across regions and take great measures to reduce the
impacts of price dierences such as increasing medical
insurance coverage and providing more nancial protec-
tion to patients in low-income regions.
is study also has the following limitations. Firstly,
the comparability is an important principle in regional
price comparisons. We matched the same services using
national codes and detected the outlier price data to
ensure the comparability. However, the dierential char-
acteristics of medical services may still exist due to dif-
ferent price schedule across regions, which may lead to
misestimation of regional price levels. Furthermore, the
same service may still have dierent quality in dierent
regions, which could lead to the underestimation or over-
estimation of price levels. Future studies should measure
and account for dierences in the quality of medical ser-
vices to estimate more accurate spatial price index. Sec-
ondly, we used the price data in the medical service price
schedule of each region, which was the ceiling price for
public medical institutions. However, the medical ser-
vice price may vary within regions due to the decentral-
ization, which may lead to price data used in this study
deviating from actual prices. Future studies should con-
sider prices from hospital charge data. irdly, the ideal
weights data should be from the same year as price data,
while this study used the estimated weights data for 2021
due to data availability. Similarly, we used the MSCPI
to extrapolate the spatial price index due to lack of the
historical price data. In future studies, the periodic price
and weights data need to be gathered to facilitate the esti-
mation of the spatial price index.
Conclusion
Measuring price dierences of medical services across
regions has signicant importance for ensuring the equal
access and future price adjustments. is study estimated
the spatial price index of medical services to measure
price dierences and found that regions in China had sig-
nicant gaps in medical service price levels. What’s more,
price levels tended to have a negative association with
income levels and the price dierences may have further
impacts on utilization of medical services and medical
expenditure. is would raise concerns about the access
to medical services and the health equity especially for
patients in low-income region. Findings from this study
may suggest that countries including China should pay
more attention to regional price levels of medical ser-
vices and promote the regional coordination to further
optimize the price levels and keep reasonable price dif-
ferences across regions. Great measures should be taken
including increasing medical insurance coverage and
enhancing nancial protection to ensure patients have
equal access to medical services regardless of the income
level and better achieve the universal health coverage.
Abbreviations
GDP Gross Domestic Product
ICP International Comparison Program
SPI Spatial Price Index
PPPs Purchasing Power Parities
NHSA National Healthcare Security Administration
CI Condence Interval
CPD Country Product Dummy
GEKS Gini-Éltetö-Köves-Szulc
MST Minimum Spanning Tree
ICC Intraclass Correlation Coecient
MSCPI Medical Services Consumer Price Index
THE Total Health Expenditure
SD Standard Deviation
IQR Interquartile Range
Acknowledgements
Not applicable.
Author contributions
B.L. designed the study and contributed to the data analysis, data
interpretation, reviewing and editing. L.L. contributed to the literature search,
data collection, data analysis, data interpretation, and writing. All authors read
and approved the nal manuscript.
Funding
Bao Liu acknowledges nancial support from the National Natural Science
Foundation of China (No. 72074050). The funding of the study had no role in
study design, data collection, data analysis, data interpretation, and writing of
the article. All authors had nal responsibility for the decision to submit it for
publication.
Data availability
The price schedules data of certain provinces are available from the
corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Received: 4 June 2024 / Accepted: 7 August 2024
References
1. Brandt L, Holz CA. Spatial price dierences in China: estimates and implica-
tions. Econ Dev Cult Change. 2006;55(1):43–86.
2. Laureti T, Rao DP. Measuring spatial price level dierences within a country:
current status and future developments. Stud Appl Econ. 2018;36(1):119–48.
3. Aten BH, Figueroa EB, Vengelen BM. Real personal income and regional price
parities for states and metropolitan areas, 2008–2012. Survey of Current
Business 94 (6). 2014. https://apps.bea.gov/scb/pdf/2014/06%20June/0614_
real_personal_income_and%20_regional_price_parities_for_states_and%20
metrpolitan_areas.pdf. Accessed January 21, 2024.
4. Bailey S, National Statistics. Relative regional consumer price levels of goods
and services, UK: 2016. Oce for. 2018. https://www.ons.gov.uk/economy/
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 10 of 10
Li and Liu BMC Public Health (2024) 24:2353
inationandpriceindices/articles/relativeregionalconsumerpricelevel-
suk/2016. Accessed January 12, 2024.
5. Rao DP, Hajargasht G. Stochastic approach to computation of purchasing
power parities in the International Comparison Program (ICP). J Economet-
rics. 2016;191(2):414–25.
6. Aten BH. Regional Price parities and Real Regional Income for the United
States. Soc Indic ReS. 2017;131(1):123–43.
7. Mishra A, Ray R. Spatial variation in prices and expenditure inequalities in
Australia. ECON REC. 2014;90(289):137–59.
8. Cheung Y, Fujii E. The Penn eect within a country: evidence from Japan. Oxf
Econ Pap. 2014;66(4):1070–89.
9. Janský P, Kolcunová D. Regional dierences in price levels across the
European Union and their implications for its regional policy. Ann Reg Sci.
2017;58(3):641–60.
10. Chen M, Wang Y, Rao DP. Measuring the spatial price dierences in China
with regional price parity methods. World Econ. 2020;43(4):1103–46.
11. Samuelson PA. Facets of Balassa-Samuelson Thirty years Later*. REV INT ECON.
1994;2(3):201–26.
12. Barber SL, Luca L, Paul O. Price setting and Price Regulation in Health Care
Lessons for Advancing Universal Health Coverage: lessons for advancing
Universal Health Coverage. Paris/WHO, Geneva;: OECD Publishing; 2019.
13. Martin AB, Hartman M, Washington B, Catlin A, Team NHEA. National health
care spending in 2017: growth slows to post–great recession rates; share of
GDP stabilizes. Health Aair. 2019:10–1377.
14. Lorenzoni L, Dougherty S. Understanding dierences in Health Care spend-
ing: a comparative study of prices and volumes across OECD Countries.
Health Serv Insights. 2022;15.
15. Sloan FA, Edmunds M. Geographic Adjustment in Medicare payment: phase I:
improving Accuracy. Washington DC: National Academies; 2012.
16. Monitor. A guide to the market forces factor. 2013. https://assets.publishing.
service.gov.uk/government/uploads/system/uploads/attachment_data/
le/300859/A_guide_to_the_Market_Forces_Factor.pdf. Accessed August 21,
2023.
17. Petrou P, Vandoros S. Pharmaceutical price comparisons across the Euro-
pean Union and relative aordability in Cyprus. HEALTH POLICY TECHN.
2016;5(4):350–6.
18. Vogler S, Vitry A. Cancer drugs in 16 European countries, Australia, and
New Zealand: a cross-country price comparison study. Lancet Oncol.
2016;17(1):39–47.
19. Thai LP, Vitry AI, Moss JR. Price and utilisation dierences for statins between
four countries. Expert Rev Pharm Out. 2018;18(1):71–81.
20. Dunn A, Grosse SD, Zuvekas SH. Adjusting health expenditures for ination:
a review of measures for health services research in the United States. Health
Serv Res. 2018;53(1):175–96.
21. Fu H, Li L, Yip W. Intended and unintended impacts of price changes
for drugs and medical services: evidence from China. SOC SCI MED.
2018;211:114–22.
22. Chai K, Zhang Y, Chang K. Regional disparity of medical resources and its
eect on mortality rates in China. Front Public Health. 2020;8:8.
23. World Bank. Measuring the real size of the World Economy: the Framework,
Methodology, and results of the International Comparison Program (ICP).
Washington, DC: World Bank; 2013.
24. Thompson KJ, Sigman RS. Statistical methods for developing ratio edit toler-
ances for Economic Data. J OFF STAT. 1999;15:517–36.
25. Balk BM. Price and quantity index numbers: models for measuring aggregate
change and dierence. Cambridge University Press; 2008.
26. World Bank. Purchasing Power parities and the size of World economies:
results from the 2017 International Comparison Program. Washington, DC:
World Bank; 2020.
27. Hajargasht G, Rao DP. Multilateral index number systems for international
price comparisons: properties, existence and uniqueness. J MATH ECON.
2019;83:36–47.
28. Summers R, INTERNATIONAL PRICE COMPARISONS, BASED UPON INCOM-
PLETE DATA*. REV INCOME WEALTH. 1973;19(1):1–16.
29. Hill RJ. Comparing price levels across countries using minimum-spanning
trees. REV ECON STAT. 1999;81(1):135–42.
30. Chen C, Barnhart HX. Assessing agreement with intraclass correlation
coecient and concordance correlation coecient for data with repeated
measures. COMPUT STAT DATA AN. 2013;60:132–45.
31. Sun L, Huang Y, Ger T. Two-way cluster-robust standard Errors—A meth-
odological note on what has been done and what has not been done in
Accounting and Finance Research. Theoretical Econ Lett. 2018;8(09):1639.
32. Li J, Li Z, Sun P. Does the razor’s edge exist? New evidence of the law of one
price in China (1997–2012). World Econ. 2018;41(12):3442–66.
33. Hengpeng Z, Mengting S. Impact of local Government revenue on surpluses
in Basic Medical Insurance funds [In Chinese]. Chin Social Secur Rev.
2023;7(06):71–88.
34. Ellis RP, Martins B, Zhu W. Health care demand elasticities by type of service. J
HEALTH ECON. 2017;55:232–43.
35. Zhou Z, Su Y, Gao J, Xu L, Zhang Y. New estimates of elasticity of demand for
healthcare in rural China. Health Policy. 2011;103(2):255–65.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional aliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com