Access to this full-text is provided by Springer Nature.
Content available from BMC Public Health
This content is subject to copyright. Terms and conditions apply.
Liu BMC Public Health (2022) 22:835
https://doi.org/10.1186/s12889-022-13240-7
RESEARCH
Determining theeect ofair quality
onactivities ofdaily living disability: using
tracking survey data from122 cities inChina
Huan Liu*
Abstract
Background: Current research on activities of daily living (ADLs) disability has mostly focused on the analysis of
demographic characteristics, while research on the microcharacteristics of individuals and the macroenvironment is
relatively limited, and these studies solely concern the impact of air quality on individual health.
Methods: This study innovatively investigated the impact of air quality on ADL disability by matching micro data of
individuals from the China Health and Retirement Longitudinal Study with data of urban environmental quality from
122 cities. In this study, an ordered panel logit model was adopted for the benchmark test, and the two-stage ordered
probit model with IV was used for endogenous treatment.
Results: This innovative study investigated the impact of air quality on ADL disability by matching individual micro
data from the China Health and Retirement Longitudinal Study with urban environmental quality data for 122 cities.
The results showed that air quality significantly increased the probability of ADL disability. The positive and marginal
effect of air quality on moderate and mild disability was higher. Generally, the marginal effect of air quality on residents’
health was negative. In terms of group heterogeneity, the ADL disability of individuals aged over 60 years, those in the
high Gross Domestic Product (GDP) group, females, and those in the nonpilot long-term care insurance group was
more affected by air quality, and the interaction between air quality and serious illness showed that the deterioration of
air quality exacerbated the ADL disability caused by serious illness; that is, the moderating effect was significant.
Conclusions: According to the equilibrium condition of the individual health production function, the ADL disability
caused by a 1% improvement in air quality is equivalent to the ADL disability caused by an 89.9652% reduction in
serious illness, indicating that the effect of improved air quality is difficult to replace by any other method. Therefore,
good air quality can not only reduce ADL disability directly but also reduce serious illness indirectly, which is equiva-
lent to the reduction of ADL disability. This is called the health impact.
Keywords: Air quality, ADL disability, CHARLS, Pollutants, Ordered logit
© The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco
mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Introduction
Since the beginning of the twenty-first century, the rapid
development of China’s economy has been accompa-
nied by a considerable increase in Gross Domestic Prod-
uct (GDP). e per capita GDP reached 72,371 yuan in
2020 [1]. Consequently, the living standards of residents
have also significantly improved. However, air pollu-
tion caused by economic development in all parts of
Open Access
*Correspondence: zcliuhuan@126.com
School of Public Administration, Zhejiang University of Finance &
Economics, No. 18 Xueyuan Street, Xiasha Higher Education Park, Hang
Zhou 310018, Zhejiang, China
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 2 of 16
Liu BMC Public Health (2022) 22:835
China also increased, negatively impacting the health of
the Chinese people. Outdoor air pollution was included
in the list of carcinogens published by the International
Agency for Research on Cancer of the World Health
Organization in 2017 because dense particulate matter in
the air can cause a significant impact on human health
[2]. Both in China and globally, environmental protection
is increasingly becoming a major issue for society as a
whole. In 2017, Comrade Xi Jinping prioritized protect-
ing the environment and maintaining harmony between
man and nature in the 19th major report of the commit-
tee party [3]. Currently, it is necessary to adhere to the
development concept of “Green mountains and green
waters are golden mountains and silver mountains” and
follow the basic state policy of conserving resources and
protecting the environment. Individuals recognize that
environmental protection is related to their fundamental
wellbeing. erefore, the study of air quality as it relates
to environmental protection has important theoretical
and practical significance.
Furthermore, from the perspective of China’s ageing
population, disability has increasingly become a major
livelihood problem. Existing research on the disabled
population mostly focuses on the analysis of public and
social policies or is conducted from a medical perspec-
tive. ese studies include the analysis of the effective-
ness of long-term care insurance (LTCI) for the disabled
population [4, 5]; the analysis of the social characteris-
tics of disabled people and their average life expectancy
[6–8]; and the analysis of the internal physical changes
that occur due to disability using the disability evalua-
tion scale [9, 10].. On the other hand, from the perspec-
tive of air quality, the study of residents’ disability is rare.
However, existing research has shown that changes in
air quality have an important impact on human health.
e change in individual health, especially the impact of
serious illness, is usually the key factor or even the only
direct factor for the impairment in activities of daily
living (ADLs). erefore, to address these gaps in the
research, this study aimed to assess the impact of air
quality on ADL disability in Chinese residents. e find-
ings discussed here will provide evidence for prioritiz-
ing government programs to deal with the issues of ADL
disability.
Literature review
ere is abundant research concerning the impact of
air pollution on health. From the macro perspective of
health impact, Usmani etal. clearly gave the definition
of air pollution, the motivation to study air pollution,
and the impact and source of air pollution and climate
change [11]. Han et al. provided a new measurement
standard for evaluating global health inequality from the
perspective of climate change and air pollution control
efficiency (abbreviated as APCI) [12]. In general, air pol-
lution is closely related to the national or regional aver-
age health level. If emission reduction efforts are shared
by all countries, in all scenarios, the benefits of common
health would far exceed the political costs [13]. Based on
the exposure response function of epidemiology, it was
revealed that the impact of future temperature changes
on citizens’ health is more significant than the change
in air pollutant concentration [14]. Among the environ-
mental indicators, cultivated land is the indicator that
shows the greatest impact on health and wealth in the
next 10 years, while air pollution has the least impact on
health and wealth for low-income countries [15]. How-
ever, it was found that environmental and air pollution
impose a great threat on the health and wealth of resi-
dents in low-income countries. Moreover, there are sig-
nificant differences in the effects of different pollutants.
From the perspective of the impact pathway of pollution,
NO2 and O3 are more important, and their AR (added
health risk) decreases significantly in urban areas with
crowded traffic, but no significant change in AR was
found in other areas with low urbanization [16].
Among the research on individual health impacts,
on the one hand, air pollution indeed has an impact
on individual health [17–21]; on the other hand, it
also affects potential medical consumption [22, 23].
In detail, (1) as one of the primary outcomes of the
impact of air pollution, the death rate of respiratory
diseases is increasing significantly [24], and this eco-
nomic cost even exceeds the economic benefits. As a
result, production efficiency decreased. For instance,
based on the HAQI (health risk-based AQI), it was
estimated that 20% of the population in the study area
was exposed to polluted air. e total mortality rates
caused by PM10, PM2.5, SO2, O3, NO2, and CO were
3.00, 1.02, 1.00, 4.22, 1.57, and 0.95%, respectively [25].
In addition, inhalable particles in air pollutants affect
individual health mainly in two ways: one is the short-
term effect on the human respiratory tract, which can
cause respiratory tract infection, chronic obstructive
pulmonary disease, lung cancer, and other respiratory
diseases [26–29]; the other is the long-term impact on
the respiratory tract that involves the triggering of the
inflammatory cascade through local inflammatory fac-
tors, ultimately leading to a significant increase in the
risk of cardiovascular and nervous system diseases
[30–34]. As the research revealed, when PM10 and O3
in air pollutants increase by 10 μg/m3 and 10 ppb, the
number of visitors to respiratory hospitals in 1 day will
increase by 10.39 and 10.93%, respectively. is would
bring about additional medical expenses of $67 mil-
lion and $70 million, respectively [35]. Furthermore,
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 3 of 16
Liu BMC Public Health (2022) 22:835
the health effects of air pollution vary under different
socioeconomic statuses. For example, self-rated air pol-
lution has the greatest impact on the self-rated health
of low socioeconomic groups, while with the improve-
ment of socioeconomic status, the impact of self-rated
air pollution on self-rated health decreases [36]. (2)
Air pollution indirectly affects residents’ medical con-
sumption. Sun etal. demonstrated that air pollution is
also the main factor that influences residents’ expen-
ditures on health management [37]. eoretically, air
pollution affects health mainly in two ways: first, the
reduction in sleep time caused by ambient air pollution
is not conducive to health; second, people spend more
time on sedentary activities to avoid exposure to air
pollution, which will indirectly lead to an increase in
personal medical expenditure [38]. Additionally, from
the empirical results, air pollution will lead to a sig-
nificant increase in medical expenses, hospitalization
expenses and extrabudgetary expenses [38]. For exam-
ple, Liu etal. estimated age- and cause-specific prema-
ture deaths and quantified related health damage with
the measurement of the age-adjusted value of statistical
life (VSL). eir results suggest that while premature
deaths fell as a result of China’s clean air actions, the
health costs of air pollution remained high [39].
Most of the existing studies on residents’ ADLs are
based on the micro viewpoints of individual disease
risk. For example, in ADL disability assessment, based
on the diagnosis rate of major diseases, individual
disease risks are defined by establishing the relevant
Disability Assessment Scale [5, 6]. However, even in
countries or regions with long-term implementation
of health care insurance, the impact of air pollution
on residents’ ADL disability has rarely been investi-
gated, neither in practice nor in theory. is also illus-
trates the major significance of this study. Current
research in this field focuses on the factors that influ-
ence the population’s health via urban green spaces,
the ecological environment and air quality. e find-
ings from such studies show that the deterioration of
the ecological environment negatively impacts human
health. However, there are some gaps in the existing
research. First, although there are relatively abundant
studies on the impact of the ecological environment
on individual health, the majority of these focus on
direct health effects, ignoring the cumulative indirect
effects of changes in environmental quality. Further-
more, these studies focus only on medical expenses.
Second, in the measurement of air quality, the tradi-
tional air pollution index (API) or the concentration of
a single pollutant are often used for testing. Although
it is suitable to investigate the impact of a single pol-
lutant, for estimates that are closer to the real-world
impact, testing should include a comprehensive list
of pollutants. ird, existing studies mainly focus on
the impact of air quality on individual health without
fully considering internal transmission mechanisms
through which air quality affects health. To address
these gaps, this study focused on the following points.
First, we investigated the indirect impact of air pollu-
tion by assessing the decline in residents’ basic activi-
ties of daily living (ADLs). Second, sulfur dioxide (SO2),
nitrogen dioxide (NO2) and inhalable particles (PM10)
were included as proxy variables, and China Health
and Retirement Longitudinal Study (CHARLS) data
from 2015 and 2018 were matched with macro regional
air quality data to construct panel data. Heterogene-
ity analysis and endogenous problem processing were
used to ensure the reliability of the test results. e air
quality index (AQI) was introduced to investigate the
robustness of the results, considering the heterogene-
ity of a single air quality index and the overall impact.
ird, by constructing the health production function,
we investigated the substitution effect of air quality and
serious illness on individual ADL disability and tested
the transmission mechanism of air quality impacting
individual ADL disability.
Methods
Theoretical hypothesis: impact ofair quality onhealth
e health demand model was first proposed by Gross-
man [40], and the health production function, which
is the core of the supply model, is derived from it. e
health production function can be divided into macro
and micro parts, which are interrelated. Among them,
the microhealth production function emphasizes the
relationship between family- or individual-level medical
and health input and individual health output through
macro policy intervention [41, 42]. e macrohealth pro-
duction function considers the overall output effect of
national health from the perspective of macroeconomics,
government health expenditure, and medical insurance
[43]. is study investigated air quality effects from a
macro perspective by analysing the macro health produc-
tion function. e theoretical mechanism of the impact
of air pollution on residents’ health is shown in Fig.1.
Based on Grossman’s health demand model, Filmer
etal. [44] constructed a macro health production func-
tion model. Health needs are formed by the correlation
between health and related factors that improve health.
e core of the health production function is composed
of output factors and health inputs. Due to the relevant
hypothesis bias in the micro field, there is an estima-
tion bias in the analysis of medical and health policy
inputs and outputs using the perfect competition mar-
ket model. erefore, more nonendogenous factors must
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 4 of 16
Liu BMC Public Health (2022) 22:835
be explained. When health economists use the general
production function theory, combined with health char-
acteristics, they put forward that in the process of main-
taining or improving health, the input and output of
medical and health resources are included in the basic
health production function. erefore, the general health
production function can be expressed as:
Equation (1) is the national health level at a certain
time point, where S represents the input of social fac-
tors, Y is the input of economic variables, E is the input
of educational variables, P is the input of medical and
health policies and Z is the social health investment.
However, the existing health production function does
not consider the impact of the natural environment or
air quality. erefore, this study used individual ADL
(1)
H=F(S,Y,E,P,Z)
disability as a proxy for health variables and assumed
that ADL disability is influenced by sociodemographic,
regional environmental and individual health charac-
teristics [45]. Here, sociodemographic characteristics
include gender, age, household-registered marital sta-
tus, etc. Regional environmental characteristics include
regional financial expenditure, per capita GDP, popula-
tion density, sunshine duration and rainfall. Individual
health characteristics include serious illness, depression
and self-reported health. erefore, the health produc-
tion function can be adjusted as follows:
In Eq. (2), ADL _ disability is calculated; R on the
right side of the equation represents the regional envi-
ronmental characteristics, H represents the individual
health characteristics, and S represents the individuals’
(2)
ADL_disability =F(R,H,S)
Fig. 1 Theoretical mechanism of the impact of air pollution on Residents’ ADL disability
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 5 of 16
Liu BMC Public Health (2022) 22:835
sociodemographic characteristics. Based on existing
research and the objectives of this study, air quality was
considered the primary factor of ADL disability, while
other influencing factors were taken as control variables.
erefore, Eq. (2) can be adjusted as follows:
e pilot for China’s LTCI showed that the most
important cause of disability for most severely disa-
bled persons was the occurrence of serious illness [5].
erefore, this study considered the rate of serious ill-
ness (i.e., diagnosis rate of serious illness) as an impor-
tant regulatory index to investigate the detrimental
effect of air quality on individual ADL. e Chronic on
the right of Eq. (3) is the serious illness rate. In addi-
tion, after controlling for other factors, we can further
investigate the substitution relationship between air
quality and serious illness, which can be derived from
Eq. (3). When the individual ADL disability remains
unchanged, it should be equal to 0, that is:
(3)
ADL_disability =F(Air, Chronic,Other)
(4)
dADL
_disability =
∂ADL_disability
∂Air
•dADL_disability+
∂ADL_disability
∂Chronic
•dChronic =
0
en, the marginal substitution rate between air quality
and residents’ serious illnesses can be:
Equation (5) shows the substitution relationship between
air quality and individual serious illness under the condi-
tion of constant ADL disability. erefore, the reduction in
individual serious illness by a one-unit improvement in air
quality represents the health impact of air quality, which
is measured by the changes in ADL disability due to air
quality. e empirical method testing the impact of air
pollution on residents’ health is shown in Fig.2.
Test model
Based on the above theoretical analyses of the health
impact of air quality, this study further constructed
an empirical test model. Considering that the core
explanatory variable of this study was residents’ ADL
(5)
MRS|Air =
dChronic
dAir
=−
∂ADL_disability/∂Air
∂ADL_disability/∂Chronic
Fig. 2 Effect of air pollution on the ADL disability of residents
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 6 of 16
Liu BMC Public Health (2022) 22:835
disability, we classified ADL disability. Please refer to
the definitions of core explanatory variables and clas-
sifications in the data section for specific explanations.
This implied that the traditional OLS estimation would
result in bias; therefore, the ordered panel logit model
was selected for the test:
In Eq. (6), ADL _ disability represents the ADL disabil-
ity of individual i living in city j in year t, which is the pri-
mary explained variable of this study; Airjt on the right
side of the equation represents the air quality of city j in
year t, which is another primary explanatory variable of
this study. In this study, SO2, NO2, and PM10 in the API
were selected as proxies of air quality, and the AQI was
selected for the robustness test. In the data processing
step, to avoid the influence of nondimensional values,
logarithmic processing was used. Hijt represents indi-
vidual health characteristics, including individual seri-
ous illness rate, self-reported health and physical pain. Rjt
represents the environmental characteristics of j city in t
year, including annual rainfall and annual sunshine dura-
tion. S indicates sociodemographic characteristics such
as gender, age, marital status, etc. Since the panel logit
model only provides the test results of random effects,
to ensure reliable results, the individual effect, regional
effect, and year effect were controlled simultaneously in
the model, which were λi, δj and ηt in Eq. (6), respectively.
εijt represents random error. Furthermore, the health pro-
duction function of Eq. (6) is nonlinear; therefore, it satis-
fies the following conditions:
where ADL _ disabilityijt
∗ is the unobservable continu-
ous variable of ADL _ disabilityijt, which is the latent vari-
able and satisfies the assumption of linearity. In Eq. (7),
r0, r1, r2... denote the parameters to be estimated. To keep
the ADL disability of residents unchanged, we can inves-
tigate how serious illness was impacted when air quality
deteriorates. Based on the above analysis, the marginal
substitution rate between air quality and serious illness
can be adjusted to Eq. (8) based on Eq. (5), where |α/β| is
the substitution rate between serious illness and air qual-
ity, as given below:
(6)
ADL_disability
ijt
=F
αln Airjt +βChronicijt +κHijt +χRjt +ϕSijt +i+δj+ηt+εijt
(7)
F
ADL_Disability ∗
ijt =
1, ADL_Disability ∗
ijt ≤r0
2, r0<ADL_Disability ∗
ijt ≤r
1
3, r1<ADL_Disability ∗
ijt ≤r
2
J,rJ−1≤ADL_Disability ∗
ijt
Considering the characteristics of the health produc-
tion function, we should determine the substitution rela-
tionship between air quality and serious illness and how
to improve air quality and reduce serious illness at the
same time when the overall ADL disability is reduced.
is is for determining the scale effect of the health pro-
duction function and verifying the marginal effect of each
variable in the real test, which will be discussed later.
Data
Individual ADL disability data
e individual micro data of this study were obtained
from the CHARLS surveys of 2015 and 2018. e data
that support the findings of this study are openly avail-
able at the following URL/DOI: http:// charls. pku. edu.
cn/. In this dataset, there were 12,520 participants from
2015 and 13,358 from 2018. By controlling for individ-
ual and time effects, as well as for sociodemographic
characteristics of the population and the macro char-
acteristics of the city, the reliability and accuracy of the
estimated effect of air quality on individual ADLs were
improved.
e core explanatory variable for the analysis was the
ADL disability of residents, and the specific indicators
were defined as follows: ADLs were determined based
on the question “whether you have difficulties in dress-
ing, bathing, eating, getting up and out of bed, going
to the toilet, controlling defecation and defecation”.
e score for this question was based on the selection
of options from 1-no difficulty, 2-difficulty but still can
be completed, 3-difficulty and need help, and 4-unable
to complete. In total, six basic self-care ability indica-
tors were used, and the total score ranged from 6 to 24.
Based on the existing classification of disability, ADL
disability was divided into five levels: serious disability,
severe disability, moderate disability, mild disability, and
healthy [6]. rough data processing, a total score of 6
was recorded as 5, which represented “healthy”; a score
of 7–9 was defined as 4, indicating a mild disability; a
score of 10–14 was recorded as 3, indicating moder-
ate disability; a score of 15–20 was defined as 2, which
indicated severe disability; and a score of 21–24 was 1,
which indicated serious disability. erefore, a higher
ADL disability score indicated a lower degree of ADLs.
(8)
MRS|Air =
∂ADL_disability/∂Air
∂
ADL_disability
/∂
Chronic
=−
α
β×
Chronic
Air
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 7 of 16
Liu BMC Public Health (2022) 22:835
e statistics of the probability of ADL disability are
presented in Table1. As shown in Table1, the rates of
serious disability, severe disability and moderate dis-
ability increased from 2015 to 2018. e proportion of
people with severe and mild ADL disability in the total
population increased from 6.29 to 7.93%, but the pro-
portion was still lower than that with mild disability.
In addition, the proportion of the healthy population
increased by a small degree during this period.
Air quality data
ere are many measurement indicators of air pollution,
such as the air quality index (AQI) and air pollution index
(API). While the main pollutants in exhaust gas were
mainly industrial emissions, the API indicator was not a
comprehensive measure of air quality [46]. e AQI is a
more comprehensive measure, and its data are released
once an hour. erefore, it is advantageous to use the
annual average AQI value to investigate the impact of air
quality on ADL disability [47].
Control variables
In addition to air quality, the main factors of ADL
disability include sociodemographic characteris-
tics and other factors. The definition and statistics
of the control variables in this study are shown in
Table 2, including the regional natural environment,
economic environment, and individual and family
characteristics.
Table 2 shows that the variation coefficients of ADL
disability in 2015 and 2018 were 0.110 and 0.128, respec-
tively. e degree of dispersion was small, and mild dis-
ability and health were the main parts. On the other
hand, the variation coefficients of the concentrations of
SO2, NO2 and PM10 were 0.652, 0.651, and 0.355 in 2015,
respectively, and changed to 0.406, 0.434, and 0.449 in
2018. us, the variations in NO2 and PM2 were simi-
lar, while the dispersion of SO2 was relatively larger. e
statistical values of the AQI in 2015 and 2018 were 85.76
and 72.14, respectively, which means that the air quality
apparently improved in 2018.
Results
Benchmark regression
In the benchmark regression, the effects of differ-
ent pollutant concentrations were tested, and the
results are presented in Table3. Models (1)–(3) are the
results of the stepwise test of air pollutant concentra-
tion effects, controlled by individual and time effects,
whereas Model (4) is based on the AQI. e results
show that both SO2 and PM10 have significant and neg-
ative effects on ADL disability. e significance level of
SO2 was low, whereas the results for the coefficient of
PM10 were more robust. In other words, higher concen-
trations of SO2 and PM10 in the air have brought about
a higher degree of ADL disability. ese results dem-
onstrate that an increased concentration of air pollut-
ants aggravates the degree of ADL disability and that
PM10 plays a more important role. e results of Model
(4) show that air quality has a significant and negative
impact on residents’ ADL disability; the worse the air
quality is, the higher the degree of residents’ ADL dis-
ability. is result proves the robustness of the results
of pollutant concentrations.
In terms of control variables, population density,
annual rainfall and annual average temperature had sig-
nificant effects on ADL disability. Population density
and annual rainfall had positive effects: the higher the
population density and annual rainfall were, the lower
the degree of ADL disability. On the other hand, annual
average temperature had negative effects: the higher the
annual average temperature was, the higher the degree
of ADL disability. Regarding individual characteristics,
household registration, depression, self-reported health
and serious illness had positive effects on ADL disability,
but marital status, disability, physical pain, gender and
education had significant and negative effects on ADL
disability.
ese results demonstrate that the concentration of
air pollutants has a significant impact on ADL disability,
and among the control variables, the basic health status
of individuals is the primary factor affecting ADL disabil-
ity. Moreover, by looking into the marginal substitution
Table 1 Probability statistics of ADL disability
ADL disability 2015 2018
Relative frequency Frequency (%) Relative frequency Frequency (%)
Serious disability 23 0.18 92 0.69
Severe disability 101 0.81 174 1.30
Moderate disability 664 5.30 794 5.94
Mild disability 2664 21.28 2482 18.58
Healthy 9068 72.43 9816 73.48
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 8 of 16
Liu BMC Public Health (2022) 22:835
effect of air quality and serious illness, to maintain the
level of ADL disability, the decrease in ADLs caused by
a 1% increase in SO2, NO2, PM10 and the AQI needs to
be compensated by a 1.2325, 0.0346, 2.087, and 2.826%
reduction in the serious illness, respectively. e substi-
tution relationship between air quality and other health
variables can also be investigated; however, they were not
of interest to this study.
Marginal eect analysis
Based on Table3, the marginal effect of air quality on ADL
disability can be further estimated, and the results are
shown in Table4. Because the ordered logit model can only
provide limited information on the signs and significance of
parameters, it is necessary to estimate the marginal effect
of air quality on ADL disability. When all explanatory vari-
ables are at the mean value, the influence of the exogenous
explanatory variables can be expressed as Eq. (9):
Table 4 shows the marginal effects of air quality on
the ADL disability of residents. PM10 is the primary
factor affecting ADL disability, and when the PM10
concentration is increased by 1 unit, the probability
(9)
∂prob
(
ADL
=
i/Air
)
∂Air
Air=Air
(i=1, 2, 3, 4, 5
)
Table 2 Descriptive statistics of main variables
Abbreviations: ADL Activities of Daily Living, AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The model controls for both the year and individual eects to consider the inuence of
unobservable factors
Variable Denition 2015 (12520) 2018 (13358)
Mean SD Mean SD
ADL disability 1 ~ 5; higher score indicated lower ADL disability 4.651 0.513 4.636 0.595
SO2SO2 content in air (μg /m3) 27.53 17.96 16.26 10.58
NO2NO2 content in air (μg /m3) 32.74 11.62 39.23 15.93
PM10 PM10 content in air (μg /m3) 94.38 40.94 89.74 40.32
AQI Dimensionless air quality; greater value indicated poorer quality 85.76 25.79 72.14 16.55
Fiscal expenditure Total annual financial expenditure of the region (million yuan) 544.9 729.6 688.0 1030
Sunshine duration Total sunshine duration in the whole year, (hour) 1814 469.0 1903 354.4
Rainfall Annual total rainfall (mm) 1067 624.8 997.3 441.2
Per capita GDP Annual regional GDP to population ratio, (yuan / person) 49,467 34,418 56,468 35,992
Population density Annual area to population ratio (Person / m2) 490.1 479.4 492.6 473.1
Average temperature Annual average temperature (centigrade) 15.24 3.867 15.08 3.926
GDP growth Regional GDP growth compared with the previous year 8.078 2.081 7.054 1.823
Green space coverage Ratio of green area to total area (in built up area) 39.54 9.130 39.96 5.022
Relative humidity Percentage of water vapor pressure in air to saturated vapor pres-
sure at the same temperature 64.65 12.39 65.03 10.64
Household register Urban = 1, rural = 0 0.401 0.490 0.405 0.491
Income 1 ~ 5 respectively represent high income, middle-high-income,
middle income, lower-middle-income and low income 2.605 0.783 2.754 0.803
Basic medical insurance Enjoying basic medical insurance = 1, no = 0 0.945 0.137 0.971 0.168
Marital status Widowed = 1, no = 0 0.103 0.304 0.125 0.330
Serious illness Number of serious illnesses diagnosed; higher value indicates a
greater number of illnesses 0.0294 0.286 0.724 1.052
Depression 1–4; higher score indicates more severe depression 2.468 0.740 2.275 0.783
Self-reported health 1 ~ 5; higher value indicates better health 2.955 0.721 2.946 0.986
Body disability 0–5; higher score indicates more severe body disability 0.154 0.444 0.145 0.445
Physical pain 1–5; higher score indicates more severe pain 1.705 0.456 2.159 1.267
Age Actual age of the individual in the survey year 59.14 10.32 58.74 10.32
Gender Male = 1, female = 0 0.478 0.500 0.474 0.499
Education level 1–11 respectively represent No formal education (illiterate),Did
not finish primary school, Sishu/home school, Elementary school,
Middle school, High school, Vocational school, Two−/Three-Year
College/Associate degree, Four-Year College/Bachelor’s degree,
Master’s degree, Doctoral degree/Ph.D.
3.390 1.001 3.477 1.935
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 9 of 16
Liu BMC Public Health (2022) 22:835
of serious disability, severe disability, moderate dis-
ability, mild disability and healthy status of residents
is significantly increased by 0.005, 0.02, 0.20, 0.79 and
1.94%, respectively. e marginal effect of NO2 is very
weak and nonsignificant. In comparison, when the SO2
concentration was increased by one unit, the increase
in the probability of serious disability, moderate dis-
ability and mild disability was 0.013, 0.12 and 0.45%,
Table 3 Impact of air quality on ADL disability: Benchmark regression
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; * p < 0.01, * p < 0.05, * p < 0.1. The pseudo log-likelihood value in the table is log pseudolikelihood
Variable (1) (2) (3) (4)
lnSO2− 0.0599*(0.0355)
lnNO2−0.0017(0.0533)
lnPM10 − 0.1056**(0.0510)
lnAQI − 0.1543**(0.0721)
Fiscal expenditure − 0.0057(0.0276) 0.0015(0.0273) − 0.0027(0.0274) − 0.0018(0.0271)
Sunshine duration − 0.0204(0.0865) − 0.0311(0.0861) − 0.0364(0.0864) −0.0219(0.0857)
Rainfall 0.1173*(0.0620) 0.1353**(0.0620) 0.1015(0.0624) 0.1232**(0.0611)
Per capita GDP 0.0403(0.0330) 0.0361(0.0357) 0.0473(0.0334) 0.0344(0.0328)
Population density 0.0596***(0.023) 0.0513**(0.0238) 0.0744***(0.025) 0.0666***(0.0237)
Average temperature −0.4412***(0.093) −0.4373***(0.093) − 0.4503***(0.094) − 0.4203***(0.0925)
GDP growth 0.0139(0.0090) 0.0159*(0.0090) 0.0174*(0.0090) 0.0173*(0.0089)
Green space coverage 0.0007(0.0022) 0.0007(0.0022) 0.0005(0.0022) 0.0004(0.0022)
Relative humidity −0.0010(0.0029) −0.0007(0.0029) − 0.0014(0.0029) −0.0017(0.0029)
Household register 0.1030***(0.036) 0.1015***(0.036) 0.0968***(0.036) 0.0701*(0.0358)
Income −0.0190(0.0223) −0.0186(0.0223) − 0.0200(0.0223) −0.0191(0.0222)
Basic medical insurance 0.2187(0.1489) 0.2232(0.1488) 0.2269(0.1489) 0.2268(0.1485)
Marital status −0.1834**(0.072) −0.1827**(0.072) − 0.1837**(0.0721) −0.2654***(0.0725)
Serious illness 0.0486*(0.0271) 0.0491*(0.0271) 0.0506*(0.0271) 0.0546**(0.0270)
Depression 0.1194***(0.025) 0.1183***(0.0251) 0.1194***(0.025) 0.1513***(0.0252)
Self-reported health 0.0930***(0.020) 0.0933***(0.0198) 0.0937***(0.020) 0.0866***(0.0197)
Body disability −0.6474***(0.049) −0.6473***(0.049) − 0.6472***(0.049) −0.6378***(0.0490)
Physical pain −0.0369*(0.0215) −0.0375*(0.0215) − 0.0370*(0.0215) −0.0415*(0.0215)
Age −0.0017(0.0020) −0.0017(0.0020) − 0.0017(0.0020) −0.3707***(0.0347)
Gender −0.3703***(0.035) −0.3705***(0.035) − 0.3707***(0.035) 0.0010(0.0020)
Education −0.0350***(0.011) −0.0350***(0.011) − 0.0349***(0.011) −0.0132(0.0110)
Individual / Year Yes Yes Yes Yes
|α/β| 1.2325 0.0346 2.0870 2.8260
sigma2_u 1.7866***(0.119) 1.7859***(0.119) 1.7901***(0.119) 1.7249***(0.1194)
Pseudo log likelihood − 27,316.835 − 27,318.331 − 27,316.157 − 27,260.191
Observations 26,218 26,218 26,218 26,218
Table 4 Marginal effect of air quality on ADL disability
Abbreviations: ADL Activities of Daily Living, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: The standard error is in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
ADL disability lnSO2lnNO2lnPM10 lnAQI
Serious disability 0.00003(0.00002) 8.42e-07(0.00003) 0.00005*(0.00003) 0.00008*(0.00004)
Severe disability 0.00013*(0.0001) 3.77e-06(0.0001) 0.0002**(0.0001) 0.0003**(0.0002)
Moderate disability 0.0012*(0.0007) 0.00003(0.0010) 0.0020**(0.0010) 0.0030**(0.0014)
Mild disability 0.0045*(0.0026) 0.0001(0.0040) 0.0079**(0.0038) 0.0115**(0.0054)
Healthy −0.0110*(0.0063) − 0.0003(0.0098) − 0.0194**(0.0094) − 0.0284**(0.0133)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 10 of 16
Liu BMC Public Health (2022) 22:835
respectively, whereas the health reduction probability
was − 1.10%. From the test of the marginal effect of the
AQI, the above results are robust. e marginal effect
of the AQI on severe, mild severe, moderate and mild
disability is positive, and the marginal effect of the AQI
on moderate and mild disability is higher. If the AQI
is increased by 1 unit, the probability of moderate and
mild disability increases by 0.30 and 1.15%, respectively.
Meanwhile, the marginal effect of the AQI on health
reaches 2.84%, which means that a 1 unit increase in
the AQI leads to a 2.84% decrease in the probability of
residents’ health.
Analysis ofgroup heterogeneity
To investigate the variations in the impact of air qual-
ity on ADL disability between different groups, analysis
models were stratified according to age, regional econ-
omy (GDP), gender and LTCI policy pilot. ese results
are shown in Table5.
Regarding age, we used the elderly population with
higher ADL disability risk as the division reference; thus,
those aged 60 years and above were divided from others.
e results show that compared with the age group under
60 years, air quality has a significantly higher impact on
ADL disability of residents over 60 years. SO2 and PM10
have a significant impact on the ADL disability of resi-
dents over 60 years. is indicates that under the same
conditions, the probability of ADL disability in elderly
individuals brought by air quality deterioration is higher
than that of the nonelderly population. However, there
was no significant difference in the effect of the AQI on
ADL disability by age.
In terms of regional economy, we selected the regional
economic aggregate as the grouping standard; that is, the
regional GDP lower than the average GDP was the low
economic group, whereas the regional GDP higher than
the average GDP was assigned to the high economic
group. e results showed that compared with the low
economic group, air quality had a more significant and
negative effect on ADL disability in the high economic
group. is is probably because the areas with stronger
economies tend to promote better quality of life. Areas
of strong economic development also have higher pop-
ulation density and more urban automobile pollution
and industrial pollution, thus resulting in a significantly
higher impact of air quality on ADL disability. In the low-
level economic development area, the situation is the
opposite. However, there was no significant difference
in the effect of the AQI on ADL disability of different
regional economic groups.
Moreover, compared with male residents, air quality
had a more significant impact on ADL disability in female
residents. is is because the life expectancy of female
residents is generally higher than that of male residents,
and in daily life, female residents are mainly engaged
in household activities. erefore, females experience
more ADL disability related to cooking fume inhalation
at home than males. However, the impact of the AQI on
ADL disability was more significant for male residents
since in general, workers in the mining industry are
mostly men. erefore, the impact of outdoor air pollu-
tion is higher for males, which increases the probability
of ADL disability.
For the LTCI pilot group, the dummy variable of the
pilot policy was constructed according to the imple-
mentation time of the LTCI policy in 15 pilot cities
in 2016, whereby the nontreatment group and treat-
ment group were determined. e results show that
compared with the pilot areas, the air quality in the
nonpilot areas had a more significant impact on ADL
disability; that is, the LTCI pilot reduced the risk of
ADL disability caused by air quality and promoted the
prevention or rehabilitation of ADL disability among
residents.
Table 5 Heterogeneity of ADL disability among different groups of residents affected by air quality
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
Grouping Indicators lnSO2lnNO2lnPM10 lnAQI Observations
Age group Under 60 years 0.0170 (0.0452) 0.0917 (0.0671) −0.0078 (0.0653) − 0.1352 (0.0921) 15,526
Over 60 years old −0.1530*** (0.0567) −0.1069 (0.0869) − 0.2208*** (0.0801) −0.1531 (0.1140) 10,692
Regional economic
status Low GDP group −0.0074 (0.0408) 0.0112 (0.0576) −0.0771 (0.0611) −0.1275 (0.0896) 18,952
High GDP group −0.2994*** (0.0907) −0.1410 (0.1680) − 0.3922*** (0.1316) −0.2325 (0.1521) 7266
Gender Male −0.0067 (0.0494) 0.0739 (0.0739) −0.0127 (0.0712) −0.2606*** (0.0999) 12,225
Female −0.1121** (0.0503) −0.0696 (0.0753) − 0.1812** (0.0720) −0.0627 (0.1038) 13,993
Long term insurance
pilot Pilot was launched 0.3958 (6.8008) −2.5769 (44.2767) −0.3120 (5.3616) −2.2673 (38.9568) 419
No pilot was con-
ducted
−0.0597* (0.0358) −0.0048 (0.0535) − 0.1097** (0.0515) −0.1475** (0.0727) 25,799
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 11 of 16
Liu BMC Public Health (2022) 22:835
Analysis oftheinteraction betweenair quality andserious
illness
Among the individual characteristics that affect ADL
disability, serious illness was the most important factor.
Previous theoretical research on LTCI shows that the
disabled population is mainly affected by serious illnesses
such as cerebral haemorrhage and cerebral infarction.
erefore, it is of great theoretical significance to inves-
tigate the interaction between serious illnesses and air
quality. e test results of the interaction items are pre-
sented in Table6. e interaction terms of serious illness
and SO2 and the interaction of serious illness and NO2
play a significant and positive role in ADL disability, and
the serious disease rate has a significant and negative
effect on ADL disability. However, from Table3, which
shows the estimation results for the models without
interaction items, the impact of serious illness on ADL
disability was significantly positive, which is contrary to
reality and theory. e results for Model (4) in Table 6
also show that the interaction terms have a positive mod-
erating effect but are not significant.
e estimation results of the interaction terms suggest
that air quality aggravated ADL disability caused by seri-
ous illness, and the interaction terms of serious illness and
the concentrations of SO2 and NO2 were the main factors
in the positive promotion effect on ADL disability. e pri-
mary reason for this might be that the increase in air pol-
lutants increases the probability of residents suffering from
serious illness, thus aggravating the risk of ADL disability.
Extensive analysis
e effect of air quality on ADL disability has been analysed.
Furthermore, to fix the problems of self-selection bias and
missing variables in samples, we used control samples and
considered two-way fixed effects in a more robust model.
Bias processing oftheself‑selection sample
Due to the environmental migration in the process of
air pollution, the estimation results are likely biased.
To reduce the estimation bias caused by environmental
migration, in the sample processing step, a subsample
test was conducted for the participants whose residence
location and groups did not change. e results are
given in Table7. It becomes clear that SO2 had a nega-
tive impact on ADL disability at the 10% significance
level, NO2 had a negative impact on ADL disability at
the 5% significance level, and the AQI had a negative
impact on ADL disability at the 10% significance level.
erefore, the findings of previous models were robust.
Treatment ofbidirectional xed eects ofpanel data
Although the above analysis synchronously controlled
for the corresponding individual sociodemographic
characteristics and urban environmental character-
istics, missing variables might still exist and result in
estimation bias. erefore, we first used a two-way
fixed effects model to address the endogeneity problem
caused by missing variables. is was referred to by Liu
and Hu [17], who viewed classified variables as con-
tinuous variables and employed a linear two-way fixed
effects model. In this case, ADL disability was con-
sidered a continuous variable, and the test results for
this model are presented in Table 8. As a result, SO2,
NO2 and the AQI did not show a significant effect on
ADL disability. e significance levels of SO2 and the
Table 6 Estimation of the effects of the interaction between air quality and serious illness on ADL disability
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
X is the rate of serious illness
Variable (1) (2) (3) (4)
lnSO2−0.0882**(0.0367)
lnNO2−0.0538(0.0555)
lnPM10 −0.1246**(0.0530)
lnAQI −0.1548**(0.0724)
Serious illness −0.2328*(0.1317) −0.4058**(0.2038) − 0.1697(0.2315) 0.0126(0.0442)
X × lnSO2/
X × lnNO2/
X × lnPM10/
X × lnAQI
0.1043**(0.0478) 0.1266**(0.0557) 0.0496(0.0517) 0.0225(0.0185)
Individual / Year Yes Yes Yes Yes
/sigma2_u 1.7819***(0.1183) 1.7848***(0.1186) 1.7891***(0.1186) 1.7806***(0.1183)
Pseudo log likelihood −27,313.284 −27,314.254 − 27,315.495 −27,315.248
Observations 26,218 26,218 26,218 26,218
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 12 of 16
Liu BMC Public Health (2022) 22:835
AQI were decreased in the fixed effects model, but they
were still significant at 15%. PM10 had a significant and
negative effect on ADL disability at a significance level
of 1%, and the significance of PM10 was higher than the
results of the benchmark model. erefore, air quality
still had a significant impact on ADL disability in the
panel two-way fixed effects model, which means that
the result was robust.
Instrumental variables
We further adopted the instrumental variable method for
endogenous processing. An ordered probit instrumental
variable method was selected. According to previous
studies, the abundance of regional mineral resources
and the proportion of mining industry employees in
the total population could be used as instrumental
variables of air quality [17]. Therefore, we chose the
proportion of mining industry employees in the total
population as the proxy variable of regional mineral
resources and constructed the two-stage method of IV
for the ordered probit model. The results are given in
Table9.
From the results of the first-stage test in Models (1) to (3),
mineral resources have a significant and positive effect on
Table 7 Effect of air quality on ADL disability of permanent residents
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
Variable (1) (2) (3) (4)
lnSO2−0.0684*(0.0362)
lnNO2−0.0183(0.0544)
lnPM10 −0.1307**(0.0521)
lnAQI −0.1368*(0.0772)
Serious illness 0.0563**(0.0279) 0.0570**(0.0279) 0.0586**(0.0279) 0.0477(0.0291)
Depression 0.1265***(0.0256) 0.1255***(0.0256) 0.1268***(0.0256) 0.1261***(0.0267)
Self-reported health 0.0850***(0.0203) 0.0855***(0.0203) 0.0860***(0.0203) 0.0864***(0.0212)
Body disability −0.6325***(0.0495) −0.6323***(0.0495) − 0.6324***(0.0495) −0.6282***(0.0516)
Physical pain −0.0342(0.0219) −0.0349(0.0218) − 0.0343(0.0219) −0.0449**(0.0227)
Individual / Year Yes Yes Yes Yes
/sigma2_u 1.7967***(0.1216) 1.7962***(0.1216) 1.8010***(0.1218) 1.8027***(0.1296)
Pseudo log likelihood −26,292.693 −26,294.516 −26,291.386 −24,363.95
Observations 25,169 25,169 25,169 25,169
Table 8 Impact of air quality on ADL disability: Based on fixed effects
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
Variable (1) (2) (3) (4)
lnSO2−0.0130(0.0095)
lnNO2−0.0238(0.0153)
lnPM10 −0.0584***(0.0185)
lnAQI 0.0002(0.0003)
Serious illness −0.0157***(0.0054) −0.0155***(0.0054) − 0.0149***(0.0054) −0.0155***(0.0054)
Depression 0.0135**(0.0057) 0.0134**(0.0057) 0.0139**(0.0057) 0.0134**(0.0057)
Self-reported health −0.0071(0.0050) −0.0070(0.0050) − 0.0065(0.0050) −0.0072(0.0050)
Body disability −0.0651***(0.0119) −0.0652***(0.0119) − 0.0658***(0.0119) −0.0649***(0.0119)
Physical pain −0.0126***(0.0036) −0.0129***(0.0036) − 0.0128***(0.0036) −0.0128***(0.0036)
Individual / Year Yes Yes Yes Yes
F Test 5.87 5.89 6.26 5.81
R20.0127 0.0127 0.0135 0.0126
Observations 26,218 26,218 26,218 26,218
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 13 of 16
Liu BMC Public Health (2022) 22:835
air quality. e validity test of instrumental variables shows
that the F value in the first stage is significantly greater than
10, indicating that the problem of weak instrumental vari-
ables did not exist. In other words, the selection of instru-
mental variables was effective. e results of second-stage
tests in Models (1) to (3) show that air quality had a signifi-
cant and negative impact on residents’ ADL disability at the
significance level of 1%, which further demonstrates that
the results of this study are robust. e results for Model
(4) suggest that the AQI still had a significant and nega-
tive effect on ADL disability. is further proves that poor
air quality significantly aggravates ADL disability. In addi-
tion, it can be seen from Model (4) that to keep ADL dis-
ability unchanged, ADL disability caused by a 1% increase
in the AQI requires an 89.9652% reduction in serious dis-
ease to compensate for ADL damage. is means that the
reduction amount of ADL disability brought by a 1-unit
improvement in air quality equals the amount caused by a
89.9652-unit decrease in severe illness.
Discussion
e air quality index at a certain time point is a compre-
hensive indicator of pollutant accumulation, which is also
an accurate reflection of air pollution at a specific time
point. us, this study takes the annual average value of
the AQI in a region as the proxy index to reflect long-term
air pollution. At the same time, a multidimensional meas-
urement of ADL could be established by dividing the disa-
bility level into five levels: health, mild disability, moderate
disability, severe disability and serious disability [6]. Based
on this, this study empirically tests the impact of the AQI
on residents’ ADL disability. e results demonstrate that
air quality has a significant impact on residents’ ADL dis-
ability, which is mainly manifested by the health reduction
effect and increasing effect on ADL disability. Compared
with existing studies, this study breaks the mould by
exploring the impact of the AQI on residents’ ADL dis-
ability from the perspective of air pollution and enriches
the research perspective of the social cost of air pollution.
Table 9 Estimation results of the IV ordered probit model
Abbreviations: AQI Air Quality Index, SO2 Sulfur Dioxide, NO2 Nitrogen Dioxide, PM10 Inhalable Particles
Note: Standard errors are in brackets; *** p < 0.01, ** p < 0.05, * p < 0.1. The control variable results are not listed here
Variable (1) (2) (3) (4)
First‑stage
lnSO2
Second‑stage
ADL First‑stage
lnNO2
Second‑stage
ADL First‑stage
lnPM10
Second‑stage
ADL First‑stage
lnAQI Second‑stage
ADL
Mineral endow-
ment 18.0074***(0.3361) 6.4223***
(0.2497) 7.9056***
(0.2517) 1.8217***
(0.1610)
lnSO2−0.2086***
(0.0481)
lnNO2−0.5079***
(0.1249)
lnPM10 −0.4492***
(0.1048)
lnAQI −1.8083***
(0.3437)
Serious illness 0.0195* 0.0192* 0.0225** 0.0201**
(0.0106) (0.0104) (0.0105) (0.0095)
Depression 0.0744*** 0.0722*** 0.0896*** 0.0811***
(0.0106) (0.0104) (0.0106) (0.0105)
Self-reported
health 0.0366*** 0.0360*** 0.0331*** 0.0295***
(0.0089) (0.0087) (0.0088) (0.0080)
Body disability −0.3027*** −0.2968*** −0.2951*** −0.2653***
(0.0163) (0.0164) (0.0163) (0.0210)
Physical pain −0.0214** −0.0210** −0.0243*** −0.0216***
(0.0084) (0.0082) (0.0083) (0.0076)
Individual / Year YES YES YES YES
|α/β| 10.6974 26.4531 19.9644 89.9652
First stage F
value 2870.47 661.23 976.54 128.13 128.13
Adjust R20.0736 0.0180 0.0263 0.0035 0.0035
Observations 26,218 26,218 26,218 26,218 26,218 26,218 26,218 26,218
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 14 of 16
Liu BMC Public Health (2022) 22:835
In addition, this study uses the average value of the AQI
as the proxy variable of air quality, which can empirically
reflect the impact of long-term exposure to air pollution
on residents’ ADL disability.
Most existing studies have investigated the impact
of air pollution from the perspective of social risk cost.
For example, several scholars have estimated the impact
of air pollution on residents’ health outcomes [11–15].
e indicators of health include changes in individual
health level or changes in the incidence of disease and
the incidence rate of diseases in the whole region (such
as lung cancer mortality or respiratory disease mortality
per 10,000 people) [17–21], as well as the increase in the
cost of treatment due to air pollution, which indicates the
social cost of air pollution [22, 23]. erefore, the reli-
ability of the conclusions of this study is a further expan-
sion of the scope of existing findings. First, this study not
only investigated the direct health outcome of air pollu-
tion but also investigated the changing paths of residents’
health influenced by air pollution, for instance, by analys-
ing the change in the prevalence of major diseases; thus,
discussion of the changing path of residents’ ADL disabil-
ity under the influence of air pollution could be extended.
is study also empirically reveals the theoretical basis
for the social governance of residents’ ADL disability and
the optimization of long-term care insurance. Specifi-
cally, the findings of this study provide insights into envi-
ronmental governance of residents with ADL disability
[47–49]. For example, investigating the changes in the
disability rate and factors of disability risk of local resi-
dents and the effective regulations of air pollution could
be undertaken for environmental governance. In this
study, this is mainly explained by investigating the impact
of different pollutants on residents’ ADL disability. e
results of this study clarify that the concentrations of
SO2 and PM10 pollutants are the main elements affecting
residents’ ADL disability. In addition, there are significant
group differences in the impact of air quality on ADL
disability. For example, air quality has a more significant
impact on the ADL disability of elderly residents aged
60 and above, female residents, residents in regions with
low economic levels, and residents in areas without pilot
long-term insurance. Additionally, the negative effect of
air pollution is stronger on these groups. ese findings
demonstrate that the health damage effect of air pollu-
tion can be effectively reduced after the implementation
of effective social policy intervention. Furthermore, few
studies or practices have calculated the cost of disability
treatment caused by pollution or the cost borne by the
entire society. is is also one of the main innovations of
this study. Ultimately, we should not only realize the sig-
nificant impact of air pollution on ADL disability but also
consider the differences between different groups and
take the most effective measures to control air pollution
and reduce its long-term social cost, that is, the long-
term care cost of treating disabled residents.
Despite the practical significance of the findings of this
study, especially the results of the analysis, which have
been proven valid after a series of robustness tests and
endogenous treatments, this study still has some limita-
tions. First, the sample of this study mainly comes from
122 cities in China, but the sample is limited to people over
45 years old and generally excludes those under 45 years
old, which may affect the applicability and reliability of the
conclusions to a certain extent. Second, as one of the most
important purposes of this study, ADL disability caused
by air pollution and the cost of disability treatment are the
main focus of this study. However, due to the complex-
ity of factors that cause disability and the indeterminacy
of actual nursing costs, this study was not able to meas-
ure the social cost of ADL disability caused by current
air pollution completely and accurately, which would also
be an important direction for further research. By gradu-
ally fixing the above problems, we can further clarify the
marginal effect and social cost of air pollution governance
and theoretically provide important support for optimiz-
ing regional policy for disability prevention and long-term
care service security. e main advantages of this study are
that it not only explores the direction of the impact of the
AQI on residents’ ADL disability but also investigates the
specific effect of the AQI on residents’ ADL disability, in
addition to the changing trend of the long-term ADL dis-
ability rate caused by the joint influences of the AQI and
the rate of serious disease. us, the logical relationship
and mutual effects between natural environmental factors
(AQIs) and individual health characteristic factors (rate of
serious disease) are entangled.
Conclusions
is study used tracking data from the CHARLS data-
base from 2015 and 2018 to construct panel data for
investigating the impact of air quality on ADL disability
and its marginal effect. e results show that air quality
has a significant impact on ADL disability, and the main
impacts were from the concentrations of SO2 and PM10.
Second, in terms of the marginal effect, the main effects
of air quality on ADL disability appear to have a positive
effect on disability increment, and it also shows that air
quality plays a leading role in the negative effect of health.
Moreover, it was revealed that air quality has a more sig-
nificant impact on the ADL disability of residents aged
60 years and above, female residents, residents with poor
economic status and residents in areas without LTCI. e
results of the interaction between air quality and serious
illness showed that air quality worsened the impact of
serious illness on ADL disability. Finally, we confirmed the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 15 of 16
Liu BMC Public Health (2022) 22:835
robustness of our findings by controlling subsamples and
using two-way fixed effects and instrumental variables.
Our findings are also strongly relevant to policy deci-
sions. First, social and economic development should
be “environmentally friendly” and should not only con-
sider the short-term increase in GDP but also consider
the basic quality of life of local residents, especially the
health of vulnerable groups such as the elderly popula-
tion and those living in economically underdeveloped
areas. Second, controlling air pollution should be prior-
itized. e impact of air quality on the natural environ-
ment of a country or region has been remarkable, and its
impact on the health of individuals is also gradually being
recognized. e increase in ADL disability caused by the
increase in the incidence of individual serious illness influ-
enced by air quality also indicates that the social cost of
environmental pollution is increasing. ird, when inves-
tigating ADL disability in theory, we should not only pay
attention to the causes of disability from the perspective
of traditional medicine or socioeconomic environments
but also note the influences of ecological environment
changes and the negative impacts of changes in air quality.
erefore, intervention policies could be implemented to
prevent ADL disability and improve quality of life.
Abbreviations
LTCI: Long-term Care Insurance; CHARLS: China Health and Retirement
Longitudinal Study; AQI: Air Quality Index; HAQI: Health Risk-based AQI; API:
Air Pollution Index; PM2.5: Fine Particulate Matter; PM10: Inhalable Particles;
O3: Ozone; CO: Carbon Monoxide; SO2: Sulfur Dioxide; NO2: Nitrogen Dioxide;
GDP: Gross Domestic Product.
Acknowledgements
The authors are very grateful for the financial support of National Natural Sci-
ence Fund of China (71904167) and Dr. Wang Meng’s and Tiantian Hu help in
research design, revising the article and polishing the language.
Author’s contributions
H.L. drafted the manuscript and approved the version to be published,and
carry out language retouching, modification. And he made a substantial
contribution to the concept and design of the work, interpretation of data,and
revised the article.
Funding
National Natural Science Fund of China (71904167).
Availability of data and materials
The data that support the findings of this study are openly available at the
following URL/DOI: http:// charls. pku. edu. cn/
Declarations
Ethics approval and consent to participate
We declare that all methods were carried out in accordance with relevant
guidelines and regulations. All experimental protocols were approved by
Institutional Review Board at Peking University. And we confirmed that
informed consent was obtained from all subjects or, if subjects are under 18,
from a parent and/or legal guardian. The IRB approval number for the main
household survey, including anthropometrics, is IRB00001052–11015; the IRB
approval number for biomarker collection, was IRB00001052–11014.During
the fieldwork, each respondent who agreed to participate in the survey was
asked to sign two copies of the informed consent, and one copy was kept in
the CHARLS office, which was also scanned and saved in PDF format.
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Received: 17 October 2021 Accepted: 18 April 2022
References
1. Nation Bureau of Statistics.(2021) Available from: http:// www. mofcom.
gov. cn/ artic le/i/ jyjl/l/ 202102/ 20210 20303 8237. shtml. According to the
preliminary statistics of GDP in 2020.
2. Olsson A, Schubauer-Berigan M, Schüz J. Strategies of the International
Agency for Research on Cancer (IARC/WHO) to reduce the occupational
cancer burden. Med Tr Prom Ekol. 2021;61(3):140–54 https:// doi. org/ 10.
31089/ 1026- 9428- 2021- 61-3- 140- 154.
3. Xi Jinping’s report at 19th CPC National Congress(2017) Available from:
http:// www. xinhu anet. com/ engli sh/ downl oad/ Xi_ Jinpi ng’s_ report_ at_
19th_ CPC_ Natio nal_ Congr ess. pdf
4. Chen L, Xu X. Effect evaluation of the long-term care insurance (ltci)
system on the health care of the elderly: a review. J Multidiscip Healthc.
2020;13:863–75. https:// doi. org/ 10. 2147/ JMDH. S2704 54.
5. Liu H. Formal and informal care: complementary or substitutes in care for
elderly people? Empirical evidence from China. SAGE Open. 2020;11(2):1–
16. https:// doi. org/ 10. 1177/ 21582 44021 10164 13.
6. Liu H. Research on disability grading based on icf functional framework:
empirical evidence from Zhejiang province, China. Front Public Health.
2021;9:616180. https:// doi. org/ 10. 3389/ fpubh. 2021. 616180.
7. Li ZH, Lv YB, Kraus VB, Yin ZX, Shi XM. Trends in the incidence of activities
of daily living disability among chinese elderly from 2002-2014. J Geron-
tol Ser A Biol Sci Med Sci. 2019;75(11):2113–8. https:// doi. org/ 10. 1093/
gerona/ glz221.
8. Wang Q, Granados J. Household heating associated with disability in
activities of daily living among Chinese middle-aged and elderly: a longi-
tudinal study. Environ Health Prev Med. 2020;25(1):49. https:// doi. org/ 10.
1186/ s12199- 020- 00882-5.
9. Janca A, Kastrup M, Katschnig H, López-Ibor JJ, Mezzich JE, Sartorius
N. The world health organization short disability assessment schedule
(who das-s): a tool for the assessment of difficulties in selected areas of
functioning of patients with mental disorders. Soc Psychiatry Psychiatr
Epidemiol. 1996;31(6):349–54. https:// doi. org/ 10. 1007/ BF007 83424.
10. Habtamu K, Alem A, Medhin G, Fekadu A, Hanlon C. Validation of the
world health organization disability assessment schedule in people with
severe mental disorders in rural Ethiopia. Health Qual Life Outcomes.
2017;15(1):64. https:// doi. org/ 10. 1186/ s12955- 017- 0647-3.
11. Usmani R, Saeed A, Abdullahi AM, Pillai TR, Hashem I. Air pollution and
its health impacts in Malaysia: a review. Air Quality Atmosphere Health.
2020;13(4):1093–118. https:// doi. org/ 10. 1007/ s11869- 020- 00867-x.
12. Han C, Xu R, Zhang Y, Yu W, Guo Y. Air pollution control efficacy and
health impacts: a global observational study from 2000 to 2016. Environ
Pollut. 2021;287:117211. https:// doi. org/ 10. 1016/j. envpol. 2021. 117211.
13. Markandya A, Sampedro J, Smith SJ, Dingenen RV, Pizarro-Irizar C,
Arto I, et al. Health co-benefits from air pollution and mitigation costs
of the Paris agreement: a modelling study. Lancet Planet Health.
2018;2(3):e126–33. https:// doi. org/ 10. 1016/ S2542- 5196(18) 30029-9.
14. José RS, Pérez JL, Pérez L, Barras R. Effects of climate change on the
health of citizens modelling urban weather and air pollution. Energy.
2018;165:53–62. https:// doi. org/ 10. 1016/j. energy. 2018. 09. 088.
15. Xiang Z, Azam M, Islam T, Zaman K. Environment and air pollution like
gun and bullet for low income countries: war for better health and
wealth. Environ Sci Pollut Res. 2020;23(4):3641–57. https:// doi. org/ 10.
1007/ s11356- 015- 5591-3.
16. Msha B, Hcfa B, Pkkla C, Akhla D. Combined effects of increased o 3 and
reduced no 2 concentrations on short-term air pollution health risks in
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Page 16 of 16
Liu BMC Public Health (2022) 22:835
•
fast, convenient online submission
•
thorough peer review by experienced researchers in your field
•
rapid publication on acceptance
•
support for research data, including large and complex data types
•
gold Open Access which fosters wider collaboration and increased citations
maximum visibility for your research: over 100M website views per year
•
At BMC, research is always in progress.
Learn more biomedcentral.com/submissions
Ready to submit your research
Ready to submit your research
? Choose BMC and benefit from:
? Choose BMC and benefit from:
Hong Kong. Environ Pollut. 2021;270:116280. https:// doi. org/ 10. 1016/j.
envpol. 2020. 116280.
17. Guo Y, Teixeira JP, Ryti N. Ambient particulate air pollution and daily
mortality in 652 cities. N Engl J Med. 2019;381(8):705–15. https:// doi. org/
10. 1056/ NEJMo a1817 364.
18. Zhang X, Chen X, Zhang XB. The impact of exposure to air pollution on
cognitive performance. Proc Natl Acad Sci. 2018;115(37):201809474.
https:// doi. org/ 10. 1073/ pnas. 18094 74115.
19. Xue T, Zhu T, Zheng Y, et al. Author correction: declines in mental health
associated with air pollution and temperature variability in China. Nat
Commun. 2019;10(1):1–3. https:// doi. org/ 10. 1038/ s41467- 019- 11660-5.
20. Morris GP, Beck SA, Hanlon P, Robertson R. Getting strategic about the
environment and health. Public Health. 2006;120(10):889–903. https://
doi. org/ 10. 1016/j. puhe. 2006. 05. 022.
21. Kunzli N, Medina S, K aiser R, Quenel P, Horak F, Studnicka M. Assessment
of deaths attributable to air pollution: should we use risk estimates based
on time series or on cohort studies? Am J Epidemiol. 2017;11:1050–5.
https:// doi. org/ 10. 1093/ aje/ 154. 10. 974.
22. Di Q, Wang Y, Zanobetti A, Wang Y, Schwartz JD. Air pollution and mortal-
ity in the medicare population. N Engl J Med. 2017;376(26):2513–22.
https:// doi. org/ 10. 1056/ NEJMo a1702 747.
23. Song C, Wu L, Xie Y, He J, Chen X, Wang T, et al. Air pollution in China: sta-
tus and spatiotemporal variations. Environ Pollut. 2017;227(aug.):334–47.
https:// doi. org/ 10. 1016/j. envpol. 2017. 04. 075.
24. Halkos G, Argyropoulou G. Modeling energy and air pollution health
damaging: a two-stage DEA approach. Air Qual Atmosphere Health.
2021;14(8):1221–31. https:// doi. org/ 10. 1007/ s11869- 021- 01012-y.
25. Luo H, Guan Q, Lin J, Wang Q, Wang N. Air pollution characteristics and
human health risks in key cities of Northwest China. J Environ Manag.
2020;269:110791. https:// doi. org/ 10. 1016/j. jenvm an. 2020. 110791.
26. Camarinho R, Garcia PV, Rodrigues AS. Chronic exposure to volcanogenic
air pollution as cause of lung injury. Environ Pollut. 2013;181(10):24–30.
https:// doi. org/ 10. 1016/j. envpol. 2013. 05. 052.
27. Berend N. Contribution of air pollution to COPD and small airway dysfunc-
tion. Respirology. 2016;21(2):237–44. https:// doi. org/ 10. 1111/ resp. 12644.
28. Bowatte G, Lodge CJ, Knibbs LD, et al. Traffic-related air pollution expo-
sure is associated with allergic sensitization, asthma, and poor lung func-
tion in middle age. J Allergy Clin Immunol. 2017;139(1):122–9. https:// doi.
org/ 10. 1016/j. jaci. 2016. 05. 008.
29. Schultz ES, Litonjua AA, Erik M. Effects of long-term exposure to traffic-
related air pollution on lung function in children. Curr Allergy Asthma
Rep. 2017;17(6):41. https:// doi. org/ 10. 1007/ s11882- 017- 0709-y.
30. Alderete TL, Habre R, Toledo-Corral CM, et al. Longitudinal associations
between ambient air pollution with insulin sensitivity, β-cell function,
and adiposity in Los Angeles Latino children. Diabetes. 2017;66(7):1789–
96. https:// doi. org/ 10. 2337/ db16- 1416.
31. Bourdrel T, Bind MA, Béjot B, et al. Cardiovascular effects of air pollution.
Arch Cardiovasc Dis. 2017;110(11):634–42. https:// doi. org/ 10. 1016/j. acvd.
2017. 05. 003.
32. Cacciottolo M, Wang X, Driscoll I, et al. Particulate air pollutants, APOE
alleles and their contributions to cognitive impairment in older women
and to Amyloidogenesis in experimental models. Translational. Psychiatry.
2017;7(1):e1022. https:// doi. org/ 10. 1038/ tp. 2016. 280.
33. Munzel T, Gori T, Al-Kindi S, et al. Effects of gaseous and solid constituents
of air pollution on endothelial function. Eur Heart J. 2018;39(38):3543–50.
https:// doi. org/ 10. 1093/ eurhe artj/ ehy481.
34. Toledo-Corral CM, Alderete TL, Habre R, et al. Effects of air pollution expo-
sure on glucose metabolism in Los Angeles minority children. Pediatric.
Obesity. 2018;13(1):54–62. https:// doi. org/ 10. 1111/ ijpo. 12188.
35. Kim MJ. Air pollution, health, and avoidance behavior: evidence from
South Korea. Environ Resour\ Econ. 2021;79(10):63–91. https:// doi. org/ 10.
1007/ s10640- 021- 00553-1.
36. Jiao K, Xu M, Liu M. Health status and air pollution related socioeconomic
concerns in urban China. Int J Equity Health. 2018;17(1):18. https:// doi.
org/ 10. 1186/ s12939- 018- 0719-y.
37. Liao, L.P. , Du M.Z., & Chen ZF. (2021). Air pollution, health care use
and medical costs: evidence from China - ScienceDirect. Energy Econ,
95:105132. https://doi.org/https:// doi. org/ 10. 1016/j. eneco. 2021. 105132
38. Sun H, Leng Z, Zhao H, Ni S, Huang C. The impact of air pollution on
urban residents’ health expenditure: spatial evidence from Yangtze river
delta, China. Air Qual Atmosphere Health. 2020;14:343–50. https:// doi.
org/ 10. 1007/ s11869- 020- 00939-y.
39. Liu J, Yin H, Tang X, Zhu T, Yi HH. Transition in air pollution, disease burden
and health cost in China: a comparative study of long-term and short-
term exposure. Environ Pollut. 2021;277:116770. https:// doi. org/ 10. 1016/j.
envpol. 2021. 116770.
40. Grossman M. On the concept of health capital and the demand for
health. Journal of political. Economics. 1972;80(2):223–55. https:// doi. org/
10. 2307/ 18305 80.
41. Muller C. A health production function for quasi-autarkic agricultural
households in Rwanda. Eur J Dev Res. 2001;13(1):87–105. https:// doi. org/
10. 1080/ 09578 81010 84267 82.
42. Bichaka F, Paulos G. Estimating a health production function for sub-
Saharan Africa (SSA). Appl Econ. 2005;37(2):155–64. https:// doi. org/ 10.
1080/ 00036 84041 23313 13521.
43. Thornton J. Estimating a health production function for the US: some
new evidence. Appl Econ. 2002;34(1):59–62. https:// doi. org/ 10. 1080/
00036 84001 00256 50.
44. Filmer D, Pritchett L. The impact of public spending on health: does
money matter? Soc Sci Med. 1999;49(10):1309–23. https:// doi. org/ 10.
1016/ S0277- 9536(99) 00150-1.
45. Liu H, Hu T. Evaluating the long-term care insurance policy from medi-
cal expenses and health security equity perspective: evidence from
China. Arch Public Health. 2022;2022:80,3. https:// doi. org/ 10. 1186/
s13690- 021- 00761-7.
46. Tian Y, Jiang Y, Liu Q, Xu D, Zhao S, He L, et al. Temporal and spatial trends
in air quality in Beijing. Landscape Urban Plann. 2019;185:35–43. https://
doi. org/ 10. 1016/j. landu rbplan. 2019. 01. 006.
47. Liu H. Health depreciation effect and medical cost effect of air
pollution:based on multidimensional health perspective. Air Qual Atmos-
phere Health. 2022;3:1–16. https:// doi. org/ 10. 1007/ s11869- 022- 01189-w.
48. Liu H, Hu T. How does air quality affect residents’ life satisfaction?
Evidence based on multiperiod follow-up survey data of 122 cities in
China. Environ Sci Pollut Res. 2021;28:61047–60. https:// doi. org/ 10. 1007/
s11356- 021- 15022-x.
49. Liu H, Hu T, Wang M. Impact of air pollution on residents’ medical
expenses: a study based on the survey data of 122 cities in China. Front
Public Health. 2021;9:743087. https:// doi. org/ 10. 3389/ fpubh. 2021. 743087.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub-
lished maps and institutional affiliations.
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
