-
Short-term exposure to ambient ozone (O3), a weather-driven photochemical pollutant, has been found to be associated with increased risk of mortality in previous epidemiological studies (1-2). Most of these studies analyzed O3-mortality associations by controlling for meteorological factors in a model fitting process. Regarding the high correlation between O3 and temperature, a recent area of interest is whether the observed O3-mortality associations can be modified by temperature. Jhun et al. found both high and low temperature could strengthen acute effects of O3 on mortality (3), while Chen et al. and Shi et al. reported that O3-mortality associations were strengthened in high temperature but not in low temperature settings (4-5), and Liu et al. and Chen et al. only found modifications by low temperature (6-7). In summary, evidence on temperature-modification was inconsistent and needed to be supplemented by regional epidemiological studies involving various meteorological characteristics. Note that in the new air quality guidelines issued by the World Health Organization (WHO), O3 limits have been distinguished between warm and cold seasons. Beijing Municipality and Tianjin Municipality, along with 26 cities distributed in Hebei, Shandong, Shanxi and Henan Provinces, have formed the regional air pollution transmission channel, and experienced a challenge of regional O3 pollution increasing steadily. Therefore, additional efforts are needed to better quantify the local health risks of O3 by considering the influence of temperature in the Beijing, Tianjin, Hebei and surrounding areas.
This study used daily counts of deaths from the Disease Surveillance Point System of China CDC and included 39 counties in the Beijing, Tianjin, Hebei and surrounding areas from January 1, 2013 to December 31, 2018. Three major causes of deaths were classified according to the 10th Revision of the International Statistical Classification of Diseases (ICD-10): non-accidental disease (A00–R99), cardio-cerebrovascular disease (I00–I99), and respiratory disease (J00–J99). Daily ambient O3 concentrations were collected from the National Urban Air Quality Real-Time Release Platform and calculated to a daily 8-hour moving average maximum (O3 8 h-average), 1-hour maximum (O3 1 h-max), and 24 hour-average of O3 (O3 24 h-average). Daily average temperature and relative humidity were obtained from the China Meteorological Data Network.
The study applied two time-series approaches with a two-stage statistical analysis to estimate whether and how temperature modified acute effects of O3 on mortality in the Beijing, Tianjin, Hebei and surrounding areas. The first approach, a temperature-adjusted approach, aimed to control the cumulative temperature impacts with a cross-basis function using a generalized linear model (GLM) and analyze associations between O3 and death without considering interactions.
The second approach, a temperature-stratified approach by a Pick-A-Point technique centering on changes of the conditional effect of O3 across the designated levels of the modifier (8), aimed to construct interaction terms between O3 and a stratification variable of temperature in the GLM and analyze differences of associations under three different temperature levels: low, moderate, and high temperature. In this model, we used three cutoffs to categorize daily average temperature, including the 10th and 90th (P10/P90), 20th and 80th (P20/P80), and 25th and 75th (P25/P75) percentiles. The model of the temperature-stratified approach was set up as follows:
$$ \begin{aligned} Log\left[E\left({Y}_{t}\right)\right]= &intercept+\beta {O}_{3}+{\beta }_{1}Tem+{\beta }_{2}({O}_{3}:Temstrata)\\ & +ns\left(RH,df\;\right)+dow+ns(time,df\;) \end{aligned}$$ Where was the expected value of death on day t;
$ Tem $ represented the daily value of temperature;$ ({O}_{3}:Temstrata) $ was the interaction term between O3 and temperature, in which temperature was divided into low, moderate, and high levels of the categorical variable by cutoffs. Both approaches estimated effects of the 2-day average of current and previous-day concentrations (lag 01) of O3 8 h-average and controlled for seasonal and time trends [$ time $ , natural smoothing function of 8 degrees of freedom (df )], day of the week$ \left(dow\right) $ , and relative humidity ($ RH $ , natural smoothing function of 5 df ). The effect estimate was expressed as a percent increase (PI) in mortality risk per 10 μg/m3 increase in O3 exposure.This study examined the sensitivity of key findings for non-accidental mortality with respect to using the following: 1) the specification of df in the smoothing functions of time trend (df=6 or 7/year) and relative humidity (df=3) in the temperature-adjusted approach to observe model stability; 2) the other two metrics (O3 1 h-max and O3 24 h-average) with different lagged exposure [the same day as deaths (lag 0), the previous day (lag 1), and lag 01] in the temperature-adjusted approach to observe impacts from different exposure assessments for the study population; and 3) O3 1 h-max and O3 24 h-average with lag 01 exposure in the temperature-stratified approach to observe whether the modification effect of temperature on different ozone metrics was robust. Statistical analyses were conducted in the R Statistical Software (version 4.0.2, the Free Software Foundation’s GNU Public License, Vienna, Austria). Statistical significance was considered at a P-value <0.05.
From 2013 to 2018, residents in the Beijing, Tianjin, Hebei and surrounding areas were exposed to a concentration of O3 8 h-average of (95.2±61.4) μg/m3. Approximately 11 deaths for non-accidental disease, 6 for cardiocerebrovascular disease, and 1 for respiratory disease per day per county were recorded (Table 1).
Variable Mean±SD P25 P50 P75 O3 24 h-average (μg/m3) 56.3±38.7 25.3 49.5 80.2 O3 8 h-average (μg/m3) 95.2±61.4 48.2 83.8 135.4 O3 1 h-max (μg/m3) 111.0±71.5 59.7 95.0 155.0 Temperature (°C) 13.3±11.1 2.8 14.6 23.4 Humidity (%) 0.6±0.2 0.4 0.6 0.7 Non-accidental diseases 11±8 6 9 14 Cardio-cerebrovascular diseases 6±4 3 5 7 Respiratory diseases 1±1 0 1 2 Abbreviations: SD=standard deviation; P25=25th percentile; P50=50th percentile; P75=75th percentile. Table 1. Summary for ambient O3, meteorological factors, and causes of death in the Beijing, Tianjin, Hebei and surrounding areas, 2013 to 2018.
Based on the temperature-adjusted approach without considering interactions, a per 10 μg/m3 increase in exposure to O3 8 h-average would increase daily mortality risks of non-accidental [PI=0.15%, 95% Confidence Interval (CI): 0.06%, 0.24%], cardio-cerebrovascular (PI=0.20%, 95% CI: 0.07%, 0.33%), and respiratory diseases (PI=−0.08%, 95% CI: −0.42%, 0.25%) in the Beijing, Tianjin, Hebei and surrounding areas. Based on temperature-stratified approach, relatively higher temperature (>75th 24 h-average temperature) significantly strengthened O3-mortality associations, with a 0.57% risk increase of non-accidental disease, 0.64% risk increase of cardio-cerebrovascular disease, and 1.17% risk increase of respiratory disease (Figure 1). Under extreme temperature (>90th 24 h-average temperature) modification, the associations between O3 and human mortality has further increased: a 0.98% risk increase of non-accidental disease, 1.19% risk increase of cardio-cerebrovascular disease, and 1.76% risk increase of respiratory disease (Figure 1). Moreover, extreme low temperature (<10th 24 h-average temperature) was also found to strengthen the acute effects of O3 on mortality (Figure 1).
Figure 1.Mortality risk estimates associated with Lag 01 exposure of O3 by using the temperature-adjusted and temperature-stratified approach in the Beijing, Tianjin, Hebei and surrounding areas, 2013–2018.
Abbreviations: P10=10th percentile; P90=90th percentile; P20=20th percentile; P80=80th percentile; P25=25th percentile; P75=75th percentile.Associations between short-term exposure to O3 8 h-average and mortality peaked at Lag 01 exposure. Analyses with different metrics of O3 exposure and more or less stringent time trends and relative humidity controlled by varying df’s did not meaningfully change our findings (Table 2). For the three cutoffs of P10/P90, P20/P80, and P25/P75, the associations between the other two O3 metrics and morality were both increased under high temperature levels (Table 3). The extreme low temperature was found to only significantly modify the association between O3 24 h-average and mortality.
Pollutants Lag dftem/rh=3, dftime=6 dftem/rh=5, dftime=8 dftem/rh=5, dftime=7 dftem/rh=3 dftem/rh=3, dftime=7 O3 8 h-average Lag 0 0.11
(0.04, 0.19)0.14
(0.06, 0.22)0.14
(0.07, 0.22)0.13
(0.05, 0.21)0.13
(0.05, 0.20)Lag 1 0.03
(−0.05, 0.10)0.07
(−0.01, 0.14)0.06
(−0.01, 0.14)0.05
(−0.02, 0.13)0.04
(−0.04, 0.11)Lag 01 0.09
(−0.00, 0.18)0.15
(0.06, 0.24)0.14
(0.04, 0.23)0.13
(0.04, 0.22)0.10
(0.01, 0.20)O3 1 h-max Lag 0 0.07
(0.01, 0.13)0.08
(0.03, 0.14)0.08
(0.03, 0.14)0.08
(0.02, 0.14)0.08
(0.02, 0.13)Lag 1 0.06
(−0.00, 0.11)0.08
(0.02, 0.13)0.07
(0.01, 0.13)0.07
(0.01, 0.12)0.06
(0.00, 0.12)Lag 01 0.10
(0.02, 0.17)0.13
(0.06, 0.21)0.12
(0.05, 0.20)0.12
(0.04, 0.19)0.10
(0.03, 0.18)O3 24 h-average Lag 0 0.13
(0.01, 0.24)0.14
(0.02, 0.25)0.17
(0.06, 0.28)0.13
(0.02, 0.24)0.15
(0.04, 0.26)Lag 1 0.12
(0.01, 0.23)0.16
(0.05, 0.27)0.17
(0.06, 0.28)0.14
(0.03, 0.25)0.14
(0.03, 0.25)Lag 01 0.17
(0.03, 0.30)0.21
(0.07, 0.35)0.24
(0.10, 0.38)0.19
(0.05, 0.32)0.20
(0.06, 0.34)Note: dftem/rh is the degree of freedom of natural smoothing function for temperature or relative humidity, dftime is the degree of freedom of natural smoothing function for the seasonal and time trends.
Abbreviation: CI=confidence interval.Table 2. Percent increase (95% CI) in mortality risks associated with short-term exposure to O3 for sensitivity analysis by using the temperature-adjusted approach in the Beijing, Tianjin, Hebei and surrounding areas, 2013–2018. (%)
Pollutants Temperature P10/P90 P20 /P80 P25 /P75 O3 1 h-max Low 0.30 (−0.01, 0.61) 0.23 (−0.09, 0.54) 0.24 (−0.12, 0.6) Moderate 0.14 (0.02, 0.25) 0.13 (−0.01, 0.27) 0.13 (−0.03, 0.28) High 0.81 (0.66, 0.97) 0.55 (0.43, 0.68) 0.49 (0.37, 0.61) O3 24 h-average Low 0.61 (0.11, 1.12) 0.31 (−0.24, 0.86) 0.32 (−0.3, 0.94) Moderate 0.14 (−0.11, 0.4) 0.08 (−0.22, 0.39) 0.06 (−0.28, 0.41) High 1.47 (1.22, 1.73) 1.02 (0.81, 1.24) 0.90 (0.68, 1.12) Abbreviations: CI=confidence interval; P10=10th percentile; P90=90th percentile; P20=20th percentile; P80=80th percentile; P25=25th percentile; P75=75th percentile. Table 3. Percent increase (95% CI) in mortality risks associated with short-term exposure to O3 for sensitivity analysis by using the temperature-stratified approach in the Beijing, Tianjin, Hebei and surrounding areas, 2013–2018. (%)
HTML
Citation: |