In general, there are few systematic studies on the relationship between air pollutants and pneumonia hospital admissions, this time-series study provides more systematic insight into the association between air pollutants and pneumonia morbidity by using DLNM model in Lanzhou from 2014 to 2019. The findings showed statistically significant associations between short-term exposure to all six air pollutants and pneumonia hospital admissions and the associations had apparent lag effects: PM2.5, PM10, SO2, NO2, and CO had adverse effects on pneumonia, but O38h had a negitive association. And the exposure-response curves for six pollutants were linear and without obvious thresholds. The effects of the six pollutants all reached the maximum value on the lag07 day. In the subgroup analysis, the association between six air pollutants and pneumonia were stronger in females, individuals aged 0–14 years and in cold season. Our results provides systematic evidence to build the relationships between air pollution and pneumonia and offers basal data for the prevention of pneumonia and the control of air pollution in Lanzhou.
Our findings showed that, particulate matter (PM) were responsible for increasing pneumonia hospital admissions: the largest effect estimates (relative risks, RRs) of pneumonia hospital admissions with a 10µg/m3 increment of PM2.5 and PM10 were found at lag07 day with 1.044(1.029,1.060) and 1.009(1.005,1.013), respectively, the effects in single-day lags were lower than in multi-day lags, which were consistent with most previous studies. A case–crossover in New York State, America [11] indicated that increase rete of pneumonia was associated with increased PM2.5 concentration during the previous week, an interquartile range increase in PM2.5 was associated with pneumonia admissions, with RR of 1.025(95% CI: 1.017, 1.032). Luo et al [19] applied ecological time-series study using generalized additive model to explored the effect of air pollutants on pneumonia hospital admissions in Taiyuan, China, and they found that a 10µg/m3 increase in PM2.5 and PM10 was associated with pneumonia admissions, with RR of 1.010(95% CI: 1.004, 1.016) and 1.008(95% CI: 1.005, 1.012), respectively. Another nationwide time series study in 184 Chinese cities [10] showed short-term elevations in PM2.5 and PM10 concentrations were associated with increased pneumonia admissions, with RR of 1.003(95% CI: 1.002, 1.005) and 1.002(95% CI: 1.001, 1.003). And study in Kandy, Sri Lanka [15] and Qingdao, China [6] also found similar results: PM had adverse effects on pneumonia, moreover the harmful effect of PM2.5 on pneumonia is slightly greater than that of PM10. We also found differences in effect estimates (RR) for different cities. PM can absorbs various harmful substances such as heavy metals and pathogenic bacteria and sneak into the lower respiratory tract and even the lungs, thus damaging human respiratory. Meanwhile, compared to PM10, PM2.5 has a smaller particle size and longer residence time in the air, therefore, it is more likely to causes harm to the human respiratory system than PM10 [15, 20, 21]. Estimates of the impact of PM on pneumonia vary across cities, which may be related to diversity of climatic and topographic conditions, as well as differences in research methods, study population, PM concentration, etc. [22].
In this study, pneumonia hospital admissions risk was 1.086(1.053,1.121), corresponding to per 10µg/m3 rise in exposure to SO2 at lag07 days. This was consistent with most previous studies [6, 19, 23]. A study in Taiyuan [19] concluded that after controlling the effects of meteorological and other factors, the estimated effect of SO2 on pneumonia hospital admissions was statistically significant, the RR95%CI = 1.083(1.045,1.122). Another study conducted by Tao et al. in Lanzhou city (2001–2005) [23] indicated that SO2 was significantly associated with male pneumonia hospital admissions, the RR95%CI = 1.096(1.004, 1.196). However, other study conducted in Nis, Serbia [14] showed that there was no significant association between SO2 concentration and pneumonia hospital admissions. It may be related to different concentration of SO2 in different cities. the mean concentration of SO2 in Nis, Serbia [14] was 8.38 µg/m3, which was lower than that in Taiyuan (69.34 µg/m3) [19], and Lanzhou (2001–2005, 79.09µg/m3) [23], the mean concentration of SO2 in this study (Lanzhou, 2014–2019) was 21.13 µg/m3. SO2 mainly comes from the combustion of coal and other fossil fuels. In recent year, the concentration of SO2 showed a downward trend in Lanzhou by deploying flue gas desulfurization at power plants, controlling SO2 industrial emissions and using clean energy in Lanzhou [24], and the average concentration of SO2 meets the requirements of the secondary standard (60 µg/m3) in Chinese Ambient Air Quality Standard. But SO2 still increased the risk of pneumonia hospital admissions in this study. This practically meant that the lower concentration of SO2 is also significant. Therefore, stricter and more effective measures must be taken to control the pollution of SO2 and prevent its adverse effects in Lanzhou city.
In line with previous studies, we found that short-term exposure to NO2 (the mean concentration: 47.36µg/m3) is positively associated with the daily rates of pneumonia hospital admissions. Furthermore, the adverse effect of NO2 on pneumonia hospital admissions was greater than PM, every 10 µg/m3 increase in NO2 corresponded to a RR (95%CI) = 1.073(1.052,1.093) increase in pneumonia hospital admissions. An ecological time series analysis in São Paulo [25] using generalized additive model (GAM) to estimate the association between air pollutants and hospital admissions for pneumonia in children, and found that NO2 (the mean concentration: 56.5µg/m3) represented a risk factor for lag 1and lag5, and 10 µg/m3 increments of this pollutant lead to increase of 7% in relative risk. Another time-series study using GAM in Taiyuan [19] to evaluate the association between NO2 (the mean concentration: 43.41µg/m3) and respiratory disease hospitalization, found that the highest pneumonia hospital admissions risk was 1.027(1.010,1.044), corresponding to per 10µg/m3 rise in exposure to NO2. Yang et al. [6] used DLNM to estimate the effect of NO2 (the mean concentration: 34.6µg/m3) on adult pneumonia hospitalization in the coastal city of Qingdao, found that every 10 µg/m3 increase in NO2 corresponded to a RR (95%CI) = 1.05(1.01, 1.10) increase in adult pneumonia hospital admissions. Another case-crossover study in Qingdao [26] also found the pneumonia admissions risk was 1.067(1.010, 1.127), corresponding to per 10µg/m3 rise in NO2 (the mean concentration: 35.7µg/m3). The results of above studies are all similar to the present study, that is, NO2 is a risk factor for pneumonia, but the estimated effects (RR) of the different studies are different. The differences may be due to different study designs, different study areas, various climate, study periods, target population, NO2 concentration, etc., and it may also be related to different NO2 emission sources in different regions [22]. However a study conducted in Nis, Serbia [14] showed that NO2 had no significant effect on pneumonia hospital admissions, the average concentration of NO2 in Nis, Serbia was 29.24µg/m3, which is lower than all syudies mentioned above and this study.
At present, studies findings for the effects of O3 on pneumonia are inconsistent. This study found that 10µg/m3 increase in O38h was associated with decreased of pneumonia hospital admissions, the RR95%CI = 0.949(0.935, 0.962), and the association was statistically significant in the cold season, but not in the warm season. this is consistent with studies São Paulo, Brazil [25] and Shenyang [27], Qingdao [6], China. However, an ecological study on the city-specific association between short-term O3 exposure and hospital admissions for pneumonia in Hong Kong and Taipei [5], a nationwide time-series study on the impact of O3 on pneumonia hospitalizations in China [7] and a meta-analysis study on short-term association between air pollution and pneumonia hospitalization in children [28] all showed positive association between O3 and pneumonia hospital admissions, meanwhile, the nationwide study in China [7] found the association of O3 with pneumonia hospital admissions during warm season were stronger than in cold season. And there are also some studies [6, 19, 21, 29] showing no association between O3 and pneumonia hospitalizations. In light of the conflicting findings, more work are needed to ascertain the relationship between O3 and pneumonia hospitalizations.
The toxicity of CO deriving from its stronger ability to bind emoglobin than oxygen, and CO can cause damage to health at high-exposure levels or lower-exposure concentrations [8]. This time series study also found strongest effects of CO on pneumonia hospital admissions in six air pollutants, every 1 mg/m3 increase in CO corresponded to a RR (95%CI) = 1.157(1.094, 1.223) increase in pneumonia hospital admissions. The findings are similar to previous studies in North Carolina, United States [30], Bangkok, Thailand [31], Ankara, Turkey [32], and other Chinese cities. For instance, a study in Qingdao [6] suggested a statistically significant positive association between CO and pneumonia hospital admissions, and RR(95% CI) of pneumonia hospitalizations for CO at the 95th percentile compared to the 25th percentile was 1.08(1.03,1.14). In Shenyang study conducted by Chang et al. [27], CO showed significant positive association with pneumonia hospital admissions (RR = 1.025%, 95% CI: 1.020, 1.029). A case-crossover study in Shijiazhuang [33] showed significant effects of increased CO on hospitalization for pneumonia, and pneumonia hospital admissions risk was 1.087 (95%CI:1.030–1.148), corresponding to per 1mg/m3 rise in exposure to CO at lag3 days. The mean concentration of CO was 1.24 mg/m3 in this study, 0.8 mg/m3 in Qingdao [6], 1.02 mg/m3 in Shenyang [27], and 2.04 mg/m3 in Shijiazhuang [33], the CO concentrations in these cities are all meet the requirements of the primary standard (4 mg/m3) in Chinese Ambient Air Quality Standard. Like SO2, CO also had adverse effects on pneumonia at low concentrations.
Rarely studies from China have explored the exposure-response curves between air pollutants and pneumonia hospital admissions. In this time series study, the exposure-response curves showed pneumonia hospital admissions were positively correlated with PM2.5, PM10, SO2, NO2, and CO, but negatively correlated with O38h. All six exposure-response curves were approximate linear with no thresholds. A study in Dongguan [8] examined the association between ambient CO and hospital outpatient visits for pneumonia dieases, and suggested that a general linear relationship between CO concentration and outpatient visits risk for pneumonia. A national time series analysis in 184 cities in China [10] found a slightly nonlinear exposure-response curves between PM and adult pneumonia admissions (aged ≥ 18 years), the curve of the association between PM2.5 and adult pneumonia admissions appeared to be a plateau at higher concentrations, and the curve of PM10 increased sharply at concentrations below 100 µg/m3 and then climbed relatively moderately as concentration increased. Another study in Qingdao [6] showed exposure-response curves of six air pollutant and adult pneumonia hospital visits (aged ≥ 14 years) exhibit nonlinear characteristics, and the slopes are steep at low concentrations and level off at high concentrations. The differences in the exposure-response curves are likely associated with different population (inclusion-exclusion criteria), air pollutant characteristics, socioeconomic factors in different regions and periods [6], which worth further study in future.
Existent results on gender difference in air pollution epidemolgy were not uniform, the subgroup analysis in this study suggested that female pneumonia patients were more sensitive to the effects of air pollutants. Some previoue epidemological studies also supported this result [21, 23, 27, 29] which could be related to genetic and biological differences, females have smaller respiratory tract diameters and female may have weaker lung immunity and be more vulnerable to air pollution. However, some inconsistent results have been reported, some studies confirmed that male for pneumonia were more vulnerable to air pollution than female [14, 15, 33], which could be related to male’s occupation such as industry or taxi driving or more outdoor activities, poorer awareness of occupational protection, and smoking, drinking, and other unhealthy habits, therefore there is higher susceptibility to pneumonia after exposure to air pollution. Even some other studies showed that air pollution effects on pneumonia hospital admissions were similar for males and females [3, 7, 10]. Gender differences in associations of air pollutants with pneumonia hospitalization remains to be further explored.
Age-stratified analyses in this study suggested that 0–14 years age group (children) was more sensitive to all six air pollutants than other two age groups, which is consitent with other similar studyies conducted in China [3, 26, 34], as well as in Euroupean country [14]. This result maybe related to children spending more time at outdoor activities which make them exposed to more atmospheric pollution, and their high breathing rate, narrow airways, and developing lungs and immune systems [34], which finally leads to their high sensitivity to the air pollutants. However, there are also different findings showing that the elderly (≥ 65 years old) are more sensitive to air pollutants due to poor immune function and comorbidity [5, 27, 33]. Most studies had inconsistent results in terms of age-stratified analyses, the possible reason for this might be due to the different demographics in different regions, the inclusion-exclusion criteria, the different composition of air pollutants, diversity of climatic and topographic conditions, and the number of cases in various studies. And age differences in associations of air pollutants with pneumonia hospitalization remains to be further explored too.
The season-specific analysis in this study indicated that the relationship between pneumonia hospital admissions and PM2.5, PM10, SO2, NO2, and CO were statistically significant in the cold season, but not in the warm season, that is to say, the associations of air pollutants with pneumonia hospital admissions during the cold season were stronger than in the warm season. The findings are consitent with most studies on the relationship between air pollutants and pneumonia hospitalization, such as studies in Shenzhen (PM2.5) [3], Qingdao (five pollutants except O3) [6, 26], Shijiazhuang (PM2.5 and PM10) [33], Taiwan (PM2.5 and NO2) [21], China, and in Turkey [32]. However a study in Jinan [34] showed that children was at statistically significant risk of hospital admission for pneumonia due to PM, particularly on warm days. Lanzhou is located in northwest China with stable stratification especially inversion and coal using (from November to March of the following year) in cold season, which blocks the air streams and makes the pollutants difficult to disperse [17], those increase the level of air pollution in cold season. And this may explain why we found the greater harmful effects of PM2.5, PM10, SO2, NO2, and CO on pneumonia hospital admissions in cold season.
The present study has a few advantages compared with previous studies. For example, this study systematically analyzed whether the relationship between all six air pollutants and pneumonia hospital admissions were affected by gender, age or season, and the exposure-response curves of air pollutants with pneumonia hospital admissions were also analyzed in this study. Those may provide more convincing evidence in the association between air pollutants and pneumonia. In addition, compared with the traditional model, we used DLNM to explore the relationships among exposure, lag effect, and pneumonia hospital admissions comprehensively, and this was the first study to analyze the relationship between air pollutants and pneumonia hospital admissions in Lanzhou, China.
Some limitations should nevertheless be recognized. Firstly, this time-series study is an ecological study, so there may be ecological fallacy for the following reasons: the exposure levels to air pollutants were only averaged levels and not the true exposure levels of individual; the individual-level confounders such as smoking, drinking, occupational history, indoor stay time, and daily ventilation time, etc. cannot be controlled. So this study is helping us to generate the etiological hypothesis, but caution should be used in causal inference. Secondly, these data only come from Lanzhou city, and the potential spatial heterogeneity of the health effects of air pollutants in multiple cities is worthy for further study.