We found a short-term positive effect of PM2.5 on mortality in this multi-city time-series design for seven major Korean cities from 2006 to 2019. Considering continuous exposure to prolonged high concentrations that exceeded the daily mean of 35 µg/m3, the effects of PM2.5 on daily all-cause, respiratory, and cardiovascular mortality were higher on the first and fourth consecutive days with high concentrations. In addition, the effect was mainly shown in the elderly population aged 65 years or more.
There are more reports of short-term effects of PM2.5 on mortality compared to long-term effects in Korea; however, the effect sizes vary greatly depending on the spatiotemporal background of each study, the applied statistical model and exposure assessment . Moreover, to date, only two multi-city designs have been published for the effect of particulate matter (PM10) on mortality [3, 16]; to the best of our knowledge, no studies have reported the pooled effect of PM2.5 through multi-city design in Korea.
Two studies reported the short-term effect of PM2.5 on daily mortality for each city. Jung et al.  reported that the total daily mortality risk in individuals aged ≥60 years old increased by 0.36% per increase of 10 µg/m3 from 2000 to 2012 in Seoul, Republic of Korea, using Poisson GAM applied lag 0 exposure; however, no estimates were presented for the entire population. Kim et al.  found that the risk of total daily mortality increased with an increase of 10 µg/m3 with 0.34%, 1.18%, and 0.43% in Seoul, Busan, and Incheon, respectively, from 2006 to 2012 using quasi-Poisson GAM applied lag 0-1 exposure. We also found that total daily mortality increased by 0.44% (95%CI, 0.18–0.69%, per an increase of 10 µg/m3) in the ≥65 years old group in Seoul, which is somewhat consistent with the result of Jung et al.  (See Table S6, Additional File 1). However, we found that the risk increases in Seoul, Busan, and Incheon were 0.24%, 0.25%, and 0.13%, respectively, in the lag 0-1 models (Table S3), which was not consistent with the results of Kim et al. . This previous study used actual monitoring of PM2.5; however, we used a simulated dataset. Furthermore, there is a difference in the statistical models that we did not consider barometric pressure to covariate and in the study period that we expanded to 2019. Finally, the number of reports of PM2.5 problems in the mass media in Korea has increased rapidly since 2012 ; thus, we probably assumed that this was reflected in the increase in the rate of wearing health masks due to increased risk awareness .
Two studies have focused on the short-term effects of prolonged continuous exposure to high PM2.5. The first study presented the excess risk of cardiovascular and respiratory mortality for continuous exposure to high concentrations in Beijing, China, from 2010 to 2012. Using GAM, the PM2.5 concentration variable was not added, and instead, the categorical variable specified how long the high concentration lasted based on the daily mean 75, 85, 105, 115 µg/m3 criteria applied to the models. Focusing on the susceptible group rather than the entire population, the main result was that on the ninth day of the consecutive high concentration duration of 105 µg/m3 or more, a 53% increase in the risk of cardiovascular mortality was reported in outdoor workers .
The second study reported the short-term durational effect of prolonged exposure to high concentrations of PM10 in 28 cities in China, Japan, and Korea. Using quasi-Poisson GAM, the effect of the change in PM10 concentration and duration (the number of consecutive days with 75 µg/m3 or more) was separated. In Korea, the estimated increases in risk for each additional consecutive day with high concentration were 0.48% for non-traumatic all-cause mortality, 0.48% for cardiovascular mortality, and 1.13% for respiratory mortality, and minimal effect modification by two age groups based on 65 .
Both studies suggested their own health effects, and the magnitude of the effects could be increased when high concentrations were continuously prolonged. However, the health effect of consecutive high concentration duration was also associated with exposure to PM2.5. Therefore, we thought it was necessary to investigate whether the short-term mortality effect of PM2.5 was modified to a greater size when the high concentration persisted and apply the effect modification term between PM2.5 concentration and the high concentration consecutive day variable in our GAMs. To the best of our knowledge, this study is the first to report the effect modification of the short-term effect of PM2.5 when a high concentration persists.
A new finding of this study was that the short-term effects of PM2.5 on daily non-traumatic all-cause, respiratory, and cardiovascular mortality on the first and fourth consecutive days with high PM2.5 concentrations were greater than the estimated effect for the entire period. From a short-term perspective, it was interesting to note that the risk did not increase linearly as the continuous exposure to high concentrations increased but increased from the first transition to the day with high concentration and at some point, i.e., the fourth day of consecutive duration.
As a result of age stratification, in the 20−64 years old group, we could not observe noticeable significant effect change due to the relatively low number of deaths in the group. If the high concentration lasts approximately 4 days, the respiratory effect may increase, future research with greater statistical power will be required., In the ≥65 years old group, the effect on the first consecutive day with high concentration was remarkable. The finding suggests that the health effect is greater when the level of exposure to high concentrations of PM2.5 increases dramatically in a short period of time, and it can be thought that the population more susceptible to PM2.5 was greatly affected on the first day of consecutive duration.
We found that a high concentration of PM2.5, lasting ≥7 days, was still observed in 2019. However, there is still insufficient understanding of intermittent high concentrations of PM2.5 episodes in Korea, and there is also a spatial difference in the contribution sources and components of PM2.5 . In general, when the influence of China is large, sulfate increases, and nitrate and organic particles are known to be generated locally. Recently, in most high-concentration consecutive duration episodes without yellow dust storms in the western, central, and southeastern parts of the Korean peninsula, the increase in sulfate, nitrate, and ammonium ion concentrations, rather than organic carbon, inorganic carbon, and heavy metals, was more pronounced than that of the PM2.5 mass concentration [2, 21].
A study suggested null findings on the short-term mortality effects of sulfate, nitrate, and ammonia ions in Korea . However, in other studies, the effects of sulfate, nitrate, and ammonia ions on cardiovascular mortality from 2008 to 2009 and emergency hospital visits for cardiovascular disease from 2010 to 2013 in Seoul were greater than those of PM2.5 mass concentration [23, 24]. In a multi-city study of six cities in Korea from 2013 to 2015, the pooled effect size of nitrate and sulfate was not greater than that of the mass concentration of PM2.5 . With the findings reported in Korea, the evidence for these three water-soluble ions is still insufficient and inconsistent.
As a result of a meta-analysis of studies published before August 2018, the short-term risk increases of nitrates and sulfate for cardiovascular mortality were 0.58% and 0.33% (per interquartile range(IQR) increase), respectively, with statistical significance even after adjusting for the mass of PM2.5. For hospitalization for respiratory disease, the short-term risk increase of nitrates was 0.68% (per IQR increase) with statistical significance . We carefully speculate whether the effect of PM2.5 on daily mortality increases with the high concentration consecutive duration was probably due to a variation in PM composition, especially the increase in nitrates and sulfates.
The exposure-response curve for the effect of cardiovascular mortality according to the duration and intensity of exposure to PM2.5 has been reported to have a downward curve shape . Pope et al.  examined the curve of the high concentration range by extrapolating from the results of smoking or specific circumstances of high PM2.5 exposure. Meanwhile, a recently reported study from 2009 to 2012 in Beijing, China (annual average PM2.5 of 84.9 µg/m3) showed that the relative risks of total and cardiovascular mortality increased as the background concentration increased (50 µg/m3 or more). Conversely, in the case of respiratory mortality, the relative risk decreased as the background concentration increased . In this study, we estimated the relative risks for all-cause mortality across the 10 µg/m3 intervals of background PM2.5 concentration (See Figure S3, Additional File 1). We found higher effect estimates on the first and fourth consecutive days than those derived in the high PM2.5 concentration intervals of more than 40 µg/m3 (See Figure S3, Additional File 1); therefore, it could be possible to see that the effect modification we observed was not simply an effect of the background effect of high concentration.
This study had several limitations. There may be misclassifications in the exposure assessment. In other words, the local variation in PM2.5 concentration in the same metropolitan area cannot be reflected. In addition, the specific cause of the regional and temporal variation in the effect of PM2.5 cannot be explained in our study. Demographic characteristics such as age, sex, population density, traffic volume, and access to medical institutions are possible effect modifiers [29, 30]. Finally, the short-term effect of PM2.5 on daily mortality increases and the variation in effect size within a short period of time with a high concentration consecutive duration cannot be explained by significant scientific evidence. Future studies are needed to investigate the effect in susceptible groups in detail and apply the components of PM2.5 to the models with reflected effect modification by a high concentration consecutive duration, as we proposed in our study.
Despite several shortcomings, we also have several strengths. Focusing on the change in the effect size of PM2.5, when the high concentration persists, which is problematic in East Asia, we found that the mortality effect of PM2.5 may increase short-term even more during the period of high concentration. Moreover, among the time series for short-term mortality effect reported in Korea, this study has the longest period of 14 years, and is the first multi-city study for the short-term effect of PM2.5 on mortality, including more than half of the entire Korean population.