Study design
To estimate the effects of AD on respiratory function in school children and adults with asthma, panel studies were conducted, in which the daily peak expiratory flow (PEF) values were measured, from March to May 2013. One hundred and thirty-seven adults with asthma, aged >18 years, and 384 school children aged 9 and 10 years were enrolled. The adults resided in Yonago, Matsue, Sakaiminato, Yasugi, or Saihaku, all of which are located within 25 km of Tottori University Hospital in Yonago, Tottori, western Japan as shown in Fig. 1 [23]. Based on the criteria of the Global Initiative for Asthma (GINA) [25], these participants were diagnosed as having asthma if they presented with a history of intermittent wheezing and exhibited airway hyperresponsiveness to methacholine, or exhibited reversible airflow limitations (12% and 200–mL variability in the forced expiratory volume in 1 s [FEV1]). The study was approved by the institutional ethics committee (Ethics Committee of Tottori University, approval no. 1656), and all the patients provided written informed consent.
School children from four of a total of 35 elementary schools in Matsue, Shimane, were enrolled, in 2013 (Figure 1) [24]. The four elementary schools were within 10 km of each other and all the participants lived within a radius of 1 km from the schools. Data on the participants’ age, gender, height, and weight, as well as the presence of asthma, allergic rhinitis, allergic conjunctivitis, atopic dermatitis, and food allergies, were recorded in March 2013. The study was approved by the institutional ethics committee (Ethics Committee of Tottori University, Approval Number 1764). The children and their parents were informed of the study by teachers, and the school children and their parents provided written informed consent for participation.
Recording daily PEF values
The PEF values were measured using a peak flow meter (Mini-Wright, Harlow, England, American Thoracic Society scale). Every day, from February to May 2013, all the participants recorded their morning PEF values. February was assigned as the trial period. PEF values were measured three times in the morning, and participants recorded the best of the three obtained values. Participants with asthma measured their PEF values before taking their medication.
Measurement of air pollutants and meteorological variables
Data on the concentrations of nitrogen dioxide (NO2), ozone, particulate matter smaller than 2.5 μm (PM2.5), sulfur dioxide (SO2), and suspended particulate matter (SPM) are monitored at many locations in Japan, by the Japanese Ministry of the Environment. For the current analysis, data on the concentrations of SPM, SO2, and NO2 were obtained from the Yonago and Matsue observatories. The Matsue observatory, under the Japanese Ministry of the Environment, has a LIDAR system and so, LIDAR data were obtained from there. Values that were measured at 120–150 m above the ground, which is the minimum altitude required by LIDAR systems to measure AD particles, were used. Data on daily temperature, humidity, and atmospheric pressure in Yonago and Matsue were obtained from the Japan Meteorological Agency (JMA).
Definition of heavy–AD
At present, in Japan, there is no standardized definition for heavy–AD. The JMA, with reference to data obtained from meteorological satellites (http://www.jma.go.jp/jp/kosa/), defines a heavy–AD day as a day on which visibility is reduced to <10 km, due to sand dust arising from the deserts of East Asia. LIDAR systems measure the levels of AD particles at 15-min intervals [18, 19]. Usually, the daily levels of the AD particles were on the median value of 96 measurements collected over a 24 h period, but they were omitted when the number of available measurements fell below 50% of the total number of measurements. Ueda et al. reported that a cut-off level of 0.07 km−1 in the median value within 24 h was associated with an increase in the number of emergency ambulance dispatches [26]. The hourly value is presented on the Japanese Ministry of the Environment’s website (http://soramame.taiki.go.jp/dss/kosa/). An hourly value of 0.1 km−1, which corresponds to a visibility range of 10 km [27] and 0.1 mg/m3 of sand dust particles [19], is the lowest detectable value in real time. In the present study, in order to examine whether differences in the definition of heavy-AD lead to different health outcomes, four cut-off levels of heavy-AD days were defined: (1) the JMA’s aforementioned definition (2) 24–h median AD particle levels greater than 0.07 km−1, (3) hourly AD particle levels greater than 0.1 km−1, and (4) hourly AD particle levels greater than 0.07 km−1. As per the JMA’s definition, there were five heavy–AD days– March 8–10, and March 19 and 20, between March and May 2013, in both Matsue and Yonago. There were 5 days during the study period– March 7, March 9, March 20, and May 30 and 31– in which the 24-h median AD particle levels were more than 0.07 km−1. There were 8 days– March 7–10, March 17, March 20, April 6, and April 30, which met the criteria (hourly AD particle levels greater than 0.1 km−1). There were 20 heavy–AD days– March 7–10, March 16-20, April 5, 6, 9, 10, 16, 29, and 30, and May 30 and 31– on which the hourly AD particle levels were greater than 0.07 km−1.
Statistical analysis
For the analyses of longitudinal repeated measurement data, linear mixed models were used to estimate the effects of heavy–AD on the daily PEF values of the adults with asthma as well as the school children [28, 29]. Similarly, the relationship between PEF values and the daily median levels of AD particles, according to LIDAR data, were estimated using the mixed effects models. In the analyses of the adults with asthma, the individual characteristics were included as adjustment factors in the regression models; age, sex, smoking status, the presence of allergic rhinitis and atopic disposition, treatment step, and respiratory function. In addition, the meteorological data (temperature, humidity, atmospheric pressure) of Yonago were included as adjustment factors. Moreover, in the analyses of the data on school children, the potential confounding factors were also adjusted in the regression models: age, sex, height, weight, the presence of asthma, allergic rhinitis, allergic conjunctivitis, atopic dermatitis, food allergies, and the meteorological variables measured in Matsue. After that, we involved random intercepts, which adequately assess intra-individual correlations among the repeated measurements.
The effects of heavy–AD days on PEF values were compared to those of non–heavy AD days. To evaluate potential persistent effects of heavy–AD on respiratory function, the lagged cumulative effects were analyzed for lag day 0 (the day of the heavy–AD event), lag day 0 to day 1 (defined as lag 0–1), lag day 0 to day 2 (lag 0–2), and lag day 0 to day 3 (lag 0–3). Estimates were provided as the regression coefficient of PEF values per interquartile range (IQR) change in AD particle levels, with 95% confidence intervals (CIs). The multiple imputation method was used to treat missing data that adequately addressed the uncertainty of the imputed values, based on multiple generated prediction values for missing data [30]. R version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria) was used for the analyses using the linear mixed models. The t–test was performed to compare the levels of air pollutants between heavy–AD days and non–heavy AD days in each heavy-AD definition. One-way analysis of variance (ANOVA) was used to determine whether there were any statistically significant differences in the levels of NO2, ozone, SO2, PM2.5, and SPM among the four definitions of heavy–AD days. The t–test and ANOVA were performed with SPSS software (Japanese version 22.0 for Windows; SPSS Japan Inc., Tokyo, Japan). All the P values were two-sided, and the significance level was set at <0.05.