Introduction

Air pollution is one of the greatest health risks and is estimated to have caused 4.2 million premature deaths worldwide in 2019 (World Health Organization 2022). More than 91% of the world’s population lives in areas that exceed the WHO air quality guideline limits. (World Health Organization 2013). Particulate matter (PM) refers to inhalable particles, composed of carbon monoxide (CO), sulfur dioxide (SO2), nitrogen dioxide (NO2), and ozone (O3) (World Health Organization 2021b). PM is separated into three main groups: coarse particles (PM10), fine particles (PM2.5), and ultrafine particles (PM0.1), with PM2.5 and PM10 being the most common in the regulatory framework (Zaharia 2014). PM can enter the respiratory tract through blood flow and infiltrate the processes of various organs. The health effects related to PM exposure include acute upper and lower respiratory diseases, asthma, chronic obstructive pulmonary disease, pneumonia, bronchiolitis, lung cancer, stroke, dermatitis and allergy, neurodegenerative disease, hypertension in cardiovascular disease, premature birth, premature mortality, decreased birth weight, child height and body mass index (Cordova et al. 2020; Jiang et al. 2020; Kai et al. 2020).

Spirometry measures the volume and flow rate of exhaled breath or breathing in the patient (American Thoracic Society 2000; Summacheeva Foundation 2018). Previous studies have demonstrated the association between exposure to respirable dust and a decrease in lung function parameters. Johanson et al. found that PM2.5, PM10, and NO2 were related to the lower mean percent predicted FVC in patients with idiopathic pulmonary fibrosis (Johannson et al. 2018). Dauchet et al. have reported that short-term exposure to PM10 was associated with decreased FEF75% in healthy individuals (Dauchet et al. 2018). Yang et al. found that PM1 and PM2.5 had significant associations with FVC and FEV1 in primary school children (Mo et al. 2020). Ma et al. found that short-term PM2.5 exposure was associated with decreased pulmonary function, including FEV1/FVC (Ma et al. 2021). Yang et al. reported that an interquartile range increase of PM2.5 concentrations was linked with lower FEV1, FVC, and FEF25 − 75% in adults aged 20 years or older (Yang et al. 2021). Syahira et al. found the personal exposure level of PM2.5 was associated with lung function parameters of traffic policemen (FEV1 and FVC) (Syahira et al. 2020). An average annual concentration of PM2.5 in Bangkok and adjacent urban provinces in Thailand is 25 µg/m3 (range 6–40 µg/m3) (Pollution Control Department 2022), which exceeds both Thailand’s and WHO standards (15 µg/m3 and 5 µg/m3, respectively) (Pollution Control Department 2022).

In Thailand, motorcycle-taxi services are currently more popular than cars or public buses for daily paratransit (Raphiphan et al. 2014). The total number of motorcycle taxi drivers was estimated to be 188,178 people in Thailand (Department of Land Transport 2021). Approximately 72% of them work in the central region of Thailand (Department of Highways 2022). They work outside for an average of 10 h per day (Pannipa et al. 2014). Therefore, their risks of being exposed to ambient air pollution are higher than those of other occupations (Poldongnok and Polyong 2020). However, up to date, the impact of respirable dust exposure on the lung function of motorcycle taxi drivers has not been well-documented.

The objective of this study is to investigate the impact of respirable dust exposure on the lung function parameters of motorcycle taxi drivers in Bangkok and adjacent provinces. For this purpose, we study the relationship between lung function parameters of motorcycle taxi drivers in these regions and their exposure to respirable dust.

Materials and methods

Study area

The study area of this investigation was located in six provinces including Bangkok and its adjacent provinces: Nonthaburi, Pathum Thani, Samut Prakan, Samut Sakhon, and Nakhon Pathom. These provinces are located in the central region of Thailand, with a population of 10,899,786 people and 5.6 million inhabitants (National Statistical Office 2021). More than half of motorcycle taxi driver license holders in Thailand are concentrated in these six provinces (Department of Land Transport 2021). Moreover, this area has a higher PM2.5 concentration than Thailand’s air quality standards (Pollution Control Department 2022).

Study design and participants

A cross-sectional study was conducted among motorcycle taxi drivers who worked in ambient outdoor air between May and June 2022. As shown in Table S1, the total number of motorcycle taxi drivers in the study areas was estimated to be 135,231 people (Ministry of Transport 2021b). The total sample size of subjects was determined so that the 95% confidence interval of the estimated proportion of an attribute became within ± 5% from the estimate using the following equation

$${n = {{p\left( {1 - p} \right)} \over {{{{e^2}} \over {{Z^2}}} + {{p\left( {1 - p} \right)} \over N}}}}$$
(1)

where n is the sample size, N = 135,231 is the population size, e = 0.05 is the level of precision, p = 0.5 is the estimated proportion of the attribute of interest, Z = 1.96 is the Z score of 97.5 percentile of the standard normal distribution (Yamane 1973). As a result, the sample size n was calculated to be 383 persons. The number of subjects in each province was determined in the manner of proportionate stratified sampling using the numbers of motorcycle taxi drivers in the six provinces (Table S1). We informed the motorcycle taxi drivers about inclusion and exclusion criteria at their workplace before they participated. For this reason, all participants satisfied the inclusion criteria, and no one met the exclusion criteria. We do not have data on the number of motorcycle taxi drivers who met the exclusion criteria.

For each province, motorcycle taxi drivers who worked within five kilometers of the closest air quality monitoring station were randomly selected in the manner of purposive sampling. As a result, a total of 343 subjects were recruited. The recruitment was conducted based on the following inclusion and exclusion criteria. The inclusion criteria we used were persons with (1) over 20 years of age, (2) motorcycle taxi drivers who have worked more than one year in Bangkok and adjacent provinces, and (3) Thai nationality and ability to speak the Thai language. Exclusion criteria were persons with (1) diagnosis of acute or chronic respiratory disease by a medical doctor, (2) disease symptoms of thoracic, abdominal or cerebral aneurysms, recent eye hemoptysis, pneumothorax, unstable cardiovascular status, recent myocardial infarction, or acute disorders affecting test performance, such as nausea or vomiting, (3) recent medical treatment of pulmonary embolism surgery, thoracic or abdominal surgical procedures, and (4) activity of drinking alcohol within 4 h, eating a large meal within 2 h, doing vigorous exercise within 30 min, or smoking within 1 h. In most provinces, the number of recruited subjects was smaller than the calculated sample size (Table S1). This is because we have to recruit motorcycle taxi drivers during the large epidemic of COVID-19 by the Omicron variant. Although the total number of recruited subjects did not reach the calculated sample size, the effect of the shortage on the subsequent analyses is small because the total number of recruited subjects is still larger than 266, which is the sample size when we set the level of precision e = 0.06 instead of e = 0.05 in Eq. (1).

All motorcycle taxi drivers recruited in this study received rapid antigen screening tests for COVID-19, and all of them tested negative before the interview and personal air sampling. The staff and subjects strictly followed the procedures to prevent the spread of COVID-19 according to the Public Health Policy (Milanese et al. 2021; Ministry of Public Health 2021; Simaporn and Phattharaphan 2020).

Data collection

Data on participants

All participants completed their questionnaires during face-to-face interviews before examining their lung function tests. Questions in the questionnaires include: demographics, working characteristics, and respiratory symptoms (cough, phlegm, wheeze, and chest tightness). The questionnaires were applied from the Thai version of ATS-DLD-78 A, American Thoracic Society Division of Lung Diseases and Department of Disease Control, Ministry of Public Health, Thailand (Doungrutai 1999). The participant’s height, weight, waist circumference, and blood pressure were measured on-site.

Lung function test

Lung function tests were conducted using calibrated portable spirometers (FIM MEDICAL model Q13 SPIROLYSER). The lung function tests were performed in the standing position (Carrie et al. 2014; Milanese et al. 2021). We followed the American Thoracic Society recommendations on the acceptability and reproducibility of lung function tests (Graham et al. 2019). We used acceptability criteria of a back-extrapolated volume less than or equal to 5% of FVC or 0.100 liter, no cough in the first second of expiration, no evidence of a leak, and an expiratory time longer than 15 s to achieve a plateau. The reproducibility criteria were that the difference between the two largest FVC values must be less than 0.150 L and that difference between the two largest FEV1 values must be less than 0.150 liter. We performed the lung function test for each participant three to eight times, and the maximum of three tests with the best volume-time and flow-volume graphs was recorded. The forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1), the ratio of FEV1 to FVC (FEV1/FVC), and forced expiratory flow between 25% and 75% expired volumes (FEF25 − 75%) were measured. Predicted values of FVC, FEV1, FEV1/FVC and lower limit of normal of FVC, FEV1, and FEV1/FVC (LLN-FVC, LLN-FEV1, and LLN-FEV1/FVC, respectively) were calculated from the gender, age, and height of each participant. We used the National Health and Nutrition Survey III equations adjusted for Asian ethnicity (Dejsomritrutai et al. 2000, 2002). Percent predicted FVC, percent predicted FEV1, and percent predicted FEV1/FVC were calculated by dividing the measured values by the predictive values. Regression equations for the predicted normal values of FVC, FEV1, and FEV1/FVC for males and females are as follows:

$$\eqalign{ FVC\left( L \right)Male = & - 2.601 + 0.122{\rm{A}} - 0.00046{A^2} \cr & + 0.00023{H^2} - 0.00061{\rm{AH}} \cr}$$
(2)
$$\eqalign{ FVC\left( L \right)Female = & - 5.914 + 0.088{\rm{A}} - 0.0003{A^2} \cr & + 0.056H - 0.0005AH \cr}$$
(3)
$$\eqalign{ FE{V_1}\left( L \right)Male = & - 7.697 + 0.123{\rm{A}} + 0.067{\rm{H}} \cr & - 0.00034{A^2} - 0.0007{\rm{AH}} \cr}$$
(4)
$$\eqalign{ FE{V_1}\left( L \right)Female = & - 10.603 + 0.085{\rm{A}} + 0.067{\rm{H}} \cr & - 0.00034{A^2} - 0.0007AH \cr}$$
(5)
$$\eqalign{ FEV1/FVC\left( L \right){\mkern 1mu} Male = & 19.362 + 0.49{\rm{A}} + 0.829{\rm{H}} \cr & - 0.00023{H^2} - 0.00041AH \cr}$$
(6)
$$\eqalign{ FEV1/FVC\left( L \right){\mkern 1mu} Female = & 83.126 + 0.243{\rm{A}} + 0.002{A^2} \cr & + 0.08{\rm{H}} - 0.0036{\rm{AH}} \cr}$$
(7)

A participant with FEV1/FVC ratio greater than or equal to LLN-FEV1/FVC and FVC lower than LLN-FVC was interpreted as a restrictive pattern. A participant with FEV1/FVC ratio lower than LLN-FEV1/FVC and FVC greater than or equal to LLN-FVC was considered as an obstructive pattern. A participant with FEV1/FVC ratio lower than LLN-FEV1/FVC and FVC lower than LLN-FVC was considered as a combined pattern (Paraskeva et al. 2011).

Personal air sampling during service

Motorcycle taxi drivers’ exposure to respirable dust during their service was measured using personal air sampling pumps (UNIVERSAL PCXR8, SKC, Inc., USA) with an aluminum cyclone and PVC membrane filter of 5.0 μm pore size. Personal air sampling pumps were calibrated and set up at 2.5 L/minute. The mass of respirable dust in the air was measured by NIOSH Method 0600 (National Institute for Occupational Safety and Health 1998).

Subjects from the same province were considered to belong to Similar Exposure Groups (SEGs). The number of subjects who underwent personal air sampling in each SEG was determined from the number of subjects in the SEG according to Occupational Exposure Sampling Strategy Manual (Nelson et al. 1977). For each SEG subjects were selected randomly. Table S2 shows the number of subjects who were undergone personal air sampling in each SEG with its breakdown. The measured value of respiratory dust for each motorcycle taxi driver was recorded along with driving distance and the number of passengers on the day of personal air sampling.

The personal exposures to respirable dust concentrations were measured by NIOSH 0600 personal air sampling using a sampler (SKC, PCXR8, USA) operated at 2.5 L/min, which was connected to an aluminum cyclone (SKC, Impact Sampler, USA) and polyvinyl chloride (PVC) filters (10-mm diameter, 5 μm pore size). We installed personal air sampling equipment at the motorcycle taxi drivers’ breathing zone (National Institute for Occupational Safety and Health 1998). We collected air samples during their working time. The mean sampling duration was 290.4 (minutes) with an SD of 32.3 (minutes). Smoking was not allowed during personal sampling. The samples were wrapped with parafilm and sealed in a box at the sampling site. They were transferred to the analytical lab. After collecting air samples, the humidity of the filters was absorbed in a desiccator by silica gel, and then the filters were passed through a static master (Mettler Toledo, Haug Deionizer) to eliminate the static electricity. Blank filters for the air samples were extracted and analyzed using the same protocol as the samples. Filters were weighed with microbalance (Mettler Toledo, XPR2U, Switzerland) at an average of three times, and the respirable dust concentration of each sample was calculated using Eq. (8).

$${C = {{\left( {{W_2} - {W_1}} \right) - \left( {{B_2} - {B_1}} \right)} \over V} \times {{10}^3}}$$
(8)

Where C is the concentration (mg/m3), W1 is the tare weight of the filter before sampling (mg), W2 is the post-sampling weight of the sample-containing filter (mg), B1 is the mean tare weight of blank filters (mg), B2 is the mean post-sampling weight of blank filters (mg), V is the sampled air volume (liter).

Air pollution data from air quality monitoring stations

Air pollution data from Air Quality Monitoring (AQM) stations measured six parameters, including Particulate matter with an aerodynamic diameter less than or equal to a nominal 2.5 micrometers (PM2.5), Particulate matter with an aerodynamic diameter less than or equal to a nominal 10 micrometers (PM10), Ozone (O3), Carbon Monoxide (CO), Nitrogen Dioxide (NO2), Sulfur Dioxide (SO2) in Bangkok, Nonthaburi, Pathumthani, SamutPrakan, Samutsakhon, and Nakhonpathom during May 1, 2022 to June 30, 2022 were retrieved from the database of the Pollution Control Department in Thailand (Pollution Control Department 2022). This period covered the all days when personal air sampling was conducted.

Statistical analysis

The data were analyzed using RStudio (version 2023.09.0). For each categorical variable, hypothesis testing was performed using a null hypothesis assuming that means of lung function parameters were the same across participants belonging to different categories. The t-test was used when the categorical variable had two values and the one-way ANOVA was used when there were more than two categories. For each numerical variable, Pearson’s correlation coefficients between the variable and lung function parameters were calculated. The significance level was set to 0.05 for all statistical analyses.

Ethical approval

Ethical approval to conduct this study was received from the Human Research Ethics Committee of Thammasat University, No.3. Ethical approval number 031/2565, and the approval date was 29 April 2022.

Results

Demographics, work characteristics, and respiratory symptoms of participants

Table 1 shows the demographics and working characteristics answered in questionnaires by recruited motorcycle taxi drivers. Statistics on numerical variables and disease symptoms are found in Table S3 and Table S4, respectively. Of 343 subjects in the six provinces, 298 (86.9%) were males and 45 (13.1%) were females. Most of the participants (81.9%) were older than or equal to 40 years old. The 37.9% of the participants were current smokers, 45.8% never smoked before, 30.3% had a history of COVID-19 infections, 58.3% had been working as motorcycle taxi drivers for more than 10 years, 84.8% worked more than or equal to 8 h a day, 85.7% wore surgical masks, 13.4% wore cotton masks while driving, and N95 masks were rarely used. The 41 (12%) participants had the symptoms of chronic bronchitis and 25 (7%) had the symptoms of acute bronchitis (Table S4), although these diseases were included in the exclusion criteria. This indicated that these motorcycle taxi drivers had the symptoms of chronic or acute bronchitis without medical diagnosis.

Table 1 Demographics and working characteristics of participants (n = 343)
Full size table

Lung function tests

Table 2 shows statistics on lung function parameters of motorcycle taxi drivers. The percent predicted FVC of all participants had a mean of 94.45 with a standard deviation (SD) (Punyaratabandhu et al. 2006) of 18.10. Percent predicted FEV1 of all participants had a mean of 97.04 with an SD of 18.42. The percent predicted FEV1/FVC of all participants had a mean of 99.40 with an SD of 12.15. Table 3 shows lung function abnormality suggested by percent predicted FVC and percent predicted FEV1/FVC. Of 343 participants from six provinces, 47 (13.7%) showed the restrictive pattern, 42 (12.2%) showed the obstructive pattern, and 7 (2.0%) showed a combined pattern. Other 247 (72.0%) showed the normal pattern.

Participants from PTT had the lowest mean value of percent predicted FVC (85.74) and percent predicted FEV1 (87.55) among six provinces (Table 2). The mean of percent predicted FEV1/FVC in PTT was 99.67, which was the third lowest among the six provinces. Of 37 participants in PTT, nine (24.3%) showed a restrictive pattern (Table 3). This proportion of restrictive patterns in PTT was the highest among the six provinces.

Participants from SPK had the lowest mean value of percent predicted FEV1/FVC (96.82), which was the lowest mean among the six provinces (Table 2). The mean of percent predicted FVC in SPK was 97.02, which was the second highest among the six provinces. The mean of percent predicted FEV1 in SPK was 97.04, which was the third lowest among the six provinces. Of 52 participants in SPK, twelve (23.1%) showed an obstructive pattern (Table 3). This proportion of obstructive patterns in SPK was the highest among the six provinces.

Table 2 Statistics of lung function parameters of participants in six provinces
Full size table
Table 3 Lung function abnormality suggested by percent predicted FVC and percent predicted FEV1/FVC
Full size table

Personal air sampling during service

Table 4 shows statistics on the concentration of respirable dust collected by air personal sampling. The median, mean, and SD of respirable dust concentration for all participants were 9.8 (µg/m3), 19.3 (µg/m3), and 37.6 (µg/m3), respectively. The means of respirable dust concentration were different among provinces (p-value = 0.016 with one-way ANOVA). The respirable dust concentration in BKK had a mean of 31.0 (µg/m3) and SD of 56.9 (µg/m3), and these were the highest among the six provinces. On the other hand, the median of respirable dust concentration was the highest in PTT at 26.3 (µg/m3), which was 9.7 (µg/m3) more than the median in BKK.

Table 4 The concentration of respirable dust from air personal sampling
Full size table

Table S5 shows the driving distance of motorcycle taxi drivers on the day of personal air sampling. The median, mean, and SD of driving distance for participants in six provinces were 22.0 (km), 26.3 (km), and 16.7 (km), respectively. The means of driving distances were different among provinces (p < 0.001 with one-way ANOVA). The mean of driving distances was the highest in BKK (33.3 km), while the median was the highest in SKN (30.0 km). Pearson’s correlation coefficient between respirable dust concentrations and driving distances was 0.422 with 95%CI from 0.282 to 0.544 (p < 0.001) (Fig. 1).

Fig. 1
figure 1

Relationship between respirable dust concentrations and driving distances

Full size image

Air pollution data from air quality monitoring stations

Table S6 shows the 24-hour average concentration of PM2.5 and PM10 at the AQM stations from May 1 to June 30, 2022. The means of PM2.5 and PM10 concentrations were different among provinces (p < 0.001 with one-way ANOVA for both). The highest median and mean of PM2.5 and PM10 concentrations were recorded in BKK. The second-highest median and mean of PM2.5 and PM10 concentrations were recorded in SKN. The maximum of 24-hour average concentrations of PM2.5 and PM10 in six provinces during the study period were lower than Thailand’s ambient air quality standards (PM2.5 ≤50 µg/m3, PM10 ≤ 120 µg/m3 before June 1, 2023). However, the median and mean of 24-hour average concentrations of PM2.5 in BKK, SPK, and SKN were higher than the WHO air quality guidelines recommend levels (15 µg/m3) (World Health Organization 2021a). The median and mean of 24-hour average concentrations of PM10 in BKK were higher than the WHO’s guidelines (45 µg/m3).

Variables Associated with lung function parameters

Table 5 shows categorical variables significantly associated with a lung function parameter. The means of percent predicted FVC, percent predicted FEV1, and percent predicted FEV1/FVC were different among the provinces of participants (p-value = 0.027, 0.001, 0.003, respectively) (Figure S1). The means of percent predicted FEV1 and percent predicted FEV1/FVC were different depending on whether participants took a break or not (p-value = 0.046 and 0.028, respectively) (Figure S2). Other categorical variables found to be associated with lung function parameters were respiratory symptoms and the status of other diseases answered in the questionnaire. Percent predicted FVC was significantly associated with a history of allergy, acute bronchitis, and chronic bronchitis (p-value = 0.016, 0.050, and 0.029, respectively). Percent predicted FEV1 was significantly associated with nasal allergy, chest pain, acute bronchitis, and chronic bronchitis (p-value = 0.013, 0.039, 0.007, and 0.002, respectively). Percent predicted FEV1/FVC was significantly associated with diabetes mellitus, hypertension, neuromuscular, chest pain, history of asthma, history of pneumonia, history of allergy, acute bronchitis, and persistent cough (p-value = 0.045, 0.023, 0.039, 0.035, 0.023, 0.038, 0.003, 0.014, and 0.050, respectively). A list of p-values for the association between each lung function parameter and all categorical variables is provided in Tables S8, S9, and S10.

Table 6 shows numerical variables significantly associated with a lung function parameter. A list of Pearson’s correlation coefficients between each lung function parameter and all numerical variables is provided in Tables S11, S12, and S13.

Distance of driving was positively correlated with the percent predicted FVC with Pearson’s correlation coefficient of 0.1119 (95%CI: 0.006, 0.215) (p = 0.038) and the percent predicted FEV1 with 0.1242 (95%CI: 0.019, 0.227) (p = 0.021). The break period was positively correlated with the percent predicted FEV1/FVC with Pearson’s correlation coefficient of 0.1352 (95%CI: 0.030, 0.238) (p = 0.010). The number of cigarettes per day was negatively correlated with the percent predicted FEV1 with Pearson’s correlation coefficient of − 0.1270 (95%CI: −0.230, − 0.021) (p = 0.019) and the percent predicted FEV1/FVC with Pearson’s correlation coefficient of − 0.1159 (95%CI: −0.219, − 0.010) (p = 0.032). These results indicated that people who smoked a large number of cigarettes tended to have a small percent predicted FEV1 and a small percent predicted FEV1/FVC.

Table 5 Categorical variables associated with lung function parameters
Full size table
Table 6 Numerical variables associated with lung function parameters
Full size table

Relationship between lung function parameters and respirable dust concentration

Individual level

The percent predicted FVC and the respirable dust concentration collected by personal air sampling had a correlation coefficient of 0.097 (95%CI: −0.062, 0.252), and the correlation was not significant (p = 0.231) (Table S11 and Figure S3A). The percent predicted FEV1 and the respirable dust concentration collected by personal air sampling has a correlation coefficient of − 0.012 (95%CI: −0.170, 0.147), and the correlation was not significant (p = 0.883) (Table S12 and Figure S3C). The percent predicted FEV1/FVC and the respirable dust concentration collected by personal air sampling has a correlation coefficient of − 0.084 (95%CI: −0.240, 0.075), and the correlation was not significant (p = 0.299) (Table S13 and Figure S3C). None of the percent predicted FVC, percent predicted FEV1, or percent predicted FEV1/FVC was significantly correlated with the respirable dust concentration collected by personal air sampling at the individual level.

Station level

Table S14 shows statistics on lung function parameters of participants in each of the twenty-one workstations. Workstation PTT_3 had the lowest mean value of percent predicted FVC at 74.70, followed by PTT_1 at 78.27. Workstation PTT_3 also had the lowest mean value of percent predicted FEV1 at 80.69, also followed by PTT_1 at 81.94. On the other hand, the mean of percent predicted FEV1/FVC was the lowest in SPK_3 at 95.76 and the second lowest in SPK_2 at 95.88.

Figure 2 shows the relationship between the mean of lung function parameters in each workstation and the mean of respirable dust concentrations measured by personal air sampling and the mean of driving distances at the workstation on the day of sampling (Table S15). Workstations’ means of percent predicted FEV1 and workstations’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.440 (95%CI: −0.733, − 0.010) and the correlation was significant (p = 0.046) (Fig. 2C). This result indicated that motorcycle taxi drivers working at workstations with the higher respirable dust concentration tended to have the smaller percent predicted FEV1. Workstations’ means of percent predicted FVC and workstations’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.232 (95%CI: −0.603, − 0.222) but the correlation was not significant (p = 0.312) (Fig. 2A). Similarly, workstations’ means of percent predicted FEV1/FVC and workstations’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.360 (95%CI: −0.685, − 0.085) but the correlation was not significant (p = 0.109) (Fig. 2E).

Fig. 2
figure 2

Relationship between lung function parameters (percent predicted) and respirable dust concentrations and driving distance at the workstation level. (A) FVC and respirable dust concentrations. (B) FVC and driving distance (C) FEV1 and respirable dust concentrations. (D) FEV1 and driving distance (E) FEV1/FVC and respirable dust concentrations. (F) FEV1/FVC and driving distance

Full size image

Province level

Figure S4C shows the relationship between the mean of lung function parameters in each province and the mean of respirable dust concentrations measured by personal air sampling and the mean of driving distances in the province on the day of sampling. Provinces’ means of percent predicted FVC and provinces’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.432 (95%CI: −0.921, − 0.584) but the negative correlation was not significant (p = 0.392) (Figure S4A). Provinces’ means of percent predicted FEV1 and provinces’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.700 (95%CI: −0.964, − 0.258) but the negative correlation was not significant (p = 0.122) (Figure S4C). Provinces’ means of percent predicted FEV1/FVC and provinces’ means of respirable dust concentration have Pearson’s correlation coefficient of − 0.569 (95%CI: −0.944, 0.451) and the correlation was not significant (p = 0.238) (Figure S4E).

Air quality monitoring stations

Figure S5 shows the relationship between the mean of lung function parameters in each province and the 61-day mean of respirable dust concentrations measured at the AQM station in the province. None of the correlations between provinces’ means of lung function parameters and the mean respirable dust concentration of PM2.5 or PM10 was significant (Figure S5).

Discussion

In this study, we recruited a total of 343 motorcycle taxi drivers in Bangkok and adjacent provinces, and their lung function parameters were analyzed using spirometry. Of these, 153 participants were selected, and their exposure to respirable dust during service was measured by personal air sampling. We analyzed the association between lung function parameters of participants and exposure to respirable dust as well as other variables answered in the questionnaire.

According to the Ministry of Transport, Thailand, 93.7% of license holders of motorcycle taxi drivers in the studied six provinces were male (Ministry of Transport 2021a). Among 343 motorcycle taxi drivers recruited in this study, 298 (86.9%) were male and 45 (13.1%) were female. The male-to-female ratio in this study significantly differs from that of license holders reported by the Ministry of Transport (p-value < 0.001). This result indicated that the number of working female motorcycle taxi drivers was significantly larger than expected from the statistics of license for folders. As far as we know, no previous research has reported such a difference in the male-to-female ratio between working motorcycle taxi drivers and license holders. The predominance of male motorcycle taxi drivers was reported by previous studies in Pakistan (87.6%), Vietnam (97.3%), and Brazil (98.7%) (Felix et al. 2016; Mansoor et al. 2023; Truong et al. 2021).

Among 343 motorcycle taxi drivers in this study, 157 (45.8%) were non-smokers. The percentage of non-smokers in the Thai population is 72.4% (National Statistical Office Thailand 2017). The number of non-smokers in motorcycle taxi drivers is significantly smaller than expected from the percentage of non-smokers in the Thai population (p-value < 0.001). Among 298 male motorcycle taxi drivers in this study, 117 (34.1%) were non-smokers. The percentage of non-smokers in the male population in Thailand is 45.8% (National Statistical Office Thailand 2017). The number of non-smokers among male motorcycle taxi drivers is significantly smaller than expected from the percentage of non-smokers in the male population in Thailand (p-value = 0.023). Among 45 female motorcycle taxi drivers, 40 (88.9%) were non-smokers. The percentage of non-smokers in the female population in Thailand is 97.3% (National Statistical Office Thailand 2017). The number of non-smokers among female motorcycle taxi drivers is significantly smaller than expected from the percentage of non-smokers in the female population in Thailand (p-value = 0.007).

Of 343 participants, 294 wear surgical masks, 47 wear cotton masks, one wears a N95 mask, and one wears no masks (Table 1). There was no significant relationship between the type of mask and lung function parameters. P-values for the one-way ANOVA tests using a null hypothesis assuming that the mean of (percent predicted) FVC, FEV1, and FEV1/FVC are the same for all types of masks were 0.705, 0.194, and 0.268, respectively (Table S11, S12, and S13). This result suggested that wearing a surgical mask did not contribute to protecting their lung function from respirable dust significantly compared to wearing a cotton mask. The effect of wearing a surgical mask on exposure to respirable dust during the service of motorcycle taxi drivers should be investigated further. This study was conducted during the COVID-19 epidemic by the Omicron variant. As shown in Tables 1 and 85.7% of motorcycle taxi drivers wore a surgical mask. The purpose of wearing masks may be to protect from COVID-19 infectious or to filter out respirable dust. The percentage of subjects who wore cotton masks is only 13.4%. The large p-value may be attributed to the large number of subjects with surgical masks. Future research should include a sufficient number of subjects with cotton masks to evaluate the effect of surgical masks over cotton masks.

We found that the means of (percent predicted) FVC, FEV1, and FEV1/FVC were different among the provinces of participants (Table 2). The means of PM2.5 and PM10 concentrations monitored at the AQM stations were significantly different among provinces (Table S6). However, none of the correlations between province means of lung function parameters and the province mean respirable dust concentration of PM2.5 or PM10 was significant (Figure S5). Similarly, the province mean of respirable dust concentrations collected with personal air sampling was not significantly correlated with the province mean lung function parameters (Figure S4). Among motorcycle taxi drivers in six provinces, those who were in Pathum Thani had the most severe lung function impairment. Motorcycle taxi drivers in Pathum Thani had the lowest mean percent predicted FVC and mean percent predicted FEV1 among six provinces (Table 2). Motorcycle taxi drivers in Pathum Thani also had the highest lung function abnormality (Table 3). Among 37 motorcycle taxi drivers in Pathum Thani, nine (24.3%) showed restrictive patterns, three (8.1%) showed obstructive patterns, and two (5.4%) showed combined patterns (Table 3). Considering that Pathum Thani had the highest median and highest minimum of respirable dust concentrations among the six provinces, respirable dust exposure may have caused restrictive lung function impairment.

The workstation mean of respirable dust concentrations collected with personal air sampling was negatively correlated with the workstation mean percent predicted FEV1 significantly (Fig. 2). The linear regression analysis using workstations’ means of percent predicted FEV1 as a dependent variable and workstations’ means of respirable dust concentrations as an independent variable resulted in a coefficient of − 0.238 and intercept of 100.762. the p-value for the coefficient to be zero is 0.045. These results suggested that motorcycle taxi drivers working at workstations where they were exposed to respirable dust at high concentrations tended to have a low percent predicted FEV1. In fact, PTT_3 has the lowest mean of percent predicted FEV1 and the highest mean of respirable dust concentration (Table S14, Table S15). The respirable dust concentration measured by personal air sampling was not significantly correlated with the lung function parameters at the individual level (Figure S3). The reason for this may be because personal air sampling was conducted only one day for each participant. The damage in lung functions is not determined by the respirable dust exposure in one day. A longitudinal study rather than a cross-sectional study is needed to investigate respirable dust exposure at the individual level.

Some respiratory symptoms were significantly associated with lung function parameters (Table 5). Percent predicted FVC was significantly associated with symptoms of acute and chronic bronchitis (p-value = 0.050 and 0.029, respectively). Percent predicted FEV1 was significantly associated with symptoms of acute and chronic bronchitis (p-value = 0.007 and 0.002, respectively). Percent predicted FEV1/FVC was significantly associated with symptoms of acute bronchitis and persistent cough (p-value = 0.014 and 0.050, respectively). These results are consistent with previous studies (Hoesterey et al. 2019; Worrall 2008). Our results indicated lung function tests can detect the initial stage of airway disease. So, medical check-ups of motorcycle taxi drivers using spirometers have a crucial implication for the health and well-being of motorcycle taxi drivers. Motorcycle taxi drivers with abnormal lung function will receive medical treatment, and they may think of working in another area with less respirable dust to prevent the disease from progressing.

Lung function parameters were significantly correlated with several numerical variables at the individual level (Table 6). The cigarette amount is negatively correlated with the percent predicted FVC and FEV1. Since cigarette smoking is a risk factor for lung function impairment, it is important to control cigarette smoking as a confounding factor when we evaluate the effect of respirable dust exposure on lung function impairment. We will discuss this point in the next paragraph. Distance of driving was positively correlated with the (percent predicted) FVC and FEV1 at the individual level. People with large (percent predicted) FVC and large FEV1 tended to drive long distances, and people who have small (percent predicted) FVC and FEV1 tended to drive short distances. Abnormality of lung functions may be the cause of short driving distances. The break period was positively correlated with the percent predicted FEV1/FVC. The reason for the positive correlation was not clear and required further investigation. It may be possible that having longer break periods can increase the percent predicted FEV1/FVC. Future research must be conducted to uncover the underlying mechanism of correlation between the break period and FEV1/FVC.

The number of cigarettes per day was negatively correlated with the percent predicted FEV1 (p-value = 0.019) and the percent predicted FEV1/FVC (p-value = 0.032) (Table 6). These results are consistent with previous studies that found cigarette smoking was associated with lung impairment (Alobaidi et al. 2022; Fuertes et al. 2018; Gomes et al. 2015; Medbø & Melbye 2007; Oelsner et al. 2020). For this reason, motorcycle taxi drivers who smoke cigarettes have an increased risk of lung function impairment in addition to those from respirable dust exposure.

Smoking and age are risk factors for lung function impairment as well as respirable dust concentration. To control these confounding factors, we conducted a multiple linear regression analysis of the mean percent predicted FEV1 at the workstation level using mean respirable dust concentration, mean cigarette amount, and mean age as independent variables. The coefficient of mean respirable dust concentration, mean cigarette amount, and mean age were − 0.2987 (p-value = 0.023), − 0.8537 (p-value = 0.132), and − 0.4860 (p-value = 0.449), respectively. The workstation mean of respirable dust concentration was the only independent variable with a significant p-value.

In this study, we have data only from motorcycle taxi drivers, all of whom are the exposed group. We have no data from the unexposed group, who is not working on the road. We can not directly compare the lung function of motorcycle taxi drivers. However, we used percentages of predicted FVC, FEV1, and FEV1/FVC, which were calculated based on data from the general population in Thailand. Predicted FVC, FEV1, and FEV1/FVC can be considered as data from the unexposed group.

Conclusions

The means of (percent predicted) FVC, FEV1, and FEV1/FVC of motorcycle taxi drivers were different among the six provinces investigated in this study. Respirable dust concentrations collected with personal air sampling suggested that motorcycle taxi drivers working at workstations where they were exposed to respirable dust at high concentrations tended to have a low percent predicted FEV1. This is the main finding of this study. The 12% and 7.3% of motorcycle taxi drivers recruited in this study had the symptoms of chronic bronchitis and acute bronchitis, respectively, but they were not diagnosed as bronchitis by a medical doctor. These results suggested that some of motorcycle taxi drivers around Bangkok and adjacent provinces had a high risk of occupational exposure to ambient air pollution. This study highlighted the importance of reducing the respirable dust exposure of motorcycle taxi drivers. To achieve this, the government first needs to monitor the concentration of respirable dust on the roads in Bangkok and the adjacent provinces. Next, the government needs to find a way to reduce the respirable dust in vehicle exhausts. Finally, the government needs to establish a medical check-up system for motorcycle taxi drivers to monitor their health status.