Table of Contents
Study site
The study site was a reinforced concrete products plant located in the industrial zone of Almaty that produces parts for bridges and other products of reinforced concrete. The plant occupies a territory of 7.52 hectares (14,838 m2). The plant has 198 permanent staff, but more people may be hired on short-term contracts depending on the workload. Following expert consultation with a plant occupational hygienist, we identified four typical locations of exposure in the production cycle and additionally one more office location as control.
The first location was a concrete-mixing unit, where cement is mixed with other components in rotating tanks, and the PM is generated when loading dry cement and mixing it with water, macadam and other components. Exposures in this location include cement dust, occupational noise and who-body vibration. This process is mostly automatized and does not require human presence in close proximity to the aerosol generation source. Personnel working in this unit enter the location once the mixture is ready, but residual air pollution is present and visible. Compared to other processes, this unit employs the smallest number of workers.
In the second and third locations the primary exposures were to metal aerosols. The second location was the armature shop, where workers knit the structures as future skeletons of reinforced concrete parts from the metal armature. The third location was metalworking workshop, where the structures created in the armature unit are welded. Aerosol in this location is generated mostly from welding, but some metal cutting using plasma cutting machines is also present.
The fourth location was the molding workshop, which employs most of the plant workforce, and in which welded metal skeletons are filled with concrete and molded into the needed shape, then dried and tested. Production occurs both indoors and outdoors, and the outdoor work is prioritized in summer and for large reinforced concrete parts. These workplaces expose workers to cement dust, steam, which is used to consolidate reinforced concrete, and noise. Additionally, final product polishing may generate aerosol.
Finally, the office accommodating accounting, human resources, procurement and other related personnel, was also included in this study as control. There is no vapor, gas, dust or fume exposure in these workplaces, but the office building is located near the main production sites. The plant also accommodates other ancillary departments, units and workshops, such as the canteen, security, boiler plant (steam and hot water are needed to accelerate concrete solidification in the final product) and others.
Exposure assessment
Exposure data were collected in summer 2023, because otherwise ambient air pollution could have biased the true concentrations given that ambient air pollution levels in Almaty in the heating season are high [4]. In each of the five locations in the plant, we collected ten personal samples using portable PM2.5 cyclones and filters. Sampling occurred throughout the full workday, independent on the amount of work and the overall load. Because smoking was not allowed inside production buildings, occupational sources were the only sources of aerosol. Overall, we collected 50 personal samples from workers from five typical locations at the production site. Sampling duration ranged from 6 to 8 h.
Personal samples were obtained from one or two workers from each selected workplace using a portable, battery-powered pump (TM30A-B, TOPSFLO, China) set to provide a flow of 2 l/min for a pre-calibrated cyclone. This portable pump provided a constant air flow through the portable PM2.5 cyclone AE2.5-mini (Alaric Electromechanics, USA) to a cassette containing a pre-weighted AФA-BП-20–1 filter (Soyuzhimprom, Russian Federation), and powered from an external battery. This set was placed on a belt of a worker, whereas the edge of the hose was fixed in the breathing zone of a worker. Filters were weighed prior to and after the sample collection using HR-60 (AND, Japan), and the mass concentration of collected PM2.5 was calculated as the difference between the filter mass after and prior to sample collection and divided by the overall air volume pumped (in µg/m3). For all technical requirements, including the range of acceptable flow rate, recalculation for standard conditions, etc. we followed the actual State Standards (ГOCT) [11, 12].
Questionnaire and lung function testing
All workers provided informed written consent prior to participation, and the study was approved by the Committee on Bioethics of al-Farabi Kazakh National University. In compliance with the local legislation, all workers, including the office staff, undergo pre-employment and annual medical screening, including spirometry when needed. For the current analysis, we offered a structured validated questionnaire either in Russian or Kazakh to all workers who agreed to participate in this study. Workers completed the questionnaires and performed spirometry in a specially designated office in the administrative building of the plant. This questionnaire aimed to collect a detailed working history ascertaining duration of employment in all previously and currently held positions. In addition to occupational history, we collected information on age, sex, place of residence, smoking history, alcohol use, regular physical activity, and respiratory symptoms. Cigarette smoking status ascertainment yielded stratification into three categories, including current smoking, former smoking and never smoking. Those exercising at least 3 times a week for at least 40 min off work were considered physically active.
Respiratory symptoms were assessed using validated COPD Assessment Test (CAT), which yielded the score from 0 to 40, when the score above 10 was usually indicative of “many symptoms” [13]. The test is useful for chronic obstructive disease management and in cases with already known diagnosis. Dyspnea was measured with widely used mMRC scale, in which the score varies from 0 to 4 [14]. We preferred these two tools in our study for its simplicity and clear score interpretation. We also asked whether a subject had ever had ever been diagnosed with chronic bronchitis, COPD, asthma or allergic rhinitis by a physician.
On the day of examination, workers also asked to refrain from smoking for at least two hours prior to the spirometry test. Three or more reproducible (difference between tests 100 ml or below) maneuvers of vital capacity (VC), followed by the same number of reproducible maneuvers of forced VC (FVC) were completed. We also measured forced expiratory volume in one second (FEV1). In subjects with FEV1/FVC below 0.7 we had the second set of similar maneuvers following 15 min after inhalation of 300 µg of salbutamol (“Binnopharm”, Russian Federation), and recorded post-bronchodilator readings. We recorded actual values of these volumes, then computed percent predicted values for each subject, except FEV1/FVC, using Global Lung Function Initiative (GLI-2012) equations. There were no participants regularly using any bronchodilator, inhaled steroids or other respiratory medication in the study. All tests were performed using MAS 2S office spirometer (Belintelmed, Belarus).
Statistical analysis
The primary outcome of interest in the association of cumulative exposure to PM2.5 in the workplace with current respiratory symptoms and lung function. We were also interested in the quantitative analysis of the magnitude of exposure in range of workplaces within this production.
All variables were evaluated for normality using Shapiro–Wilk test, and the majority of primary variables were non-normally distributed. Hence, we present data as medians with the corresponding interquartile ranges (IQR) or, alternatively, as means with standard deviation (SD). Unless otherwise stated, we utilized and presented the outcomes of non-parametric analytical methods across the manuscript, such as p-values from the Mann–Whitney U-test for continuous two-group data comparison or from χ2 test for binary data. In case of normally distributed data analysis, such as for lung function indices, we used t-tests. When more than two groups of non-normally distributed data were compared, we used Kruskall-Wallis test.
Exposure data were summarized as the value in µg/m3 for each of 50 days of sampling, including: 10 days from the armature workshop, 10 days from the molding workshop, 10 days from the metalworking workshop, 10 days from the concrete-mixing unit and 10 more days from the office. We then calculated the medians of PM2.5 concentrations for each location. We tested whether between-workshop variance exceeded the one within workshops using non-parametric Kruskall-Wallis test and presented the corresponding p-value from the test.
The computed arithmetic mean concentrations for a given workshop were then used to calculate the cumulative dose as a metric of exposure throughout the entire employment for each included worker. Because this specific production needs skilled workers who would unlikely change their occupation throughout their career, we assumed that the current position in the company would be similar to any past employment elsewhere. Therefore, current exposures in the workplace will reflect the picture of exposure a worker have had in the workplace. Hence, exposure metric used and analyzed in this study is the function of PM2.5 concentration in the current position and the overall work duration (cumulative PM2.5 dose, mg/m3-year).
The final stage in the analysis was testing the association of cumulative exposure with the outcomes of interest, including CAT score as a surrogate of respiratory symptoms and FEV1/FVC using linear regression models. First, we confirmed the effect of selected variables in the univariate analysis, including age, female sex, cumulative dose, chronic bronchitis and current cigarette smoking with CAT as an outcome. For FEV1/FVC, the range of such predictors was narrower and only comprised cumulative dose and production group. We then included them in the adjusted models adjusted for the total of three, except age (in case of CAT) and two (in case of FEV1/FVC) variables. We report beta coefficients for significant (p < 0.05) predictors from the adjusted for confounders models with the corresponding 95% confidence intervals (CI). All tests were completed in NCSS 2021 (Utah, USA), and p-values below 0.05 were considered significant.
