The association between sources of air pollution and respiratory health in Pretoria, South Africa

Abstract

BACKGROUND: The United Nations Sustainable Development Goals address, amongst other current issues, air pollution, namely Goal 3.9: "By 2030, substantially reduce the number of deaths and illnesses from hazardous chemicals and air, water and soil pollution and contamination"; Goal 11.6: “By 2030, reduce the adverse per capita environmental impact of cities, including by paying special attention to air quality and municipal and other waste management” and Goal 11.6.2: “Annual mean levels of fine particulate matter (e.g. PM2.5 and PM10) in cities (population weighted)”; and Goal 13: "By 2030, take urgent action to combat climate change and its impacts". The South African (SA) Air Quality Standards are not as stringent as the WHO guidelines. The WHO guidelines recommend ambient PM2.5 levels to be 15 µg.m-3 over 24 hours and 5 µg.m-3 over a 1-year period while the SA standards recommend 40 µg.m-3 over 24 hours and 20 µg.m-3 over a 1-year period. The 2005 WHO guidelines are based on epidemiological evidence conducted and are continuously being reviewed. Epidemiology studies researching the health effects related to short- and long-term exposure to criteria air pollutants have become important in South Africa for input into more stringent national air quality policies. METHODOLOGY: The first objective of the project was to establish a trend in the sampled PM2.5 and constituent trace elements. PM2.5 samples were collected for 24 hours (from 9 a.m. to 9 a.m.) every third day between 23 April 2017 to 28 February 2020. Duplicate samples were collected every fifteenth day. The sampling site is located on the roof of the HW Snyman building at the School of Health Systems and Public Health, Faculty of Health Sciences, at the University of Pretoria. The campus is located next to the main arterial road (Steve Biko) running alongside the Prinshof Campus and the Tshwane Hospital. The seasons were classified into autumn (18 April – 31 May and 1 March – 17 April), winter (1 June – 31 August), spring (1 September – 30 November) and (1 December to 28 February). When averaging the duplicate samples, a total of 350 observations for the 34-month dataset were recorded. Gravimetric analysis was used to determine PM2.5 mass. The data was captured on Microsoft Excel (Microsoft Corporation, Redmonton, WA, USA) which was also used to calculate air concentrations from the XRF data. All other calculations were done using STATA 15 statistical software. X-ray fluorescence spectrometry (XRF; XEPOS 5, SPECTRO Analytical Instruments GmbH, Germany) was used to determine concentrations of elements between aluminium and uranium in the periodic table. Results which were 30% below the limit of detection (LoD) were excluded. The duplicate sample values were averaged. Precision was estimated using the root mean square method on the duplicate sample concentration. RESULTS: PM2.5 is composed of other constituents such as organic matter, soot, black carbon and secondary constituents.1-6 The sum of the trace elements is 0.99 µg.m-3 and makes up 12.10% of the total PM2.5 over the 34-month study period. The mean PM2.5 concentration for the 34-month study period was 23.2 ± 17.3 µg.m-3 with a range of 0.69 – 139 µg.m-3. The 34-month mean concentration was above the yearly WHO guideline (5 µg.m-3) and the yearly South African NAAQS (20 µg.m-3). During the study period, the 24-hour WHO guideline of 15 µg.m-3 was exceeded on 217 out of the 350 days and the 24-hour South African NAAQS of 40 µg.m-3, on 53 days. The average PM2.5 levels were significantly higher during autumn and winter than during summer and spring (p<0.001). The most abundant trace element was S (1175 ng.m-3), followed by Si (646 ng.m-3) and K (250 ng.m-3). The median levels of these 3 trace elements and Cl were significantly higher in winter than in spring and summer (p<0.001); most likely due to higher rainfall and humidity during the warmer seasons. The positive matrix factorisation model (PMF) was run three times for three factor configurations (4, 5 and 6) for the 34–month study. The 6-factor source contribution is assigned to the campaign with resuspended matrix (24%), mining (33%), exhaust (12%), industry (15%), biomass burning (4.2%) and vehicular emissions (12%), contributing to the modelled total annual PM2.5 concentration. Two sensitivity tests were performed. A test for the winter season whereby As, Se and Pb was included in the dataset and a full year test whereby Ni is adjusted. The geographical origin of air masses that passed the sampling site, also known as, transport clusters, was determined using the HYbrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT). The model indicated four main transport clusters for the 34-month period: 42% of the wind trajectories were from the westerly direction (W). The remaining three main transport clusters were north easterly (NE) (29%), the south easterly (SE) (15%) and from the long-range Indian Ocean (LRIO) (14%) direction. A health risk assessment was performed for adults, children, and infants by performing a standardised health risk assessment. For children and infants, the HQ values were approximately 15 and 16 when using the yearly WHO guideline as a reference concentration. The HQ values for PM2.5 when using the yearly SA NAAQ standards were between 3.47 for children and 3.78 for infants. When using the WHO guideline, the HQ values were 6, 18 and 20 for adults, children and infants, respectively. When using the SA NAAQS the HQs were 1, 4 and 5 for all three age groups during the 34-month study period. The HQ demonstrated the same seasonal patterns, with a higher mean concentration in winters, when evaluating for both WHO (5 µg.m-3) and SA NAAQS (20 µg.m-3) values. The case-crossover epidemiology study investigated the association between hospital admissions for all respiratory diseases (J00-J99) and PM2.5 as analysed at the School of Health Systems and Public Health. A full dataset including total PM2.5, soot, BC, UV-PM and nine trace elements is assessed. Respiratory hospital admissions increased significantly by 2.7% (95% CI: 0.6, 4.9) per 10 µg.m-3 increase in PM2.5. When analysing the trace elements, an increase in hospital admissions increased significantly for Ca by 4.0 % (95% CI: 1.4% - 6.8%), Cl by 0.7 % (95% CI: 0.0% - 1.4%), Fe by 3.3 % (95% CI: 0.5% - 6.1%) and for Si by 1.3 % (95% CI: 0.1% - 2.5%) per 1 µg.m-3 increase. For model 2 which included controlling for total PM2.5, to limit overestimation, hospital admissions increased significantly for Ca by 5.2 % (95% CI) (1.5% - 9.1%) for age groups 0 – 14 and by 0.1% (95% CI) (6.4% - 7.1%) for >65, respectively. Four regression models were run to determine the association with sources as determined in the PMF model. In model 1, Tapp, public holiday and month year were addressed, the respiratory hospital admissions increased significantly by 2.9 % (95% CI: 0.1% - 5.7% for re-suspended dust matrix and by 1.6 % (95% CI: 0.1% - 3.2%) for biofuel burning sources per 1 µg.m-3 increase per sources. For the City of Tshwane air pollution dataset PM10 and NO2 had a significant positive correlation (rho = 0.31, p<0.001), and PM10 and NO2 had a negative correlation with Tapp (rho = -0.18, -0.31, -0.14, p<0.001). Also, SO2 had significantly positive correlation with NO2 (rho = 0.83, p<0.001). There was no association found with respiratory hospital admissions. CONCLUSION: It can be concluded that the rise in non-communicable diseases including respiratory disorders was exacerbated by high levels of air pollutants in a 34-month study in central Pretoria. The WHO guideline which is based on epidemiological evidence is stricter and raises the alarm sooner for dangerous PM2.5 and trace elements. Legislation and policy can address sources of pollution in urban areas such as vehicle emissions and biofuel burning. Efficient and precise monitoring of ambient PM2.5 alongside national air monitoring stations is beneficial for environmental health studies. Results of the health risk assessment show that there is a considerable risk to PM2.5 and the trace elements for children and infants.