A global review of air quality and public health research: geo-temporal patterns, methods and issues

Air quality and its effects on public health have long been a serious global concern, with pollutant sources, levels, and transport increasingly becoming subjects of research and public interest. As the amount of multi-source air quality datasets grows, research efforts have also increased, showing varying degrees of success, challenges, and coverage across different regions worldwide. Therefore, a study that synthesises this knowledge is urgently required for multiple planning and monitoring purposes. Here, we present a systematic review of studies on air quality and their potential impact on public health, sourced from the Scopus, Web of Science and PubMed databases. Our search identified 123 peer-reviewed articles that met the search criteria between 2005 and 2026, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The review identified an increasing trend in articles, with most studies spatially concentrated in Asia (53), where China (23) and Iran (9) account for the highest numbers. Africa follows with 23 articles, with South Africa (9) having the most. North America has 20 articles, led by the United States (17). Europe has few articles (10), and South America (5), and no more than four articles for other countries. About half of these studies (50.4%) used ground-based stations, while almost one-fifth (20.7%) used MODIS satellite data, with growing studies using various sensors coupled with machine learning methods. The most studied pollutants include PM 2.5 (66 articles), NO 2 (38), PM 10 (36), SO 2 and O 3 (28 each), CO (13), and relatively few examined NO X (4) and BC (1). The results revealed that relatively few articles were published within the Remote Sensing and Geoinformation journal category. This finding further underscores a significant research gap at the intersection of remote sensing and air quality studies. Air quality monitoring across sub-Saharan Africa continues to be constrained by sparse ground-based monitoring networks and a limited number of studies that integrate satellite-derived and in situ observations. Consequently, there is a pressing need for more region-specific research to enhance pollution assessment, improve exposure characterization, and advance understanding of air quality dynamics within the regional context. Overall, studies successfully associated poor indoor and outdoor air quality with adverse effects on human health such as respiratory disorders, hospital admission and mortality to both children and adults.

Indoor and outdoor air quality is a serious global challenge which affects, inter alia, human health and cause complicated sicknesses which ultimately lead to death. A report by the State of Global Air (SoGA), (2024) revealed that poor air quality is ranked as the second leading cause of death to children globally, following malnutrition. The World Health Organization (WHO), through the 2012 Global Burden of Disease (GBD) report indicated that ambient air pollution was estimated to have contributed to approximately 3 million premature deaths globally during that year, highlighting the disproportionate vulnerability of children compared to other population groups (1; https://www.healthdata.org/research-analysis/health-topics/air-pollution-research-library). Children are particularly vulnerable to air pollution due to their developing respiratory systems and increased exposure. Air quality levels are influenced by multiple factors, including meteorological conditions, race, sex, economic status, vulnerability, age, and underlying health issues (2–9). Additionally, land use activities, urban vs. rural settings, and socio-economic factors play a crucial role in determining pollutant concentration and their health impacts (10, 11). The impact of indoor air pollution is equally concerning, with respiratory diseases, chronic obstructive pulmonary disease (COPD) risk, and adverse pregnancy outcomes being prevalent in exposed populations. Vulnerable groups, including women, children, and the elderly, are disproportionately affected, necessitating urgent action for cleaner, healthier environments (12). However, the vast majority of exposure to air pollutants occurs indoors.

As population growth continues, especially in urban centres, it drives increased infrastructure development, energy demand, and industrial activity which leads to higher emissions from vehicles, factories, and power generation facilities, contributing to elevated pollutant concentrations and wider spatial dispersion. Consequently, densely populated urban areas often experience significantly poorer air quality and greater associated health risks (13–15). Various datasets including ground-based observations and satellite-derived datasets can be used to investigate air quality at local, regional, national, and global scales. These datasets enable the assessment of key atmospheric pollutants, including nitrogen dioxide (NO 2), sulphur dioxide (SO 2), particulate matter (PM 2.5 and PM 10), and carbon monoxide (CO), providing comprehensive insights into their spatial and temporal variability. In parallel, spatial analysis methods have been increasingly applied in studies exploring air quality and its impact on human health. These methods ranging from Geographic Information System (GIS) based mapping and spatial interpolation to advanced geostatistical modeling and remote sensing techniques play a crucial role in identifying pollution hotspots, understanding exposure patterns, and assessing health outcomes across different populations and regions.

Given the growing recognition of air quality as a critical environmental and public health concern, reflected in the increasing volume of research worldwide, there is a need to systematically synthesize and evaluate the patterns, trends, and contributions of these studies. Such an assessment requires consideration of not only the scientific and public health contexts but also the evolution of methodologies, technological advancements, and their applications across local, regional, and global scales. Within this framework, the present study explores how insights from the global body of literature can inform and advance air quality research in South Africa and the broader Sub-Saharan African region. Specifically, it investigates how varying levels and characteristics of pollutant exposure influence public health outcomes and examines the extent to which existing research contributes to our understanding of spatial and temporal trends, vulnerable populations and regions, methodological developments, and persistent knowledge gaps within the air quality literature.

Literature search and data extraction

We conducted a literature search (Web of Science, Scopus and PubMed) of peer-reviewed articles that investigated air quality and public health from 2000 to 2026. The study followed a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. The search included the following keywords: “air quality”, “air pollution”, “human health”, “respiratory diseases”, “satellite imagery”, “ground monitoring”, “spatial analysis”, “GIS”, “machine learning”, and yielded a total of 3,320 articles (Table 1). All articles were vetted as illustrated in Figure 1, and 123 articles met the criteria and included in the analysis. We excluded articles that (a) did not focus on both key pollutants and human health, (b) studies published before 2005, (c) was published in other languages than English. The number of literatures reviewed and included in this study was 123 in total, 101 articles and 12 reports (see Figure 1).

Data analysis

The selected studies were analyzed using a structured narrative synthesis approach with thematic categorization to identify key trends, research gaps, and emerging themes in air quality and associated health impact research. While this review adopts a global perspective to capture broader trends and developments in air quality and public health research, particular attention is given to South Africa and the wider Sub-Saharan African region. This additional focus is motivated by the relatively limited representation of studies from these regions in the existing literature. Consequently, global findings are critically examined and contextualized within the South African and Sub-Saharan African settings to identify region-specific challenges, research gaps, and opportunities for advancing air quality monitoring, exposure assessment, and public health research.

Bibliometric characteristics and article trends

The results show an increasing trend in the number of articles over time, with more studies conducted in 2022 (16) than in the entire period (Figure 2). Fewer than 2 studies per year were published between the years 2005 and 2014. From 2015 onward, articles increased, with more than 3 articles published annually until 2018. It was also observed that the majority of studies (n = 18) conducted before 2019 relied on ground-based observations from air quality monitoring stations, with spatial analysis techniques such as interpolation used to estimate pollutant concentrations in areas lacking monitoring coverage. This growth was followed by a decline in 2019, where only 4 studies were published. Figure 2 further showed a significant rise in articles observed between 2020 and 2022, with 2022 recording the highest number of articles of 16 articles. This increase coincides with the COVID-19 pandemic, during which there was heightened interest in air quality and its relationship to human health, particularly due to changes in human activities and increased focus on respiratory outcomes. Furthermore, the finding of the study shows that more studies focused on hospital admission and mortality cases as a result of poor air quality and most studies used surveys to collect the data. Another decline in article was noticed in 2023, this decline may be attributed to the subsiding effects of the COVID-19 pandemic, delays in article processes, and restricted access to articles; however, an upward trend re-emerged in 2024 and 2025, indicating continued relevance of the research topic. The results further indicates that the use of machine learning and satellite-based observation such as MODIS AOD and Sentinel-5P gained momentum from the year 2020 to the year 2025 (Table 2). The results indicate that although machine learning approaches have been increasingly adopted in recent studies, traditional methods such as regression continue to be widely used and remain relevant across many studies (Table 2). Overall, the analysis showed that regression was the most commonly used method, followed by machine learning and correlation-based approaches.

The distribution of articles across journal categories indicates a strong emphasis on air quality focused research. Air Quality and Atmospheric Environment journals accounted for the highest number (56 articles) of articles in the literature (see Figure 3), reflecting the central role of atmospheric science in studies related to air pollution. This was followed by journals under the human health impact and epidemiology, which has 31 articles and this makes it the second highest publishing journals. This highlights the growing interest in the health implications of air quality. The study observed that fewer studies were published in journal relating to sustainability and energy (11) which suggest that there is a relatively limited integration of air quality research within broader sustainability and energy studies. The Remote Sensing and Geoinformation category recorded the lowest number of articles (6) as indicated in Figure 3, indicating underrepresentation of spatial, satellite-based, and geospatial approaches in the existing literature. Overall, the results suggest that while air quality and health-related disciplines dominate the research field, there is a notable gap in studies leveraging remote sensing technologies and sustainability-focused perspectives.

The distribution of articles per country shows that a small number of countries contribute a large share of the total articles. China emerged as the leading contributor indicating the highest number of published articles of 23 as indicated on Figure 4. This was followed by United States of America (n = 17), indicating strong research activities concentration. Iran and South Africa contributed an equal number of articles (n = 9), followed by India (n = 5). Most of the remaining countries recorded only one or two articles, reflecting broader global participation but lower individual research output. Most studies conducted in China, Iran, the United States of America, South Africa, and India focused on the use of ground-based observations and satellite imagery to assess air quality and its impacts on human health, hospital admissions, and mortality related to air pollution. Furthermore, the number of countries without any published studies is substantially higher than the number of countries with at least one article. Continentally, Asia leads in articles, followed by North America, while Africa shows a moderate contribution. Europe has a smaller but diverse output, South America shows limited activity, and Australia has the lowest contribution, highlighting regional differences in research focus and capacity (Figure 4).

Tools and data used

Ground stations data and devices are the most commonly used data sources appearing in 51 studies, which highlights their key role in air quality research (Figure 5). Among satellite products, MAIAC AOD from MODIS is widely used (21 studies) and reflect its importance for aerosol monitoring. Sentinel-5P TROPOMI and other satellite or reanalysis datasets such as MISR, MERRA-2, VIIRS, and SeaWiFS were used less frequently, indicating growing but limited adoption. Models and specialized tools like CAMS, CALIPSO, SILAM, RegCM-CHEM, and CMAQ with Downscaler are used sparingly, suggesting emerging applications.

Current global and South African air quality trends: impacts of poor air quality indoor and outdoor

To minimize the impacts of poor air quality, coordinated efforts are required at multiple levels, ranging from individuals and households to national governments. In support of these efforts, various standards and guidelines have been developed at both global and national levels to reduce air pollution and its associated burden on human health, addressing sources from household emissions to ambient air pollution. Notably, the World Health Organization (WHO) has established global air quality guidelines that different countries can adopt or adapt to support air quality management and public health protection (83). In line with these guidelines, the South African government has pronounced its commitment to address air quality issues through the National Environment Management: Air Quality Act (NEMAQA; No. 39 of 2004). Despite the existence of these guidelines, laws, and policies, air pollution remains to be a significant public health challenge globally. Many lives continue to be lost to air quality related illnesses, and communities experience complex and chronic health conditions resulting from prolonged exposure to poor air quality (61). For instance, in 2005, it was found that 69% of the global population were exposed to higher concentration of PM 2.5 which was more than WHO air quality guideline set at that time (9) and 90% of the global population exposed to higher concentration of PM 2.5 in 2019 (83). Studies show that PM 2.5 concentrations in many countries exceeded the respective national air quality standards during the assessment period (21, 38, 56).

South Africa was ranked among the top global emitters, ranking 12th in 2010, 14th in 2018, and 15th in 2023 (84, 85). Although CO 2 emissions declined from about 463 million metric tons in 2010 to 401 million in 2023, levels remain high and continue to exceed minimum air quality standards (86). Additionally, Southern Africa is a hotspot for pollutants such as NO 2, VOCs, and SO 2, largely driven by the energy sector (87). Black carbon (BC) is mainly emitted from residential fuel burning, making household air pollution a major concern (87). In 2019, household air pollution caused 697,000 deaths in Africa which is more than the 394,000 deaths from ambient air pollution (88). Studies have shown rising levels of SO 2 and NO 2 in Mpumalanga, Limpopo, and Gauteng, with these areas identified as pollution hotspots largely due to multiple coal-fired power stations (55, 87). Within Gauteng, the City of Johannesburg, Ekurhuleni and some parts of the City of Tshwane have the highest concentrations of these pollutants (24). While the highest pollution levels occur near direct emission sources, Mahesh et al. (60) found that in Durban, South Africa, areas without local sources were affected by particulate matter transported from surrounding regions.

Spatial variability of pollutants

The spatial distribution of pollutants is presented in Figure 6, which shows the quantification different pollutants published in countries worldwide. The majority of articles are concentrated in Asia (55), followed by Africa (23), North America (20), Europe (10) and relatively few in South America (5). Several studies have investigated PM 2.5 levels (66 articles), mostly in China and several other countries; where it has been shown to peak in the morning and evening due to household activities and traffic emissions in countries such as Nigeria, Cote ^ d'Ivoire and Ghana (89, 90). Higher concentrations of NO 2 (38) and PM 10 (36), SO 2 and O 3 (28) were reported across different countries, while Nox and BC had relatively few publications. Pollutants such as NO 2, PM 10 and PM 2.5 were also observed in winter, as well as during weekdays, nights, and evenings (4). Studies in Equatorial Asia and Central Mexico observed higher PM 10 levels in summer (13, 91). SO 2 and CO levels typically increase in winter and decrease in summer in Kenitra city Morocco (68). Furthermore, CO and NO 2 peaked during rush hours, in the morning and evening driven primarily by local traffic emissions (68). Similar winter increases in SO 2 have been reported in multiple regions (17, 27, 91, 92). Additionally, PM 2.5 levels are higher in the dry season than in the wet season in some parts of West Africa (89, 90).

In South Africa, CO, NO 2, and SO 2 concentrations were highest in winter, especially in June, and lowest in autumn, with March showing the cleanest air quality (93). NO 2 levels increased in winter and remained elevated in spring (93). Mpumalanga, Limpopo, and Gauteng were identified as major pollution hotspots for CO, NO 2, and SO 2 (55), with high PM 10 levels reported in areas such as Johannesburg, the Vaal Triangle, and Highveld towns (31). PM 10 concentrations were highest in Gauteng and Mpumalanga compared to Western Cape and KwaZulu-Natal (31). Olivenhoutbosch area in Gauteng province recorded the highest PM 2.5 concentrations in winter than in summer in 2017 (30). In this study, PM 2.5, PM 10, SO 2, and NO 2 emerged as the major pollutants studied, with China having the highest number of studies across all these pollutants (17, 14, 13, and 11, respectively). In contrast, South Africa had a limited number of studies on PM 2.5, PM 10, NO 2, and SO 2 (4, 3, 4 and 4, respectively) (Figure 6). The study further shows that CO, ozone (O 3), NO x, and BC were the least studied pollutants across different countries. China still had the highest number of studies for CO and O 3 (6 and 8, respectively). NO x and BC showed the largest gap and were the least studied pollutants overall, with a maximum of 2 and 1 studies, respectively (Figure 6).

Air quality impact on human health

PM 2.5 is a major global health risk and was ranked among the top 5 causes of air pollution-related mortality worldwide, contributing to over 4 million deaths in 2015 (94). Both short and long-term exposure to pollutants such as PM 10, PM 2.5, and NO 2 are strongly linked to increased hospital admissions and risks of cardiovascular and respiratory diseases, including asthma (21, 42, 48, 50, 51, 56–58, 73, 77). Long-term PM 2.5 exposure is also associated with higher prevalence of respiratory infections in children, reduced life expectancy, and increased infant mortality in sub-Saharan Africa (49, 59, 95). Additionally, prenatal exposure to air pollution has been linked to adverse developmental outcomes, including a higher risk of autism spectrum disorder (67). Indoor and outdoor air pollution is not just a respiratory problem; it also affects the cardiovascular system, the heart and circulation, and mental health.

Exposure to PM 10, SO 2, and NO 2 resulted in 28 057 respiratory hospital admissions in Lanzhou, China, from 2001 to 2005 (43). Globally, poor air quality has been linked to mortality (44, 45) and premature deaths (69, 76, 77, 96). In Poland, PM 2.5 exposure was associated with more premature deaths than NO 2 (69). In 2005, high PM 2. 5 levels were estimated to cause about 2 million premature cardiovascular deaths and 186,000 lung cancer deaths globally (9). Higher CO 2 and PM 2.5 concentrations are linked to increased lung cancer prevalence in Southeast Asia (82). Long-term exposure to PM 2.5 and NO 2 is associated with increased cancer mortality in the United States, particularly among vulnerable populations (97). Additional evidence showed that PM 2.5 related mortality were observed across all 50 states, Washington DC, and Puerto Rico (6).

PM 2.5 is a major air pollutant linked to significant health impacts, including premature deaths, respiratory diseases, and hospital admissions. In Poland, it caused more premature deaths than PM 10 and NO 2, with 3,222 deaths reported between 2015 and 2020 (71). Several studies have found it to be the primary pollutant responsible for respiratory illness, mortality (15, 49, 65, 71, 78, 96, 98), and hospital admissions (58), particularly during winter when concentrations are higher (58). Globally, PM 2.5 accounted for over 90% of air pollution-related deaths in 2021 (99) and caused an estimated 8,134 annual adult deaths in 21 cities in Kazakhstan (100). Long-term exposure to PM 2.5 is associated with asthma prevalence, increased hospitalisations, and premature mortality (53, 54, 62). Furthermore, exposure to PM 2.5, PM 10, and NO 2 increased adult asthma risk (101). In Tehran, PM 10 was observed to cause greater health impacts than SO 2 and NO 2 (91). PM 10 and NO 2 caused most respiratory and cardiovascular hospital admissions in Ho Chi Minh City (46).

This observation indicates that high concentration of NO 2 can lead to respiratory related sickness and death (65). South Africa has experienced a rise in the number of deaths by air quality related illness, an estimated increase in the number of deaths in South Africa was observed between the years 2000, 2006 and 2012 as a result of exposure to ambient air pollution attributed to PM 2.5 (80). Of all the death cases reported in all the provinces in South Africa, Gauteng was found to have high estimated number of deaths caused by PM 2.5 and O 3 in 2012 (80). Furthermore, 80% of deaths cases related to Asthma were reported from South Africa (102). The study further found that South Africa was ranked as the 5th country to have the largest asthma mortality globally (102). The main gases responsible for causing public health concerns were identified to be PM, CO, O 3 NO 2 and SO 2 (103). The Global Burden of Disease report further found more deaths cases by air pollution in low- and middle-income countries including South Africa (1).

Trends of spatial analysis methods

Studies on air quality and health impacts increasingly integrates ground-based and satellite data with methods such as surveys, spatial analysis, GIS mapping, geostatistics, and remote sensing to identify pollution hotspots, exposure patterns, and health outcomes. Surveys help assess public perceptions and health effects as shown by studies in Kansas City (52), India linking pollution to respiratory symptoms (40), and Italy examining asthma and rhinitis (62). The integration of survey data with satellite-derived observations enabled the estimation of long-term PM 2.5 exposure levels for children by assigning satellite-modelled monthly average concentrations to survey clusters (59).

Recent advances in machine learning have demonstrated strong reliability in estimating and forecasting air quality, particularly when integrated with remote sensing data to assess pollutant concentrations, health impacts, and future trends. These methods improve spatial estimation of pollutants such as PM 2.5 and NO 2 (36) and effectively model health outcomes like asthma incidence, with Random Forest (RF) frequently outperforming models such as Xtreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and Multiple Linear Regression (MLR) (39, 62). Machine learning has also been used to analyse spatiotemporal pollution patterns and identify disease hotspots, with PM 2.5 often showing the strongest association with asthma occurrence (28, 29). In forecasting, models such as RF, Gradient Boosting (GB), Support Vector Regression (SVR), and XGBoost have shown strong predictive performance, sometimes outperforming alternative approaches (26, 32–34, 37). Comparative studies further highlight the effectiveness of machine learning, with XGBoost outperforming statistical and deep learning models, as well as RF and SVR in some cases (31, 104). Machine learning has demonstrated strong performance even in data-limited regions such as South Africa's Highveld (30). Additionally, regression models were used to examine association between environmental exposures and health outcomes, including respiratory and cardiovascular diseases, hospitalisation, and mortality, while capturing temporal trends and complex effects (3, 40, 41, 51, 57, 63, 64, 67, 75).

Spatial correlation methods are used to assess relationships between meteorological variables, climate and air quality, and its impact on human health. For example, Mitra and Czajkowski (79) used Moran's I to analyse PM 2.5 in Bangladesh, identifying a clustered spatial pattern with positive spatial autocorrelation. A similar pattern was observed in China by Zeng et al. (19). Pearson correlation has also been used to examine associations between PM 2.5, hospital admissions, and meteorological variables (17, 20). In South Africa, Global Moran's I showed significant clustering of Early Childhood Development (ECD) centres near NO 2 and SO 2 pollution sources, indicating proximity to pollution hotspots (24). Moran's I was used to assess spatial autocorrelation between air pollutants (SO 2, NO 2, CO, and PM) and asthma emergency visits, revealing clear spatial patterns in health outcomes (23).

Ground monitoring stations provide reliable data on pollutant concentrations and spatial distribution; however, their limited coverage restricts their ability to capture full spatial variability, resulting in data gaps, especially in countries like South Africa where ground monitoring stations are concentrated in priority areas (14, 115). Despite this limitation, ground-based data have been widely used to successfully estimate pollutants such as PM 2.5 (33, 78, 104), NO 2 (26, 36, 51, 56, 71), and PM 10 (25, 28), producing reliable results. Ground observed data have also been applied to assess health impacts, including respiratory hospitalisations in China (43) and associations between PM 2.5 exposure and health outcomes in the United States (52). To address spatial data gaps, studies often use GIS-based interpolation methods such as Inverse Distance Weighting (IDW), Kriging, and Voronoi Neighbourhood Averaging (VNA) (13, 14, 28, 34, 72). For example, in China, VNA outperformed IDW and Kriging in simulating PM 2.5 exposure and effectively captured concentration levels (72).

Spaceborne technologies offer broad spatial coverage, including inaccessible areas, and capture spatial variability at high resolution (6). Commonly used satellite data include Moderate Resolution Imaging Spectroradiometer (MODIS) derived Aerosol Optical Depth (AOD) and Fine AOD (FAOD) (13), as well as Sentinel-5P and VIIRS, all of which provide high spatial and temporal resolution air quality information. MODIS has a resolution of 250 m to 1 km and a temporal resolution of 1–2 days and has 36-band data and is onboard of Terra which was launched in 1999 and Aqua launched later in 2002. Sentinel-5 Precursor satellite was launched on 13 October 2017 and has the capability of collecting high spatial resolution (3.5 km × 7 km (trace gases) and 7 km × 7 km (CO and CH 4) air quality information with a temporal resolution of less than a day. VIIRS has a spatial resolution of 370–750 m and a temporal resolution of less than a day and is onboard of Suomi National Polar-orbiting Partnership (SUOMI NPP) and National Oceanic and Atmospheric Administration (NOAA) offering 22-band data. MODIS-derived AOD has been widely applied to estimate pollutants such as PM 2.5 and PM 10 (19, 25, 30, 31, 37, 53, 59, 63, 79, 95, 105), while FAOD has been shown to improve PM 2.5 estimation accuracy (38). Sentinel-5P TROPOMI data have been used to map pollutant distributions and air quality indices in countries such as South Africa and Senegal (24, 75, 93). Additionally, integrating multiple satellite datasets, including MODIS, MISR, and SeaWiFS, enhances the accuracy of pollutant estimates like PM 2.5 (49, 105, 106). For example, Sentinel-5P data combined with meteorological and land-use variables in a RF model has been used to estimate daily surface-level NO 2 concentrations in East Asia (27).

Satellite-derived observations are often compared with ground-based measurements to evaluate their reliability in representing near-surface pollutant levels and to assess temporal trends, spatial patterns, and potential biases. These comparisons support the use of satellite data as a complement to ground monitoring. For example, a positive correlation between satellite-derived and ground-based estimates was found by Amiri and Shahne (37), while Shikwambana et al. (55) reported that satellite observations closely matched ground-based trends of SO 2 and NO 2 in Mpumalanga, South Africa. Consistent findings across large-scale studies, such as Yu et al. (63), further confirm agreement between observed and modelled PM 2.5 exposure and mortality. Integrating satellite and ground-based data has been shown to improve the accuracy of pollutant estimates (25, 31, 34, 35, 53, 60, 104). For instance, combined remote sensing and ground-based datasets revealed strong links between PM 2.5 exposure and mortality across the United States (6), while in South Korea, integration improved PM 2.5 and NO 2 estimates by reducing errors and better capturing spatial variability, especially in areas with limited monitoring (36).

Article trend

In recent years, studies have increasingly been published on air quality and human health, with the latest cohort adopting machine learning approaches, particularly tree-based models such as RF, due to their strong predictive performance, flexibility, and robustness in handling complex datasets. This coincides with the period from 2015 when several satellite-based air quality sensors such as CAMS, Sentinel-5 were launched. In environmental health research, machine learning is especially advantageous as it can capture non-linear relationships and interactions between air pollutants and health outcomes. It has the capacity to integrate multiple heterogeneous data sources and to model and predict associations between air pollution and health outcomes. Despite these advantages, traditional methods such as regression remain widely applied and continue to demonstrate their relevance particularly for trend analysis, exposure assessment, and in understanding the relationship between different variables. The study further revealed a significant gap in pollutants such as BC and Nox worldwide, in sub–Saharan Africa, gaps were identified across all the pollutants studied (PM 2.5, PM 10, SO 2, NO 2, CO, O 3, NO x and BC). The overall findings indicate that Africa has a moderate level of research output led by South Africa; however, studies focusing on townships and informal settlements where air pollution exposure may be high remain limited in this region. Filling these gaps is essential to better understand air quality and its health impact across sub-Saharan Africa.

Spatial and seasonal variability

The study demonstrated that both annual and daily pollutant concentrations remain elevated across many countries, despite ongoing global and regional mitigation efforts. This persistent trend underscores the need for strengthening existing air quality legislations and policy frameworks. Among the pollutants examined, PM 2.5 has consistently been reported to exceed the WHO daily and annual set standard as reported in several previous studies (2, 20, 35, 61, 65, 66, 69–71, 73, 78, 89, 90, 95, 106–108). Within the South African context, PM concentrations were similarly found to be persistently high, frequently surpassing the national ambient air quality standards, particularly in Mpumalanga Province (61). In addition, Gauteng, Mpumalanga, and Limpopo provinces have been identified as pollution hotspots for NO 2 and SO 2 (24, 55). These elevated levels are largely attributable to emissions from coal-fired power stations and other industrial activities concentrated in these regions. The overall poor air quality observed in South Africa can be largely linked to the country's carbon-intensive energy profile, with approximately 90% of energy production reliant on fossil fuels. This persistent reliance on fossil fuels continues to hinder progress toward achieving sustained improvements in air quality, as stipulated by NEMAQA 39 of 2004.

The study revealed notable variability of different pollutants across different seasons and among countries. PM 2.5 concentrations was observed to be higher in winter and dry season and lower in summer and wet season in different countries including Delhi, China, Ulaanbaatar, South Korea, southern Poland, north-central France, northern Italy, Bangladesh, India, Morocco and South Africa (2, 4, 5, 10, 13, 16–18, 26, 30, 31, 35, 38, 60, 64, 68, 79, 81, 91). Higher observation of PM 2.5 may be linked to reduced atmospheric mixing, lower temperatures, and limited precipitation which reduce pollutant dispersion and removal, thereby allowing PM 2.5 to accumulate in the atmosphere. SO 2 was also observed to be higher in winter season which may be attributed to increased emissions and reduced atmospheric dispersion (17, 27, 91, 92).

Air quality and health impact

Studies have shown that women, children, elderly and those with underlying sicknesses are more vulnerable to air quality related sicknesses (7, 9). Among the pollutants assessed, evidence indicates that PM 10, PM 2.5, SO 2, CO 2, and NO 2 are major contributors to adverse health outcomes, with most studies identifying PM 2.5 as the primary driver of respiratory effects, including increased symptoms, hospital admissions, and mortality (6, 9, 15, 49, 53, 54, 58, 59, 62, 65, 69, 71, 78, 80, 94–96, 98, 100). Evidence further indicates a clear association between air quality and the prevalence of asthma symptoms, hospital admissions, and mortality (8, 22, 23, 47, 54, 62, 66, 101, 102, 109, 110). These findings underscore the substantial public health burden posed by ambient and indoor air pollution, particularly PM.

Vulnerable populations and impact of air pollution on human health

The findings indicate that air pollution arises from a range of economic and non-economic sources operating across both urban and rural environments. Major contributors include emissions from industrial activities such as power generation, refining, and mining, which release substantial quantities of gaseous and particulate pollutants into the atmosphere. In urban settings, transportation emerges as a dominant source, driven by fuel combustion and traffic congestion. Additionally, household cooking and heating practices contribute significantly to local pollution levels, particularly in densely populated areas. Furthermore, household air pollution emerged as a critical concern, studies have shown that it can be more harmful to human health than outdoor air pollution, particularly in regions where solid fuels are used for cooking and heating (88). Agricultural activities contribute through the burning of crop residues, use of fertilizers, and livestock emissions, whereas natural sources such as wildfires can episodically release substantial quantities of pollutants over an area. The combined influence of these sources plays a critical role in determining air quality levels and their variability across space and time.

The results indicate that in low-income settings, women's domestic roles, particularly cooking and heating, increase exposure to indoor air pollutants (111). Women and children in rural areas were most affected due to reliance on solid fuels emitting pollutants such as PM, VOCs, NO 2, CO, and SO 2 (5); consistent with this, the use of unclean fuels in Ghana was associated with a 73% higher risk of child mortality (105). Air pollution exposure was also linked to asthma outcomes, with NO 2 increasing hospital admissions in Berlin, particularly among women (109), while higher childhood asthma visits were observed among males and those living near coal-fired plants in Edmonton (47). Children exhibit increased vulnerability to air pollution, with a higher prevalence of asthma observed in densely populated urban environments characterized by elevated concentrations of PM 2.5 (8, 22, 25, 110). Additionally, higher pollution levels in urban and industrial areas were linked to increased respiratory hospital visits (27, 41, 106, 112). Although contrasting evidence from Nigeria indicates higher PM 2.5 concentrations in less urbanized areas due to local emission sources (90), which highlight that health impacts vary across demographic and geographic contexts.

In different study locations, air pollution exposure and associated health outcomes vary by age group and geographic context. In South Africa, children under 5 years attending ECD centres within 5 km of pollution sources were exposed to higher NO 2 and SO 2 levels than those farther away (24). In China, children (0–14 years) and adults over 60 were more affected by respiratory diseases linked to poor air quality, with PM 2.5 and NO 2 contributing to asthma and morbidity in children in China and California (42, 66, 113). In Vietnam, people aged 5–65 years had higher risks of respiratory and cardiovascular hospitalizations than those over 65 (46). Studies in Buffalo and Bangkok reported greater risk among adults ≥65 years, likely due to differences in exposure and activity patterns (51, 64). In South Korea and Finland, SO 2, NO 2, CO, and PM 10 were associated with increased asthma visits and greater vulnerability in adults over 30 years of age (23, 96). In China and Senegal, children were highly exposed to PM 10, PM 2.5, SO 2, and NO 2, leading to acute respiratory infections and hospital admissions, with over 99% of children in China living in high-pollution areas (48, 50, 66, 75). Overall, Children are more vulnerable to asthma because their lungs are still developing and PM can easily penetrate and damage the lungs (22).

In Kenya, prolonged PM 2.5 exposure was linked to Acute Respiratory Infection (ARI) symptoms in children, with higher prevalence in males than females (59). In Ahvaz, Iran, population under 60 years experienced high asthma-related hospital admissions (54). Similarly, adults aged 18–64 in Denmark living near highways and busy traffic areas were exposed to higher NO 2 and PM 2.5 levels, leading to respiratory infections and hospitalization (74). In Kanpur, emissions from industries, household fuel use, and vehicles increased SO 2, PM, and NOx exposure, increasing respiratory illness risk among nearby residents (41). This is consistent with the evidence that populations near major roads and traffic zones are most vulnerable (40, 78). Higher asthma rates were also observed within 0.5 km away from the main roads, and PM 2.5 remained elevated up to 1 km (16, 40). In contrast, some studies reported higher male vulnerability to air pollution related hospital admissions and chronic obstructive pulmonary disease (COPD), linked to smoking patterns and occupational exposure (43, 57, 109). Black communities have also been shown to face higher exposure to these pollutants and their impact than white communities (3, 6, 97). Overall, these studies collectively highlight that younger populations, particularly children, adults over 65 years and populations living near pollution sources are most vulnerable to air pollution related health effects.

Methodological quality and risk of bias

The methodological quality and risk of bias of the included studies were assessed using the Risk of Bias Assessment Instrument for Systematic Reviews Informing the WHO Global Air Quality Guidelines (114). This evaluates potential sources of bias across key domains, including study design, exposure assessment, outcome measurement, confounding, statistical analysis, selective reporting, and overall study quality. Given the diversity of study designs in this review, ranging from epidemiological investigations and spatial analyses to remote sensing and machine-learning-based studies, the assessment identified methodological strengths and limitations that could influence the interpretation of findings. Most studies showed a low to moderate risk of bias, particularly those using validated exposure datasets, robust statistical methods, and well-defined health outcomes. However, common sources of potential bias included limited ground-monitoring coverage, uncertainty in satellite-derived exposure estimates, incomplete adjustment for confounding factors, and variability in health data quality across regions. The risk-of-bias assessment was considered when interpreting results and identifying research gaps and future research priorities.

This review presents global research on air quality and public health published between 2005 and 2026, and reveals a considerable rise in scholarly attention on this subject over time. Most studies made use of ground-based station data, while satellite-based estimates have risen recently; these incorporated machine learning which provide valuable insights into the spatial and temporal quantification and patterns of air pollutants. The results demonstrate that air pollution remains a major environmental and public health problem worldwide, with research activity concentrated primarily in Asia, North America, and selected African countries, particularly South Africa. The review showed that PM 2.5 is the widely studied pollutant, consistently associated with adverse health outcomes, including respiratory and cardiovascular conditions leading to reduced life expectancy. Other pollutants, including PM 10, NO 2, SO 2, CO, and O 3, were also linked to significant health impacts, while NOx and BC remain relatively less reported. Levels of these pollutants were elevated during winter and dry seasons, reflecting the combined influence of meteorological conditions as well as household, transport and industrial activities. Results further show that children, women, individuals with pre-existing health conditions, and populations living near industrial areas, major roads, and other pollution sources are particularly vulnerable to the effects of poor air quality. The growing use of remote sensing data to for air quality monitoring is supported by articles published in the Remote Sensing and Geoinformation journal category. These articles further highlight the research gap in remote sensing and air quality literature, as air quality monitoring in data-poor regions such as sub-Saharan Africa remains limited by sparse ground monitoring networks and few published studies integrating satellite and in-situ observations. This underscores the need for more region-focused research to improve pollution assessment and understanding of exposure within a regional context. Moreover, several studies focused on a larger area coverage rather than smaller rapidly growing areas like townships. The limitation of the study was that some articles were written in other languages and there were access restrictions to certain peer-reviewed articles which resulted in the exclusion of some potentially relevant peer-reviewed studies. Taken together, the reviewed studies greatly contributed to improved accuracy of air quality estimates, while helping to reduce bias and provide directions on future focus of research especially in sub-Saharan Africa.