Why did we conduct this review?
Globally, outdoor air pollution is a serious public health problem. In 2016, approximately 4 million deaths were attributable to air pollution, mostly from cardiovascular and respiratory diseases. Air pollution has also been linked to other health problems, like asthma. It is of much concern both in low- and middle-income countries, where air quality may still be worsening, as well as in high-income countries, where pollution levels have decreased over several decades.
Many different policies and programmes have been put into place to reduce air pollution; examples include vehicle restrictions to reduce traffic, fuel standards for cars, buses and other motorized transport, industrial regulations to limit pollution from factories, and the replacement of inefficient heating stoves with more efficient, cleaner burning stoves. So far, no review has investigated systematically whether these measures have impacted air pollution and health as intended.
What is the aim of this review?
We investigated whether measures put into place to reduce outdoor air pollution have actually reduced air pollution and improved health.
What were the main results of this review?
We found 42 studies evaluating a broad range of measures to reduce air pollution in different countries around the world, although most were from high-income countries. Most aimed to reduce air pollution from cars and other vehicles. However, we also identified measures addressing heating and cooking, industry, or a combination of different sources.
We wanted to know whether these measures led to a reduction in the overall number of deaths, and in the number of deaths from cardiovascular and respiratory causes. We also investigated whether the measures led to fewer people going to hospitals for cardiovascular and respiratory problems. We also examined whether there were any changes in outdoor air quality, looking at different pollutants, such as particulate matter, fine particulate matter and other criteria pollutants.
Studies were very diverse with respect to the policies or programmes they assessed, the settings and contexts in which they were implemented, and the methods used to evaluate them.
The evidence we identified was of low and very low certainty, which means we cannot be very confident in the overall findings. Questions around certainty arose because of how studies were designed, conducted and analyzed. While some studies applied rigorous methods, others did not.
Overall, we observed mixed results across studies. Many studies observed no clear changes in health or air quality associated with the measures, while others did observe clear improvements. We identified very few studies that reported worsened health or air quality associated with the measures.
How do we interpret these results?
Differences in the studies make it difficult to draw general conclusions about whether the measures worked. Detecting changes in population health and air pollution levels is challenging, and assessing whether changes that occur are due to a specific measure is complex. Air pollution levels are changing constantly and often unpredictably due to weather and other factors, and other changes happening at the same time could also impact population health and air pollution. When regulations to limit industrial pollution are introduced, one must keep in mind that several other changes may be occurring in the background: an increase in traffic and an upgrade of residential heating systems, for example, or an economic downturn that leads to reduced pollution. It can sometimes take a long time before improvements in health become apparent. In interpreting the review’s findings it is important to remember that just because a study did not detect an improvement does not mean that there really was no improvement.
Further evaluations of measures to reduce outdoor air pollution in different countries, in particular in low- and middle-income countries, are needed. Wherever possible, future evaluations should apply more reliable and standardized methods to analyze the data. This should help improve the quality of individual studies as well as our confidence in the findings across studies.
How up to date is this review?
This review includes studies up to 31 August 2016; any studies that were published after that date are not included in this review.
Given the heterogeneity across interventions, outcomes, and methods, it was difficult to derive overall conclusions regarding the effectiveness of interventions in terms of improved air quality or health. Most included studies observed either no significant association in either direction or an association favouring the intervention, with little evidence that the assessed interventions might be harmful. The evidence base highlights the challenges related to establishing a causal relationship between specific air pollution interventions and outcomes. In light of these challenges, the results on effectiveness should be interpreted with caution; it is important to emphasize that lack of evidence of an association is not equivalent to evidence of no association.
We identified limited evidence for several world regions, notably Africa, the Middle East, Eastern Europe, Central Asia and Southeast Asia; decision-makers should prioritize the development and implementation of interventions in these settings. In the future, as new policies are introduced, decision-makers should consider a built-in evaluation component, which could facilitate more systematic and comprehensive evaluations. These could assess effectiveness, but also aspects of feasibility, fidelity and acceptability.
The production of higher quality and more uniform evidence would be helpful in informing decisions. Researchers should strive to sufficiently account for confounding, assess the impact of methodological decisions through the conduct and communication of sensitivity analyses, and improve the reporting of methods, and other aspects of the study, most importantly the description of the intervention and the context in which it is implemented.
Ambient air pollution is associated with a large burden of disease in both high-income countries (HICs) and low- and middle-income countries (LMICs). To date, no systematic review has assessed the effectiveness of interventions aiming to reduce ambient air pollution.
To assess the effectiveness of interventions to reduce ambient particulate matter air pollution in reducing pollutant concentrations and improving associated health outcomes.
We searched a range of electronic databases with diverse focuses, including health and biomedical research (CENTRAL, Cochrane Public Health Group Specialised Register, MEDLINE, Embase, PsycINFO), multidisciplinary research (Scopus, Science Citation Index), social sciences (Social Science Citation Index), urban planning and environment (Greenfile), and LMICs (Global Health Library regional indexes, WHOLIS). Additionally, we searched grey literature databases, multiple online trial registries, references of included studies and the contents of relevant journals in an attempt to identify unpublished and ongoing studies, and studies not identified by our search strategy. The final search date for all databases was 31 August 2016.
Eligible for inclusion were randomized and cluster randomized controlled trials, as well as several non-randomized study designs, including controlled interrupted time-series studies (cITS-EPOC), interrupted time-series studies adhering to EPOC standards (ITS-EPOC), interrupted time-series studies not adhering to EPOC standards (ITS), controlled before-after studies adhering to EPOC standards (CBA-EPOC), and controlled before-after studies not adhering to EPOC standards (CBA); these were classified as main studies. Additionally, we included uncontrolled before-after studies (UBA) as supporting studies. We included studies that evaluated interventions to reduce ambient air pollution from industrial, residential, vehicular and multiple sources, with respect to their effect on mortality, morbidity and several air pollutant concentrations. We did not restrict studies based on the population, setting or comparison.
After a calibration exercise among the author team, two authors independently assessed studies for inclusion, extracted data and assessed risk of bias. We conducted data extraction, risk of bias assessment and evidence synthesis only for main studies; we mapped supporting studies with regard to the types of intervention and setting. To assess risk of bias, we used the Graphic Appraisal Tool for Epidemiological studies (GATE) for correlation studies, as modified and employed by the Centre for Public Health Excellence at the UK National Institute for Health and Care Excellence (NICE). For each intervention category, i.e. those targeting industrial, residential, vehicular and multiple sources, we synthesized evidence narratively, as well as graphically using harvest plots.
We included 42 main studies assessing 38 unique interventions. These were heterogeneous with respect to setting; interventions were implemented in countries across the world, but most (79%) were implemented in HICs, with the remaining scattered across LMICs. Most interventions (76%) were implemented in urban or community settings.
We identified a heterogeneous mix of interventions, including those aiming to address industrial (n = 5), residential (n = 7), vehicular (n = 22), and multiple sources (n = 4). Some specific interventions, such as low emission zones and stove exchanges, were assessed by several studies, whereas others, such as a wood burning ban, were only assessed by a single study.
Most studies assessing health and air quality outcomes used routine monitoring data. Studies assessing health outcomes mostly investigated effects in the general population, while few studies assessed specific subgroups such as infants, children and the elderly. No identified studies assessed unintended or adverse effects.
The judgements regarding the risk of bias of studies were mixed. Regarding health outcomes, we appraised eight studies (47%) as having no substantial risk of bias concerns, five studies (29%) as having some risk of bias concerns, and four studies (24%) as having serious risk of bias concerns. Regarding air quality outcomes, we judged 11 studies (31%) as having no substantial risk of bias concerns, 16 studies (46%) as having some risk of bias concerns, and eight studies (23%) as having serious risk of bias concerns.
The evidence base, comprising non-randomized studies only, was of low or very low certainty for all intervention categories and primary outcomes. The narrative and graphical synthesis showed that evidence for effectiveness was mixed across the four intervention categories. For interventions targeting industrial, residential and multiple sources, a similar pattern emerged for both health and air quality outcomes, with essentially all studies observing either no clear association in either direction or a significant association favouring the intervention. The evidence base for interventions targeting vehicular sources was more heterogeneous, as a small number of studies did observe a significant association favouring the control. Overall, however, the evidence suggests that the assessed interventions do not worsen air quality or health.