Air pollution and human cognition: A systematic review and meta-analysis

This systematic review summarises and evaluates the literature investigating associations between exposure to air pollution and general population cognition, which has important implications for health, social and economic inequalities, and human productivity.

The engines MEDLINE, Embase Classic+Embase, APA PsycInfo, and SCOPUS were searched up to May 2022. Our inclusion criteria focus on the following pollutants: particulate matter, NOx, and ozone. The cognitive abilities of interest are: general/global cognition, executive function, attention, working memory, learning, memory, intelligence and IQ, reasoning, reaction times, and processing speed. The collective evidence was assessed using the NTP-OHAT framework and random-effects meta-analyses.

Eighty-six studies were identified, the results of which were generally supportive of associations between exposures and worsened cognition, but the literature was varied and sometimes contradictory. There was moderate certainty support for detrimental associations between PM_2.5 and general cognition in adults 40+, and PM_2.5, NOx, and PM₁₀ and executive function (especially working memory) in children. There was moderate certainty evidence against associations between ozone and general cognition in adults age 40+, and NOx and reasoning/IQ in children. Some associations were also supported by meta-analysis (N = 14 studies, all in adults aged 40+). A 1 μg/m³ increase in NO₂ was associated with reduced performance on general cognitive batteries (β = -0.02, p < 0.05) as was a 1 μg/m³ increase in PM_2.5 exposure (β = -0.02, p < 0.05). A 1μgm³ increase in PM_2.5 was significantly associated with lower verbal fluency by -0.05 words (p = 0.01) and a decrease in executive function task performance of -0.02 points (p < 0.001).

Evidence was found in support of some exposure-outcome associations, however more good quality research is required, particularly with older teenagers and young adults (14-40 years), using multi-exposure modelling, incorporating mechanistic investigation, and in South America, Africa, South Asia and Australasia.