Table 3 Characteristics of included systematic reviews (sorted by publication year)
Author, year, synthesis type | PEOS of review | Number and type of studies included | Main findings | Risk of bias / quality assessment of individual studies | Risk of bias findings | Body of evidence assessment method | Body of evidence assessment findings |
|---|---|---|---|---|---|---|---|
Suades-González, E., et al. (2015) [68], Systematic review | Population: Children 0–18 years old Exposure: Outdoor air pollution during pregnancy, around birth or during childhood, using direct or indirect assessment methods Outcome: Neuropsychological development, incl. cognition, behavior, neurodevelopmental disorders, psychomotor outcomes Studies: Cohort, case–control, or cross-sectional design, “original research articles”. Published after 2012 in English language | 20 cohort, 6 case–control, and 6 cross-sectional studies | Pre- or postnatal exposure to PAH was associated with a lower global IQ, pre- or postnatal exposure to PM2.5 was associated with an increased risk of ASD, and prenatal exposure to NOx was associated with higher risk of ASD. All other associations showed mixed findings | Own criteria: study design, sample size, exposure and outcome assessment, confounder control | 25 out of 32 included studies were rated as a “good quality studies” | Modified IARC (2006) – authors differentiate between “inadequate” and “insufficient” evidence based on whether or not studies “reported an association” | Pre- or postnatal exposure: PAH and global IQ, PM2.5 and ASD: Sufficient evidence Prenatal exposure to NOx and ASD: Limited evidence All other associations: Inadequate or insufficient evidence, due to few studies, low quality, or inconsistency |
Lam, J., et al. (2016) [5], Systematic review and meta-analysis | Population: Humans Exposure: Indoor/ outdoor air pollution exposure during any developmental period (maternal or paternal, in proximity to conception, during pregnancy, or childhood), prior to outcome assessment Outcome: Any clinical diagnosis or other continuous or dichotomous scale assessment of ASD, based on ICD 9/ 10, or DSM 4/ 5 criteria Studies: Original data | 17 case–control studies, 4 ecological, 2 cohort studies | Per 10 μg/m3 increase in PM10: (SOR: 1.07, 95% CI: 1.06–1.08, n = 6 studies) and PM2.5 (SOR: 2.32, 95% CI: 2.15- 2.51, n = 3 studies) All included studies generally showed increased risk of ASD with increasing exposure to air pollution, but with some inconsistency across chemical components | Modified instrument, developed based on the Cochrane Collaboration’s tool and the AHRQ domains (i.e., selection bias, confounding, performance bias, attrition bias, detection bias, and reporting bias). Also, an approach for rating exposure assessment methods was newly developed | Majority of studies were rated as “low” or “probably low” risk of bias, besides for domains confounding and exposure assessment | Navigation Guide | Moderate quality of evidence across all air pollutants, due to small number of studies in meta-analysis and unexplained statistical heterogeneity |
Morales-Suárez-Varela, M., et al. (2017) [69], Systematic review | Population: Humans Exposure: PM2.5, PM10, or diesel PM exposure during pregnancy/ early childhood Outcome: Measures of ASD symptoms or diagnosis Studies: Published after 2005 in English | 4 cohort and 9 case–control studies | 9 out of 13 studies suggested positive associations during specific exposure windows Authors conclude there is an increased risk of ASD due to PM exposure, with varying magnitude according to the particle size and composition, with the association with PM2.5 and diesel PM being largest | Scottish Intercollegiate Guideline Network (SIGN) used to assign levels of evidence based on study design Additional considerations were sample size, specified inclusion and assessment criteria, exposure and outcome assessment, confounder control | 5 studies at a high risk of confounding or bias and a significant risk that the relationship is not causal, 8 studies at a low risk of bias and a moderate probability of causal association | SIGN levels of recommendation | Level of recommendation: C-D C: A body of evidence of studies at a low risk of bias, directly applicable to the target population, with consistent results D: Extrapolated evidence from studies at a low risk of bias An association between PM exposure and ASD cannot be ruled out. Data is insufficient to reach a consensus about ASD risk |
Tenero, L., et al. (2017) [70], Systematic review | Population: Children Exposure: Indoor or outdoor air pollution Outcome: Sleep disordered breathing, sleep apnea, excluding asthma or SIDS Studies: Observational and intervention studies included, published in English language | 4 cohort studies, 2 cross-sectional studies, 2 "prospective surveys” / intervention studies | Results suggest an involvement of environmental pollution in the worsening of sleep-disordered breathing in children | Centre for Evidence Based Medicine guidelines (2009, 2011) | Evidence level 3B for all studies | Centre for Evidence Based Medicine guidelines (2009, 2011) | Grade C (“cohort or case–control studies”) |
King, C., et al. (2018) [71], Systematic review | Population: Children < 3 years old Exposure: Criteria air pollutantsa, at any time period before hospitalization, categorized as acute (less than 7 days), sub-chronic (1 month prior), or lifetime exposure Outcome: Hospital admission, emergency department visits, unscheduled primary care visits, or critical care admission for bronchiolitis Studies: Cohort, case–control, time series, case-crossover designs included, ecological designs excluded | 4 case–control and 4 case-crossover studies | Long term exposure to PM2.5, and acute exposure to SO2 and NO2 may be associated with increased risk of hospitalization for bronchiolitis. Results for other pollutants were inconsistent | NOS, in addition to considerations of specific aspects of: selection bias, exposure and outcome assessment, adjustment for confounders. Studies were rated at low risk of bias if they adjusted for at least two prespecified confounders. Further, studies were classified as higher quality if infants were < 2 years old | NOS score 7–8 (good quality) 2 studies rated as “unclear/ high” risk of bias across all domains, 6 others were rated as “low” across all domains | GRADE | Low to moderate, frequent downgrading for inconsistency, upgrading for precision, and up- or downgrading for study quality |
Fu, L., et al. (2019) [72], Systematic review and meta-analysis | Population: Fetuses, infants at birth Exposure: Criteria air pollutantsa Outcome: Fetal growth indicators during pregnancy and anthropometric measurements at birth. Studies reporting exposure windows, sample size, and partial regression coefficient with 95% CI or standard errors were included Studies: English or Chinese language | 11 prospective and 4 retrospective studies | Higher PM2.5 exposure during entire pregnancy negatively associated with head circumference at birth (-0.30 cm, 95% CI: -0.49 to -0.10), and NO2 exposure during entire pregnancy associated with shorter length at birth (-0.03 cm, 95% CI: -0.05 to -0.02). All other associations were not significant or inconclusive | ACROBAT-NRSI, with modifications | 7 studies rated at low, 5 at moderate, and 3 at high risk of bias. Common concerns were exposure or outcome assessment methods, lack of confounder control, or study design (retrospective studies) | GRADE | PM2.5 and head circumference: Moderate PM10 and head circumference: Low NO2 and birth length: Low NO2 and head circumference: Very low Downgraded for risk of bias and inconsistency, upgraded for dose–response gradient |
Rappazzo, KM., et al. (2021) [73], Systematic review and meta-analysis | Population: Any population capable of becoming pregnant, including those at increased risk of preterm birth Exposure: O3 exposure during 1st or 2nd trimester. Exposure assessment must have covered entire trimester. Only studies reporting a continuous exposure contrast included Outcome: Preterm birth Studies: Cohort, case–control studies. Reviews and abstract-only references excluded. Only English language articles included | 16 cohort, 3 case control studies | Increased risk of PTB from exposure during 1st trimester (SOR per 10 ppb increase in O3: 1.06, 95 CI: 1.03–1.10), and during 2nd trimester (SOR per 10 ppb: 1.05, 95% CI: 1.02- 1.08) | Modified OHAT framework, domains included “participant selection, outcome, exposure, confounding (consideration of co-pollutants), analysis, selective reporting, sensitivity, and overall quality” | High n = 1, medium n = 9, and low n = 9 confidence. Common concerns included insufficient reporting or exposure assessment methods | OHAT | Moderate (no up- or downgrading in any domain) |
Ravindra, K., et al. (2021) [74], Systematic review and meta-analysis | Population: Children < 5 years old. Live births, stillbirths, and terminations eligible Exposure: Outdoor or indoor air pollutants Outcome: Congenital anomalies, meaning anomalies of prenatal origin present at birth. Neurological defects, ASD, ADHD included Studies: Case–control, cohort, or ecological studies included. Published after 1950 in English | 16 case–control, 9 cohort and 1 ecological study | N/A (due to serious concern with the data synthesis methods, results not reported here) | Risk of bias tool modified for prevalence studies from tool by Leboeuf-Yde and Lauritsen (Hoy et al. 2012), and ROBINS-E preliminary tool | Hoy et al.: Moderate (n = 4) to low (n = 22) ROBINS-E: Moderate (n = 11), low (n = 13), high (n = 1), unclear (n = 1) | Navigation Guide | NO2 and atrial septal defects, ventricular septal defects: High All other associations Low to very low Downgrading was due to risk of bias, imprecision, or inconsistency |
Stenson, C., et al. (2021) [75], Systematic review | Population: Children and adolescents Exposure: TRAP pollutants exposure during pregnancy, childhood, or adolescence Outcome: Academic performance measured using standardized tests, exam results, GPA, other, (but not tests of cognitive function) Studies: Cross-sectional, ecological, prospective or retrospective cohort, panel and case–control studies. Only peer reviewed articles in English included. Abstract-only or conference materials excluded | 7 cross-sectional designs, 2 retrospective cohort studies, and 1 longitudinal study | 9 studies reported link between higher TRAP exposure and poorer student academic performance. Effect sizes were generally small | Adapted version of the OHAT tool. Key confounders and co-exposures were pre-determined, and adjustment for these was considered | Serious concerns about insufficient reporting and exposure assessments | OHAT approach. Also, an overall data assessment visualization table was developed to assess potential publication bias | Low Due to serious risk of bias, imprecision, and potential publication bias. 9 papers stated an absence of conflict of interest or reported funding sources which did not raise concerns |
Uwak, I., et al. (2021) [76], Systematic review and meta-analysis | Population: Pregnant women Exposure: Prenatal exposure to ambient particulate air pollution (PM2.5, PM10, PM2.5–10) Outcome: Birth weight measured as a continuous variable. Studies reporting birth weight as z-scores were excluded Studies: N/R | 51 cohort, 2 cross-sectional studies | Increased risk observed for PM2.5 exposure in the 2nd or 3rd trimester: -5.69 g (95% CI: − 10.58, − 0.79, I2: 68%) and -10.67 g birth weight (95% CI: − 20.91, − 0.43, I2: 84%), respectively. Over the entire pregnancy: -27.55 g (95% CI: − 48.45, − 6.65, I2: 94%) PM10 exposure in the 3rd trimester and the entire pregnancy: -6.57 g (95% CI: − 10.66, − 2.48, I2: 0%) and -8.65 g (95% CI: − 16.83, − 0.48, I2: 84%), respectively | Navigation Guide: recruitment strategy, blinding, confounding, exposure assessment, incomplete outcome data, selective outcome reporting, conflicts of interest, other concerns | Risk of bias generally “low”/ “probably low” for most domains. 43% of studies at “probably high” risk of bias for exposure assessment method (reliance on county-level monitoring data without adequate temporal coverage/ spatial resolution) | Navigation Guide | PM2.5: Low, due to imprecision and/or unexplained heterogeneity PM10: Low (imprecision) to moderate PM2.5–10: Very low/low (risk of bias, inconsistency, and imprecision) |
Gong, C., et al. (2022) [77], Systematic review and meta-analysis | Population: Full-term, singleton neonates Exposure: Ambient PM2.5 Outcome: Change in grams of term birth weight (≥ 37 weeks of gestation) as a continuous variable Studies: Epidemiological studies | 61 cohort, 1 case–control | Per 10 μg/m3 increase in PM2.5: − 16.54 g (95% CI: − 20.07 g, − 13.02 g, I2 = 96%) | NOS (only performed for studies included in meta-analysis) | Most studies (n = 22) obtained high NOS scores (7 + stars) Others (n = 9) obtained fair/low scores (5–6 stars) | GRADE (only for studies included in meta-analysis) | Overall: Very low (downgraded for high heterogeneity) Subgroup of studies using LUR-models: Moderate, upgraded for reasonable residual confounding which could reduce the effect estimates |
Lin, LZ., et al. (2022) [78], Systematic review and meta-analysis | Population: Children 0–18 years old Exposure: Ambient PM exposure (pre-conception, prenatal, postnatal) Outcome: Neurodevelopmental disorders (ASD, ADHD, dyslexia, etc.). Confirmed by clinician or structured interview Studies: Case–control, cross-sectional, cohort, case-crossover, time-series, and panel studies. Full-text must be available; in English; did not include abstracts, reviews, conference materials | 16 case–control, 13 cohort, 2 cross-sectional studies | Increased risk of ASD linked to exposures to PM2.5 during prenatal periods (SOR: 1.32, 95%CI, 1.03–1.69), 1st year after birth (SOR: 1.62, 95%CI, 1.22–2.15) and 2nd year after birth (SOR: 3.13, 95%CI, 1.47–6.67). Inconsistent evidence found for other types of PM and neurodevelopmental outcomes Some heterogeneity explained by exposure assessment period and study country | NOS, AHRQ, and Cochrane ROB tool | NOS score for case–control and cohort studies: 6–8. AHRQ score for cross-sectional studies: moderate and low quality. Cochrane ROB tool for all studies: 25 studies rated at low risk of bias, 5 as unclear, 1 as high | The Best Evidence Synthesis (BES) System for synthesis without meta-analysis. GRADE for meta-analytical results | BES: PM2.5 exposure and the risk of ASD in 1st year after birth: Strong evidence PM2.5 exposure and the risk of ASD in 3rd year after birth: Moderate evidence 63 other associations: insufficient or no evidence GRADE: PM2.5 exposure and ASD in 2nd year after birth: Moderate (upgraded for magnitude of effect), 12 other associations: low (n = 3) or very low (n = 9) (due to heterogeneity) |
Yu, Z., et al. (2022) [79], Systematic review and meta-analysis | Population: Pregnant women without specific medical conditions Exposure: Ambient PM, short- and long-term exposures. Studies using proxies for exposure (e.g., traffic density) excluded Outcome: PTB diagnosed by clinical standardization or definite medical records Studies: Case–control, cross-sectional, time-series, or cohort studies. Excluded articles without full-text | 84 case–control and cohort studies | Long-term exposure to PM2.5 and PM10 during entire pregnancy (SOR per 10 μg/m3: 1.084 (95% CI: 1.055–1.113) and 1.034 (95% CI: 1.018–1.049), respectively. Positive associations were also found between PM2.5 exposure in 2nd trimester and PTB subtypes Short-term exposure to PM2.5 (SOR per 10 μg/m3: 1.003 (1.001–1.004, I2: 65%) and 1.003 (1.001–1.005, I2: 77%) and PM10 (SOR: 1.001 (95% CI: 1.000–1.001). PM10 exposure in 2 weeks prior to birth also increased PTB risk | Risk of bias: Tailored OHAT approach Quality: NOS and Mustafic et al. (2012) for case-crossover and time-series studies | Most studies found to have moderate to high quality. Concerns were related to exposure assessment, confounding, and exclusion bias (missing data) | Navigation Guide | Moderate (Concerns regarding inconsistency, but still reached an overall rating of “moderate”) |
Zhu, W., et al. (2022) [80], Systematic review and meta-analysis | Population: Pregnant women that conceived naturally (no IVF/ ET) Exposure: Chronic PM exposure (preconception to prenatal) Outcome: Spontaneous abortion, defined as a loss of fetus within 180 days gestation Studies: Only English language articles considered. Conference abstracts and reviews not included | 4 case–control/ case-crossover studies, 3 cohort studies | Increased risk after PM2.5 and PM10 exposure (SOR per 10 μg/ m3: 1.20 (95%CI: 1.01–1.40) and 1.09 (95%CI: 1.02–1.15), respectively | NOS | All studies were rated as “high quality” (score ≥ 7) | GRADE (GRADEpro App) | Association with PM2.5 and PM10: Moderate |
Ziou, M., et al. (2022) [81], Systematic review and meta-analysis | Population: Children/ adolescents < 19 years old. Studies exclusively conducted in asthmatic, allergic, or immunocompromised populations excluded Exposure: Ambient PM (short- or long-term exposure) Outcome: Upper respiratory tract infections, including otitis Studies: English, French, Spanish language articles from peer-reviewed journals included. Abstracts, reviews, case studies excluded | 19 time-series, 4 cohort, 4 case-crossover, 4 meta-analyses, 2 cross-sectional, and 1 interventional study | Both PM2.5 and PM10 associated with hospital presentations for URTIs (SRR: 1.010, 95%CI: 1.007–1.014), SRR: 1.016, 95%CI: 1.011–1.021) in meta-analyses. Narrative analysis found total suspended particulates to be associated with URTIs, but mixed results were found for PM2.5 and PM10 | Quality of studies: NOS for cohort studies, modified NOS (Modesti et al. for cross-sectional studies, and Mustafic et al. for case-crossover/ time-series) Risk of bias: Mixture of modified OHAT tool and Navigation Guide, as reported by previous reviews | Quality and risk of bias ratings ranged from low to high. Concerns included outcome and exposure assessment methods, missing data, recall bias, confounding, and selection bias | OHAT, with modifications to address case-crossover/ time-series studies | PM2.5 and PM10: Moderate (upgraded for confounding, as studies were conducted in hospital or ambulance settings, where only a minority of children with URTIs present. Thus, there may be a bias towards null) Total suspended particles/ PM unspecified: Low (no change) |
Blanc, N., et al. (2023) [82], Systematic review and meta-analysis | Population: Humans Exposure: Maternal or paternal exposure to outdoor air pollutants in the preconception period. Preconception period defined as 1 month to 1 year before conception Outcome: All child health outcomes Studies: English language articles | 16 cohort and 6 case–control | Exposure to outdoor air pollutants during maternal preconception period were associated with various health outcomes, of which birth defects showed the most consistent findings Large sample sizes, however inconsistencies observed in air pollutant levels and reported associations | NOS | All studies rated as high to moderate quality | Modified GRADE (starting observational studies at low) (Morgan 2016). Available human and experimental animal data and mechanistic data additionally considered Publication bias not assessed. Study funding sources also considered | Preconception exposure to PM10 and PM2.5 and birth defects Moderate PM2.5 and birthweight: Moderate All other associations: Low |
Liang, W., et al. (2023) [83], Systematic review and meta-analysis | Population: Pregnant women Exposure: Ambient air pollutants. Excluded studies assessing exposure exclusively to indoor pollutants, traffic, or extreme natural environments. Categorical comparisons not included Outcome: Risk of GDM, with explicit diagnostic criteria reported Studies: Cohort studies included. Conference abstracts and studies without accessible data excluded | 15 prospective, 16 retrospective cohort studies | Exposure to NO2, SO2, PM2.5, PM10, BC, and nitrate may significantly increase the risk of GDM. PM2.5 exposure had largest effect on GDM risk | NOS | All studies rated as high quality (7–9 stars) | Modified GRADE for air pollution studies by WHO (starting observational studies at moderate) | NO2 exposure during entire pregnancy, and BC exposure in first trimester: Low All other pollutants and exposure periods (32 associations): Moderate (14) to high (18) Main reasons for downgrading were high risk of bias and wide 80% prediction intervals |
Tandon, S., et al. (2023) [84], Systematic review and meta-analysis | Population: 10–19-year-old children/ adolescents Exposure: Short- or long-term exposure to ambient air pollutants Outcome: Blood pressure (BP) Studies: Longitudinal (cohort, panel) or cross-sectional studies | 5 cohort, 3 cross-sectional studies | Non-significant associations were observed for cohort studies assessing long-term exposure to PM10, PM2.5, and NO2. Significant positive associations were observed for cross-sectional studies assessing long-term exposure to PM10 (0.34 mmHg, 95% CI: 0.19, 0.50) and NO2 on diastolic BP (0.40 mmHg, 95% CI: 0.09, 0.71), and PM10 on systolic (0.48 mmHg, 95% CI: 0.19, 0.77) BP | NOS for cohort studies, adapted NOS for cross-sectional studies (from Herzog et al.) | 3 studies rated as good quality, 4 as fair quality, 1 as poor quality | GRADE | Low to very low Reasons for downgrading: risk of bias (due to potential selection bias, loss of follow-up, issues with exposure and outcome assessment methods), and inconsistency |