검색 상세

Prenatal Exposure to Ambient PM2.5 and NO2 and Early Childhood Neurodevelopment: The Mediation by Maternal Sleep - Evidence from the Korean Children’s Environmental Health Study （KO-CHENS）

초록/요약

산전 미세먼지（PM2.5） 및 이산화질소（NO2） 노출이 영유아 신경발달에 미치는 부정적 영향은 보고된 바 있으나, 이 연관성을 연결하는 생물학적 경로는 아직 충분히 규명되지 않았다. 매개 경로 후보 중 산모 수면 장애는 특히 주목할 만한데, 대기오염물질은 수면 연속성을 방해하는 인자로 이미 확립되어 있으며, 손상된 산모 수면은 시상하부-뇌하수체-부신（HPA） 축 교란을 통해 태아 신경내분비 프로그래밍을 독립적으로 저해하기 때문이다. 그러나 이 경로가 반사실적 인과 분석 틀 내에서 통계적으로 검증 가능한 매개 기전을 구성하는지에 대해서는 기존 연구에서 검토된 바 없다. 본 연구는 2015년부터 2020년까지 한국 어린이 환경보건 코호트（KO-CHENS）에 등록된 단태아 출생아 3,619명을 대상으로, 검증된 토지이용 회귀모형（LUR）을 활용하여 임신 분기별 및 전체 임신기간 PM2.5·NO2 농도를 거주지 주소 수준에서 추정하였다. 신경발달 상태는 생후 6개월에 덴버 발달 선별 검사 II（DDST-II）와 다변량 로지스틱 회귀모형으로, 생후 12·24·36개월에는 베일리 영유아 발달 검사 3판（BSID-III）과 다변량 선형 회귀모형으로 각각 평가하였다. 임신 중기 산모 수면은 피츠버그 수면의 질 지수（PSQI） 전체 점수 및 자가 보고 수면 시간을 통해 수집하였으며, 두 변수 모두 매개변수 후보로 설정하였다. 오염물질의 총 효과는 5,000회 비모수 부트스트래핑을 통해 평균 인과 매개 효과（ACME）와 평균 직접 효과（ADE）로 분해하였다. 임신 1분기 PM2.5 10 μg/m³ 증가당 생후 6개월 신경발달 지연 위험이 88% 증가하였으며（OR: 1.88; 95% CI: 1.15, 3.11）, 일반화 가산 모형 스플라인 분석을 통해 선형 용량-반응 관계가 확인되었다（edf = 1.00; p = 0.012）. 임신 1분기 NO2는 완전 보정 모형에서 생후 24개월 인지 복합점수의 유의한 감소와 독립적으로 관련되었다（β: −1.10; 95% CI: −1.91, −0.28）. 임신 2분기 PM2.5는 산모 수면 시간 연장과 유의하게 관련되었고（β: 0.19시간/10 μg/m³; 95% CI: 0.08, 0.30）, 이는 수면 분절보다 사이토카인 매개 수면 유도 기전에 부합하는 양상이었다. 반사실적 매개 분석 결과, PM2.5 노출이 산모의 수면 시간을 연장시키고 그 연장된 수면 시간이 신경발달에 영향을 미치는 경로에서, 생후 36개월 인지 복합점수（ACME: −0.08; 95% CI: −0.18, −0.01） 및 언어 복합점수（ACME: −0.09; 95% CI: −0.20, −0.01）에 대해 통계적으로 유의한 ACME가 확인되었다. 임신 2분기 PM2.5의 생후 24개월 언어 복합점수에 대한 유의한 직접 효과도 관찰되었다（ADE: −1.80; 95% CI: −3.10, −0.43）. PSQI 전체 점수에 대한 유의한 매개 효과는 어느 평가 시점에서도 확인되지 않았다. 본 연구 결과는 임신 1분기를 PM2.5 귀인 신경발달 위험의 가장 중요한 취약 시기로 규명하며, 저자가 아는 한 반사실적 분석 틀을 통해 산모 수면 시간이 산전 PM2.5와 영유아 초기 인지 및 언어 발달 결과 사이의 통계적으로 유의한 매개 경로임을 공식적으로 입증한 최초의 연구이다. 이 경로에서 산모 수면이 잠재적으로 중재 가능한 중간변수로 확인됨에 따라, 고노출 도시 환경 임산부에서 대기오염 귀인 신경발달 부담을 경감하기 위한 보완 전략으로서 수면 중재의 효과 평가가 지지된다. 주요어: 산전 대기오염; 신경발달 지연; 산모 수면; 인과 매개 분석; 출생 코호트

more

초록/요약

Although prenatal exposure to ambient fine particulate matter （PM2.5） and nitrogen dioxide （NO2） has been linked to adverse neurodevelopmental trajectories in early childhood, the biological mechanisms connecting these exposures to downstream developmental outcomes remain poorly understood. Among plausible intermediary pathways, maternal sleep disruption warrants particular attention: air pollutants are well-established disruptors of sleep continuity, and compromised maternal sleep independently perturbs fetal neuroendocrine programming through hypothalamic-pituitary- adrenal （HPA） axis dysregulation. Whether this pathway constitutes a statistically verifiable mediating mechanism has not previously been examined within a counterfactual causal framework. Drawing on 3,619 singleton births enrolled in the Korean Children's Environmental Health Study （KO-CHENS） between 2015 and 2020, this study estimated trimester-specific and whole-pregnancy PM2.5 and NO2 concentrations at each participant's residential address using a validated land-use regression（LUR） model. Neurodevelopmental status was evaluated at 6 months via the Denver Developmental Screening Test II （DDST-II）, modelled with multivariable logistic regression, and at 12, 24, and 36 months via the Bayley Scales of Infant and Toddler Development, Third Edition （BSID-III）, modelled with multivariable linear regression. Mid-pregnancy maternal sleep was captured through the Pittsburgh Sleep Quality Index （PSQI） global score and self-reported sleep duration, both treated as candidate mediators. The total pollutant effect was decomposed into the average causal mediation effect （ACME） and average direct effect （ADE） via nonparametric bootstrapping across 5,000 iterations. Each 10 μg/m³ increment in first-trimester PM2.5 was associated with an 88% increase in the odds of neurodevelopmental delay at 6 months （OR: 1.88; 95% CI: 1.15, 3.11）, with a linear dose-response confirmed by generalized additive model spline analysis （edf = 1.00; p = 0.012）. First-trimester NO2 was independently associated with reduced cognitive composite scores at 24 months in the fully adjusted model （β: −1.10; 95% CI: −1.91, −0.28）. Second-trimester PM2.5 was positively associated with longer maternal sleep duration （β: 0.19 hours per 10 μg/m³; 95% CI: 0.08, 0.30）, a pattern consistent with cytokine-driven somnolence rather than sleep fragmentation. Counterfactual mediation analysis identified statistically significant ACMEs for the PM2.5 exposure disrupting the maternal sleep duration and that in turn having an impact on neurodevelopmental pathway at 36 months for both cognitive （ACME: −0.08; 95% CI: −0.18, −0.01） and language composite scores （ACME: −0.09; 95% CI: −0.20, −0.01）. A significant direct effect of second-trimester PM2.5 on language composite scores was additionally observed at 24 months （ADE: −1.80; 95% CI: −3.10, −0.43）. The PSQI global score produced no significant mediation effects at any assessment point. These findings establish the first trimester as the most biologically critical gestational window for PM2.5-attributable neurodevelopmental risk in a Korean birth cohort setting, and, to the authors' knowledge, represent the first formal counterfactual demonstration of maternal sleep duration as a statistically significant mediator between prenatal PM2.5 and early childhood cognitive and language outcomes. The identification of maternal sleep as a potentially modifiable intermediate in this pathway supports investigation of sleep-targeted interventions as a complementary approach to reducing air-pollution-attributable neurodevelopmental burden among pregnant populations in high-exposure urban environments. Keywords: Prenatal air pollution; Neurodevelopmental delay; Maternal sleep; Causal mediation analysis; Birth cohort

more

목차

I. Introduction 1
A. Background 1
1. Global Burden of Air Pollution 1
2. Prenatal and Early-Life Air Pollution Exposure and Neurodevelopmental Delay 3
3. Biological Mechanisms Linking Air Pollution to Fetal Neurodevelopment 6
4. Maternal Sleep as a Proposed Mediating Pathway 8
B. Statement of the Problem 11
C. Objectives and Questions of Research 11
D. Significance of the Study 12
II. Literature Review 14
A. Introduction 14
B. Epidemiological Evidence: PM2.5 and NO2 and Child Neurodevelopment 15
1. Study Designs and Exposure Assessment Approaches 15
C. Cognitive Delay and Psychomotor Development 16
D. Language Delay 17
E. Biological Mechanisms of Pollutant-Attributable Neurodevelopmental Toxicity 17
1. Neuroinflammation 17
2. Oxidative Stress 19
3. Epigenetic Reprogramming 20
4. HPA Axis Dysregulation and Fetal Programming 21
F. Air Pollution and Maternal Sleep During Pregnancy 22
1. Evidence from the General Population 22
2. Evidence in Pregnancy 23
G. Maternal Sleep Disruption and Fetal Neurodevelopment 24
1. HPA Axis as the Neuroendocrine Bridge 24
2. White Matter Microstructure and Neonatal Brain Development 25
3. Postnatal Pathways: Caregiving Quality and Child Sleep 26
H. Theoretical Framework and the Mediation Model 27
1. The concept of mediation in epidemiology 27
2. Foundations and limitations of traditional mediation approaches 28
3. Counterfactual causality and the formalization of mediation 31
4. Identification assumptions and their application to the present study 33
5. Appropriateness of the counterfactual framework for the present investigation 36
I. Identified Gaps and Contribution of the Present Thesis 37
J. Chapter Summary 39
III. Methods and Materials 45
A. Study design and population 45
B. Exposure assessment 46
C. Mediator assessment 48
D. Outcome Assessment 49
E. Covariates 50
F. Statistical analysis 53
1. Descriptive analysis 53
2. Verification of Linearity Assumption 54
3. Exposure-outcome analysis 54
4. Exposure-Mediator Analysis 56
5. Subgroup analysis 58
6. Sensitivity analysis: prenatal exposure model 58
7. Software 59
IV. Results 60
A. Baseline Characteristics of Study Participants 60
B. Prenatal PM2.5 Exposure and Neurodevelopmental Outcomes 78
1. Neurodevelopmental delay at 6 months （DDST-II） 78
2. Continuous Neurodevelopment Outcomes at 12, 24, and 36 months （BSID-III） 81
C. Prenatal NO2 exposure and Neurodevelopmental Outcomes 83
1. Neurodevelopmental delay at 6 months （DDST-II） 83
2. Continuous Neurodevelopment Outcomes at 12, 24, and 36 months （BSID-III） 84
D. Postnatal PM2.5 and NO2 Exposure and Neurodevelopmental Outcomes 88
E. Linear Mixed-Effects Model for Repeated Neurodevelopmental Assessments 91
F. Exposure-Mediator Associations: PM2.5, NO2 and maternal sleep 94
1. PM2.5 and PSQI Global Score 94
2. PM2.5 and Individual Sleep Components 95
3. NO2 and Maternal Sleep Variables 96
G. Causal Mediation Analysis 98
1. Mediation by PSQI Global Score 98
2. Mediation by sleep duration 100
H. Verification of Linearity Assumption 103
I. Subgroup Analysis 106
J. Sensitivity Analysis: prenatal exposure 112
1. Kriging Model as Alternative Exposure Assessment: PM2.5 112
2. Kriging Model as Alternative Exposure Assessment: NO2 115
3. Sensitivity Analysis: E-values for Unmeasured Confounding 118
K. Summary of Findings 121
V. Discussion 124
A. Principal Findings 124
B. Primary Exposure–Outcome Associations 125
1. Prenatal PM2.5 and Neurodevelopmental Delay at 6 Months 125
2. Prenatal NO2 and Cognitive Development at 24 Months 128
C. Comparison with Findings from Prior Birth Cohort Studies 130
D. Trimester-Specific Mechanisms 136
1. First Trimester: Structural Disruption of Neurulation and Radial Migration 137
2. Second Trimester: Synaptic Scaffolding, BDNF Epigenetic Programming, and the Physiological Opening of the Sleep Mediation Window 139
3. Third Trimester: Myelination Initiation and the Absence of Significant Associations in the Present Follow-Up Window 141
E. PM2.5 and Maternal Sleep: Exposure-Mediator Associations 143
F. Causal Mediation Analysis: Sleep Duration as a Mediating Pathway 145
1. Principal Mediation Findings and Their Interpretation 145
2. Biological Mechanisms of the Sleep Duration Mediation Pathway 148
3. Alternative Explanations and Honest Appraisal of the Mediation Results 150
G. Subgroup Analysis and Effect Modification 153
H. Sensitivity Analyses and Robustness of Findings 155
I. Strengths and Limitations 157
1. Strengths 157
2. Limitations 158
J. Public Health and Clinical Implications 161
K. Conclusion 163
References 165
국문요약 182

more
반출 Meta View 목록