Associations of Multiple Environmental Pollutants with the Risk and Mortality of Viral Infectious Diseases: Differential Infection Spectrums and Trends

Technical Summary: Associations of Multiple Environmental Pollutants with the Risk and Mortality of Viral Infectious Diseases

Problem Statement
Existing epidemiological research has predominantly focused on the association between single infectious diseases and individual meteorological or chemical pollution factors. A significant gap remains regarding the individual and combined (mixture) effects of multiple environmental pollutants on the incidence and mortality of a broad spectrum of viral infectious diseases. Furthermore, while overt infectious diseases with clear clinical features are well-studied, diseases characterized by asymptomatic or subclinical infections (latent infections) often suffer from under-surveillance and underestimated transmission risks. This study addresses the need to quantify both single and mixed effects of various pollutants on different types of viral infections to inform environmental health prevention and control strategies.

Methodology
The study utilized surveillance data from 2005 to 2019 covering six viral infectious diseases in Beijing, China: Acquired Immunodeficiency Syndrome (AIDS), Influenza, Hand, Foot, and Mouth Disease (HFMD), Viral Hepatitis, Hemorrhagic Fever with Renal Syndrome (HFRS), and Measles. These diseases were categorized into two groups based on clinical presentation and detectability:

• Explicit Diseases: Influenza, HFRS, and Measles (characterized by significant, recognizable symptoms).
• Latent Diseases: AIDS, HFMD, and Viral Hepatitis (characterized by high proportions of asymptomatic or subclinical infections).

The researchers integrated this with annual data on meteorological factors and four categories of pollution: chemical, domestic, water, and noise.

The analytical framework involved three primary steps:

1. Epidemic Threshold Estimation: The Moving Epidemic Method (MEM) was employed to estimate epidemic thresholds and intensity levels (medium, high, very high) for each disease. This allowed for the identification of alert weeks and risk classifications.
2. Pollutant Screening: Spearman correlation coefficients were used to screen pollutants and climate factors associated with incidence thresholds and mortality. LASSO regression was additionally applied for HFMD where Spearman correlation yielded no significant factors.
3. Mixture Effect Analysis: A Weighted Quantile Sum (WQS) regression model was developed to assess the individual and combined effects of the pollution mixtures. The model estimated a co-exposure index (WQS) to identify which specific pollutants carried the highest weights for incidence thresholds and mortality, while accounting for correlations among high-dimensional mixtures. Survey-weighted linear regression was also used for single-pollutant analyses.

Key Contributions

• Methodological Application: The study successfully applied the MEM model to estimate epidemic thresholds and intensity levels across a diverse spectrum of viral diseases, including those with declining trends and latent infection patterns.
• Mixture Analysis: It advanced the field by moving beyond single-pollutant analysis to evaluate the combined effects of complex pollution mixtures (water, domestic, chemical, and noise) on viral disease outcomes.
• Spectrum Differentiation: The research highlighted distinct "infection spectrums," demonstrating that different pollutants drive risks for explicit versus latent diseases and that these effects vary between incidence thresholds and mortality rates.
• Quantification of Weights: The study identified specific pollutants with the highest positive and negative weights for disease outcomes, providing a data-driven basis for prioritizing control measures.

Results

• Epidemic Trends: Explicit diseases (HFRS, Measles, Influenza) and latent diseases (Viral Hepatitis, HFMD, AIDS) exhibited distinct temporal patterns. HFRS and Measles showed high intensity in the early and intermediate stages of the study period, respectively, while Influenza peaked in the later stages. Latent infections followed a similar temporal order.
• Pollution Associations (Incidence):
  • AIDS: The WQS index showed a positive association with incidence thresholds. Domestic waste output had the highest positive weight (0.47).
  • HFMD: Industrial wastewater discharge had the highest positive weight (0.66).
  • HFRS: SO₂ had the highest positive weight (0.37).
  • Measles: NO₂ showed a significant positive association (βWQS = 0.13).
  • General Trend: Each decile increase in the pollution mixture was associated with elevated incidence thresholds. Incidence thresholds were predominantly influenced by domestic and water pollution.
• Pollution Associations (Mortality):
  • AIDS: The WQS index was positively associated with mortality. The harmless treatment ratio for household refuse had the highest positive weight (0.32).
  • Influenza: Industrial wastewater discharge had the highest positive weight (0.16).
  • General Trend: Mortality was mainly driven by chemical and water pollution.
• Specific Findings: Single-pollutant analyses revealed that influenza mortality increased with domestic waste output and sewage treatment rates but decreased with NO₂ and industrial wastewater discharge. For latent diseases, AIDS incidence increased with domestic waste output but decreased with road traffic noise.

Significance and Claims
The paper claims that targeted control of domestic waste, industrial wastewater, and air pollutants (specifically SO₂, NO₂, and PM₁₀) should be prioritized to reduce pollution-driven risks of viral infectious diseases. The authors assert that integrating the MEM model into routine surveillance can provide accurate early warnings for both explicit and latent viral infections.

The study emphasizes the necessity of cross-sectoral collaboration between environmental and public health sectors to implement long-term pollution interventions. It posits that understanding the differential effects of pollution mixtures on various infection spectrums is crucial for precise surveillance and risk assessment.

Limitations Acknowledged by the Authors
The authors explicitly note several limitations:

• Ecological Fallacy: As an ecological study, the findings reveal statistical associations at the population level and cannot establish causal relationships at the individual level.
• Confounding: The study did not systematically control for key confounders such as vaccination rates, healthcare resources, and personal hygiene behaviors.
• Data Resolution: The mismatch between annual/monthly pollution data and weekly/daily disease monitoring data limits the identification of short-term lag effects.
• Generalizability: The focus on Beijing limits the applicability of results to regions with different climates, pollution profiles, and population mobility.
• Model Constraints: The WQS regression assumes uniform effect directions among mixture components and cannot address effect heterogeneity; future research could utilize methods like Bayesian Kernel Machine Regression (BKMR).