Modern Epidemiological Methods

Advanced Study Designs - Tricky Trials & Clever Cases

• Nested Case-Control (NCC):
  • Design: Cases from cohort; controls sampled at case occurrence (risk-set).
  • Adv: Cost-effective. Lmt: No incidence. Use: Biomarkers.
• Case-Cohort (CCH):
  • Design: Cases from cohort; controls (subcohort) from baseline.
  • Adv: Multiple outcomes, incidence. Lmt: Complex analysis. Use: Multiple diseases.
• Case-Crossover (CCO):
  • Design: Cases are own controls; hazard vs. control period exposure.
  • Adv: Controls fixed confounders. Lmt: Transient exposure, recall bias. Use: Acute triggers.
• Mendelian Randomization (MR):
  • Design: Genetic variants as instrumental variables (IVs) for causality.
  • Adv: Reduces confounding. Lmt: Pleiotropy, valid IVs. Use: Causal inference.

⭐ Mendelian Randomization helps mitigate confounding and reverse causation in observational studies.

Molecular & Genetic Epi - Cell Sleuths & DNA Detectives

• Biomarkers: Indicators of biological states, crucial for modern epidemiology.

  Type	Description	Example
  Biomarker of Exposure	Measures contact with an agent	Cotinine (tobacco smoke)
  Biomarker of Effect	Indicates biological response to exposure	DNA adducts (carcinogens)
  Biomarker of Susceptibility	Identifies predisposition to disease	BRCA1/2 genes (breast cancer)

• Omics Technologies: High-throughput analysis.
  • Genomics: Study of entire genomes (DNA).
  • Proteomics: Study of proteomes (proteins).
• Genetic Association Studies (GWAS):
  • Identify common genetic variants (e.g., SNPs) associated with disease risk across the genome.
  • Principle: Compare allele frequencies between cases and controls.
• Gene-Environment (GxE) Interactions:
  • Recognizes that genetic and environmental factors often interact to influence disease risk.
  • Essential for understanding multifactorial diseases.

⭐ Biomarkers of susceptibility can identify high-risk individuals for targeted prevention strategies.

Spatial Epi & AI - Geo-Gurus & Data Dynamos

• Geographic Information Systems (GIS): For visualizing, mapping, and analyzing spatial patterns of disease. Identifies geographical hotspots.

  ⭐ GIS is instrumental in identifying geographical hotspots for targeted public health interventions.

• Cluster Analysis: Statistical methods to detect disease clusters.
  • Spatial: Identifies geographic concentrations (e.g., SaTScan).
  • Temporal: Detects unusual occurrences over time.
• Syndromic Surveillance: Early outbreak detection using pre-diagnostic data (e.g., symptoms, emergency calls, OTC sales).
• Big Data & AI/Machine Learning (ML):
  • Sources: EHRs, social media, genomic data, environmental sensors.
  • AI/ML Applications: Predictive modeling for outbreaks, automated surveillance, risk factor identification, personalized prevention.

Evidence Synthesis - Review Roundups & Stat Stacks

• Systematic Review: Formulate question, search, select, extract data, assess quality, synthesize findings.
• Meta-Analysis: Statistically combines study results for a precise pooled estimate.

  ⭐ The $I^2$ statistic quantifies the percentage of total variation across studies due to heterogeneity rather than chance.

• PRISMA Guidelines: Preferred Reporting Items for Systematic Reviews and Meta-Analyses. 📌 Mnemonic for flow: "I-S-E-I" (Identification, Screening, Eligibility, Included).

• Forest Plot: Visualizes individual study results & overall pooled estimate.

  • Vertical line: Line of no effect (e.g., OR=1). Diamond: Pooled estimate; if crosses line → non-significant.

• Heterogeneity: Variation between study results beyond chance.

  • Chi-square test (p < 0.10 suggests heterogeneity).
  • $I^2$ statistic: Quantifies extent. $I^2 = (Q - df)/Q \times 100%$.
    • <25%: Low
    • 25-75%: Moderate
    • >75%: High heterogeneity.

• Publication Bias: Tendency to publish positive/significant results.

  • Funnel Plot: Scatter plot of effect size vs. study precision. Asymmetry suggests bias.
  • Egger's test: Statistical test for funnel plot asymmetry.

• Effect Models:

  Model	Assumption	Use When	CI Width
  Fixed-effect	One true effect; variation by chance	Low heterogeneity	Narrower
  Random-effects	Effects vary; incorporates heterogeneity	High heterogeneity	Wider

High‑Yield Points - ⚡ Biggest Takeaways

• GIS enables spatial mapping of diseases and hotspot identification.
• Remote sensing aids in tracking environmental risk factors and vector habitats.
• AI/Big Data revolutionizes disease surveillance, outbreak prediction, and risk modeling.
• Molecular epidemiology uses biomarkers for pathogen subtyping and transmission dynamics.
• Genetic epidemiology explores genetic susceptibility and gene-environment interactions in disease.
• Digital epidemiology utilizes online data sources for early warning systems and syndromic surveillance.
• Causal inference methods like Mendelian Randomization strengthen evidence from observational studies.