Epidemiologists use various measures to understand how exposures relate to health outcomes. These tools, like relative risk and odds ratios, help quantify the strength of associations between factors and diseases.

Interpreting these measures requires careful consideration of statistical significance, potential biases, and confounding factors. By understanding these concepts, researchers can draw more accurate conclusions about cause-and-effect relationships in public health.

Measures of Association in Epidemiology

Calculation of risk measures

• Relative Risk (RR)
  • Ratio of incidence in exposed group to incidence in unexposed group quantifies association strength
  • $RR = \frac{Incidence_{exposed}}{Incidence_{unexposed}}$
  • Interpretation: RR = 1 no association, RR > 1 increased risk, RR < 1 protective effect
  • Used in cohort studies and randomized controlled trials
• Odds Ratio (OR)
  • Ratio of odds of exposure in cases to odds of exposure in controls measures association in case-control studies
  • $OR = \frac{Odds_{exposed}}{Odds_{unexposed}}$
  • Approximates RR when outcome is rare (<10% prevalence)
  • Useful when incidence cannot be directly calculated (retrospective studies)
• Attributable Risk (AR)
  • Difference in risk between exposed and unexposed groups quantifies absolute increase in risk
  • $AR = Incidence_{exposed} - Incidence_{unexposed}$
  • Expressed as cases per 1,000 or 10,000 population
  • Population Attributable Risk (PAR) estimates proportion of disease attributable to exposure in entire population

Absolute vs relative associations

• Absolute measures
  • Quantify actual difference in risk between groups (cases per 1,000)
  • Provide context for public health impact and individual risk assessment
  • Guide resource allocation and intervention planning
  • Examples include Attributable Risk and Number Needed to Treat (NNT)
• Relative measures
  • Express proportional difference in risk between groups
  • Facilitate comparison across populations with different baseline risks
  • Often used in scientific communication and meta-analyses
  • Examples include Relative Risk and Hazard Ratio
• Choosing between measures
  • Consider study design (case-control vs cohort)
  • Align with research question (individual risk vs population impact)
  • Tailor to target audience (clinicians vs policymakers)

Role of statistical significance

• Statistical significance
  • Quantifies likelihood observed results occurred by chance
  • P-value represents probability of obtaining results assuming null hypothesis
  • Significance level (α) typically set at 0.05 determines rejection of null hypothesis
  • Confidence Intervals (CI) provide range of plausible values for true effect
• Assessing measures of association
  • Guides interpretation of observed associations (real effect vs random variation)
  • Informs decision-making in research and public health policy
  • Helps determine need for further studies or interventions

Sources of bias and confounding

• Selection bias
  • Distorts sample representativeness of target population
  • Examples: volunteer bias (healthier individuals participate), loss to follow-up (differential attrition)
  • Mitigation: proper sampling techniques, minimizing non-response
• Information bias
  • Introduces systematic errors in data collection or measurement
  • Examples: recall bias (differential memory of past exposures), observer bias (subjective outcome assessment)
  • Mitigation: standardized data collection, blinding of assessors
• Confounding
  • Third variable associated with both exposure and outcome distorts true relationship
  • Leads to over- or underestimation of effect
  • Control methods: randomization, matching, stratification, statistical adjustment
• Effect modification
  • Exposure effect varies across levels of third variable
  • Represents true biological interaction
  • Addressed through stratified analysis or interaction terms in models