Epidemiology study designs

Ecological
1. Features
  1. Advantages
    1. If data is difficult to measure at an individual level
    2. To study group-sepcific events
    3. When data are readily available
      1. Quick
      2. Less costly
  2. Disadvantages
    1. Cannot infer causality
    2. Confounders likely
  3. Population-based
  4. Prevalence or incident cases
  5. Data Collection
    1. Routine data sources
      1. Vital registration data
      2. Demographic data
      3. Disease registries
  6. Analysis
    1. Standardization
      1. Direct
        Standardized Rate
      2. Indirect
        SMR
  7. Interpretation
    1. Ecological fallacy
    2. Bias
    3. Mixing/Migration
    4. Confounding
  8. Study Design
    1. Time-trend design
    2. Mixed studies
    3. Multi-group study
Cross-sectional
1. Features
  1. Advantages
    1. Rapid
    2. Easy to collect/analyse
    3. Relatively cheap
    4. Can help policy to see how many ppl need treatment
    5. May be possible to assume causality if exposure constant (i.e. genetics)
    6. May be only plausible study design to answer question
  2. Disadvantages
    1. Did exposure precede outcome (information bias)?
    2. Medium probability of selection bias
    3. Medium confounding
    4. Causality not possible in exposures that change (i.e. diet)
    5. Can't see time relationship or incidence rate
  3. Study Design
    1. Descriptive
      1. Estimate the frequency of outcomes or exposure without the intention of investigating causal associations.
        Individual level data on outcome, exposure or both
      2. Secular
      3. Seasonal
    2. Analytical
      1. Collects data on both outcome and exposure at the same time-point for a given study subject. Good for testing hypothesize of association where exposure status does not change over time.
  4. Sampling
    1. Questionnaires
    2. Pilot study
2. Data collection
  1. Period Prevalence
  2. Point Prevalence
3. Interpretation
  1. Reverse causality
4. Analysis
  1. **Prevalence Ratio
  2. Outcome rare= odds ratio
Cohort Studies
1. Features
  1. Advantages
    1. Low selection bias
    2. Low recall bias
    3. Low confounding
  2. Disadvantages
    1. Loss to follow up
    2. Logistically difficult
    3. Expensive
  3. Study Design
    1. Descriptive
      1. Observational
    2. Analytical
      1. Prospective
      2. Retrospective
    3. Sampling
      1. Occupational cohort
    4. Analysis
      1. Descriptive
        Risk or rate
      2. Analytical
        Risk, rate, attributable risk, attributable fraction, population attributable risk, population attributable fraction
      3. Time-series analysis
    5. Data collection
      1. Retrospective= routine sources
      2. Prospective = exposures that aren't going to change
  4. Interpretation
    1. Confounding/bias
    2. Loss to follow up
    3. Observer bias
  5. Screening method
    1. Lead-time bias
    2. Length-time bias
Case Control
1. Features
  1. Sampling
    1. Case definition
      1. Inclusion/exclusion criteria
  2. Data Collection
    1. Almost always retrospective
    2. Reporting bias
  3. Advantages
    1. Low loss-to follow up
    2. Rapid
  4. Disadvantages
    1. Selection bias
    2. Information bias
    3. Confounding
  5. Study Design
    1. Matching
      1. Nested case-control
      2. Overmatching
      3. Individual matching
      4. Frequency matching
        Control
  6. Analysis
    1. Odds ratio of exposure
    2. Outcome rare measure approximates risk ratio
  7. Interpretation
    1. Watch for discussed biases and confounders
    2. Reverse causality
  8. Screening method
    1. Selection bias
    2. Information bias
Intervention
1. Advantages
  1. Low selection bias
  2. Low information bias
  3. Low Confounding
  4. Infer causality
2. Disadvantages
  1. Loss to follow up
  2. More complex
  3. Expensive
3. Features
  1. Study Design
    1. Ethical approval
      1. Declaration of Helsinki
    2. Efficacy
    3. Effectiveness
      1. Plausability studies
        Historical controls
        Geographical controls
        Opportunistic controls
    4. RCT
      1. Cluster-randomized trials
      2. Factorial design
      3. Crossover design
    5. Non-randomized trial
      1. selection bias
  2. Sampling
    1. Representative population
    2. Choice of Control
    3. Allocation
      1. Simple randomization
      2. Systematic randomization
      3. Blocked Randomization
      4. Stratified randomization
      5. Matched-pair randomization
      6. Envelopes (allocation concealment)
      7. Minimization
    4. Blinding
      1. Single
      2. Double blinded
        Gold standard: Double-blinded randomized placebo-controled trial
      3. ascertainment bias
  3. Data Collection
    1. Same as Cohort and Cross-sectional
  4. Analysis
    1. Intervention allocation
    2. Interim analysis
    3. Intention-to-treat v.s per-protocol
    4. Intervention efficacy
  5. Interpretation
    1. Bias and confounders
  6. Screening methods
    1. Information bias
    2. Misclassification bias

Next up

Epidemiology study designs

Description

Resource summary

Similar

	Created by jillian.kowalchuk over 10 years ago