Please enable JavaScript.

Coggle requires JavaScript to display documents.

Study designs (analytical studies (randomised controlled trial…

- - - - inexpensive
      - good for rare diseases with long latencies
      - can look for several exposures per disease
    - - risk of selection and recall bias
      - not possible to measure prevalence, incidence, or risk
      - can only look for one disease
      - not feasible for rare exposures
      - temporality is difficult to assess
  - - - least prone to bias
      - temporaility - look at the exposure before the outcome
      - can study several exposures
      - and can study several outcomes for each exposure
    - - expensive and time-consuming
      - loss to follow-up may lead to bias
      - poor design for rare diseases or those with long latencies
  - - - randomisation
        
        ensures that the only difference in groups is due to trial treatment
        
        bias
        
        confounding
        
        all participants have the same chance of being assigned to each of the study groups
        
        assume that clinicians will thwart every attempt to maintain randomisation
        
        investigators who intellectually grasp the concept may have contradictory interests in clincial practice - trying to get the best treatment for a particular patient
        
        clincians must have equipoise
        
        treatment effects found are likely to be much larger without adequate concealment of treatment allocation or double-blinding
        
        true randomisation
        
        must be unpredictable
        
        need to separate the person who generates allocation from those who assess eligibility
        
        sequentially numbered, sealed, opaque envelopes
        
        third-party schemes
        
        telephone randomisation
        
        pharmacy randomisation
        
        web-based service
      - blinding
        
        knowledge of clinicians, staff, patients etc to the actual assignment
        
        occurs during and after enrollment
        
        safeguards the assignment sequence after allocation (patients, clinicians, trial investigators)
        
        not always possible - financial burden - often requires more staff
        
        consider the importantce of blinding with respect to outcome-level bias - subjective vs objective outcomes
        
        if subjective, blinding is even more important
        
        reporting bias
      - stopping rules
        
        if serious adverse effects, stop for safety
        
        futtility - if interim differences between groups are minimal, and positive result with the sample size is unlikely
        
        external new information
        
        apparent benefit
      - intention to treat analysis
      - allocation concealment
        
        recruitment and assignment to groups
        
        occurs before and during the enrollment process
    - - control group
      - minimises bias
      - minimises confounding
      - statistically powerful
    - - expensive
      - patient enrolment can be difficult
      - drop-out rate
      - ethical problems in testing new therapies in humans
      - long time scale
      - complex to interpret
  - - - prevalence not incidence
      - for frequent, non-fatal chronic disease
      - quick and easy
      - useful for hypothesis generation
    - - can't measure incidence
      - prevalent cases are survivors
      - can't assess temporality
      - don't know when the disease occured
      - quickly emerging diseases are a problem
      - not good for hypothesis testing
- - - - undescribed disease, unexpected link between diseases, unexpected new therapeutic effect, adverse events
    - - recognition of new diseases
      - formulation of hypotheses
    - - high risk of chance
      - lack of appropriate comparison group