Shared Flashcard Set


Confounding (T Pierce)
Health Care

Additional Health Care Flashcards




confounding (definition and impact)


  • mixing of effects; association between exposure and disease is distorted because it is mixed with the effect of another factor that is associated with the diseases
  • impact- results of confounding is to distort the true association toward the null (negative confounding)or away from the null (positive confounding)


criteria for confounding variables

  • must be more or less common in the exposed than unexposed group
  • must be an independent cause or predictor of the disease
  • must not be an intermediate step in the causal pathway between disease and exposure
  • can be risk factor, preventative, or correlate for a cause of disease
  • must be associated with disease independently of exposure

assessing for presence of confounding

  • is variable related to both exposure and outcome
  • does the association seen in the crude analysis have the same direction and similar magnitude to associations seen within strata of confounding variable

how to control for confounding in general

  • design and analysis phase
    • control for confounding, you must have info on variables that are potential confounders
    • this is true for all methods controlling confounding except randomization 
Usually, confounders are risk factors for disease

Name ways to control for confounding in design phase


  • randomization
    • with large enough sample size, likely to control for known and unknown sample size
  • restriction
    • restrict admissibility criteria for study subjects and limit entrance to individuals who fall within a specified category of the confounder
    • advantage- straightforward, convenient, inexpensive
    • disadvantage- limit generalizability
  • matching
    • select study subjects so that potential for confounding is distributed in an identical manner among exposed and unexposed groups (cohort) or among cases and controls (case control study)


goals of controlling for confounding in design phase


  • goal- eliminate or reduce variation in the level of confounding factor between compared group


methods for controlling for confounding in analysis
  • multivariate analysis
    • simultaneously adjusts for several variables
    • involves construction of math model that describes association between exposure, disease, and confounders
  • stratification
    • evaluate the association within homogeneous categories of the confounding variable
    • if stratum specific estimates differ appreciably from each other = presence of effect modification
    • purpose- control for confounding and to assess effect modification
    • limits- difficult to multiple variables simultaneously due to large number of strata that will be generated relative to number of study subjects


Supporting users have an ad free experience!