Cause ● Characteristic that preceded to event of disease ● Without this event, disease wouldn’t have occured ● Reason of health related event ○ Must precede the effect ○ Cause could be either… ■ host ■ Agent ■ Environmental factors ○ Two types ■ Positive=presence of an exposure ■ Negative= absence of an exposure Aim of an Epidemiologic Study Distorting factors may distort exact cause ● Confounding is another way of distortion for causal conclusion. Hill’s Criteria 1. Temporality 2. Strength of association 3. Biological gradient 4. Consistency 5. Specificity 6. Experimental evidence 7. Plausibility 8. Coherence 9. Analogy Temporality ● Exposure must be precede the onset of disease ● Only condition necessary for establishing causality ● Some critiques… ○ Cross sectional studies ○ Bi-directonal associations ○ Disease onset determination Strength and Association Increasing sample size will just make larger bias ● Stronger associations less likely to result of bias ● Confounding errors or chances ● Critiques ○ Easier to see larger association with larger sample size ○ Smaller association does not mean no causation Biological Gradient ● Association is more likely to be causal if the strength of the association increases (decreases) with amount of exposure -dose response effect ● Both the level and duration of exposure ● Critiques ○ May not be linear Consistency ● Association is more likely to be causal if found in different studies ○ Different study design ○ Different populations ● From single study might be unreliable ○ Constant threats to internal validity ● Critiques ○ Consistency does not mean accuracy ■ Only some types of methodology, study will use the same methods on different populations. ■ If flaws are not noticed, will be biased in same way ○ Publication Bias ■ When publishes statistically significant results or finding in a certain direction Specificity ● Exposure cause a unique effect ○ When exposure only cause one disease ● Critiques ○ Few risk factors actually is within it ■ One risk factor = one disease is very rare ○ Few disease only have one risk factor Experiment ● Randomized controlled trials offers the best protection against systematic error ● Should be valid in determining causal associations ● Animal experiments/ randomized trials ● Critiques ○ Can be unethical Plausibility, Coherence, Analogy ● Often treated as interchangeable ● Association more likely casual ○ Biologically plausible and supported by existing theory ● Comparison to known casual relationship chan bolster arguments that an observed association is causal ● Critiques ○ Lot we don't know ■ Can't necessarily use as judging causality