Causation

(aetiology)

(pp, p. 2-1)

• the study of what causes, increases or decreases the risk for a disease or condition.
• occurs when exposure definitely results in the corresponding outcome
  • association occurs when exposure makes it more or less likely
• studies of causation generally look at:
  • Exposures to causative/protective agents
  • Outcomes
  • Time component
  • Other factors that can affect outcomes

    ---

CRITERIA FOR ASSESSING CAUSATION

• Correct Temporal Sequence
  • Exposure must proceed outcome
  • in cross-sectional, exposure and outcome are assess concurrently (temporality not established reliably)
• Strength of the Association
  • hi relative risk, less likely assoc explained by unrecognized factors, bias, or confounders
  • so, several criteria are used to judge causality
• Consistency of the Association
  • more likely a cause when similar findings are found in different 
    • populations
    • times
    • places
    • methodologies
• Dose-Response Relationship
  • increased exposure > greater frequency of outcome
  • absence of dose-response does not mean noncausal, but a threshold may exist
• Biological Plausibility
  • is association credible based on natural history of the disease or possible pathogenic mechanisms?
• Experimental Evidence
  • strengthens case for a causal relationship
  • RCY and randomized community trials reduce bias and confounders > thus provide strong corroboration of suspected causal association

Risk Factors

(pp, p2-1)

• Risk factor = an exposure (agent or event)

Temporality

(termporal sequence)

(pp, p2-2)

• Exposure comes before outcomes 
• Most important criteria in studies of causation/etiology/harm

dose-response relationship

(pp, p.2-3)

• The greater the exposure, the more frequent/severe the outcome

analytical observational study

(pp, p.3-1)

(text p179-180)

• This study design examines risk factors (exposures) for certain diseases (outcomes)
• Types of observational studies
  
  • Cohort 
  • Nested Case-control 
  • Case-control 
  • Cross-sectional 
  • Case reports; case series

• Study of risk factors/exposures for certain diseases/outcome
• Researcher does NOT manipulate the exposure 
• Exposure is measured, and its association with the outcome is calculated 

cross-sectional study

(pp, p4-3)

(text 180)

• Information on exposure and outcome are collected at the same time, and at only one point in time (e.g., surveys)
• No comparison group
• Cannot determine temporality

case-control study

(pp, p. 4)

(text, 180)

• Involves participants with and without outcome
• Evidence of exposure is collected retrospectively
• Useful for rare or very long-term outcomes

• **case-control studies can also be 'nested' within an existing cohort study
• Nested Case-Control/Design
  • Identified within cohort study 
  • Participants with and without outcome
  • Still prospective

cohort study

(pp, p.3-3)

(text, p180)

• Involves people with and without exposure
• Prospective – followed over time for evidence of outcome

  ---

• Eg. id a group of LTC residents without pressure ulcers and collect measurements (BMI)
• follow this group to determine how many, and which, develop pressure ulcers.

power

(pp, p5-2)

(text p183) not added yet

• the ability of the study to detect a statistically significant finding if it is there.

• minimun number of Ss required for sufficient _____ to avoid false negative (Type II)
• Statistical significance is r/t to the _____ of a study, which is linked to sample size.  If the sample size is not large enough to detect an effect of an exposure on an outcome, then there is a risk that we may calculate a false-negative result.

Sample Size Calculation

(pp, p5-2)

• Look for evidence that the researchers determined the minimum number of participants needed. This is required to give the study sufficient power to avoid a false negative. (What type of error is this?)

bias

(pp, p. 5-3)

• Results from systematic (i.e., regularly occurring) error in estimates of association between the exposure and the outcome
• Can be difficult to detect and interpret
• Its effect is dependent on the relationship to the exposure or outcome or both

information bias

(pp, p6-1)

• Information bias - Error in the information collected for the study
• Misclassification bias - A type of information bias
  • Participants are incorrectly labeled with the exposure or outcome
  • The effect depends on the design and whether or not the misclassification relates to the outcome

recall bias

(pp, p. 6-2)

• A common type of misclassification bias
• Refers to inaccurate participant memory in giving responses
• Leads to overestimates the association between exposure and outcome. Why?

selection bias

(pp, p6-3)

• Errors in the process of identifying study participants 
• Has less of an effect in a cohort study because it is prospective and the outcomes are not known in advance

confounding factor

(confounder)

(pp, p. 7-1)

• Key factors that affect the outcome
• Confounders are plausible alternative explanations for the association between exposure and outcome
• May be known or unknown but if known should be excluded or measured and accounted for in the data collection and/or analysis

matching

(pp, p.7-2)

• A technique used in case-control studies to control for or minimize confounders
• Participants are selected as controls based on known confounders in the cases
• If matched, appropriate analysis must be used

strength of the association

(pp, p. 8-1,2)

• A mathematically calculated estimate of the relationship between the exposure and the outcome
• May be (and generally should be) statistically altered/adjusted to control for the effect of confounding factors
• Relative Risk: risk of adverse effect in exposed group divided by risk of adverse effect in unexposed group
• RR=1: no difference
• RR>1: increase in risk associated with exposure
• RR<1: decrease in risk associated with exposure
•  
•  

relative risk (RR)

(pp, p. 8-3)

(text p180)

• Relative Risk: risk of adverse effect in exposed group divided by risk of adverse effect in unexposed group
• RR=1: no difference
• RR>1: increase in risk associated with exposure
• RR<1: decrease in risk associated with exposure

• An estimate of association between an exposure and the outcome
• Used for RCTs and cohort studies (prospective)
• Compares rates of developing the outcome (i.e., disease/condition) in two groups of people; those with and those without the exposure/risk factor

Odds Ratio

(pp, p.9-1)

(text, p182)

• OR and RR are similar when frequency is low
  • * become divergent as outcomes become more frequent*
• Another estimate of association between an exposure and an outcome
• Used for case-control studies (retrospective)
  • ORs is not confined to case-control studies
• Comparison of rates of having the exposure/risk factor in two groups of people; those with and those without the outcome(s)

harm

(pp, 1-2)

• Studies of harm evaluate the negative outcomes or side effects of treatments

Number Needed To Harm

(pp, p. 10-1)

• If the ARR is a negative number, the incidence rate of the Exposed Group is GREATER than the incidence rate of the Unexposed Group (only works if you are sure to subtract Re from Rc vs. the reverse)
• If this is the case, then we are doing more harm than good
• When the ARR is a negative number, the NNT becomes a Number Needed to Harm (NNH)