epidemiology IV Flashcards
risk factor
- An environmental, behavioural or biological factor confirmed by temporal sequence, usually increasing the probability of a disease occurring, and, if absent or removed, reduces the probability.
- Risk factors are part of the causal chain, or expose the host to the causal chain. Once disease occurs, removal of a risk factor may not result in a cure.”
concept of cause
- Few diseases have single ‘cause’
- Most result from exposure of susceptible individuals to one of more causal agents
- multifactorial
- Exposure to causal agent(s) does not inevitably result in disease
- Investigation of cause complex:
- characteristics of susceptible/resistant individual
types of exposure to external agent
key to remember in concept of cause
never single dimension - duration, frequency, types etc
bradford hill criteria is used
define causal risk factor
selected criteria for casual association
i.e. factor under study likely to cause disease
8 bradford hill criteria
- Strength of association
- Dose response
- Change in risk factor – reduction
- Time sequence
- Consistency
- Specificity – defined exposures
- Biological plausibility
- Experimental preventive trial
Multiple studies need to show same data
cause needs separated from
Concept of cause must be distinguished from concept of association
- Not all factors associated with occurrence of disease are causes
- Some factors may be associated independently with a causal agent but do not themselves cause disease or increase risk of developing disease
cause is a higher bar than association
cause =
external agent which results in disease in susceptible individuals
confounding variables =
common things between variables
Underlie
- Particular type of extraneous variable which for some reason has been left uncontrolled.
- The result is that on looking at the findings of an experimental study, rather than only one possible variable exerting influence on outcome, there are found to be others, which are said to be confounding the results
2 examples of cause Vs confounding variable
- association seen between playing bingo and oral cancer, i.e. higher prevalence of oral cancer among bingo players compared to non-bingo players.*
- But does playing bingo cause oral cancer??
- Confounding variables include:*
- Smoking status (bingo players more likely to smoke than non-bingo players)
- Age (bingo players tend to be older and oral cancer more common in the elderly)
- Drug trial for control of hypertension:*
- Test group (drug); Placebo group (control)*
- Result: test group lower blood pressure than placebo group*
- Confounding: average age of test group significantly lower than control group. Hypertension age related, therefore result may be due to age difference rather than effect of drug.*
- Age difference has confounded the findings
limitations of descriptive epidemiology
- Can only go so far
- Patterns and trends not causes
- Hypothesis generating
- Ecological fallacy
risk factor hypothesis is
- Suggestion that exposure to a particular agent may cause the development of a particular disease if susceptible individual exposed to agent in question
OR
- Suggestion that possession of certain characteristics (e.g. socio-economic status, ethnicity, genetics) may make disease outcome more likely if exposed to certain agents
- From : descriptive epi / clinical impression / lab studies
examples of risk factor hypothesis
- More periodontal disease seen in smoker
- Oral cancer seen more frequently among those consuming high quantities of alcohol
social determinants impact on health outcomes
what to do when examine description epidemiology data
ask question why? -> Generation of hypothesis
e.g.
- Decrease in caries levels associated with War-time sweet rationing
- Higher incidence of oral cancer in West of Scotland compared with rest of UK
- Caries more prevalent in children from low socio-economic groups
investigations in human populations
- Complex in humans – can’t use experimental approach (e.g. can’t give one group of people chemical suspected of causing stomach cancer)
Start with generation of hypothesis
- Should be biologically plausible
Analytic Observational Study rather than experimental studies then required
- related to measures of risk
in animals can conduct experiment
3 common indices of risk
- absolute
- relative
- attritibutable
cohort studies
- Prospective studies
- Recruit group of people who have not manifest the disease at time of recruitment and assess risk factors
- Individuals observed over period of time to measure frequency of occurrence of disease among:
- people exposed to risk factor
- people not exposed to risk factor
e.g. cohort study
Hypothesis:
- placement of amalgam restorations in the teeth of pregnant women increases risk of neurological problems in developing child
Design:
- follow-up infants (all free of ‘disease’ at outset), one group whose mothers received amalgam restorations during pregnancy and the other group whose mothers did not receive amalgam restorations during pregnancy
Follow-up children over a period of years and compare occurrence of ‘disease’, i.e. neurological problems, in the two groups
case-control studies
Retrospective studies
- Compare individuals with disease (cases) with those without disease (controls)
Trace back to assess risk factors:
- Past histories and exposure to suspected harmful agents compared
Less robust than cohort studies
- May be used for preliminary investigation of hypothesis, followed by cohort (if possible)
Value of study profoundly affected by method of selection of cases and controls
- Controls should be random sample of population from which cases selected
- Stratification of control selection important (don’t want them to be significantly different – age, socioeconomic)
