Epidemiology Flashcards
Define prevalence
Define incidence
Define Mortality
prevalence- no of people with a problem in a define population at one time.
Incidence- no of new cases of a problem arising in a define population in a defined period of time.
Mortality- no of people dying in a defined population in a defined period of time
Epidemiology definition
study of disease in populations
Concepts of causality:
What is the difference between the deterministic approach vs the stochastic approach
1) Deterministic approach - A causes B: inevitabilty.
Validation of hypothesis by systemic observations to predict with certainty future events. i.e Tubercle bascilis causes TB.
2) Stochastic approach- A increases likelihood of B - probability:
Assessment of hypothesis by systemic observations to give the risk of future events
Overcrowded accomodation increases risk of Tb
What are the differences between the deterministic and stochastic concepts of causality?
Deterministic:
Newtonian thinking, objective quantifiable and certain , the whole is the sum of the parts, useful in thinking about single cause for a single disease
Stochastic:
quantum thinking, whole greater than the sum of the parts, whole is not predictable from knowledge of the parts, probabilities of certainties, systems theory, observer influences the observed.
What is a confounding factor?
A confounding factor describes interference by a third variable so as to distort the association being studied between the two other variables, because of a strong relationship with both other variables.
For example, smoking and obesity are closely linked, therefore when taking into account the effects of smoking on heart disease, obesity could distort this relationship.
What is a mediating variable?
A mediating variable is the variable that explains the relationship between the dependent variable (the outcome being measured) and the independent variable (The factor changed to measure outcome).
Mediating variable = variable through which exposure (independent variable) wholly or partially exerts its effect:
E.g. obesity is our measured factor and directly increases risk of heart disease- is measured. But sugar intake linked to obesity and is directly linked to heart disease via diabetes. Sugar intake = mediating variable.
What is reverse causality?
Where the factor you are measuring has a direct effect on the outcome, but the outcome itself has a direct effect on the factor you are measuring. I.e unemployment and mental illness.
Unemployment can lead to mental illness, yet mental illness could lead to unemployment.
Can epidemiological studies prove causality?
Epidemiological studies cannot prove causality but in assessing prbability they can make a case “beyond reasonable doubt”.
What are the three association features are included in bradford hills criteria for inferring causality (1965)
1) strength of association
2) specificity of association
3) consistency of association
What is strength of association?
where a causal link is more likely with strong associations - commonly measure by rate ratio/ odds ratio e.g. heavy smokers have 20 x higher risk of mortality from laryngeal cancer than non smokers.
Strong associations unlikely to be explained by undetected confounding or bias (note-not always true.)
What is specificity of association?
Where a causal link is more likely when a disease is associated with one specific factor and vice versa. E.g. asbestos and mesothelioma.
However remember lack of specificity does not necessarily weaken case, e.g. tobacco and multiple cancers/ diseases.
What is consistency of association?
where a causal link is more likely if the association is observed in different studies and different subgroups.
- Consistency of association between studies or groups is unlikely to be due to the same confounding factor or bias
- Lack of consistency can be due to features of study design
E.g. many different studies demonstrated association between smoking and ischaemic HD.
In bradford hill’s criteria for inferring causality (1965) what three exposure/ outcome features are included?
Temporal sequence
Dose response
Reversibility
What is temporal sequence?
Where a causal link is more likely if the exposure to the putative cause has been shown to precede the outcome.
(i.e the causal link cannot exist if the outcome preceded the exposure to the putative factor).
Good study designs for temporal sequence are prospective cohort study and randomised controlled trials
Weak study designs are cross sectional, and case control study
What is the dose response (biological gradient)
where a causal link is more likely if different levels of exposure to the putative factor lead to different risk of acquiring outcome.
Dose response (biological gradient) is unlikely to be due to unknown or confounding bias.
- note lack of biological gradient does not rule out causal link (e.g. think threshold effect or U shaped relationship).
What is reversibility?
Where a causal link is very likely if the removal or prevention of the putative factor leads to a reduced or non existent risk of acquiring the outcome.
Probably the strongest evidence for a causal link but it is often difficult to demonstrate: - many diseases have long time lags (e.g. cancer development after asbestos).
Ethical issues (e.g. RCT on smoking vs non smoking).
What are three other factors/ other forms of evidence (not association features or exposure/ outcome features) that are included in bradford hills (1965) criteria for inferring causality?
1) coherence of theory
2) biological plausibility
3) Analogy
What is the coherence of theory?
Where a causal link is more likely if the observed association conforms with current knowledge
Coherence with current paradigms / constructs/ theories strengthens the case for a causal link.
Problem: can lead to inappropriate rejection of “unfavoured” associations
Lack of coherence does not rule out a causal link.
What studies are included in observational studies?
- Cross sectional surveys
- Cohort studies
- Case control studies
What are cross sectional studies?
- observational study is a study that analyses data from a population, at a specific point in time (snap shot of a population at a certain time). (e.g. prevalence of breast cancer in a population.)
- Prevalence of specific disease or distribution of specific disease in a population
- monitoring health over time
- medium cost
What are case controlled studies?
- In a case controlled study, patients who have developed a particular disease are identified and their past exposure to suspected aetiological factors is compared with that of controls that do not have the disease.
- Always set up w specific purpose, to investigate the suspected determinants
- quick and cheap
- However recall and selection bias is an issue (Selection bias where you are unaware you have selected biasly as the study shows the outcome predicted.)
What is a cohort study?
- Cohort study –> Study where one or more cohorts are followed and subsequent evaluations with respect to disease/ outcome are measured to determine which exposures are associated with it.
- may be set up for specific purpose, often multipurpose e.g. effects of smoking or abestos
- Determinants of common conditions and their relative importance
- very slow and very expensive
- issues with confounding and unknown risk factors
What are some experimental study designs?
- Randomised controlled trials
- controlled studies
- natural experiments –> may be only way to measure something, individuals exposed to experiemental and control conditions via nature/ factors outside control of investigators, issues with confounding factors
- uncontrolled studies –> where a factor is measured but there is no control data, do not know what happens without intervention
What is the hierarchy of evidence?
- Systematic reviews highest
- then experimental studies –> RCT and controlled trials
- Observational studies –> cohort studies and case control studies
- Descriptive studies –> cross sectional and qualitative studies
What types of bias are there in epidemiological studies?
- Selection bias –> during design phase and execution (e.g admission to study, prevalence/ incidence, detection, volunteers, loss to follow up).
- information bias –> data collection phase (e.g. due to interviewer, questionnaire, recall, diagnostic suspicion and exposure).
- confounding bias –> where the relationship between independent variable and dependent variable is altered by another closely related factor.
