Flashcards in Epidemiology study designs and measures Deck (54)
What are the defining features of a cross sectional study?
A survey of a random sample of people (aimed to be representative of the population) at a given point of time. This is usually to find associations and risk factors for disease by a questionnaire on behaviour. Cannot be used to find causation.
What is an example of a cross sectional study?
How many people in area X have heart disease? Taking a sample of the population can find prevalence of disease at that time and use questionnaires to find associations (Do you smoke?/How much red meat do you eat per day?/Do you exercise?/What are your feelings towards your local health service?)
What are the advantages of a cross sectional study?
-Many variables compared at ones
-Quick and cheap
-Useful when routine data isn't available
What are the disadvantages of a cross sectional study?
-Not useful for a disease with a short duration (many will be missed in a snapshot study/will get different results every time you check)
-Cannot find causality
-Easy to introduce sampling and collection bias
-Can't find incidence
-Questionnaires can often require a pilot study
What are the defining features of a cohort study?
Used to find the effect of an exposure. Take 2 groups of people with a disease and varying degrees of exposure to something you suspect is affecting the disease and follow over time. Longitudinal study where comparisons are taken continuously. Association between risk and outcome usually measured as risk ratio/odds ratio.
Name a type of descriptive observational epidemiological studies
Name 2 types of analytical observational epidemiological studies
Name 3 types of experimental epidemiological studies
Experiments, randomised control trials, intervention studies
What are cohort studies used for?
To test if an exposure variable is causally related to health
To measure incidence and incidence rate over a period of time
What are the 2 types of cohort study?
Prospective (2 groups, 1 exposed, follow over time to see effect)
Retrospective (2 groups, 1 has been exposed, look back in records to see effect)
What are the advantages of a cohort study?
-Allows outcomes to be explored over time
-Possible to study a harmful risk factor/exposure (it would be unethical to expose a group to this risk factor in an experimental setting)
-Useful for rare exposures (because you choose based on exposure)
-Can look at multiple exposures
-Can measure incidence and prevalence
-Samples aren't biased by knowing the outcome status - samples are defined before the disease is onset
-High validity, true incidence known
What are the disadvantages of a cohort study?
-Can be very time consuming and expensive
-Behaviours can change, control groups can become exposed or exposed group stop being exposed
-Change in measurements/classifications etc can occur over the time the study takes (can cause misclassification
-Selection bias, consent bias
-Loss to follow up, people might bail (might introduce bias, by which people bail)
-Not suitable for rare diseases, might never come up
What is the Framingham Heart Study and why was Framingham a good place for it?
Longest running cohort study of cardiovascular disease in USA.
-Stable employment and population
-Local authority kept a list of residents
-One health facility so easy to follow up
-Variety of socioeconomic and ethnic subgroups
-Previous successful study of TB
What analyses are associated with cohort studies?
Number of instances of illness commencing during a specified period
What is incidence rate?
Number of people who develop a disease in 1 year/Average number of people in one population
What is cumulative incidence?
Number of people who develop disease in period/Number of people at risk of getting the disease at start of period
You want to find out if there are patterns, relationships & associations. Which study?
You want to know how widespread the event occurrence (prevalence) is. Which study?
You want to know how quickly an event is occurring (incidence). Which study?
Which factors are causing the event?
Case control study
How effective is an intervention at controlling/preventing/managing an event?
What's the difference between a case-control study and a retrospective cohort study?
In a case-control study the investigator splits the group by disease and examines how exposures differ.
In a retrospective cohort study the investigator splits the group by exposure and examines how outcomes differ.
What are the usual sources of controls for a case-control study and what are their limitations?
Hospital - ill by definition, increased number of smokers and alcohol abusers
Relatives - very (too) willing to participate, genetic similarity, anxiety related to the case individual
General population - unwilling to participate, uncooperative
What is the structure of a case-control study?
Group with outcome + group without outcome and you look back in time to see who is exposed to the variable of interest. Used to test if an exposure variable is related to health problem.
Which analyses can you do from case-control data and which can you NOT do?
Can test the null hypotheses (chi-sq and fishers exact test) and measure the magnitude of the difference (odds and odds test)
Can NOT measure risk because we chose case and control so there's no "at risk population"
What do you have to think of when choosing controls for a case-control study?
Without the disease but would have been chosen for the cases if they had had the disease (e.g. totally comparable to cases)
What do you need to think of when defining cases for a case control study?
Look for previous definitions in literature
Needs to be reproducible by anyone so needs to be a clear definition
Precision needs may vary (epidemic vs clinical trial)
Clearly list the criteria needed for a positive diagnosis
Include certainty of diagnosis
Include severity of disease
What is the difference between a risk and a hazard?
A hazard is something that if you are exposed to it, increases your chance of a negative outcome (e.g. being bitten by a plasmodium infected mosquito).
A risk is the probability that an outcome will occur within a given time period (e.g. the risk of developing malaria is lower for children that sleep under a mosquito net).