Introduction/ Measures and Frequency of Association. Flashcards
What does a disease determinant result in?
A change in disease status between populations.
Who is widely considered the fist epidemiologist?
Hippocrates (460-370bc).
If a disease is present at stable frequencies what can it be described as?
An endemic.
If something is classed as ‘not deadly’, what can it also be described as?
Trivial.
Acute events rarely occur with a chronic condition. True or False?
False, they can occur.
John Gaunt is widely known for being the first person to systematically collect data, including birth and death rates of populations. What did this collection quantify the death rate of children to be?
What are two other examples of data that were collected?
1/3 dead by the age 5.
This is still the case in developing countries.
- Plague deaths.
- Incidence rate of rickets.
Who is known for the first ever clinical trial? What was it for?
James Lund and scurvy.
What was the first big geograohical study to be conducted?
John Snow/ Cholera.
What observation found in the distribution of colon cancer showed the importance of the environment in this disease?
The migration of the Japanese to Hawaii. Incidence went from being exceptionally rare to the same as the caucasian population within 40 years.
What type of cancer can be caused by Hep C?
Liver.
What is the main determinant of liver cancer in Europe? What did it used to be?
Obesity high alcohol consumption leading to fatty liver disease.
What is a time series?
The distribution and prevalence of something in a population over a period of time.
In the 60’s cardiovascular mortality was much higher in Europe than today. What percentage has it decreased by?
25%.
Despite increased levels of obesity there has been a reduction in systolic bp over the last few decades. Why?
Improved hypertension treatments and reduction of hidden salts in food.
What chronic disease has not changed in levels in the last 40ys?
Cancer.
Is there a law regarding salt content in food in the UK?
No.
In the mid 70’s alcohol was linked to colorectal cancer. What chemical compound was responsible for this link and where was this compound specifically found?
Nitrosamine, found in dark beer.
One the link between alcohol and colorectal cancer was made what intervention(s) was made to prevent this link from occurring in the future?
- Gas ovens used to produce beer were replaced by electric (gas was reacting with natural amines producing the nitrosamine).
- Beer was tested from the 1980’s for nitrosamine.
What three things make up the epidemiological triad?
- Host/ internal environment.
- External environment.
- Agent/ potential cause.
Which is more descriptive/analytical, classical or modern epidemiology?
Classical.
Which is more experimental, classical or modern epidemiology?
Classical.
Which is focused on risk factors opposed to causes, classical or modern epidemiology?
Modern.
Which is focused more on chronic disease, classical or modern epidemiology?
Modern (but classical is also).
Since the development of modern epidemiology diseases have been able to be spit into what?
Subtypes.
Breast cancer has how many known subtypes?
12.
Does breast cancer before or after menopause have an increased risk of being Her2+?
After.
What four diseases combined were responsible for 80% of all NCD’s in 2013?
- CVD.
- Cancer.
- Respiratory disease.
- Broad sense diabetes.
What disease has virtually no known cause apart from high levels of radiation exposure?
Brain cancer.
What type of studies did data from EPIC result in?
Etiological studies.
How many additional people have become obese in the last 25 years?
1 billion.
What does your BMI have to be to be considered obese?
30-35.
What does your BMI have to be to be considered overweight?
25-30.
What does your BMI have to be to be considered malnourished?
<18.5.
What is the immediate problem of the obesity crisis?
The rise of diabetes.
100 million had diabetes in 1980. How many had it in 2010?
400 million.
Most cancers seem to be a ____% higher disk in diabetics.
30-40.
What percentage of women over 50 in Qatar did the Imperial Qatar study identify as diabetic?
50%.
Blood pressure control measures have been successful in Europe and North America, partially due to pharma. Where in the world have they not been successful?
Asia and South America.
Longer life is not thought to correlate with the number of years of healthy life. True or false?
False.
Obesity is not a ‘medical problem’. What is is instead?
A societal problem.
What does causal inference aim to do?
Identify determinants of disease and eliminate the casual effects of the exposure of a disease.
What type of measures do continuous variables provide?
Summary measures.
What type of measures do discrete variables provide?
Frequency measures.
Name 3 frequency measures.
Proportion, rate, odds.
Do the two quantities used to calculate a ratio need to be related?
No.
What is the prerequisite for something to be a proportion?
N needs to be in D.
What does the denominator need to be for the measurement need to a rate?
A measure of person time.
What does prevalence measure?
The extent of existing cases of a health condition of interest by providing a snapshot of a situation.
What is the most common type of prevalence?
Point.
What type of prevalence takes into account the entire period of interest?
Period.
What is point prevalence often expressed as?
A percentage.
What is prevalence good at doing?
Comparing populations.
What needs to be stated when stating a point prevalence?
A given time point.
What is the equation for point prevalence?
Number of existing cases/ total population.
Why would the prevalence of the common cold be calculated as a point prevalence rather than a period prevalence?
The time course is too short.
What sort of disease is it hard to find a prevalence of?
Diseases with a high fatality.
Why is the prevalence of smoking in diabetics less?
As a diabetic who smokes is more likely to die than a diabetic who does not smoke.
What can you not infer from prevalence?
Causation.
When will a case be including in period prevalence?
If it became a case before or during the time period.
What can a person ‘at risk of developing a condition’ be described as?
Disease free.
What two things does prevalence reflect?
The incidence rate (IR) and the duration of the disease (D).
Name three scenarios that will decrease the prevalence of a disease?
- Occurrence decreases.
- Cases are cured.
- Cases die earlier.
What can prevalence be used to assess?
The health status of a population.
What is cumulative incidence defined as?
Proportion of a population with a new event during a given time period.
What does CI measure?
Occurrence of a disease within a population.
Why does CI need a timepoint?
As it is cumulative? Ie it will eventually reach 1.
What is the equation for CI?
Number of new cases during a given time period/ Number of disease free individuals at the start of the time period.
To be included in a CI calculation what must a person have?
The potential to develop the disease.
What is CI also known as?
Incidence proportion/ risk.
What is the definition of ‘Attack Rate’.
Proportion of individuals that develop an infectious disease during an outbreak. (Not a true rate).
What is the definition of ‘Case Fatality Rate’.
Proportion of cases who die among those who develop the disease. (Not a true rate).
CI cannot be estimated from cross sectional studies or surveys. What can?
Prevalence.
What is the survival proportion equal to?
1- CI.
What is the definition of a competing risk?
An event that removes a subject from being at risk from the outcome of interest.
What is the definition of person time?
The sum of the individual time period spent in the cohort by each individual.
Is person time a rate?
Yes.
What is the definition of incidence rate?
Number of new cases during follow up/ total person time contributed by disease free individuals.
How can you minimise competing risk (2 ways)?
- Look at incidence rate/ person time instead of just CI.
2. Reduce the time of follow up.
How can you work out the average density of CI during a short time period?
C/ NΔT. = CI/ΔT.
What does NΔT equal?
Approximate person time during Δt ( a short time interval).
Why is the actual person time slightly less than NΔT ?
As some people will develop the disease in this periodn of time despite the short time period.
What is the equation for working out odds/ proportion?
Odds = Proportion/ 1- proportion.
Why is there more discrepancy between odds and proportions when both get larger?
As proportions can not exceed 1.
Presuming that CI (IR) and D are constant over time and that the population is in steady state what can prevalence be defined as?
Prevalence = IR x D.
CI can be calculated from the incidence rate by what equation?
CI= 1-e^(-IR x t).
What two scales can quantify the strength of an association?
- Multiplicative.
2. Additive.
What is the definition of a counterfactual?
Potential outcomes that would have been observed under each possible exposure/ treatment value. For some subjects these are outcomes that they would not have experienced.
At the start of a study what does each subject have a chance of experiencing?
All counterfactuals.
What would need to happen for a causal relationship to actually be determined?
Each individual would need to experience both counterfactuals. This can’t happen.
What does ‘the endpoint of a casual mechanism’ define?
An effect.
In terms of the Bradford Hill criteria does a small association mean no effect?
No.
What are the 9 points in the Bradford Hill criteria?
- Strength.
- Consistency.
- Specificity.
- Temporality.
- Biological gradient.
- Plausibility.
- Coherence.
- Experiment.
- Analogy.
What does the Temporality part of the Bradford Hill Criteria state?
Effect must be after cause with the expected time delay.
What does the Biological gradient part of the Bradford Hill Criteria state?
Greater exposure should lead to greater incidence. In some cases the presence of a specific factor can trigger an effect.
Absolute effects are differences in incidence rates, cumulative incidences and prevalence. All of these have an additive scale. What sort of effects are relatives of these measures and have a multiplicative effect?
Ratios.
What sort of measure is a risk difference?
An absolute measure.
What is the equation for risk difference?
CID = CI1- CI2.
What is the equation for risk ratio?
RR= CI1/CI2.
What sort of effect does a odds ratio show?
An individual effect.
What is the equation for the odds ratio?
O1/O2.
