Module 10 Flashcards
Identifying Causes of Disease and Evaluating Treatment
What is the Epidemiologic Triad
Host (e.g. genetics, age)
Agent (e.g. bacteria, chemical)
Environment (e.g. climate)
All diseases are multifactorial
What are the types of causes
Cause
Necessary cause
Sufficient cause
What is cause
condition/event/characteristic that contributes to a disease
What is Necessary cause
condition/event/characteristic that MUST be present for disease to occur
What is Sufficient cause
set of MINIMAL conditions/ events to produce disease
Modern criteria for causality
• Coherence with known facts • Consistency with repetition • Time sequence • Specificity • Strength of association • Dose-response relationship
What is Deductive reasoning
Assuming something with a known fact
– All men are mortal
– Socrates is a man
– Therefore Socrates is mortal
Inductive reasoning
– requires qualitative and quantitative evidence to suggest that rejecting the supported conclusion is irrational
“All crows we have seen so far are black. Therefore, all crows are black.”
What is the relationship between hypothesis and data
Deduction and induction reasoning are always done
What is causality
The real reason of the cause
What is the strongest way to prove causality
Casual judgement, stronger as there are more evidence
What is syndrome diagnoses
– fixed combination of characteristics e.g. idiopathic cardiomyopathy in dogs, sudden death in ruminants
What are the types of evidence
• Personal observation and anecdote
– “I tried this in 3 animals and they all got better”
– POOR source of information
• Scientific studies
– best source of information
– properly conducted
– different types provide different information
What are the clinical research, what are the 3 types and their quality
• Goal - infer both causation and treatment effects • Descriptive studies – poor quality • Experimental studies – highest quality • Analytical epidemiological studies – provide good quality
What are descriptive studies
essentially describes what happened, without the analytical advantages of cases vs controls or other factors that can be compared and tested statistically.
• Case reports
– rare condition
– unusual presentation of a disease
– cause/treatment effects?
• Case series
– number of animals
• Survey
– prevalence within a population
NO CONTROL GROUP OF ANIMALS
What are experimental studies
specifically designed for the purpose, with controls, animals allocated (randomly) to treatments, often blinded so that there is no observer bias. • Compare two groups – disease cause – treatment effect • Laboratory experiments • Clinical trials • Planned, controlled allocation • Expensive • Ethics
What are analytical eqpidemiological studies
Analytical epidemiological study – takes advantage of naturally-occurring circumstance (or circumstances not directly designed by the researchers). Three main types exist:
• Deriving information from natural cases • Planned comparison between two groups – not allocated by experimenter • 3 types – cohort studies – case-control studies – cross-sectional studies • sample population for disease/exposure at same time
What is cohort study
Cohort study – animals exposed or not exposed to the hypothesised cause (or protectant) of interest are followed forward through time to see if there are differential rates of disease in the two (or more) groups
– select puppies and follow over time
– disadvantages?
– cf. experimental study
What are case-control studies
Case-control study – animals with or without the disease are tracked backwards in time (such as through their medical records) to see if they had differential exposure to the hypothesised cause of interest
– identify animals with and without mammary tumours
– investigate if obese as pups
What are hypothesis
assertion or conjecture concerning one or more populations
What about hypothesis
• Cannot be proven, but can be rejected
• Knowledge accumulates by falsification
– accepting a hypothesis means we do not have enough evidence to reject it (NOT that it is true)
What is null hypothesis
• No difference exists between groups
– hope to reject
• Alternate hypothesis
– a difference exists
• Underlies scientific method
– not just in clinical research
When can we reject the null hypothesis?
• Almost always a difference between groups
– chance (random error)
• sample of population
– bias (systematic error)
• poor study design
– real effect
• Require sufficient evidence to rule out bias and chance to reject the null hypothesis
What is bias
Systematic error in design, conduct or analysis
–a valid study is (relatively) free from bias
– implies accuracy
What are the three types that cause bias
– selection – study groups not representative of target population
• randomisation
– information
• blinding
– confounding
• only variation should be the factor you are trying to test the effect of
How do we rule out bias?
Examine materials and methods section – was it a “fair” comparison
– allocation of treatment?
– comparable treatment groups?
– procedures well defined?
– outcome well-defined and assessed without considering treatment?
– follow-up identical?
– masking of treatments?
What is P-value
• Is the observed difference between two groups likely to have occurred due to chance?
– probability of having obtained the data if the null hypothesis is true
• small P values (<0.05)
– reject chance as a likely explanation for observed difference
How to increase precision in a study
- Precise study relatively free from random error
- In general, increasing study size will increase precision
• Precise study
– favours rejection of null hypothesis (smaller P-value)
• Imprecise study
– favours acceptance of null hypothesis (larger P-value)
How to assess precision
• 95% confidence interval
– 95% chance that the real population parameter lies within this range
• Smaller the 95% CI(confidence interval), the more precise the estimate
– i.e. the higher the level of variance, the wider the 95% CI
What does precision affect
affects P value and rejection/ acceptance of null hypothesis
Power of study
• Accept null hypothesis does not necessarily mean treatments are equivalent
– biological significance
– size of CI
– actual P value (0.08 vs. 0.8)
• Power of the study
– probability the null hypothesis will be rejected, when there is a REAL difference of a given magnitude between treatments
How should you estimating the power before the study begins
– specific alternate hypothesis
– decide what magnitude of difference is important e.g. 60% for treatment Avs 80% for treatment B
• Power of 80% considered reasonable
How do you determine biologically significant
Consider the point estimate – e.g. 60% vs. 75%
How to you determine the precision of the study
Look at confidence interval
How do you rule out the chance for the observed treatment?
Look at P-value
affected by size of study as well as magnitude of difference between groups
What is cross-sectional studies
Cross-sectional study – animals with or without the disease at this point in time are also checked to see if they have the hypothesised cause of interest or not.
Example of cross-sectional studies
n a cross-sectional study, the animals already have the disease (or not) and you are testing to see whether they also have the hypothesised causative factor (or not). An example would be where you are interested in the relationship between skin pigmentation and skin cancer in older cats. You divide cases at your clinic into cats with or without skin cancer and also categorise them into white cats and non-white cats. You analyse to see if there is a greater prevalence of skin cancer in white cats.
Example of cohort study
the animals are divided into exposed and non-exposed (to the hypothesised causative factor), but they do not have the disease at the start. You then track them over time, to see if they develop the disease at different rates. In this example, you would identify young cats that are white in colour, plus (otherwise matched) young cats that are not white. You then follow them over time (assuming that most keep visiting your vet clinic) to analyse whether they differ in their rates of developing skin cancer.
