The Role of Clinical Trials Flashcards
HRT Case
Link to Coronary Heart Disease
Studies showed that women taking hormone replacement therapy had lower incidence of CHD.
RCTs demonstrated that HRT caused a slight increase in risk of CHD.
What is the criteria for confirming causalty?
Bradford Hill Criteria
What are the criteria in the Bradford Hill Criteria?
1) Strength
2) Consistency
3) Specificity
4) Temporal Sequence
5) Dose response
6) Experimental evidence
7) Biological plausability
8) Coherence
9) Analogy
Bradford Hill Criteria
What does strength mean
The higher the OR or RR, the stronger the association.
Bradford Hill Criteria
What does consistency mean
The association is consistent when results are replicated in studies.
Using different settings and methods.
Bradford Hill Criteria
What does specificity mean
When single putative cause produces a specific effect.
Bradford Hill Criteria
What does temporal sequence mean
Exposure always precedes the outcome
Bradford Hill Criteria
What does dose response mean
An increasing level of exposure (amount/time) increases the risk
Bradford Hill Criteria
What does experimental evidence mean
Experiments make the inference more plausible
Bradford Hill Criteria
What does biological plausibility mean
The association agrees with current accepted understanding of biological processes
Bradford Hill Criteria
What does coherence mean
Association should be compatible with existing theory and knowledge
Bradford Hill Criteria
What does analogy mean
A finding of analogous associations between similar factor and diseases.
Why may results found be wrong?
Chance
Bias
Confounding effects
How do we protect from chance effecting results?
Statistical significance
Confidence intervals
What is statistical significance?
Influenced by sample size
p<0.05 typically considered to be benchmark of significance.
What is confidence intervals?
Range of values, estimated from a sample within which the true population value is likely to be found.
Greater the CI, less confidence we have about the data.
What is bias?
Systematic error that leads to an incorrect measure of association.
Unlike chance, sample size does not effect bias, and stats cannot be used to adjust for it.
What are the different types of bias?
Selection bias
Information bias
How does selection bias occur?
Systematic difference between:
- characteristics of individuals sampled and population from which the sample is taken from.
- comparison groups within the study population
How does information bias occur?
Systematic difference between:
- comparison groups in way the data was collected
- those measuring the outcome (observer bias)
- study participants (responder bias)
- measurement tools e.g. questionnaires or weighing scales.
How is observer bias overcome?
Blinding
How is responder bias overcome?
Using placebos
What are confounding factors?
May be other factors that are associated with exposure and independently affects the risk of developing the disease.
If prevalence of these other factors differ between groups, the will distort the observed association between disease.
Examples of confounding factors
Gender Age Diet Exercise History of heart disease Other underlying conditions Socio-economic status
How do we deal with confounding factors?
Set restriction criteria for the population.
Use matching of subjects when feasible (although this is difficult and costly)
What are analytical studies?
Observation - observed the natural development without interfering.
Intervention - investigators aim to change the natural course of outcome by manipulating exposure
What types of observation studies are there?
Ecological studies
Cross-sectional studies
Cohort Studies
Case control studies.
What are ecological studies?
- Analyse associations at population or group level.
- Population represents a group of individuals with shared characteristics (e.g. geographical, employment, etc.)
- Utilises data usually collected for other purposes
What type of analysis does ecological studies have?
Data is often continuous (e.g. mortality rate)
Correlation and regression analysis can be applied to the data.
If outcome data can be grouped as two comparison groups, the relative risk can be estimated.
What are the advantages of ecological studies?
Quick
Inexpensive
Generally make use of routine data therefore individuals do not need to be contacted.
What are the disadvantages of ecological studies?
Unable to control for confounding factors as data is often collected for other purposes.
Geographical comparisons may suffer from migration of populations.
Information bias
Ecological fallacy
What is ecological fallacy?
The problem of trying to draw conclusions about individuals from a group-level study.
Basics of cross-sectional study
Present/Future
Present: Disease and exposure
Future: N/A
Basics of cohort study
Present/Future
Present: Exposure
Future: Disease
Basics of case control study
Present/Future
Present: Disease
Past: Exposure
What are cross sectional studies?
Measure frequency of outcomes and/or exposures at one point in time.
Useful for measuring the frequency of chronic outcomes as they include prevalent rather than incidence.
Individuals with acute outcomes may die or recover before detection by survey.
Advantages of cross-sectional studies
- Quick and relatively inexpensive
- everything done in present, taking sample from pop. of interest.
- Divide that sample into cases and controls and exposed and unexposed.
- Assess association between exposure and disease.
- Can be used as a pilot study.
Disadvantages of cross-sectional studies
- Cannot distinguish between exposure effects on duration versus incidence. Unknown if association with incidence or duration.
- Everything measured at one specific time point.
- Not good for rare diseases / short survival as limitation in finding large number of cases.
- One may have to study an extremely large population to get enough cases to have any statistical power.
- Subject to selective survival bias
What is selective survival bias?
If people that are alive have a different risk factor for the disease than those that die early, the sample may be enriched for a particular sub-type of your disease.
What are the two types of cohort studies?
Prospective
Retrospective
What is prospective cohort studies?
Identify ppts and then follow them over time until they acquire outcome or limit for study reached.
Data on exposures are collected before data on the outcome.
Ppts must be free of the outcome at the start of the study.
What are retrospective cohort studies
Look backwards and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of study.
More vulnerable to bias and confounding factors.
Advantages of cohort studies
- Allows the study of a wide range of outcomes that may be associated with single exposure of interest.
- An outcome that was not anticipated can be included in data collection.
Biases can be minimised.
Disadvantages of cohort studies
- Expensive
- Can be long term e.g. decades / lifetime
- Can suffer from loss of ppt from study.
What are case control studies?
Opposite of cohort.
Identify those with and without outcome and then determine their previous exposure to potential risk factors.
Advantages of case control studies
- Quick
- Inexpensive
- Study multiple causes of disease
- Useful for studying rare diseases
- Useful for studying diseases with long latency periods
Disadvantages of case control studies
Vulnerable to bias e.g. recollection bias
Difficulties associated with obtaining reliable data.
