What is a census useful for?
Allocation of resources
Population projections
Trends in population eg age, ethnicity
What is a census?
Simultaneous recording of demographic data for all persons in a defined area.
Crude birth rate is..?
Number of live births per 1000 population
What is general fertility rate?
Number of live births per 1000 fertile women aged 15-44
Total period fertility is…?
Average number of live births born to a hypothetical woman in her lifetime
What is considered a more accurate measure of birth rate, crude birth or general fertility rate?
GFR
What is a crude mortality rate?
Average number of deaths per 1000 of the population
What is the age specific death rate?
Number of deaths per 1000 of the population in a specific age group.
What is an incidence rate?
Number of new cases/events per 1000 people per year
What is prevalence?
The amount of people who currently have the disease in a set population with no time frame (snapshot of population at particular time)
What is a confounding factor?
Something associated with both exposure and outcome of interest, but is not the causal pathway between them
What effect do confounding factors have on results of your study?
Lead to distorted and misleading results
What does SMR stand for?
Standardised mortality rate
Why is SMR useful?
Takes into account confounding factors to provide a summation figure describing mortality experienced by a local population compared to general population.
How do you calculate SMR?
Observed number of death (local) divided by expected number of deaths (general) multiplied by 100.
What does an SMR of >100 or <100 suggest?
> 100 suggests excess mortality with confounding factors controlled for and vice versa
What is an incidence rate ratio?
Incidence rates of two populations of varying exposure compared to see if exposure causes certain diseases.
Define ‘confounder’
Something which is associated with both exposure and outcome but is not the causal pathway between them.
What is variation in an epidemiological study?
Difference between observed and ‘real’ value
Why are error factors and confidence intervals produced?
To allow for variation
What is a confidence interval?
A range of values around the observed that we can have 95% confidence the real value lies within.
How is error factor used to produce confidence intervals?
Lower bound - value/error factor
Higher bound - value*error factor
What are the 4 values we calculate error factor for?
Incidence rate
Incidence rate ratio (1/d + 1/d)
Prevalence
SMR
What is the p value?
The probability of obtaining a test statistic
What is the significance of p<0.05?
Indicates there is enough evidence to discard the null hypothesis, data cannot be put down to chance.
If there is insufficient evidence to discard null hypothesis what will he p-value be?
p>0.05
What is biasing?
The deviation of results from the truth due to various processes.
What is selection bias?
Give 2 examples
Error due to systematic differences in the way two groups are collected in a study.
Healthy worker effect
Allocation bias
What is information bias?
Give 2 examples
Error due to systematic differences in the way subjects are classified in a study
Recall bias
Publication bias
What is a cohort study?
Recruiting disease free individuals and classifying them by different exposures and filling them up for a long period of time, disease progress is monitored and incidence rate is calculated.
What are cohort studies good for studying?
Rare exposures or diseases that take a long time to develop
What are the two types of cohort studies?
Prospective - disease free individuals recruited at present time and followed up
Retrospective - disease free individuals recruited and historical records are used to classify exposure and followed up
What is the difference between internal comparison and external comparison in a cohort study?
What calculations are used in each?
Internal comparisons occur when you have sub-cohorts in your sample and compare exposed to unexposed
Relies on IRR calculations
External comparison occurs when you have an exposed population which you compare against your reference population instead.
Relies on SMR calculations
What is the Healthy worker effect?
Biasing of results when a study involves workers that are compared to a reference population, this is a form of selection bias as a working individual is more likely to be healthy than an unemployed worker.
How might you control for the healthy worker effect?
Any comparisons made should be between two groups of working individuals to prevent bias
What is a Case Control Study?
Involves recruiting disease free individuals (controls) and diseased individuals (cases) and then determining exposure status and comparing.
Uses an Odds ratio to determine the risks conferred by an exposure.
How do you calculate odds ratio?
(exposed cases x unexposed controls) / (exposed controls x unexposed cases)
What is the null hypothesis for a Case Control Study?
Odds ratio will be 1 giving the same odds for developing disease in exposed and unexposed
What is a Case Control Study useful for studying?
Rare disease
How is error factor calculated for a Case Control Study?
(1/a) + (1/b) + (1/c) + (1/d) is square rooted to represent all groups in the calculation
Sorry I cant put in the full equation, kinda hard on this crappy card creator.
Why is it useful to find a large number of controls for a Case Control Study?
What is the normal ratio of cases to controls?
Used to decrease the error factor and hence confidence intervals as the case group may be rather small due to having rare disease (fun fun fun)
Usually a 5:1 ratio of controls to cases to minimise the error factor as much as possible
What bias is a Case Control Study particularly sensitive to?
Selection bias
Controls may not be representative of gen. pop.
for example in a CCS about rare heart disease controls may be recruited from the same ward and so will likely have similar risk factors/exposures.
How are confounders controlled for in Case Control Studies?
Matching up cases and controls with similar details
Outline the process of performing a randomised control trial
Identify a source of eligible patients
Invite eligible patients to trial
Randomly allocate patients to control or treatment
Follow up patients in identical manner
Attempt to minimise losses to follow up and maximise compliance
Analyse data and obtain results
What is the effect on results of randomisation in a randomised control trial?
Removes any confounders
What is the process of double-blinding and why is this important?
The patient and doctor are both unaware of which treatment is being given to who
Removes selection bias
Prevents doctors placing healthier patients in new drug group to produce more positive, statistically significant results
If no comparative therapy is available for comparison against a new therapy, then how is a control group established?
Placebos are used to compare against new therapy
What is intention to treat analysis and why is it advantageous?
ITT analysis includes non-compliers and crossover patients in the results
Advantages:
No loss of randomisation
Presents a more accurate picture of how the new therapy performs in clinical practice
What are the Bradford-Hill criteria used for?
To determine wether a causal-effect relationship exists between exposure and outcome (once confounders, bias and chance have been removed)
What are the 9 Bradford-Hill criteria?
Strength of association Specificity of association Consistency of association Temporal sequence Dose response Reversibility Biological plausibility Coherence of theory Analogy
A trial’s fulfilment of any Bradford-Hill criteria indicates what?
The relationship between outcome and exposure more likely to be causal
What can be deduced from the strength of association between outcome and exposure?
The stronger the association (eg. Higher IRR or OR) the more likely the relationship is causal
Why does specificity of association affect our assessment of whether a relationship is causal?
If outcome is associated with a specific factor/exposure then the relationship is more likely causal
What is meant by ‘consistency of association’?
The association occurs in other studies too
This suggests causality
What is meant by temporal sequence and how does this help show causality?
Temporal sequence is the order of exposure and outcome
If exposure doesn’t precedes outcome then the relationship will not be causal
What is meant by ‘Dose Response’ and what does it suggest?
Dose response is differing levels of exposure leading to different levels of outcome, this suggests causality
What is meant by ‘Reversibility’? (BH criteria)
Removal of the causative factor causes reduced risk of outcome in a causative relationship
What is meant by ‘Biological Plausibility’? (BH criteria)
Plausible biological mechanism should support observations, this suggests causality
What is meant by ‘Coherence of Theory’? (BH criteria)
If observations confirm scientific thinking then they suggest a causative relationship
What is meant by ‘Analogy’? (BH criteria)
If another similar disease has similar outcomes then it suggests causality
When undertaking a randomised control trial how can patient losses be minimised?
Follow up at practical times
No coercion or inducements to join the study in the first place
Honesty to patients
How can compliance be maximised in a randomised control trial?
How can compliance be monitored?
Simple instructions
Patient allowed to ask questions
Trial made simple and accessible for the patient
Compliance can be monitored by blood tests, urine tests etc
What are the main ethical principles underlying the conducting of a randomised control trial?
Clinical equipoise Scientifically robust Ethical recruitment Valid consent Voluntariness
What is clinical equipoise and why is it a good ethical principle?
What criticism could be made of this principle?
Clinical equipoise is the reasonable uncertainty into which drug administered to patients in a trial is better for the patient
Therefore we are not knowingly subjecting patients to a known less effective treatment
This is not always possible, for example in placebo controlled trials
What is meant when we say ‘this trial is scientifically robust’
A scientifically robust trial will be in pursuit of knowledge that will benefit the general population
What two factors comprise ethical recruitment to a study?
Recruitment must be from a population that any gained knowledge will benefit
There are no unethical exclusions from the trial
What is valid consent?
Consent which is given with sufficient knowledge, a cooling of period and the ability to withdraw from a trial at any time
What is meant by ‘voluntariness’? (ethically)
No coercion or manipulation used to gather subjects for a study
What are the 4 basic principles of any medical role?
Beneficence (Do good)
Non-maleficence (Do no harm)
Autonomy (Let the patient decide)
Justice (Be fair)
What is a systematic review?
What two factors are systematic reviews reliant on to produce unbiased results?
A compilation of primary studies (normally RCT, could be case control or cohort) for the purpose of reviewing the best evidence available on a particular topic
Lack of publication bias
High quality of primary studies
Studies used in a systematic review should be?
Relevant and credible
Transparent
Reproducible
Explicit
What is a meta-analysis and how are the results of a meta-analysis shown?
Provides quantitative assessment of studies used in a systematic review
Provides an overall value associated with confidence intervals
Results shown via forest plots
In a forest plot, what is shown by:
Squares
Lines
The diamond
Squares show the odds ratio of individual studies, size of square is proportional to the weigh given to the study
Horizontal lines associated with squares show the 95% confidence interval
Vertical line in the centre shows null hypothesis
The diamond represents pooled estimate odds ratio, with it’s width representing the 95% CI
What is meant by homogeneity and heterogeneity in the context of systematic reviews?
The degree to which studies in the systematic review are similar is expressed as either:
Homogeneity: Degree to which studies identical
Heterogeneity: Degree to which studies are varied
What are the two models can be used to analyse studies within a systematic review?
Fixed effect model
Random effects model
Describe the fixed effects model
Assumes the studies used in a systematic review are homogenous and any variation in data comes from within-study variation
Describe the random effects model
Assumes the studies used in a systematic review are heterogenous and variation between dat comes from within study variation and between study variation
What is publication bias?
How is publication bias relevant to a systematic review?
How is publication bias assessed?
Publication bias is whereby studies are only likely to be published if the results are statistically significant or have a large sample size.
If systematic reviews only include published studies, results will be biased, therefore they should also use data from unpublished studies
Publication bias is determined using funnel plots
What are the two axes of a funnel plot?
How will a funnel plot vary between biased and unbiased systematic reviews?
y = Sample size of study x = OR
Unbiased = Funnel shape when all studies plotted
Biased = Vary in shape/not funnel shaped
What are the advantages of conducting a systematic review?
Explicit methods can reduce bias and exclude poor quality studies
Meta-analysis provides overall figure for the studies
Large amounts of information can be assimilated quickly by healthcare professionals
Reduction in time between research discovery and implementation of clinical use
Used in evidence based practice guidelines
What can be said about the test for heterogeneity?
How is homogeneity/heterogeneity expressed numerically?
Weak statistical test
Expressed as a P value
p=>0.1 indicates homogeneity
p=<0.1 indicates heterogeneity