HaDPop Flashcards
What is a census useful for?
Allocation of resources
Projections of populations
Trends in populations
What is crude birth rate?
Number of live births per 1000 population
What is general fertility rate?
Number of live births per 1000 fertile women aged 15-44
What is total period fertility rate?
Average number of children born to a hypothetical woman in her lifetime
What is the difference between fecundity and fertility?
Fecundity is the physical ability to reproduce whereas fertility is the realisation of this as births
What is age standardised death rate?
Number of deaths per 1000 in age group
What is standardised mortality rate?
Comparison of observed and expected death rates if the age-sex distribution is adjusted for
What is incidence?
Number of new cases a year
A rate b/cod the time dimension
How is incidence calculated?
New events/(person x time)
What is prevalence?
Number of people in a population with existing disease
Proportion
How is prevalence estimated?
Incidence x length of disease
How is incidence rate ratio calculated?
Exposed/unexposed
What are the different types of risk associated with rate and ratio?
Rate = absolute Ratio = relative
What is an observed value the best estimate of?
True/underlying tendency
What does a null hypothesis assume?
No difference b/w two outcomes
What does p>0.05 indicate?
Hypothesis can neither be rejected or accepted
What is the 95% confidence interval?
Range within which we can be 95% certain the true value of the underlying tendency lies
Is the observed value always in the 95% CI?
Yep
How is the confidence interval calculated?
Calculate observed value
Calculate error factor
Lower limit = observed/e.f.
Upper limit = observed x e.f.
How do you conduct a cohort study?
Recruit disease-free cohort –> follow over time –> incidence ratio –> calculate incidence rate ratio (relative risk)
What are the advantages of cohort studies over routinely available data?
Study unusual exposures and personal characteristics
More detailed info on outcomes and exposures
Additional data on confounding factors can be collected after trail starts
What is a census?
Simultaneous recording of demographic data to all persons in a defined area
How is a concurrent/prospective cohort study conducted?
Recruit outcome free individuals and classify by exposure status –> follow up counting p-y and cases
When can data collection be carried out for a prospective cohort study?
Immediately or delayed
Why may there be a delayed start to a concurrent cohort study?
Organisation
Ensure participants are disease free
How is a retrospective/historical cohort study conducted?
Recruit disease free individuals and classify on initial exposure status –> count p-y and cases using data records
Can a historical cohort study be continued into a concurrent study?
Yes
What is internal comparison in a cohort study?
Comparison of sub-cohorts
Why may there be a large error factor in an internal comparison?
Sub cohorts may be radically different sizes
Sub cohorts may not be comparable due to confounding
What is used to numerically evaluate internal comparison?
IRR
What must be large to give a small error factor when conducted an internal comparison?
Both sub-cohorts
What is external comparison in a cohort study?
Comparison of a cohort with a reference population after standardisation methods have been used
What is used to compare numerical data in external comparison of cohort studies?
SMR
What is used to calculate the expected cases in the reference population for an SMR in external comparison of a cohort study?
Lexis diagram
What determines the size of the SMR in external comparison of cohort studies?
The smallest cohort
Why is the error factor for internal comparisons larger than for external comparisons in cohort studies?
IRR = two terms in e.f. expression SMR = one term
Is the confidence interval larger in an external or internal comparison of a cohort study?
Larger for IRR therefore larger in internal
What are the disadvantages of an external comparison in a cohort study?
Limited data for reference population
Often no incidence data
Usually compromise with mortality data
Selection bias - healthy worker effect
What is the healthy worker effect?
Occupational cohorts yield SMRs
What are the advantages of using a concurrent cohort study?
Allow detailed and prospective assessment of exposure, outcomes and confounders
What type of conditions are cohort studies better suited to?
Fluctuate randomly or systematically w/age
What advantages do cohort studies have over case-control?
Better for studying a range of outcomes
Better for studying a rare exposure
Establishes exposure preceded outcomes
What are the disadvantages of using a cohort study?
Large and resource intensive Take a long time Survivor bias to follow up Ethical dilemma and political change effects due to long duration Not good for rare outcomes Difficulty w/confounding
How is a case-control study conducted?
Identify group of cases
Ascertain previous exposure status of everyone
Compare level of exposure in cases and controls
What numerical method is used to examine case-control studies?
Odds ratio
How is an odds ratio calculated?
ad/cb
What is the rare disease assumption?
If both unexposed and exposed case numbers are much smaller than control numbers we can assume IRR ~= ad/cb
How does a case-control study compare outcomes?
Based on outcome status
How does a cohort study compare outcomes?
Based on exposure status
What is the economic advantage of case-control studies over cohort?
Quicker therefore cheaper
Is a cohort study or case-control study better for rare exposures?
Cohort
Is a cohort study or case-control study better for rare outcomes or diseases?
Case-control
Which two types of bias are case-control studies prone to?
Selection
Information
How does a cohort study limit bias?
By ascertaining exposure status
What is a nested case-control study?
Collection of data from evolving outcome and exposure database of concurrent cohort study
Which type of study can directly measure incidence?
Cohort
What factors suggest a case-control study would be suitable?
Rare diseases
Look into many different exposures at once
Absolute incidence rates not needed
Why can you not use an IRR to evaluate a case-control study?
Cannot be sure you have the right denominator
What is the precision of an OR affected by which does not affect an IRR?
Number of healthy people
How many times more controls than cases do you typically use in a case-control study?
5
What gives the largest error factor in calculating the odds ratio for a case-control study?
Smallest of a,b,c or d
What is the most difficult aspect of a case-control study to control?
Selection bias - participants not representative of general population
Which bias are case-control studies susceptible to?
Selection
Recall
Systematic
Why are case-control studies subject to recall bias?
Exposure status is incorrectly determined due to looking back at history to determine exposure
How does systematic bias arise in case-control studies?
Assessor bias
Data collection methods differ
How are confounders removed from case-control studies?
Match up cases and controls with similar details
What is a necessary exposure?
Always precedes disease
What is a sufficient exposure?
Can cause disease in its own
In epidemiology, what is a cause?
Exposure of factor that increases the probability of disease
What can apparent associations arise from?
Chance Selection bias Information bias Confounding Reverse causality
What is selection bias?
Error due to systematic differences in the characteristics of the groups being studied due to differences in the way they were selected
What is information bias?
Error due to systematic differences in the measurement or classification of subjects in the group being studied
What is confounding?
Something which has an effect on both the exposure and the outcome
What is reverse causality?
Cause-effect relationship exists in wrong direction
What is association when referring to epidemiology?
Statistical dependence b/w 2 or more events
Which types of studies are classed as analytical studies?
Cohort
Case-control
Which study is an example of an experimental study?
Random control trial
As well as analytical studies, what do observational studies encompass?
Prevalence studies
What were Henle-Koch’s postulates?
Agent not found in other disease cases
Agent in every case by isolation in pure culture
Once isolated, agent can reproduce and be recovered from experimental disease
Which is the strongest Bradford-Hill criterion?
Reversibility
What are the 9 Bradford Hill Criteria for inferring causality?
Reversibility Coherence of theory Specificity of association Analogy Consistency of association Biological plausibility Dose response Temporal sequence Strength of association
What is reversibility?
Removal/prevention of putative factor –> reduced/non existent risk of outcome
What is specificity of association?
Association specific to exposure - outcome association more likely to be causal
How does analogy infer causality?
If one exists –> more plausible
What is consistency of association?
Demonstrated by different studies, different people, different times and different places more likely to be causal as unlikely to be explained by same confounding or bias
What is biological plausibility?
Biologically plausible mechanism infers causality
What is biological plausibility limited by?
Current knowledge
What is does response?
Varying exposure gives varying outcome - confounding/bias unlikely to operate at same degree
What is temporal sequence?
Association in which putative factor precedes outcome more likely to be causal
What is strength of association?
Strong association –> more likely to be causal
What is coherence of theory?
Conforms w/current knowledge and theory
What are 2 fundamental assumptions in any epidemiological investigation into a disease?
Disease does not occur at random
Disease has causal and preventable factors that can be identified through systematic investigation
What are the four steps of epidemiological reasoning?
Hypothesis
Analytical study
Observed association - exclude alternative associations
Does statistical association represent cause-effect?
What scheme is used for the assessment of causation?
Description of the evidence
Internal validity: non-causal associations
Internal validity: features of causality
External validity: generalisation of results
Comparisons w/other evidence
What 3 questions are considered when assessing non-causal associations in internal validity?
Has methodology minimised bias?
Any potential confounders?
How likely is it results are due to chance?
What are four questions considered when assessing features of causality?
Strong relationship?
Dose response relationship?
Specificity of association?
Correct time relation/reverse causality?
What is considered using the study results during generalisation of results?
Applied to source population
Applied to other population
Relevant to your population?
What is asked when comparing results with other evidence when assessing causation?
Consistent w/more powerful study design?
Specificity?
Biological mechanism?
What is the basic reproductive rate of infection, R?
R = B x D
B = transmission coefficient D = time period of infectivity
What just a clinical trial be?
Fair - unbiased w/out confounding
Controlled - comparison of interventions
Reproducible - in experimental conditions
What aspect of a RCT that differs from a cohort study makes the RCT unethical?
Treatment allocated by trial investigator so the investigator is taking part in the trial
What is efficacy when talking about health care interventions?
Ability to improve the health of a defined group under specific conditions
What are the disadvantages of non-randomised clinical trials?
Allocation bias
Known and unknown confounding
How could clinicians cause allocation bias in a non-randomised clinical trial?
They would not include patients with multiple morbidities with the new treatment
What are problems caused by comparison with historical controls in NRCTs?
Selection less rigorous
Treated differently from ‘new treatment’ group
Unable to control for confounders
Almost always overestimate benefit of new treatments
Less information available on bias and confounders
What is compared in a NRCT?
New treatment v.s. standard treatment
What are the three main steps involved in RCT?
Define
Conduct
Comparison
What must be defined in a RCT?
Disease Treatments to be compared Outcomes to be measured Possible bias and confounders Patients eligible for trial Patients to be excluded
How is an RCT conducted?
Identify eligible patient source and invite Consent willing patients Fairly allocate to treatments Minimise losses and maximise compliance Follow up identically
What is considered in the comparison stage of an RCT?
Observed difference? –> arisen by chance?
Big observed difference?
Is observed difference attributable to treatment?
How does random allocation minimise confounding?
Treatments groups are likely to be similar in size and characteristics by chance
How can random allocation of treatment be achieved?
Toss a coin
Random number tables
Computer generated random number
How can knowing the treatment allocation bias a trial?
Patient alters behaviour
Clinician alters treatment/care/interest (non-treatment effects)
Investigator alters approach when measuring/assessing (measurement bias)
What is the difference between single, double and triple blinding?
Single = patient/clinician/assessor does not know allocation Double = two of above do not know Triple = non of above know
How is blinding achieved?
Aim to make treatments appear identical in every way
What treatments can blinding be difficult for when using an RCT?
Surgery Psychotherapy v.s. antidepressant Alternative medicine Lifestyle interventions Prevention programmes
What is the placebo effect?
Knowing you are receiving a ‘new’ treatment is enough to make you feel better
What are the ethical implications of the placebo effect?
Placebo = deception
Only used when no standard treatment available
Must inform patient is receiving a placebo
What criteria must suitable outcome measures for clinical trials meet?
Appropriate and relevant Valid and attributable Sensitive and specific Reliable and robust Simple and sustainable Cheap and timely
What are outcome measures?
Defined before start of trial
Prevent data dredging
Protocol for data collection
Agreed criteria for measurement and outcome assessment
What do secondary outcomes of clinical trials include?
Occurrence of side-effects
Other outcomes of interest
How many primary outcomes is preferable in a clinical trial?
One
What outcome measures do you need during a clinical trial?
Baseline
During
Final
What three characteristics do outcome measures have?
Patient focussed
Clinically defined
Patho-physiological
What can cause non-compliance?
Patient believes treatment is not working Prefer other treatment CBA Misunderstood instructions Dislike treatment
How are losses to follow up minimised?
Make follow up practical and convenient
Be honest about commitment
Avoid coercion
Maintain participant contact
What is an explanatory trial?
‘As treated’
Non compilers excluded
What does an explanatory trial establish?
Physiological potency
What is not preserved in an explanatory trial?
Random allocation - no longer immune to confounding
What is a pragmatic trial?
‘Intention to treat’
If participants don’t like taking drug then future patients won’t either
Non-compliers included
What is preserved in a pragmatic trial?
Randomisation - no confounding problems
What was the declaration of Helsinki (2000)?
‘The health of my patient will be my first consideration’
What are considerations of the individual ethic in an RCT?
Principle of autonomy Principle of justice Principle of nonmaleficence Principle of beneficence RCTs do not guarantee benefit - may result in harm as for benefit of future patients, not participants RCTs allocate treatment by chance
What five issues should be considered for a clinical trial to be regarded as ethical?
Clinical equipoise Scientifically robust Ethical recruitment Valid consent Voluntariness
What is clinical equipoise?
Reasonable uncertainty/genuine ignorance about better treatment
What does clinical equipoise allow?
Valid consent
What are the issues with clinical equipoise?
Uncertainty of individual clinician or scientific community
“Reasonable uncertainty”?
“Better” for patient or society?
What makes a clinical trial scientifically robust?
Relevant/important issue Asks valid question Appropriate study design and protocol Potential to reach sound conclusions Justify comparator/placebo treatment Acceptable risks of harm Provision for safety and well being monitoring Arrangements for appropriate reporting and publication
What is inappropriate inclusion in ethical recruitment?
Participants unlikely to benefit
High risk participants
Participants likely to be excluded
What is inappropriate exclusion in ethical recruitment?
People who differ from an ideal homogenous group
People difficult to get valid consent from
What is needed for valid consent?
Knowledgable informant Appropriate information Cooling off period Informed participant Competent decision-maker Legitimate authoriser
Does a signed consent form equate with valid consent?
No!
What is Voluntariness?
Decision to participate is free from coercion or manipulation
What is the role of the Research Ethics Committee?
Ensuring dignity, rights, safety and well-being of participants is the primary consideration in any research study
Who manages research governance in clinical trials?
NHS Trust
Who governs financial management of clinical trials?
PCT
Who governs resource implication of clinical trials?
R+D office
Why are literature reviews using expert reviews biased and subjective?
Have to make the same assumptions as the expert, can’t be sure you’d reach the same conclusions
Why is a literature review using systematic reviews unbiased and objective?
Explicit assumptions
Transparent methodology
Reproducible
When is a systematic review used?
When an overview of primary studies that used explicit and reproducible methods is needed
What should a systematic review set out to identify as completely as possible?
All relevant evidence, published or not that meet certain predefined criteria with clear aims
What should a systematic review analyse carefully?
Results of all studies identified
Quality of all studies identified
What does a systematic review include explicit statements about?
Types of outcome measures
Types of interventions
Types of participants
Types of study
What does a systematic review investigate?
Clearly focused question with a single outcome
What do you have to use to carry out a systematic literature search?
Professional researchers
What three criteria must a systematic review meet?
Explicit
Transparent
Reproducible
What is a meta-analysis?
Quantitative synthesis of the results of two or more primary studies that addressed the same hypothesis in the same way
What is the purpose of a meta-analysis?
Facilitate synthesis of a large number of study results
Systematically collate study results
Reduce problems of interpretation due to variations in sampling
Quantify effect sizes and their uncertainty as a pooled estimate
What 4 quality criterion must be met in order for a meta analysis to be used?
Compilation of a complete set of studies
Identification of common variable or category definition
Standardised data extraction
Analysis allowing for sources of variation
What numerical analysis is carried out using a computer programme in meta analysis?
ORs and 95% CI for all and pooled data
Pooled estimate OR
What are studies in a meta analysis weighted according to?
Size
OR uncertainty
How does a small error factor affect the weighting of a study in meta analysis?
Increases it
How are the individual odds ratios in a meta analysis displayed on a forest plot?
Square with size in proportion to its weighting
Horizontal line indicates 95% CI
What does the diamond on a Forest plot represent?
Pooled estimate
Centre indicates pooled odds ratio
Width represents pooled 95% CI
What is the vertical line on a Forest plot?
Null hypothesis odds ratio
What does it mean if the 95% CI toughest the null hypothesis line?
Results are not statistically significant
What are the two common difficulties in systematic reviews and meta analysis?
Heterogeneity b/w studies
Variable quality of studies
What is the fixed effects model?
Assumes studies are estimating exactly the same effect size by measuring one true value using any resources
What is the random effects model?
Assumes estimating similar but not same effect size by trying to calculate mean for all effects
What can be used to analyse variation between heterogenous studies?
Sub-group analysis
What can the random effects model account for but not explain?
Variation
In which effects model in systematic review and meta analysis is the 95% CI wider?
Random
Is weighting more equal in random effects or fixed effects model?
Random
Why does the hypothesis test for heterogeneity have low statistical power?
Uses 10% significance for p value
What can cause variable quality of studies in systematic review and meta analysis?
Poor study design
Poor design protocol
Poor protocol implementation
List the types of study in order of susceptibility to bias and confounding starting with the least susceptible.
RCTs
NRCTs
Cohort
Case-control
How is variable quality of studies in systematic review and meta analysis approached?
Define basic quality standard and only include studies that satisfy it or weight studies on how well they fit it
What is the effect of publication bias?
Studies w/ statistically significant or favourable results are more likely to be published leading to biased selection for systematic review and meta analysis
What size of study is more susceptible to publication bias?
Small - bug are almost always published
How can publication bias be analysed in a systematic review?
Check meta analysis
Funnel plot
Use weak statistical test
What will the funnel plot of a more reliable systematic review which is less affected by publication bias look like?
Closer to central dotted line with studies equidistant from line
What kind of stats analysis is usually better to assess variable quality of studies in a systematic review or meta analysis?
Simple
