HADPOP Flashcards
Define crude birth rate
Number of live births per 1000 population
Describes impact of birth on size of population
Define general fertility rate
Number of live births per 1000 females aged 15-44
compares fertility of fertile female populations
Affected by age-specific fertility rate & age distribution
Define total period fertility rate
Average number of children born to a hypothetical woman in her life
Compares fertility of fertile females without being influenced by age or group stucture
What are the determinants of fertility
Fecundity: physical ability to reproduce
Fertility: realisation of this potential as births
Define crude death rate
Number of deaths per 1000 population
Define age-specific death rate
Number of deaths per 1000 people in a given age group
What is standardised mortality ratio
How do you calculate it
What are its advantages
Compares number of deaths with expected number of deaths (if age-sex distribution of the population in identical)
Observed number of deaths / expected (x 100)
Provides overall mortality ratio adjusted for different distributions of factors e.g. Age, sex in populations
Compares death rates by applying same population age structure
How do you calculate incidence / incidence rate
What are its advantages
Number of new cases / person-years (no of people x no of yrs)
Number of cases / population number x 1000
Focuses on new events, monitors epidemics
What is prevalence
How do you calculate it
What are its advantages
A proportion not a rate
Number of people affected (old & new) / total population number
(X 100 for percentage)
Describes burden
Measure service needs
What is incidence rate ratio
How do you calculate it
How do you interpret it
Compares number of cases per population per unit time
Rate b (exposed) / rate a (unexposed)
Compare IR between groups with different levels of exposure
Make sure person years are the same for both
If rate b higher = difference in exposures associated with different rates of disease
b > a = IR > 1
b
What figures do you use in error factor calculation for: IR / prevalence IRR SMR OR
d= no of events observed (not a rate)
d1 & 2 = no of events in each population
O = observed no of cases
a= cases exposed, b= controls exposed, c= cases unexposed, d= controls unexposed
Interpreting confidence intervals:
If range doesnt include null hypothesis (1)
If range does include null hypothesis (1)
Can reject null
Data inconsistent with hypothesis
Cannot reject null
Data consistent with hypothesis
Describe:
Systematic variation
Random variation
Variation attributed to particular factors
Variation cant be explained (e.g. By chance)
Cohort studies:
Describe characteristics
Advantages
Disadvantages
Recruit disease free individuals
Follow them over time & count how many develop disease (person yrs)
Comparative: risk one group vs another
Prospective or historical
Study rare exposures/characteristics
More detailed info on outcomes & exposures
Collect additional info on potential confounders
Large/resource intensive Time consuming Risk loss to follow up Not good for rare outcomes Confounding (unknown)
Cohort studies:
How is risk estimated
How can comparisons be made
Incidence rate in each group
Exposed & unexposed:
Subdivide levels of exposure & compare with relative risk (IRR)
Compare to external reference population (SMR)
(Error factor for internal comparison larger than for external)
(External comparison usually have less random variation)
Case control studies:
Characteristics
Advantages
Disadvantages
Identify group already ill (cases) & group that are not (controls)
Look back to compare exposure status (generally retrospecitve)
Identify exposure & compare disease outcome
Prevalence fixed by study design/number of participants;
Estimates relative (not absolute) risk
Good for rare diseases
Can look @ different exposures once, for single outcome/disease
Quicker, cheaper
Not good for rare exposures
Selection bias: recall, finding representative comparison population
Info bias: info on exposure, randomly inaccurate measurement, systematic (e.g. Different data collection), interviewer bias
Confounding: minimise with matching for confounders & logistic reg
Odds ratio (for case control & cohort):
How to calculate
What it shows
Effects on OR if:
Cases underestimate more than controls
Controls underestimate more than cases
Both cases & controls underestimate randomly
Table: cases & controls across top, exposed & unexposed down side
(axd) / (bxc)
Estimates relative risk (like IRR)
compare oods of being exposed in cases vs odds of being exposed in controls
Increase precision by increasing number of controls (reduced EF); usually 5 x number of cases
Bias OR towards underestimation of affect of outcome
Bias OR towards overestimation of affect of outcome
Shrinkage towards null
Define chance: 95% confidence interval & P value
95% chance that true value lies within range
How likely results would have occurred by chance if the null was true
Define bias
What are the types of bias
Deviation of results / inferences from truth; invalidates conclusins
Selection bias:
correct info, wrong people
Characteristics of groups & the way they are selected
Information bias:
Incorrect info, right people
Measurement/classification into groups
Recall bias
Each may be:
Systematic (results skewed in particular direction)
Non-systematic (random errors, CI wider, IR towards 1)
Define confounding
Characteristic of the population, linked to exposure & outcome
Measure of effect of exposure on risk of outcome being distorted
Define reverse causality
Cause-effect relationship exists in the opposite direction
How do you distinguish causal from non-causal association
Hypothesis: from observation / clinical practice & studies
Analytical study: test hypothesis
Observed association: test validity, exclude alternative explanations (chance, bias, confounding)
Does statistical association represent cause-effect relationship?:
Infer beyond data in light of current knowledge
Census: definition
Simultaneous recording Of demographic data By government @ particular time Re: all people with live in a particular country
Cause-effect relationship:
What are Henle-Koch’s Postulates
What are their limitations
Agent present in every case by isolation in pure culture
Agent not found in any other diseases
Agent able to reproduce disease in experimental conditions in experimental animals (once isolated)
Diseases are multifactorial; exposure can cause many diseases
Evaluating strength of evidence in the cause-effect relationship:
What are the Bradford Hill criteria for inferring causality
Strength of association Specificity of association Consistency of association Temporal sequence Dose response (biological gradient) Reversibility (strongest criterion) Coherence of theory Biological plausibility Analogy
Randomised control trials:
What is the purpose of a clinical trial
Why do you not use historical controls
Fair, controlled, comparative, reproducible study to clarify the most appropriate treatment of future patients
Selection criteria less well defined/rigorous
May have been treated in different/other ways
Less information about patients, therefore difficult to adjust for confounders
Randomised control trials:
What are the steps involved
Why are participants randomly allocated to groups
Why use blinding (& what are the types of blinding)
Answer question: is new treatment better of worse than the usual/standard treatment?
Identify, recruit, consent & maintain 2 comparable groups
Allocate, follow up, assess & compare
Minimise allocation bias & confounding (known & unknown)
Minimise non treatment effect & measurement bias
(Single or double blind)
Randomised control trials:
What is the definition of placebo
Why is it used
Inert substance that looks/tastes/is packaged identically
Eliminate placebo effect (psychological benefit of being treated)
Randomised control trials:
Features of suitable outcome measures
Measured by assessor
Blind to treatment status
Relevant, objective, valid, sensitive, reliable, practical
Randomised control trials:
How do you minimise losses to follow up
How do you maximise compliance
Sufficient/clear info on what is involved
Follow up not too onerous
Clear, simple instruction
Check on compliance
Randomised control trials:
What are the types of analysis
As treated: but no longer randomly allocated / immune to follow up (e.g. Assess drug physiology)
Intention to treat: realistic indication of impact of treatment
Randomisation preserved & confounding avoided
Randomised control trials:
What are the ethical principals of research
What are the issues in making a clinical trial ethical
What is the role of the Research & Ethics committee
Declaration of Helsinki: health of patient 1st consideration
Collective ethic: all patients should have treatment tested for efficacy/safety
Individual ethic: RCTs for benefit of future patients
Voluntariness Valid consent Ethical recruitment Scientifically robust Clinical equipoise (uncertainty about better treatment)
Involved NHS trust & PCT research & development office
Focus on: scientific design/conduct, recruitment, care & protection, confidentiality, informed consent, community consideration
