Module 1 (GATE frame) Flashcards
Epidemiology
The study of how much ‘dis-ease’ occurs in a population and of the factors that determine differences in dis-ease occurrence between populations
The main goal of epidemiology
Measure the frequency of health and dis-ease in different populations to find out the causes of poor health and how to improve it
Measuring epidemiology
The occurrence of dis-ease (N) in groups of populations (D) at one point or over a period of time (T) - E = N/D/T
Occurrence
Describes the transition from a non-diseased state to a diseased state; big change over a period of time usually due to a single component
Age standardisation
Process of converting the different age structures in each population into one (standard) population age structure and working out the disease rates
Categorical data
Quantitative data grouped into categories
Numerical data
Quantitative data which takes on numerical values; can be grouped into categorical data by making a large range of values classify as a group/outcome
Triangle in GATE frame
Population - overall study/participant population representing the denominator; study setting, eligible population and participants willing to take part
Circle in GATE frame
Study-specific sub-denominators including the exposure group(s) and comparison group
Exposure group
Group(s) of participants subject to a certain exposure; studies can have multiple exposure groups
Comparison group
Group of participants not subject to the certain exposure; studies can only have one comparison group
Square in GATE frame
Outcomes - dis-ease outcomes; can be split into more than 4 squares where multiple exposures/dis-ease outcomes can be classified to more than 2 categoris
Arrow in GATE frame
Time - horizontal or vertical representing the time when or during which the outcomes are measured
Cohort study
Follow-up study; investigates associations between risk or prognostic factors (exposures) and dis-ease incidence in different groups of individuals; commonly used for investigating risk factors or dis-ease with large effects; incidence or prevalence
Cohort study advantages
Usually cheaper than RCTs; exposure usually measured before outcome, avoiding recall bias and providing clear time sequence between exposure and dis-ease outcomes
Cohort study disadvantages
Confounding is common, can hide effects when they are small; maintenance error is very common in long term studies because the exposure is not controlled by investigators (loss due to follow-up)
Cohort study main design features
Longitudinal, observational (non-experimental) where participants are allocated to EG and CG by measurement and dis-ease outcomes are measured during follow-up
Cross-sectional study
Measures the population, exposure and dis-ease frequency at the same time; investigates associations between risk factors and dis-ease prevalence in the groups; prevalence
Cross-sectional study advantages
Generally cheaper and can be completed more quickly than RCTs or cohort studies; best design for assessing the prevalence of dis-ease in a population; no maintenance error because there is no follow-up
Cross-sectional study disadvantages
Uncertain time sequence (possible reverse causality) which limits interpretation of cause and effect so not useful for causes of disease; confounding common so not useful for benefits of drugs
Cross-sectional study main design features
Cross-sectional, observational (non-experimental) where participants are allocated to EG and CG by measurement and dis-ease outcomes are measured at the same time
Randomised controlled trial
Participants are randomly allocated to EG and CG; need to always check for differences between the two groups at the beginning of the study (baseline comparison); can be single-blind or double-blind
Randomised controlled trial advantages
Blind studies prevents participants and measurers from acting differently due to their knowledge of the intervention, decreases chance of confounding; best study for effect of treatments if practical and ethical; groups similar at beginning of study; any difference on outcomes can be attributed to the intervention
Randomised controlled trial disadvantages
Not useful for risk factors of disease, can be unethical or impractical (do a cohort study instead); logistically difficult (long-term follow-up) and costly; maintenance error is common - random error
Randomised controlled trial main design features
Longitudinal, experimental; participants randomly allocated to either study exposure or comparison exposure and dis-ease outcomes measured during a follow-up period
Ecological study
Comparisons of groups of populations (cities, regions or countries) instead of groups of individuals; can be longitudinal or cross-sectional
Ecological study advantages
Generally cheaper and quicker than all other study designs because it uses data already collected; useful when the majority of some populations are exposed but others are not; efficient for rare outcomes
Ecological study disadvantage
Confounding is very common
Ecological study main study design
Longitudinal or cross-sectional, non-experimental or experimental; exposure and comparison allocated to groups rather than individuals
Incidence
Involves counting the number of onsets of a dis-ease occurring during a period of time; done when it is possible to observe when the disease occurs (observable onset points); dis-ease outcome needs to be categorical; arrow is vertical
Incidence calculation
Number of persons in group who have dis-ease outcome/ total number of persons in group, during study time
Incidence strengths
Determined only by dis-ease risk in a population (clean measure of occurrence); measures include event, population and time
Incidence weaknesses
Can be difficult to measure as you have to observe events over time
Prevalence
Involves counting the number of people with a dis-ease at a point in time; done when it is not easily possible to observe when the disease occurs (no observable onset points); can be point (horizontal arrow) or period (diagonal arrow)
Prevalence calculation
Number (sum) of dis-ease states/ total number of persons, at a point in time
Point prevalence
Where dis-ease status can be easily measured at one point in time (does not take previous time period into account); horizontal arrow
Period prevalence
Where dis-ease status cannot be easily measured at one point in time so must be measured by the number of people who experienced in the past; diagonal arrow
Prevalence strength
Relatively easy to measure as you ‘stop time’ and count
Prevalence weaknesses
Measures only include events and population (less information than incidence); determined by incidence, death rate and cure rate (dirty measure of occurrence); can lose data
Meta analysis
A way to help decide if there is likely to be a real treatment effect; combining results from several studies together mathematically, generally a summary estimate of the effect; reduces random error and is an alternative to conducting one large study
Risk ratio
Relative risk; calculated by dividing EGO by CGO; no difference between the occurrences in each groups means an RR of 1 (no effect); if they are calculated as averages, it is the relative mean; no units
Relative risk reduction
When the relative risk is less than 1 (no effect value), usually expressed as a percentage decrease
Relative risk increase
When the relative risk is greater than 1 (no effect value), usually expressed as a percentage increase
Risk difference
Absolute risk difference; calculated by subtracting CGO from EGO; no difference between the occurrences in each group means a RD of 0 (no effect); if they are calculate as averages, it is the mean difference
Absolute risk reduction
Risk is lower in EG
Absolute risk increase
Risk is higher in CG
Non-random error
Can occur due to poor study design, execution, process or measurement; RAMBOMAN
R in RAMBOMAN
Recruitment; who was recruited into the study and is it possible to describe the population that the participants represent?; external validity error (not representative); allocation bias (EG and CG different); non-response bias
A in RAMBOMAN
Allocation; were the study participants correctly and successfully allocated into EG and CG and were they similar at the beginning of the study?; confounding (EG and CG differ in ways other than the exposure); measurement error if allocated due to measurement
Confounders
Factors which cause EGO and CGO to differ, makes it impossible to know whether the study exposure has an effect on these occurrences and occurs when CG and EG differ in ways other than the study exposure
M in RAMBOMAN
Maintenance; were most of the participants maintained throughout the study in EG and CG to which they were initially allocated?; participants need to maintain exposure/comparison status, not be exposed to other influential factors and not drop out; maintenance error (loss due to follow-up)
BOM in RAMBOMAN
Blind and objective measurement; were the people who measured the dis-ease outcomes unaware of the participant’s exposure status and were these measurements made objectively?; dis-ease outcome must be measured correctly for results to have meaning; objectivity and blindness
Objectivity
Outcome must be measured with well-defined methods, ideally by machines to eliminate human error; categorical outcomes such as death are easier to judge objectively
Blindness
Knowledge of a participant’s exposure may influence their behaviour and can lead to confounding, knowledge can also influence the practitioner’s perception or interpretation of signs/symptoms of the study outcome
AN in RAMBOMAN
Analysis; if there were differences in the characteristics or participants in EG and CG that could affect the study dis-ease outcomes (i.e. confounders), were they adjusted for in the analyses?; can be reduced by dividing study population into strata
Random error
Occurs due to the way studies are designed and conducted, due to chance and can be related to the dartboard analogy; bullseye and darts (random error); includes sampling, measurement/assessment, biological phenomena and allocation errors
BOM in RAMBOMAN
Blind and objective measurement; were the people who measured the dis-ease outcomes unaware of the participant’s exposure status and were these measurements made objectively?; dis-ease outcome must be measured correctly for results to have meaning; objectivity and blindness
Objectivity
Outcome must be measured with well-defined methods, ideally by machines to eliminate human error; categorical outcomes such as death are easier to judge objectively
Blindness
Knowledge of a participant’s exposure may influence their behaviour and can lead to confounding, knowledge can also influence the practitioner’s perception or interpretation of signs/symptoms of the study outcome
AN in RAMBOMAN
Analysis; if there were differences in the characteristics or participants in EG and CG that could affect the study dis-ease outcomes (i.e. confounders), were they adjusted for in the analyses?; can be reduced by dividing study population into strata
Random sampling error
Error that occurs in every sample study due to chance, every study population can only be a sample of the total population of interest; bigger the sample, the smaller the differences between the sample and population; result from any study will be an estimate of the truth rather than actually truth
Random measurement/allocation error
Error that occurs due to our lack of ability to measure biological factors in exactly the same way every time we measure them (particularly if the measurement instrument requires a human operator); reduce by taking multiple measurements and average them or use an automatic, more objective instrument
Randomness inherent in biological phenomena
Error that occurs due to the inherent variability in all biological phenomena (and therefore in all measurements of this), when the thing being measured changes from moment to moment; reduce by taking multiple measurements and averaging them
Random allocation error
Error that occurs due to the EG and CG in an RCT differing by chance alone, particularly when the trial is small; reduce by undertaking a larger study
Confidence interval
A way to describe the amount of random error in a measurement or calculation of the truth in the whole population; the wider the CI, the more random error; if EGO=CGO, the study will show ‘no effect’ (RR=1 and RD=0), no effect line; shows whether EGO and CGO are statistically significantly different
95% confidence interval
There is about a 95% chance that the true value of EGO in the whole population of interest, from which the study participants were recruited, lies within the upper and lower values of the CI - assuming no non-random error
Statistically significant
No overlap in the CIs for EGO and CGO; CIs for RR and RD do not cross the no-effect line
Borderline statistically significant
No or slight overlap in the CIs of EGO and CGO; CIs for RR and RD nearly cross the no-effect line
Not statistically significant
Overlap in the CIs for EGO and CGO; CIs for RR and RD cross the no-effect line; there may be too much random error to determine if there is a real difference between EGO and CGO
Clinical significance
A result is considered to be clinically significant if a clinician would make a similar clinical decision whether the result was near one end of the CI or the other