Module 1/2/3 Flashcards
Denominator?
the number of people in a study population
EGO and CGO?
Exposure group occurrence (EGO) = a/EG
Comparison group occurrence (CGO) = b/CG
Epidemiologists measure and compare?
Epidemiologists measure and compare dis-ease occurrences in different groups of people
group and population are used…?
interchangeably.
numerator?
the number of people from the study population in whom dis-ease occurs
numerical measures can be represented as… by..?
Numerical measures can be represented as categorical measures by dividing into different categories
Quanitive data can be?
Categorical or numerical
what do the arrows represent?
Arrows represent time (vertical is incidence, horizontal is prevalence )
what does ‘occurence’ measure?
An ‘occurrence’ describes the transition from a ‘non-dis-eased state’ to a ‘dis-eased state’
What does the circle and square (A,B,C,D) represent? (8)
The circle represents the study-specific denominators.
One exposure group (EG) and comparison group (CG)
Some studies have multiple exposure groups.
The square represents the numerators or dis-ease outcomes A, , b, c, d. (from left to right) A = EG and disease B= CG and disease C= EG and no disease D= CG and no disease
What happens when the study exposure is in numerical measures?
the numerical measures are often converted into categorical measures (two or more). When they are changed it is possible to calculate the occurance
what does the traingle represent?
The triangle represents the Participant Population (P)
What is a population?
A population is any group of people who share a specified common factor
What is Epidemiology?
The study of how much ‘dis-ease’ occurs in groups (populations) and of the factors that determine differences in dis-ease occurrence between two groups
what is incidence?
If the transition from a non-dis-eased state to a dis-eased state is an easily observable ‘event’ then epidemiologists count the occurrences as the number of events over a period of time ( incidence)
what is prevelance?
However if the transition is not easily observable , like transitioning from a non-diabetic to a diabetic state, then epidemiologists count the occurrences as the number of people with the dis-ease ‘state’ at a point in time (prevalence)
What question is involved in epidemiological thinking?
‘what’s the denominator?’
Why do epidemiolgists study negative events?
Epidemiologists tend to study negative events or states, like death or disease, because they are easier to measure than positive states of health such as degrees of wellbeing
why do we use Dis-ease instead of disease?
Use ‘dis-ease’ instead of ‘disease’ to encompass any health-related event or health-related state.
Eg. defining the prevalence of significant asthma as the proportion of a group of people who at the time of asking have had at least two severe asthma attacks in the previous one-year period
This is an example of? why?
This is called period prevalence because the outcome definition depends on the time period specified
How can people leave the prevalence of diseases? Therefore…?
People can leave the prevalence pool either by dying or if they are cured
Therefore a population with a high incidence of disease could have a low prevalence if the death rate or cure rate is also high (vice versa)
How do we often measure the prevelance of diseases?
Often measure the prevalence of diseases at two points of time and calculate the change in prevalence
The difference in prevalence between the two time points is in fact a measure of the incidence of disease over the period between the two time points. (e.g 10/100 had diabetes at one point in time and 20/100 people had diabetes 10 years later, then the prevelance of diabetes increased by 10/100 over 10 years)
how is incidence calculated?
Incidence is calculated by counting the number of onsets of disease (events) occurring during a period of time, and then dividing the numerator by the number of people in the study population.
EGO= a÷EG (/T) CGO= b÷CG (/T)
If the numerator is a count of a categorical then prevelance…?
if the numerator is the sum of the scores for a numerical..?
- If the numerator is a count of a categorical disease states, then prevalence will be a proportion
- If the numerator is the sum of the scores for a numerical outcome measured on everyone then the mean is similar to a measure of prevalence
period prevelance does not include?
Period prevalence calculations do not include the actual number of episodes an individual person has in the calculation
Prevelance measures never include?
A unit of time in their description of the measurement (e.g 10 per 100 people)
There can be more than one…… but always only one….?
More than one EG possible but always only one CG
What are the two epidemiological measures of dis-ease ocurrence?
incidence and prevelance
What does incidence requires? (2)
It requires the disease outcome to be a categorical variable
vertical arrow in the GATE frame
what happens when numerical outcomes are not converted into categories?
We often calculate the mean or median level of the outcome. so
EGO= Σa/EG
CGO= Σb/CG
What is a cohort study? (1)
It is a type of longitudinal study that follows research participants over a period of time. During the period some of the cohort will be exposed to a specific risk factor/charectersitics by measuring outcomes over a period of time.
What is a cross sectional study? (4)
The study takes place at a single point in time.
Used to describe what is happening at the present. (think of a cross sectional study as a snapshot of a particular group of people at a given point of time)
it does not involve manipulating variables
it’s often used to look at the prevailing charactheristics in a given population
What is prevelance calculated by?
Prevalence is calculated by counting the number of people with a dis-ease at one point of time and then dividing by the number of people in the study group at that point in time
What is the most appropriate measure of disease occurrence?
incidence as it has an easily observable onset
what’s always included in calculation of incidence?
Time
Why are some observable events still measured with prevelance?
Some diseases that do have observable onsets or events are still best measured as prevalence if the events come and go frequently - eg asthma attacks
Why is period prevelance a mix of incidence and prevelance?
Period prevalence is a mix of incidence and prevalence because it initially involves defining the presence of dis-ease based on the number of onsets that have occurred over a period of time, and then converting this into a single measure of disease at a point in time
Why is prevelance less useful? What type of arrow in the gate frame?
Prevalence is less useful than incidence because it is dependent on the death and cure rates
Horizontal arrow in gate frame
Allocation- What are measurement errors?
Inaccurate measures of exposures are usually known as measurement errors but as they can result in participants being allocated to the wrong exposure/comparison group , we consider them to be a type of allocation error ( allocation measurement error )
Allocation- What happens in observational studies (EG or CG) (2)
In observational studies the EG and CG are frequently quite different from each other in many respects and we usually try to adjust for these differences in the analyses ( age standardisation etc )
Important to collect sufficient information about the differences between EG and CG that can be used for the adjustments
Allocation- What happens in some small RCT’S? (2)
In some small RCTs, randomly allocating participants may not produce groups with similar characteristics just by chance alone
So always important to check for differences between EG and CG at the beginning of a study- called a ‘baseline comparison’ and should be done whether the study has allocated participants by randomisation or by measurement
Most epidemiological studies are designed to do what?
Most epidemiological studies are designed to investigate whether there are differences in disease occurrence between exposure and comparison groups within a study population
Recruitment- What is confounding error (caused by allocation)?
When the participants who are allocated to the exposure group are recruited from a different source than the participants allocated to the comparison group
Recruitment- What is involved in gate triangle (6)
Top represents setting
Rest of the triangle represents the eligible population
Tip of the triangle represents those from the eligible population who agree to take part
Often only a small proportion of the eligible population agree to participate in a study
If the non-responders are different from the responders, this can cause a recruitment error
Response rate of less than about 70-75% could cause significant recruitment error
relative risk greater than 1.0? What is it usually represented as? (2)
Relative risk increase. The RRi is usually expressed as a percentage increase, calculated by sibtracting 1.0 from the relative risk and then multiplying by 100.
RRI= (RR-1) x 100
e.g if the RR= 2.0 then the RRI= (2.0-1) x 100= 100% higher risk of…
Relative risk less than 1.0 can be expressed as? (3)
Relative Risk Reduction (RRR) because it is reduced below 1.0 (i.e the no-effect value).
The RRR is usually expressed as a percentage and is calculated by subtracting the relative risk from 1.0 and then multiplying by 100.
e.g heart attack (not taking the drug)= 10/100
(taking the drug)= 7/100
RR= 7/100 ÷ 10/100= 0.7
RRR= (1.0-0.7) x 100= 30%
What are non random errors?
Errors caused by problems with how the study is designed or conducted are called non-random errors (biases, systematic errors or validity problems)
What are random errors?
Errors caused by chance are described as random errors
What are the two main ways to compare two-disease occurrences?
EGO/CGO = relative risk (may be called a risk ratio)
EGO - CGO = risk difference (absolute risk difference to distinguish it from a relative risk (more appropriate measure))
What are two main ways of allocation?
- Randomly (reduces confounding ) ( experimental )
2. Allocate participants by measurement ( observational )
What happes in s ‘double blind’ randomized controlled trial (RCT)?
neither participants nor investigators know which intervention was given to which participant. They are like cohort studies except participants are allocated randomly to EG or CG.
What is an RCT?
A randomized controlled trial (RCT) is the best type of study to answer questions about the effect of treatments (but only if it is both ethical and practical)
What is in allocation random error? (2)
“Were the study participants successfully allocated to the Exposure Group (EG) and the Comparison Group (CG)?”
- If EG and CG differ in ways apart from the study ‘exposure’, and if these other differences also have an effect on the study outcome, then it’s not possible to know whether the study exposure or the other factors caused EGO and CGO to differ ( confounders )
What is it called when calculating the difference between two means?
if outome measures are calculated as means (averages) the difference between two means are called the mean difference (MD).
What is recruitment error? (3)
When the study findings are not applicable to a wider population
Commonly occurs when the main objective of the study is to measure the characteristics of a specified eligible population but the Participants (P) who are recruited are not representative of the Eligibles
In many studies it is unnecessary to recruit participants who are representative of a specified external population ( only when measuring prevalence in entire population)
What is the 95% confidence interval?
The 95% confidence interval is used to describe the amount of random error in the study results.
“there is about a 95% probability that the true value of EGO in the whole population of interest, from which the study participants were recruited, lies between 8.0 and 10.0”
What is used within the gate frame to indicate that there may be more than 2 exposure groups? (2)
Dotted horizontal and vertical lines are used within the GATE frame to indicate that there may be more than two exposure groups and more than two outcome groups
Alternatively either or both exposures and outcomes can be numerical measures
When is the RD considered an ARR or ARI?
If the risk is lower in the exposure group (ARR) or if the risk is higher in the exposure group (ARI)
Adjustment- stratified analysis- What is direct age-standardisation?
Direct age-standardisation - when disease incidence or prevalence in different populations with different age structures are compared. Each population is stratified into comparable age groups and disease incidence or prevalence is calculated for each strata. Then each population’s age structures are standardised and a standardised disease measure is calculated for the combined age strata
Adjustment- what is strata?
Confounding can be reduced by dividing participants into ‘strata’ , and then analysing the data as if there were two separate studies
The results of the analyses in the different strata can then be combined , if they give reasonably similar results
how can confounding in allocation occur? how is it solved and when is it possible? (3)
Confounding can still occur in a large RCT if the random allocation process is not done properly
Solved by concealment of allocation
Randomisation is only possible when the exposure being investigated is considered to be safe
What are the strengths of cohort studies? (2)
Cheaper than RCTs
Exposure measured before outcome, avoiding recall bias and providing clear time sequence between exposure and dis-ease outcomes
What are the weaknesses of cohort studies? (3)
How accurate was allocation (measurement)
How similar are EG and CG (confounding)
How well maintained in EG and CG
What happens if results are not statistically significant? (2) What is the width of the CI dependant on?
“Too much random error to determine if there is a real difference between EGO and CGO”
The width of the CI is dependent on the number of events that occur in a study
What happens in statistically significant results (large overlap)?
When there is a large overlap between 95% CIs for EGO and CGO - the study is unable to determine if EGO is different from CGO in the population from which the study participants were recruited.
What happens in statistically significant results (no overlap)? (2)
When there is no overlap in the CIs for EGO and CGO it is reasonable to assume that EGO and CGO are truly different from each other
The confidence intervals for RR and RD will not cross the no-effect line
What happens when maintanence errors are small? How is it improved? (2)
As long as any maintenance errors are small and similar in both EG and CG, the error will underestimate the true effect of the exposure on the study outcomes (this is considered preferable to not knowing whether the error will exaggerate or underestimate the true study effect measures)
Maintenance error improved by keeping participants/study practitioners blind
What is a 95% confidence interval statement?
“There is about a 95% chance that the true value in a population lies within the 95% confidence interval”
“In 100 identical studies using samples from the same population, 95/100 of the 95% confidence intervals will include the true value for the population”
What is blind or objective measurement error?
Can cause outcome measurement errors
Blinding or double blinding improves outcome allocation when the measurement is not very objective
Or use objective measurements wherever possible eg blood tests instead of questionnaires
What is clinical significance?
If a clinician would make the same clinical decision whether the true result was at one end of the confidence interval or the other
What is cohort study commonly used for? (2)
Commonly used for investigating risk factors for disease with large effects, as usually unethical to do RCTs
Not very useful for studying benefits of interventions with small effects because confounding may hide a small harm
What is maintenance error? (4)
Maintenance error caused when…
Participants do not maintain their exposure and comparison status throughout the study ( switch groups )
Are exposed to other factors that could influence the study outcomes
Drop out of the study
What is meta-analyses? (3)
Combining the results of a number of small studies is similar to conducting a larger study and reduces the amount of random error
Can change conclusion of study (from statistically not significant to statistically significant)
When none of the individual studies are large enough to give a precise enough estimate
What is random allocation error?
The EG and CG in a randomised controlled trial may differ by chance alone, particularly if the trial is small