MODULE 1 - POPHLTH 111 Flashcards
What is the main goal of epidemiology?
To measure the occurrence of dis-eases in populations and measure the factors of causes in different populations.
What is the key feature of epidemiological study?
It ALWAYS starts with a population.
What is the numerator and denominator of epidemiological studies?
Denominator: The TOTAL number of people in a population. Numerator: The number of people who dis-ease occurs.
Why epidemiologists often age standardise measures of dis-ease frequency before making comparisons between populations?
Age-standarise is a way to make a fair and equal comparison of dis-ease occurrence between populations. This is because different populations have different age structures, which will affect the outcome of the dis-ease. (E.G older people are more likely to die from a dis-ease than younger people).
What does population/ group mean?
Sharing one or more common feature . E.g: Same age, gender or ethnicity.
What is the death rate in NZ and what does this mean?
100% or one person per lifetime. This is because everyone dies once in their life. What is important it to understand that without a specific time frame, the measure will not be meaningful. E= (n/d) / t. So instead, you must say what is the death rate in NZ per year?
When to use E= n / d and when to use E = (n/d) / t.
- Use that when looking at the proportion. 2. Looking at the specific time rate.
What are the common design features of PECOT on the Gate Frame?
P - Total participant population (Triangle)
E - Exposure groups (Sub- denominator, circle)
C- Comparison groups (Sub- denominator, circle )
O - Outcome (People who got the dis-ease from the exposure group and comparison group, square)
T - Time (Arrow)
Difference between EG and EGO
EG is the no. of people who got exposed (sub-denominator) and EGO are the people who got the outcome from the people that got exposed ( a / EG).
2 key Features of cohort study
- Researchers allocate people in CG and EG by MEASURING (observing what they are already doing).
- Cohort studies follow up OVER TIME.
What is the difference between incidence & prevalence?
Incidence is a type of measure that measures the disease events over a period of time whereas prevalence measures the number of people that have the disease at one point in time.
What are the equations for incidence and prevalence?
- Incidence = (a/ EG) over t and (b/EG) over t
- Prevalence = ( a / EG) over 1 and (b/EG) over 1.
When should you measure prevalence and incidence?
To measure incidence, the events must be easily observable (the number of new COVID-19 cases) and to measure prevalence that is hard to observe if the dis-ease has occurred (people who are obese).
Why must dis-ease outcomes in incidence measures be categorical (yes/no)?
So we can count how many people do and do not develop the outcome over time — needed for calculating incidence.
What are cross-sectional studies?
Studies that measure the outcome and exposure at the SAME time. It can only measure prevalence.
In which situation is it difficult to measure prevalence and how to overcome this?
When it is difficult to measure the outcome at that specific point in time (such as are you suffering from an asthma attack right now?). So instead, we count the people who have had it in a previous time period, but not ALL the cases. This shows that the time arrow is BACKWARDS.
What does it mean when the time arrow is horizontal vs backwards?
Both mean that is it a measure of prevalence but horizontal arrow occurs when you are measuring dis-ease occurrence right now and backwards means you are measuring dis-ease occurrence in a past period of time.
What types of measure must be death rate?
Incidence
What is another way to present numerical health data like cholesterol levels in terms of prevalence?
By reporting the average (mean) cholesterol level in the population.
Describe in what situations it would be
most appropriate to measure incidence rather than prevalence and vice versa
For incidence, it would be the most easiest to measure incidence when it is easiest to count when the dis-ease occurs.
What are RCTs?
Randomised controlled trials (RCTs) are like cohort studies except participants are allocated randomly to EG or CG. This means participants have equal chance of being allocated to EG or CG.
How can we measure prevalence in cohort studies?
During the beginning of a cohort study.
List strengths and weaknesses of incidence and prevalence.
Strengths of incidence= incidence is determined only by the dis-
ease risk in a population (its a ‘clean’
measure of dis-ease occurrence), incidence measures include events (N),
population (D) and time (T). Weakness = Measure OVER time (time-consuming).
Strengths of prevalence= prevalence is relatively easy to measure
as you ‘stop time’ and count
Weakness of prevalence = prevalence measures include only events
(N) and population (D) – less information than incidence
* prevalence is determined by the
incidence, cure rate and death rate (its a ‘dirty’ measure of dis-ease occurrence).
How do you calculate RD and RR?
RD is EGO - CGO. It MUST include units. RR is EGO / CGO. It doesn’t have units.
What are ecological RCTs?
Ecological RCTs are a type of study where the participants are geographical regions AND these regions are randomised.
When do you use PECOT and PEOT?
- You use PECOT when comparing 2 groups (Is depression more common in university students who sleep less than 6 hours a night compared to those who sleep more?) and you use PEOT when there is NO comparison being made which sometimes are measures of prevalence in cross-sectional studies. (What percentage of university students currently have depression).
What does RD = 0 mean and RR = 1 ?
RD = 0 there is no difference in the risk of the outcome between the exposed group and the comparison group. This is also known as the ‘no-effect value;’ remember that the equivalent no-effect value for a RR = 1.0). RR =1 means the risk is the same in both groups.
What does it mean if the Relative Risk (RR) is large but the Risk Difference (RD) is small?
It means the exposure might sound like it has a big effect (e.g. “doubles the risk”), but the actual number of people affected is very small. This makes the finding less meaningful in practical or clinical terms. Always check both RR and RD to understand the true impact!
What is RRR and how do we calculate it?
RRR is known as relative risk reduction which indicates that the exposure reduces the risk of the outcome. RRR = (1.0 - RR) x 100. ONLY IF THE RR is less than 1, we minus it from 1 to get RRR.
What is RRI?
Relative risk increase (RRI) = (RR - 1) x 100%
What is RD and RR?
Two main ways to compare two dis-ease occurrences NOT the measures of dis-ease occurrences.
What are non-random errors and how do they occur?
Non-random error occurs because of poor study design, poor
study processes or poor study measurement it is called
a non-random error (or bias). They are often called biases or systematic
errors.
What is the R of RANBOMAN?
Are the participants a representative sample from a relevant & defined population? Recruitment error occurs when the participants in a study are not representative of the larger population the study aims to measure, leading to external validity issues.
For the R in RANBOMAN, how is it presented?
The triangle in the GATE
frame is divided into 3 overlapping levels, where open top of triangle
represents the setting in which the eligible population was recruited, the next one is Eligible Population an identifiable, meaningful population? (Meeting the criteria). The third layer (the one that ends) represents eligible population who actually agree to participate in the study. These are the study participants.
Why are non-responders important in recruitment error?
Non-responders are individuals who meet the eligibility criteria for a study but choose not to participate. If non-responders differ from responders in ways that affect the study’s outcome (e.g., health status, behaviors), this can introduce bias, known as non-response bias or recruitment error. This is especially important in prevalence studies, as the study findings may not accurately represent the wider population.
Describe allocation error. Random allocation error only affects RCTs.
Allocation occurs when were the study participants correctly and successfully allocated to the Exposure Group (EG) and the Comparison Group
(CG)? and ii. were the EG and the CG similar at the beginning of the study. There are two types of allocation error: measurement error (cross-sectional & cohort studies) and confounding error which are external factors that can disrupt the outcome of the study (diet, age, stress levels).
What is the difference between random allocation and random sampling?
Random allocation refers to the idea of allocating people to EG and CG randomly whilst random sampling is choosing participants from a larger population. Only random allocation is involved in RCTs, NOT random sampling.
Where is confounding error found and when is it not found?
Confounding will be present in almost every observational study (cross-sectional, cohort). The best way to reduce the likelihood of confounding is to conduct an RCT. However, confounding can still occur in a large RCT if the random allocation process is not done properly.
What is the maintenance of RAMBOMAN and what studies do this?
Were most of the participants maintained throughout the study in the groups (EG & CG) to which they were initially allocated? No loss to follow-up in x-sectional studies because participants not followed up. The best way to keep the degree of
maintenance error similar EG and CG is to keep the participants and study practitioners ‘blind’ to whether the participant is receiving the study exposure (E) or the comparison
exposure (C). Cohort studies have also of maintenance error. Its not just about it being hard to follow people over time but also whether people stay in their exposure group or not.
What is random error and non-random error? Why are similar studies produce different results ?
Random error are error due to chance and non-random errors are errors due to the study design being poor. These differences in results from similar
studies are also random errors
Why is it hard to measure the exact reason for dis-ease occurrence in biology?
- People are moving targets so identical measurements of exposures and outcomes change over time. (Random measurement error)
- We only measure a sample of the people we want to study, NOT the entire population.
Why is it that smaller the random sample, the greater the random sampling error?
The smaller the sample, the less likely it is to be representative of the whole population — just due to random chance.
Define 95% confidence interval.
The actual definition of a 95% confidence interval is ‘in 100 identical studies using samples from the same
population, 95 out of 100 of the 95% confidence intervals will include the true value in the population that
the studies were sampled from’
Why type of epidemiological measure has random error?
Every epidemiological measure (EGO, CGO, RR, RD, NNT) has random error which can be estimated by a confidence interval.
What does a wider CI and narrow CI tell us?
Wider CI tells us there is more uncertainty in the data and narrower means less uncertainty in the data.
What does not statistically significant mean?
It means that there’s no strong evidence that the exposure (like a treatment or behaviour) caused a change in the outcome. For example, there is NO strong evidence that a drug worked better than the placebo.
What does it mean when 95% CI overlaps RD = 0 and RR=1.
If 95% CIs for the RD (EGO – CGO) overlaps 0, or 95% CI overlaps with RR=1, then there is
probably no statistically significant difference between
EGO and CGO. In simple terms, there is no strong evidence that the exposure changed the outcome.
When does it mean when the CI is statically different between EGO and CGO?
When the 95% CIs for EGO & CGO do not overlap OR the 95% CI for the Risk Difference does not overlap the no effect line (RD=0) .
How do we reduce random sampling error and random measurement error?
- Do a bigger study random sample error.
- Combining studies in a meta-analysis is the next best thing to doing a larger study & reduces random error. Another thing is to do multiple measurements.
When do we use the different types of study?
- It is ethical, affordable and practical?
- Depends on the question.
Finally, at the end of the
6-month follow-up period, participants would receive another Covid-19 test. Why wouldn’t there be outcome measurement error?
Covid-19 test is an objective and more accurate, so there would not be any error.
