Analytical Epidemiology And Observational Studies Flashcards
Descriptive epidemiology
Describes occurrence of disease
Determinants within a population
Analytical epidemiology
Explores the quality and amount of influence that determinants have on the occurrence of disease.
Explores “how” health events occur
Usually through studies with group comparisons
What is reverse causation?
Reverse causation is mistakenly assuming that variable A influences variable B when it is actually B that influences A.
Define Cross-sectional study (prevalence study)
Observational study that looks at the relationship between health related characteristics and other variables of interest within a defined population at one particular time
Define case-control study
Observational study that looks at persons with a disease or condition and a suitable control group of persons without the condition of interest, and comparing how frequently a suspected attribute or risk factor is present in each group
Define cohort study (incidence study)
Observational study that follows two groups of people those with and those without an exposure over time, comparing how frequently an outcome occurs in each group
What provides a way to visualise both the quality of evidence and the amount of evidence available?
The levels of evidence pyramid
What happens as you go down the evidence hierarchy pyramid ?
As you go down the pyramid, the amount of evidence will increase as the quality of the evidence decreases.
Vice versa as you go up
Define innate factors
simply born with - their sex, their race, and ethnicity -basically their genetic composition (non-modifiable).
Define ‘acute exposures’
by which we simply mean those that are relatively brief (e.g. infectious agent SARS-CoV-2 during a covid-19 epidemic; it intrauterine exposure for a foetus; brief physically or mentally stressful events; medication, environmental factor, v accine, food etc)
Define ‘chronic exposures’
refer to things like pollution, social factors (poverty or policies/laws that might have an impact on health).
Define ‘time-varying exposures’
would apply to our behaviours – how/what we eat, exercise, smoke or drink alcohol and how much. All of these things that might be changing over the life course
Factors that can impact on health outcomes
Innate factors
Acute exposures
Chronic exposures
Time-varying exposures
What are possible health outcomes or health indicators?
Binary
Ordinal
Continuous outcomes
What are binary health outcomes ?
factors or outcomes that either occurred or did not occur, e.g., diseased or not, living or dead, MI or not.
What are ordinal health outcomes?
meaning simply graded categories, grading say from very poor, poor… to good and excellent (self reported, subjective).
What are continuous outcomes?
measurement such as systolic blood pressure, serum cholesterol levels, and so forth.
Why is the process of “counting people” useful?
important basic measure of disease frequency that is essential to detecting trends or the sudden occurrence of a problem, such as an epidemic.
Simple counts of the number of diseased people are also important to public health planners and policy makers for assessing the need for resources in a population.
When measuring disease frequency what two elements are helpful in comparing groups?
Proportion
Rates
Define ‘proportion’
Ratio relating a part to a whole, often expressed as a percentage (%)
Define ‘rates’
Ratio in which the denominator also takes into account the dimension of time
What are the two fundamental measures of disease frequency?
Prevalence & incidence
Define ‘prevalence rate’
The proportion of the population that has disease at a particular time
Point prevalence equation
𝑷𝒐𝒊𝒏𝒕 𝑷𝒓𝒆𝒗𝒂𝒍𝒆𝒏𝒄𝒆= 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐶𝑎𝑠𝑒𝑠 𝑎𝑡 𝑆𝑖𝑛𝑔𝑙𝑒 𝑇𝑖𝑚𝑒 𝑃𝑜𝑖𝑛𝑡/ 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑃𝑒𝑜𝑝𝑙𝑒 𝑖𝑛 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑎𝑡 𝑆𝑎𝑚𝑒 𝑇𝑖𝑚𝑒 𝑃𝑜𝑖𝑛𝑡
What does Prevalence indicate?
Prevalence indicates the probability that a member of the populationhasa given condition at a point in time. It includes all cases who still have the condition and are still members of the population.
When is calculating prevalence of various conditions across different geographical areas or amongst different sub-groups of the population and then examining prevalence of other potential risk factors or health outcomes beneficial?
When planning health services
Prevalence is a way of assessing the overall burden of disease in the population. True or false?
True
When is Orevalnce not useful?
Prevalence is not a useful measure for establishing the determinants of disease in a population (causes and other factors that influence the occurrence of disease and other health-related events)
What is Incidence risk?
Incidence risk is the total number of new cases divided by the population at risk at the beginning of the observation period
What is Risk?
Risk is a way of quantifying the probability of a particular outcome within a specified period of time.
Risk is the proportion of individuals in a population (initially free of disease) who develop the disease within a specified time interval. Incidence risk is expressed as a percentage (or if small as per 1000 people).
Incidence risk equation
Incidence Risk = 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛𝑒𝑤 𝑐𝑎𝑠𝑒𝑠/𝑇𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑒𝑜𝑝𝑙𝑒 𝑎𝑡 𝑟𝑖𝑠𝑘(𝑑𝑖𝑠𝑒𝑎𝑠𝑒 𝑓𝑟𝑒𝑒 𝑝𝑒𝑜𝑝𝑙𝑒) 𝑎𝑡 𝑡ℎ𝑒 𝑏𝑒𝑔𝑖𝑛𝑛𝑖𝑛𝑔 )
What does the incidence rate assume?
The incidence risk assumes that the entire population at risk at the beginning of the study period has been followed for the specified time period for the development of the outcome under investigation.
In a cohort study participants may be lost during follow-up due to:
Develop the outcome under investigation
Refuse to continue to participate in the study
Migrate
Die
Enter the study some time after it starts
What is Incidence rate?
Incidence rates also measure the frequency of new cases of disease in a population. However, incidence rates take into account the sum of the time that each person remained under observation and at risk of developing the outcome under investigation.
To account for variations during follow up what more precise measure can be calculated ?
Incidence rate
Incidence rate formula
𝑰𝒏𝒄𝒊𝒅𝒆𝒏𝒄𝒆 𝑹𝒂𝒕𝒆=𝑵𝒖𝒎𝒃𝒆𝒓 𝒐𝒇 𝑵𝒆𝒘 𝑪𝒂𝒔𝒆𝒔 𝒅𝒖𝒓𝒊𝒏𝒈 𝒔𝒑𝒆𝒄𝒊𝒇𝒊𝒆𝒅 𝒑𝒆𝒓𝒊𝒐𝒅/𝑻𝒊𝒎𝒆 𝒆𝒂𝒄𝒉 𝒑𝒆𝒓𝒔𝒐𝒏 𝒘𝒂𝒔 𝒐𝒃𝒔𝒆𝒓𝒗𝒆𝒅, 𝒕𝒐𝒕𝒂𝒍𝒍𝒆𝒅 𝒇𝒐𝒓 𝒂𝒍𝒍 𝒑𝒆𝒐𝒑𝒍𝒆
If incidence rate remains the same, in which situations can the prevalence rate decrease?
If Cure rate increases
If Recurrence decreases
Suppose we were interested in the prevalence of diabetes in a nursing home with 800 residents.
If 50 of the residents were found to be diabetic, what is the prevalence at this point in time?
50/800= 0.0625
62.5 per 1000 resident or 6.25 per 100 residents or 0.0625%
Suppose we were interested in the prevalence of diabetes in a nursing home with 800 residents.
50 of the residents were found to be diabetic.
All the residents completed a 1 year follow up.
Over the next 12 months, 25 were found to be diabetic by the end of the year. What is the new incidence risk?
25/750= 0.0333 or about 3.3 per 100 or 3.3% over a year
What are biases?
Biases are systematic differences between the data that has been collected and the reality in the population.
How do we reduce bias?
Randomisation (experimental analytical studies)
Masking
Define selection bias
error in the process of selecting participants for the study (non representative to the real population) or/and assigning them to particular arm of the study (e.g. in RCTs)
Define attrition bias
when those patients who are lost to follow-up differ in a systematic way to those who did return for assessment or clinic.
Define observer bias
when variables are reported differently between assessors
Define procedure bias
subjects in different arms of the study are treated differently (other than the exposure or intervention)
Define response bias
occurs when the research materials (e.g., questionnaires) prompt participants to answer or act in inauthentic ways through leading questions.
Define misclassification bias
occurs when a variable is classified incorrectly
Define recall bias
systematic error due to differences in accuracy of recall to recall of past events / experiences; subjective interpretation
Why is randomisation important in reducing bias?
Random sampling methods help ensure that your sample doesn’t systematically differ from the population.
In addition, if you’re doing an experiment, use random assignment to place participants into different treatment conditions. This helps counter bias by balancing participant characteristics across groups.
Why is masking important in reducing bias?
Wherever possible, you should hide the condition assignment from participants and researchers through masking (blinding).
Participants’ behaviours or responses can be influenced by experimenter expectancies and demand characteristics in the environment, so controlling these will help you reduce systematic bias.
What are cross sectional studies?
Cross sectional studies look at measurements on a subject at more or less one point in time.
What is a longitudinal study?
A longitudinal study measures a subject over a period of time.
Imagine you wanted to look at how many
Problems with cross-sectional studies?
- is the data valid?
Sampling method: bias, representativeness
Where can bias arise in cross-sectional studies?
Bias might come from an unrepresentative sample, a badly worded questionnaire, unstandardized interviewers, non-response, conditions on that day that are unusual.
The study may even be affected by things such as the weather! If we are to look at ED attendances on hot and rainy days then the type of patients presenting might be drastically different. If we are enticing patients to come on 1 day to the GP surgery for a study and it snowed that day it might affect the type of patient that turned up.
Problems with longitudinal studies?
- higher costs
- higher admin burden
- subject migration or withdrawal
- changes to medical practice
What type of study can be used to compare prevalence in different times?
Cross-sectional studies
Advantages of cross sectional studies
- Commonly conducted to find out the prevalence and to some extent, association of factors
ability to rapidly generate data for policy makers or for generating hypotheses on the topic or for further research. - Contains multiple variables at the time of the data snapshot
- The data can be used for various types of research
- Many findings and outcomes can be analysed to create new theories/studies or generate hypothesis))
- Easy to conduct, require less time and financial resources
Disadvantages of cross sectional studies
- Cannot be used to analyse behavior over a period to time
- Does not help determine cause and effect
- The timing of the snapshot is not guaranteed to be representative. Variables e.g. environmental factors change over time (seasons of the year) and this factor should also be considered at the time of conducting such a study design.
- They are inefficient for rare diseases or diseases with a long latency period between exposure and disease manifestation.
- Prone to different types of bias such as non-response bias, recall bias, interviewer bias
- It measures exposure and outcome at a single point in time. Often it is problematic to assess the time-directionality of an effect – how can we be sure that an exposure did precede an outcome, if both are being measured at a single point in time? Also, it may even be that the outcome is causing the exposure, a problem in cross-sectional studies that we call reverse causality
What are cohort studies?
Subjects are selected based on a known exposure to the factor of interest e.g.
Babies whose mothers smoked whilst pregnant
Patients with a certain disease given a certain medication
Dentists exposed to drilling noise
Subjects are followed up to a specific end point and the exposure groups compared
Some groups may have graded exposure (e.g. smoking)
Some cohorts may need a control group
There should be objective evidence of exposure e.g. maternity records, disease register, certain profession
In cohort studies, subjects are selected based on a known exposure to the factor of interest. True or false?
True
In cohort studies, analysis is based on?
Measurement of risk (relationship of end point to proposed causal factor)
Comparison of risk in different exposure groups is?
Relative risk
What is the Relative Risk (RR) or Risk Ratio(RR)?
Is the measure of the incidence risk in one group, relative to the incidence risk in another (exposed vs unexposed).
Is RR>1
There is a positive association between exposure and outcome (exposure is harmful)
If RR<1
There is a negative association between exposure and outcome (exposure is protective)
If RR=1
There is no association between exposure and outcome
Formula for RR
Relative Risk = Incidence risk among exposure/ incidence risk among non exposed group
Incidence risk amon an exposed group = A/ A+B
Incidence risk among non exposed group= C/C+D
What is the Absolute risk?
The risk of an individual developing a disease over a time period
If absolute risk is 10 in 100 of developing heart disease in non-smokers and relative risk is 50% for smokers compared to non-smokers, how many smoker will get the disease?
15 in 100
What is a key to making an informed medical choice?
A key to making an informed medical choice is weighing potential benefits versus potential harms.
What is the Relative Risk Reduction (RRR)?
Is the relative decrease in the risk of an adverse event in the exposed group compared to an unexposed group.
RRR tells you by how much the treatment reduced the risk of bad outcomes relative to the control group who did not have the treatment.
Formula for RRR?
RRR = IRunexposed – IRexposed / IR unexposed (IR-incidence rate or relative risk)
What is Absolute Risk Reduction (ARR)?
Is the actual amount that the risk is reduced by the treatment.
The proportion of untreated (unexposed) persons who experience an adverse event minus the proportion of treated (exposed) persons who experience the event.
Formula for ARR?
ARR = IRunexposed – IRexposed
Study showed that 1.9% of people taking Lipitor (exposed) suffered a heart attack, while 3.0% of the placebo group (unexposed) had one.
What is the Relative risk reduction (RRR)
RRR = IRunexposed – IRexposed / IR unexposed (IR-incidence rate or relative risk)
RRR= (3.0-1.9) / 3.0 = 0.36 = 36%
Study showed that 1.9% of people taking Lipitor (exposed) suffered a heart attack, while 3.0% of the placebo group (unexposed) had one.
What is the Absolute risk reduction(ARR)?
ARR = IRunexposed – IRexposed
ARR= 3.0% -1.9% = 1.1%
What is the Number Needed to Treat (NNT) ?
This is the theoretical number needed to ‘treat’ (or enrol on intervention programme) in order to prevent one adverse outcome (e.g. death). It reframes the treatment’s benefits in a more user-friendly way.
Clinical decisions should be based on NNT alone, true or false?
False
Advantages of prospective cohort studies
People are recruited into cohort studies regardless of their exposure or outcome status. This is one of their important strengths. People are often recruited because of their geographical area or occupation, for example, and researchers can then measure and analyse a range of exposures and outcomes.
Cohort studies should include groups that are identical EXCEPT for their exposure status. As a result, both exposed and unexposed groups should be recruited from the same source population.
Incidence (temporal relationship between exposure and outcome) is clear.
Best evidence for causality among observational studies.
Good for assessing causality, prognosis, risk factors and harm.
Important when RCT are unethical to conduct
Disadvantages of cohort studies?
Key exposures of interest may go hand-in-hand with confounding factors.
Long follow-up period while waiting for events or diseases to occur. Often replaced by case-control studies
Ineffective to investigate diseases with low incidence rate, outbreaks, long latency diseases (retrospective cohort or case-control studies are more suitable for this)
Attrition. If a significant number of participants are not followed up (lost, death, dropped out) then this may impact the validity of the study. Not only does it decrease the study’s power, but there may be attrition bias – a significant difference between the groups of those that did not complete the study (increasing sample size could remedy this)
Can be expensive and time-consuming, especially if a long follow-up period is chosen or the disease itself is rare or has a long latency.
Subject to bias (confounding, associations, selection, observer, etc. bias)
What are case-series?
A series of cases
e.g.
Each patient that comes to neuro ICU with an SAH is monitored for early cerebral blood flow then outcomes in 3 months
OR
For each patient that comes with SAH the relatives are asked about potential risk factors leading up to the haemorrhage
What are the problems with case-series?
Problems are usually surrounding the selection of the matched control group. Study is stronger with several control groups. Questioner bias. Sequence of events may be unclear.
Advantages of case-series
Usually cheaper than cohort and more than one variable/exposure could be tested
The degree of matching of the participants will influence the results gained but may also may selection of participants much more difficult. If too much matching is achieved then this might affect the relative risk calculated.
Results are likely to be better if the questioner is blinded (doesn’t know which group the patient is in) to avoid unintentional questioner bias
When might you use a case-control study?
Useful when investigating cause of disease, because it helps determining if an exposure (risk factor) is associated with an outcome (disease, condition of interest)
Useful to study conditions with low incidence rate e.g.<5%
acute outbreaks,
rare diseases,
long latency diseases (time-lag between exposure & outcome)
When it might be unethical to expose the patient or do nothing following exposure
An example of a case-control study would be interviewing a group of mothers of children born with birth defects (cases, with the outcome of interest) to mothers of children born without birth defects (controls, without outcome the interest) and comparing their dietary and supplemental intakes of folic acid just prior to and during their pregnancies (exposure).
If a lower level of exposure to folic acid is found in the case group as compared to the control group, the hypothesis that lower levels of folic acid are related to certain birth defects can be supported.
Odd ratio formula
Odds of exposure in disease cases= Number of cases with exposure /Number of cases without exposure = A/C
Odds of exposure in controls= Number of controls with exposure/ Number of controls without exposure = B/D
Odds of exposure in cases/ Odds of exposure in controls = (A/C) / (B/D) = AD / BC = ODDS RATIO
To investigate the association between smoking and pancreatic cancer in female adult population, researchers conducted a case-control study. 100 cases and 400 matching controls we included in the study. Their findings are presented below:
Total with Cancer = 100
Total No cancer = 400
Smokers with cancer = 60
Smoker with no cancer = 100
Non-smoker with cancer = 40
Non -smokers with no cancer = 300
What is the odds ratio? AxD/BxC = ratio
OR= (60/40) / (100/300) = 4.5
OR- the odds of developing the disease is for 4.5 times higher in those who smoke compared to those who do not
Advantages of case control studies?
Slide 62
Disadvantages of case control studies?
Slide 62
LOOK AT WEEK 3 LECTURE FOR QUESTIONS
On blackboard
