EBH Flashcards
What is the James Lind Alliance and what is it used for?
- non profit organization
- used for identifying research priorities (prioritizes the top 10 unanswered questions)
How is the target population derived?
from eligibility criteria
How is the accessible population derived?
from the target population + limited by TIME and GEOGRAPHY
The accessible population is a subset of the target population. How is the accessible population derived?
Limited by TIME and GEOGRAPHY
What are the 2 MAIN subgroups of studies?
- Intervention
2. Observational`
What are the 2 main subtypes of Observational Studies?
- Analytic (has comparison group)
2. Descriptive (no comparison group)
What are the 2 types of descriptive studies?
- Cohort Study
2. Cross-Sectional Study
List the 4 different types of measurement scales
- Numeric
- Ordinal
- Nominal
- Dichotomous
What is an ordinal scale?
It is used to measure a relative quantity (e.g. a patient’s severity of pain or quality of life)
List 3 examples of ordinal scales that are used to measure PAIN
- Verbal Descriptor Scale
- Wong-Baker Facial Grimace Scale
- Activity Tolerance Scale
List 2 types of graphs that frequencies (frequency data) can be presented on
- Stacked Histogram
2. Bar Chart/Graph
Dot plots can be used to display data if 2 conditions are met. What are the 2 conditions?
- Continous scale
- Small amounts of data
If there are larger data sets, there will be too many dots and so we rely more on summaries
What is another name for a quantile?
a percentile
What is the percentile of the median?
50th percentile
List 2 summary measures that are ROBUST
- Median
2. Interquartile Range
List 2 summary measures that are NOT ROBUST
- Mean
2. Range
What is the definition of variance?
The average of the squared differences from the mean
How is standard deviation calculated?
it is the square root of the variance
What is the definition of skewness?
an excess of low or high values
What is the definition of Point Prevalence?
the proportion of persons w/ a particular disease or attribute on a SPECIFIC date
e.g. how many people have the disease?
What is the formula for point prevalence? (i.e. how do you calculate the point prevalence?)
(Number of Cases) / (Total Number of People at Risk at that Point in Time)
What is the definition of Period Prevalence?
the proportion of persons w/ a particular disease or attribute within a SPECIFIED TIME PERIOD
What is the formula for period prevalence? (i.e. how do you calculate the period prevalence?)
(Number of Cases at Any Time in the Study Period) / (Total Number of People at Risk at the Start of the Study Period)
What is the prevalence useful for? (what is it used to determine?)
- tells us the burden of disease
- useful for measuring the demand for health services
How is the cumulative incidence calculated?
(# of new cases) / (total # of people at risk)
[this is all during a designated time period]
What is the advantage and disadvantage of cumulative incidence?
Advantage: tells you how many patients will emerge over time
Disadvantage: not useful for conditions that may happen repeatedly to the same person
What is the incidence density? How is it calculated?
Incidence Density: incidence rate; tells us the rate of events per person-time (usually per person-year)
= (# of new cases) / (total person-time at risk)
What is the main advantage of calculating incidence density rather than cumulative incidence?
it is useful for diseases that recur
Do short-lasting diseases have a high or low prevalence?
low prevalences (e.g. the common cold)
Do long-lasting diseases have a high or low prevalence?
high prevalences (e.g. angina, HIV/AIDS)
List 2 values that can be used to evaluate a SCREENING test
- Sensitivity
2. Specificity
Sensitivity can be used to evaluate screening tests. What is sensitivity?
it is the percentage of people WITH the problem who are correctly identified by the test
Specificity can be used to evaluate screenings tests. What is specificity?
it is the percentage of people WITHOUT the disease who are correctly identified by the test
How is sensitivity calculated?
Sensitivity = (True Positive) / (All People with the Disease)
How is specificity calculated?
Specificity = (True Negative) / (All People without the Disease)
List 2 values that can be used to evaluate a DIAGNOSTIC test
- Positive Predictive Value
2. Negative Predictive Value
What is the positive predictive value and how is it calculated?
PPV: the percentage of all positive tests that are true positive tests
PPV = True Positive / All Positive Tests
What is the negative predictive value and how is it calculated?
NPV: the perventage of all negative tests that are true negative tests
NPV = True Negative / All Negative Tests
What is another name for the Relative Risk?
Risk Ratio
What is the general formula for calculating the Risk Ratio?
RR = (Incidence of Disease in EXPOSED Group) / (Incidence of Disease in UNEXPOSED Group)
How is the Incidence Rate Ratio calculated? (what is the formula?)
IRR = (Incidence Density in EXPOSED Group) / (Incidence Density in UNEXPOSED Group)
To calculate the relative risk, you need to know the….
Incidence
To calculate the relative risk, you need to know the incidence and this cannot be determined in which type of studies?
Case-Control Studies
What is the purpose of the Odds Ratio?
- it measures the association b/w exposure and outcome
- usually done when the incidence cannot be calculated
- it is a good approximation of the relative risk
For which studies (predominantly), is the prevalence ratio calculated for?
Cross-Sectional Studies
What is the formula for the Prevalence Ratio?
PR = (prevalence of outcome in EXPOSED group) / (prevalence of outcome in UNEXPOSED group)
predominantly used for Cross-Sectional Studies
What is the formula for the Relative Risk Reduction?
1 - Relative Risk
What is the Risk Difference (definition)?
Risk Difference: the difference b/w the risk of an outcome in the exposed group vs the unexposed group
- can be calculated using either the cumulative incidence or incidence density
- unit = PER PERSON YEARS
How can the Risk Difference be calculated?
Risk Difference = (Incidence in EXPOSED Group) - (Incidence in UNEXPOSED Group)
can be calculated with either cumulative incidence or incidence density
What unit is used for Risk Difference?
per person years
How can the Number Needed to Treat (NNT) be calculated? (most likely/common method?)
NNT = 1 / (risk difference)
Note: for NNT, always round up
The relative risk (risk ratios) can be calculated for what type(s) of studies?
- Cohort Studies
2. Intervention Studies (e.g. RCT)
What measure of exposure effect (e.g ratio) can be calculated from a CASE-CONTROL study?
Odds Ratio
What are the 2 types of error?
- Random Error = by chance
2. Systematic Error = bias
What is precision? (give the definition in regards to random error)
Precision is the lack of random error (therefore values fall within a narrow range of values)
Does a smaller or larger sample size reduce the random error?
A larger sample size reduces the random error and gives you a more PRECISE estimate
To get an ACCURATE estimate of values, what is required of the sample population?
the sample should be REPRESENTATIVE
List the 2 types of error that can arise during the selection of participants in a study
- Sampling Error - random error simply due to chance
2. Selection Bias - arising from a non-representative sampling method
What type of error is sampling error?
Random Error (due to chance)
List 2 types of error that can arise when using instruments for measurement in studies
- Inaccuracy - Bias/Systematic Error
2. Poor Reliability - mainly due to Random Error
What is validity? (definition)
it is the capacity of a test, instrument or other to give a TRUE RESULT
What is reliability? (definition)
it is when repeated measurements give the same or similar values
What is the type of error that may arise when there are MULTIPLE observers recording measurements in a study?
Systematic Error - Bias
What type of error is present if there is inconsistencies in values measured by the SAME observer?
most likely Random Error (chance)
To minimize random error, what should be done?
Increase the number of participants or the amount of measurements
To minimize bias, what should be done?
- representative sampling
- use of valid measurement instruments and techniques
List the 5 different types of probability sampling methods
- Simple
- Stratified
- Cluster
- Multistage
- Systematic
List the 2 different types of non-probability sampling methods
- Convenience
2. Snowball
What is a Sampling Frame? (definition)
it is an accurate list of all the people in the population of interest (e.g. including traits like age, sex, contact details)
What is Simple Probability Sampling?
when individuals are randomly selected from the population
What is Stratified Probability Sampling?
when individuals are organized into “strata” (according to characteristics) then randomly selected
What is Multistage Random Sampling?
random sampling that is carried out in stages
e.g. simple random sampling to select random SS schools, then randomly select 50% of children from each school
What is Systematic Sampling?
is the selection of individuals based off of a set system/plan
e.g. selecting every 5th member of the population
What is Convenience Sampling?
no structure to identifying people
e.g. asking everyone waiting in a waiting room to take part
What is Snowball Sampling?
finding people in the population who can use their network and contacts to find other people
What is Statistical Inference? (definition)
it is when the sample is used to make inferences about the population
(small group of people used to make inferences about the larger group)
What is the Null Hypothesis (H0) and Alternative Hypothesis (H1)? (definitions)
Null Hypothesis: a claim of no difference or no association; the probability that the pattern in the data arose by CHANCE
Alternative Hypothesis: a claim of a difference or an association; the probability that the pattern in the data did NOT arise by chance
What test can be done to see:
How likely it is that the difference between the MEANS of 2 groups arose by chance?
t-test
What test can be done to see:
How likely it is that the VARIATION between the MEANS of a number of groups arose by chance?
Analysis of Variance (ANOVA)
What test can be done to see:
How likely it is that the association between the rows and columns of a table arose by chance (qualitative variables)?
Chi-squared test (if sparse data, Fisher’s exact test)
What test can be done to see:
How likely it is that the difference in ordinal response data of 2 groups arose by chance?
Mann-Whitney test
What test can be done to see:
How likely it is that the differences in survival between several groups arose by chance?
Logrank test, Cox regression (Kaplan-Meier used to graph data)
If a regression coefficient is 0, what does this mean in regards to null and alternative hypothesis?
The null hypothesis can be accepted as there is no relationship/association
If a regression coefficient is NOT 0, what does this mean in regards to the null and alternative hypothesis?
The alternative hypothesis is accepted (the null hypothesis is rejected) as a relationship exists
List 3 factors that affect the range/width of a confidence interval
- Sample Size - a larger sample will lead to a better estimate
- Variability - greater levels of variance yields larger confidence intervals (more uncertainty)
- Degree of Confidence - the greater the confidence, the wider the interval
If the sample size is larger, what will happen to the confidence interval?
It will get narrower/smaller
Big samples give more precise measurements (less uncertainty)
If the sample size is smaller, what will happen to the confidence interval?
It will get larger
Small samples give less precise measurements (more uncertainty)
What is the general formula for calculating confidence intervals?
Size of CI = (variability of thing being measured) / (amount of measurements made)
List 2 measures of exposure effect that can be calculated from cross-sectional studies
- Prevalence Ratio
2. Odds Ratio
Why can’t the relative risks be calculated in cross-sectional studies?
RR needs the incidence and you are not looking at new cases in cross-sectional studies
What does a Pearson’s Correlation Coefficient of 0 mean?
there is no LINEAR association b/w the 2 variables - however, it doesn’t mean that they are not related, just that they’re not linearly related
What is confounding? (definition)
there is an apparent relationship between the exposure and outcome, but this is due to a third variable
List some of the limitations of cross-sectional studies
- prone to bias
- results may be due to confounding
- difficult to establish causality
What is the ecological fallacy?
- for ecological studies
- these studies use aggregate (grouped) data and so, they give you information on the group, but cannot be applied to individuals
- the associations seen b/w aggregated data does NOT exist at the individual level
List the 3 components of a cohort study
- Follows a defined group of participants
- Over a period of time
- Records the incidence of events of interest
What is a population cohort study?
They follow up population groups to document the incidence of diseases (and try to find the associated risk factors)
What is a clinical cohort study?
They follow up patient groups to document the natural history of a disease
What is Survival Time and what type of studies are they used for? (definition)
Survival Time: the time until the event of interest (outcome) occurs
used in cohort studies to account for variable follow-up
What are the 2 types of cohort studies?
- Population Cohorts
2. Clinical Cohorts
List some of the strengths of cohort studies
- can study multiple different endpoints/outcomes
- can record incidence
- less biased
- external validity
List 2 of the main disadvantages of cohort studies
- Impractical for diseases with a LONG pre-clinical phase
2. Impractical for RARE diseases (e.g. rare outcomes)
(Statistical) power is the probability… (definition)
that we reject the null hypothesis when its FALSE
Case control studies are not suitable when the exposure is…
RARE
List some of the advantages of case control studies
- useful for rare diseases
- useful for diseases with long latent periods
- cheap, quick
- can study multiple exposures
List 4 sources of error in case control studies
- Selection Bias (e.g. hospital controls)
- Observer Bias (aka Interviewer Bias)
- Recall Bias
- Confounding
What can and cannot be calculated in a case-control study?
- prevalence
- incidence
- relative risk
- odds ratio
CANNOT calculate:
- prevalence
- incidence
- relative risk
CAN calculate:
- Odds Ratio
What is Simpson’s Paradox?
- a possible effect of CONFOUNDING
- it reverses the direction of the association (makes a positive one negative and vice versa)
Why may a multivariable regression be used?
- to deal with CONFOUNDING
- it adjusts the estimated effect of an exposure on an outcome for the effect of other potentially confounding factors
What are Preclinical Studies and what occurs in them?
- in vitro and in vivo (animal) studies
- assessment of pharmacology and toxicity
What are Phase 0 Trials and what occurs in them?
- short duration
- < 15 volunteers
- ‘microdosing’
- initial eval. of pharmacokinetics
What are Phase 1 Trials and what occurs in them?
- healthy volunteers
- 20 to 100 ppl
- assesses drug metabolism, bioavailability, toxicity
What are Phase 2 Trials and what occurs in them?
- patients
- 50 to 300 ppl
- measures effectiveness
- identifies side effects
2a) determines clinical efficacy
2b) determines optimal (therapeutic) dose
What is determined in Phase 2a of a clinical study?
the clinical efficacy
What is determined in Phase 2b of a clinical study?
the optimal (therapeutic) dose
What are Phase 3 Trials and what occurs in them?
- large patient groups (300 - 3000)
- longer duration
- definitive assessment of efficacy
*List the 4 methodological features of a clinic trial?
- Randomized
- Controlled
- Blinded
- Intention to Treat Analysis
What is the Hawthorne Effect and how is it compensated for?
Hawthorne Effect: individuals modify their behaviour in response to the awareness of being observed
- patients don’t know what treatment they are given in clinical trials to reduce this effect
What is Intention-to-Treat analysis?
All persons randomized to a treatment are counted in the analysis whether or not they complete it or not.
Evaluating the treatment POLICY.
What is Per-Protocol Analysis?
In clinical trials, this analyzes only those who completed treatment as required (doesn’t include those who are not compliant)
What is Equipoise?
Equipoise: there must be a genuine uncertainty as to whether the new treatment is better or not; but there should be NO grounds for believing it causes harm
What is a Sensitivity Analysis and when may it be done?
a sensitivity analysis is when a meta-analysis is re-done with low quality studies removed
A meta-analysis can provide misleading results if…
the methodological quality is poor
A Forest Plot is used to display the results of…
a meta-analysis
What is Statistical Heterogeneity? (definition)
it is the extent to which results from different studies differ from each other in a systematic review
– if the results differ too much, pooling results may be misleading
What is the I2 statistic used for?
it quantifies the amount of statistical heterogeneity
What is Publication Bias and how is it visualized?
Publication Bias: when the outcome of a study influences whether or not it will be published or not
Publication bias can be determined by Funnel Plots
An asymmetrical funnel plot is evidence of..
possible publication bias
List the 5 different types of qualitative research designs
- Ethnography
- Narrative Research
- Phenomenology
- Case Study
- Grounded Theory
What is Ethnography? (definition)
is the study of individual cultures (e.g. social group, geographical, tribal, religious, lifestyle)
a form of qualitative research
What is Narrative Research? (definition)
the use of a narrative to collect + interpret info about the patient’s experience of illness (e.g. expressive writing interventions - EWI)
a form of qualitative research
What is Phenomenology? (definition)
it is the meaning of an illness for individuals, through lived experience
a form of qualitative research
What is a Case Study? (definition)
it is an intensive study about a person, group or a unit - it is aimed to generalize
a form of qualitative research
List the 4 forms of sampling in qualitative research
- Purposive
- Convenience
- Quota
- Snowball
What is Data Saturation?
- qualitative research
- when new data no longer brings in additional insights
- “informational redundancy”
List the 3 types of data collection in qualitative research
- Interview (unstructured, semi-structured)
- Focus Group Discussions
- Observational (participant vs. non-participant observer)
List the 4 types of qualitative data analysis
- Content (thematic and framework)
- Grounded Theory
- Narrative
- Conversational & Discourse Analysis
What is the COREQ checklist used for?
used for reporting qualitative research
List the 4 different types of mixed methods designs
- Convergent Parallel
- Explanatory Sequential
- Exploratory Sequential
- Embedded
List some vulnerable groups
- children
- adults that lack capacity (e.g. dementia, intellectual disability, mental illness)
- pregnant women
- elderly
What are the 3 requirements to do research on a vulnerable population?
- Addressing a research need/priority
- Can’t be done in a non-vulnerable population
- Has to be beneficial to the (vulnerable) group
List the 6 types of vulnerability
- Cognitive
- Juridic
- Deferential
- Medical
- Allocational
- Infrastructure
What is Assent? (definition)
is the agreement of someone who is not able to give legal consent (e.g. not of legal age)
What is Proxy Consent? (definition)
in which people with the legal right to consent delegate that right to another person (e.g. adults who lack capacity)
- personal representative
- professional representative
What are human challenge trials?
when the person is deliberately exposed to an infectious pathogen (usually done in the context of vaccine development)
What is the 10/90 Gap?
<10% of global health research funding goes to low income countries
but the low income countries is where there are 90% of preventable deaths
What are the 3 Rs in animal research?
- Replacement
- Reduction
- Refinement
