evidence based medicine terminology Flashcards
Research and evidence-based medicine. In: Murtagh J, Rosenblatt J, Coleman J, Murtagh C. eds. Murtagh's General Practice, 7e. McGraw-Hill; Accessed February 24, 2021. https://murtagh.mhmedical.com/content.aspx?bookid=2471§ionid=199053198
EBM - what is a ‘valid’ method
a method that measures what it claims to measure
EBM - difference between internal and external validity
internal validity refers to how well it measures the things in the studied population
external validity refers to how generalisable it is to the general population
EBM - what is a ‘true positive’ in a diagnostic test
a positive test result in a patient(s) who have the disease
EBM - what is a ‘true negative’ in a diagnostic test
a negative test result, in a patient without disease
EBM - what is a ‘false positive’ in a diagnostic test
a positive test result in a patient who does not have disease
EBM - what is a ‘false negative’ in a diagnostic test
a negative test result in a patient who has a disease
EBM - how do you calculate the positive predictive value of a diagnostic test
A / (A + B)
where
A = True Positive
B = False Positive
EBM - how do you improve the positive predictive value of a diagnostic test
reduce the amount of false positives
EBM - how do you calculate the negative predictive value of a diagnostic test
D / (C + D)
where
D = True negative
C = False negative
EBM - how do you improve the negative predictive value of a diagnostic test
reduce the amount of false negatives
EBM - what are the implication of a positive test result in a test with high positive predictive value
higher PPV = lower chance of false positive -> likelier that disease is present
EBM - what are the implications of a negative test result in a test with high negative predictive value
higher NPV = lower chance of false negative -> likelier that disease is not present
EBM - how do you calculate sensitivity of a diagnostic test
A / (A + C)
where
A = True Positive
C = False negative
EBM - how do you calculate specificity of a diagnostic test
D / (B + D)
where
D = true negative
B = false positive
EBM - what does sensitivity of a diagnostic test measure
proportion of diseased people with a positive test result
EBM - what does specificity of a diagnostic test measure
how often a test is negative, when there is no disease present
EBM - how do you improve the sensitivity of a diagnostic test
increase the amount of true positives measured
reduce the amount of false negatives measured
EBM - how do you improve the specificity of a diagnostic test
increase the amount of true negatives measured
reduce the amount of false positives measured
EBM - why is a negative result in a sensitive test good at ruling out disease
sensitive tests have few false negatives. So if the result is negative, its more likely to be a true negative
EBM - why is a positive test result in a specific test good at ruling in disease
specific tests have few false positives. So if the result is positive, its more likely to be a true positive
EBM - define incidence
- number of new cases measured,
- in a defined population
- over a defined time period
EBM - define prevalence
- number of known cases in a population / number of people in that population
EBM - what is bias
any effect that occurs during research, that leads to measurements that depart systematically from the true values
EBM - what is selection bias
investigating a sample that is not representative of the population as a whole
EBM - how does bias affect the validity of a study
it loses internal validity
EBM - what is recall bias
bias that occurs from sample population’s recollection; can be
- people with the outcome being more likely to recall an exposure
- people with an exposure being more likely to identify the outcome as occurring
EBM - what is observer bias
bias that occurs from the investigator’s measurements; by measuring the outcome differently given they are aware of the studied individual’s exposure status
EBM - what type of study is most likely to suffer from recall and observer biases
case-control studies
retrospective cohort studies
EBM - what is a systematic error in measurement
an error that has a predictable, constant or proportional effect on measuring the true value
EBM - what is random error in measurement
deviations from true value caused by fluctuations in measurement
EBM - how do you compensate for random error in measurement
do multiple measures and take average
EBM - how do you compensate for systematic error in measurement
change study design so the error can’t occur
EBM - what effect do systematic errors have on the validity of a study
reduces the internal validity of a study
EBM - what is a type I error
an error where a study concludes there is a difference between 2 groups, but there isn’t in reality
EBM - what is a type 2 error
an error where a study concludes there is no difference between 2 groups, but there is in reality
EBM - clinical study design - broadly, the 2 types of study designs classified by how they handle exposures
observational studies - exposure is not modified by scientists
experimental / interventional studies - the exposure is modified by scientists
EBM - clinical study design - what is a case report
a report on a single patient with a particular disease or outcome, may go backwards to explore their exposures
EBM - clinical study design - what is a case series
a report on a group of patients with a particular disease or outcome, may go backwards to explore their exposures
EBM - clinical study design - what is a cross-sectional study
prevalence study ie looks at how many people in a population have a disease, and characteristics of the population
EBM - clinical study designs - what is the difference between, and similarities between, descriptive and analytic studies
descriptive and analytic are both observational ie you don’t mess with any of the exposures, just record whats observed
analytic studies will compare 1 group against another, whereas a descriptive study just tells you what happens in a particular group of people
analytic studies allow you to make inferences (hypotheses) to compare 1 group against another
EBM - clinical study design - what is a case control study
Take a group of people with a disease
Take a matched group of people without the disease
look back and check their exposures
EBM - clinical study design - what type of analytic study is ideal for rare disease, why
case controls; their design is outcome -> exposure, so you can recruit sufficient people
EBM - clinical study design - what is a diagnostic accuracy study
a type of study that compares the diagnostic accuracy of a new investigation test, against a gold standard test
EBM - clinical study design - what type of study are diagnostic accuracy studies usually classed as
case-control studies
- you have a group of people with the disease, and without the disease
- you are then comparing test results between the 2
outcome -> exposure
EBM - clinical study design - what is a cohort study
a study where you take 2 or more populations with / without an exposure, and assess whether they have a particular outcome over a time period
EBM - clinical study design - what is an audit
an audit is a study that measures current standard of care against an ‘ideal’ standard, usually a guideline.
Can also measure whether the ‘right results’ are produced ie outcomes are changing
EBM - what is a ‘null hypothesis’
means that the relationship between 2 data points is occurring by chance ie they are not related
EBM - what does it mean to ‘reject the null hypothesis’
rejecting it means that that the 2 data points measured have a relationship to one another
EBM - how to calculate risk
(number of events in a group) / (number of members of a group)
EBM - how to calculate odds
(number of members who had the event) / (number of members who did not have the event)
EBM - how to calculate a risk ratio
consider 2 groups - one had an intervention (I), the other did not (C)
risk in intervention = (number of intervened people with outcome) / (number of people intervened)
risk in control = (number of controlled people with outcome) / (number of people in control)
risk ratio = risk in intervention / risk in control
EBM - how to calculate an odds ratio
consider 2 groups - one had an exposure (E), the other had no exposure (N)
odds in exposure group = (number of people exposed with outcome) / (number of people exposed without outcome)
odds in non-exposure group = (number of people unexposed with outcome) / (number of people unexposed, without outcome)
odds ratio = odds in exposure group / odds in non-exposure group
EBM - why can’t you calculate a risk ratio in a case-control study
we’re going from outcome -> exposure, so we don’t actually have an idea of how many people are in the exposure group
