outcomes Flashcards
what should the primary outcome be?
whatever marks the greatest theraputic importance to the studies research hypothesis
what problems do we see in clinical trials regarding the outcome measures they use?
- studies measuring the effectivness of intterventions for the same condition use different outcome measures
- DEFINITION of outcome measure
different definitions exist for the same outcome measure. E.g., “singleton live birth” some definitions include and others exclude still births in it. LOOK UP THE DIFFERENT DEFINITIONS FOR OUTCOME MEASURES. inconsistent definitions big problem.
- measure used
different methods might be used to measure the outcome measure which migh lead to diferent results - reporting
bias reporting of outcome measure
examples of different poutcome measures used
clinical measurment: blood test, urine test
clinician assessed outcomes: clinical global impression fo change, cause-specific mortality
- patient reported outcomes
what do different definition of the outcmome measure impact?
sample size
meta analyses
policy
how do different outcome measures affect the sample size?
because part of the sample size calculation requires the estimated effect size.
inconsistent definitions of the outcome measure increases the variability in the estimated effect size and variability of outcome
makes it hard to calculate the required sample size
can lead to both underpowered studies (not enough power to detect effect) or overpowered (unecessarily large or costly)
how do inconsistent definitions of the outcome measure affect meta analyses
can lead to hetrogeneuity in the included analyses. harder to estimate the overall effect size of the intervention
impacts the implication
what are patient reported outcome measures?
self report measure by n about the status of their health condition
different PROMS for a specific condition e.g., anxiety has lots
what are the psychometric properties of different scales
reliability
validity
resposivness - doesthe scale detect changes over time that matter to N?
ideally what do we want when picking an outcome measure?
use the same outcome as other studies addressing hte same research question
even when using hte same outcome, use the same measurment e.g., specific PROM
different ways to measure medication adherence
objective vs subjective
objective - pill count, biochemical measurees
sbjective - self report
problem when different types of outcome measures or different measures of the same outcome measure is used. What can we do to solve this?
use core outcome sets (COS)
these are a collection of outcome measures that are considered to be the minimum set of outcomes that should be reported for certain condition
these are chosen by patients, cllinicianc and researchers. ook through all outcoem measures then decide on the core outcome set.
what is a CRF
case report form
used to collect thedata and information collected over the course of the trial
why do we need to check the CRF to ensure statistical function is met?
make sure data collected will address the research question
make sure the outcomes specified in the protocol are collected
data is formatted in way that ensues analyses can be performed
validated PROMS are not altered and have no errors
what are different outcome categories?
objective vs subjective
surrogate vs substantive
primary vs secondaty
composite primary outcome
surrogate vs substantive outcome
substantive - what we want to know e.g., cardiovascular disease
surrogate - what we actually measure e.g., blood pressure
what must the surrogate measure do?
capture the effect of the intervention on the clinical outcome
why use a surrogate outcome?
reduces duration and size of trial - thus cheaper
less invasive
primary vs secondary outcomes
primary outcom - the varaiable capturing the most clinically relvant and convincing evidence related to the primary trial objective
secondary - measures that support the primary objective of measures related to econdary objective
what things do we need to know about the primary outcome for the sample size calculation?
- how it will be measured
- how often it will be measured
- is it measured at baseline
- do we know its distribution and variability
what is the problem with having more than 1 primary outcome measure
issues of multiple testing
increases type 1 error
need to consider correlation between outcomes - result of one outcome might influence the result of another. if the outcomes are correlated it makes it harder to interpret the results.
why is it a problem if two outcomes are correlated
- impairs interpretation of the study
- reduces the power in the study - might not be able to detect treatment effect even if it exists
composite primary outcome
combinig omre than two variables into a single outcome
must be pre-defined
why use a composite outcome measure
- increases numbe rof events detected reducing the required sample size
- avoids issue of picking an outcome measure when 2 areequally important
- avoids issues of multiplicity
problems of composite outcoem measures
- difficultl to interpret
- sometime s only vertain variables within are significant - attributes innaprioriate treatment effect across all
- if some are not as strong as others can dilute treatment effect
what do we need in an outcome measure for the trial protocol
- specific measurment used e.g., systolic blood pressure
- analysis metric - time-to-event? change from baseline
- method of aggregation e.g., proportion, mean
- time point for each outcome
- explain clinical relevance - how it relates to the study objectivesand how it evaluates the effecivness of an intervention
- explain how it measures the potential adverse effects in an intervention
problems with dichotomising a continuous outcome variable?
loss of information - reduces power
n close to either side of the cut off judged as different rather than simmilar (which they are)
do not use the sample median as the cut off
what are estimands?
a target of estimation for the treatment efefct.
what are the four attributes of estimands?
- the population (n with high blood pressure)
- outcome - BP at 12 weeks
- specify how to accound for intercurrent events
- population summary of outcome (what we are comparing between groups e.g., change in BP between groups)
what are intercurrent events?
events that occur during th ecourse of the trial that can affect the outcome measure
how can we properly account for intercurrent events?
estimand in the trial design should specify how events that occur during the trial affecting outcomes will be addressed
e.g., rescue medication (given if treatment is not effective or leads to advrese events) might improve the outcome measure inflating the treatment effect. account for the effect e.g., include the use of rescue medication as a covariate
