Study types Flashcards
Definition of systematic review
review which identifes, appraises & synthesizes all evidence that meets pre-specified eligibility criteria for a specific research question
Definition of meta-analysis
statistical analysis that combines results of multiple scientific studies
Definition of RCT
comparison outcomes of group receiving intervention with control group
Definition of Prospective vs retrospective Cohort study
groups classified based on exposure followed up prospectively to analyse outcomes
groups classified based on outcome look at retrospectively to analyse exposures
Definition of Case-control study
groups classified based on outcome and analysed retrospectively regarding exposure
Definition of Cross-sectional study
survey of a defined population at a single point in time
Phase I Clinical Trial
These are “first in human” studies usually conducted on a small number of healthy volunteers to determine the tolerability and safety of a new drug, as well as its pharmacokinetics and pharmacodynamics
Phase II Clinical Trial
These are “dose-finding” trials usually on patients - a lower number than Phase III but larger than Phase I - assessing optimal dosing (giving different doses and following up to determine the efficacy and tolerability/toxicity) and efficacy (biological effect of the drug)
Phase III Clinical Trial
These trials assess the effectiveness of a new drug/ intervention, as compared to the gold standard therapy (whilst always ensuring its safety), to inform the value of the new intervention in clinical practice. They are usually multi-centre randomised control trials on large numbers of patients.
Post-marketing surveillance (Phase IV)
Monitors safety of the drug after it is being marketed (pharmacovigilance), to detect any rare or long-term adverse events or side effects that were not detected during previous trials
Randomisation
The process of assigning trial subjects to treatment or control groups using an element of chance, so that each participant has an equal chance of being allocated to either group.
This is to reduce allocation bias and balance known and unknown possible confounding factors, such that any difference between the two groups is purely by chance.
Stratified randomisation
Prevents imbalance between treatment groups for known factors that influence prognosis or treatment responsiveness. As a result, stratification may prevent type I error and improve power for small trials (<400 patients), but only when the stratification factors have a large effect on the prognosis.
Allocation concealment
Refers to masking the randomisation code or sequence before patients are recruited so that investigators don’t know what group the next patient will be randomised to, thus avoiding selection bias
Blinding
Involves masking the allocation of the treatment group after randomisation, so one or more parties involved in the trial are not aware whether the participants are in the intervention or control group. This aims to remove bias, including performance and observer bias.
Efficacy
The capacity for the therapeutic effect of a given intervention, e.g. whether a drug demonstrates a health benefit over a placebo or other intervention when tested in an ideal situation. These are called explanatory trials.
Effectiveness
How well an intervention works in practice, i.e. in the real world outside a clinical trial. This tends to be lower than the efficacy. There are called pragmatic trials
Placebo
Traditionally refers to a substance or treatment with no active therapeutic effect. In medical research, it is unethical to give a substance without therapeutic effect when one is available. Hence, a placebo refers to the control used, for example, the gold-standard intervention in which the intervention of interest is being measured against.
Surrogate endpoint
A variable that is relatively easily measured and which predicts a rare or distant outcome of the intervention tested. It is not a direct measure of either harm or clinical benefit.
Surrogate outcome
Markers that are substituted in for a clinical outcome e.g. ejection fraction as an indicator of disease severity in heart failure
Composite outcome
pros vs cons
A combination of variables e.g. a scoring tool. Useful when each variable in the scoring tool is rare as it gives power to the study. Difficult to interpret.
Internal validity
the degree to which the causal relationships found in a study can be trusted (with bias reducing this).
Secondary research
E.g. systematic review/literature review. This means research that amalgamates the results of many studies, largely negating the influence of bias on the results.
Advantages of SR and MA
More accurate assessment of the ‘true’ population effect.
Larger patient numbers lead to increased power and reduce type II error (which is the erroneous conclusion that an intervention has no effect).
Less bias (funnel plot → publication bias assessment, heterogeneity analysis → method and data collection)
Disadvantages of SR and MA
Direct comparisons between studies are difficult, due to differing methodology.
It can introduce types of bias (selection, publication (consider grey literature), and language bias (non-English articles), selective reporting bias)
Steps followed in a SR
Check for existing reviews/protocols
Identify your research question
Define inclusion/exclusion criteria
Search for studies
Select studies for inclusion based on pre-defined criteria
Extract data from included studies
Evaluated risk of bias of included studies
Present results and assess quality of evidence
Steps followed in a MA
Frame a question (based on a theory):
PICOK method.
Run a search
Screen
Extract data
Determine the quality of information in the articles (assess internal validity but also use GRADE criteria). Critically appraise the information contained within each study. You must decide if the study meets the internal validity criteria. A method to ascertain the quality of each outcome within an article is to use the GRADE criteria (grading recommendations assessment,
development, and evaluation).
Determine the extent to which these articles are heterogeneous.
Estimate the summary effect size in the form of ORs and using both fixed and random effects models.
Construct a forest plot
Determine the extent to which these articles have publication bias and run a funnel plot.
Conduct subgroup analyses and meta regression to test if there are subsets of research that capture the summary effects.
What is the problem with surrogate outcomes?
May not reflect outcomes that are important to patients