20. Clinical Trials Flashcards
What is a clinical trial in which the allocation is not random called?
quasi-experimental
what bias is removed by the investigator randomly allocating subjects to receive (or not) the treatment of interest in a RCT?
selection bias (if benefit is seen, it’s less likely to be due to confounding or other biases and thus more likely to be a true association)
heirarchy of study designs - what comes first vs last in investigation?
first: ideas, opinions, editorials, anectodal
- case series, case reports
- cross- sectional studies
- case control
- cohort
- RCTs
- systematic reviews (metanalysis)
ethical issues in investigator-assigned interventions
- when to conduct a RCT?
- clinical equipoise (at start of trial, there must be astate of clinical equipoise regarding the merits of the regimens tested, and teh trial must be designed in such a way as to make it reasonable to expect that, if successfully conducted, a clinical equipoise will be disturbed)
requirements of ethical trials
- social or scientific value
- scientific validity
- fair subject selection
- favorable risk-benefit ratio
- independent review
- informed consent
- respect for potential and enrolled subjects (*** PROTECTION OF HUMAN SUBJECTS MUST BE CONSIDERED BEFORE AND TRIAL STARTS)
nuremberg code (1947)
- in research “voluntary consent of the human subject is absolutely essential”
- subjects must have CAPACITY to consent, freedom from coercion, and comprehension of risks and benefits involved
- study must minimize risk and harm with a favorable risk/benefit ratio, have qualified investigators using appropriate study designs, and allow subjects to withdraw at any time
belmont report (1978)
- respect for persons (recognize personal dignity and autonomy of individuals, w/special protection of those with diminished autonomy)
- beneficence
- justice
IRB
- oversight to human subjects research and ensures conduct is ethical and scientifically sound
- empowered to approve (or disapprove) research
- period reviews and monitoring
Phase I drug/product development
- initial human trials
- small sample (10-100)
- healthy volunteers
- no comparison group
- primary goal to evaluate safety (adverse effects, maximum dose/toxicity, evaluate treatment mechanism)
phase II drug/product development
- expanded human trials
- small sample (100s)
- patients with disease
- comparator group (new drug vs placebo or standard)
- goals (expanded safety, determine optimal dosing/duration, efficacy)
Phase III drug/product development
- RCT
- large sample (100s-1000s) with disease
- comparator group
- expensive and costly (up to 50$ million)
- primary goal to determine efficacy (additional info on safety like adverse effects)
Phase IV drug/product development
post-market surveillance
- after FDA approval
- ongoing monitoring in real world setting
- identify additional adverse effects (low frequency reactions, high-risk groups, long-term effects)
- DRUGS CAN FAIL IN POST-MARKET SURVEILLANCE (eg thalidomide)
efficacy vs effectiveness
efficacy: benefit of the intervention or drug under ideal conditions
- RCT w/placebo comparison
- highly select population, strict inclusion/exclusion
- intervention is highly standardized
effectiveness: benefit of the drug/intervention in the real-world setting
- comparison to current standard
- generalize pop, few exclusions
- intervention implementation flexible
RCT “random allocation”
- what bias is eliminated?
must be unpredictable and concealed (allocation concealment) - investigator cannot know if the next person will be intervention vs control or bias can be introduced
- eliminates selection bias (removes potential for bias in allocation of subjects, on average makes groups comparable with respect to known and unknown confounders, but groups may be uneven by chance)
blinding
withholding info about allocation of interventino from those involved in the study (limits bias)
effect of un-blinding patients?
- reporting symptoms, side effects
- reporting benefits
- drop out; alternative tx
effect of un-blinding doctors?
- administer alternative treatments
- reporting symptoms/benefits
major advantage of RCT vs observational studies?
elimination of selection bias
effect of un-blinding research staff?
- data collectors
- outcome assessments
- data analysis
post-randomization changes
- study participants drop out, switch treatment groups, fail to comply with assigned treatment
- benefits of randomization are loss
- can adjust for this in analysis
RRR (relative risk reduction)
(risk in standard care group - risk in treatment group)/risk in standard of care group
- careful, can over-inflate benefits (or risks) if the risk is super rare
absolute risk reduction (ARR)
risk of outcome in control group - risk of outcome in treat group
number needed to treat
NNT = 1/ARR
intention-to-treat analysis
analysis according to original group assignment
- simulates the “real world” which has high drop-out and non-compliance
- we do this to keep the benefits of random allocation
- How does the treatment work in the people for whom it was targeted
per-protocol analysis
- how does treatment work in people who actually received treatment?
- need to interpret knowing there might be selection bias
- if huge % dropped out, the study failed no matter what.
critical appraisal of RCT?
CONSORT statement:
CONsolidating Standards of Reporting Clinical Trials
advantages and disadvantages of RCT vs observational studies?
advantages: elim selection bias, if well done can be definitive trial
disadvantages: high resource (cost, time, effort), requires equipoise, may not be ethical
randomized cluster trial
- experiemnts in which “clusters” of participants (rather than single indivs) are randomly allocated to intervention groups
- use when interventions cannot be applied to an individual or when there is significant contamination across individuals
- disadvantages: complex design/analysis, needs more people for same power
meta-analyses
combines results from small studies to create one larger study
measures of association in RCT?
RRR, ARR, NNT
