Introduction to Clinical Trials Flashcards
What are the three key features of an RCT broadly speaking?
Controlled
Unbiased
Adequately sized
List some different types of intervention that may be evaluated by a clinical trial
Drug therapy Medical procedures Educational intervention Screening Trial Vaccine trial Medical therapy Complex interventions (e.g effects of educational programmes)
What is the Declaration of Helsinki?
International recommendation on the conduct of medical research on human participants to which all clinical trials should comply
What is the national ethical committee responsible for reviewing of trial protocols in the UK to ensure safety of future participants?
IRAS (Integrated Research Application System)
What is the role of the MHRA in the UK and FDA in the USA
Regulatory authorities approve trials of investigational medicinal products and medical devises to ensure trials have been conducted with sufficient quality and standards and further to licence products which are effective and safety has been properly evaluated
Outline the role of an Observational Trial
Non-experimental study where the individuals are observed and data collected with nothing being done to influence either the exposure or the course of events.
What is the major flaw of observational trials?
Can be subject to considerable bias if not designed well. Often have very different outcomes to RCTs
Outline the different phases of clinical trials
Phase I - Testing effect(s) of novel therapy on healthy sample
Phase II - Dose-response curve
Phase III - large sample size
Phase IV - post-marketing (long-term safety and interactions)
What is the most common trial design?
Parallel trial - New vs standard therapy
List some other trial designs other than parallel
Cross-over Factorial Equivalence Bioequivalence Non-inferiority Superiority Exploratory Confirmatory Cluster
Outline the utility of single vs multi-centre trials
Single centre - used for Phase I and II trials
Multi-centre trials - good for large recruitment over short periods of time and increases generalisability of results
Outline three reasons why trial results may not represent the ‘truth’?
Biased in a predictable fashion
Contimainated with a confounding factor
Random chance
What are the most common sources of bias in trials?
Most commonly at selection and outcome measurement.
Also includes omission of data and loss to follow-up
How are confounding errors minimised?
Adequate planning and exploration of variables with sound scientific principles and reasoning.
Randomisation (e.g. stratified) to account for known dependent variables
Stratified analysis and regression analysis
How is the effect of random chance minimised in trials?
Large sample sizes, appropriate of results statistical analysis and meta-analysis
What are the advantages of uncontrolled trials?
Used to generate hypotheses and justifications of PIII trials
Can assess pharmacokinetic properties of novel agents
Might be the only study design possible due to ethical considerations
What are the disadvantages of uncontrolled trials?
Investigator bias
Susceptible to over-interpretation
Much less informative than RTCs
What is the purpose of the clinical trial protocol?
- Defines the clinical question
- Outlines how it is answered
- How will it be compliant to ICH-GCP, EUCTD, FDA/MHRA regulations
- What analyses should be reported
- How will results be interpreted?
What is the purpose of the clinical trial protocol?
- Defines the clinical question
- Outlines how it is answered
- How will it be compliant to ICH-GCP, EUCTD, FDA/MHRA regulations
- What analyses should be reported
- How will results be interpreted?
How are study objectives stated differently in a priori (proscpective) versus retrospective research?
Prospective (a priori) = stated as a hypothesis
Retrospective research = stated as a question
What are primary endpoints in a trial?
Measure outcomes that answer the primary question of the trial
What are secondary endpoints?
Ask other questions relevant to the study
What are the main type of endpoints investigated in RCTs?
Binary (e.g. death of any cause) Disease specific death Occurrence of symptoms QOL outcomes Utilisation of healthcare resources
What qualities must endpoints have to be clinical relevant?
Clinically meaningful
Answer the main question
Can be practically assessed
Occur frequently enough within the time frame of the study to be powerful enough
Death as an endpoint can be impractical for what reasons?
May be rare in the timeframe of the study
Death can results from factors other than those under scrutiny in the trial
What are composite endpoints?
Multiple outcomes measured
What stage of trials are composite endpoints more common?
Exploratory / Early-phase trials
How can composite endpoints be expressed in a trial?
Can be expressed as total numbers or in a hierarchical fashion (only counting the most serious endpoint)
What are the advantages of composite endpoints?
- Increases event rate therefore decreases participants required
- More comprehensive evaluation can be achieved including clinical, pathologic, radiographic and biochemical can be measured simultaneously
Limitations of composite endpoints are:
Assumption that avoidance of any one endpoint is as clinically significant as any other
Assumption that each outcome related to disease progress are equally important
Inconsistent results if some outcomes improve and some deteriorate.
What are surrogate endpoints?
A lab measurement / physical sign used as a substitute for a clinically meaningful endpoint that directly measures how a patient feels, functions or survives
Give some common examples of surrogate endpoints
Pharmacokinetic measurements e.g. conc/time curves / active metabolites
In vitro measurements e.g. ABx concentration (mean) required to inhibit microbial growth.
Radiological appearance
Disease markers
Macroscopic appearance of pathology
What are the advantages of surrogate endpoints?
Decrease sample size
Particularly used in phase II trials where changes in surrogate endpoints occur before adverse clinical events
What are the limitations of surrogate endpoints?
Relationship between surrogate and clinical outcomes must be well understood otherwise it can produce misleading results
What are the two main health-economic outcomes?
- QUALYs - measurement of the duration of an individuals life in the context of clinical significance
- Cost
What measurement can be made from both QUALY and cost?
Cost-effectiveness ratio in terms of QOL
What are the advantages of health-economic endpoints?
Creates a common variable with which to compare unlike outcomes such as sight-years gained, symptom free time etc.
What are the limitations of health-economic outcomes?
Economic analyses not easily transferrable across countries since cost of care and access to resources vary massively.
Effects of acute conditions may not have an effect on long term impact of QOL if not fatal therefore economic analysis may be inappropriate in these scenarios
Why is the process of patient selection so important? How do we define selection?
A set of eligibility criteria is essential to ensure that the results of the trial have external validity.
What considerations need to be taken into account when reviewing eligibility criteria?
Source of patients - sample size and clinical outcome balanced against practical considerations
Investigators and centres - when estimating potential participants a screening log is useful
What are eligibility criteria usually based on?
Patient characteristics
Diagnostic test criteria
Disease duration
Disease severity
Convention is to have 1-2 inclusion criteria and more exclusion criteria
What is the general aim of eligibility criteria?
TO recruit the most number of patients without including patients with significant load of co-morbidity which may affect results of the trial
IF specific diagnostic tests are required for patient selection when must these tests be done in relation to randomisation?
Before randomisation - stratification/minimisation techniques may be done to ensure balance across groups
How does disease status effect patient selection?
May be necessary for a washout period pre-trial beginning
In trials it may be unethical to withold treatment and therefore more patients are recruited to the treatment in 1:2 ratio
What ethical issues are considered with regards to patient selection?
Consider age ranges that reflect the general population
Special considerations
Consider the inconvenience of the trial (e.g. time off work, repeat visits, invasive tests e.g.) - properly counsel patients
Why might patients not be selected?
Patients unavailable during the screening patients
Study personelle not available to screen patient
Drug/intervention not available
Procedure unavailable e.g. broken ECG machine
Patient lost to follow up
If a large number of patients are excluded by the criteria what course of action may be indicated?
Review of criteria and potentially a protocol amendment and ethical review
What is the definition of clinical trial bias?`
Systematic distortion of the estimated intervention effect away from the true value caused by inadequacies of design conduct or analysis / publication of a trial.
List some common biases
Selection bias Bias in study management Observer ascertainment Bias from exclusion post-randomisation Publication bias
How is selection bias controlled?
Randomisation and concealment
How is bias in study management controlled?
Standardised procedures, equipment and training of staff across sites
How is observer ascertainement controlled?
Blinding
How is bias from exclusion post-randomisation controlled?
Intention to treat analysis or worst case scenario analysis
How is publication bias controlled?
Prospective registration of trials and publication of negative trials
What is allocation concealment?
Decreases selection bias by shielding randomisation code before and until treatments are administered.
What is blinding?
Decreases observer bias by shielding treatment after administration
What is the most common cause of accidental unblinding?
Variation in side effect profile between treatment and placebo
How may accidental unblinding due to side effect profile variation between treatment and placebo arms be controlled?
Three arm parallel trials
What is intention to treat analysis?
All participants who are intended to be treated are included in analysis despite non-compliance (may or may not have missing data)
What is worst case scenario analysis?
If dichotomous data. Assume drop-outs did badly and therefore if the analysis is still in favour of the intervention then must not be due to loss to follow-up
What are the issues associated with worst case scenario analysis?
May underestimate the true magnitude of effect of loss to follow-up
If the results are negative then the data is uninterpretable
It ignores all data except the final analysis
What is meant by per-protocol analysis?
Any missing data points are then not included in the final analysis
What is publication bias?
Editors favour positive findings and therefore pre-registration of the trial with the agreement to publish whatever results are found is good practice
List some common methods of randomisation?
Simple
Block
Stratified
Minimisation/adaptive
What is the definition of simple randomisation?
Allocates participants without regard to those who have already been enrolled in trial
What is block randomisation?
Periodically reinforces a balance in patients for each arm of the trial.
When is forced balance most essential for a trial?
- When recruitment is slow
- When the types of patients reviewed over time changes
- When the trial may be stopped early
- Routine practice changes over the course of the trial.
What is stratified randomisation?
Within each stratum block randomisation is used to allocate patients to each group.
Different strata account for a dependent variable and balance this between each arm of the trial.
What is minimisation / adaptive randomisation?
Assigning patients to each arm in a way that adjusts the probability of being assigned to each arm depending on existing imbalances in the groups. Starts in a simple randomisation fashion
When is minimisation a fvoured technique for randomisation?
In diseases w many prognostic factors and is becoming increasingly popular w computer / interactive voice response systems
What five factors affect sample size required for a trial?
- Expected event rate in control arm
- Smallest treatment effect that is clinically worth measuring
- Determining the significance level of rejecting the null hypothesis
- Choosing the power of the study
- Study design
How do we approximate the event rate in the control arm?
By retrospective research and review of previous studies/ registries
How do you work out absolute treatment effect?
Event rate in control arm - event rate in treatment arm = absolute event rate
How does one determine significance level for type 1 error?
Conventionally set at 95% (P-value =0.05) confidence however increasingly up to 97.5% (P-value = 0.025)
What is the conventional type 2 error rate? Why is it higher than conventionally accepted type 1 error rate?
80% power = 20% type 2 error rate.
Conventionally less stringent than type 1 error rate as the burden of evidence required to change common practice to a new therapy is high.
What variable directly affects the power of a study?
Number of participants
What effect does the study design have on the power and sample size of a study?
Parallel studies require larger sample size.
How is drop out rate accounted for?
Conventionally 1 year drop out rate is assumed to be 20% therefore requirement of sample size is increased by at least 20% from calculated minimum to adjust for this occurrence.
Are negative result trials simply due to small sample sizes?
Yes and no. Be careful with assumptions for alpha, beta and event rates between the arms of a trial. Meta-analysis can be helpful in increasing power of results.
