-
BENEFITS
-
Provide credible evidence
- Statistical tests can verify
- Investigator control over sudy aspects/interventions
- Potentially impacts health care practice
- FDA requirement fore new drug entities
-
DISADVANTAGES
- More complicated study design and reporting needed for detailed methodology
-
Published clinical trial is NOT always clinically useful information
-
CONSORT statement
- Designed to improve quality of reporting clinical trials in medical
- helps get most accurate and unbiased results → checklist and flow chart
-
Positive vs negative
- See negative results less in lit. Info can still be useful however
-
STRUCTURE
- Topic
-
Sections
-
ABSTRACT
-
Study overview
- May have a word limit
-
INTRODUCTION
- Study rationale, purpose, objective
-
Background info
- Literature review possible
-
METHODS
-
Controlled clinical trial
- PROSPECTIVELY determines difference in effect between two therapies
-
TYPES
-
historical controls
- aka "before and after studies"
- Rationale
- Compares results from new intervention with those from patients treated with paste intervention
- used with alternative therapy is unavail and placebo control considered unethical
- Advantages
- Only one group is needed for study
- useful for rare diseases or with high death rates
- Disadvantages
- Overestimating the effect of the intervention
- Not true randomization
- The participants and procedures may differ in new study vs the old study
- People healthier back then
-
randomized control
- PREFERRED METHOD
- Advantages
- Minimize bias
- Can examine multiple interventions
- Can help determine CAUSE AND EFFECT relationships
- CONSORT guidelines promote study quality
- Disadvantages
- cant used to dermine impact of harmful interventions
- that would be unethical!
- Can be costly and labor intensive
- Obviously
- dropout can compromize randomization benefits
- Multicenter groups may not be comparable to individual clinician patient groups
- Commercial sponsors can affect the outcomes of clinical trials
- TYPES
- Placebo-controlled
- rationale
- to document changes in dependent outcome variables due to factors other than the treatment under study
- Placebo =
- no pharmacologically active ingredient
- Placebo effect =
- change in dependent variable due to factors other than treatment effects
- "feel improvement that is not part of the actual treatment"
- Hawthorne effect =
- Increase in performance due predominantly to being chosen for the study
- "better results when the participants are aware that you are studying them"
- Active-controlled
- Rationale
- To compare safety/efficacy of intervention with active ingredients
- Used when denying effective therapy is unethical
- If we tested warfarin, we cant just give placebo to the others
- subtypes
- Equivalence trials
- No significant difference in safety/efficacy between standard and alternative treatment
- Noninferiority trials
- Newer treatment is NO WORSE than another treatment
- Superiority trials
- to establish if the new treatment is better than the standard treatment
- common randomized/controlled trial designs
- Parallel design
- def.
- Fixed # of study participants randomized to one of N treatments, followed for some set times, and then discountinued
- Advantages
- MOST COMMONLY USED TRIAL DESIGN
- Not subject to carryover effects
- Not affected by natural history of disorder
- Potentially shorter trial duration
- Disadvantages
- Potential for unbalanced groups
- Between-patient variability
- Used to assess
- Effects when treatment is curative
- Therapeutic effects with prolonged treatment
- When effects of one treatment persist after therapy (carry over)
- Comparative efficacy of large #s of treatments
- Crossover design
- def.
- Participants assigned to one treatment and then switched to alternate treatment at some point
- general
- Order of treatment should be random to avoid confounding of results
- Advantages
- Each person serves as his own control
- Less variability b/w treatment groups
- Sample sizes are much smaller → less costly
- Disadvantages
- Carry over effects
- Disease course can change over the study period → effects interpretation of the results
- Data analysis much more affected by patient dropouts
- Lose 2 patients instead of 1
- Utilized when
- When trial has SHORT term carry over effects
- Treatment PERIOD does not alter difference between treatment effects
- A baseline measurement cannot be made
- Treatment does not result in an ↑ dropout rate
-
Study Population
-
Terms
-
Target population
- Pop which investigator desires to generalize
-
Study population
- Population under study, population from which SAMPLE is drawn
-
Sample
- Subset of the population under study
-
Pic
- Topic
-
Participant selection
- Need to make sure that they represent the target population
-
Inclusion/Exclusion criteria
-
Must be clearly defined
- INCLUDE those who would benefit from intervention
- INCLUDE population that will support hypothesized results
- EXCLUDE those who may be harmed
- EXCLUDE those who will not comply
-
Typically adderss
- Pregnant, lactation, comorbid, renal and hepatic function
- Direcly correlate with study results
-
Blinding
-
def.
- Treatment status for experimental and control groups is concealed from patients and/or investigators
-
Advantages
- ↓ possible bias
-
Disadvantages
- double bind are more difficult to conduct
- Potential ethical issues with triple blind
- What happens if adverse rxn occurs and no1 knows what they are taking
- Risk of bias with single and unblinded
- Difficult to mask adverse event unique to a specific treatment arm
-
Types
- No Blinding
- investigator and participator are aware of the treatment
- Simpler and may reflect clinical practice more accurately
- Single blind
- Either investigator or participator (but not both) know about the treatment
- May provide better care to the patient
- Double blind
- Neither if them know what the treatment is
- GOLD STANDARD
- Triple blind
- participator, investigator and personnel do not know what the treatment is
-
Double dummy technique
- used
- to maintain blind in which each API treatment has a placebo identical to it
- rationale
- Needed when the API has a different rout of administration
- Ex. Tylenol..testing to see how sprinkles vs liquid works
- To make the taste or smell of treatment
-
Random Assignment/randomization
-
Goal
- to balance intervention and groups by providing study objects with equal chance to be in any treatment group
-
Advantages
- Minimizes potential for bias
- Promotes comparable study groups
- Ensures researchers dont create significantly different groups
- ensures that statistical difference is due to the treatment and NOT external factors
- Guarantees validity of statistical results
-
Disadvantages
- Can lead to unequal numbers in treatment groups
- Ethical considerations
- You CANNOT NOT treat an asthmatic
- Typically not possible with rare diseases
- Would take lots of time and $ to do this
-
Intervention vs control
- Intervention
- NEW therapy under investigation
- Control
- Therapy used in trial to measure difference in effect compared to study treatment
- Ideally should come from same target population as intervention
- Intervention-control group comparisons
- Statistical analysis
- Want P value less than 0.05 to be significant
- Unbalanced allocation
- Unequal # of subjects in intervention group compared to control
- Ex. 2:1 allocation
- Usually happens to gather more data on the therapeutic group compared to the control
-
Common randomization protocols
- Simple random assignment
- General
- uses outcome of random event to determine participant placement
- Ex. Day of week, birthday, random number table
- Advantages
- Easy, minimizes selection bias
- Disadvantages
- # of subjects can be unbalanced, especially in small groups
- Potential for reduced ability to detect differences and loss of trial credibility
- Stratification
- General
- Subjects have equal probability of being distributed within treatment groups
- Groups contain similar proportion of CONFOUNDING Variables
- Advantages
- Breaks trial into "smaller trial" and reduces variability in groups comparisons
- Disadvantages
- Cannot stratify for all variables that may affect study
- May not stratify for confounding variables found during the course of study
- Block randomization
- General
- Subjects have equal probability of being distributed EVENLY within blocks of equal size
- Advantages
- ensure similar participation numbers in experimental and control groups
- Maintains balance in confounding factors if a block/group needs to leave the study early
- Disadvantages
- Data analysis is more complicated
- Investigators may be able to predict the next allocated treatment
- TYPES
- nonstratified
- Block size is chosen (4,8,16,24)
- Number of patients in each is proportional within the block
- Stratified
- subjects are grouped into homogenous blocks on the basis of a variable
- NOT the intervention variable
- Should be influencing the outcome
- Ex. Blocks of eight stratified by age and sex
-
Bias
-
Def.
- Systematic error presented into some aspect of study by experimental design or methods that promotes one outcome over the other
-
TYPES
-
Publication bias
- Inclination to publish more trials with positive results and omit those with negative results
-
Investigator bias
- Investigator interpretations favor treatment results that are expected or desired and reject other treatment results
-
Selection sampling bias
- When participants selected are not equally balanced or objectively represented
- Results in attributes that are over/under represented
- TYPES
- Study population charac.
- ex. Of the birthing room
- Those who chose special room volunteered for the experience
- Had lower complications
- Those in the standard room probably had complications before or were older
- Run-in-phase
- Short term period (2-4 weeks) prior to study initiation to assess incidence of a predetermined outcome defined by investigators
- Problem =
- May end up with more motivated group and not reflect the patient population
- Subject recruitment
- Sending out letters to recruit for participants in raffle
- Had more motivated students = bias
-
RESULTS
-
Subject demographics
- Dropout/compliance
-
Data analysis methods
-
TYPES
- Intention to treat (ITT)
- Data analysis based on ALL study participants randomized (including those that dropped out
- Advantage
- Provides more realistic estimate of safety and efficacy of GENERAL population
- Disadvantage
- The dropout can BIAS the trial results
- Modified Intention to Treat
- allows some exclusions if they are unlikely to bias the results
- Advantage
- Useful when outcomes are not measured in all participants
- Disadvantage
- Not clearly defined and lack of guidelines can introduce bias
- Per-protocol method
- Data analysis is based on only those who COMPLETED the trial
- Advantage
- Accurate estimate of treatment effects on participants who fully comply with protocol and finish
- Disadvantage
- Study dropout due to adverse effects may cause overestimation of results
-
Study Outcomes
-
Primary Endpoint
- The outcome the clinical trial is specifically trying to measure
- Composite endpoint
- Combination of end points into one main endpoint
- Ie. Look at mortality, stroke AND MI
- Does not hold as much weight as the primary end point
-
Secondary Endpoint
- Results results of less interest that might contribute supporting evidence
-
Subgroup analysis
- Evaluation of subsets of results based on specific demographics (age, disease)
- Prequisites
- Clinical trial should be well designed
- Multiple subgroup analysis should be avoided
- Primary endpoint should be statistically significant
- Subgroup should be designed PRIOR to trial initiation
-
Ancillary vs adjunctive therapies
- Ancillary
- Additional therapy that is NOT equally distributed between groups
- Ex. Allowed to take other meds for additional BP control
- Adjunctive
- additional theray that can affect study outcomes but is distributed EQUALLY between the groups
- Ex. All patients told to get on a diet during trial
-
IX Discussion/Comments/Conclusion
- Study strengths/limitations
- Comparison to previous trial results
-
X Acknowledgements
- Author information/qualifications of researchers
-
XI References
- Peer reviewed, $$ support