Lecture 25: Overview of Research Methodology Flashcards
Primary Research Question
– The one the investigators are most interested in answering
– The question on which the sample size of the study is based
– The outcome that must be emphasized in the reporting of results
Secondary Research Question
– Related to the primary question but are “less important”
• Explore a different outcome than the primary question
• Explore a different patient subgroup
Is drug effective (primary) and safe (secondary)
what research do clincians encounter most?
phase 3 or 4
typically most helpful for clinical decision making.
describe pre-clinical research
In vitro or animal studies
molecules to be tested to determine efficacy and safety
15-20 year process
describe Phase 1
Human pharmacology studies: drug tolerance, metabolism, interactions, pharmacokinetics, pharmacodynamics, dose finding
tests how we move from tube to pts, how body tolerates with healthy volunteers
describe Phase 2
Therapeutic exploratory studies: effects of various doses, surrogate (biomarker) outcomes
research for group of small patients, surrogate endopoint trials come into play, effect of a med on biomarkers or lab tests
describe Phase 3
Therapeutic confirmatory studies: demonstrate clinical use & establish safety profile.
– “Prospective studies comparing the effects and value of interventions() against a control
in human beings.”
prove drug is beneficial, safe, efficacious for a large pop
describe Phase 4
Therapeutic use studies: examine drug in broad or special populations, seek to identify uncommon adverse events.
– ”Long term studies or studies conducted after regulatory agency approval of the drug or device”
on market, look at special poplations (Chronic kidney disease), uncommon AE
what is the 1st q to ask to decide what type of study should be used?
descriptive (PO) vs analytic (PICO)
Q1: was the aim to describe a population (PO) or quantify the relationship b/w factors (PICO)
Quantify relationship b/.w exposure and outcome (drug and effect) = analytical type of research
what is the 2nd q to ask to decide what type of study should be used?
analytic - experimental vs observational analytic
Was the intervention allocated randomly?
- Randomized to active treatment and control, cross over studies
- Observational research: if not being controlled by research and not randomly assigned
○ Cohort vs cross sectional and case control studies
what is the 3rd q to ask to decide what type of study should be used?
cohort vs cross sectional vs case control study
Q3: when were the outcomes determined in relation to exposure we are trying to determine (after, same time, before exposure)
- Cohort study: AFTER the exposure - Cross sectional (analytic): exposure and outcome AT THE SAME TIME - Case Control Study: looking at outcome and know it before the information about exposure is available; identifying cases who we know have outcome of interest and look back to see whether or not they've had exposure in the past
look at Oxford levels of evidence
Benefits Level 1: therapy related qs with systematic reviews or randomized systemic revies Level 2: RCT L3-4: Cohort study or control study Level 5: mechanism based reasoning
Harms
Similar RCTs, sometimes only have observational research, not ethical to expose someone
If asking about how common a problem is, we won’t do a RCT, but survey or census
what are fundamentals to a study?
smoking cessation example
Exposure: early NRT nicotine replacement therapy
- Usually for course of 12 wks
Outcome: did the subject quit
Must include a comparison
Research q: In smokers, does NRT started 2 wks prior to the quit date improve cessation rates at 6 months
Quit rate is outcome
Revised research q: in smokers, does early NRT in comparison to NRT started on the quit date improve cessation rates at 6 months or 1 year
define confounding variable
“A variable other than the risk factor & outcome variable under study which is related independently to each of the other two and can create an apparent association or mask a real one.”
smoking, coffee and cancer
Randomized experimental studies will control confounding variables
observational vs experimental (randomized studies)
- Observational Studies
– No manipulation of study groups
• Only observe patients and outcomes
– Good for questions relating to harm (causation) - Experimental (Randomized) Studies
– Comparable to Cohort studies EXCEPT the exposure is determined by
the researchers
• Follow the 2 groups to see who has outcome of interest
– Good for questions of therapy/prevention
Randomized Controlled Trials
goal
The goal of a randomized controlled trial is to make exposed and control similar in all respects other than the intervention at the start of the study and maintain that balance as the trial progresses.
defined pop gets new treatment or control/current treatment
- outcome with disease or no disease
Exposure is determined by the researchers
– Researchers expose one group to a therapy & not the other group
– Then follow the 2 groups to see who has outcome of interest
Randomized Controlled Trials
strengths and weaknesses
• Gold standard to answer “therapy” or “prevention” questions
• Strengths
– Low susceptibility to bias vs. other designs
– Why? reduce confounders
• Weaknesses
– Expensive & time consuming
– Less generalizable (i.e., less applicability)
– May introduce “hidden bias”
efficacy vs effectiveness research
which does RCT does?
efficacy: does treatment work under ideal conditions?
effectiveness: does treatment work under ordinary circumstances?
RCT gives efficacy rather than effectiveness
Superiority RCTs
Superiority trial to show one intervention is better than another
– Most common RCT done
– Comparator most often is placebo
– Assesses whether the new intervention is better or worse than control
– Aim to rule out equality between treatments
Equivalence or Non-inferiority RCTs
Equivalence: show tha 2 drugs are same in terms of effectiveness and efficacy
Non-inferiority: New drug is no worse than an existing gold standard
– Sometimes it is unethical to do a placebo controlled RCT
– Aim to rule out differences between two treatments
“Quasi Experimental” Designs
– Definition: look like experimental design but lack true randomization
but there is some aspect of randomization
Cohort Studies
design
group of pts
exposed or not exposed
do they develop disease or not?
Randomization is not the process of detemrining when ppl are exposed or not exposed
Ppl are exposed because a pt decides to take an otc or hcp recommended
prospective vs retrospective study
Prospective study that happens forwardf in time, in the future
- Not always feasible
Prospective study that happens forwardf in time, in the future
- Not always feasible
- Retro uses data that is already collected like in a pt med record or alberta health services
- Define a cohort back in the past to see whether there are ppl exposed to a certain med and see if they got an outcome of interest
Cohort Studies
When are they used?
• Used when it is not reasonable to expose people to something
• Can be used to answer questions about causation (therapy)
• Good for answering questions about prognosis
- harm questions, can be therapy tho not highest level of evidence
Inception cohort: group od people with same in disease, stage, early in disease
Follow them in time
Cohort Studies
what are they
• Two or more groups of people are selected
• Basis for selection is EXPOSURE to a particular agent
– (e.g., Occupational exposure, environmental toxin)
• Groups are followed-up to see how many develop the outcome of interest
Key Points
• Hallmark is the comparison of EXPOSED to NON-EXPOSED
• No intervention – Only observe patients during course of study
Case-Control Study
design
Associations b/w exposure and outcome rather than starting with exposure status
Outcome determined first; recall exposure
study group - identify cases or controls
determine if they were exposed or not exposed previously
Case-Control Study
When are they used?
- Generally concerned with establishing the causes of disease
- Usually the only option for studying rare conditions
key pts
• People are selected on the basis of outcome/disease NOT exposure
• Cannot demonstrate causality, just associations
Case-Control Study
what are they?
• Type of retrospective observational study
• Identify a group of people with a disease of interest (cases) and
compare them to a similar group of people without the disease
(controls)
• Data are then collected on past exposure to a possible causal agent
for the disease
Cross - Sectional Survey
what are they?
– A representative sample of participants is recruited
– They are interviewed, examined or otherwise studied
– Defining feature: data collection at a single time point
• e.g. What do first year students think about PHARM 324?
• e.g. What is the prevalence of Peripheral Arterial Disease in adults over
the age of 50 in Alberta?1
Cross - Sectional Survey
advantages, disadvantages
simple, cheap
no causation
Hopw prevalent is a disease
Sample survey: we get some people
Census survey: everyone in populaiton
Cannot look at causation
Case Reports & Case Series
define
Case report – description of an individual patient
• Case series – descriptions of a series of patients
• No comparison group
• AKA descriptive studies
• Used to report rare/unusual occurrences
- Commonly reported in literature and can describe cases or unsual expeirence/scnearios
- Landmark future research
- Case series: multiple cases in a group of people
- Limited in use for practice
name 3 systematic errors
selection bias, info bias, confounding
Qualitative Research
when is it useful?
• Sometimes the goal is in-depth understanding of human behavior rather than numerical outcomes
• These types of questions are better suited to qualitative investigation
– E.g., Interviews, focus groups, etc.
Research Question:
• Why do continuing smokers not access Stop Smoking Services offered by clinicians?
Biased RCT vs. Good Quality Cohort Study
“Nuanced” Hierarchy of Evidence - The GRADE Approach
what can decrease confidence in evidence (5)
see picture for more
RCTs have high confience evidence
Observational studies have low confience because of confounding and other facotrs
risk of bias: If a RCT is biased because of factors, this lowers confidence
Inconsistency: conflicting evidence, lower confidence
Indirectness: If evidence is not directly applicable to practice, lower confidence. If trial participant results are different that target population
Imprecision: Results not precise and large range of data, less confident
Publication bias: showing research that is shows benefit for a med
Observational studies with low confidence can be increased with factors
how can we differentiate cohort, cross sectional, and case control studies?
looking at when the study outcomes were determined in relation to when the exposure of
interest occurred
