Research & Ethics Flashcards
(151 cards)
1
Q
What is evidence-based medicine?
A
- Clinical judgment
- Best available research/evidence
- Patients’ values and preferences
“use of data from research on populations to inform decisions about individuals”
2
Q
What is research?
A
focused systematic inquiry aimed at generating new knowledge
3
Q
5 A’s of EBM
A
Ask Acquire Appraise Apply Assess
4
Q
What type of study is the first step in answering a clinical question?
A
observational
5
Q
Reasons to choose an observational study?
A
Ethical issues, studying a rare disease, not practical or too costly to do an RCT, long time to develop the condition
6
Q
Descriptive studies
A
Subtype of observational studies
There is no comparison group and you cannot assess casual associations
Answers who, what, why, when, and where
7
Q
Three types of observational studies that include a comparison group?
A
- cross-sectional
- case control
- cohort
8
Q
What is a cross-sectional study?
A
A snapshot at one point in time; data on exposure and outcome are collected at the same time period
** cannot assess casual associations
9
Q
What is a case-control study?
A
A study where you start with an outcome, find cases with that outcome, and then find controls without that outcome. “Think backwards” and do a retrospective investigation
*** Can measure an odds ratio (OR)
10
Q
What are the advantages of case-control studies?
A
Retrospectively identify risk factors, efficient for rare diseases and long latency periods, quick & inexpensive, allows you to look for multiple potential risk factors or exposures
11
Q
What are the disadvantages of case-control studies?
A
Selection bias (hard to find true controls), recall bias, interview bias, inability to calculate incidence, relative risk & hazard ratio
12
Q
What is an odds ratio (OR)?
A
A comparison of the odds of exposure among cases divided by the odds of exposure among controls
Used as a measure of association for case-control studies
13
Q
What is a cohort study?
A
Start with an exposure and then follow over time to look for an outcome to occur or develop
(aka follow-up, incidence, or longitudinal studies)
14
Q
What are the three types of cohort studies?
A
- prospective - forward in time from exposure to outcome
- Retrospective - define the outcome and then go back in time to exposure.. then follow foward
- Ambidirectional - prospective and retrospective combined
15
Q
What are the advantages of cohort studies?
A
determine incidence, study natural history of disease, efficient for rare exposure & multiple outcomes from single exposure, minimize recall bias, allows calculation of more measures of association
16
Q
What are the disadvantages of cohort studies?
A
expensive, time consuming, attrition bias, inefficient for rare disease & long development time, selection bias
17
Q
What types of intervention may a researcher use in an experimental study?
A
drug, device, procedure, preventive strategy, or diagnostic study
18
Q
Why are non-randomized experimental studies performed?
A
Often to determine feasibility for enrollment or acceptability of interventions
19
Q
What are the advantages of RCTs?
A
distribution of variables that might influence the study are randomly distributed (i.e. confounding variables), thus a difference in outcome would be attributed to the treatment/intervention
20
Q
Non-validated practices
A
novel or innovative practices that have not been subjected to rigorous study
should be subjected to research in order to validate them before widespread dissemination
21
Q
What is the goal of clinical practice?
A
to provide the best care for the individual patient
22
Q
What is the goal of clinical research?
A
to produce generalizable knowledge to benefit future patients
23
Q
Does clinical research use means that are acceptable in clinical practice?
A
No. Clinical intervention is based on best interests of the patient & patient choice, do not involve placebos, and all parties are aware of the selected intervention
24
Q
What is a background question?
A
One that asks for general knowledge about the disorder, test, treatment, etc
25
What are the two components of a background question?
Root verb + condition
Root + verb = who, what, where, when, why, how + verb
Condition = disease or outcome
Examples: "What causes high blood pressure?," "Who is at risk for high blood pressure?"
26
What is a foreground question?
One that asks about specific knowledge to inform clinical decisions and actions
4 main components: PICO
27
What does PICO stand for?
```
P = Patient, problem, or population
I = Intervention, exposure, prognostic factor or maneuver
C = Comparison (if relevant)
O = Outcome (including time frame, if relevant)
```
28
What are the four main types of foreground questions?
PDT - E
1. Prognosis
2. Diagnosis
3. Therapy
4. Etiology/Harm
29
Therapy questions
Used when we want to find the best treatment (intervention) for a specific disease or condition
(can include preventative treatments)
30
Diagnosis questions
Used when we want to know how accurate a diagnostic test is in a specific group of patients
Compare new test/procedure to a known gold standard (reference standard)
31
Prognosis questions
Used when we want to know which factors change the course (prognosis) of a given disease or condition
Estimates progression, likelihood, and complications
32
Etiology/harm questions
Used when we want to know about negative impact from an exposure or intervention
(public health issues or exposure to environmental agent increases risk of bad outcome)
33
What type of evidence will help you answer therapy and diagnosis questions?
RCTS and clinical trials
34
What type of evidence will help you answer prognosis and etiology/harm questions?
Cohort studies
| If no cohort then look for case-control... no case-control then cross-sectional, case-series or report
35
Is it legal for a physician to prescribe an FDA-approved drug for a use other than the non-approved use?
Yes
36
Can drug, device, and biologic manufacturers market items for unapproved (off-label) uses?
Nope
Unless its in reference texts (full article or as a complete chapter), clinical practice guidelines, or provision of journal articles
37
What are the four primary phases of FDA testing?
1. Safety and dosage
2. Safety and effectiveness in small numbers
3. Safety and effectiveness with statistical power
4. Post-market data collection to assess adverse effects in larger general population
38
FDA Phase 1
20 to 100 healthy volunteers (unless its testing something like chemo, then affected people) are used to investigate the safe dosage range
About 70% of drugs move on
39
FDA Phase 2
Up to several hundred people with the disease/condition
About 33% of drugs move on
40
FDA Phase 3
300 - 3000 affected volunteers are monitored for side effects, effectiveness, and safety
About 25-30% of drugs move on
41
FDA Phase 4
Post-marketing studies conducted after a treatment is approved for use by the FDA
Conducted in several thousand volunteers with the disease/condition
42
How long does a patent grant exclusive use?
20 years
43
What do generic drug competitors have to demonstrate to the FDA?
Bioequivalence to the patented drug
| generic releases active ingredient at the same speed and in the same amounts as the original
44
What variability from the original, patented drug does the FDA require for generics?
5%
45
Who regulates OTC drug advertising?
FTC
| FDA is only Rx drug advertising
46
The FDA "true statement" advertising requirement is not met if...
advertising is false or misleading, does not present a fair balance, and fails to reveal material facts
47
What is the adequate provision requirement?
Manufacturers must make "adequate provision" for the consumer to get full labeling information in language they can comprehend
Four sources:
1. toll-free phone #
2. website
3. Referral to concurrent print ad or brochure
4. Referral to a healthcare practioner
48
What is a CONSORT diagram?
A standard way to report RCTs; accounts for what happens to all the subjects from beginning to the end of the trial
Flow diagram from enrollment to study conclusion
49
What kind of bias does randomization protect against?
Selection bias
Produces a balance comparison between groups if done properly
50
Simple randomization
Repeated fair coin flip
51
Permuted block randomization
Randomization is a sequence of blocks. Each block contains a pre-specified number of treatment allocations in random order
At the end of the trial there are equal numbers in each group
52
Stratified randomization
Achieves balance among groups in terms of subject's baseline characteristics (covariates)
Ex: stratified by treatment center, age, sex, disease stage
53
What is allocation concealment?
A method of preventing those enrolling participants from (unconsciously or otherwise) influencing which participants are assigned to a given treatment group
Done before the subject enrolls; always feasible
54
Why is allocation concealment important?
Inadequate or unclear allocation concealment can lead to overestimating the benefit of the treatment effect
55
Blinding
Process by which subjects, care-givers, data collectors, outcome adjudicators, etc are unaware of which treatment is given to an individual subject
Done after a subject enrolls; not always feasible
56
What kind of bias does blinding prevent?
1. Ascertainment bias - treating subjects differently based on treatment allocaiton
2. Observer bias - data collectors performing subjective assessments
3. Recall Bias - during follow-up
57
Intention-to-treat anaylsis
Statistical concept to address non-adherence & missing outcomes
Provides an unbiased, conservative estimate of treatment effect
58
Per protocol anaylsis
Excludes those not adherent to treatment, switched groups, missed measurement, violated protocol
59
Individual clinician equipoise (theoretical equipoise)
If a physician believes one arm of an RCT to be better than another, then it is unethical for that physician to randomize her patient in an RCT. Since her obligation is to recommend what she thinks is in the patient's best interest.
60
Community of physicians: clinical equipoise
When the relevant community of physicians (such as a field or specialty) is sufficiently uncertain about which or two or more interventions is best and an RCT will resolve the ambivalence
Absence of evidenced-based validation; indifference or uncertainty
61
Patient/subject equipoise
When patients are sufficiently uncertain as to which or two or more interventions is best that they are willing to be randomized
62
Kantian imperative
Always treat persons as ends in themselves, never as means to an end.
Denying best individual care for patients risks using them for the sake of future patients and society
63
Surrogate endpoints
Lab measurements, tumor size, physiological measurements
64
Clinically-meaningful endpoints
MI, functional status, disease-related quality of life, death
65
Data safety monitoring boards
Periodically look at interim results. If there is too high a rate of adverse events then they stop the trial
If the trial will dissolve clinical equipoise then they also stop the trial
66
Definitions for misconduct in research
fabrication, falsification or plagiarism in proposing, performing, reviewing research, or reporting research results
67
Fabrication
Making up data/results and recording or reporting them
68
Falsification
Manipulating research materials, equipment, processes, or changing or omitting data or results such that research is not accurately represented in the research record
69
"research misbehavior"
failing to confirm research results that your name is attached to
70
First author
The one who does the most work and the order goes from greatest contribution to the least
First author is usually the corresponding author for issues with journal editor, etc, and the one most responsible for accuracy and authenticity of content
71
Plagiarism
the appropriation of another person's ideas, processes, results or words without giving appropriate credit
"with the intention that the words be taken as the work of the deceiver"
72
Cryptomnesia
unconscious plagiarism; individuals previously exposed to others' ideas will often remember the idea, but not its source
73
Four fair use exceptions to copyright protection
1. Purpose and character of use: educational or commercial, spontaneous os repetitive, transformative or reproductive
2. Amount and substantiality of the portion used: small amount vs. larger; selection is not considered "heart of the matter"
3. Nature of the copyrighted work: technical or artistic; factual or imaginative; published or unpublished
4. Effect of the use on the market: avoiding payment of royalties, ready market for the original, alternative to students purchasing original work?
74
Open access publication
Access is free of charge and in some forms may be free of restrictions on various uses
Publisher will charge the author(s) to publish, but makes access free
75
Creative commons license
Authors may use this to give varying degrees of commercial or non-commercial use
Most forms require attribution, but one form does not
76
iThenticate
Allows you to scan research articles and sources to ensure that you have properly cited your research or grant proposal and do not have any plagiarism issues present
77
Deja vu
A database of extremely similar Medline citations; used to detect duplicate publications and potential plagiarism
78
Adaptive randomization
Begin with randomization, but once you have experience, you begin to randomize to the treatment that is working best because you believe equipoise is being resolved
79
Disadvantages of adaptive randomization
Must e only one outcome of interest, outcomes must be apparent in a short period of time requires more patients (& thus prolonging the study), may suffer from accrual bias by participants who want to be recruited into the trial later
80
FDA "gold standard"
Placebos!
Testing new therapies against a placebo rather than a proven effective similar therapy
Remember, the FDA's concern isn't the best option, its safety and effectiveness
81
Crossover design
mitigates issues with placebo controls
Switch from placebo to active control or vice-versa
Data to compare person's condition under two treatments
All subjects get both placebo and active intervention at different points
82
Absolute risk difference
Difference between two event rates.
Experimental Event Rate - Control Event Rate
Sometimes also called the "absolute difference"
83
Relative risk or risk ratio (RR)
EER/CER
84
Relative risk reduction (RRR)
| EER - CER | / CER
85
Number needed to treat (NNT)
The number of patients who need to be treated to prevent 1 additional bad outcome, calculated as 1/ARR, rounded up to the next highest whole number and accompanied by its 95% CI
86
Number needed to harm NHH)
The number of patients who, if they received the experimental treatment, would lead to 1 additional person being harmed compared with patients who receive the control treatment.
Calculated as 1/ARI (absolute risk difference)
87
How do we improve accuracy?
Minimize systematic error
88
Systematic error
deviations from "truth" that are not due to chance alone (i.e. improperly calibrated measuring device)
89
Precision
how close the measured values are to each other; consistency of repeated observations
90
Random error
refers to variability (random variation); corresponds to imprecision
91
Screening tests
Done in apparently healthy people to identify those at increased risk of disease/disorder
Can lead to subsequent diagnostic test/procedure or treatment
92
Perfect discrimination
All patients who have the disease would have a positive test and all patients who do not have the disease would have a negative test
93
Sensitivity
The proportion of persons with disease who have a positive test; also known as the true positive rate
The ability of a screening test to correctly identify all screened who truly have disease
94
What is the most important characteristic for a screening test?
Sensitivity because we want to rule out disease
95
Specificity
Is the proportion of persons without the disease who have a negative test result; true negative rate
Probability that a healthy, no disease, person will have a negative test
96
What is important when confirming a diagnosis?
Specificity
97
If a test has 90% sensitivity what does that tell us?
The test positively identifies 90 out of 100 people with the disease
98
If a test has 75% specificity what does that tell us?
The test will be negative in 75 out of 100 people without the disease
99
False positive rate (FPR)
1 - specificity
Proportion of persons without disease who have a positive test result
Ex: A specificity of 75% means the test is wrong, falsely positive, in 25%
100
False negative rate (FNR)
1 - sensitivity
Proportion of persons with disease who have a negative test result
Ex: A sensitivity of 90% means that the test is wrong, falsely negative, in 10%
101
Positive Predictive Value (PPV)
Disease present among those with positive test result
Probability that those with a positive test have the disease
102
Negative predictive value (NPV)
Disease absent among those with negative test result
Probability that those with a negative test don't have the disease
103
Do sensitivity and specificity change with changes in prevalence?
No, only predictive values change with changes in prevalence
Sensitivity and specificity are characteristics of the test; they are independent of the population subjected to the test
104
What happens to PPV and NPV as prevalence decreases?
PPV decreases, NPV increases
105
What decreases as prevalence decreases?
TP and FN
106
What increases as prevalence decreases?
FP and TN
107
Prevalence
of cases of disease/population at risk for the disease
108
Receiver Operating Characteristic (ROC) plot
Shows the trade-off between sensitivity and specificity
Calculate the sensitivity and specificity of every observed data value and then plot the sensitivity (true positive rate) against 1 - specificity (false positive rate)
109
Ideal area under the ROC curve (AUC)?
Greater than or equal to 0.7
110
Likelihood ratio
Provide a numerical measure of the effect of a result on probability
Incorporates both the sensitivity and specificity of the test to provide a direct estimate of how much a test result will change the odds of having a disease
111
LR for positive result (LR+)
Tells you how much the odds of disease increases when a test is positive
112
LR for negative result (LR-)
Tells you how much the odds of disease decreases when a test is negative
113
Likelihood ratios to rule out and rule in disease?
Rule out if LR- less than 0.1
Rule in if LR+ is greater than 10
114
Lead-time bias
Appears that survival is improved but this is really an overestimation of survival because the diagnosis was made earlier in the course of the disease; the time from disease onset to death is still the same
115
Length bias
Appears that survival is improved but it is due to more slowly growing tumors having more opportunity to be detected by screening than faster growing tumors
116
Over-diagnosis bias
Persons who screen positive and are truly disease free (false positive) will have a better long-term outcome
"pseudo-disease" gives the appearance of an effect screening program
117
Type 1 Error (alpha error)
Accepting false positive result as disease when there is not disease; also known as the false positive rate
Only possible in a positive study
Incorrectly reject the null hypothesis
Ex: Positive pregnancy test but no pregnancy
118
Type 2 Error (beta error)
Accepting false negative result as no disease when there is disease; also known as the false negative rate
Incorrectly accepting the null hypothesis
Only possible in a negative study
Ex: Negative pregnancy test but actually pregnant
119
When does equipoise exist (in terms of errors)?
When there is no difference/null hypothesis exists
120
Non-inferiority study
Not designed to determine superiority, but to determine that X drug was within the same treatment range as the other drug(s) being tested
(Part of the Dan Markington case)
121
Systematic reviews
Comprehensive method to summarize the evidence from primary studies
They may or may not include a meta-analysis
122
Meta-Analysis
Combines sample estimates of treatment effects to get a total overall estimate of population parameter to reduce large amount of information into something more manageable. Increases power and precision.
A significant effect may be seen overall that was not seen in any of the individual studies
Failure to include all studies biases the population parameter
123
Reporting bias
Collective term for several kinds of biases that occur when the reporting of study findings is driven by nature and direction of those results
Types: publication, selective outcome reporting, and selective analysis reporting
124
Funnel plot
Graph to assess for reporting bias, but you can't tell which one is present
X-axis is estimated treatment effect, Y-axis is standard error or reciprocal of the standard error
If subjectively assessed to be asymmetrical then concerns that the authors of the SR&MTA may not have included all possible studies. If all relevant trials that were conducted were included in MTA, then the funnel plot would be symmetrical in shape
Need to have at least 10 studies for funnel plot to be reliable
125
As you increase precision (increase the sample size) what happens to the standard error?
(in reference to funnel plots)
It decreases
Or if reciprocal of SE then it would increase
126
Egger's Test
Formal statistic to assess for asymmetry
Null hypothesis = symmetry
Alternative hypothesis = asymmetry
If p-value for Egger's is not significant (over 0.05) then there is no evidence to reject the null
127
Other reasons for asymmetry in funnel plots?
Poor quality studies, fraud, inadequate analysis, true heterogeneity, chance
128
Forest plot
graphic representation of MTA results
129
Advantages of meta analysis
Increase power & precision (combine & aggregate results)
130
Heterogeneity may arise from differences in what?
Populations, interventions, co-interventions, study designs, study quality, subgroup difference, random error in included studies
131
Heterogeneity vs. homogeneity: Cochran Q (chi-square)
Null hypothesis: Statistical homogeneity
Alternative hypothesis: Statistical heterogeneity
If p value is not significant (over 0.05) then we do not reject the null and conclude that there was minimal variation in included studies & that pooling was reasonable
If p value is significant (less than 0.05) then statistical heterogeneity is present and we conclude that the treatment effects of the included studies different substantially & we question whether the results should have been pooled
132
Heterogeneity vs. homogeneity: Higgins I^2 statistic
represents the percentage of variation between estimates due to heterogeneity
0% (homogeneity) to 100% (heterogeneity)
Most consider more than 50% statistical heterogeneity
133
Publication bias
Failure to include all relevant studies because they were not published (not accessible)
Positive results more likely to be published, English-language bias, cited by others, published in higher-impact journal, influenced by funding source or research group, etc
134
Selective outcome reporting
authors report only certain outcomes that they found to be the most interesting
135
Selective analysis reporting
authors present only the results from analyses with the most significant P value or with the largest treatment estimate
136
What kind of study will scatter at the bottom of a funnel plot?
Studies with smaller sample sizes because they have less precise estimated effects & will scatter more widely at the bottom of the plot
137
What estimate has a greater contribution in a forest plot?
Those with a more precise estimate. Have a narrower CI and a greater weight
% weight determined by the precision of the sample estimate
138
Random effects meta-analysis
Used if statistical heterogeneity is found. Results in wider confidence intervals than the fixed effects approach; less accurate overall estimate
139
Heterogeneity due to differences in
patient populations studied, interventions used, study design, study quality, random error, difference in subgroups
140
When can a meta analysis mislead?
When it is done outside of a systematic review
141
PRISMA
Preferred Reporting Items for Systematic Reviews and Meta-Analyses
142
Three prerequisites for obtaining informed consent
1. disclosure of information
2. comprehension of information
3. voluntariness (absence of coercion)
143
Are waivers and alterations of informed consent applicable to studies involving FDA-regulated products?
No
144
Three situations where waivers and alterations of informed consent is valid
1. No more than minimal risk
2. Emergency exemption
3.
145
What should an informed consent contain?
Explanation of purpose, duration of participation, description of procedures
146
Is risk is greater than minimal risk what must be provided?
Explanation of any compensation to be paid or explanation fo any medical care to be provided if injury occurs
147
Therapeutic misconception
Deep-seated default presumption that physician or other health care provider would not be offering this intervention if it were not a medical treatment
Issue is that while clinical equipoise may exist about which treatment is best overall, the individual patient may respond better to one treatment over another
148
What is the most effective strategy in improving participant understanding in the informed consent process?
Person-to-person interactions (i.e. extended discussions and possibly test or feedback)
149
Under what conditions can an IRB waive or alter the requirement of informed consent?
1. Research involves no more than minimal risk
2. Waiver or change will not adversely affect the rights and welfare of the participants
3. Research couldn't practically be carried out without a waiver or change
4. When possible participants will be provided additional pertinent information after participation
150
Under what conditions does the FDA permit an IRB to approve a clinical investigation without participants' informed consent?
Emergency exemption for investigational drug, biologic, or device
Exception for planned emergency research
Informed consent for studies using leftover human specimens that re not individually identifiable
Circumstances when the US president may waive informed consent for military personnel for investigational product of the armed forces
151
Waiver of documentation of consent?
Not for FED regulated research
Potential participants or parents of children who are potential children are presented the informed consent but do not sign because the principal risk would be potential harm resulting from a breach of confidentiality
key: research presents no more than minimal risk of harm to participants
