Aubrey- Epidemiology Test 2 Flashcards
Order of Research Evidence Pyramid
Most evidence to least
- Not important ones
- Randomized, DB, controlled trials
- Cohort
- Case Control
- Cross sectional
- Not important ones
Study design selection is based on (6)
Perspective of research question (hypothesis)
Ability/ Desire to force group allocation (randomization)
Ethics of methodology
Efficiency & Practicality (time/resources)
Costs
Validity of acquired info (internal/external)
External validity and Internal validity
Ext: how well can i take the findings from the studies and apply them
Int: Is the study group design and the methods used valid?
Null Hypothesis (Ho)
Researchers either reject or accept this based on results
States that there will be NO true difference between groups being compared
Statistical perspectives that can be taken by the researcher (3)
Superiority
Noninferiority
Equivaency
Alternative Hypothesis (H1)
A research perspective which states there WILL BE a (true) difference between the groups being compared
Type I error
false positive
pregnant male
Type II error
False negative
Ho True p>0.5
Correct
Ho True p
Type I error
H1 True p>0.5
Type II error
H1 True p
Correct
Two types of study designs
Observational and Interventional
Observational studies
Natural
Elements occurring naturally or that individual freely picks
Can observational studies prove causation?
No
Interventional studies
Experimental
Researcher force group allocation
Randomization process
Types of observational studies (5), list in increasing strength of evidence
Cases reports/series, ecological, cross-sectional, case-control, cohort
What observational studies are analytical
Cross sectional
Case control
Cohort
What intervential studies are analytical
ALL
Study population
The final group of individuals selected for a study
Are human studies observational or intervential?
Both
Study Population Selection based on (4)
-Research Hypothesis/ Question
-Inclusion & Exclusion selection criteria (inventional) & Case and Control group OR Exposed and Non-Exposed group selection criteria (observational)
[Desired vs logical vs plausible]
-Ethics
-Equipoise
Equipoise
Genuine confidence that an intervention may be worthwhile (risk vs benefit) in order to use it in humans
4 Key principles of Bioethics
[Be able to demonstrate 4 principles]
- Autonomy
- Beneficence
- Justice
- Nonmaleficence
Autonomy (3)
[4 key principles]
Self-rule/ Self-determination
Participants must..
-Have full & complete understanding of the risks and benefits
(No misinformation, incomplete info, or ineffectively-conveyed info (language or education level))
-Decide for ones-self, without outside influences
-No coercion, reprisal, financial manipulation
Beneficence
To benefit or do good for, the patient (not society)
Has to be some benefit even if getting placebo (getting care from doctor, assessment)
Justice
Equal and fair treatment regardless of patient characteristics
Nonmaleficence (3)
Do no harm. Researchers must not..
- Withold info
- Provide false information
- Exhibit professional incompetence
Belmont Report (What is it and 3 guiding principles)
Issued by National commission for protection of human subjects of biomedical and behavior research, decides whether of not research conducted is ethical
1) Respect of persons
- Research should be voluntary, subjects autonomous
2) Beneficence
- Research risk are justified by benefits
3) Justice
- Risk and benefits are equally distrubted
Consent
Agreement to participate
-Based on being fully and completely informed
[mentally-capable and legal consenting age]
Assent
Agreement to participate
- Based on being fully and completely informed, give by mentally capable individuals NOT able to give legal consent (children)
- Requires consent of parent or legal guardian
Ethical Conduct of Research, Who Determines
IRB (before)
OHRP (enforce during)
DSMB (during)
Interstitutional Review Board (IRB) levels (3)
all human subjects must be reviewed
- Full board: ALL interventional trials with more than minimal/ no risk to patients, all medication studies
- Expediated: minimal risk and/ no patient identifiers
- Exempt: no patient identifiers, low/no risk, de-identified dataset analysis, environmental studies, use existing data/ specimens
Office of Human Research Protections (OHRP)
agency that administers and enforces regulations
Data Safety & Monitoring Board (DSMB)
Semi-independent committee not involved with the conduct of the study but charged with reviewing study data as study progresses, to asses undue risk or benefit
DURING
Can stop study early for either overly-positive or overly negative findings
Best type of outcome?
Patient oriented
Outcomes (2)
Patient-Oriented
Individual vs. Combined
Surrogate: decreases blood pressure
Patient benefit: keeps patient out of hospital
Internal Validity
Assessment (measurements)
Scientifically-rigorous and standardized
Objective better than subjective assessments
Valid, accurate, and reproducible
Inventional Study Designs alternate names (5)
clinical trial, clinical study, experimental study, human study, investigational study
Key Difference in Investigational vs Observational
Investigator selects “interventions”
Allocates subjects into groups
More “rigorous” in ability to show cause and effect
Can investigational studies show causation?
Yes
Phases of Interventional studies trends (3)
Phase 1 to Phase 4
Population tends to increase
Tends to get longer in duration
Change in focus as you move up
Phases of Interventional studies (4)
Phase 1: Safety
Phase 2: Effectivness
Phase 3: Effectivness
Phase 4: Safety
Pre-Clinical Phase
Prior to human investigation
Bench and animal research
Phase 1
New drug/ device/ procedure Small N (20-80) Healthy volunteers Assess safety and toxicity, dosing, and pharmcokinetics Short duration (few days to week)
Phase 2
New drug/ device/ procedure, indication/ study population
Large N (100-300)
Utilize patients of interest with condition
Expands purpose of Phase 1 safety but assess efficacy
Short to medium duration (several months)
Narrower inclusion criteria
Major flaw is limitation, want any changes to be due to intervention
Phase 3
New drug etc, indication or study population
Even large N (1,000-3,000)
Patients with condition
Determine safety and primary purpose efficacy
Longer duration (months to years)
Superiority vs non-inferiorty vs equivalence formats
Phase FDA mandates before drug approval
[Drug with name could still be phase 3 if new condition or drug used in group not approved of when in phase 3 before]
Phase 4
Post marketing
Long-term effects (risks/benefits) in large population of diseased
Registries, Surveys
Drug will always have name if in Phase 4
Advantages of Interventional Trials (2)
Cause precedes effect (shows causation)
Only design used by FDA for approval
Disadvantage of Interventional (4)
Cost
Complexiy/Time
Ethical consideration
Generalizability (external validity)
Which Study design is described by the following?
Systemic review of all published literature of topic up to a specific point in time
Systemic Review
What is entailed with a meta-analysis?
It takes all studies and put them together so that it appears that there was one large, new study and build a consensus from the meta-analysis
Describe Case reports and Case series
Case reports and case series are just reports of one or more individuals that have had a unique experience; usually hypothesis generated
Inclusion & amp; exlusion criteria?
Desired vs logical vs plausible selection criteria
These absolutely impact generalizability
Control groups needs to be identical to the case group in every way except for the presence of disease
True or False, observational and interventional studies must be reviewed by an IRB prior to study initiation
True, observational and interventional studies use human subjects
Explanatory Interventional Studies
Explain impact talk about causation
A lot of restrictions
These studies not for clinical practice, no room for adjustment
So restrictive can’t even adjust dosage
Pragmatic interventional studies (3 key elements)
Allow flexibility to change dosage and add other drugs
Still Interventional studies
Key elements
-No placebo: practicality, no actually physician uses placebos
-Let in the regular people: people with multiple morbidities and drugs
-Use own clinical judgements on how they treat their patients
Limitations of Pragmatic studies (3)
1) Lose the researches control of how the physician prescribes that prescription or manages the patients
2) Lose all of the advantages of the explanatory method
[Lock, removal of confounding]
3) Loss of control and rigidity
Designs of Intervential Studies (4)
Simple
Factorial
Parallel
Cross-Over
Simple
Divides (randomizes subjects in +2 groups)
A single randomization process
Commonly used to test a single hypothesis
Any number of groups only one randomization
Factorial
Divides in greater than or equal to 2 groups
Further subdivides into each group
Randomizes at least twice
Takes more people but answers more questions
Tests usually at second randomization point
Components of Factorial Studies (5)
Improves efficiency for answering clinical questions
Increase study populations sample size
Increase complexity (which may be a barrier to recruitment)
Increase risk of drop outs
May restrict generalizability results
Parallel
Groups simultaneously and exclusively managed
No switching after initial randomization
Can be simple or factorial
Cross-Over (self control)
Groups serve as own control by crossing over from one intervention to another during study
Allows small N
Between & Within- Group comparison possible
-comparison of A vs B
-comparison of seiners on A and seiners on B
Forced switching, everybody switches
Washout Period
Period of time to remove first drug from system
Could be 1 day to 1 month
Can be at beginning or middle
Can be part of lead-in phase
Run in/ Lead in phase
All study subjects blindly given one or more placebos for initial therapy to determine “new” baseline elf disease (standardization)
- Can assess study protocol
- Can “wash out” existing medication
- Can determine among to placebo-effect
Disadvantages of Cross-Over design (6)
- Only suitable for long term conditions which are not curable or which treatment provides short-term relief
- Duration of study for each subject is longer
- Carry-over effects during cross over
- Treat-by-period interaction
- Small N
- Complexity in data analysis
Treat-by-period interaction is?
-Differences in effects of treatments during different time periods
Outcomes/Endpoints
Primary
Secondary/Teritary/ etc
Composite
Primary Outcome
Most important key outcomes
Main research question (hypothesis)
Can have multiple
Secondary/ Tertiary Outcomes
Lesser importance yet still valuable
Possible for future hypothesis question
Other related info generated from data
Composite
Combines multiple endpoints into a single outcome
Could be considered primary outcome, and if so, then secondary outcomes may be the individual outcome elements from composite
Measures of Association, how to count a person
One person can contribute toe ash of the three individual elements but only counted once in overall amount
Usually secondary outcomes are listed as the breakdown composite
Followed by non composite secondaries
Examples of Patient-Orientated Endpoints
Death Stroke or Myocardial infarction Hospitalization Preventing need for dialysis [Patient centered outcomes are more impactful]
Examples of Surrogate Markers
[Elements used in place of evaluating patient0oriented (direct endpoints] Blood pressure (risk of stroke) Cholesterol (risk of heart attack) Change in SCr (worsening renal function)
Sample Selection & Group allocation
Can be non-random or random
If random then randomly picked (not randomization)
Randomization
baseline characteristics
Purpose: to make groups as equal as possible
Based on known and unknown confounders
equality of groups is determined by p values
[Can be shown in table format, text statement, key of table]
Types of Randomization (3)
Simple
Blocked
Stratified
Simple Randomization
Equal probability for allocation to one of study groups
Blocked Randomization
Ensures balance within each interventional group
Ensure groups equal in size
Last person in each block is put where they need to belong to even out groups
Someone outside investigation must control this
Stratified randomization
ensures balance with known confounding variables
examples: gender, age, disease severity/duration, comorbidities
Can also pre-select levels to be balanced within each interfering factor (confounder)
Masking types (3)
Single Blind
Double Blind
Open-Label
Single Blind Masking
study subjects are not informed of intervention
Clinicans know
Only okay if objective study, researchers have no interaction with patients
Double Blind Masking
No one knows
Someoneoutside investigation controls this
Open-Label
Everyone knows
Some studies you can’t blind or doesn’t matter if patients know
Comparing injectable drug with inhaler
[would have to have double dummy-placebos]
What type of survey can be used to assess adequacy of blinding?
Post-hoc surverys
Forms of Blinding (3)
Placebo (Dummy)
Placebo-effect
Hawthorne effect
Placebo (Drummy)
Inert treatments made to look identical to active treatments
Double dummy if two placebos
Placebo-effect
improvement in conditions, power of suggestion
Hawthrone effect
desire of patients to “please” investigators by reporting positive results
Post-hoc sub-group analysis
[Not accepted unless prospectively planned]
Data dredging or fishing
Managing Drop-outs/Lost to follow up
Include them
Ignore them
Treat them “as treated”
Incude them Management
Intent to treat
Example: last known assessment carried forward
Impact:
-Preserves randomization process
-Preseves baseline and group balance at baseline which controls for known and unknown confounders
-Maintains statistical power (original size sample)
Ignore them Management
Include only compliant or completing subjects
Per-Protocol or Efficacy Analysis
-Compliance predefined (80-90%)
-Impact of Per-protocol
Biases estimates of effect (over-estimates)
[reduces generalizability since not including those that could only do 65%]
Treating them “as treated”
Ignores group assignments
Allows subject to switch groups and be evaluated in grips they moved to, ended in, or stayed in most
If they switch or primary care switches to other group by accident
Assessing Adherence (3)
Drug levels
Pill counts
Bottle counter-tops
Methods of improving adherence
- Frequent follow up visits/ communication
- Treatment alarms/ notification
- Medication blister packs or dosage containers
Case-Control studies allow the research to what?
Be a passive observer of natural events occurring in the individuals with the disease/condition of interest (cases) who are compared with people who do not have conditions (controls)
Commonlly generates an Odds Ratio as measure of association
The control group supplies information about what?
The expected baseline risk factor profile in the population from which the cases are drawn
Case-control studies have group assignments based on what?
Disease status
What is case-control useful?
When studying a rare disease or investigating an outbreak
Case-Control Study Designs Table
2x2 table
Know column totals
What we don’t know is of those A + C people which will end up in box A and which in box C
Compare odds of exposure in the disease compared to nondiseased
Reasons to select Case-Control design (5)
- Unable to “randomize” (unethical, Illegal, not feasible)
- Limited resources (Time, money, subjects)
- The disease of interest is rate
- Prospecive exposure data, derived from prospective Cohort study is difficult/expensive to obtain and/or very time inappropriate
Case controls studies are prospective or retrospective?
Retrospective (always)
Retrospective study
Know outcome
Strengths of Case-Control Studies (7)
- Good for assessing multiple exposures of one outcome
- Useful when disease are rare
- Useful in calculating odds and Ors
- Less expensive than interventional trials and prospective cohorts
- Useful when ethical issues limit interventional studies
- Useful when dynamic populations [Dont have to worry about loss to follow up]
- Useful when disease has a long induction/latent period
Weakness of Case-Contrl Studies (4)
Limited by amount of data
Worry about misclassifications
Worry about have enough data
Worry about if disease has changed or exposure has changed
The way controls are selected for is what?
Major determinant in whether any conclusion is valid
Internal validity
Type of study that is built from investigators taking people with disease and finding people without disease
Case-control study
Case-Crossover Design
Subjects are their own controls during the other times they don’t have the acute change in risk
Levels of risk changes over the year
What study design is able to adequately attempt to address the issue of “temporality”
Case-Crossover Design
Nested Case Control studies
These are case-control studies conducted after, or out of, a prospective COHORT study
Subjects in cohort study ultimately develop disease are defined as cases
These defined cases used in a new (different) study design
Used to evaluate other exposures
Cohort set up on what and hunts for what?
Set up on exposure and hunts for outcome
Sampling of controls is used for?
Nested Case-control studies or when “sampling” necessary from Cohort
Three types of sampling of controls
Survivor sampling
Base sampling
Risk-set sampling
Survivor sampling
Sample of non-disease individuals (survivors) at end of study period
Base sampling
Sample of non-disease individuals at start of study
Risk-Set sampling
Sample of non-disease individuals during study period at same time when Case was diagnosed
Matching Schemes
Individual matching
Group matching
Don’t match anything that might be a risk factor
Individual matching
Matches individuals based on specific patent based characteristics
Used when have unique and important characteristics
Group matching
Proportion of cases and proportion of controls with identical characteristics are matched
41% cases are males so 41% of controls are males
In Group matching does case or control need to be selected first?
Case
Cohorts are stronger than case controls only applies to what?
Prospective NOT RETRO
Cohort and caes controls are equal in terms of retrospective
Cohort Study allow researcher to?
To be a passive observe of natural events occurring in naturally-exposed and unexposed groups
Useful when studying rare exposure
Commonly generates Risk Ratio (RR) as measure of association
Group allocation of Cohort studies is based on what?
Exposure status OR group membership (something in common)
Cohort studies also termed what?
Incidence studies/ Follow up studies/ Longituidinal studies
Cohort Study Designs Table
2x2 Table
Know who were exposed or not, what we don’t know is if they got the disease or not
Reasons to select Cohort design (4)
- Unable to “randomize”
- Limited resources (more so for prospective)
- The exposure of interest is rare
- More interested in incidence rates/ predictors of risk for outcome
Cohort Study Designs can be directed in what direction?
Prospective, Retrospective, Ambidirectional fashion
Group assignment still based on exposure
Prospective Cohort Studies
Exposure group is selected on the basis of past or current exposure and both groups (exposure and non-exposure) followed into future to assess for outcomes and then compare
- Takes longer
- Still group allocated based on exposure
- It is possible to lose people
- Incidence study cause can look at newly diseased
Retrospective Cohort Studies
At the start of the study, both the exposure and the outcome of interest have occurred
Retrospectively start at time of exposure and follow forward to the point of outcome occurrence in the present
Exposure still have to occur before outcome, group allocation based on exposure
Know outcomes when starting group but ignore them
Downside: at mercy of data available, accuracy and detail
Ambidirectional Cohort Studies
Uses retrospective design to assess past differences but adds all data collect on additional outcomes prospectively from start of study
Looking for outcomes in past and future
Only am bidirectional study
Cohort is what? Types?
A group with something in common
Birth Cohort: individuals born in same region
Inception Cohort: individuals coe to same point (work)
Cohort sizes (3)
Fixed
Closed
Open (dynamic)
Fixed cohort
can’t gain members but can have loss to follow up
911 firefighters
Close cohort
No gain or loss
Walled off room with no doors
Short term
Open (dynamic) cohort
has new additions and some loss
Framingham heart study, babies added, elderly die
Unexposed group can come from 3 sources
Internal (best): same cohort
General population: randomly selected
Comparison: least acceptable
Comparison cohort source
Least acceptable
Simple attempt to match groups as close as possible on numerous personal characteristics
(can’t control for other potentially harmful exposures)
Find them where you can, make sure not exposed
Strengths of Cohorts (7)
- Good for assessing multiple outcomes of one exposure
- Useful when exposure are rare
- Useful in calculating RISK and RR
- Less expensive than interventional
- Good when ethical issues limit use of interventional
- Good for long induction/latent periods (Retrospective)
- Able to present temporality (prospective)
Advantages of Prospective Cohort (5)
-Can obtain greater amount of study info
[more control over specific data collection, get more data, can educate patients on what to avoid]
-Follow-up/ Tracking of patients may be easier
-Better at giving answer to “temporality”
-May look at multiple outcomes for one exposure
-Calculate incidence and incidence rates
Disadvantage of Prospective Cohort (4)
- Time, expense, lost to follow up
- Not efficient for rare disease
- Not suited for long induction
- Exposure may change over time
Advantages of Retrospective Cohort (4)
- Best for long induction/latency conditions
- Able to study rare exposures
- Useful if the data already exists
- Saves time and money compared to prospective
Disadvantages of Retrospective Cohort (4)
- Requires access to charts, database
- “Information” may not factor in or control for other exposures
- Patients may not be available for interview or contact
- Exposure (or amount) may have changed over time
Issues affecting outcome occurrence in groups (3)
Levels of exposure (stratify) Induction period (interval between exposure and onset) Latency period (interval between onset and clinical diagnosis)
Key biases in Cohort Studies
Healthy work effect
Selection bias
Cross-sectional studies examines?
relationships of health/ disease to other variables of interest at same time
aka PREVALENCE
Called cross sectional because information gathered represents what is occurred at a point in time or time frame a-cross a large population
SNAPSHOT in time
Any study with word NATIONAL in it
Cross sectional studies
Focueses simultaneously on disease & population characteristics, including exposure, health status, health care, utilization etc
- seeks associations
- generates and tests hypothesis
- by repetition in different time periods, measure trends
Cross section studies studies what 3 things
Person, Place, TIme
Cross sectional advantages
- Fairly quick and easy to perform for researcher using data (data already collected)
- Useful for determining prevalence and risk factors across populations
- useful for measuring current health status & planning health services
- Useful for evaluation differences in subgroups within populations
- Require IRB review but low level expediated
Disadvantages of Cross sectional
- Prevalent cases may represent survivors (bias)
- Difficult to study disease of low freq
- Problems determing temporal relationship of a presumed cause & effect (exposure and disease histories taken at same time)
2 Cross sectional approaches
- Collect data on each member of population
2. Take same of pop and draw inferences
NHANES
Assess the health and nutritional status of adults & children
Combines interview and physical examinations
Interview includes demographic, socioecominic, diet
Oversamples greater than 60, african america/ hispancis
NHIS
CDC interview, personal household contacts, non impatient, dr. offices
Not physical screening but data is direct (not from chart)
Health of the civilian, non-institutionalized population
NAMCS
reliable information about the provision and use of ambulatory medical care services in US
Based on sample of visits to non-federal, office based physicians
Dr. offices, free clinics, walk in and out, no inpatient
Info from charts
NHCS
combined study designed to describe national patterns of healthcare delivery in non-federal hospital based settings
-discharges of impatient departments, visits to ER, outpatient department, ambulatory surgery centers
intergrates 3 previous studies (NHDS, NHSMCS, DAWN)
BRFSS
A state based telephone health survey that collects info on health risk behaviors, preventative health practices, and health care access
Monthly collection
Sensitivity
How well a test can detect presence of disease when in fact disease is present -positivity of test -accuracy in those with disease Sens= YP/ (YP+FN) Sens= YP/ (all diseased) Sens=A/ (A+C)
Specificity
How well a test can detect absence of disease in those that disease is absent
Spec= NN/ (NN+NP)
Spec= NN/ all non disease
Spec= D/ (B+D)
Positive Predictive Value (PPV)
how accurately a positive test predicts the presence of disease
PPV=YP/ (YP+NP)
PPV= YP/ (total positive test)
PPV= A/ (A+B)
Negative Predictive Value (NPV)
how accurately a negative test predicts the absence of disease
NPV=NN/ (NN+YN)
NPV= NN/ (all negative tests)
NPV= D/ (B+D)
Diagnostic Accuracy (DA) or Diagnostic Precision (DP)
proportion of time that a patient is correctly identified as either having disease or not having disease with positive or negative test
DA/DP= (YP+YN)/ (All patients)
Likeihood Ratio’s (LR)
ratio of the proportion of a given test result (pos or neg) for a person with the disease/ propbabiliy of the same result (pos or neg) for a person without the disease
Likeihood Ratio Positive (LR+)
Probabily of a positive test in the presence of disease/ probability of a positive test in the absence of disease
Sensitivity/ (1-specificity)
[(A/A+C)/(B/(B+D)]
Likeihood Ratio Negative (LR-)
Probability of a negative test in the presence of disease/ probability of a negative test in the absence of disease
1-sensitivity/ specificity
[(C/(A+C)/(D/D+B)]
LR bottomline
LR+ >10 to demo most bene
LR-
For screening tests with numerical values we use what?
ROCs receiver operator curves
a more efficient way to show a relationship between sensitivity and specificity for test with numerical (continuous) outcomes
Medical Screenings (2)
Validity
Reliability
Validity
ability to accurately discern between those that do and those that do not have the disease
Telling the truth
Reliability
Ability of a test to give the same result on repeated uses
Analogous to reproducibility/ consistency
A valid test is always reliable, but a reliable test is not always valid
True or false, selection bias is less of, or not, a significant issue during case-crossover study designs
True
When is inception cohort studies useful?
For single-group non-comparisons for incidence rate determination
What is the effect of loss to follow up?
Lower sample size
Increase risk of type 2 error
Study populations may not be equal between groups
True of false, most cross sectional studies are surveys of or databases related to different aspects of US populations
True
True of false, probability samples are NOT the most common type of sampling scheme
False
When are probability samples most even?
When you use a multiprobabilty, multistage selection
List 2 categories of sampling schemes of cross sectional studies
Probability samples and systemic (convience) samples
Systemic samples of cross sectional sampling schemes
Decide on what fraction of population is to be sampled and how they will be sampled
2 approaches to collection of new info in cross sectional studies
Questionaire/ Surveys
Physical assessments
What should be considered when determining an appropriate sample size?
Take into consideration the number of drop outs
What is the danger of the selection process?
Misclassification
T/F: Most observation studies designs are NOT able to prove causation
True
What type of study design is described below? Researchers observe subject elements occurring naturally or selected by individuals (natually or freely)
observational
What type of study is commonly era and the only type to prove causation
Interventional studies
Case-control studies commonly generate the ____ ____ as measure of association
Odds ratio (OR)
Cohort studies commonly generate the ____ _____ as measure of association
Risk Ratio (RR)
Cross-sectional studies seek ___, not causation
Assocations
A highly sensitive test has:
a low false negative rate
A high specificity test has:
a low false positive rate
