1
Q
What are the 2 Main Types of Errors?
A
- Random Errors (Chance)
- errors or divergence of observation from truth due to chance alone
- no trend, but there is departure nonetheless
- 2-tailed presentation of errors
- can never be completely eliminated, but can be estimated/accounted for by statistics
- Systematic Errors (Bias)
- errors or divergence of observation from truth due to system or procedure utilized
- a trend of departure
- 1-tailed presentation of errors (usually)
- can be avoided, eliminated, minimized, and recognized
- Selection
- Measurement
- Confounding
Note:
- “Random” refers to the PROCESS, not the RESULT
2
Q
True or False:
Effects from Chance (random) and Bias (systematic) are isolated
A
False!
Effects from Chance (random) and Bias (systematic) are cumulative
3
Q
What are the 3 main types of Systematic Errors (Bias)?
A
Basics about bias:
- Occurs at any stage or part of process
- A trend in the collection, analysis, interpretation, publication, or review of data
- Force or tendency to produce results that depart systematically from true values
- Lead to conclusions systematically different from the truth
- Usually 1-tailed or unidirectional
- (i.e. trend of over or under estimation)
- Selection bias
- groups differ in outcome determinants other than 1 being studied
- Measurement bias
- methods of measurement not the same between or among groups
- Confounding bias
- 2 factors are associated (packaged, mated)
- the effect of 1 distorts or is interpreted as the effect of the other
4
Q
What are 3 Ways to help Alleviate / Mitigate Measurement Bias?
A
- Masking/Binding
- Ensure persons recording outcomes/events unaware of participant group
- Standard Definition of Event
- Use standards, criteria, definitions for what will classify as the outcome
- Equitable Effort
- Look for outcomes or events equally for each participant in every group
5
Q
What are 3 Ways to help Alleviate / Mitigate Selection & Confounding Bias?
A
- Randomization
- Ensure equal distribution of confounding variables in each study group
- Restriction
- Restrict study or case admission criteria to prohibit variation of known or suspected confounder in study groups
- Matching
- According to known or suspected confounders to ensure equal distributions of confounders in each study group
6
Q
Association vs Causation
A
7
Q
What are the 3 Primary Objectives of Descriptive Epidemiology?
A
- Evaluate trends
- monitoring of known disease
- identification of emerging problems
- Plan, provide, evaluate health services
- data for resource allocation
- Identify problems or areas for further study
- fruitful areas for investigation
8
Q
What are the 3 Subtypes of Descriptive Studies?
A
- Case Report
- unusual case by clinician
- astute observations
- foster additional valence, questions
- Case Series
- larger number of similar cases
- summary of characteristics from several settings
- case definition, typical features
- Cross-Sectional Studies
- surveys of population
- estimate disease prevalence or distribution of exposures
- snapshot at one point in time
9
Q
What are 5 reasons to use Descriptive Studies?
A
- Generate etiologic hypotheses
- “what is causing this?…”
- Estimate magnitude of problem
- prevalence in a population
- Snapshot in time
- Exposure and Disease data collected simultaneously
- Rapid, Inexpensive
Note:
- Cannot specifically determine cause and effect; only possible associations
10
Q
Descriptive vs. Analytical
A
Descriptive
- Asking Questions
- amount, distrib of disease
- Identify or show problem
- Ecological
- Cross-sectional
- Hypothesis generating
- Studies:
- Case reports
- Case series
- Cross-sectional
Analytical
- Getting Answers
- Determinants of disease
- Reason why high or low frequency of disease in pop.
- Identify association or causation
- Studies
- Clinical Trials
- Case-Control
- Cohort
11
Q
Secular Trend vs. Cohort Effect
A
Secular Trend
- changes in disease freq over time
- across the board
- ‘vertical’ and ‘horizontal’ rates
Cohort Effect
- group or block of persons that share experience or exposure
- (age of birth)
- patterns of ‘cells’ or cohorts as move through time
- ‘diagonal rates’
12
Q
What are the pros/cons for Ecologic Analytic Studies?
A
Ecologic study
- unit of analysis = a group
- city, census tract, etc
Pros/Uses:
- generate etiologic hypothesis
- test specific hypothesis
- suggest cause & effect
Cons/Limits
- Ecological fallacy
- does not represent exposure-dx relation at individual level
- not have # of exposed cases
- assoc seen at aggregate level
- not necessarily true at individ level
13
Q
What are the pros/cons for Case-Control Analytic Studies?
A
Case-Control Study
- Retrospective
- Start with cases, then pick controls
Uses/Pros
- good for rare conditions/diseases
- vaccine effect, treatment efficacy, screening, outbreaks
- can suggest causation
Limits/Cons
- not ideal for rare exposures
- unknown exposure-disease time relationship
- cannot provide direct evaluation of risk (OR = Odds Ratio)
14
Q
What are the pros/cons for Cohort Analytic Studies?
A
Cohort
- Prospective, retrospective, or historical prospective
- Start with population group or subgroup
- Follow that cohort through time
Uses/Pros
- good for rare exposures
- can select cohort based on known exposures (i.e. occupational)
- can suggest causation
- direct evaluation of risk (RR = Relative Risk)
Limits/Cons
- not ideal for rare conditions/diseases
- must wait for outcomes to occur
- significant effort, time, costs
15
Q
What are the 4 Important Questions to ask in Research?
A
- What is the research question / hypothesis?
- What are the eligibility criteria?
- How were data collected?
- How was/were the outcome/s measured?
