DESIGN AND BIAS Flashcards
Study Design
specific plan/protocol for conducting a study
can be quantitative or qualitative
Descriptive study
a type of observational study
observes distribution of disease, identifies associating characteristics
When, where, who
case reports, case series, cross sectional surveys
Case report
Detailed presentation of single case; reports new/unique findings
Case series
summarizes experience of group of patients with similar diagnosis
single/multiple sources
reports on new/unique condition
Helpful for hypothesis generation, informative for very rare diseases with few established risk factors
Not helpful for cause and effect relationship, or assessing frequency
Quantitative study design
quantifies problem
generates numbers to turn into statistics
makes generalizations from sample to pop
Qualitative study design
exploratory
gains insight into problem; seeks to understand underlying reasons and opinions
Analytical study
type of observational study inclusive of ecological study, cross-sectional study, case-control study, cohort study
Ecological study
looks for associations between occurrence of disease and exposure to known or suspected cases
unit: group of people/population
Cross sectional study
Snapshot of data available at a certain point in time eg point prevalence
Surveys exposures and disease status of population at single point
Observational
Finds and describes associations and prevalence
Assesses burden of disease in pop, need for health services, trends of prevalence/severity over time, compares prevalence in diff populations
Present time frame
Case control study
starts with cases, match with controls (no disease), looks back to assess exposures
Retrospective (what led to this? what has happened? contributing factors?)
Present to past
Measures Odds Ration (OR)
Cohort
group of people sharing common experience/characteristic within defined time period (age, occupation, exposure)
Measures Relative Risk (RR)
Start with disease free, classify as exposed/unexposed, record outcomes in both groups, compare outcomes using RR
Prospective (what is happening?)
Incidence/Follow-up study
Experimental Study
Intervention study (deliberate action, manipulation)
Similar to cohort, but exposure is under direct control of investigator
Seeks to know effect of intervention (what is happening after intervention)
Participation requires meeting entry criteria
Provides scientific/statistical significance; scientific proof of etiological factor
Measures effectiveness and efficacy of health services for prevention, control, treatment
Animal/Human study
Randomized controlled (intervention vs placebo; groups not directly chosen by investigator) / Non randomized controlled
Randomization
Participants allocated into 2 groups: study vs control
Study group is given intervention
Control group is given standard/no intervention
Clinical Trial/RCT Phase I
tests new therapy for first time in small group
checks safety in humans, right dose, side effects, toxicity
Clinical Trial/RCT Phase II
tests how well new therapy treats disease
tests safety
more individuals
Clinical Trial/RCT Phase III
Larger groups
Better idea of side effects
Compares to existing therapies
Clinical Trial/RCT Phase IV
Approved by US FDA
Released to gen pop
Effects monitored and evaluated over a long period of time in large specific groups
Adverse effects seen on larger scale
Rare phenotypes exposed
Blinding
neither researchers’ nor participants’ expectations can influence study results
Meant to ameliorate placebo effect
Levels playing field
Single vs Double
Single blinded
Only participants have no idea what group they are in
Double blinded
No one knows which participants are in which group
Gold standard for testing new inventions
Placebo effect
physical/mental health appears to improve with dummy treatment (appears real, no therapeutic effect), eg sugar pill, saline injection, fake surgical procedure
Triggers person’s belief in benefit from treatment and their expectation of feeling better
Hawthorne Effect
Subjects of experimental study attempt to change/improve behaviour due to knowledge of being evaluated
Apparently unavoidable with human subjects
(Reality is up for debate)
Systematic review
Review conducted according to clearly stated scientific research methods
Designed to minimize biases and errors
Meta Analysis
Statistical analysis of collection of studies
Focuses on contrasting and comparing results
Anticipates identifying consistent patterns and sources of disagreements
Based on but not necessarily included in systematic reviews
Qualitative research
provides insight into problem
exploring and understanding meaning of the world and why things are the way they are
Error
difference between observed results and reality
random vs systematic
Random Error
affects precision of instruments (controlled by statistical tools–optimal sample size, hypothesis tests, confidence intervals)
directly related to study sample and size (decreased influence when study sample/pop size increases)
Not constant with each measurement
Does not always occur in same direction
Systemic Error/Bias
Affects internal validity of study, thus external validity of results obtained
Occurs in every measurement, in same direction (greater/smaller)
Classified according to selection, classification, confounding (controlled for in analysis phase)
Sources of Bias
Selection
Exclusion
Inclusion
Information/Wrong Classification
Recall
Observer
Confounder/Confounding Variable
Types of Selection Bias
Survivorship
Non-response
Under coverage
Neyman
