Statistical Analysis Flashcards
Steps for Statistical Analysis
- Pick an analytical model
- Control for confounding
- Look at subgroups for differences
- Test for effect modification
- Assess assumptions (Sensitivity analysis)
- Propensity scores
Prevalence
- Definition
Case Control Study
- Whether or not subject has a disease
- Does not care about time
Prevalence
- Kinds
Point Prevalence (Cross Sectional)
- Who had a disease at a specific point in time
Period Prevalence
- Who had a disease over a specific measure of time
Prevalence
- Formula
Total number of people with disease / Total number of people in population
Incidence
- Definition
Cohort Study
- Proportion of people in a group who have developed a disease during a period of time
Incidence
- Kinds of
Incidence Proportion (Must specify follow up duration)
- Proportion of at risk population who has developed disease
Incidence Rate
- Number of people at risk and the time of risk
Incidence
- Formula
Incidence Proportion
- Total number of people with new disease / Total number of people in population at risk
Incidence Rate
- Total number of people with new disease / Total Person-Time at risk
2x2 Tabels
In RCTs we know which groups are exposed and which are not
- We try to emulate this in observational studies
Analytical Model
- Cross Sectional
Start with exposure and outcome
- Amount of exposed and unexposed are guessed
Analytical Model
- Case-Control
Start with outcome
- Can look at odds of an outcome
- No incidence rate
Analytical Model
- Cohort
Start with exposure
- No randomization
- Can look at exposure rates and outcomes
Odds Ratio
- Definition
The odds of an outcome happening in one group vs another (Exposed vs Unexposed)
- (a/b) / (c/d)
Odds Ratio
- Use
Cross Sectional, Case-Control, Cohort Studies
Odds Ratio
- What model
Logistic Regression Model
Relative Risk
- Definition
How more or less likely for an event to happen in one group vs another
- (A/A+B) / (C/C+D)
Relative Risk
- Use
Cross Sectional and Cohort Studies
- Used when we can calculate incidence
Survival Time
Patient important outcome that measures time to an event
- Don’t just focus on probability of an event occuring
Kaplan Meier
- Definition
X-Axis: Time In Study
Y-Axis:
- Time to Event Plot
- Survival Plot
Kaplan Meier
- Time to Event Plot
Cumulative proportion of people who had event of interest
- Goes up
Kaplan Meier
- Survival Plot
Remaining proportion of people who have not had event of interest
- Goes down
Hazard Ratio
- Definition
Takes into consideration occurrence of events and time interval after follow up to when event occurs (or end of follow up if event does not happen)
Hazard Ratio
- What model
Cox Proportional Hazards Model
Sensitivity Analyses
- Examples
Change duration of assessment window
- Longer or shorter
Change definition to identify exposure or outcomes
Add or remove covariates
