Epidemiology

Bias

Bias Subject → Stop with single blinding Howthorne → Subjects change behavior when observed Recall → Sick patients remember more Reporting → Subjects don’t report due to culture Researcher Selection Attrition → lost to follow up Ascertainment → Sample not representative of population Berkson → Sample from hospital Prevalence bias...

Study Design

Clinical Studies Experimental → Have intervention Open labeled Single blind Double blind Observational → No intervention With comparison Longitudinal Case-control Cohort No comparison Case report Case series Cross-sectional

Confidence Interval

Diagnostic Tests

Sensitivity Specificity Positive Predictive Value Negative Predictive Value

Hypothesis Testing

Null Hypothesis P Value → Is always compared to Alpha (0.05) 1-Alpha → Confidence Interval Alpha → Type 1 Error Beta → Type 2 Error 1-Beta → Power

Risk

Relative Risk Odds Ratio Absolute Risk Reduction → Number needed to Treat Attributable Risk → Number needed to Harm Relative Risk Reduction