Statistics and Other Maths Flashcards
What is the formula for A-a Gradient?
(FiO2 x (Atmospheric pressure - PH2O) - PaCO2/0.8) - PaO2
Simplified (150 - PaCO2/0.8) - PaO2
Formula for Number Needed to Treat
1/ARR (Absolute Risk Reduction)
Formula to convert from risk to odds
Odds = risk ÷ (1 - risk)
Formula to convert from odds to risk
Risk = odds ÷ (odds + 1)
Formula for sensitivity
Sensitivity = True Positive ÷ (True Positive + False Negative)
i.e. proportion of patients with disease who test positive
More sensitive = fewer false negatives = exclude a disease with negative test
Formula for specificity
Specificity = True Negative ÷ (True Negative + False Positive)
i.e. proportion of patients without disease who test negative
More specific = fewer false positives = rule a condition in with positive test
Formula for Positive Predictive Value
PPV = True Positive ÷ (True Positive + False Positive)
i.e. what proportion of patients with a positive test have the disease
NOTE: depends in prevalence of disease
Formula for negative predictive value
NPV = True Negative ÷ (True Negative + False Negative)
i.e. what proportion of patients with a negative disease do not have the disease?
NOTE: dependent on disease prevalence
Positive Likelihood Ratio
Positive Likelihood Ratio = Sensitivity ÷ (1 - Specificity)
Negative Likelihood Ratio
Negative Likelihood Ratio = (1 - Sensitivity) ÷ Specificity
Post-test odds if test is positive
pre-test odds x positive likelihood ratio
Post-test odds if test is negative
Pre-test odds x negative likelihood ratio
What study should be used to determine prevalence?
Cross sectional study
What study should be used to determine incidence?
Cohort study
What study type can be used to determine prognosis?
Cohort study
Selection Bias
Means of selecting patients into arms of the study is not the same between arms
Measurement Bias
Measurements of exposures and outcomes different in different arms
Lead Time Bias
Survival post diagnosis is longer because disease is diagnosed earlier in its course
Length Time Bias
Diagnose more indolent disease with screening versus the cases that present clinically.
i.e. the cases picked up by screening are milder/slower growing and likely have better survival than cases that develop between screening interval which are likely to be aggressive given their rate of development
Health Volunteer Bias
Healthy people are more likely to participate in screening programs
