USMLE Step 2 CK: Statistics and Epidemiology

distinction between case control and retrospective cohort study

case control - identifies the outcome first and looks back for associated risk factors

retrospective cohort : identifies risk factor exposure first, then looks for the outcome.

likelihood ratio purpose

measures performance of diagnostic tests independent of disease prevalence

accuracy of a diagnostic test in equation form

(TP + TN) ÷ (TP + FN + TN + FP)

positive predictive value

go horizontally; TP/ (TP + FP)

negative predictive value

go horizontally; TN/(TN+ FN)

insufficient power runs the risk of committing a.....

type II error - saying they're the same when they are actually different.

number needed to treat (NNT) = ?

NNT = 1/ARR

ARR = (rate of disease in the control group) - (rate of disease in the intervention group)

the most common sources of metastases to the brain

Lung > Breast > Unknown Primary > melanoma > colon

effect modification vs. comfounding

the association actually has a direct indluence over the outcome whereas the confounder is only associated with no influence on causality.

selective survival bias

a study on cancer survival is a case control study that is taking into account all cancers - not jist newly diagnosed cases and will include largely benign cases

Qualitative Study

Studying social or other less quantifable phenomena, like how pts think about disease.

Phenomenology-based

A qualitative study often includes interviews of individuals, or focus groups, hearing perspective

Case report

Disease course and outcome o fsingle pt (NOT analyitical) since no hypothesis testing

Case series

Descriptive study describing hx, possible exposure, and clinical findings of group pts w siliar dx. Not testing hypothesis.

Matched case control

one group of pts w outcome under study (cases), against second matched group without the outcome (controls). ID associated exposure w each group. Prospective or retrospective, always observational.

Randomized trial

prospective, interventional, assigned randomly to receive intervention. Compared against placebo or standard, depending non design/disease treatment. Gold standard when combined w blinding, more powerful evidence than other observational or descriptive studies