Biostats Cheat Sheet

positive predictive value A X (A + B) prevalence 100 (t) Negative Predictive Value D : (4 + (t : prevalence a X 100 (D + C) 2 Xi : (POPULATION E(Xi : [SAMPLe) : N Highest IQR Q3 : QI : = mean) - N STANDARD DeVIATION IQR - value Q3 - - - - Lowest mean) 2 1 Value Q1 half median lower half median Upper : + () blopsy do not have disease (thm blopsy) +: C ) have disease thru - STANDARD DeVIATION : +: Or d (+ ) RANGe : - : disease no disease - mean RISK/Benefits Measures of of Clinical Interventions RISK RISK # of events = # of Relative RR * * = RR >1 * RR RR risk (control) (90) event event of of event (ARR) reduction absolute risk verity RRR 1 = ex . among higher among is group /protective effect) intervention the same between is absolute risk reduction = risk Int lower is two groups intervention group Reduction (RRR) Relative Risk RRR (Intervention) (90) of risk , patients risk risk , I > risk , 1 = (RR) Risk 10090 x risk = (control) risk 2 % (RR) risk - (Intervention) * multiply treat to 10090 nonfatal >Risk of (intervention Number Needed 100% x (control) -relative answer : (ARR) M1 get 90) to was 29o lower group) compared among to placebo aspirin (control (NNT) = NNT R NNT Ideal If NNT NNH to NNT . is for every 20 , = Needed Number 1 IS Harm often 20 rounded up patients treated (nearest #) whole by Intervention , . I patient will . benefit (NNH) = ARI absolute Ideal I riskIncrease NNH NNH = is III D , (ARI) . for = NNH every risk (Intervention) is often 111 - rounded risk (control) down patients treated by (nearest #) whole intervention , I . patient will be harmed . ODDS RATIO case-control : OR odds = odds odds exposure (cases) exposure (controls) probability exposed probability not being exposed = OR-odds : conort studies exposure odds odds outcome outcome of : OR = odds of = odds odds outcome outcome probability = OR1 Or · · <1 on-1 outcome having of probability = , Increased odds , decreased odds odds , for point estimate 959C1 mean = of outcome outome having a outcome event of among is the = & ingas group active between same group active of event among two groups mean Ccritical value) (standard - = SE mean event of Confidence Interval Cl outome (control # of those wo · unexposed # of those w/outcome 2 outcome # those (intervention) outcome of I- odd exposed those outcome having # of those wo RCTs # # of those w/outcome · odd having probability I- = (unexposed) outcome of odds exposure (exposed) outcome probability = 2 error of mean) (1 96) (SEmean) . SD = M · calculate both z-critical · 95 - - - from upper and lower bound z-table %, one-talled have all find 0 . w/ Of talls test in one-tall 9500 in table If choosing 0 interval 21 & : 5%0 . get rangel to (10090 (there 9505 , It - is 5% none corresponds > to z find 1 = 9590 = . 0 . 9500) closest value = (left-column) 6 1 = . 65 0 + . 9505 0 . 05 or 0 . 9495) (top row) · 95 - - - %, two-talled test each tall find 0 0 . . 9750 has 2 . : 590 (100 % 2 - ex : from corresponds (left-column) 1 9 to : / -table tcritical (2 If ex 97 590 = . 0 = +critical) 1 It . n = 5 sample %, two-talled 24 n = = 5So = of talls , 85 , , df df = n . 06 Hop row) 84 calculation = given value size mean standard deviation . CI o , 2 = . 069 , n = 85) (since no 1 98 = . 84 d8 , = choose 120 nearest > z-score - 1 96 = . 24) = 0 23 = two-taled , , + tari 1 z-score 9750) 1 96 = > z-score . 9750 z t-critical 590 . [+) = (-) = above below mean mean = higher = 1 980 . . 550 df)