Public Health Flashcards
Cross Sectional
Case-Control
Cohort
Cross-section - looks at CURRENT freq disease and freq risk
Case-Control - retro; start w/ disease then look at who had exposure (odds ratio)
Cohort - retro or prospective; start w/ exposure then look at disease (relative risk)
4 Phases of Clinical Trial
SWIM
I - safety (just healthy volunteers)
II - does it Work? (small # pts w/ disease)
III - Improvement (test against current drug or placebo)
IV - Monitor (post-marketing surveillance)
Sensitivity
Specificity
Sensitivity - SCREEN
1 - rate false neg
TP/ (TP + FN)
Specificity - CONFIRM
1 - rate of false pos
TN / (TN + FP)
PPV
NPV
PPV - if test is pos how likely to have disease
TP / (all pos) or TP / (TP + FP)
PPV increases w/ inc prevalence
NPV - if test is negative how likely to not have disease
TN / (all neg) or TN (TN + FN)
NPV decreases w/ inc prevalence
Likelihood Ratio
Multiply LR by pre-test probability to estimate post-test odds
pos LR = true pos / false pos = sensitivity / (1-spec)
neg LR = false neg / true neg = (1-sens) / specificity
What are ideal pos and neg likelihood ratios?
pos > 10
neg < .1
Odds Ratio
odds if exp / odds if no exposure
a/b) / (c/d
Relative Risk
Proportion of ppl w/ disease in exposure group / proportion of ppl w/ disease in non-exposure group
RR = 1 (no risk)
RR > 1 then inc risk if exposure
RR < 1 then dec risk if exposure
Attributable Risk
Diff in risk b/n groups
Risk in exposed - Risk in unexposed
AKA how much risk can be explained by exposure
Relative Risk Reduction
1 - RR
Proportion of risk reduction attributable to intervention compared to control
Absolute Risk Reduction
Risk in intervention group - risk in control group
Number Needed to Treat
Number Needed to Harm
NNT = 1/ARR
NNH = 1/ AR
Standard Error Equation
SD / square root of sample size
So std error dec as inc sample size
Type I Error
Type II Error
Type I - false pos (reject null when null is true)
alpha - prob of type I (usually set at .05)
Type II - false neg (do not reject null when there is an effect)
beta - prob of type II error
Power (what inc it?)
1 - beta
Inc by …
- inc sample size
- in effect size
- inc precision of measurement
When are confidence intervals significant?
If CI for mean diff between groups does not cross 0
If CI for odds ratio or relative risk does not include 1
If CIs between groups do not overlap
Coefficient of Determination
r^2 in correlation
amount of variance in one variable that can be explained by variance in the other variable
Order of Surrogate Decision Making
Spouse - adult children - parents - siblings
Health Maintenance Organization
Restricted to limited panel of providers in network
Denied payment for services that do not meet established guidelines
Need referral to see specialist
Point of Service
Can see out-of-network providers but higher copays and costs
Need referral to see specialist
Preferred Provider Organization
Can see in or out of network but all have higher cost
do NOT need referral to see specialist
Exclusive Provider Organization
Limited to a network
do NOT need referral to see specialist
Medicare Parts A-D
A - hospital and hospice
B - basic medical bills (doctors office fees and diagnostic tests)
C - combo of A and B from private company
D - prescription drugs
hospitAl, Basics, Combo, Drugs