board review intro/biostatistics

Question 1

diagnostic error

Answer 1

missed opportunity

Question 2

types of studies

Question 3

cross-sectional

Answer 3

moment in time

Question 4

case control

Answer 4

generally retrospective

Question 5

cohort

Answer 5

2 groups - based on presence or absence of risk factor

Question 6

RCT

Question 7

systematic review with meta analysis

Question 8

which study is best for studing rare disease

Answer 8

case control

Question 9

strongest experimental design

Answer 9

rct

Question 10

best to determine disease prevalence

Answer 10

cross-sectional

Question 11

studying outcomes that develop over time or survival analyses

Answer 11

cohort

Question 12

heterogeneity

Answer 12

most important limitation of meta-analysis

Question 13

confounding

Answer 13

try to account for it with randomization

Question 14

validity

Answer 14

internal – how well study error is minimized (ie, errors in sampling, measurement, and analysis)

external – aka generalizability
–extent to which study results can be applied to other settings

threats to validity
–systematic error and random error

Question 15

systematic error

Answer 15

attempt to min via study design
1. confounders - factors other than variables being studied that are associated with the study population
mitigate: matching, statistical technique

2. bias – presence of factors that skew results in a specific direction
   mitigate: study design
   —ie, appropriate selection of particpants

Question 16

randomization

Answer 16

minimize systematic error
– try to ensure the groups being studied differ only in terms of intervention

Question 17

random error

Answer 17

error introduced by chance
–assess via CI and p-values

mitigate: increase study size, use precise measurement techniques

Question 18

statistical analysis

Answer 18

indicates likelihood of study result being caused by chance alone

does not compensate for bias

the larger the sample size the more likely you are to detect a difference
-always consider practical significance

Question 19

CI

Answer 19

range of values which the true results fall
-95% confidence that true value is within range

a CI that crosses the null value (ie, 1 for RR and OR) is not statistically significant

Question 20

assessing the value of a diagnostic test

Answer 20

sensitivity
specificity
PPV
NPV
LR

Question 21

sn and sp

Answer 21

inherent to test themselves and does not change with prevalence of disease in population

sensitivity - proportion of patients with dz who test positive snout
ie, high sensitvity –> low false neg

spec – proportion of patients without dz who test negative
high sp –> low false positive
-spin

Question 22

as sensitivity increases

Answer 22

false negative ratio decreases
npv

Question 23

as sensitivity increases

Answer 23

false negative ratio decreases
npv increases

Question 24

as specificity increases

Answer 24

false positive rate decreases
ppv increases

Question 25

predictive values

Answer 25

reflects the validity of a positive or negative test result
-used to interpret a test result
-determined by sensitivity, specificity, and prevalence of disease

PPV–% of pt with pos test who have dz
NPV–% of pt with neg test who do not have dz

Question 26

as prevalence increases

Answer 26

PPV increases and NPV decreases

Question 27

LRs

Answer 27

also assess performance of a diagnostic test
-independent of dz preva

Question 28

LRs

Answer 28

also assess performance of a diagnostic test
-independent of dz prevalence

how much more or less likely is a disease based on the test result

pret test probability xLR = post-test probability
the magnitude of LR suggests how strongly a test will increase or decrease the likelihood of the disease

LR+ used when test is positive
sens/(1-sp)
2,5,10 increases post test probability by 15, 30 and 45%

LR- used when test is neg
1-sen/sp
0.5, 0.2, 0.1 decreases post test probability by 15, 30, and 45%

Question 29

risk estimates–NNT/NNH

Question 30

risk estimates ARR and RRR