Epidemiology and Stats Flashcards
primary levels of prevention
prevent onset of dz
- vaccines, diet/exercise
secondary levels of prevention
prevent progression of dz
- screening, hypertension meds
tertiary levels of prevention
prevent complications of dz
- surgery, rehab
colon cancer screening
50-75
colonoscopy q10y
flex sig q5y + FOBT q3y
FOBT q1y
breast cancer screening
40-49 if + famhx
- only BRCA1/2
50-74
- mammo q2y
cervical cancer screening
21-65
Pap smear q3y
lung cancer screening
55-80 +; 30 pack-yr +; quit < 15yrs
low dose CT q1y
cholesterol screening
female 45, male 35, 20 if at risk
fasting lipid q5y
HTN screening
everybody
check BP every visit
DM screening
HTN
A1c
osteoporosis screening
female 65, high risk at 60
one time DEXA
AAA screening
male >50, smoker
one time u/s abdominal aorta
STDs
sexually active
rapids tests at least once, pref q1y
hepatitis c screening
baby boomers
one time ab screen
depression screening
all adults
PHQ9, clinical judgment
alcohol misuse screening
all adults
interview
diagnostic tests 2x2 table
Test +, Disease + = true positive = A
Test +, Disease - = false positive = B
Test -, Disease + = false negative = C
Test -, Disease - = true negative = D
sensitivity
A/(A+C)
ability to detect true positives
specificity
D/(B+D)
ability to confirm true negatives
PPV
A/(A+B)
probability of a disease in pt with + result, increase prevalence increase PPV
NPV
D/(C+D)
probability of no disease in pt with - test result, increase prevalence decrease NPV
observational studies
case series
cross sectional
cohort
case control
randomized controlled trial
gold standard, uses intervention vs. control groups & tracks dz outcomes, uses odds ratio
intervention = treatment
control = placebo, standard of care, nothing
case series
qualitative, narrative
cross sectional
retrospective, snapshot of dz and exposure in a given time, uses prevalence
cohort
prospective, starts with exposed vs. unexposed and tracks dz outcomes over time, uses relative risk
case control
retrospective, starts with dz vs. no dz and looks at exposures, uses odds ratio
observational studies f/u
cannot establish causation, only correlation
methods to eliminate bias
randomization, blinding, standardization, statistical controlling