Sensitivity and Specificity
specificity formula
= [# of true negative] / [total # w/o disease]
sensitivity formula
= [# of true pos] / [total # w disease]
-LR equation
= [false negatives] / [true negatives] 0-1.0
NPV equation
= [true negative] / [all that tested negative]
PPV equation
= [true positive] / [all the tested positive]
+LR equation
= [true positives] / [false positives] can range from 0-10+
Negative likelihood ratio (-LR)
How much less likely a negative test is in a client with the disease compared to a client w/o the disease
clinical question of specificity
The client does not have the disease. Will they test negative?
clinical question of sensitivity
The client has the disease. Will they test positive?
Tyle II error
false negative
complement of true positive
false negative (Type II error)
Type I error
false positive
complement of true negative
false positive (Type I error)
what do you want out of a screening test?
high sensitivity (aka a low false negative rate so that you can rule out negatives--SnNout)
what do you want out of a diagnostic test?
high sensitivity and specificity
what do you want out of a confirmatory test?
high specificity (aka a low false positive rate so that you can rule in positives--SpPin)
positive likelihood ratio (+LR)
how much more likely a positive test is in a client with the disease compared to a client w/o the disease
what's the most useful clinical measure?
likelihood ratios
true negative
number of people with negative result and without the disease
false negative
number of people with negative test results and disease
true positive
number of people with positive test results and disease
false positive
number of people with positive tests results and without the disease
what are PPV and NPV dependent upon?
prevalence of condition
negative predictive value (NPV)
proportion of people with negative test results who are correctly diagnosed, aka probably that a pt does not have a disease given a negative test
Positive predictive value (PPV)
proportion of people with positive test result who are correctly diagnosed, aka probability that person has the disease given a positive test
sensitivity
the true positive rate, OR probability that an affected person will test positive
Specificity
true negative rate, OR probability that an unaffected person will test negative
SnNout
when sensitivity is high, a negative results helps to rule out the disorder (b/c the true positive rate is so high it means the rate of false negatives is very low)
SpPin
when specificity is high, a positive test helps to rule in the disorder (b/c high specificity means high true negative rate therefore low false positive rate)
