Sensitivity & Specificity
What would a low specificity lead to?
If a test has a LOW Specificity, it would lead to many false diagnosis (FALSE POSITIVE: test stating a truly HEALTHY patient is SICK)
What would 100% specificity mean?
If a test has a Specificity of 100%, it would identify EVERY healthy patient and have no FALSE POSITIVES
If the prevalence goes up, the PPV and NPV will....
PPV goes up NPV goes down
What is specificity?
W•Looking at a test's ability to detect people who DO NOT HAVE Disease X •Percentage of people who are correctly identified as HEALTHY •Of all people WITHOUT Disease X, how well does Test Y do at correctly identify them as disease free?
X axis
disease X
Bottom left box
false negative
Top right box
false positive
What would a 100% sensitivity test do?
identify every sick patient
How are specificity and sensitivity related?
inversely related
What would a low sensitivity test do?
lead to many missed diagnoses
Side of the chart
positive, negative
SNOUT
sensitivity--a test is good at ruling OUT disease
Top of the chart
sick, healthy
SPIN
specificity--a test is good at ruling IN a disease
Y axis
test result
Bottom right box
true negative
Top left box
true positive
How will a high prevalence affect predictive values?
•Higher prevalence will INCREASE the PPV and LOWER the NPV
What is sensitivity?
•Looking at a test's ability to detect people who DO HAVE disease X •Percentage of people who are correctly identified •Of all people with Disease X, how well does test Y do at correctly identify them
Why will a low prevalence have a low PPV
•Lower prevalence (IE: a rare disease) will have a low PPV due to an increased False Positive amount
What do the predictive values depend on?
•Predictive Values vary according to the prevalence of a disease
What do sensitivity and specificity depend on?
•Sensitivity & Specificity depends solely on the characteristics of the test
What is the sensitivity formula?
•Sensitivity= (TRUE POSITIVE / (TRUE POSITIVE + FALSE NEGATIVE)) * 100
What is the specificity formula?
•Specificity= (TRUE NEGATIVE / (TRUE NEGATIVE + FALSE POSITIVE)) * 100
What is a true positive?
•Test result correctly identifies a SICK patient with Disease X •SICK ID'd as SICK
What is a true negative?
•Test result correctly identifying a HEALTHY patient without Disease X •HEALTHY ID'd as HEALTHY
What is a false negative?
•Test result incorrectly identifying a SICK patient without Disease X •SICK MISS ID'd as HEALTHY
What is a type I error?
•Test result incorrectly identifying a SICK patient without Disease X •SICK MISS ID'd as HEALTHY •aka false negative
What is a false positive?
•Test result incorrectly saying a HEALTHY patient has Disease X •HEALTHY ID'd as SICK
What is a type II error?
•Test result incorrectly saying a HEALTHY patient has Disease X •HEALTHY ID'd as SICK •aka false positive
Negative predictive value
•What proportion of negative results are TRUE NEGATIVES •NPV= (TRUE NEGATIVE / (TRUE NEGATIVE + FALSE NEGATIVE)) * 100
Positive predictive value
•What proportion of positive results are TRUE POSITIVES •PPV= (TRUE POSITIVE / (TRUE POSITIVE + FALSE POSITIVE)) * 100