Week 9 Flashcards
mass screening
whole population or subset
multiple or multiphasic screening
many screening tests at the same time
targeted screening
groups with a particular exposure
case-finding/ opportunistic screening
tested while at the doctors for another purpose
3 ways to distinguish between normal and abnormal
- normal as common
- abnormal associated with disease
- abnormal as treatable
- normal as common
Values that occur frequently are normal
Values that occur infrequently are abnormal
Normal distribution: 2.5% abnormal using + 2SD cut off
Percentile (95%): 95% normal and 5% abnormal
- abnormal associated with disease
Use of the distribution of measurements for both healthy/diseased people with an attempt to define the cut-off that separates the two groups
Results in some healthy people on the ‘abnormal’ side and some diseased people on the ‘normal’ side
- Abnormal as treatable
Definition of abnormal changes over time based on changing treatment thresholds
before screening, disease must be
Relatively common (prevalent) disease
Several consequences
Early treatment produces better outcomes
Considered a problem by people
Natural history well-understood
Relatively long preclinical phase when disease could be detected by screening
the screening test should be
Good accuracy OR high sensitivity and/or specificity Safe simple/logistically manageable Relatively cheap Acceptable to ‘healthy’ people
positive results threshold- screening test
tend towards high sensitivity not to miss potential disease
positive results threshold- diagnostic test
tend towards high specificity (true negatives)
more weight given to accuracy and precision than to patient acceptability
positive result- screening test
indicates suspicion of disease (often used in combination with other risk factors) that warrants confirmation (diagnosis)
positive result- diagnostic test
result provides a definite diagnosis
in tests, we are most concerned about
false negatives
true positive
disease present
test positive
false positive
disease absent
test positive
false negative
disease present
test negative
true negative
disease absent
test negative
sensitivity
probability of a positive test in people with the disease
sensitivity F
true positive/total diseased
a/a+c
specificity
probability of a negative test in people without the disease
specificity F
true negative/ total non disease
d/b+d
positive predictive value (PPV)
probability a person has the disease when the test is positive
PPV F
true positive/total positive
a/a+b
negative predictive value NPV
probability a person does not have the disease when the rest is negative
we want high (near 100%) NPV to avoid false negative
NPV F
true negative/total negative
d/c+d
sensitivity and specificity
relate to the features of the test
Do not change with the prevalence of the disease
PPV and NPV
do change with the prevalence of the disease
PPV will be low when prevalence is low
O’briens labral tear test accuracy
Diagnosed via physiotherapists using the Active Compression Obrien test
DST test accuracy
Used to diagnose depression
Screening- length time bias
Screening often diagnoses diseases that are less aggressive than those present clinically
Thus treatment needs to be safe and effective- as sometimes disease detected through screening wouldn’t present/be diagnosed clinically
Screening- lead time bias
Early detection: disease diagnosis by screening before clinical presentation
As a result there is lead time, extra time that you know you have the disease
Thus ‘survival time’ should not be used to evaluate a screening test- as survival in screened individuals will appear longer.
This is irrespective of whether screening affects the survival course or not
chi squares
two categorical variables
expected cell frequency
(row total x column total) / grand total
df
(row - 1) x (column - 1)
chi square test not valid if
more than 20% of the cells have expected frequency smaller than 5
if more than 20% of the cells have expected frequency less than 5
use Yates continuity corrected chi squares
