Screening Flashcards
What is screening
Systematic testing to detect disease at a stage when cure/control is possible.
Screening vs Diagnosis
Traditional diagnosis involves individuals with symptomatic disease.
Screening involves asymptomatic individuals with low pretest probability of disease.
(Diagnosis almost always happens when patients come in and coplain about symptom. While screening happens when individuals without obvious symptom come voluntarily.)
When to screen
Ideally, we want to screen population at “Biologic onset” stage.
Why screening?
- Early detection of disease precursors/signals and disease susceptibility in individuals who without diagnosed disease.
- Control disease development and recurrence, and its associated disability and mortality.
Important considerations at screening
- Ethical (Harms vs Benefits)
− Subject anxiety and/or stigmatization
− Invasive procedure
+ Improved quality of life
+ Effective treatment can be administered
- Characteristics of the condition (disease)
The things we screen on should be : Serious for the individual and the community; Sufficient prevalence in the population
- Characteristics of the screening test
- Reliability and validity (accuracy)
- Feasibility (burden, cost)
- Effective (screening reduces severity, mortality)
Test reliability: what? How? Types? Measures?
- Ability of a measuring instrument to give consistent results on repeated trials (precision)
- Established by repeated measurements
- Technical reliability (measurement, instrument); Intra- and Inter-examiner reliability
- Intraclass correlation coefficient (RIC); Kappa coefficient (k)
Test validity: what? How? Measures?
- Ability of the measuring instrument to give a true measure of the phenomenon being measured (accuracy)
True measure = ‘gold standard’/criterion measure
- Established by comparison with a “gold standard” measurement or established diagnosis criteria determined by biopsy, surgery, autopsy, etc.
We calculate the degree to which the measure is free of error.
- Two measures: sensitivity (Se), specificity (Sp)
Measures of test validity: sensitivity (Se), specificity (Sp)
Work on the colums
Both in Percentage
Sensitivity: proportion of those with disease testing positive on screening. Se = TP / (TP+FN) = A/A+C
Specificity: proportion of those without disease testing negative on screening. Sp= TN / (TN+FP) = D/B+D
Measures of case yield (predictive values): Positive (PV+), Negative (PV-)
Work on rows
In percentage
Positive predictive (PV+): the proportion of individuals screening positive who actually have disease.
PV+ = TP / (TP+FP) = A/A+B
Negative predictive (PV-): the proportion of individuals screening negative who actually do not have disease.
PV- = TN / (TN+FN) = D/C+D
Relationship between measures of validity and case yield
If there are a lot of false negative (FN or C), may have low sensitivity (Se), and PV- may be high.
If there are a lot of false positive (FP or B), may have low specificity (Sp), and PV+ may be low.
Interpretations of Se, Sp, PV+, PV-
Se: xx% of those who actually have the conditions/disease, we screened for positive.
Sp: xx% of those who don’t have the conditions/disease, we screened for negative.
PV+: xx% of those positive screens are true positive. This is low because there are a lot of false positive (B).
PV-: xx% of those negative screens are true negative This is low because there are a lot of false negative (C).
Cut-point Selection for Continuous Factors
- Lower cut-point – more screen positive:
FN lower; FP higher
Se enhanced; Sp reduced
We lower cut-point when:
- Identifying every case is necessary (HIV; PKU)
- Early treatment is important (breast cancer)
- Costs for additional tests is low
- Higher cut-point – more screen negative
FP lower; FN higher
Sp enhanced; Se reduced
We higher cut-point when:
- Costs & risks of re-screening high (colorectal cancer)
- Low false positive is desired (gallbladder disease)
How to enhance Se and Sp?
Use multiple tests!
What is Simultaneous testing (parallel testing) and how to define positive/negative
Complete multiple different tests
Screen positive = either test positive
Screen negative = all tests negative
Tends to enhance Se by lower FN (results in lower Sp)
What is Sequential testing (serial testing) and how to define positive/negative
Positives on test #1 proceed to test #2 (more expensive)
Screen positive = all tests positive
Screen negative = either test negative
Tends to enhance Sp by lower FP (results in lower Se)
