Validity and Reliability of Diagnostic and Screening Tests Flashcards
Lead time bias
- Primary determinant of success of early treatment
- Interval from detection with screening to time which diagnosis would have been made without screening
- When survival is counted from the point in time when early diangosis was made.
Different for different individuals
Can introduce bias false impresion of improved survival in a screened population.
Survival outcomes look better because of early detection
The patient is diagnosis bc of screening but the patient still dies of disease
Sensitivity
Proportion of a test to detect positive cases
Specificity
Probability of a test to detect negative cases
Positive predictive value
From those positive who actually has the disease
If prevalence increase, PPV increase, NPV decrease
Negative predictive value
From those negative who actually does not have the disease
Validity
Ability of a test to measure what is intended to measure. Distinguish who has a disease and who does not.
TAbility of test or measuring instrument to give a true measure.
Help us determine the comparison test or gold standard
Reliability
The likely hood that on repetition, they deliver the same results in same situation or condition
Repeatability or if results can be replicated
Measures like overall agreement and Kappa statistic
Steps to calculate percent agreement
- Table with comparisons (a,b,c,d)
- Cardiologist or observations on totals horizontal and vertical with percentages
- Sum only agreements of TP and TN
- Divide them by Total observations
Steps to calculate percent agreement due to chance
- Take the total of vertical observation or specialist, multipliy by horizontal percentage (14.2 = 0.142)
- Repeat for both TN
- Repeat for TP
- Then add both results and divide by total observations or population
Kappa statistic
> 0.75 excellent agreement
0.40 to 0.75 good agreement
<0.40 poor agreement
