Lecture 3B

Question 1

What is predictive value

Answer 1

Probability that a diagnos.c test accurately reflects the presence or absence of disease

Question 2

What is predictive values defined using

Answer 2

Can be defined for both positive and negative tests

Question 3

How do you calculate the positive predictive value

Answer 3

– For positive tests predicting disease
• True positives / total number of positive tests
• True positives/true positives + false positives

Positive predictive value = TP/(TP + FP)

Question 4

How do you calculate the negative predictive value

Answer 4

– For negative tests predicting healthy status
• True negatives / total number of negative tests
• True negatives / true negatives + false negatives

Negative predictive value = TN/ (TN + FN)

Question 5

What can positive predictive value be use for

Answer 5

When the cost of treatment is high

Question 6

What can negative predictive value be use for

Answer 6

when cost of missing sick animal is high

Question 7

What is the efficiency of the test

Answer 7

Proportion of animals that were correctly classified as having the disease or being healthy
– (True positives + True negatives) / total number of animals

efficiency = (TP + TN)/(TP + FP + TN + FN)

Question 8

What is prevalence

Answer 8

Proportion of animals that have the disease in a population
– Diseased animals / total number of animals
– (True positives + False negatives) / total number of animals
Prevalence = (TP + FN)/(TP + FP + TN + FN)

The disease prevalence is not affected by the test characteristics
– Dependent on the disease
– Dependent on population considered
– Changes over time

Question 9

What is the use of prevalence

Answer 9

The disease prevalence is not affected by the test characteristics
– Dependent on the disease
– Dependent on population considered
– Changes over time

Question 10

What is PV+

Answer 10

Predictive value
TP/(TP+FP)
Can also be calculated by
PV+ = (Pr x Se)/((Pr x Se) + [(1-Pr) x (1-Sp)])

Question 11

What types of errors are there

Answer 11

• Random errors
• systematic errors
• Human errors

Question 12

What is random errors

Answer 12

Affect reproducibility of the test
– Caused by addition of small factors in multi-step procedure
• Pippetting errors that accumulate as test proceeds
• Instability of calibration curve / degradation of key reagents
• Conditions on the day (i.e. temperature in the lab)

Question 13

What is systematic errors

Answer 13

Consistently high or low results
– Caused by interfering substance in all samples or reagents that affect the test
• Enzyme inhibitors in enzymatic assays
• Improper blanking of the sample

Question 14

What are human errors

Answer 14

Mistakes made by operators
– Caused by poor standard operating procedures, boredom, inexperience
• Mislabeling of tubes
• Errors in measurement of reagents quantities (for example decimal point errors)

Question 15

How to minimise human error

Answer 15

fool-proofing of procedures
√ Build in fail safe systems at planning stage
√ “It is difficult to fool-proof a procedure because fools are so inventive…”
• Example: duplicate sti ckers one for report and another for the sample

Question 16

What is swiss cheese model

Answer 16

Building in safety nets in every level to prevent error

Question 17

How to improve diagnostic certainty

Answer 17

• Repeat the same test on same samples but only for random errors
• Perform independent test based on different principle, you can multiply both probability together. must be independent for each test
• Set cut-off at appropriate levels
• Minimise human error