Home / Summaries / Users' Guides to the Medical Literature A Manual for Evidence-Based Clinical Practice, Second Edition / lr-prob-posttest

Diagnostic tests

Q: How to calculate the posttest probability with the likelihood ratio?

1. Using nomogram; figure with 3 vertical lines, 1 pre 2 LR 3 post prob. draw a line from pre prob trough LR and read the post prob. 2. Calculate: convert pretest prob to odds multiply this odds by the LR convert the posttest odds to posttest prob this calculation is complicated, but online tools can help. note: the pre test probability is an estimate and influences the posttest probability. Likelihood ratios from tests regularly used in daily practice are easily accessable in the Rational Clinical Examination via JAMA

Q: What are the 2 advantages of using multi LR method, instead of dichotomized scores?

1. Resembling the true testresults better; more info is retained 2. When knowing the LR of a test result; one can easily move to posttest prob by using a nomogram.

Q: What is important by assessing the reproducibility of testresults

- poor reproducibility is caused by the test itself - poor reprod. Is caused by interpretation interpretation: high: + , test is simple to use and unambigious - , for interpretation highly skilled clinicians are needed low: - mediocre: +, when still dicriminating well between those with and without disease.

Q: What factors are important to take into account in applying the results in patient care?

1. Judge the reproducibility of testresults 2. Judge the applicability to practice; representativeness of spectrum of target + and target- patients and setting 3. Judge if your management strategie will change - by calculating the proportion of patients were a change in management strategie takes place (below test treshold/ above treatment treshold) using LR>10 and LR <0,1. - when using sequential tests; determine the added value of this test (clinical prediction rules can help with that). 4. Judge if the patient is better of (patient important outcomes)

6 important questions on Diagnostic tests

How to calculate a likelihood?

- ratio of the number of positive test result in patients with disease
divided by
- ratio of the number of positive test result in patient without disease

How to interpret a likelihood ratio?

- 1 = pre and post test are the same
- >1 = increase in probability of presence of disease (the higher the LR, the greater the increase)
- <1 = decrease in pob of presence of disease (the smaller LR, the greater the decrease)

What to say about the usefulness of the size? (only a rough gide line):

conclusive changes: if LR > 10 or LR < 0,1
moderate change: if LR 5 -10 or LR 0,2 - 0,5
small change: if LR 1-2 or LR 0,5 - 1

How to calculate the posttest probability with the likelihood ratio?

1. Using nomogram;

figure with 3 vertical lines,
1 pre 2 LR 3 post prob.
draw a line from pre prob trough LR and read the post prob.

2. Calculate:

convert pretest prob to odds
multiply this odds by the LR
convert the posttest odds to posttest prob

this calculation is complicated, but online tools can help.

note: the pre test probability is an estimate and influences the posttest probability.

Likelihood ratios from tests regularly used in daily practice are easily accessable in the Rational Clinical Examination via JAMA

What are the 2 advantages of using multi LR method, instead of dichotomized scores?

1. Resembling the true testresults better; more info is retained
2. When knowing the LR of a test result; one can easily move to posttest prob by using a nomogram.

What is important by assessing the reproducibility of testresults

- poor reproducibility is caused by the test itself
- poor reprod. Is caused by interpretation

interpretation:
high:
+ , test is simple to use and unambigious
- , for interpretation highly skilled clinicians are needed

low:
-
mediocre:
+, when still dicriminating well between those with and without disease.

What factors are important to take into account in applying the results in patient care?

1. Judge the reproducibility of testresults
2. Judge the applicability to practice; representativeness of spectrum of target + and target- patients and setting
3. Judge if your management strategie will change
- by calculating the proportion of patients were a change in management strategie takes place (below test treshold/ above treatment treshold) using LR>10 and LR <0,1.
- when using sequential tests; determine the added value of this test (clinical prediction rules can help with that).
4. Judge if the patient is better of (patient important outcomes)

The question on the page originate from the summary of the following study material:

Users' Guides to the Medical Literature A Manual for Evidence-Based Clinical Practice, Second Edition | 9780071590358 | Gordon Guyatt, et al

View summary