Research articles can be filled with confusing statistics terminology. Here we define some of the more common terms used in research papers related to physical therapy, especially those more relevant to the orthopedic special tests we describe elsewhere on the site.
Interexaminer Reliability, also known as Inter-rater Reliability, is the degree of consistency between different examiners when they observe the same occurrence or test. For example, when multiple examiners perform the same special test on a patient, this term relates to how similar their reported observations are. The results are reported on a percentage scale from 0% (no examiners report the same result) to 100% (all examiners report the same result).
The Likelihood Ratio is used to communicate how likely it is that a person has a specific condition based on a positive result from the test being performed. The ratio is calculated by taking the probability that a patient with a specific condition will have a certain result on a given test and dividing it by the probability that a patient without that specific condition will have a given result on the same test. Likelihood ratios can be either positive or negative depending on whether you are describing the likelihood based on a positive test result or the likelihood based on a negative test result.
LR of 1 or lower generally means there is little probability that a patient has the condition being tested for.
LR of 1 is non-informative. It essentially is of no diagnostic value.
LR of greater than one means it is likely that a patient has the condition. The higher above 1, the more likely the patient has the condition being tested for.
Sensitivity in medical testing is the ability of a test to determine the presence of a condition; a “true positive”. This means that a test with 80% specificity will correctly diagnose 80 out of 100 people with a condition, but it will also incorrectly diagnose that 20 out of 100 have a condition when they actually don’t have the condition.
ie. How many with a torn rotator cuff does the test diagnose with a torn rotator cuff?
Specificity is a measure of how accurate a test is at determining that a patient does not have a given condition. That is, if a negative result for a given test means there is no chance the person has the condition being tested for, that test has very high specificity. Therefore, a test with 80% specificity will result in 20% of people being found to be negative for a condition who actually do have the condition. Specificity is said to determine “true negative”.
ie. How many people without a torn rotator cuff does the test diagnose as not having a torn rotator cuff?
Likelihood Ratio can also be expressed in terms of Sensitivy and Specificity using the following equations.
Positve LR = Sensitivity / (100 – Specificity)
Negative LR = (100 – Sensitivity) / Specificity