What are sensitivity and specificity?
Sensitivity and specificity are measures of the effectiveness of a diagnostic test. Most often they are used as part of medical research when doctors (or others) try to determine if a patient has a disease or not. This leads to four possible results:
- True positive – The patient has the disease and you say he does
- True negative – The patient does not have the disease and you say he does not
- False positive – The patient does not have the disease and you say he does
- False negative – The patient has the disease and you say he does not
Sensitivity is defined (see Dictionary of Statistics) as the conditional probability of having a positive test result, given that the patient has the disease. That is:
True positive/(true positive + false negative)
Specificity is defined as the “conditional probability of a negative test result, given that the patient does not have the disease”. That is
True negative/(true negative + false positive)
What are good values of sensitivity and specificity?
This varies by the state of knowledge in the field. Higher values of both measures are always better. But some diseases already have excellent diagnostic tools and some do not. Sometimes, sensitivity and specificity are used to compare two tests, one of which may be a “gold standard” and the other may be new and possibly better. Or the new test may be less expensive or easier to administer or have fewer side effects. Then researchers must decide if the sacrifice of sensitivity and specificity are worth the cost.
How can sensitivity and specificity be adjusted?
Many, if not most, diagnostic measures give a quantitative result. For example, a person’s blood pressure is not high or low, it is defined in millimeters of mercury. Researchers can increase sensitivity (and decrease specificity) by using different numbers as cut-offs for diagnosing disease.
Taking it to ridiculous extremes, if you simply say that every patient has the disease, then sensitivity will be 1, which is perfect. But then specificity will be 0, which is as bad as it can be.
Which is more important, sensitivity or specificity?
This depends on the disease, its prognosis, the effectiveness of treatment and the side effects of treatment. Is a false negative worse than a false positive? Sometimes it is, sometimes it is not.
If the disease is easily treated but very harmful if not treated then sensitivity is more important. In this situation, you do not want to miss anyone who has the disease. But if the disease is difficult to treat and not harmful, then specificity is more important. In this case you do not want to tell people who do not have the disease that they do.
Can sensitivity and specificity be used outside medicine?
Certainly. They can be used whenever a diagnosis is being made. This could be a promotion decision at work, a decision to issue a credit card or any decision that has two choices and can be right or wrong.
Sources: B.S. Everitt, Dictionary of Statistics, Cambridge University Press.