Sensitivity vs Specificity of a Lab test
I have been studying microbiology lately and I was having trouble understanding the difference between specificity and sensitivity. Since I have figured it out I would like to share the concept.
For a Lab test, you will always hear about it's high/low specificity and high/low sensitivity... What's the difference between the two? And then what does false positive and false negative mean? There was a Yahoo answer to this question.... But I will add my own explanation in the end as well
The yahoo answer says
"sensitivity = probability of a positive test among patients with disease
specificity = probability of a negative test among patients without disease
So a test with low sensitivity does a relatively poor job at detecting occurrences of a condition when the condition exists and a test with low specificity does a relatively poor job at detecting non-occurrences of a condition when the condition doesn't exist.
For example, let's say that I had a group of 10 people: 5 men and 5 women. The first five people in the group are men (1-5) and the next five are women (6-10). I develop some sort of test that I think will let me identify men. The results of my test come in and I determine that person 1, 2 , 3 , 4, 7, and 9 is a man.
The test sensitivity would be: 4/5 = 80% (because I correctly identified 4 of the 5 men as men).
The test specificity would be 3/5 = 60% (because I correctly identified 3 of the 5 women as "not men").
A test can have both high sensitivity and specificity, both low sensitivity and specificity, or a combination"
This answer basically explains everything except the false positive and false negative.
False positive: as the name says it! A result which is actually not positive ie a person with no disease is diagnosis with one
False negative : once again name implies it. It means that the person has a disease but due to an error in lab investigation he is given a thumbs up sign for not having the disease.
So now taking the same example from Yahoo answer,
The number of false positive results in this example are two (7 and 9 are females but they are "diagnosed" wrongly as men; false positive)
The number of false negative is only one (5 was actually a man but falsely assumed a woman)
This gives us an important relationship :
As the specificity of a test decreases(as in this example) the number of false positives increases.
On the other hand , as the sensitivity of a test increases the number of false negative decreases.
Now let's make some sense of it..
Let's make an analogy
You are taking an MCQ test, if you have low specificity it means you are not good at eliminating or ruling choices out therefore every choice seems a potential answer to you (even though there are some choices which are clearly wrong)
And conversely if you have good specificity you will eliminate better you will have less false positive results.
Now, taking this mcq example further, if you have high sensitivity you have a greater chance of picking the right answer than a wrong one
Conversely, if you have low sensitivity you have a less chance of picking the right answer and a very high chance of pickin a wrong one (false - positive)
I hope I am clear and didn't confuse more!
|The above post was thanked by:|
Here is another explanation that I hope helps.
Sensitivity (Se) and specificity (Sp) are fixed properties of a diagnostic test.
Sensitivity (also known as the true positive rate) is the proportion of all people with the disease who test positive, or the probability that when the disease is present, the test is positive (Se = TP/(TP+FN) or 1-FNR). A highly-sensitive test (e.g. when the test is usually positive in the presence of actual disease or Se approaches 100%) is desirable when there is a significant penalty for an incorrect negative test (e.g. a false negative that results in missing curable colon cancer). A highly SeNsitive test is used as an initial screening test and is most helpful when Negative to rule OUT a disease (SN-N-OUT).
Specificity (also known as the true negative rate) is the proportion of all people without a disease who test negative or the probability that when the disease is absent the test is negative (Sp= TN/(TN+FP)) or 1-FPR). A highly-specific test (e.g. when the test is usually negative when the actual disease is absent or Sp approaches 100%) is desirable when there is a significant penalty for an incorrect positive test (e.g.the physical, emotional, and financial cost of surgery for a positive biopsy). A highly SPecific test is used for confirmation after a positive screening test and when Positive it rules IN actual disease (SP-P-IN).
In the example below, a 2x2 table is used to visualize the evaluation of the FOBT as a diagnostic test. The test (FOBT) may be either positive (for blood) or negative for a given patient, who may be positive for actually having a disease (has colon cancer based on endoscopy and biopsy) or negative (does not have colon cancer). In the example, 20 patients tested positive and had the disease while 10 tested negative and had the disease so the sensitivity is 20/(20+10) or approximately 67%. The false-negative rate is FN/(TP+FN) or 10/(20+10) ~ 33%, which is also 100%-Se. 180 patients tested positive and did not have the disease while 1820 patients tested negative and did not have the disease so the specificity is 1820/(180+1820) or approximately 91%. The False Positive Rate is 180/(180+1820) or approximately 9%, which is 100%-Sp.
Le, Tao; Bhushan, Vikas. First Aid for the USMLE Step 1 2016 (Page 33).
Fletcher, R. Clinical Epidemiology: The Essentials (Page 111)
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|Specificity Sensitivity Question!||steps_8186||USMLE Step 1 Forum||5||08-10-2016 06:59 AM|
|sensitivity vs specificity||medicalbiology||USMLE Step 1 Forum||1||01-20-2014 05:12 AM|
|Specificity Sensitivity Question!||tyagee||USMLE Step 1 Forum||3||03-09-2012 07:48 AM|
|parameters of sensitivity and specificity||manphnx||USMLE Step 1 Forum||6||01-31-2011 01:35 AM|
|Sensitivity Specificity Equations||step1||USMLE Step 1 Forum||4||12-26-2009 11:11 PM|