I have been told many times recently by GPs and other Microbiologists that there is no point doing a throat swab from patients with pharyngitis because there is now the Centor score that allows you to identify the Group A beta-haemolytic streptococcus, Streptococcus pyogenes, without having to send any tests to the microbiology laboratory. But is this true or is Centor like the half-man half-horse Centaur - a myth?
Click for larger image
The Centor score is a set of “bed side” criteria that is used in the diagnosis of streptococcal pharyngitis. It was developed in the US in an emergency department setting and has become widely used in the UK as a replacement for routinely taking throat swabs from patients with pharyngitis. The resulting Centor score (0-4) correlates to the percentage of patients who actually have S. pyogenes by comparing to positive bacterial culture from throat swabs.
The Centor score gives 1 point for each of the following characteristics:
The Centor system then makes recommendations about what to do in terms of investigation and treatment relevant to each score. If the score is 2 or 3 a rapid antigen test is then used; a bit like a pregnancy test, where a throat swab is taken and applied to the test device. The test gives either a positive or negative result and guides treatment.
Added to this Centor score, there is a modified Centor score (or McIsaac score), which includes patient age as a criterion and modifies the Centor score; S. pyogenes is far more common in those patients aged 3-14 years of age so they gain a point (even though they can still only score 4 as a maximum!) and far less common in the over 45s so they lose a point:
The Centor score gives 1 point for each of the following characteristics:
- Tonsillar exudates
- Tender anterior cervical lymphadenopathy
- Fever by history
- Absence of cough
The Centor system then makes recommendations about what to do in terms of investigation and treatment relevant to each score. If the score is 2 or 3 a rapid antigen test is then used; a bit like a pregnancy test, where a throat swab is taken and applied to the test device. The test gives either a positive or negative result and guides treatment.
Added to this Centor score, there is a modified Centor score (or McIsaac score), which includes patient age as a criterion and modifies the Centor score; S. pyogenes is far more common in those patients aged 3-14 years of age so they gain a point (even though they can still only score 4 as a maximum!) and far less common in the over 45s so they lose a point:
- +1 if aged 3-14 years
- +0 if aged 15-44
- -1 if aged ≥45
Centor score |
Patients who are actually S. pyogenes positive |
Recommendations |
0 |
0% |
No test, no treatment |
1 |
7% |
No test, no treatment |
2 |
21% |
Test with rapid antigen test. Treat positives |
3 |
38% |
Treat if rapid antigen test is positive OR treat empirically if rapid test not available |
4 |
57% |
Treat empirically |
Rapid antigen tests for S. pyogenes haven’t really caught on the UK yet but they are coming. Watch this space! In fact, they are now part of the National Institute of Health and Care excellence (NICE) guideline for the diagnosis and management of pharyngitis. These particular rapid antigen tests are said to be about 70% to 90% sensitive and 95% specific by NICE, which means there is up to 30% false negatives (undiagnosed and untreated) and 5% false positives (over diagnosed and over treated).
This isn’t very good. Now I may sound like an old Microbiologist not wanting to move with the times of cheap and rapid point of care testing making “us” old dinosaur Microbiologist’s redundant…but I’m not, I actually really support this technology but in my opinion before it is rolled out fully e.g. in NICE guidelines, it does need to improve!
The alternative is to send a bacterial throat swab for culture to the microbiology laboratory (there is no sensitivity and specificity for throat swabs as they are the “gold standard”, in theory they are 100% sensitive and 100% specific but in reality they are probably not). This method may be more costly and it takes longer (up to 2 days) but will still identify S. pyogenes quickly enough to allow for successful treatment with either penicillin or macrolides.
Why use the Centor score?
So if, as I believe, the Centor score is not that good at predicting S. pyogenes pharyngitis what is its use? Well it may surprise you, but I believe its real use lies in the same reason we treat S. pyogenes pharyngitis in the first place; we don’t give antibiotics to treat pharyngitis but rather to prevent complications.
S. pyogenes pharyngitis has a high complication rate, about 27% overall. These complications can be split into suppurative (relating to direct infection with the bacterium) and non-suppurative (due to an autoimmune type reaction to the bacterium). The complications and their percentages (where they are known) are as follows:
Suppurative 2%
Non-suppurative 25%
If you multiply the rate of complication (27% overall) with the percentage of patients who are S. pyogenes positive for a particular Centor score it starts to give you an idea why the Centor score might be helpful, e.g. for a Centor score of 1: 27% complication of 7% S. pyogenes gives 1-2% total complication rate.
This isn’t very good. Now I may sound like an old Microbiologist not wanting to move with the times of cheap and rapid point of care testing making “us” old dinosaur Microbiologist’s redundant…but I’m not, I actually really support this technology but in my opinion before it is rolled out fully e.g. in NICE guidelines, it does need to improve!
The alternative is to send a bacterial throat swab for culture to the microbiology laboratory (there is no sensitivity and specificity for throat swabs as they are the “gold standard”, in theory they are 100% sensitive and 100% specific but in reality they are probably not). This method may be more costly and it takes longer (up to 2 days) but will still identify S. pyogenes quickly enough to allow for successful treatment with either penicillin or macrolides.
Why use the Centor score?
So if, as I believe, the Centor score is not that good at predicting S. pyogenes pharyngitis what is its use? Well it may surprise you, but I believe its real use lies in the same reason we treat S. pyogenes pharyngitis in the first place; we don’t give antibiotics to treat pharyngitis but rather to prevent complications.
S. pyogenes pharyngitis has a high complication rate, about 27% overall. These complications can be split into suppurative (relating to direct infection with the bacterium) and non-suppurative (due to an autoimmune type reaction to the bacterium). The complications and their percentages (where they are known) are as follows:
Suppurative 2%
- Tonsillar abscess (also known as a Quinsy)
- Otitis media
- Sinusitis
- Pyomyositis
- Necrotising fasciitis
Non-suppurative 25%
- Rheumatic fever (3%)
- Reactive arthritis
- Scarlet fever (up to 10%)
- Toxic shock syndrome
- Glomerulonephritis (10%)
- PANDAS syndrome
If you multiply the rate of complication (27% overall) with the percentage of patients who are S. pyogenes positive for a particular Centor score it starts to give you an idea why the Centor score might be helpful, e.g. for a Centor score of 1: 27% complication of 7% S. pyogenes gives 1-2% total complication rate.
Centor score |
Patients who are actually S. pyogenes positive |
Complication Rate |
0 |
0% |
0% |
1 |
7% |
1-2% |
2 |
21% |
5-6% |
3 |
38% |
10% |
4 |
57% |
15% |
So a Centor score of ≥2 is associated with a complication rate of ≥5% and this is a common cut off in science for when something is worth doing or not worth doing (NB this is what a p value of 0.05 actually means in a statistical test). So a Centor score of ≥2 is a “call to action”; the Centor score dictates whether you test and treat or just treat with empirical antibiotics.
Problems with the Centor score
So what do I see as the main problem with the Centor score? Well whatever your Centor score, if the cause of your pharyngitis is S. pyogenes then you have the same complication rate of 27%. Let me put it another way. Of all the patients with a Centor score of 1, 7 patients will have a complication rate of 27% and 93 will have a complication rate of 0%; it’s not the Centor score which dictates whether you have a complication it is the presence of S. pyogenes.
To my way of thinking if you want a method to reduce the complication rate as low as possible without treating lots of patients unnecessarily, then you have to have a good method which detects the main risk factor for the complication i.e. a throat swab for S. pyogenes rather than one (Centor score) that at best only predicts 57% (Centor score of 4 and even worse at scores 2 or 3). So if the Centor score is 1 or 4 then the patient does not get tested; in these cases a score of 1 has no test and no treatment and a score of 4 has no test and empirical treatment. This actually means that 7 in 100 patients with a score of 1 would not get the treatment they required, whereas 43 out of 100 patients with a score of 4 would get unnecessary antibiotics. In an era when we are trying to reduce the use of antibiotics this would seem counter-productive.
What would I do?
Personally I don’t see the Centor score as being particularly useful. I think if S. pyogenes is part of the differential diagnosis of pharyngitis then the patient should be tested for S. pyogenes and treated if the test is positive. The current gold standard test would be a bacterial throat swab for culture but I think a rapid antigen test with a high sensitivity and specificity, say around 92-98% would be equally acceptable (there will always be better tests emerging on the market, currently there is one declaring this level of accuracy, it may post date the NICE guidelines of 70% to 90% sensitive and 95% specific). This would strike a good balance between not under treating low Centor scores, not over treating high Centor scores, and not having more costly and slower laboratory tests for S. pyogenes.
Problems with the Centor score
So what do I see as the main problem with the Centor score? Well whatever your Centor score, if the cause of your pharyngitis is S. pyogenes then you have the same complication rate of 27%. Let me put it another way. Of all the patients with a Centor score of 1, 7 patients will have a complication rate of 27% and 93 will have a complication rate of 0%; it’s not the Centor score which dictates whether you have a complication it is the presence of S. pyogenes.
To my way of thinking if you want a method to reduce the complication rate as low as possible without treating lots of patients unnecessarily, then you have to have a good method which detects the main risk factor for the complication i.e. a throat swab for S. pyogenes rather than one (Centor score) that at best only predicts 57% (Centor score of 4 and even worse at scores 2 or 3). So if the Centor score is 1 or 4 then the patient does not get tested; in these cases a score of 1 has no test and no treatment and a score of 4 has no test and empirical treatment. This actually means that 7 in 100 patients with a score of 1 would not get the treatment they required, whereas 43 out of 100 patients with a score of 4 would get unnecessary antibiotics. In an era when we are trying to reduce the use of antibiotics this would seem counter-productive.
What would I do?
Personally I don’t see the Centor score as being particularly useful. I think if S. pyogenes is part of the differential diagnosis of pharyngitis then the patient should be tested for S. pyogenes and treated if the test is positive. The current gold standard test would be a bacterial throat swab for culture but I think a rapid antigen test with a high sensitivity and specificity, say around 92-98% would be equally acceptable (there will always be better tests emerging on the market, currently there is one declaring this level of accuracy, it may post date the NICE guidelines of 70% to 90% sensitive and 95% specific). This would strike a good balance between not under treating low Centor scores, not over treating high Centor scores, and not having more costly and slower laboratory tests for S. pyogenes.
Rapid antigen test |
False positive |
False negative |
Drawback |
Negative (92% sensitive) |
N/A |
8% |
Fail to prevent 2% of complications |
Positive (98% specific) |
2% |
N/A |
Unnecessarily treat 2% of patients |
So Centor isn’t really about identifying S. pyogenes, after all who wants a test which has a false positive rate of 43%! Remember: a Centor score of 4 only detects 57%. Centor is in fact about trying to reduce complication rates as simply as possible using a bed side set of criteria whilst accepting a high amount of over treatment.
Interestingly NICE don’t even recommend the Centor score! They recommend a different system called the FeverPAIN score (developed in a UK primary care setting) which gives 1 point for each of the following:
The FeverPAIN score has the same drawbacks as the Centor score even if it is slightly better at predicting S. pyogenes infection; there are still missed positive S. pyogenes pharyngitis patients and over treatment of non S. pyogenes pharyngitis patients.
Interestingly NICE don’t even recommend the Centor score! They recommend a different system called the FeverPAIN score (developed in a UK primary care setting) which gives 1 point for each of the following:
- Fever over 38°C.
- Purulence (pharyngeal/tonsillar exudate).
- Attend rapidly (3 days or less)
- Severely Inflamed tonsils
- No cough or coryza
The FeverPAIN score has the same drawbacks as the Centor score even if it is slightly better at predicting S. pyogenes infection; there are still missed positive S. pyogenes pharyngitis patients and over treatment of non S. pyogenes pharyngitis patients.
FeverPAIN score |
Patients who are actually S. pyogenes positive |
Complication Rate |
0-1 |
13-18% |
3-5% |
2-3 |
34-40% |
9-11% |
4-5 |
62-65% |
16-18% |
So at the end of the day there are a number of scoring systems out there to “help” with diagnosing S. pyogenes pharyngitis and preventing complications by over treating. Do I think they are of value? Not really. I think good history taking will lead to the suspected diagnosis and a positive throat swab is the best predictor of the risk from complications. Remember: a positive throat swab indicates a 27% risk of complication whereas a Centor score of 4 only predicts a 15% risk of complication and a FeverPAIN score of 4-5 only predicts an 18% risk of complication. The throat swab is far better at predicting complications from S. pyogenes pharyngitis. But then I would say that because I’m a Microbiologist and we do like our tests! What do you think? Do you take a swab and wait for 48 hours for the result before issuing antibiotics if it is positive or do you not bother with the swab, go on a clinical diagnosis and give antibiotics, as it saves a repeat visit? Would a point of care rapid antigen test change what you would do?
RSS Feed