The sore throat dilemma: virus, bacterium, or guesswork
Sore throat is one of the commonest reasons people consult a doctor, and it is also one of the commonest reasons antibiotics are prescribed unnecessarily. The vast majority of sore throats are viral: rhinovirus, adenovirus, and other respiratory viruses cause pharyngitis that no antibiotic can touch, and the right treatment is rest, fluids, and simple analgesia. A minority, however, are caused by group A streptococcus, Streptococcus pyogenes, and that minority matters. Strep throat responds to antibiotics, and untreated strep can lead to rheumatic fever, a serious autoimmune complication that damages the heart valves. The clinician’s task is to tell the two apart at the bedside, and the throat itself rarely cooperates: viral and bacterial pharyngitis can look identical on examination.
This is a classic problem of probability. In an unselected population of people with sore throats, roughly 5 to 15 percent of adults and 20 to 30 percent of children actually have group A strep, but those averages hide enormous variation between individual patients. A 25-year-old with a runny nose, a barking cough, and a normal-looking throat has a strep probability near zero, while a 10-year-old with a high fever, pus on the tonsils, tender neck glands, and no cough at all has a probability well over half. Treating every sore throat with antibiotics would expose millions of people to side effects and drive antimicrobial resistance; treating none would miss the cases that genuinely need treatment. The Centor score sits exactly in that gap, converting five bedside observations into a probability that guides a rational testing and treatment decision.
What the Centor score is
The original score was developed by Robert Centor and colleagues and published in 1981 in Medical Decision Making (Centor RM, Witherspoon JM, Dalton HP, et al. The diagnosis of strep throat in adults in the emergency room. Med Decis Making. 1981;1(3):239-246). Working with adults in an emergency department, the team identified four clinical features that predicted a positive throat culture for group A streptococcus: a history of fever, tonsillar exudates, tender anterior cervical adenopathy, and the absence of cough. Each feature scores one point, giving a total from 0 to 4.
In 1998, Warren McIsaac and colleagues published a modified version validated in primary care that added a fifth element: an age adjustment (McIsaac WJ, White D, Tannenbaum D, Low DE. A clinical score to reduce unnecessary antibiotic use in patients with sore throat. CMAJ. 1998;158(1):75-83). Children aged 3 to 14 score an extra point, adults aged 15 to 44 score no adjustment, and patients aged 45 and over lose a point. The age adjustment reflects epidemiology rather than clinical impression: strep throat peaks in school-age children and becomes genuinely uncommon in middle-aged and older adults. The modified score therefore ranges from -1 to 5, and it is this McIsaac version that this calculator implements, since it is the version most guidelines and most clinicians actually use.
The score then received one of the largest validations any clinical prediction rule has ever had. Fine and colleagues applied both the Centor and McIsaac scores to a very large national dataset of patients seen in retail clinics for sore throat and reported the observed prevalence of group A strep at each score level (Fine AM, Nizet V, Mandl KD. Large-scale validation of the Centor and McIsaac scores to predict group A streptococcal pharyngitis. Arch Intern Med. 2012;172(11):847-852). The probabilities this calculator reports, from under 8% at a score of -1 to over 55% at a score of 5, come from that large-scale validation.
The four clinical criteria explained
- Fever over 38°C: +1 point. A documented or reliably reported fever above 38 degrees Celsius (100.4 Fahrenheit). Fever reflects a systemic inflammatory response and is more characteristic of bacterial than viral pharyngitis, though plenty of viral sore throats also cause fever, so it is a modest predictor on its own.
- Tonsillar exudates or swelling: +1 point. White or yellow pus on the tonsils, or tonsils that are markedly swollen and inflamed. Exudates are the finding most laypeople associate with strep, and they do raise the probability, but they are not specific: Epstein-Barr virus and other viral infections can produce exudates too.
- Tender anterior cervical adenopathy: +1 point. Tender, swollen lymph nodes palpable at the front of the neck, typically under the angle of the jaw. The tenderness matters: non-tender nodes are less informative. These nodes drain the tonsils, so they enlarge when the tonsils are under bacterial attack.
- Absence of cough: +1 point. The patient has no cough. This is the negative predictor in the set: cough is a hallmark of viral upper respiratory infection, so its absence makes a viral cause less likely and strep relatively more likely. It is scored by answering Yes to absence of cough.
The age adjustment explained
The age adjustment is what turns the original Centor score into the McIsaac score, and it deserves separate attention because it is the item most often applied incorrectly. Patients aged 3 to 14 get +1 point, because school-age children have the highest incidence of group A streptococcal pharyngitis. Patients aged 15 to 44 get no adjustment, reflecting the intermediate adult risk. Patients aged 45 and over get -1 point, because strep throat becomes genuinely uncommon in this age group and most sore throats in older adults are viral.
The practical effect is substantial. A 50-year-old with fever, exudates, tender nodes and no cough scores 4 on the clinical items but 3 after the age adjustment, moving from the highest treatment band down into the middle one. That is intentional: the same clinical picture means something different at different ages. Note also the lower bound: the score is not validated in children under 3, in whom group A strep pharyngitis is rare and rheumatic fever essentially does not occur, so do not apply it to toddlers.
How to use this calculator
Enter the patient’s age and answer the four clinical questions Yes or No. The calculator needs every answer: an incomplete Centor score is misleading, because each missing item could swing the total by a point and move the patient across a management threshold. Examine the throat with a good light, palpate the anterior cervical nodes deliberately rather than asking the patient whether their neck is sore, and establish the fever history carefully, since antipyretics taken an hour ago can mask it.
Press Calculate to see the total score from -1 to 5, the items that contributed, the probability of group A streptococcal infection from the large-scale validation, and the recommended strategy: neither test nor treat, test and treat only if positive, or test with consideration of empiric antibiotics. The result cards are written for documentation and for explaining the reasoning to the patient.
Interpreting the result: scores -1 to 0
A score of -1 carries a strep probability under 8%, and a score of 0 carries about 8%. At these levels the recommendation is clear: neither test nor treat. Ordering a rapid strep test or throat culture in a patient with an 8% pretest probability is more likely to cause harm than good, because the false positive rate of testing at such low prevalence starts to rival the true positive rate, and a positive result may simply reflect asymptomatic carriage rather than active infection. Up to a fifth of school-age children carry group A streptococcus in the throat without being ill, so testing low-risk patients manufactures apparent strep cases out of carriers.
These patients need symptomatic care: paracetamol or ibuprofen, fluids, rest, and reassurance with a safety net. Explain plainly that antibiotics would not help a viral sore throat and would add side effects, and give clear return advice for worsening pain, difficulty swallowing fluids, drooling, or a muffled hot-potato voice, which can signal peritonsillar abscess.
Interpreting the result: scores 1 to 2
A score of 1 carries a strep probability of about 14%, and a score of 2 carries about 23%. This is the genuine zone of uncertainty, and the recommendation is to test, and prescribe antibiotics only if the test is positive. A rapid antigen detection test gives an answer within minutes; a throat culture is the more sensitive reference standard but takes 24 to 48 hours. A positive test in this band is actionable: treat with penicillin or amoxicillin as first line per local guidelines. A negative rapid test in a child or adolescent, where the stakes of missing strep are higher, is often followed by a confirmatory throat culture, because rapid tests miss roughly 10 to 15 percent of true cases.
This middle band is where the score earns its keep in antimicrobial stewardship. Without the score, many of these patients would receive antibiotics on clinical impression alone; with the score, roughly three-quarters of them can be shown a negative test and spared the prescription. In settings where testing is unavailable, a delayed or back-up antibiotic prescription, to be collected only if symptoms fail to improve within a few days, is the pragmatic alternative endorsed by several guidelines.
Interpreting the result: scores 3 to 5
A score of 3 carries a strep probability of about 37%, a score of 4 about 55%, and a score of 5 over 55%. The recommendation is to test, and consider empiric antibiotics while awaiting the result, following local guidelines. Even at the top of the scale, nearly half of patients do not have strep, which is why testing is still recommended rather than treating blindly: a rapid test takes minutes and can confirm the diagnosis before the first dose. Where the clinical picture is compelling and the patient is a child at higher risk of complications, starting antibiotics at once while the throat culture incubates is reasonable, with the option to stop if the culture comes back negative.
The main reason for treating confirmed strep, beyond shortening symptoms by roughly half a day to a day, is preventing complications. Rheumatic fever is the complication that shaped a century of sore throat management: it follows untreated group A streptococcal pharyngitis after a latent period of weeks and can permanently damage the heart valves. It is now rare in high-income countries but remains a real threat in many parts of the world and in crowded settings. Antibiotics also reduce the risk of peritonsillar abscess, otitis media, and sinusitis, and they shorten the period of infectivity, which matters for return to school or work.
Accuracy and evidence
The score’s calibration is unusually well documented thanks to the Fine 2012 large-scale validation, which reported observed strep prevalence at every score level in a very large real-world dataset: under 8% at -1, 8% at 0, 14% at 1, 23% at 2, 37% at 3, 55% at 4, and over 55% at 5. The monotonic rise is exactly what a good prediction rule should show, and the absolute numbers give clinicians honest probabilities to share with patients instead of vague impressions.
Discrimination, the ability to separate strep from non-strep, is moderate rather than excellent, with areas under the curve typically around 0.75 to 0.80 in validation studies. That is why the score is used as a testing and treatment threshold tool rather than a diagnostic test: it sorts patients into sensible management bands instead of declaring who has strep. The McIsaac age modification improved performance over the original Centor score in primary care populations, which is why it became the standard version.
It is worth comparing with FeverPAIN, the main alternative score used in UK primary care, which uses five items (fever, purulence, attend rapidly, inflamed tonsils, no cough or coryza) and was derived to predict streptococcal isolation specifically. The two scores perform similarly, and guidelines tend to recommend whichever the local system has adopted; the management logic of test-and-treat thresholds is the same.
Limitations
- It predicts probability, not the diagnosis. Even at a score of 5, nearly half of patients do not have group A strep. The score guides testing; it does not replace it where the result would change management.
- Not validated under age 3. Group A streptococcal pharyngitis is rare in toddlers and rheumatic fever does not occur at that age, so the score should not be applied to children under 3.
- Carriage confounds testing. Up to 20% of school-age children carry group A streptococcus asymptomatically. A positive test in a low-score patient may reflect carriage with a viral pharyngitis, not true strep throat.
- It says nothing about other causes. Glandular fever (Epstein-Barr virus), peritonsillar abscess, and other conditions need their own assessment. Atypical features such as a rash, prolonged fatigue, or unilateral severe pain should prompt a broader differential.
- Local epidemiology and guidelines vary. Rheumatic fever risk, antibiotic resistance patterns, and the availability of rapid testing differ between settings, so the score’s management bands should always be applied through local guidance.
- No score replaces clinical judgment. A toxic-looking patient, stridor, drooling, or inability to swallow fluids needs urgent assessment whatever the score says.