Centor vs FeverPAIN score for sore throat

What Centor and FeverPAIN each count

The Centor score and the FeverPAIN score share two items, fever and the absence of cough, and differ in the rest. Centor and colleagues described four findings in 1981 from adults with sore throat in an emergency room: tonsillar exudate, swollen tender anterior cervical nodes, absence of cough and a history of fever, each worth 1 point, for 0 to 4. McIsaac later added age points so the score works in children, which is the version the Centor calculator uses. FeverPAIN, from Little and colleagues in the UK PRISM study, also scores 5 items at 1 point each: Fever in the past 24 hours, Purulence, Attending rapidly (within 3 days of onset), severely Inflamed tonsils, and No cough or coryza. So FeverPAIN replaces the neck node item with how quickly the patient presented and how inflamed the tonsils look, and its no-cough item also requires no runny nose. Its total runs from 0 to 5.

Where the FeverPAIN score came from

FeverPAIN came from two cohorts of patients aged 5 or over with acute sore throat in UK general practices, 606 in the first and 517 in the second (Little and colleagues, BMJ Open, 2013). The team looked for features that predicted growth of Lancefield group A, C or G streptococci on a throat swab, not group A alone as Centor did. A first score built in cohort 1 discriminated poorly in cohort 2, so the team kept items that held up in both: rapid attendance, fever in the last 24 hours and severely inflamed tonsils were significant in both cohorts, and purulence and no cough or coryza in at least one. FeverPAIN's area under the ROC curve was 0.73 in cohort 1 and 0.71 in cohort 2. A score of 1 or less identified 38% and 36% of patients, with streptococci in 13% and 18%. A Centor score of 1 or less identified fewer patients, 23% and 26%, with streptococci in 10% and 28%.

What the PRISM trial showed

The PRISM randomized trial, published by Little and colleagues in the BMJ in 2013, tested whether targeting antibiotics with a clinical score works better than delayed prescribing for everyone. Patients aged 3 or over with acute sore throat in UK primary care were randomized to delayed antibiotics, targeted use by clinical score, or a rapid antigen test used according to the score. Partway through, an earlier score was replaced by FeverPAIN, used in 631 patients. With FeverPAIN, antibiotic use was 29% lower than in the delayed antibiotic group, and patients reported slightly less severe symptoms on days 2 to 4, with moderately bad symptoms resolving faster. The antigen test group had a similar fall in antibiotic use, 27%, but the authors found no clear advantage of adding the test over the score alone. There were no significant differences in complications or reconsultations. NICE included FeverPAIN in its 2018 sore throat guideline.

How NICE and IDSA use the scores

The 2018 NICE guideline on antimicrobial prescribing for acute sore throat (NG84) accepts either FeverPAIN or Centor, with different bands for each. With FeverPAIN, a score of 0 or 1 means no antibiotic, 2 or 3 means considering no antibiotic or a back-up prescription, and 4 or 5 means an immediate antibiotic or a back-up prescription. With Centor, 0 to 2 means no antibiotic and 3 or 4 means an immediate or back-up prescription, so Centor has no middle band. NICE uses the original Centor criteria, not the McIsaac version with age points. The 2012 Infectious Diseases Society of America (IDSA) guideline takes a different route: it recommends confirming group A streptococcus with a rapid antigen test or culture rather than clinical features alone, and advises against testing when features such as cough, runny nose or hoarseness suggest a virus. Under IDSA, clinical features decide who is tested; under NICE, the score decides who is offered a prescription.

Worked example: a patient the scores split on

Because they count different features, the two scores can put the same patient in different NICE bands. Take a 30-year-old who has had a fever for a day, attends on the second day of symptoms, and has purulent, severely inflamed tonsils, but also a cough, and no tender neck nodes. FeverPAIN gives 1 for fever, 1 for purulence, 1 for attending within 3 days and 1 for severely inflamed tonsils, with nothing for the cough item: a score of 4, which in NG84 points to an immediate or back-up antibiotic. Centor gives 1 for fever and 1 for exudate, with nothing for the nodes or the cough item: a score of 2, which in NG84 means no antibiotic. The McIsaac score is also 2, because age 30 adds no points, and in McIsaac's study a score of 2 had an 11.2% rate of positive cultures. When the scores disagree, the decision rests on the score your guideline names and on clinical judgment.

You can check the Centor and McIsaac result in the Centor score calculator.