Centor, McIsaac and FeverPAIN Score Calculator

Three sore-throat scores, computed together, with each one’s own published likelihood of isolating streptococcus and the cohort it came from. Centor was derived in adults in an emergency room, FeverPAIN in UK primary care from age 5 — and applying either outside its population is the error.

Last updated 8 Oct 2026

Sore throat scores

3 instruments, 0 to 4 or 0 to 5

Which instrument is the headline score?All three totals are computed from the items below and shown together, because the point of the page is that they are different instruments rather than three names for one. This control only chooses which total gets the headline and which published likelihood bands are shown under it.

Centor / McIsaac item — feverCentor’s item is a history of fever or a temperature over 38 °C with no time limit; FeverPAIN’s item is fever in the previous 24 hours. They are not the same observation, which is why both appear separately on this page rather than being shared. NICE prints both wordings verbatim.

Centor / McIsaac item — absence of coughScored for the ABSENCE of cough, so the reassuring answer earns the point. FeverPAIN’s equivalent item is “no cough or coryza”, which is stricter: a child with a runny nose scores the Centor item and not the FeverPAIN one.

Centor / McIsaac item — tonsillar exudate or swellingMcIsaac’s published wording is tonsillar swelling; Centor’s is tonsillar exudate. FeverPAIN splits the same territory into two separate points, purulence and severe inflammation, which is one reason its maximum is 5 rather than 4.

Centor / McIsaac item — tender anterior cervical adenopathyAnterior cervical, and tender. Posterior cervical adenopathy points the other way clinically and is not this item. In young children cervical nodes are commonly palpable without being tender, and palpable is not the item.

McIsaac item — ageThis single item is the whole difference between Centor and McIsaac, and it is the reason the age-modified version reaches children at all: Centor was derived in adults in an emergency room and McIsaac’s validation ran from age 3. A total pushed below 0 by the age point is reported as 0, and 5 is reported as 4, exactly as McIsaac’s paper specifies.

FeverPAIN item — FeverNICE’s wording, verbatim: “Fever (during previous 24 hours)”. The 24-hour window is the item; a fever three days ago does not score it, although it would score the Centor fever item.

FeverPAIN item — PurulenceNICE’s wording: “Purulence (pus on tonsils)”. Scored separately from severe inflammation below, so a child with both purulent and severely inflamed tonsils scores two FeverPAIN points where Centor gives one.

FeverPAIN item — Attend rapidlyNICE’s wording: “Attend rapidly (within 3 days after onset of symptoms)”. It is the only item on any of these instruments that is about health-seeking behaviour rather than examination, and it is in the score because early attendance tracked streptococcal isolation in the derivation cohorts.

FeverPAIN item — severely Inflamed tonsilsSeverely, not simply red. This is the most subjective of the five FeverPAIN items and the one where two examiners most readily differ by a point.

FeverPAIN item — No cough or coryzaBoth must be absent. Coryza is inflammation of the nasal mucous membranes, in NICE’s own gloss. A child with a cold and a sore throat loses this point and keeps the Centor absence-of-cough point if they are not coughing, which is a real, if small, way the two instruments diverge in paediatric practice.

4pointsExample

FeverPAIN as the headline: fever in the last 24 hours (1), no pus (0), attended within 3 days (1), severely inflamed tonsils (1), no cough or coryza (1). Centor items: fever (1), no cough (1), tonsillar swelling (1), no tender nodes (0), age 3 to 14 (+1)

Three instruments, three populations

Centor = fever over 38 °C + no cough + tonsillar exudate or swelling + tender anterior cervical nodes (0 to 4)
McIsaac = Centor + age (3–14 years +1 · 15–44 years 0 · 45 and over −1), reported within 0 to 4
FeverPAIN = Fever in 24 h + Purulence + Attend within 3 days + severe Inflammation + No cough or coryza (0 to 5)

the populations are the point: Centor was derived in adults in an emergency room in 1981. McIsaac’s validation cohort was 521 patients aged 3 to 76 in a Toronto family medicine centre, which is where the age point comes from. FeverPAIN was derived in two UK general-practice cohorts of 606 and 517 patients aged 5 and over. Applying an instrument outside the population it was derived in is the error these three exist to illustrate
isolating streptococcus is not the diagnosis: every published band on this page is a likelihood of ISOLATING group A streptococcus on a swab, not a probability of streptococcal disease. Asymptomatic carriage is common in children, so a carrier with a viral sore throat gives a positive swab. The IDSA’s position, as summarised by the AAFP, is that viral and streptococcal sore throat “cannot be differentiated using clinical features alone” and that a clinical diagnosis “should be confirmed using a rapid antigen detection test and/or culture of a throat swab”
the discrimination, stated: Centor at 3 or more, pooled across 21 studies and 4,839 patients in adults over 14: sensitivity 0.49, specificity 0.82 (95% CI 0.72 to 0.88). FeverPAIN in its two derivation cohorts: area under the curve 0.73 and 0.71. Neither is a test; both are ways of deciding how much a swab would add
no prescribing decision here: NICE NG84 attaches antibiotic strategies to the FeverPAIN and Centor bands, and this page reproduces the scores and the likelihoods and not the strategies. the sensitivity and specificity calculator is where the general 2×2 arithmetic lives; the table below applies it to this instrument at two prevalences
the age point can run off both ends: McIsaac’s score is reported within 0 to 4: a raw total of 5 is reported as 4 and a raw total of −1 as 0. In the 206,870-patient validation of the age-modified score the same normalisation was applied

Worked example

FeverPAIN as the headline: fever in the last 24 hours (1), no pus (0), attended within 3 days (1), severely inflamed tonsils (1), no cough or coryza (1). Centor items: fever (1), no cough (1), tonsillar swelling (1), no tender nodes (0), age 3 to 14 (+1)
FeverPAIN: 1 + 0 + 1 + 1 + 1 = 4 of 5, in NICE's 4–5 band at a 62 to 65% likelihood of isolating streptococcus
Centor on the same child: 1 + 1 + 1 + 0 = 3 of 4, in NICE's 3–4 band at a 32 to 56% likelihood
McIsaac: 3 + 1 for age 3 to 14 = 4 of 4. In the 206,870-patient validation, 55% of patients scoring 4 on the age-modified score had group A streptococcus
Three numbers, three ranges, one child — and the ranges only partly overlap. That is not a flaw in the arithmetic; it is three instruments derived in three populations against the same swab
Take the Centor 3 as a positive test, with its pooled sensitivity of 0.49 and specificity of 0.82. At a 5% prevalence the positive predictive value is 12.5%; at 20% it is 40.5%. The published post-test probability for exactly those two prevalences is “12% to 40%”, so the hand arithmetic and the systematic review agree. A score is not a diagnosis, and a published figure describes the cohort it came from rather than the child in front of you.

The three item sets, side by side

The FeverPAIN and Centor items are NICE NG84’s own wording. Read the third row carefully: FeverPAIN gives a point for pus and a second point for severe inflammation, where Centor gives one point for exudate, so the same throat contributes differently to the two totals. Nothing here is a different name for the same instrument.
Centor (1981)	McIsaac (1998)	FeverPAIN (PRISM)
History of fever over 38 °C	Same	Fever during the previous 24 hours
Absence of cough	Same	No cough or coryza
Tonsillar exudate	Tonsillar swelling	Purulence (pus on tonsils) — and, separately, severely inflamed tonsils
Tender anterior cervical adenopathy	Same	Not an item
Not an item	Age: 3–14 +1 · 15–44 0 · 45 and over −1	Attended rapidly, within 3 days of onset
Maximum 4	Maximum 4 (reported within 0 to 4)	Maximum 5

What a Centor score of 3 or more is worth, worked through a 2×2

Worked by hand from the pooled sensitivity of 0.49 and specificity of 0.82 for a Centor score of 3 or more in adults over 14, across 21 studies and 4,839 patients. The two prevalences are the review’s own bracket — 5 to 10% in adults, 20 to 25% in children — and the resulting post-test probabilities of 12% and 40% reproduce the figures the review itself reports, which is the check that this arithmetic is the arithmetic the authors did. The lesson is the asymmetry: in an adult primary-care population most patients with a score of 3 do not have streptococcus, while a score under 3 makes it unlikely.
Per 1,000 patients	Prevalence 5% (adults)	Prevalence 20% (children)
With streptococcus	50	200
True positives (sensitivity 0.49)	24.5	98
False positives (specificity 0.82)	171	144
True negatives	779	656
False negatives	25.5	102
Positive predictive value	12.5%	40.5%
Negative predictive value	96.8%	86.5%

Two instruments, one swab, and the population they were drawn from

Centor and FeverPAIN answer the same question — how likely is it that a swab from this throat grows group A streptococcus — and they answer it with different items, different maxima and different published likelihoods, because they were derived in different people. Centor’s four items came out of an adult emergency-room population in 1981. McIsaac’s modification adds a single age point, +1 for ages 3 to 14 and −1 for 45 and over, and that one item is what extends the instrument to children at all; its validation cohort was 521 patients aged 3 to 76 in a Toronto family medicine centre, where the proportion with streptococcus ran from 2.5% at a score of 0 to 52.8% at 4. FeverPAIN was built in UK general practice, in two cohorts of 606 and 517 patients aged 5 and over, with areas under the curve of 0.73 and 0.71.

The arithmetic matters more than it looks. Pooled across 21 studies and 4,839 adults, a Centor score of 3 or more has a sensitivity of 0.49 and a specificity of 0.82. Put that against a 5% prevalence — the review’s own figure for adults in primary care — and 1,000 patients give 24.5 true positives against 171 false ones: a positive predictive value of 12.5%. At a 20% prevalence, which is the lower end of the paediatric range, the same instrument reaches 40.5%. Nothing about the score changed; only the population did. That is the single most useful thing to know about any of these instruments, and it is the reason the same score prompts a different action in a GP surgery and in a children’s ward.

Two cautions specific to children. First, every band on this page is a likelihood of isolating streptococcus, and asymptomatic carriage is common in childhood, so a positive swab in a carrier with a viral sore throat is indistinguishable from an infection. Second, the IDSA’s guidance, as summarised by the AAFP, is that viral and streptococcal sore throat “cannot be differentiated using clinical features alone”, that a clinical diagnosis “should be confirmed using a rapid antigen detection test and/or culture of a throat swab”, and that testing “is generally not recommended in children younger than three years” — which is below the floor of every instrument here.

NICE NG84 attaches antibiotic strategies to the FeverPAIN and Centor bands. This page gives the scores and the published likelihoods and names the guideline; it renders no prescribing decision, and the strategies belong in NG84 and in local antimicrobial guidance rather than under a calculator. A score is not a diagnosis, and a published figure describes the cohort it came from rather than the child in front of you. A single value is a snapshot of one examination: a child whose illness is evolving can score differently an hour later, so a reassuring score does not exclude deterioration, and the trend across serial assessments by the same examiner carries more than any one reading. Published item wordings and severity bands differ between sources and between units, so the reader’s own unit’s chart and protocol take precedence over any banding printed here.

Frequently asked questions

What is the difference between the Centor and FeverPAIN scores?

Different items, different maxima and different derivation populations. Centor has four items and a maximum of 4, derived in adults in an emergency room; FeverPAIN has five and a maximum of 5, derived in UK general practice in patients aged 5 and over. FeverPAIN scores purulence and severe inflammation separately, counts only fever in the last 24 hours, requires no cough and no coryza, and includes attending within three days of onset, which Centor does not.

Can the Centor score be used in children?

Centor’s own cohort was adults, and the pooled sensitivity and specificity quoted for a score of 3 or more come from studies in patients over 14. McIsaac’s age-modified version is the one validated from age 3, and FeverPAIN from age 5. Below three years, IDSA guidance is that testing is generally not recommended at all.

What does a FeverPAIN score of 4 or 5 mean?

NICE gives it as a 62 to 65% likelihood of isolating streptococcus, from the two PRISM primary-care cohorts. Roughly a third of patients scoring at the top of the range still do not have it, and isolating the organism is not the same as streptococcal disease, because carriage is common in children.

Why does the same Centor score mean different things in different places?

Because predictive value depends on prevalence and the score’s accuracy does not. With sensitivity 0.49 and specificity 0.82 for a score of 3 or more, the positive predictive value is 12.5% at a 5% prevalence and 40.5% at 20%. The table on this page works both through a 2×2, and the answers match the published post-test probabilities of 12% and 40%.

Which score does NICE recommend?

NICE NG84 defines both the FeverPAIN and the Centor criteria in its own terms and prints the likelihood bands reproduced here for each. This page stops at the score and the likelihood and does not reproduce the prescribing strategies NG84 attaches to them — for those, and for the age ranges the guideline covers, read NG84 itself alongside local antimicrobial guidance.

Related calculators

References

National Institute for Health and Care Excellence. Sore throat (acute): antimicrobial prescribing. NICE guideline NG84, Terms used in the guideline — the FeverPAIN and Centor items and the likelihood bands reproduced here are NICE’s own wording (accessed 10 October 2026).
Centor RM, et al. The diagnosis of strep throat in adults in the emergency room. Med Decis Making. 1981;1(3):239–246. Bibliographic details confirmed against the journal’s own listing; the paper was not opened.
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 — 521 patients aged 3 to 76 at a University of Toronto family medicine centre (full text read).
Fine AM, Nizet V, Mandl KD. Large-scale validation of the Centor and McIsaac scores to predict group A streptococcal pharyngitis. Arch Intern Med. Published online 7 May 2012. doi:10.1001/archinternmed.2012.950 — 206,870 patients aged 3 and over and 142,081 aged 15 and over (full text read).
Aalbers J, O’Brien KK, Chan W-S, et al. Predicting streptococcal pharyngitis in adults in primary care: a systematic review of the diagnostic accuracy of symptoms and signs and validation of the Centor score. BMC Med. 2011;9:67 — 21 studies, 4,839 patients; Centor 3 or more, sensitivity 0.49 and specificity 0.82 in adults over 14.
Little P, Hobbs FDR, Moore M, et al. PRImary care Streptococcal Management (PRISM) study. Health Technol Assess. 2014;18(6). doi:10.3310/hta18060 — the FeverPAIN derivation cohorts (n = 606 and n = 517, aged 5 and over), area under the curve 0.73 and 0.71.
Infectious Diseases Society of America. Clinical practice guideline for the diagnosis and management of group A streptococcal pharyngitis: 2012 update, as summarised in Am Fam Physician, 1 September 2013, page 338 (the AAFP summary is what was read): sore throat of the two kinds “cannot be differentiated using clinical features alone”, and “testing is generally not recommended in children younger than three years”.

Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/