About the ATLAS Score for Clostridioides difficile Infection
Five bedside variables, maximum 10. Age scores 0 under 60, 1 from 60 to 79, 2 at 80 or over. Systemic antibiotics continued during CDI treatment scores 2. Leukocytes score 0 below 16, 1 from 16 to 25, and 2 above 25 ×10⁹/L. Albumin scores 0 above 35 g/L, 1 from 26 to 35, and 2 at 25 or below. Creatinine scores 0 at 120 µmol/L or below, 1 from 121 to 179, and 2 at 180 or above. The derivation paper gives a linear relationship for response: cure rate ≈ 100 − 5.08 × score. It predicts response to treatment — not mortality, not recurrence, and not severity.
Formula
A Age < 60 -> 0
60-79 -> 1
>= 80 -> 2
T Treatment with systemic no -> 0
antibiotics during CDI yes (>=1 day) -> 2
L Leukocyte count < 16 -> 0
(x10^9/L) 16-25 -> 1
> 25 -> 2
A Albumin (g/L) > 35 -> 0
26-35 -> 1
<= 25 -> 2
S Serum creatinine <= 120 -> 0
(umol/L) 121-179 -> 1
>= 180 -> 2
Total 0-10
Predicted cure rate = 100 - 5.08 x score- Treatment with systemic antibiotics
- Worth 2 points and the only modifiable item. Stopping or narrowing concomitant antibiotics is the single intervention the score directly argues for.
- cure rate = 100 − 5.08 × score
- A linear fit from the derivation cohort, giving 100% at 0 and approximately 49% at 10. It is a population-level relationship, not an individual probability.
- Albumin at exactly 35 g/L
- Falls in the 1-point band, since the 0-point band requires a value strictly above 35. A common off-by-one.
- The acronym is the score: Age, Treatment with systemic antibiotics, Leukocyte count, Albumin, Serum creatinine.
- It predicts response to therapy. It is not a severity score and does not predict mortality, recurrence, colectomy or ICU admission.
- Creatinine bands are published in µmol/L; conventional-unit equivalents are derived from those figures.
- Fulminant features — hypotension, ileus, megacolon — demand escalation irrespective of the score.
Interpreting the result
Start with the antibiotics item. If it is contributing two points, ask whether the concomitant systemic antibiotics can be stopped, narrowed or shortened — often they cannot, but the question should be asked deliberately rather than by default, and it is the only part of the score that responds to anything you do. A high total predicts poorer response to standard therapy, which should prompt review of the treatment choice and early involvement of infection specialists rather than waiting for failure to declare itself. Treat the predicted cure rate as a population-level expectation rather than an individual probability: the linear fit is a description of the derivation cohort, and a score of 6 does not mean this patient has a 69.5% chance of cure. Most importantly, do not read ATLAS as a severity score, because it is not one and was never validated as one. Hypotension, ileus and megacolon define fulminant infection and demand escalation — including early surgical involvement — irrespective of what ATLAS returns, and a patient can look unremarkable on these five variables while being in serious trouble. Alongside this, the standard work continues: isolate, review the antibiotic history, and follow current guideline treatment, which now favours fidaxomicin or vancomycin over metronidazole in most settings.
| Score | Band | What it means | Action |
|---|---|---|---|
| 0–3 | Low score | Predicted cure approximately 85% to 100% | Standard therapy; still review whether concomitant antibiotics can be stopped |
| 4–6 | Intermediate score | Predicted cure approximately 70% to 80% | Review treatment choice; reassess clinically rather than relying on the number |
| 7–10 | High score | Predicted cure approximately 49% to 65% | Poor predicted response — review the agent and involve infection specialists early |
What the ATLAS Score (C. difficile) needs (5 inputs)
- Age
- Under 60 scores 0, 60 to 79 scores 1, and 80 or over scores 2.
- Treatment with systemic antibiotics during CDI therapy
- One day or more of concomitant systemic antibiotics scores 2. The only modifiable item in the score, and it carries as much weight as the most extreme band of any other variable.
- Leukocyte count
- Below 16 ×10⁹/L scores 0, 16 to 25 scores 1, and above 25 scores 2 — that is below 16,000, 16,000 to 25,000, and above 25,000 cells/µL.
- Albumin
- Above 35 g/L scores 0, 26 to 35 scores 1, and 25 or below scores 2. Note that exactly 35 falls in the 1-point band.
- Serum creatinine
- 120 µmol/L or below scores 0, 121 to 179 scores 1, and 180 or above scores 2. The original is in µmol/L, so conventional units are derived from it rather than the reverse.
What it returns
- Total score, 0 to 10
- Five items, each worth up to 2 points.
- Predicted cure rate
- From the derivation paper's linear relationship, cure rate ≈ 100 − 5.08 × score, giving 100% at a score of 0 and about 49% at 10.
- The item breakdown
- Shown so the antibiotics contribution is visible, since it is the one item that can be acted on.
How it is calculated
The five variables split into three groups doing different work. Age and albumin describe the host: physiological reserve and, in albumin's case, both nutritional depletion and the protein-losing enteropathy that severe colitis itself causes — which is why albumin sits in so many gastrointestinal scores and why it is as much a consequence as a cause here. Leukocyte count and creatinine describe the current insult: the systemic inflammatory response and the renal effect of volume depletion, which together capture how hard the infection is hitting. The antibiotics item is different in kind from all four, and that is the point. It describes not the patient but a decision that has been made about the patient — whether the selective pressure that permitted C. difficile to expand is still being applied. Keeping that in a bedside score, weighted equal to the worst band of any physiological variable, makes it structurally impossible to calculate the score without confronting the question. Everything else in ATLAS is observation; that one item is an intervention.
Facts & figures
| Letter | Variable | 0 points | 1 point | 2 points |
|---|---|---|---|---|
| A | Age (years) | Under 60 | 60–79 | 80 or over |
| T | Treatment with systemic antibiotics | No | — | Yes (1 day or more) |
| L | Leukocytes (×10⁹/L) | Under 16 | 16–25 | Above 25 |
| A | Albumin (g/L) | Above 35 | 26–35 | 25 or below |
| S | Serum creatinine (µmol/L) | 120 or below | 121–179 | 180 or above |
The antibiotics item has no 1-point band — it is 0 or 2. Maximum total 10.
| Score | Predicted cure rate |
|---|---|
| 0 | 100% |
| 2 | 89.8% |
| 4 | 79.7% |
| 6 | 69.5% |
| 8 | 59.4% |
| 10 | 49.2% |
From cure rate = 100 − 5.08 × score. A population-level linear fit, not an individual probability.
Evidence
Derivation and validation — Miller and colleagues, 2013
2013A simple clinical bedside score derived and validated to predict response to therapy in Clostridium difficile infection, published in BMC Infectious Diseases in 2013, using data from clinical trial populations.
Produced a linear relationship between the score and treatment response, cure rate = 100 − 5.08 × ATLAS score, using five variables all available at the bedside without imaging or stool quantification.
Guideline context — ACG 2021
2021ACG clinical guidelines on the prevention, diagnosis and treatment of Clostridioides difficile infection.
Defines severity and fulminant disease on clinical and laboratory grounds independently of ATLAS, and sets out current treatment preferences — a reminder that ATLAS predicts response rather than defining severity.
Guideline context — IDSA/SHEA 2021 focused update
2021IDSA and SHEA focused update to the 2017 guideline on Clostridioides difficile infection in adults.
Moved treatment preference towards fidaxomicin, with vancomycin as an acceptable alternative — changing the standard therapy against which any predicted cure rate should now be read.
How it compares
ATLAS Score (C. difficile) vs Edinburgh Gastric Ulcer Score
Both are deliberately minimal bedside instruments built for one narrow decision — and both are easy to over-read.
EGUS uses three variables to decide whether a gastric ulcer needs a repeat endoscopy; ATLAS uses five to predict response to CDI therapy. Neither is a severity score, neither predicts mortality, and both have a specific validated purpose that is narrower than their apparent scope. The shared pitfall is the same: quoting the headline figure outside the decision it was built for. EGUS's negative predictive value is conditional on biopsy adequacy and macroscopic appearance; ATLAS's cure rate is a population-level fit that says nothing about whether this patient is developing megacolon.
ATLAS Score (C. difficile) vs Truelove and Witts criteria
Two colitides that look alike on the ward — and the scores are not interchangeable in either direction.
Both conditions present with frequent bloody or non-bloody diarrhoea, systemic upset and a raised inflammatory response, and C. difficile is a recognised precipitant of a flare in inflammatory bowel disease as well as a mimic of one. Truelove and Witts defines acute severe ulcerative colitis and triggers intravenous steroids; ATLAS predicts response to antimicrobial therapy in CDI. Getting the diagnosis the wrong way round is consequential, since steroids for undiagnosed C. difficile and antibiotics alone for a severe ulcerative colitis flare are both harmful. Testing for C. difficile in any acute colitis is the practical resolution.
ATLAS Score (C. difficile) vs Oakland score
Both are bedside scores in acute lower gastrointestinal presentations, built for opposite ends of the risk spectrum.
Oakland identifies patients with acute lower gastrointestinal bleeding at low enough risk for safe discharge, using seven variables. ATLAS predicts treatment response in C. difficile infection. The comparison is worth drawing because both are attempts to make an early ward-level decision on data that is already available, and both share the limitation that a reassuring score does not exclude a patient deteriorating for a reason the score does not measure.
Pearls & pitfalls
- ATLAS predicts response to treatment. It is not a severity score and does not predict mortality, recurrence or colectomy.
- The antibiotics item is the only modifiable one and is worth 2 points — calculating the score should trigger a review of the antibiotic chart.
- Albumin of exactly 35 g/L scores 1, not 0, since the 0-point band requires strictly above 35.
- Creatinine of exactly 120 µmol/L still scores 0; 121 is the first 1-point value.
- Leukocytes of exactly 16 and exactly 25 ×10⁹/L both sit in the 1-point band.
- The antibiotics item has no 1-point band — it is 0 or 2.
- Creatinine bands are published in µmol/L; conventional-unit equivalents are derived, not original.
- The predicted cure rate is a population-level linear fit, not an individual probability.
- Fulminant features — hypotension, ileus, megacolon — demand escalation whatever the score shows.
- Treatment has moved on since derivation: fidaxomicin and vancomycin are now preferred over metronidazole, so the absolute cure figures should be read with that in mind.
Critical actions
- Review the antibiotic chart and ask explicitly whether concomitant systemic antibiotics can be stopped or narrowed.
- Record the leukocyte count, albumin and creatinine from the same day the score is calculated.
- Assess separately for fulminant disease — hypotension, ileus, megacolon — which the score does not capture.
- Follow current guideline therapy, which favours fidaxomicin or vancomycin over metronidazole.
- Institute contact precautions and review environmental cleaning.
- Involve infection specialists early where the score is high or the patient is not responding.
- Involve surgery early if fulminant colitis is suspected, irrespective of the score.
- Reassess response clinically rather than recalculating the score, which was validated at the start of treatment.
Why this score exists
Putting a treatment decision inside a prognostic score is unusual, and it is the most interesting thing about ATLAS. Four of the five variables are things you observe about a patient: how old they are, how high their white count is, what their kidneys and albumin are doing. The fifth is something you are doing to them. Concomitant systemic antibiotics score two points — as much as being over 80, as much as a creatinine above 180 — and unlike every other item, that score can be changed this afternoon. There is a quiet argument embedded in that weighting. A score built purely from host and disease variables tells you what is likely to happen; including the antibiotics item tells you that part of what is likely to happen is a consequence of a choice still being made. It also means the calculation cannot be completed without someone reviewing the antibiotic chart, which for a condition caused by antibiotics is a useful piece of enforced procedure. The limitation follows from the same design: because it predicts response rather than severity, a patient can score low and still be developing toxic megacolon, and ATLAS will not say so.
About the creator
First author; infectious diseases and Clostridioides difficile research
Led the derivation and validation of the ATLAS bedside score.
Co-author; Clostridioides difficile clinical trials
Contributed the trial cohorts from which the score was derived.
Co-author; infectious diseases
Contributed to the derivation and validation work.
Limitations
- Derived from clinical trial populations, which are healthier and more selected than the patients who typically develop severe C. difficile infection in hospital.
- Predicts response to therapy only — not mortality, recurrence, colectomy or ICU admission.
- It is explicitly not a severity score, so a patient with fulminant colitis can score low.
- Treatment has changed substantially since 2013, with fidaxomicin and vancomycin now preferred over metronidazole, so the absolute cure figures are dated.
- The linear cure relationship is a population-level fit and overstates the precision available for an individual patient.
- Albumin is affected by many things in acute illness and is not specific to CDI severity.
- The antibiotics item is binary and takes no account of the number, spectrum or duration of concomitant agents.
- Recurrence, which is the dominant clinical problem in C. difficile infection, is not addressed at all.
If you are the patient
Clostridioides difficile, usually shortened to C. diff, is a gut infection that most often follows a course of antibiotics. The antibiotics clear away the normal bacteria in the bowel, which lets C. diff grow in their place and cause diarrhoea. The ATLAS score is a quick way for the team to estimate how likely the standard treatment is to clear it. It adds up five things, all already known or on routine blood tests: your age, whether you are still taking other antibiotics for something else, your white blood cell count, a blood protein called albumin, and a kidney blood test called creatinine. Each is worth up to two points, so the total runs from 0 to 10. A lower total means treatment is more likely to work first time. The most useful part is the antibiotics question. If you are still on antibiotics for a separate infection, that makes the C. diff harder to clear — so the team will look at whether those can be stopped, changed or shortened. Sometimes they genuinely cannot, but it is always worth asking, and it is the one part of the score that can actually be changed. One thing to be clear about: this score estimates how well treatment is likely to work. It is not a measure of how severe the infection is. If you become very unwell — a very swollen or painful tummy, a low blood pressure, or your bowels stopping altogether — that needs urgent attention regardless of what this score says, and your team will act on how you are rather than on the number.
Frequently asked questions
What is the ATLAS score?#
A five-item bedside score predicting response to treatment in Clostridioides difficile infection. The acronym is the score: Age, Treatment with systemic antibiotics, Leukocyte count, Albumin, Serum creatinine. Each contributes up to 2 points, for a maximum of 10, and the derivation paper gives cure rate ≈ 100 − 5.08 × score.
Is ATLAS a severity score?#
No, and this is the most important thing to understand about it. It predicts response to therapy and was never validated as a measure of severity. A patient can score low and still be developing fulminant colitis. Hypotension, ileus and megacolon define fulminant disease and demand escalation irrespective of the score.
Why does the antibiotics item carry 2 points?#
Because continuing systemic antibiotics through CDI treatment maintains the selective pressure that allowed C. difficile to expand, and is consistently associated with poorer response. It is also the only modifiable item in the score — weighting it equal to being over 80 or having a creatinine above 180 makes it impossible to calculate the score without reviewing the antibiotic chart.
How is the predicted cure rate calculated?#
From the derivation paper's linear relationship: cure rate = 100 − 5.08 × ATLAS score. That gives 100% at a score of 0 and approximately 49% at 10. It is a population-level fit describing the derivation cohort, not an individual probability, and it should be read as an expectation rather than a prediction about one patient.
What are the easy scoring mistakes?#
Albumin of exactly 35 g/L scores 1 rather than 0, because the 0-point band requires strictly above 35. Creatinine of exactly 120 µmol/L still scores 0, with 121 the first 1-point value. Leukocytes of exactly 16 and exactly 25 ×10⁹/L both fall in the 1-point band. And the antibiotics item has no 1-point band at all — it is 0 or 2.
Are the absolute cure figures still accurate?#
They should be read with caution. The score was derived in 2013, and treatment has since moved towards fidaxomicin and vancomycin in preference to metronidazole. The relative ordering the score produces is likely to hold, but the absolute cure rate attached to any given total reflects the therapy of its era.
Does ATLAS predict recurrence?#
No. It addresses initial response to therapy only. Recurrence is arguably the dominant clinical problem in C. difficile infection — driving decisions about fidaxomicin, bezlotoxumab and faecal microbiota transplantation — and none of that is informed by this score.
What should be done alongside calculating the score?#
Review and if possible stop or narrow concomitant systemic antibiotics; institute contact precautions; follow current guideline therapy, which favours fidaxomicin or vancomycin; assess separately for fulminant features; and involve infection specialists early where the score is high or the patient is not responding. Surgical involvement should be early if fulminant colitis is suspected, whatever the score.
References
Original / primary reference
Clinical practice guidelines
- Kelly CR, Fischer M, Allegretti JR, LaPlante K, Stewart DB, Limketkai BN, Stollman NH. ACG Clinical Guidelines: Prevention, Diagnosis, and Treatment of Clostridioides difficile Infections. Am J Gastroenterol. 2021;116(6):1124-1147.
- Johnson S, Lavergne V, Skinner AM, Gonzales-Luna AJ, Garey KW, Kelly CP, Wilcox MH. Clinical Practice Guideline by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA): 2021 Focused Update Guidelines on Management of Clostridioides difficile Infection in Adults. Clin Infect Dis. 2021;73(5):e1029-e1044.