About the Travis (Oxford) Criteria for Acute Severe Ulcerative Colitis
Applied on day 3 of intravenous corticosteroids: more than eight stools that day, or a stool frequency between three and eight together with a CRP above 45 mg/L. Either arm predicts roughly an 85% chance of colectomy during the same admission. Note the structure, because it is widely misquoted — the CRP threshold applies only within the three-to-eight stool window. The paper's own Conclusions paraphrase it loosely as 'more than 8 stools or CRP above 45', dropping that condition, and most secondary sources reproduce the loose version. The two disagree for a patient who is responding, with fewer than three stools but a CRP that has not yet fallen.
Formula
On day 3 of intravenous corticosteroids:
stool frequency > 8 per day
OR
stool frequency 3-8 per day AND CRP > 45 mg/L
-> approximately 85% required colectomy on that admission
WIDELY REPRODUCED LOOSER FORM (from the paper's Conclusions):
stool frequency > 8 per day OR CRP > 45 mg/L
The two diverge when stool frequency is under 3 with a CRP above 45.- More than 8 stools
- Sufficient on its own, whatever the CRP. A patient still passing nine or more stools on day 3 of intravenous steroids has not responded.
- 3 to 8 stools with CRP above 45 mg/L
- The combined arm. Both conditions are required — the CRP threshold is not a free-standing criterion, though it is very commonly quoted as one.
- Day 3
- Not day 2 and not day 5. The criteria were derived at this specific point, and applying them earlier or later is extrapolation.
- Derived in 51 episodes of severe colitis defined by Truelove and Witts criteria, all treated with intravenous and rectal hydrocortisone.
- The paper also defines a day 7 checkpoint: more than 3 stools daily with visible blood carries a 60% chance of continuing symptoms and a 40% chance of colectomy over the following months.
- Ciclosporin was used in 14 of the 51 episodes, and the modern rescue-therapy landscape differs substantially from 1996.
- The criteria predict colectomy on that admission — they do not indicate it.
Interpreting the result
A positive result should change what happens that day, not that week. Involve the colorectal surgical team while the patient is still relatively well, rather than as a rescue when medical therapy has visibly failed — the outcome of colectomy in acute severe colitis depends heavily on how deconditioned the patient has become by the time it happens. Make an active decision about rescue therapy with infliximab or ciclosporin rather than deferring it. Before attributing failure to refractory colitis, exclude Clostridioides difficile and cytomegalovirus superinfection, both of which reproduce steroid non-response and are treated differently. Check that thromboprophylaxis is prescribed: acute severe colitis is strongly prothrombotic, and bloody diarrhoea is not a contraindication — this is among the most common omissions on these admissions. A negative result is not a discharge decision and does not mean the admission is going well; continue daily review and reassess formally at day 7, where the paper's second finding applies. Finally, be deliberate about which version of the rule is being applied, since the looser paraphrase in wide circulation will flag some responding patients that the operational rule does not.
| Score | Band | What it means | Action |
|---|---|---|---|
| More than 8 stools on day 3 | High risk of colectomy | Approximately 85% required colectomy on that admission | Involve the surgical team today; decide on rescue therapy; exclude C. difficile and CMV |
| 3–8 stools with CRP above 45 mg/L | High risk of colectomy | The combined arm — same prediction, arising in a more ambiguous clinical picture | Same actions; the intermediate stool count makes early surgical involvement more valuable, not less |
| Criteria not met | Not high risk | Does not predict a favourable admission | Continue intravenous steroids with daily review; reassess formally at day 7 |
What the Travis (Oxford) Criteria needs (2 inputs)
- Stool frequency on day 3
- The count on day 3 of intravenous steroids, not the admission figure. More than eight satisfies the criteria on its own; three to eight opens the CRP arm.
- CRP on day 3
- The threshold is 45 mg/L, equivalently 4.5 mg/dL. It applies only when the stool frequency is between three and eight — a detail lost in most reproductions of the rule.
What it returns
- High risk or not high risk
- Binary. The criteria were designed as a trigger rather than a graded score, which is both their strength and their main limitation.
- Which arm was met
- Reported explicitly, since the stool-only arm and the combined arm carry the same prediction but arise in different clinical pictures.
How it is calculated
The two variables are not arbitrary. Stool frequency is the most direct available measure of how much colon is still inflamed and failing to absorb, and it responds quickly when steroids are working — so a patient still passing nine stools on day 3 is giving a fairly unambiguous signal. CRP measures the acute-phase response driving that inflammation and, importantly, falls with a different time course, which is why it adds information rather than duplicating the stool count. The structure of the rule follows from that difference. A very high stool frequency is severe enough to stand alone. An intermediate stool frequency is genuinely ambiguous — it might represent a partial response or a plateau — and it is precisely there that a persistently raised CRP resolves the ambiguity by showing the inflammatory drive has not been switched off. That is why the CRP condition is nested inside the three-to-eight window rather than free-standing: below three stools the patient is responding and a lagging CRP means little, and above eight the stool count has already answered the question.
Facts & figures
| Source | Wording | Effect |
|---|---|---|
| Paper's Results | More than 8 stools, or 3–8 stools together with CRP above 45 mg/L | The operational rule — CRP applies only in the 3–8 window |
| Paper's Conclusions | Frequent stools (over 8/day), or raised CRP (over 45 mg/L) | Drops the stool condition on the CRP arm |
| Where they diverge | Fewer than 3 stools with CRP above 45 | Loose form flags a responding patient; the operational rule does not |
Both forms come from the same paper, which is why the looser one is so widely reproduced. This calculator implements the Results wording.
| Timepoint | Criterion | Predicted outcome |
|---|---|---|
| Day 3 | More than 8 stools, or 3–8 with CRP above 45 | Approximately 85% colectomy on that admission |
| Day 7 | More than 3 stools daily with visible blood | 60% chance of continuing symptoms; 40% chance of colectomy over following months |
The day 7 finding is often forgotten. A patient who passes the day 3 checkpoint still needs the day 7 assessment.
Evidence
Derivation — Oxford, Travis and colleagues, 1996
1996A prospective study of 51 consecutive episodes of severe ulcerative colitis defined by Truelove and Witts criteria, affecting 49 patients admitted to the John Radcliffe Hospital, monitoring 36 clinical, laboratory and radiographic variables. All episodes were treated with intravenous and rectal hydrocortisone; 14 also received ciclosporin.
Stool frequency and CRP distinguished outcomes during the first five days. On day 3, more than eight stools, or a frequency of three to eight with CRP above 45 mg/L, predicted colectomy on that admission in approximately 85%.
The population it is applied to — Truelove and Witts
1955The 1955 criteria defining acute severe ulcerative colitis, which determine who is eligible for the Travis assessment in the first place.
Remain the entry criteria for the admission; the Travis rule is a day 3 checkpoint within it rather than a diagnostic tool.
Contemporary comparator — the Ho index
2004A 2004 Edinburgh cohort of 167 consecutive patients with severe ulcerative colitis, producing a graded score from stool frequency, colonic dilatation and albumin.
Reported roughly 85% sensitivity and 75% specificity for corticosteroid failure at a threshold of 4, and adds colonic dilatation, which Travis does not consider at all.
How it compares
Travis (Oxford) Criteria vs Ho index
Same question, same day, different shape — binary versus graded, and only Ho sees colonic dilatation.
Both are day 3 predictors of steroid failure in acute severe colitis, and they use overlapping inputs. Travis is binary and uses stool frequency and CRP; Ho is a 0–9 score using stool frequency, colonic dilatation and albumin. The practical differences are that Ho gives an intermediate band where Travis forces a yes or no, and that Ho counts colonic dilatation as four points — enough to reach high risk on its own — while Travis does not consider it at all. Running both is reasonable, and where they disagree the disagreement usually locates the clinical question rather than obscuring it.
Travis (Oxford) Criteria vs Truelove and Witts criteria
Truelove and Witts gets the patient admitted; Travis decides what happens on day 3.
The two sit in sequence and are frequently confused. Truelove and Witts defines acute severe ulcerative colitis at presentation — six or more bloody stools daily plus a systemic feature — and is what triggers admission and intravenous steroids. The Travis criteria are applied three days into that treatment and ask a different question: is this working. Travis was derived specifically in patients defined by Truelove and Witts, so applying it to a less sick population is outside what it was built for.
Travis (Oxford) Criteria vs UCEIS
Endoscopic severity at one moment against clinical trajectory over three days — related but not interchangeable.
UCEIS grades what the mucosa looks like, scoring vascular pattern, bleeding and erosions, and a high score at admission does carry prognostic weight in acute severe colitis. Travis measures how the patient is behaving over the first three days of treatment. The distinction matters because endoscopic severity is fixed at the time of the procedure while the Travis inputs are dynamic, and it is the trajectory that determines whether steroids are working. In practice a severe UCEIS at admission raises the prior, and the day 3 assessment tests it.
Pearls & pitfalls
- The CRP threshold applies only when stool frequency is between three and eight. The free-standing 'CRP above 45' version is a paraphrase from the paper's Conclusions, not its analysis.
- Apply on day 3 specifically — the criteria were derived at that timepoint and using them on day 2 or day 5 is extrapolation.
- Use the day 3 stool count, not the admission figure.
- The criteria predict colectomy; they do not indicate it. The decision remains clinical and multidisciplinary.
- A negative result is not reassurance — it predicts nothing about the day 7 checkpoint, which still applies.
- Exclude C. difficile and CMV before attributing steroid failure to refractory colitis.
- Prescribe thromboprophylaxis. Acute severe colitis is strongly prothrombotic and bloody stool is not a contraindication — one of the commonest omissions on these admissions.
- Involve the colorectal team on the day the criteria are met, while the patient is still well enough for an elective-quality operation.
- CRP can be falsely low in patients already on immunosuppression, weakening the combined arm.
- The criteria say nothing about colonic dilatation; the Ho index does, and toxic megacolon needs recognising independently.
Critical actions
- Confirm the patient meets Truelove and Witts criteria for acute severe colitis before applying this rule.
- Record the stool count on day 3 of intravenous steroids specifically.
- Send stool for C. difficile toxin and consider CMV assessment if the trajectory stalls.
- Check that thromboprophylaxis is prescribed and being given.
- Involve the colorectal surgical team the day the criteria are met.
- Make an explicit decision on rescue therapy rather than deferring it to the next ward round.
- Obtain a plain abdominal radiograph to exclude colonic dilatation, which these criteria do not assess.
- Reassess at day 7 regardless of the day 3 result.
Why this score exists
It is unusual for a paper to state its own finding two different ways, and this one does. The Results section gives the operational rule — more than eight stools, or three to eight together with a CRP above 45 — while the Conclusions compress it to 'frequent stools, or raised CRP', dropping the condition that ties the two together. Both are in the same abstract. The compressed version is easier to remember and it is the one that propagated, which is why most secondary sources, and a good many calculators, implement a rule the paper's own analysis does not support. The difference is small in the population the criteria were built for, since a patient meeting Truelove and Witts is unlikely to be passing fewer than three stools on day 3 — but it is not nothing, and it points the wrong way: the looser rule flags a patient whose stool frequency has fallen below three, which is to say someone who is getting better. Encoding the Results wording costs nothing and is what the data actually showed.
About the creator
First author; inflammatory bowel disease, Oxford
Led the study that moved the acute severe colitis decision point from day 7 to day 3.
Senior author; Oxford inflammatory bowel disease unit
Supervised the prospective series from which the criteria were derived.
Co-author; colorectal surgery
Contributed the surgical perspective that gives the criteria their purpose — earlier involvement rather than later rescue.
Limitations
- Derived in 51 episodes at a single centre in 1996, a small sample by modern standards.
- The therapeutic landscape has changed completely — infliximab was not available, and ciclosporin was used in only 14 episodes.
- Binary output, so a patient just short of the threshold and one far below it are reported identically.
- Colonic dilatation is not considered at all, despite being an independent indication for surgery.
- CRP may be falsely low in patients already on immunosuppression or biologics, weakening the combined arm.
- The rule is stated two ways in the source paper, and the looser version is more widely implemented than the operational one.
- It predicts colectomy on that admission only, and says nothing about the medium-term course of patients who avoid surgery.
- Stool frequency is self-reported or nursing-recorded and is not always accurately counted.
If you are the patient
If you have been admitted with a severe flare of ulcerative colitis, you will usually be started on steroids given through a drip. On the third day, the team will look at two simple things: how many times you are opening your bowels, and a blood test called CRP that measures inflammation. This is not an extra test — it is a planned checkpoint. The reason for checking on day three rather than waiting longer is that if the steroids are not working, it is much better to know early. It gives time to try a second medicine, such as infliximab or ciclosporin, and time to involve the surgical team for a proper discussion while you are still feeling relatively strong. In the past this conversation often happened around day seven, by which point people were weaker and any operation was harder to recover from. If the checkpoint suggests the steroids are not working, it does not mean you will definitely need surgery, and it certainly does not mean surgery is happening that day. It means the team will act now rather than wait — starting a second medicine, and making sure a surgeon has met you in case one is needed. If you pass the checkpoint, that is encouraging, but you will still be reviewed every day and assessed again around day seven. Two things are worth asking about: whether you are on a blood-thinning injection, which is important in a colitis flare even though you are passing blood, and whether infections such as C. difficile have been tested for, since they can look exactly like a flare that is not responding.
Frequently asked questions
What are the Travis (Oxford) criteria?#
Applied on day 3 of intravenous corticosteroids in acute severe ulcerative colitis: more than eight stools that day, or a stool frequency between three and eight together with a CRP above 45 mg/L. Either predicts roughly an 85% chance of colectomy during the same admission.
Is it 'more than 8 stools OR CRP above 45'?#
Not in the paper's analysis. The Results state the rule as more than eight stools, or three to eight stools together with a CRP above 45 — the CRP threshold is conditional on the stool frequency. The paper's Conclusions paraphrase it without that condition, and that looser form is what most secondary sources reproduce. This calculator implements the Results wording.
Where do the two versions disagree?#
For a patient with fewer than three stools on day 3 and a CRP still above 45. The operational rule does not flag them, because a stool frequency that low means they are responding and a lagging CRP adds little. The loose form does flag them. The divergence is uncommon in a Truelove and Witts population but points the wrong way, since it identifies patients who are getting better.
Why day 3 specifically?#
Because that is where the criteria were derived, and because it is early enough to act on. Before this work, steroid failure typically declared itself around day 7, by which time the patient was more malnourished, more deconditioned and more thrombotic, and any colectomy was performed under worse conditions. Applying the rule on day 2 or day 5 is extrapolation.
Does meeting the criteria mean the patient needs surgery?#
No. The criteria predict colectomy; they do not indicate it. What they should trigger is earlier action — involving the colorectal team while the patient is still well, making an active decision about rescue therapy with infliximab or ciclosporin, and excluding C. difficile and CMV superinfection before concluding the colitis itself is refractory.
What if the criteria are not met?#
Continue intravenous steroids with daily review. A negative day 3 result is not reassurance about the admission and is not a discharge decision. The same study defines a day 7 checkpoint that still applies: more than three stools daily with visible blood carries a 60% chance of continuing symptoms and a 40% chance of colectomy over the following months.
How do the Travis criteria compare with the Ho index?#
They answer the same question on the same day with different structure. Travis is binary and uses stool frequency and CRP. Ho is a graded 0–9 score using stool frequency, colonic dilatation and albumin, so it offers an intermediate band and — importantly — counts colonic dilatation, which Travis ignores entirely. Running both is reasonable, since dilatation alone reaches high risk on the Ho index.
What is most often missed on these admissions?#
Thromboprophylaxis. Acute severe colitis is a strongly prothrombotic state and bloody diarrhoea is not a contraindication to prophylactic anticoagulation, yet it is frequently withheld for exactly that reason. After that, failure to exclude C. difficile and cytomegalovirus, both of which mimic steroid-refractory colitis and are managed differently.
References
Original / primary reference
Clinical practice guidelines
- Rubin DT, Ananthakrishnan AN, Siegel CA, Sauer BG, Long MD. ACG Clinical Guideline: Ulcerative Colitis in Adults. Am J Gastroenterol. 2019;114(3):384-413.
- Lamb CA, Kennedy NA, Raine T, Hendy PA, Smith PJ, Limdi JK, et al. British Society of Gastroenterology consensus guidelines on the management of inflammatory bowel disease in adults. Gut. 2019;68(Suppl 3):s1-s106.