Gut 1996; 38: 905-910

Predicting outcome in severe ulcerative colitis

S P L Travis, JM Farrant, C Ricketts, D J Nolan, N M Mortensen, M GW Kettlewell,D P Jewell

AbstractBackground-Simple criteria are neededto predict which patients with severeulcerative colitis will respond poorly tointensive medical treatment and requirecolectomy.Aims-To find out if the early pattern ofchange in inflammatory markers or othervariables could predict the need forsurgery and to evaluate the outcome ofmedical treatment during one year followup.Patients-51 consecutive episodes ofsevere colitis (Truelove and Witts criteria)affecting 49 patients admitted to JohnRadcliffe Hospital, Oxford.Methods-Prospective study monitoring36 clinical, laboratory, and radiographicvariables. All episodes treated with intra-venous and rectal hydrocortisone and 14of 51 with cyclosporine.Results-Complete response in 21episodes (-3 stools on day 7, withoutvisible blood), incomplete response in 15(>3 stools or visible blood on day 7, butno colectomy), and colectomy on thatadmission in 15. During the first five days,stool frequency and C reactive protein(CRP) distinguished between outcomes(p<0.00625, corrected for multiple com-parisons) irrespective ofwhether patientsor the number of episodes were analysed.It could be predicted on day 3, that 85% ofpatients with more than eight stools onthat day, or a stool frequency betweenthree and eight together with a CRP >45mg/l, would require colectomy. Forpatients given cyclosporine, four of 14avoided colectomy but two continued tohave symptoms. After admission, com-plete responders remained in remissionfor a median nine months and hada 50/0 chance of colectomy. Incompleteresponders had a 60% chance of con-tinuous symptoms and 40%/ chance ofcolectomy.Conclusions-After three days intensivetreatment, patients with frequent stools(>8/day), or raised CRP (>45 mg/l) needto be identified, as most will requirecolectomy on that admission. The roleof cyclosporine for treating severecolitis has yet to be defined. After sevendays' treatment, patients with >3stools/day or visible blood have a 60%chance of continuous symptoms and40%/ chance of colectomy in the followingmonths.(Gut 1996; 38: 905-910)

Keywords: ulcerative colitis, cyclosporine, colectomy.

Treatment with intravenous corticosteroidsand a policy of early colectomy originallyreduced the mortality in severe episodes ofulcerative colitis from 31-61% in the 1950s' 2to 5-9% in 1962.23 Although mortality out-side specialist centres in 1974 remained alarm-ingly high (370/0, 4), in specialist or districthospitals with an interest in colitis it is now 3%or less, including operative mortality.5-7 Thesefigures are pertinent because the introductionof further medical treatment such ascyclosporine8 runs the risk of delaying colec-tomy inappropriately. It remains difficult,however, to predict at an early stage whichpatients with severe ulcerative colitis willrespond poorly to intensive medical treatmentand require colectomy.There is a need for simple clinical and

laboratory criteria that will predict outcomeand assist the decision to operate on patientswith severe ulcerative colitis. A large retro-spective study in 1975 identified a persistenttachycardia, fever, hypoalbuminaemia, andradiological features (mucosal islands or dila-tion) after 24 hour treatment in hospital asbeing associated with colectomy on thatadmission.9 A rising or persistently raised Creactive protein (CRP) was associated withurgent colectomy in six of eight patients,'0but it was not clear whether more frequentmeasurements of CRP would be of predictivevalue. It has since been reported that firstepisodes, extensive disease,6 11 and three ormore distended loops of small bowel on theinitial plain abdominal radiograph'2 areassociated with a poor response to intravenoustherapy.To assess whether the early pattern of

change in inflammatory markers or othervariables could predict the need for surgery, 36clinical, laboratory, and radiographic variableshave been measured prospectively in patientswith severe colitis. The study was performedwith the option of using cyclosporine inpatients not responding to intravenous corti-costeroids and also examined the outcomeduring a one year follow up period.

Methods

PATIENTSAll patients with severe ulcerative colitis admit-ted to the John Radcliffe Hospital betweenMarch 1992 and September 1993 wereevaluated prospectively. The diagnosis ofulcerative colitis was made on normal clinical,radiological, and pathological criteria and asevere episode defined as the passage of -6bloody stools daily with one or more of thefollowing criteria: temperature >37.8°C, pulse

Gastroenterology UnitS P L TravisJ M FarrantD J NolanD P Jewell

and Department ofSurgeryN M MortensenM GW Kettlewell

John RadcliffeHospital, Oxford

School ofMathematicsand Statistics,University ofPlymouthC Ricketts

Correspondence to:Dr S P L Travis,Gastroenterology Unit,Derriford Hospital,Plymouth PL6 8DH.Accepted for publication29 December 1995

905

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from

Travis, Farrant, Ricketts, Nolan, Mortensen, Kettlewell, _Jewell

>90/min, haemoglobin < 10-5 g/dl, or erythro-cyte sedimentation rate (ESR) >30 mm/h.3 13

MEASUREMENTSIn addition to the sex and date of birth of thepatient, the following data were collected:

Clinical details - date of diagnosis, duration ofcurrent relapse, duration of previous remission,current maintenance therapy, number andtiming ofprevious admissions, maximum extentof macroscopic disease, number of motions inthe day prior to admission, abdominal tender-ness on palpation, and rigid sigmoidoscopicappearances (loss of vascular pattern, contactbleeding, or ulceration indicated by adherentmucosal slough) on admission.

Clinical observation - daily stool frequency,consistency (unformed, semi-formed, formed),amount ofblood (visible, occult blood positive,none), pulse rate, and temperature (bothmeasured six hourly).

Laboratory investigations - blood was takendaily for five days, then as indicated forstandard laboratory measurement of full bloodcount, ESR, CRP, potassium, creatinine,and albumin. In addition, serum was collecteddaily and stored at -20°C for orosomucoidanalysis by nephelometry (Beckman Array,High Wycombe, Bucks). Plain abdominalradiographs were taken on admission and aftertwo days. At the end of the study these werereported by an experienced radiologistunaware of the outcome, for the distribution offaeces (none, up to the hepatic flexure, splenicflexure, or sigmoid colon), presence ofmucosal islands, colonic dilatation >5 5 cm,-3 distended loops of small bowel gas,12 orperforation.

Outcome - Complete response to intensivemedical therapy was defined as a stool fre-quency 6,3/day on day 7, with no visible bloodin the motions. Incomplete responders weredefined as those with a stool frequency >3 orvisible blood on day 7 who did not requirecolectomy on that admission. Indications forcolectomy were failure to respond or frankdeterioration during the first few days ofintensive medical therapy; continued diar-rhoea, abdominal tenderness or a low gradefever after intensive medical therapy; andperforation, increasing colonic dilatation, ormassive haemorrhage.-4

MANAGEMENTAll patients received standard intensive medicaltherapy for severe colitis.15 Fluid, electrolyte,and haemoglobin deficiencies were correctedand hydrocortisone 100 mg given intravenouslysix hourly, with rectal hydrocortisone 100 mgtwice daily. This was continued for five to sevendays with oral fluids until it was clear that thepatient had responded or colectomy wasneeded. Parenteral nutrition was given tomalnourished patients. Incomplete responderswere treated with intravenous cyclosporine4 mg/kg/day or further intravenous corti-costeroids for up to six days, then convertedto oral therapy (cyclosporine 5 mg/kg/day

and oral corticosteroids), or referred for colec-tomy.

STATISTICAL ANALYSISStudent's unpaired t test was used to comparedata on admission. Both the number of episodes(51) and the number of patients (49) wereanalysed. All 51 episodes were analysed first,then the 49 patients excluding the initialepisodes in the two patients who were enteredtwice and finally the 49 patients excluding thesecond episodes in these two patients. Repeatedmeasures analysis ofvariance were used to assessdifferences between outcomes and to identifypotential trends in data during admission inmean bowel frequency, pulse rate, haemoglobin,platelet count, ESR, CRP, orosomucoid, andalbumin, using the StatGraphics statistical soft-ware package. The analyses were performedinitially for five days and then for eight days afteradmission. As 97% of all potential data wascollected, algorithms that allowed for occasionalmissing values were used, rather than excludepatients with missing data. Because the analysiswas repeated for all eight measurements, the sig-nificance level was set at 0.00625 to correct formultiple comparisons (p=005/8, Bonferroni'scorrection). A classification tree16 was thendeveloped to try to predict patient outcomefrom the measures obtained on the third day oftreatment.

Results

PATIENTSFifty one episodes were treated in 49 patients(26 male, age 21-77, median 43). Twenty one(42%) responded completely, 15 (29%)required colectomy on that admission, and 15had an incomplete response. Analysing bypatients rather than episodes changes thefigures by <2%. Excluding the initial episodes,21 patients had a complete response, 14required colectomy, and 14 had an incompleteresponse. Excluding subsequent episodes, thenumbers are 21, 13, and 15. Two patients hadbeen transferred for treatment from otherhospitals, neither ofwhom required colectomy.Two patients were subsequently found to haveCrohn's disease, one at the time of surgery andanother after an ileoanal pouch had beenformed.

ANTECEDENT DATAA third of episodes involved a first episode ofcolitis, defined as the first episode of symptomsleading to diagnosis, more commonly incomplete responders (Table I). For those witha first episode (16 of 49 patients), 12 (75%)responded completely, and three (19%)required colectomy during the initial admis-sion. For those with a previous episode ofcolitis, the median duration of remissiontended to be shorter in those who requiredcolectomy, as did the time since any previousadmission, but the differences did not reachstatistical significance. The duration of relapse

906

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from

Predicting outcome in severe ulcerative colitis

did not differ significantly. Similar proportionsof patients were taking maintenance therapy ineach group, but numbers are too small toidentify any relation between outcome andtype of salicylate therapy.

ADMISSION DATAThe number of Truelove and Witts criteria'3on admission in addition to a bloody stoolfrequency ¢'6/day was similar in those whoresponded to medical treatment or whorequired colectomy (Table II). One personwho had a colectomy had no additional criteriaon admission (and therefore technically onlyfulfilled the criteria for a moderate episode),but is included because she deteriorated duringtreatment to meet these criteria. On admission,

TABLE I Patient data prior to admission

Responders Incomplete Colectomy Overall

Number of episodes 21 15 15 51Age (SD) (y) 46-7 (19-2) 47-5 (12-3) 43-2 (15-3) 45-9 (15-3)First episode (/) 57 7 20 31Previous remission (range, months) 16 (5-38) 15 (5-240) 9 (3-54) 13 (3-240)Previous admission (%) 56 43 50 49Time since last admission (range, months) 67 (24-185) 61 (8-240) 5 (1-103) 47 (1-240)Salicylate therapy (%) 89 93 83 89

Sulphasalazine (/) 75 23 60 48Mesalazine (/) 25 31 10 23Olsalazine (%) 0 46 30 29

First episode: % patients in each group presenting with a first episode of colitis; previousremission: median duration of remission in months before current relapse; previous admission:% patients previously admitted for treatment, excluding those with first episodes (n= 17 of 35);time since last admission: median time in months in these patients; salicylate therapy:maintenance treatment taken by patients before admission.

TABLE II Patient admission details

Responders Incomplete Colectomy Overall

Number of episodes 21 15 15 51Motions/day 8 (2) 8 (2) 8 (3) 8 (2)Pulse rate 106 (15) 96 (11) 101 (14) 101 (14)Temperature 37-7 (0.7) 37-3 (0.8) 37-6 (0.4) 37-5 (0.7)Haemoglobin (g/dl) 12-6 (2.6) 11-3 (2.4) 11-2 (2.0) 11-8 (2.4)ESR (mm/h) 41 (25) 48 (20) 47 (28) 45 (24)CRP (mg/l) 43 (38)* 89 (85) 116 (102) 78 (81)Orosomucoids (mg/dl) 117 (41) 144 (55) 158 (50) 137 (50)Truelove and Witts criteria 2-2 (1-0) 2-1 (0.8) 2-1 (1.3) 2-2 (1.0)Extent of disease (/)

Distal 24 20 0 16Left sided 19 13 20 18Extensive 38 13 20 25Pancolitis 19 54 60 41

Figures are mean (SD). *Significantly different from the colectomy group (p=0 005). No otherdifferences are significant when corrected for multiple comparisons (p<0 00625). The meannumber of Truelove and Witts criteria are those in addition to bloody stool frequency >:6/day.When analysed by number of patients, the means change by <5% except for the CRP inincomplete responders, which was 100 (86) mg/l if initial episodes in two patients entered twicewere excluded.

TABLE HI Analysis of serial data overfive days: significance values comparing patientswho required colectomy with those who did not

Patients (n=49) Patients (n=49)Episodes (n=S1) Excludingfirst admission Excluding second admission

Mean A time Mean A time Mean A time

Bowel frequency <0-001 <0-001 <0-001 <0-001 <0-001 <0-001Pulse rate 0-012 0-064 0-010 0-026 0-022 0.077Haemoglobin 0-041 0 947 0 045 0-967 0-076 0-965Platelet count 0-042 0-796 0-056 0-778 0 054 0-882ESR 0-119 0-017 0-155 0-013 0-155 0-013CRP 0.001 0-025 0-002 0 044 <0-001 0.018Orosomucoids 0-011 0 904 0-013 0-902 0.011 0-761Albumin 0.035 0-726 0-041 0-714 0-029 0-837

Significance values of the data are presented graphically (see Figure). Data have been analysedby episode and per patient, excluding either the first or second admission of two patientsadmitted twice. Mean: compares the means between those who required colectomy and thosewho did not. A time: compares the rate of change over time between the two groups. Over thefive day period, stool frequency was significantly higher and decreased more slowly in those whorequired colectomy; the CRP was significantly higher in the colectomy group, but the rate ofchange did not differ when corrected for multiple comparisons (p<0-00625, see text).

the presence of severe rectal inflammation onsigmoidoscopy causing the appearance ofulceration was significantly more common(93%) in those who required colectomy thanthose who only had contact bleeding or agranular rectal mucosa (39%, p=0002). Thedistribution of disease did not differ signifi-cantly between groups, although pancolitis waspresent in 60% who required colectomy andonly in 19% of complete responders. Theinitial CRP was significantly higher in thecolectomy group compared with those whoresponded completely (Table II). These dataare consistent with the CRP being the mostsensitive maker of colonic inflammation.

MANAGEMENTIntravenous and rectal hydrocortisone weregiven for five days only in all patients whoresponded completely, a median of six days(range 5-8) in incomplete responders and amedian of five days (range 2-8) in those whohad a colectomy. In 30 of 51 episodes whichdid not completely respond to hydrocortisone,four had deteriorated sufficiently by five daysto need urgent colectomy. Another 12 con-tinued with hydrocortisone for up to threemore days until it was clear whether improve-ment continued (8) or colectomy (4) wasneeded, and the remaining 14 received intra-venous cyclosporine (4 mg/kg/day) for amedian four days (range 1-6). In those whoreceived cyclosporine, seven of 14 did notimprove and proceeded to colectomy duringthat admission. The remaining seven of 14were given oral cyclosporine (5 mg/kg/day)when symptoms were under sufficient controlto leave hospital, but three required colectomywithin three months. Of the incompleteresponders, the median stool frequency on dis-charge was three (range 2-9), with visibleblood in 60% and after a median nine days inhospital (range 7-20). Of the 15 patients whohad a colectomy on the same admission, 12had a subtotal colectomy, one a proctocolec-tomy, one a colectomy and mucous fistula, andthe one found to have Crohn's colitis at opera-tion had a split ileostomy.

PATTERN OF CHANGE DURING FIRST FIVE DAYSWhether analysed by episode or by patientnumbers, repeated measures analysis of vari-ance over the first five days showed that thebowel frequency and CRP were significantlyhigher (p<000625) in patients who requiredcolectomy than in those responding partly orcompletely. The bowel frequency also declinedsignificantly more slowly in the colectomygroup (Table III, Figure). At a less rigorous5% level of significance, the mean pulse rate,haemoglobin, platelet count, serum albumin,and orosomucoids also differed between thosewho required colectomy and those who didnot. The rate ofchange in pulse rate, CRP, andESR was slower in the colectomy group whenanalysed by patients rather than episodes(Table III). When data were analysed overeight rather than five days, the bowel

907

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from

Travis, Farrant, Ricketts, Nolan, Mortensen, Kettlewell, _rewell

frequency, pulse rate, and CRP were signifi-cantly higher (p<0 00625) in those whorequired surgery. Follow up analysis showedthat incomplete responders were more similarto the colectomy group than the completeresponders for bowel frequency, but similar tocomplete responders for the change in CRP.Two days after admission, the presence of

mucosal islands on the plain abdominal

radiograph was significantly more common inthose requiring colectomy (50%) than thosewho did not (10%, p=0.013). The extent ofcolitis (pancolitis, extensive, left sided, distal,proctitis) on the admission radiograph agreedwithin one colonic segment to that identifiedby barium enema, colonoscopy, or operationin 74%. The plain radiograph overestimatedthe extent of disease by two or more colonic

-0- Colectomy

-*- Nocolectomy1101

100

a)

a)a)CL,

90

80

70

1 2 3 4 5

12.5

12.0

mo 11.5m

c

= 11.0g° 10.50)0E 10.0a)

= 9.5

1 2 3 4 5

4 51 24 5 1 2 3 4 5

mE

0C.)ci

E0c)0

0

1 2 3 4 5

-0

0-

2 3 4 3 5Days Days

Pattern ofchange (mean (95% confidence limits)) for different variables in severe ulcerative colitis during the firstfivedays of treatment (see Table IIIfor significance values).

10

9

>8

a 7

az0 6

5

0m 4

3

2

140

120

100

E 80E

[L 600

40

20

0

55

50

45=

40EE 35

U) 30

25

20

15

2

470

450

- 430

c 410CO,0 390

X 370

350

330

ou

!4.n.

908

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from

Predicting outcome in severe ulcerative colitis

segments in 18% and underestimated theextent in 8%.

PREDICTING OUTCOME ON DAY 3The simplest rule predicted with 85% successthat patients with more than eight bowelactions on day 3, or with three to eight bowelactions and a CRP >45 mg/l would needcolectomy on the same admission. Of thosemisclassified by this rule, four patients whowould have been classified as surgical cases didnot undergo colectomy on that admission, butrequired colectomy in the following months.Three patients underwent colectomy when thisclassification rule suggested that they shouldnot.

FOLLOW UP DATAPatients were followed up for a median periodof 12 months (range 3.5-2 1). For those whohad responded completely, none had continu-ous symptoms and the duration of remissionwas significantly longer than in incompleteresponders (p<0001, Table IV). Immuno-suppression with prednisolone, azathioprine orcyclosporine was still necessary at the time oflast follow up in 14% of complete responders,compared with 82% of incomplete responders.Among the complete responders, colectomywas subsequently needed in one patient (5%),compared with six of 15 episodes (400/o) inincomplete responders. Of those who had acolectomy, 10 of 15 (67%) went on to have anileoanal pouch (including one which later hadto be excised for Crohn's disease) and three of15 (20%) had a proctectomy after continuoussymptoms from the rectal stump. One patientdied from a myocardial infarction five monthsafter colectomy.

DiscussionThis study confirms that patients with severecolitis defined by the Truelove and Wittscriteria13 have a 29% chance of colectomyon the same admission.5 14 This reflects theinadequacies of medical treatment even withcyclosporine and the prospective data showhow difficult it is to predict who will requirecolectomy. Incomplete responders are shown

TABLE IV Follow up data on patients responding completelyor incompletely to medical treatment (median 12 months,range 35-21)

Responders Incomplete

Number of episodes 21 15Remission (range, months) 9 (2-21) 0 (0-12)*Continuous symptoms (/) 0 60Relapse (%) 43 86Number of relapses (range) 0 (0-4) 1 (0-2)Readmission (/) 36 50Colectomy (%/o) 5 40Immunosuppressants (n) 3/20 9/11

Prednisolone 2 8Azathioprine 1 4Cyclosporine 0 3

Remission: median duration of remission, *p<0-001; relapse:0/0 relapses in those entering remission; readmission: %patients readmitted for treatment, including colectomy;immunosuppressants: patients receiving these drugs at thetime of last follow up; five of nine incomplete responders weretaking more than one immunosuppressant.

to have a very high risk of continuing symp-toms or colectomy in the months after a severeepisode.While the criteria for a complete response

were stringent (stool frequency -3 withoutvisible blood after seven days), this is a reason-able definition of remission for patients suf-fering the symptoms. Only 42% of episodesresponded completely to intensive medicaltreatment. By the time that incompleteresponders were discharged, 73% still had astool frequency >3/day or visible bleeding. Ofthe 49 patients, 43% ultimately required colec-tomy. Although patients are not normallyentered into a study more than once, we feelthat analysis of the number of episodes ratherthan the number of patients is clinically mostrelevant. This is because a previous severeepisode of colitis influences surgical decisionmaking and the principal end point of thestudy was surgery on that admission. As ithappens, analysis by episodes or patients(Table III) made little difference.There is thus an urgent need for more effec-

tive medical treatment. Cyclosporine wasineffective in seven of 14 patients who had notresponded to intravenous corticosteroids, andin the 500/O who showed some response tocyclosporine, only four subsequently avoidedcolectomy and two of these continued to havesymptoms. This means that only two of 14patients (14%) can be considered to haveentered remission and these two were stillreceiving prednisolone at the time of follow up.These figures are not encouraging and are incontrast with those described by Present.8 Itremains possible, however, that earlier use ofcyclosporine might be of greater benefit. Thereis also a potential benefit from partial remis-sion, in that it may allow a patient time tocome to terms with the prospect of surgery,especially if the extent of disease is limited.A severe relapse with a defined course of

action (admission for intensive medical ther-apy) can be identified by a single Truelove andWitts' criterion'3 in addition to a bloody stoolfrequency ¢n6/day. Table II shows that therewas no difference in the number of criteria onadmission between responders, incompleteresponders, or those who required colectomy.Objective assessment of relapse in ulcerativecolitis is essential if the severity is not to beunderestimated, but there seems little need todifferentiate between 'severe' and 'fulminant'colitis. Several of those who met all five criteriaresponded completely to medical therapy andone patient who only had a moderate episodesubsequently deteriorated to need urgentcolectomy. The two patients who turned out tohave Crohn's colitis are included because thisreflects clinical practice. Excluding these twodid not change the significant variables in therepeated measures analysis of variance.

Only bowel frequency and CRP differedsignificantly between those responding tomedical treatment and those requiring colec-tomy at the most rigorous level of significance(Table III). The orosomucoids tended to behigher on admission and the ESR to declinemore slowly in the colectomy group, which

909

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from

910 Travis, Farrant, Ricketts, Nolan, Mortensen, Kettlewell, J7ewell

favours the CRP as the most useful inflam-matory marker. In an attempt to find a morespecific marker of colonic inflammation, theserum concentration of nitric oxide metaboliteswas measured in some patients. These metabo-lites decreased in a similar fashion to the CRP,but did not discriminate between the groups.17The pulse rate tended to be higher and declinedmore slowly in those who required colectomy,while the haemoglobin and albumin werelower. These simple measures help in the over-all assessment of colitic patients. Surprisingly,the platelet count, temperature, or amount ofsmall bowel gas did not consistently discrimi-nate between those who required colectomyand those who did not.

It might be argued that there is circularreasoning on the premise that we have definedcriteria for colectomy and then evaluated themeasures that define that decision. Indeed, ithas been suggested that the threshold forcolectomy in severe ulcerative colitis in Oxfordis lower than elsewhere. If this is the case, thenpublished experience supports the practice.3 5 6By examining the change in pattem of com-monly measured variables, we have tried topredict which patients make a poor response tomedical treatment. In Oxford, such patientsgenerally have a colectomy because delayingsurgery increases the risk of complications, orof persistent symptoms if medical treatment iscontinued. No single factor (such as diarrhoea)influenced the surgical decision. All decisionsconsidered the duration, extent and previouspattem of disease, the influence of symptomson the patient's lifestyle, the inclination of thepatient, and response to or side effects frommedical treatment.The follow up data show that there is good

reason to be optimistic in those who respondcompletely (5°/O colectomy rate, 85% in remis-sion without immunosuppressants), but everyreason to be cautious in other patients. Forincomplete responders, 40% required colec-tomy within a few months (range 3-30 weeks)and only two of 14 (14%) remained in remis-sion without immunosuppressive therapy.These data are similar to previous reports.3 5Furthermore, in five patients who had anemergency sub-total colectomy and were notcandidates for an ileoanal pouch, four had tohave a proctectomy for continuing symptoms.Even though emergency proctocolectomy hasbeen shown to be a safe operation in experi-enced hands,5 a sub-total colectomy wasinitially performed in these patients because ofconstraints on theatre time.What conclusions, then, can be drawn to

give objective advice on the management ofpatients with a relapse of ulcerative colitis?Firstly, the severity of relapse should beassessed objectively. All patients with a bloodystool frequency >6/day with any additional

feature (pulse >90, temperature >37.8°C,haemoglobin <10.5 g/dl, ESR >30) should beadmitted for intensive treatment. Stool fre-quency, six hourly pulse rate, and CRP shouldbe monitored daily as a slower rate of improve-ment in these variables distinguished thosewho required urgent colectomy from thosewho did not. Although one must be wary ofplacing values on individual variables, it ispossible with reasonable confidence to predictthe outcome on day 3. Patients who continueto have frequent stools (>8 on day 3), or anincreased CRP on day 3 (>45 mg/l with a stoolfrequency of 3-8) need to be identified early,as 85% of these will require colectomy usingthe criteria in this study. The inadequacies ofconventional treatment should be recognised,but the role of cyclosporine for treating severeulcerative colitis has yet to be defined. After aweek of treatment, those patients who have astool frequency >3/day or visible blood in thestool have a 60% chance of continuous symp-toms and 40% chance of colectomy in themonths after admission.We are particularly grateful to Sister and staff on the medicaland surgical Gastroenterology wards for their care of thepatients, to Dr Helen Chapel and June White, Department ofImmunology, for the measurement of orosomucoids, and toKaren Hayllar, King's College Hospital, London for additionalstatistical evaluation.

1 Edward FC, Truelove SC. The course and prognosis ofulcerative colitis. Gut 1963; 4: 299-315.

2 Gallagher ND, Goulston SSM, Wyndham N, Morrow W.The management of fulminant ulcerative colitis. Gut1962; 3: 306-11.

3 Truelove SC, Jewell DP. Intensive intravenous regimen forsevere attacks of ulcerative colitis. Lancet 1974; i:1067-70.

4 Ritchie JK. Results of surgery for inflammatory boweldisease: a further survey of one hospital region. BMJ1974; 1: 264-8.

5 Truelove SC, Lee EG, Willoughby CP, Kettlewell MGW.Further experience in the treatment of severe attacks ofulcerative colitis. Lancet 1978; ii: 1086-8.

6 Jarnerot G, Rolny P, Sandberg-Gertzen H. Intensive intra-venous treatment of ulcerative colitis. Gastroenterology1985; 89: 1005-13.

7 Jones HIV, Grogono J, Hoare AM. Acute colitis in a districtgeneral hospital. BMJ 1988; 294: 683-4.

8 Lichtiger S, Present DH, Kornbluth A, et al. Cyclosporinein severe ulcerative colitis refractory to steroid therapy.NEnglJMed 1994; 330: 1841-5.

9 Lennard-Jones JE, Ritchie JK, Hilder W, Spicer CC.Assessment of severity in colitis: a preliminary study. Gut1975; 16: 579-84.

10 Buckell NA, Lennard-Jones JE, Hernandez MA, Kohn J,Riches PG, Wadsworth J. Measurement of serum proteinsduring attacks of ultraviolet as a guide to patient manage-ment. Gut 1979; 20: 22-7.

11 Meyers S, Level PK, Feuer EJ, Johnson JW, Janowitz HD.Predicting the outcome of corticoid therapy for acuteulcerative colitis. Results of a prospective randomized,double-blind trial. Jf Clin Gastroenterol 1987; 9: 50-4.

12 Chew CN, Nolan DJ, Jewell DP. Small bowel gas in severeulcerative colitis. Gut 1991; 32: 1535-9.

13 Truelove SC, Witts U. Cortisone in ulcerative colitis: pre-liminary report on a therapeutic trial. BMJ 1954; 2:375-8.

14 Jewell DP, Caprilli R, Mortensen N, Nicholls RJ, Wright JP.Indications and timing of surgery for severe ulcerativecolitis. Gastroenterology International 1991; 4: 161-4.

15 Jewell DP. Medical management of severe ulcerative colitis.IntJ Colorectal Dis 1988; 3: 186-9.

16 Breiman L, Friedman JH, Olshen RA, Stone CJ. Classificationand regression trees. California: Wadsworth, 1984.

17 Rees DC, Satsangi J, Cornelissen PL, Travis SPL, White J,Jewell DP. Are serum concentrations of nitric oxidemetabolites useful for predicting the clinical outcome ofsevere ulcerative colitis. EurJ Gastroenterol Hepatol 1995;7: 227-30.

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.6.905 on 1 June 1996. Dow

nloaded from