The Role of an Open-Access Bleeding Unit in the Management of Colonic Haemorrhage

A 2-Year Prospective Study

P. N. BRAMLEY, J. W. MASSON, G. McKNIGHT, K. HERD, A. FRASER, K. PARK, P. W. BRUNT, A. McKINLAY, T. S. SINCLAIR & N. A. G. MOWAT The Gastrointestinal Unit, Wards 13/14, Aberdeen Royal Infirmary, Forresterhill, Aberdeen, Scotland

Bramley PN, Masson JW, McKnight G, Herd K, Fraser A, Park K, Brunt PW, McKinlay A, Sinclair TS, Mowat NAG. The role of an open-access bleeding unit in the management of colonic haemorrhage. A 2- year prospective study. Scand J Gastroenterol 1996;31:764-769.

Background: Major colonic haemorrhage poses difficult diagnostic and therapeutic problems and, in contrast to upper gastrointestinal bleeding, has no generally accepted plan of management. Methods: We report community-based prospective data accumulated over 2 years (1991-93) on 1602 patients referred to an open-access bleeding unit with suspected gastrointestinal haemorrhage. Results: Of 278 (17%) admissions with suspected lower GI haemorrhage, 252 were confirmed. Forty-eight per cent were defined as ‘significant’ bleeds, with a decrease in haemoglobin and cardiovascular compromise. Of 102 significant bleeds in subjects more than 60 years old, 29% rebled, and 12.6% required emergency surgery. Diverticular disease (24%) was the commonest diagnosis, with tumours, infective colitis, and inflammatory colitis each at 10%. The overall 30-day mortality for colonic bleeding was 5.1% (13 of 252), with only 1 death occurring in the group less than 60 years old. Conclusions: This study provides a unique database for the natural history of colonic bleeding and its management within the setting of a specialized bleeding unit.

Key words: Audit; bleeding unit; colonic diseases; gastrointestinal haemorrhage

Dr. N. A. G. Mowat, Gastrointestinal Unit, Ward 13/14, Aberdeen Royal Infirmary, Aberdeen, Scotland, AB9 228 cfar: +44 I224 840711)

Whereas there are generally accepted policies for the management of upper gastrointestinal bleeding (1-3), no validated policies exist for colonic bleeding. This is partly because of difficulties in visualizing the actively bleeding colon (4,s) but also because colonic bleeding is sometimes perceived to be less catastrophic, stopping spontaneously in most patients (6).

The few reported studies examine the role of new diagnostic modalities (such as arteriography (7-13), on-table lavage and colonoscopy (6,14,15), or radioisotope scintig- raphy (16, 17)) in highly selected patient groups with severe colonic bleeding. Little has been written on the natural history of colonic bleeding in a whole community (18, 19), and there are no reports of a standardized approach to the management of colonic bleeding within an ‘Open-Access’ Bleeding Unit. The few reports from such units have consistently shown low death rates for upper GI bleeding (3,20), and this encouraged us to ascertain whether similarly good results might be achieved by applying the same standardized management to patients with colonic bleeding.

MATERIALS AND METHODS

All patients with suspected upper or lower gastrointestinal

bleeding from October 1991 to September 1993 were admitted to the high-dependency, six-bed Bleeding Unit in Aberdeen Royal Infirmary. At that time the Unit provided the only acute endoscopy service for the whole of Grampian and the Northern Isles (adult population of 467,760; General Register Office for Scotland, 1994). Patients were mostly direct referrals from general practitioners, with the others from inpatients within Aberdeen Royal Hospitals or smaller cottage hospitals. The Unit is continuously manned by two endoscopy-trained staff nurses, and a resident middle-grade physician with support and supervision by gastroenterolo- gists and surgeons.

On admission, severity of bleeding is graded as follows: ‘significant’ = with one or more of the following: collapse, shock, haemodynamic disturbance, or acute decrease in haemoglobin; ‘trivial’ = with no haemodynamic disturbance or decrease in haemoglobin; and ‘No bleed’ = no evidence of blood loss, with an alternative diagnosis made.

Investigation and management policy The emphasis is on speedy resuscitation and early

diagnosis. Significant bleeds are investigated as soon as resuscitation is achieved, irrespective of the time of the day. Those with fresh rectal bleeding are sigmoidoscoped on

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.

Bleeding Unit for Colonic Haemorrhage 765

admission and, if haemodynamically unstable, proceed to upper GI endoscopy to exclude rapid upper GI bleeding. If rectal bleeding continues with haemodynamic disturbance, and the diagnosis is still unclear, emergency visceral arteriography is undertaken, day or night. If rectal bleeding stops, then the patient is prepared for colonoscopy, and if the colon is not adequately examined, barium studies are performed.

The written management protocol for both upper and lower gastrointestinal bleeding evolved after a review of recent literature (1,6) and much discussion between consultant surgeons and gastroenterologists. Blood is replaced until the patient is haemodynamically stable, aiming for a post- transfusion haemoglobin of 10-1 1 ddl. Those who continue to bleed and require more than 4 units of blood, or those who have a significant rebleed (defined as further bleeding and drop of haemoglobin of 2ddl ) are reviewed with surgical colleagues with a view to early surgery. Therapeutic endoscopy for lower GI bleeding was not attempted during this study period. In those proceeding to surgery with an unidentified bleeding source, on-table colonoscopy with intraoperative colonic lavage is undertaken and, if appro- priate, enterotomy and enteroscopy.

Patients with significant haemorrhage were kept under observation in the Unit for 48 h after cessation of bleeding before transfer to a general medical ward, with discharge at the discretion of the supervising consultant. Low-risk patients with trivial or no bleeds are discharged directly home from the Unit when possible. Data were collected prospectively, and all deaths within 30 days of admission or surgery were reviewed to analyse the sequence of events after the index bleed. Mortality is expressed as the 30-day mortality per patient admission.

Statistics Where appropriate, data are expressed as median with

mean and 95% confidence intervals (95%CI). The chi- squared test with Yates correction was used for categorical data. Differences between continuous data samples with normal distributions were determined with the two-tailed t test.

RESULTS

Patients Over the 2-year period 1602 patients were admitted with

suspected (upper or lower) gastrointestinal bleeding. Of these,

Table 1. Gender distribution by age Gender <60 year old 260 year old Total

Male 54 64 118 Female 29 105 134 Total 83 169 252

1324 had upper gastrointestinal bleeding and have been reported elsewhere (21). Of the 278 with suspected colonic bleeding, 252 had bleeding confirmed, and in 26 (9.3%) an alternative diagnosis was made. The male to female ratio was 1.9 : 1 in those under 60 years, and 0.6 : 1 in those more than 60 years old (Table I). The annual incidence of colonic haemorrhage for the Grampian region is 27 per 100,OOO adult population at risk.

Source of referral Seventy-one per cent of all confirmed lower-bowel bleeds

were by direct GP referrals, with 8% referred via the Casualty Dept. Twenty-one per cent of patients bled while already in hospital for other reasons.

Triage The median time to admission was 1.3 h (mean f 95%

CI = 1.4 f 0.1). Fifty-six per cent of those with significant bleeds and 43% of trivial bleeds underwent upper GI endoscopy to exclude blood loss within the upper GI tract. After sigmoidoscopy, 33% of trivial bleeds required colonos- copy, and 45% underwent barium studies. In the significant bleed group, after sigmoidoscopy, 30% went directly to angiography, 47% proceeded to colonoscopy and 32% required barium studies for complete visualization of the colon or small bowel. Thirty-nine per cent of those with trivial (51 of 132) and 13% of those with significant bleeds (16 of 120) were discharged directly home from the unit (see Table 111). Patients with significant bleeds stayed in hospital longer than those with trivial bleeds (9 days (12 2 2) versus 4 days (6 2 1); P < OOO1).

Severity of bleed (Tables 11-IV) One hundred and twenty patients (47.6%) were categorized

as having significant bleeds, and of these 23 (9.1% of total bleeds) presented with cardiovascular collapse. More sig- nificant bleeds than trivial bleeds (60.3% versus 39.7%) occurred in patients more than 60 years old, whereas in the under-60 group, the trend was reversed, with 78% trivial bleeds and 22% significant bleeds (P < 0.0001). Eighty-five

Table 11. Age of patients on the basis of the decade and severity of bleed. Of all significant bleeds 85% occurred in the more than 60- year group compared with 50% of trivial bleeds

Age by decade Significant Trivial Total

< 20 20-29 30-39 40-49 50-59 6 0 4 9 7&79 8-9 >90 Total

2 2 2 5 7

23 27 45

7 120

2 12 12 16 23 22 22 19 4

132

4 14 14 21 30 45 49 64 11

252

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.

766 P. N. Bramley et al.

per cent of all significant bleeds were in those more than 60 years old, and of these, 43% were more than 80 years old. By contrast, only 50% of trivial bleeds were in those more than 60 years old, and of these, 17% were more than 80 years (Table 11).

Aetiology of bleeding The final diagnostic categories after investigation, surgery,

or postmortem, their overall frequency, and the severity of bleed are recorded in Table 111. Diverticular disease bleeding was the commonest diagnosis at 24%, with infective bloody diarrhoea, colitis, and tumours each at approximately 10%. However, in 25% of the patients no definite source of bleeding was identified. Of the 26 patients who had not bled, 13 had infective gastroenteritis, including 3 with septicaemia; 6 had large-bowel perforation or obstruction, and 7 miscellaneous (for example, iron-stained stool, vaginal bleeding, and haematuria).

Risk factors for bleeding The relative roles of recognized risk factors are included in

Table IV. The percentage of patients with significant concurrent medical illness increased with age but, surpris- ingly, did not influence severity of bleed in those more than 60 years old. In the younger age group, however, there was a significantly increased percentage with concurrent illness in the significant bleed group (66% versus 39%; P < 0.001) (Table IV).

Rebleeding Those most at risk of rebleeding were the elderly (more

than 60 years) with significant bleeds, and in this group 28.5% (29 of 102) rebled, compared with 3.0% (2 of 67) of elderly trivial bleeds (P < 0.OOOl). The rebleeding risk was lower in those less than 60 years old, occurring in 2 of 18 with significant and only 1 of 65 trivial bleeds.

Surgery Only 32 patients (12.6%) required surgery for continuing

blood loss or rebleeding. These included 26 of the 120 with significant bleeds, of whom 20 were more than 60 years. The

Table 111. Diagnosis classified by severity of bleed

main reasons for surgical intervention in this group were eight with tumour (three requiring urgent surgery for leiomyoma, Hodgkin’s disease colon, and colonic carcinoma; five colonic carcinomas were operated on semiurgently), seven with angiodysplasia, five with colitis (including ischaemic), two with diverticular disease, and four with miscellaneous diagnoses. Only 6 of 132 with trivial bleeds (4.5%) required semiurgent surgery (4 patients with bleeding from colonic carcinoma, and 2 with severe ulcerative colitis).

Long-term follow-up Long-term follow-up (mean, 30.5 months) was completed

for 250 patients. Of 20 patients readmitted with lower GI bleeding, 15 rebled from the same source; in 4 the diagnosis was made or altered on the second admission, and in 1 upper gastrointestinal bleeding precipitated re-admission. The diagnosis, age, and severity of the original bleed are shown in Table V.

Mortality The 30-day mortality for confirmed lower gastrointestinal

bleeding was 5.1% (13 of 252), with only 1 death in the under- 60-year group. Five of the 32 (15.6%) patients requiring surgery died (2 of angiodysplasia, 1 of ischaemic colitis, 1 of diverticular disease bleed, and in 1 the origin of the bleed was not identified). All were elderly (median age, 82 years; range, 78-84 years). Four of these five patients had severe concurrent disease. Eight patients who died did not have surgery. They also were elderly (median age, 83 years; range, 54-93 years), and all had severe concurrent disease. The mortality for those patients referred from the community (directly from a GP or via the Casualty Dept.) was significantly less at 2.5%, compared with 14.8% in those who bled while receiving treatment in hospital for other illnesses (P < 0.OOOl).

DISCUSSION

This study provided a unique opportunity to establish a community database of gastrointestinal bleeding, free from most distortions of case selection. Audit of the referral

Approximate Disease Significant Trivial Total percentages

Origin not found Diverticular Infective Ha em o rr h o i ds Angiod y splasia Cancer

Colitis, ischaemic Colitis-Crohn’s Colitis-UC Others

Polyp

29 40

2 1

15 8 3 5 7 2 7

35 20 24 21 2 5 9 5 2 5 4

64 60 26 22 17 13 12 10 9 7

11

25% 24% 10% 9% 7% 5% 5% 4% 4% 3% 4.5%

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.

Bleeding Unit for Colonic Haemorrhage 767

Table IV. Triage through the Unit, risk factors, rebleed, and surgical rates classified by age and severity of bleed

<60 year old a50 year old

Significant, Trivial, Significant, Trivial, n = 18 n = 65 n = 102 n = 67

% Discharged directly home 11% 47% 14% 30% % Concurrent illness 66% 39% 81% 78%

% Smoking 28% 43% 10% 15% % Rebleed 11% 2% 29% 3% % Surgery 33% 3% 20% 6%

% on NSAIDs*/aspirin 17% 17% 51% 49%

* NSAIDs = non-steroidal anti-inflammatory diseases.

patterns showed that 95% of all those with upper or lower GI bleeds, whether inpatient or referred from the community, were admitted to the Unit.

Referring doctors have open access at any time for advice or direct referral, thus minimizing delays, with most patients arriving within 90 min of notification. As the Unit admits all those with suspected gastrointestinal bleeds, this removes the need for referring doctors to make the sometimes difficult distinction between upper or lower gastrointestinal bleeding.

The open-access admission policy resulted in a total of 26 no-bleeds (10.3%) being admitted in the first 2 years. The no- bleed group is a diverse diagnostic group, who are often acutely unwell with other conditions. For them the Unit acts as a high-dependency ‘receiving unit’ enabling rapid diagnosis and treatment.

There is instant access to sigmoidoscopy, flexible sig- moidoscopy, or colonoscopy as the situation dictates, with the facility for rapid access to angiography. Despite the early investigation of these patients, in 25% the origin of the bleeding was not found. A significant number of this group had diverticular disease, but we were unwilling to ascribe

bleeding to this diagnosis unless we had visual confirmation at colonoscopy. Bleeding directly attributable to angiodysplasia can also be difficult to establish, and it is sometimes found incidentally during colonoscopy (4). We attributed bleeding to angiodysplasia only if we bad convincing visual evidence. It is therefore possible that the reported incidence of bleeding from diverticular disease and angiodysplasia is underesti- mated.

We did not specifically design our study to test the relative sensitivity and specificity of the diagnostic tests used. With no accepted gold standard for diagnosis of colonic bleeding, this type of study would be difficult. However, our experience with angiography is disappointing (diagnostic, 33%; localiz- ing, 20%; and unhelpful, 47%) and does not match some previous reports (10). We are in the process of trying to improve the diagnostic yield of angiography and hope to report this in the future.

To evaluate the importance of fresh rectal bleeding as a marker of colonic haemorrhage, we have separately docu- mented the frequency of this sign in those with confirmed upper gastrointestinal bleeding. In all, 35 of 552 (6.3%) of

Table V. Late rebleeding admissions (>30 days)

Age at initial bleed (years) bleed Diagnosis at initial bleed Source of late rebleed

Severity of initial

62 74 82 93 93 76 73 87 85 70 16 23 53 93 75 93 60 57 45 80

~~

Significant Trivial

Significant Significant

Trivial Significant Significant Significant Significant Significant Significant

Trivial Trivial

Significant Significant Significant Significant

Trivial Significant

Significant

Cause unknown Angiody splasia Angiodysplasia Angiodysplasia Angiodysplasia

Diverticular Diverticular Diverticular Diverticular

Cause unknown Inflammatory bowel disease Inflammatory bowel disease

Ischaemic Ischaemic Carcinoma Carcinoma

Cause unknown Haemorrhoids

Angiodysplasia

Polyp

Angiodysplasia Angiodysplasia Angiodysplasia Angiod y splasia Angiod ysplasia

Diverticular Diverticular Diverticular Diverticular Diverticular

Inflammatory bowel disease Inflammatory bowel disease

Ischaemic Ischaemic Carcinoma Carcinoma

Aortacolonic fistula Colonic varices Haemorrhoids

Duodenal ulcer

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.

768 P. N. Bramley et al.

these patients had fresh rectal bleeding in association with to be concentrated on those patients with significant haematemesis and/or melaena. Only five (0.9%) presented haernorrhage. Most importantly, this approach produces low with fresh rectal bleeding only. death rates and for that reason deserves widespread adoption.

Those patients who bled while in hospital with another The. potential economic benefits of this approach to GI primary pathologic condition carried a significantly higher bleeding are important and are currently under detailed risk of dying (15%) than those referred directly from the evaluation. community. This reflects their’generrtl debility due to multiple pathologic conditions. Although these groups are not matched for age and sex, the high mortality underlines the importance ACKNOWLEDGEMENTS of their inclusion in mortality data.

There is only one study with which to compare our data (19). The Finnish study reported 266 lower gastrointestinal bleeds diagnosed by sigmoidoscopy, double-contract barium enema, or colonoscopy over a 5-year period (19). The commonest diagnostic groups were local anal conditions such as haemorrhoids/fissures (35%), polyps/cancers (21%), diverticular bleeds (19%), inflammatory bowel disease (8%), angiodysplasia (6%), with a similar number of undiagnosed patients (24%). Although no data are presented on the severity of bleeding, the diagnostic pattern suggests a much greater proportion of trivial bleeds. The differences between studies may reflect differing referral patterns, with Grampian general practitioners referring local anal conditions directly to surgical outpatients. The 30-day mortality for the Finnish series was 4.1% (11/266), similar to that reported in this study; however, our population’s mean age was nearly 8 years older.

Almost half of the patients in the present study (120 of 252) presented with significant bleeds with a decrease in haemoglobin and cardiovascular compromise. Our data suggest that severe bleeding occurs especially in the elderly, often in association with concurrent illness, and is not usually fatal in those less than 60 years old (median age at death, 82 years). The lack of any association between ingestion of non- steroidal anti-inflammatory agents or aspirin and increasing severity of bleeding is surprising and contrasts with what we have reported in upper gastrointestinal haemorrhage (21). Other reports have linked increasing complications, from diverticular disease to ingestion of non-steroidal anti- inflammatory agents or aspirin (22,23).

Although there are reports of selected series in the literature (6,24), these are often related to ill-defined catchment populations and fail to include subjects who bled while already an inpatient with an alternative primary diagnosis. There are no other community surveys with which to compare our mortality and surgical referral rates. The data reported are uniquely derived from unselected patients from a large population, enabling an accurate assessment of diagnosis, severity, and mortality of colonic haemorrhage to be made. This centralized approach to the management of both upper and lower gastrointestinal bleeding enables medical and nursing staff expertise to concentrate and evolve within one area, with rapid patient progress through agreed investiga- tional procedures. Bed usage is efficient, with trivial bleeds rapidly triaged through the unit, enabling time and resources

We thank all nursing staff and junior medical staff for their major contribution. We also acknowledge David Anderson, Ruth McKee, Tim O’Hanrahan, Peter Thomas, John Bagley, and Dominic Walshe for their valuable endoscopic help and all our consultant surgical colleagues for their dedicated work in making the Unit a success. We thank Jeff Hussey and Shona Campbell for their excellent radiologic support. Aberdeen Royal Hospitals NHS Trust fund the Unit and also supported a research nurse for part of the study period. J. Masson is supported by an educational grant from Glaxo UK. Part of these data have previously been published in abstract form (Gut 1993;34:S55).

REFERENCES

1. Guidelines for good practice in and audit of the management of upper gastrointestinal haemorrhage. Report of a joint working group of the British Society of Gastroenterology, the Research Unit of the Royal College of Physicians of London and the Audit Unit of the Royal College of Surgeons of England. J R Coll Physicians Lmd 1992;26:281-9.

2. Clements D, AsIan S, Foster D, Stamatakis J, Wilkins WE, Moms JS. Acute upper gastrointestinal haemorrhage in a district general hospital: audit of an agreed management policy. J R Coll Physicians Lond 1991;25:27-30.

3. Holman RA, Davis M, Gough KR, Gartell P, Britton DC, Smith RB. Value of a centralised approach in the management of haematemesis and melaena: experience in a district general hospital. Gut 1990;31:504-8.

4. Miller LS, Barbarevech C, Friedman LS. Less frequent causes of lower gastrointestinal bleeding [review]. Gastroenterol Clin

5. Boley SJ, Brandt U, Frank MS. Severe lower intestinal bleeding: diagnosis and treatment. Clin Gastroenterol 1981; 10:

6. Berry AR, Campbell WB, Kettlewell MG. Management of major colonic haemorrhage. Br J Surg 1988;75:6374.

7. Bar AH, DeLaurentis DA, Parry CE, Keohane RB. Angiography in the management of massive lower gastrointestinal tract hemorrhage. Surg Gynecol Obstet 1980;150:226-8.

8. Grace DM, Gold RE. Angiography in determining the cause and treatment of gastrointestinal bleeding. Can J Surg 1978;21: 171-4.

9. Matolo NM, Link DP. Selective embolization for control of gastrointestinal hemorrhage. Am J Surg 1979;138:8404.

10. Browder W, Cerise El, Litwin MS. Impact of emergency angiography in massive lower gastrointestinal bleeding. Ann Surg 1986;204:530-6.

11. Baum S. Angiography and the gastrointestinal bleeder. Radi-

12. Salem RR, Wood CB, Rees HC, Kheshavarzian A, Hemingway AP, Allison DJ. A comparison of colonoscopy and selective visceral angiography in the diagnosis of colonic angiodysplasia. Ann R Coll Surg Engl 1985;67:225-6.

13. Friedman HI, Hilts SV, Whitney PJ. Use of technetium-labeled

North Am 1994;23:21-52.

65-91.

ology 1982143569-72.

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.

Bleeding Unit for Colonic Haemrrhage 769

autologous red blood cells in detection of gastrointestinal bleeding. Surg Gynecol Obstet 1983;156:449-52.

14. Batch AJ, Pickard RG, De Lacey G. Peroperative colonoscopy in massive rectal bleeding. Br J Surg 1981;68:64.

15. Cussons PD, Berry AR.. Comparison of the value of emergency mesenteric angiography and intraoperative colonoscopy with antegrade colonic irrigation in massive rectal haemorrhage. J R Coll Surg Edinb 1989;34:91-3.

16. Kouris K, Abdel-Dayem HM, Awdeh M. Image subtraction in acute gastrointestinal bleeding studies using 99Tcm-DTPA. Nucl Med Commun 1985;6:717-22.

17. Nicholson ML, Neoptolemos JP, Sharp JF, Watkin EM, Fossard DP. Localization of lower gastrointestinal bleeding using in vivo technetium-99m-labelled red blood cell scintigraphy. Br J Surg 1989;76:358-61.

18. Boley SJ, DiBiase A, Brandt LJ, Sammartano RJ. Lower intestinal bleeding in the elderly. Am J Surg 1979;137:57-64.

19. Makela J, Kiviniemi H, Laitinen S, Kairaluoma MI. Diagnosis

Received 8 November 1995 Accepted 12 February 1996

and treatment of acute lower gastrointestinal bleeding. S a n d J Gastroenterol 1993;28:1062-6.

20. Hunt PS, Korman MG, Hansky J, Marshall RD, Peck GS, McCann WJ. Bleeding duodenal ulcer: reduction in mortality with a planned approach. Br J Surg 1979;66:633-5.

21. Masson J, Bramley PN, McKnight G, Herd K, Mowat NAG. Specialised bleeding units are the logical way forward in the management of upper gastrointestinal haemorrhage: a two year prospective study [abstract]. Gut 1995;36:A34.

22. Campbell K, Steele RJ. Non-steroidal anti-inflammatory drugs and complicated diverticular disease: a casecontrol study. Br J Surg 1991;78: 190-1.

23. Wilson RG, Smith AN, Macintyre IM. Complications of diverticular disease and non-steroidal anti-inflammatory drugs: a prospective study. Br J Surg 1990;77:11034.

24. Baker R, Senagore A. Abdominal colectomy offers safe management for massive lower GI bleed. Am Surg 1994;60: 578-81.

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y M

onas

h U

nive

rsity

on

10/3

1/14

For

pers

onal

use

onl

y.