gastrointestinal haemorrhage - NCBI

90

Journal of the Royal Society of Medicine Volume 73 February 1980

Survey of management in acute upper gastrointestinal haemorrhage' G E Thomas MD MRCP2

P B Cotton MD FRCP

Gastrointestinal Unit, Middlesex Hospital, London WIN 8AA

C G Clark MD ChM

P B Boulos FRCS

Departments of Surgery and Gastroenterology University College Hospital, London WCIE 6A U

Summary: The answers to a questionnaire concerning attitudes of members of the British Society of Gastroenterology to the management of acute upper gastrointestinal bleeding are analysed. In the majority of cases patients were admitted to general wards under the care of physicians. Use of intensive therapy units and venous pressure monitoring varied widely. Emergency endoscopy appeared readily available and was usually the first diagnostic procedure. Double contrast radiology and emergency angiography were available in relatively few centres. Specific nonoperative treatments (angiographic and endoscopic) were scarcely employed. Most respondents agreed that elderly patients fared badly, but there was little agreement concerning other factors which influence re-bleeding or outcome. There was a wide divergence of opinion concerning the need for surgical intervention in certain hypothetical clinical situations. Despite the difficulties involved, we believe that controlled trials are necessary to improve the management of bleeding patients. Introduction Modern techniques of investigation and treatment have apparently not reduced the overall mortality in patients with acute upper gastrointestinal haemorrhage (Schiller et al. 1970); some benefit may have been masked by the increasing age of patients (Allan & Dykes 1976). There have been no adequate trials of different treatment methods nor of the timing and type of surgical intervention. Reviews have called for more surgery (Schiller et al. 1970) or less (Allan & Dykes 1976) or for alternative nonoperative methods (Katon 1976). These disagreements have recently been highlighted by the debate concerning the role of emergency endoscopy. Whilst endoscopy can give a higher diagnostic yield than urgent barium radiology (Allen et al. 1973, Cotton 1975, Stevenson et al. 1976), there is no proof that this increases the patient's chance of survival (Dronfield et al. 1977, Eastwood 1977). Lack of benefit despite more accurate diagnosis suggests that management is faulty, and that information from urgent endoscopy should be used to formulate better treatment (Cotton 1977). In order to highlight areas of consensus and disagreement, we decided to canvas the opinions of doctors regularly dealing with acute upper gastrointestinal haemorrhage. This paper presents data obtained from members of the British Society of Gastroenterology (BSG) in response to a questionnaire.

Methods The first part of the questionnaire dealt with admission policy, diagnostic procedures and medical management, and was usually completed by a physician. The second part concerned surgical management, and was almost invariably completed by a surgeon. Answers were sought concerning general policy and individual opinions; no attempt was made to obtain detailed results of treatment. ' Accepted 2 July 1979 Present address: Kent & Canterbury Hospital, Canterbury, Kent

2

0141-0768/80/020090-06/$01.00/0

,'-.

1980 The Royal Society of Medicine


91

The questionnaire was sent to a total of 620 members of the British Society of Gastroenterology. Detailed replies were received from 154 centres, representing 262 named members. These constitute approximately half of the membership in active clinical work, others being retired or working in nonclinical specialties (radiology, pathology, basic sciences). Not everybody answered every question; results are therefore expressed as percentages of the total response to individual questions.

Questions, results and discussion Admission responsibility In 96% of the centres, patients were normally admitted under the care of physicians. The fact that the decision for or against surgery is perhaps the most crucial decision for any individual has led to the suggestion that bleeding patients should be dealt with entirely by surgeons (Hellers & Ihre 1975). However, morbidity and mortality are greatly influenced by associated medical disease (Coghill & Willcox 1960, Walls et al. 1971, Dronfield et al. 1977) which may be better handled by physicians. Most experts agree that a combined team approach is ideal (Devitt 1969). Use of intensive therapy units and venous pressure monitoring Most responders treated their patients on general wards. Availability of an intensive therapy unit (ITU) was not questioned, and the lack of a unit might account for the high proportion of respondents (40%/) who never used one; 5% admitted most of their bleeding cases to ITU, and 55% used one for the most severe cases. The suggestion that patients should be managed in an ITU for the first 48-72 hours when re-bleeding is most likely (Northfield 1971, Hegarty et al. 1973, Jones et al. 1973) has clearly gained limited acceptance. Central venous pressure (CVP) monitoring was used frequently by only 30% of responders, and during transfusion by only 70%. Northfield & Smith (1970) recommended routine CVP monitoring to provide better control of transfusion requirements, and earlier recognition of re-bleeding.

Availability ofspecialist investigations BSG members believed that endoscopy was consistently more available on an emergency basis than barium radiology and angiography (Table 1). The apparent lack of availability of emergency barium radiology probably reflects the fact that it is rarely requested (see below). If angiography is to be useful, it must be available during active bleeding (Irving & Northfield 1976, Butler et al. 1976, Athanasoulis et al. 1976). Emergency angiography clearly plays little part in current management, since it is only available in 17% of centres (Table 1). Table 1. Percentage of centres where different investigations were usually or always available within 4 or 24 hours

Endoscopy Barium radiology Angiography

< 4 hours

< 24 hours

58% 33% 17%

90% 70% 36%

Choice of investigation 82% of respondents usually or always recommended endoscopy as the first diagnostic procedure, the remainder choosing barium radiology. Of those who usually performed endoscopy first, 42% often or usually requested a barium meal later; 2% often proceeded to angiography. Of the 18% of responders who usually chose barium radiology first, 66% usually then proceeded to endoscopy. Factors which influence the respondents in their choice or sequence of investigations were not explored.

92


Emergency endoscopy has been claimed to have a diagnostic accuracy of 85-95% (Cotton 1975) and comparative studies have consistently shown that early endoscopy is more accurate than radiology (Stevenson et al. 1976, Dronfield et al. 1977). There is increasing evidence that double contrast barium radiology is more accurate than the conventional technique (Scott-Harden 1974, Laufer et al. 1975, Stevenson et al. 1976). It was therefore surprising to find that the double contrast technique was used frequently in only 33% of the responding centres, sometimes in 51% and never in 16%. Timing of investigation There was a predictable relationship between the severity of bleeding and the speed of investigation (Table 2): 98% of centres investigated severe bleeding within 24 hours, whereas only 47% felt the same urgency with minor bleeds. The practice appears to be a sensible compromise between the clinical desire for early diagnostic information and the inconvenience of arranging emergency investigations. The diagnostic yield falls off when patients are investigated after 48 hours (Allen et al. 1973, Cotton 1975). Table 2. Usual timing of investigation according to the severity of bleeding

% centres

Minor Bleed Moderate bleed Severe bleed

< 4 hours

4-24 hours

> 48 hours

3 21 61

44 60 37

24 2 0

Do BSG members control or influence management policy of bleeding patients admitted under other clinicians? The answers to this question were particularly revealing. Only 31% of respondents claimed that more than 80% of bleeding patients in their own hospital were treated as they themselves recommend; 50% claimed this for between 50% and 80% of patients admitted with bleeding. Whether gastroenterologists handle bleeding patients better than other clinicians is a moot point. It could be argued that the main value of endoscopy is to ensure that bleeding patients are seen by someone who has specialist knowledge of gastroenterology. Does a precise early diagnosis help the patient? 71% of respondents were convinced that a precise diagnosis (usually by endoscopy) benefited their patients, 27% did not know, and 2% thought that it did not. 72% of respondents felt that they would be unwilling to withhold endoscopy for the purposes of a controlled trial. Dronfield et al. (1977) showed that the increased diagnostic accuracy ot endoscopy did not decrease overall mortality in bleeding patients. This was explained by the fact that the lesions missed by radiology were usually small and unimportant in terms of re-bleeding. It was therefore suggested that centres with good radiological facilities should not hurry to provide emergency endoscopy services. The selected group of BSG members already appear to have better access to emergency endoscopy than double contrast radiology.

Use of specific nonoperative treatments Both fibreoptic endoscopy (Katon 1976) and angiography (Athanasoulis et al. 1976) have important therapeutic applications. However, these new methods of treatment were used only rarely by BSG members; only 15% of centres had used angiographic treatment 1-10 times during the previous 2 years. In the same period, only 6% of centres had used endoscopic treatment between 1-10 times, and only 1% on more than 10 occasions.


93

Is re-bleeding more likely with gastric ulcers, or with duodenal ulcers? 34% of respondents believed gastric ulcers re-bled more frequently than duodenal ulcers; 31% believed the reverse, and 35% thought that the incidence of re-bleeding was equal. Most published series have found a higher rate of re-bleeding in gastric ulcers (Jones 1956, Coghill & Willcox 1960, Ward-McQuaid et al. 1960, Northfield 1971, Jones et al. 1973), but this tendency has been difficult to separate from the effect of ageing; recently Foster et al. (1978) found that duodenal ulcers re-bled more frequently than gastric ulcers.

Do age and sex influence re-bleeding? The great majority of respondents (88%) agreed with the published statements that the risk of re-bleeding increases with age (Lewin & Truelove 1949, Jones 1956, Allan & Dykes 1976, Morgan et al. 1977). 60% of respondents considered that the sex of the patient had no influence on re-bleeding, but 37% thought that men re-bled more frequently than women (and 3% vice versa). There is scant comment on this point in published series, but Northfield (1971) and Jones et al. (1973) found the incidence of re-bleeding to be equal between men and women. Factors influencing respondents towards surgery (1) Previous dyspepsia: A previous history of dyspepsia influenced 50% of respondents a little towards surgery, 26% a lot, and 24% not at all. Published series provide no clear guidelines. A short or absent past history of dyspepsia is associated with a poorer overall prognosis (Lewin & Truelove 1949, Coghill and Willcox 1960, Jensen et al. 1972, Johnston et al. 1973). This led Coghill & Willcox (1960) to recommend surgery in such patients. Reports of a higher operative mortality when there is little or no previous ulcer history (Cocks et al. 1972, Kim et al. 1972) may reflect a tendency to operate more quickly when an ulcer is clinically apparent (Schiller et al. 1970). The complexities of this problem are further illustrated by the increased incidence of associated diseases when there is a short or absent dyspeptic history (Coghill & Willcox 1960), and the higher mortality - both operative and nonoperative - when other diseases are present (Coghill & Willcox 1960, Walls et al. 1971, Cocks et al. 1972, Dronfield et al. 1977, Kim etal. 1972). (2) Previous episode of bleeding: 80% of respondents were greatly influenced towards surgery, 16% only a little, and 4% not at all when there was a past history of acute bleeding. Published series are surprisingly free from comment or data on this point, but survival from a particular episode is not influenced by such a history (Lewin & Truelove 1949, Jones 1956, Coghill & Willcox 1960). (3) Ulcer size: Large ulcers influenced 38% of respondents a lot towards surgery, 38% a little, and 24% not at all. The evidence in favour of surgical intervention for large bleeding ulcers is anecdotal (Coghill & Willcox 1960), and the claim that mortality increases with larger ulcers (Jones 1956) is not supported by detailed analysis of published data. (4) Increasing age: Most respondents (82%) were positively influenced towards surgery by increasing age; 15% were influenced only a little, and 3% not at all. Evidence to support this belief is not powerful. Jones (1947, 1956) and Coghill & Willcox (1960) recommend surgery for the elderly on the basis of their clinical experience, but without controlled observations. Since the mortality of surgery also increases with age, Allan & Dykes (1976) have suggested that medical treatment in an intensive therapy unit might be preferable in the elderly. Jensen et al. (1972) recommended surgery on the basis of a surgical mortality of 3%, compared with a mortality of 16% in patients managed medically. However their medical group included patients rejected as unfit for surgery. Indications for surgery in hypothetical patients 'A fit patient aged 45 with previous minor dyspepsia has an episode of acute bleeding which requires a 4 unit blood transfusion. The bleeding stops.' Respondents were asked whether they would recommend surgery (when and which procedure) when urgent endoscopy reveals (1) a duodenal or (2) a benign gastric ulcer. Answers to such hypothetical clinical situations are of limited value, but the responses show

94

Journal ofthe Royal Society of Medicine Volume 73 February 1980

remarkable policy disagreements (Table 3), both for gastric and duodenal ulcer patients. In these circumstances, surgery was recommended always by 8-10%, usually by 20-21%, sometimes by 46-52% and never by 20-23% of respondents. The timing of surgery appears to be less controversial; 95% of respondents preferred to operate on a semi-elective basis if possible. Table 3. Respondents' recommendations regarding surgery in a hypotheticalfit 4S-year-old with previous minor dyspepsia who has an episode ofacute bleeding which stops after 4 unit blood transfusion and endoscopy reveals (1) a duodenal ulcer or (2) a benign gastric ulcer Recommendations regarding surgery (%)

Duodenal ulcer Gastric ulcer

Never

Sometimes

Usually

Always

20 23

52 46

20 21

8 10

For bleeding duodenal ulcers, 84% of respondents usually performed vagotomy, pyloroplasty and ulcer suture; 9% favoured Polya gastrectomy, and 7% vagotomy and antrectomy. Billroth I partial gastrectomy was preferred by 86% for bleeding gastric ulcers; vagotomy, pyloroplasty and ulcer excision by 6%, Polya partial gastrectomy by 5% and various other procedures by the remainder. The respondents choice of surgical procedure reflects published advice (Ward-McQuaid et al. 1960, Clark 1968, Cocks et al. 1972, Hegarty et al. 1973). It is widely recognized that emergency surgery carries a higher mortality than a semi-elective procedure (Ward-McQuaid et al. 1960, Schiller et al. 1970, Crook et al. 1972, Johnston et al. 1973, Heideman et al. 1977), but this also reflects the fact that emergency surgery will be chosen more often when patients are bleeding most heavily. Respondents were also asked whether they would change their recommendations if the patient was aged 70 years, all other facts being equal. There was a definite preference towards more active surgical intervention with increasing age: 45% of respondents would change their management in duodenal ulcer, and 34% in gastric ulcer. Only 5% of those who would change became more conservative.

Preference for type of operation in non-ulcer bleeding Respondents were asked which procedure they would use (when necessary) in patients bleeding from multiple acute lesions in the antrum; multiple acute lesions involving the high gastric fundus; and oesophageal varices. The lack of surgical consensus in these circumstances is vividly illustrated in Table 4. Table 4. Preference of respondents (%) for type of operation when necessary

Operation Billroth I Vagotomy, pyloroplasty Polya gastrectomy Vagotomy, pyloroplasty Antrectomy Billroth I, Roux en Y Total gastrectomy Proximal gastric vagotomy Under-running ulcer Vagotomy, antrectomy, Roux en Y

Multiple acute Multiple acute lesions involving lesions in the high gastric antrum fundus 34 24 18 24 6 2 1 1 1 0

19 38 16 38 1 1 10 3 5 1

Bleeding oesophageal Operation

varices

Boerema, Crile Milnes Walker Emergency shunt Milnes Walker Tanner transection Various other

17 25 22 25 12 10


95

What do you do when laparotomy is performedfor bleeding and no lesion is found despite gastrotomy? Responses again indicated lack of consensus in this difficult situation: 39% of respondents would usually do no formal procedure, 33% would perform truncal vagotomy and drainage, 8% would perform endoscopy on the operating table, and the remainder offered a variety of alternative procedures.

Conclusions In Britain, patients with acute upper gastrointestinal bleeding are managed by a wide range of physicians and surgeons with varying specialist interests. This fact alone makes it difficult to collect comparative data, and to assess the results of new methods of investigation and treatment. The results of our questionnaire show wide policy disagreements within the membership of the British Society of Gastroenterology. Most members now utilize early endoscopy to demonstrate the bleeding lesion, but they lack control data on which to base subsequent management. In particular, there is a marked lack of consensus concerning indications for surgical intervention. This aspect merits further study and prospective trials, particularly in elderly patients.

Acknowledgment: We wish to thank our colleagues in the British Society of Gastroenterology for their courtesy in completing the questionnaire. References Allan R & Dykes R (1976) Quarterly Journal of Medicine 45, 533-550 Allen H M, Block M A & Schuman B M (1973) Archives ofSurgery 106,450-455 Athanasoulis C A, Waltman A C, Novelline R A, Krudy A G & Sniderman K W (1976) The Radiologic Clinics of North America 14, 265-280 Butler M L, Johnson L F & Clark R (1976) American Journal of Gastroenterology 65, 501-511 Clark C G (1968) Postgraduate Medical Journal 44, 590-593 Cocks J R, Desmond A M, Swynnerton B F & Tanner N C (1972) Gut 13, 331-340 Coghill N & Willcox R G (1960) Quarterly Journal of Medicine 29, 575-596 Cotton P B (1975) Topics in Gastroenterology, vol 3. Ed. S C Truelove and M J Goodman. Blackwell Scientific, Oxford Cotton P B (1977) Lancet i, 1367 Crook J N, Gray L W, Nance F C & Cohn I (1972) Annals ofSurgery 175, 771-782 Devitt J E (1969) Gastroenterology 57, 89-94 Dronfield M W, MclIlmurray M B, Ferguson R, Atkinson M & Langman M J S (1977) Lancet i, 1167-1169 Eastwood G L (1977) Gastroenterology 72, 737-739 Foster D N, Miloszewski K J A & Losowsky M S (1978) British Medical Journal i, 1173 Hegarty M M, Grime R T & Schofield P F (1973) British Journal of Surgery 60, 275-279 Heideman M, Larsson I, Stenquist B, Zederfeldt B & Darle N (1977) Acta Chirurgica Scandinavica 143, 307-312 HeUers G & Ibre T (1975) Lancet ii, 1250-1251 Irving J D & Northfield T C (1976) British Medical Journal i, 929-931 Jensen H E, Amdrup E, Christiansen P, Fenger C, Lindskov J, Nielson J & Damgaard Nielsen S A (1972) Scandinavian Journal of Gastroenterology 7, 535-540 Johnston S J, Jones P F, Kyle J & Needham C D (1973) British Medical Journal iii, 655-660 Jones F A (1947) British Medical Journal ii, 441-447 Jones F A (1956) Gastroenterology 30, 166-190 Jones P F, Johnston S J, McEwan A B, Kyle J & Needham C D (1973) British Medical Journal iii, 660-664 Katon R M (1976) Gastroenterology 70, 272-277 Kim H, Dreiling D A, Kark A E & Rudick L (1972) American Journal ofGastroenterology 62, 24-35 Laufer I, Mullens J E & Hamilton J (1975) Radiology 115, 569-573 Lewin D C & Truelove S C (1949) British Medical Journal i, 383-386 Morgan A G, McAdam W A F, Walmsley G L, Jessop A, Horrocks J C & de Dombal F T (1977) British Medical Journal ii, 237-240 Northfield T C (1971) British Medical Journal i, 26-28 Northfield T C & Smith T (1970) Lancet ii, 584-586 Schiller K F R, Truelove S C & Gwyn-Williams D (1970) British Medical Journal ii, 7-14 Scott-Harden W G (1974) Journal ofthe Royal College of Physicians of London 8, 365-374 Stevenson G W, Cox R R & Roberts C J C (1976) British Medical Journal ii, 723-724 Walls W D, Glanville J N & Chandler G N (1971) Lancet ii, 387-390 Ward-McQuaid J N, Pease J C, Smith A McE & Twort R J (1960) Gut 1, 258-265