Endoscopy for upper gastrointestinal bleeding: how urgent is it? - Nature

10 downloads 0 Views 206KB Size Report
Jul 14, 2009 - abstract | early endoscopy has been advocated for the management of upper gastrointestinal bleeding, but the optimal timing for early ...
reviews endoscopy for upper gastrointestinal bleeding: how urgent is it? Kelvin K. F. Tsoi, Terry K. W. Ma and Joseph J. Y. Sung abstract | early endoscopy has been advocated for the management of upper gastrointestinal bleeding, but the optimal timing for early endoscopy is still uncertain. The aim of this review is to evaluate the optimal timing of early endoscopy by examining the findings of randomized clinical trials and retrospective cohort studies that used comparable outcome measures and have been reported in the literature. Outcome measurements included recurrent bleeding, surgery, mortality, length of hospital stay, and blood transfusion. studies were categorized into those in which endoscopy was performed within 2–3 h, 6–8 h, 12 h or 24 h of the patient’s presentation to hospital. we conclude that early endoscopy aids risk stratification of patients and reduces the need for hospitalization. However, it may also expose additional cases of active bleeding and hence increase the use of therapeutic endoscopy. No evidence exists that very early endoscopy (within a few hours of presentation) can reduce the risk of rebleeding or improve survival. Tsoi, K. K. F. et al. Nat. Rev. Gastroenterol. Hepatol. 6, 463–469 (2009); published online 14 July 2009; doi:10.1038/nrgastro.2009.108

Introduction

upper gastrointestinal bleeding (uGiB) is a common disorder that warrants hospital admission and emer­ gency management. the annual incidence of hospital admission for uGiB in the us is estimated to be 160 cases per 100,000 people.1 Despite advances in both pharmacologic and endoscopic therapy, mortality from uGiB remains at 5–10%.2 endoscopy has become the standard of care in the diagnosis and treatment of uGiB. this method helps to identify the source of bleeding, establish the underlying etiology (for example, Helicobacter pylori infection), achieve hemostasis and provide prognostic informa­ tion to predict the risk of rebleeding.3 most national and international guidelines recommend that upper endo­ scopy should be performed within 24 h of presentation in patients with uGiB. 3–6 However, the role of early endoscopy in uGiB has been an area of contention among gastroenterologists, surgeons, endoscopists and health­care administrators. the american society for Gastrointestinal endoscopy does not give clear recom­ mendations, although it states that “early endoscopy (within 24 h of hospital admission) has a greater impact than delayed endoscopy on length of hospital stay and requirements for blood transfusion”.4 the timing of early endoscopy is not mentioned in the British society of Gastroenterology guidelines.5 these guidelines simply state that in clinical practice, endoscopy is only rarely required out of hours. the optimal timing for endoscopy is, therefore, unclear. Competing interests J. J. Y. sung declares associations with the following companies: AstraZeneca, Bristol-Myers squibb, GlaxosmithKline, Nycomed, roche. The other authors declare no competing interests.

retrospective cohort studies and a few randomized, controlled trials have evaluated the clinical benefit of early endoscopy, yet interpretation of their results is limited by the heterogeneity of study designs. Factors such as small sample size, selection of patients, the use of different definitions of early endoscopy (which vary from 1 h to 24 h) and various modes of endoscopic therapy used may contribute to the conflicting results. a factor that must be taken into consideration is that in countries or health­care systems where therapeutic endoscopy can be done only in the presence of accredited endoscopists, early endoscopy that is required at night or at weekends could put tremendous strain on resources. of the 1,498 studies identified through database searches, 25 studies were deemed appropriate for further scrutiny. we excluded 10 studies that focused on manage ment of patients after early endoscopy 7–16 and seven that did not report comparable outcome measures or only described management recommenda­ tions. 17–23 we, therefore, base our discussion in this review on three randomized, controlled trials24–26 and five retrospective cohort studies27–31 published between 1996 and 2007. we found great heterogeneity in the study design, follow­up period, endoscopic therapy and adjuvant therapy in these eight studies (table 1). time to endoscopy is defined as the time since the patients were admitted to the emergency room or presented to hos­ pital. one crucial factor in the evaluation of timing for early endoscopy is whether hemodynamically unstable patients were excluded from the studies. among the eight trials, two specifically excluded hemodynamically unstable patients.25,26 Five studies included both stable and unstable patients24,27–30 and one study included only hemodynamically unstable patients.31

nature reviews | gastroenterology & hepatology © 2009 Macmillan Publishers Limited. All rights reserved

institute of Digestive Disease and Department of Medicine and Therapeutics, Chinese University of Hong Kong, Hong Kong special Administrative region, People’s republic of China (K. K. F. tsoi, t. K. W. Ma, J. J. y. sung). Correspondence: J. J. Y. sung, Department of Medicine and Therapeutics, Prince of wales Hospital, shatin, New Territories, Hong Kong special Administrative region, People’s republic of China [email protected]

volume 6 | auGust 2009 | 463

reviews Key points ■ endoscopy within 12 h of presentation leads to increased use of endoscopic therapy for advanced stigmata of hemorrhage, which may not be necessary ■ No evidence exists for any clinical benefit of endoscopy performed within 12 h of presentation ■ early endoscopy within 12 h of presentation does not reduce the rebleeding rate or improve survival of patients ■ endoscopy within 24 h of presentation is recommended for management of upper gastrointestinal bleeding, because it has clinical benefits ■ endoscopy within 24 h of hospitalization aids risk assessment and reduces the length of hospital stay ■ in health-care practices where a 24 h endoscopy service is not available, endoscopy should be offered to patients the next day

Endoscopy within 2–3 h

in a prospective, randomized, controlled study from the us, lee et al.25 assessed 110 hemodynamically stable adult patients with acute nonvariceal uGiB. (varices are large veins that develop across the esophagus and stomach to bypass a blockage in the portal vein; they bleed easily.) 56 patients were randomly allocated to receive endoscopy within 1–2 h in the emergency depart­ ment, while 54 received endoscopy within 48 h of admis­ sion (controls). the proportion of patients taking either aspirin or nsaiDs was the same in both groups; these drugs can cause uGiB. slightly more patients in the early endoscopy group had a history of gastrointestinal bleed­ ing (36% versus 28%). of those who received early endo­ scopy, 32% had stigmata of recent hemorrhage, compared with 21% in the control group. no difference was evident in the incidence of recurrent bleeding, mortality, need for surgery, or transfusion requirements between the two groups (table 2). However, the mean hospital stay was significantly shorter in the early­endoscopy group (1.3 days versus 1.5 days; P = 0.004). of patients in the early endoscopy group, 46% were discharged without clinically significant adverse events in the subsequent 30­day follow­up period. the authors concluded that early endoscopy, within 2–3 h of presentation, aided risk stratification of patients and reduced the need for hospitalization. this study was the first randomized, controlled trial with a head­to­head comparison of endoscopy within 2 h and endoscopy within 48 h. more than half of the patients in this study were admitted during the night or weekend (63% in the early­endoscopy group versus 72% in the control group), which in a real clinical setting might be unmanageable. the study population may reflect selec­ tion bias because two­thirds of the patients who were ini­ tially diagnosed to have uGiB were excluded because of hemodynamic instability (37%), or known or suspected variceal bleeding (33%). in summary, we conclude that this study does not show a clinical benefit of endo­ scopy within 2 h of admission, but early endoscopy does significantly shorten the length of hospital stay. in a retrospective study from switzerland, schacher et al. 29 compared the clinical outcomes of 81 adult

patients with a bleeding peptic ulcer who received either endoscopy within 3 h or endoscopy within 48 h in the emergency department. Hemodynamic parameters were monitored until the time of endoscopy. all 43 patients in the early­endoscopy group underwent endoscopy within 3 h (median 2.1 h). of the 38 patients in the late­ endoscopy group, 84% underwent endoscopy within 24 h and the other 16% within 48 h (median 12.0 h). the two groups of patients were similar with respect to demographic characteristics, history of ulcer disease, previous uGiB, admission criteria and the risk scoring system used. the two risk scoring systems used clinical and endoscopic findings to predict the risk of recurrent uGiB (the Baylor score)32,33 or mortality (the rockall score).34 no statistically significant differences were evident in the incidence of recurrent bleeding, mortal­ ity, medical complications, need for surgery or median length of hospital stay between the two groups (table 2). interestingly, significantly more patients in the early­ endoscopy group had actively spurting ulcers (Forrest ia lesions; 19% versus 3%), and hence more patients underwent therapeutic endoscopy in this group (77% versus 47%). even so, the two groups did not differ with respect to recurrent bleeding or mortality. Baylor pre­ endoscopy scores were similar in both groups (5.5 in the early­endoscopy group versus 5.7 in the control group), which suggested that risk stratification based on clini­ cal parameters alone does not correlate with endoscopic criteria used to identify ulcers as high risk. in summary, this study also shows that endoscopy within 3 h of admis­ sion does not alter the clinical outcomes of patients. early endoscopy exposes additional cases of active bleeding, and hence increases the use of therapeutic endoscopy, which may not be necessary for patients with low­risk stigmata. the increased use of therapeutic endoscopy, therefore, does not translate into reduced rebleeding or improved mortality.

Endoscopy within 6–8 h

in a prospective, randomized, blinded, multicenter study, Bjorkman et al.26 randomly allocated 93 adult patients with acute nonvariceal uGiB to receive either endo­ scopy within 6 h or endoscopy within 48 h of admittance to the emergency department. Hemodynamically stable patients without severe comorbidity (defined as a pre­ endoscopic rockall score ≤5) were recruited in this study. mean age, sex distribution, and pre­endoscopic rockall scores were comparable between the two groups. similar to the previous studies, patients who received early endoscopy had significantly more active bleeding and/ or exposed vessels (32% in early­endoscopy group versus 20% in the control group), but mortality and rebleeding rates were similar in the two groups, irrespective of the timing of endoscopy (table 2). although early endoscopy may not affect clinical outcomes, it has potential implications for hospital resources. endoscopic diagnosis enables triage of patients for hospital admission or outpatient management.

464 | AUGUST 2009 | volUme 6

www.nature.com/nrgastro © 2009 Macmillan Publishers Limited. All rights reserved

reviews Table 1 | Features of eight studies included in this review study

Design

study period (follow-up)

patients

Mean age (years) (early endoscopy vs control)

exclusion of hemodynamically unstable patients

therapy

Lin et al. (1996)24

rCT

1993–1994 (2 months)

NA

66.4 (66.0 vs 66.8)

No

endoscopic: injection or thermal coagulation Adjuvant: H2 blockers

Cooper et al. (1998)27

retrospective, cohort

1991–1993 (NA)

NA

66

No

endoscopic: NA Adjuvant: none

Cooper et al. (1999)28

retrospective, cohort

1994 (NA)

stratified into low, medium and high-risk groups

66

No

endoscopic: thermal and/or injection of bleeding ulcers Adjuvant: none

Lee et al. (1999)25

rCT

12 months (30 days)

46% low risk in early-endoscopy group

49.0 (47 vs 51)

Yes

endoscopic: NA Adjuvant: aspirin, H2 blockers, PPis

Bjorkman et al. (2004)26

rCT

NA (30 days)

rockall score ≤5

54.5 (57 vs 52)

Yes

endoscopic: thermal, injection, or both Adjuvant: none

schacher et al. (2005)29

retrospective, cohort

1997–1998 (14 days)

stratified into low-risk and high-risk groups

69.4 (68 vs 71)

No

endoscopic: combined injection of fibrin glue plus epinephrine or epinephrine alone Adjuvant: none

Tai et al. (2007)30

retrospective, cohort

2004–2004 (NA)

High risk (defined by age, comorbidity, active bleeding, shock, etc.)

65.6 (64.0 vs 67.0)

No

endoscopic: epinephrine injection, thermocoagulation, hemoclip, argon plasma coagulation, and alcohol injection Adjuvant: none

Targownik et al. (2007)31

retrospective, cohort

1999–2004 (30 days)

High risk but no clear definition

63.1 (63.6 vs 62.6)

No

endoscopic: NA Adjuvant: pre-endoscopic and postendoscopic PPis

Abbreviations: NA, not available; rCT, randomized controlled trial.

low­risk patients may be safely discharged, while high­risk patients should be offered early endoscopic hemostasis therapy and subsequent intensive care if necessary. this approach may substantially decrease the burden on health­care resources and reduce costs, par­ ticularly in relation to management of low­risk patients. nevertheless, in Bjorkman’s study, no difference in 30­day resource use was shown between the early­endoscopy and control groups (table 2). these resources included the length of hospital stay, days spent in an intensive care unit, blood transfusion requirements and the need for additional intervention (surgery, repeated endoscopy, angiography or other radiologic intervention). one pos­ sible explanation was that although 40% of patients in the early­endoscopy group met the eligibility criteria for early discharge, only 8.5% of patients in this group were not hospitalized. the authors suggested that in everyday clinical practice, the results of urgent endoscopy may not affect the final decision by the attending physician because of medical and legal considerations. Besides, the duration of hospital stay may also be affected by the availability of hospital beds, differences in hospital prac­ tice and the discretion of clinicians who look after such patients. Hence, endoscopic triage will have no effect on resource utilization unless international guidelines on its use are adopted. in a retrospective cohort study from Canada, targownik et al.31 studied the clinical outcomes of 169 hemodynamically unstable adult patients with uGiB in

a 5­year period between 1999 and 2004. in all, 77 patients received early endoscopy within 6 h of presentation, whereas 92 patients underwent endoscopy between 6 h and 24 h (control group) of their presentation to hospital. the prevalence of high­risk endoscopic stigmata was significantly higher in the early­endoscopy group than the control group (57% versus 37%), which translated into a higher rate of endoscopic hemostasis (53% versus 37%). more patients in the early­endoscopy group received intravenous PPis after endoscopy (37% versus 23%). However, the overall incidence of adverse outcomes, including mortality, rebleeding, repeated endoscopy with hemostasis, need for surgery and readmission for uGiB, was similar irrespective of the timing of endoscopy (table 2). transfusion requirements and median lengths of hospital stay were no different. this study is unique in that the researchers recruited only patients who had either hypotension or tachycardia on their initial assessment of vital signs. Despite the fact that the investigators tried to limit the analysis to high­risk patients with uGiB, the mean pre­endoscopic rockall scores were not statistically different (3.2 in the early­endoscopy group versus 3.3 in the control group), and less than half of the patients in either group had a pre­endoscopic rockall score >5. this study may, therefore, have recruited hemodynamically unstable but otherwise low­risk patients. other limita­ tions related to this retrospective study include a relatively small sample size (for a 5­year study) and unmeasured factors that might have stimulated a patient to undergo

nature reviews | gastroenterology & hepatology © 2009 Macmillan Publishers Limited. All rights reserved

volume 6 | auGust 2009 | 465

reviews Table 2 | Clinical outcomes from eight studies that compared different timings of endoscopy for UGiB studies

number of patients (early endoscopy vs controla)

early endoscopy vs controla rebleeding (%)

surgery (%)

Death (%)

repeated endoscopy (number of patients)

Mean units of blood transfused

Mean or median days of hospital stay

110 (56 vs 54)

3.6 vs 5.6

3.6 vs 1.9

0.0 vs 3.7

4 vs 4

1.2 vs 1.1

1.3 vs 1.5b

81 (43 vs 38)

14.0 vs 15.8

9.3 vs 7.9

0.0 vs 0.0

NA

NA

5.9 vs 5.1

0.0 vs 0.0

4 vs 1

2.1 vs 1.5

4.0 vs 3.3

Early endoscopy defined as ≤2 to 3 h Lee et al. (1999)25 schacher et al. (2005)

29

Early endoscopy defined as ≤6–8 h Bjorkman et al. (2004)26

93 (47 vs 46)

NA

2.1 vs 2.2

Targownik et al. (2007)

169 (77 vs 92)

11.7 vs 8.7

10.4 vs 2.2

10.4 vs 6.5

10 vs 15

3.3 vs 2.9

4 vs 4

Tai et al. (2007)30

189 (88 vs 101)

7.1 vs 10.0

0.0 vs 0.0

1.1 vs 5.9

1 vs 1

3.5 vs 3.4

5.1 vs 6.0

5.7 vs 9.3

5.7 vs 7.4

1.9 vs 1.9

NA

NA

NA

31

b

Early endoscopy defined as ≤12 h Lin et al. (1996)24

107 (53 vs 54)

Early endoscopy defined as ≤24 h Cooper et al. (1998)27

3,801 (2,240 vs 1,561)

NA

3.8 vs 6.7c

2.9 vs 2.8

NA

NA

4 vs 6c

Cooper et al. (1999)28

909 (583 vs 326)

NA

NA

3.8 vs 3.4

NA

NA

5 vs 6.4c

Controls are defined as the group in which endoscopy was performed later than in the early-endoscopy group. bP