BMJ · the Rockall score and recently described Progetto Nazionale Emorragia Digestive (PNED) score require endoscopy before calculation.4,5 The Rockall score has been widely reported

Confidential: For Review O

nly

International multicentre prospective study comparing risk-

scoring systems for patients presenting with upper gastrointestinal bleeding

Journal: BMJ

Manuscript ID BMJ.2016.033358

Article Type: Research

BMJ Journal: BMJ

Date Submitted by the Author: 07-May-2016

Complete List of Authors: Stanley, Adrian; Glasgow Royal Infirmary, GI Unit Laine, Loren; Yale University School of Medicine, Digestive Diseases Dalton, Harry; Royal Cornwall Hospital, Gastroenterology Ngu, Jing; Singapore General Hospital Schultz, Michael; University of Otago Dunedin School of Medicine Abazi, Rozeta; Odense University Hospital, Department of Gastroenterology and Hepatology Zakko, Liam; Yale University School of Medicine, Digestive Diseases Thornton, Susan; Glasgow Royal Infirmary, GI Unit Wilkinson, Kelly; Royal Cornwall Hospital, Gastroenterology Khor, Cristopher; Singapore General Hospital Murray, Iain; Royal Cornwall Hospital, Gastroenterology

Laursen, Stig; Odense University Hospital, Department of Gastroenterology and Hepatology Dunne, Philip; Glasgow Royal Infirmary, GI Unit Fleming, Jude; Glasgow Royal Infirmary, GI Unit Holloway, Amelia; Royal Cornwall Hospital, Gastroenterology Froud, Oliver; Royal Cornwall Hospital, Gastroenterology Sutton, J; Royal Cornwall Hospital, Gastroenterology Michell, Nick; Royal Cornwall Hospital, Gastroenterology Pinchin, Emily; Royal Cornwall Hospital, Gastroenterology Ong, Hock; Singapore General Hospital Chin, Yung; Singapore General Hospital Koay, Doreen; Singapore General Hospital

Steiner, Tracey; Dunedin Hospital

Keywords: Upper Gastrointestinal bleeding, risk score, Glasgow Blatchford score, Rockall score, AIMS65, PNED

https://mc.manuscriptcentral.com/bmj

BMJ


nly

Page 1 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nlyInternational multicentre prospective study comparing risk-scoring

systems for patients presenting with upper gastrointestinal bleeding

A J Stanley (Consultant Gastroenterologist)1, L Laine (Professor of Medicine)2, H R

Dalton (Consultant Gastroenterologist)3, J H Ngu (Consultant

Gastroenterologist)4, M Schultz (Associate Professor of Gastroenterology)5,7, R

Abazi (Consultant Gastroenterologist)6, L Zakko (Fellow in Gastroenterology)2, S

Thornton (Research nurse)1, K Wilkinson (Medical student)3, CJL Khor

(Consultant Gastroenterologist)4, IA Murray (Consultant Gastroenterologist)3, SB

Laursen (Postdoctoral researcher in Gastroenterology)6

on behalf of the International GI bleeding consortium (listed below*)

1. Department of Gastroenterology, Glasgow Royal Infirmary, Glasgow, UK

2. Section of Digestive Diseases, Yale School of Medicine, New Haven, and VA

Connecticut Healthcare System, West Haven, Connecticut, USA

3. Gastrointestinal Unit, Royal Cornwall Hospital, Cornwall, UK

4. Department of Gastroenterology & Hepatology, Singapore General Hospital,

Singapore

5. Department of Medicine, Dunedin School of Medicine, University of Otago,

Dunedin, New Zealand

6. Department of Medical Gastroenterology, Odense University Hospital, Odense,

Denmark

7. Gastroenterology Unit, Southern District Health Board, Dunedin Hospital,

Dunedin, New Zealand

Page 2 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

2

*International GI bleeding consortium:

P Dunne1, J Fleming1, A Holloway3, O Froud3, J Sutton3, N Michell3, E Pinchin3, HS

Ong4, YK Chin4, DS Koay4, T Steiner7,

Correspondence to:

Dr Adrian J Stanley,

Consultant Gastroenterologist

Glasgow Royal Infirmary,

Castle St, Glasgow, UK,

G4 OSF

Ph: +44 141 211 4073

Email: [email protected] and [email protected]

Page 3 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

3

Abstract

Objectives: To compare the predictive accuracy and clinical utility of five risk

scoring systems in the assessment of patients presenting with upper

gastrointestinal bleeding (UGIB).

Design & setting: Consecutive patients with UGIB presenting to six large

hospitals in Europe, North America, Asia and Oceania over 12 months were

included. Prospective data were collected to calculate and compare the pre-

endoscopy scores: admission Rockall score (ARS), AIMS65 and Glasgow

Blatchford score (GBS); and the post-endoscopy scores: full Rockall score (FRS)

and PNED.

Participants & main outcome measures: 3012 patients were included. The scores

were compared in their ability to predict pre-defined clinical endpoints:

combined need for intervention or death, endoscopic therapy, mortality,

rebleeding, and length of hospital stay. Optimum score thresholds to identify

low- and high-risk patients were determined.

Results: GBS was best (AUROC=0·89) at predicting intervention or death

compared with PNED (0·71), AIMS65 (0·70), FRS (0·69) or ARS (0·69) (all

P<0·0001). GBS≤1 was the optimum threshold to predict survival without

intervention (sensitivity=98·6%, specificity=34·6%). GBS was better at

predicting endoscopic therapy (AUROC=0·75) than AIMS65 (0·63) and ARS

(0·61) (both P<0·0001). GBS≥7 was the optimum threshold to predict

endoscopic therapy (sensitivity=80%, specificity=57%). PNED (AUROC=0·79)

Page 4 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

4

and AIMS65 (0·78) were best at predicting mortality, with PNED superior to ARS

(0·76), FRS (0·72) (both P<0·05), and GBS (0·69; p<0·0001). Score thresholds of

PNED ≥4, AIMS65 ≥2, ARS ≥4, and FRS ≥5 were optimal at predicting death with

sensitivities 65·8-78·6% and specificities 65·0-65·3%. No score was helpful at

predicting rebleeding or length of stay.

Conclusions: GBS has high accuracy at predicting need for hospital-based

intervention or death. GBS≤1 appears the optimum threshold for directing

patients to outpatient management. AUROCs of scores for the other end-points

are <0·80, therefore their clinical utility for these outcomes appears limited.

Trial registration: The study was registered with ISRCTN (reference no.

16235737)

Funding: Data collection in Glasgow, Singapore and Dunedin was funded by a

Health Board endowment, Hospital Research and guthealthnetwork grant

respectively. The funders had no input to the study design, data collection or

interpretation, writing of report or submission for publication

Page 5 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

5

What is already known on this subject?

Several pre- and post-endoscopy risk scores have been reported to predict a

variety of endpoints in patients presenting with upper GI bleeding. Directing

patient care to outpatient management, or urgent endoscopy and higher-level

care, using low and high scores respectively on presentation has been suggested.

However the accuracy and generalizability, and the optimum thresholds of

scores to predict low and high-risk patients which may help direct care remain

unclear.

What this study adds

This international study suggests that the GBS is accurate (AUROC=0.89) in

predicting need for intervention or death at all centres, with GBS≤1 the optimal

threshold for directing low-risk patients to outpatient management with

sensitivity 98·6%, specificity 34·6%, PPV 96.6% and NPV 56.0%. All scores have

low predictive accuracy for other specific outcomes including endoscopic

therapy and mortality, therefore their clinical utility to direct management of

high-risk patients appears limited.

Widespread use of GBS in Emergency Departments of many countries may allow

a large minority of patients with upper GI bleeding to be safely managed as

outpatients.

Page 6 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

6

Introduction

Upper gastrointestinal bleeding (UGIB) is a common cause of admission to

hospital worldwide, with a UK incidence of 103-172 per 100 000 adults per year

and mortality of 8-14%.1-3 Many risk assessment scores have been developed to

predict clinically relevant outcomes, including mortality, need for hospital-based

intervention, rebleeding, and length of hospital stay.4-9 Several of these including

the Rockall score and recently described Progetto Nazionale Emorragia Digestive

(PNED) score require endoscopy before calculation.4,5 The Rockall score has

been widely reported and it has been suggested that PNED is superior to the

Rockall score, although it has not been externally validated.5 Other endoscopy-

based scores have been described, but are not appropriate for unselected

patients with UGIB, have not been externally validated, or have been shown to be

inferior to the scores above.10 However requiring endoscopy to calculate a score

may delay risk assessment in some healthcare settings, as there can be

significant delays in performing endoscopy out-of-hours or on weekends.11

There has been much recent interest in pre-endoscopic risk scores for UGIB,

which can be calculated shortly after presentation to hospital. The most widely

studied pre-endoscopic scores are the abbreviated “admission” Rockall score

(ARS), the Glasgow-Blatchford score (GBS) and the recently described AIMS65

score.7,8 These use clinical, haemodynamic and (for the latter two) readily

available laboratory parameters. It has been suggested that these scores can be

used to identify very low risk patients who could be managed as outpatients.12-18

Studies have also suggested these scores could identify higher risk patients who

Page 7 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

7

may require urgent endoscopy or management in high dependency or intensive

care units.19-20

Several studies have compared scores in their ability to predict various

outcomes.12,21-23 Although GBS appears useful in identifying very low-risk patient

in the UK, many studies have been small and single centre and no international

study has compared all the commonly used scores. The aim of this prospective

international study was to compare five endoscopic and pre-endoscopic risk

assessment scores in their ability to predict clinically relevant endpoints. In

addition, we assessed the clinical utility of these scores, by determining optimal

thresholds for identifying very low risk patients who could be managed as

outpatients, and higher risk patients who may require specific management

strategies aimed at improving outcome.

Page 8 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

8

Methods

The study was undertaken in Yale-New Haven Hospital (USA), Glasgow Royal

Infirmary (Scotland UK), Royal Cornwall Hospital Truro (England UK), Odense

University Hospital (Denmark), Singapore General Hospital (Singapore) and

Dunedin Hospital (New Zealand). Ethical approval was obtained from West of

Scotland Ethics committee (reference 14/WS/0012; project number: 145837)

and each centre obtained approval from their local research committee or

review board. Due to the observational, non-interventional nature of the study,

the ethical committee (and local committees and boards) agreed that individual

patient consent was not required. The sponsor was NHS Greater Glasgow and

Clyde Health Board, Scotland, UK.

The predetermined clinical endpoints were:

1. combined outcome of need for hospital-based intervention (red blood cell

transfusion, endoscopic therapy, interventional radiology or surgery) or death; 2.

endoscopic therapy; 3. 30-day mortality; 4. rebleeding within 7 days; 5. length of

hospital stay.

Patient management

Patients were included in the study if they presented to the hospital with

evidence of UGIB defined by haematemesis, coffee-ground vomiting, or melaena.

Patients who developed UGIB while already an inpatient for another reason were

not included.

Page 9 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

9

All patients presenting to each hospital with UGIB were initially assessed in the

emergency department or acute assessment unit. Two centres (Glasgow and

Odense) had a policy of non-admission for those with GBS≤1 and one centre

(Truro) for those with GBS≤2 and age <70years, unless required for other

reasons. Out-patient endoscopy was arranged in all three centres. Proton pump

inhibitors (PPIs) were not routinely given to all admitted patients prior to

endoscopy. Following endoscopy, the policy in all centres was to administer high

dose PPIs by intravenous bolus followed by infusion to patients with high-risk

ulcer stigmata who required endoscopic therapy and to other selected patients

depending on clinical judgement.15-18 For patients with suspected variceal

bleeding, the policy in all centres was to give intravenous vasopressors and

antibiotics prior to endoscopy.16,24

The endoscopic practice in all centres for patients with high-risk stigmata of non-

variceal bleeding was to administer injection therapy, thermal contact and/or

clips, but not adrenaline alone.15-18 Band ligation or injection of tissue glue +/-

transjugular intrahepatic portosystemic stent-shunt was performed in cases of

oesophageal or gastric variceal bleeding respectively.16,24 In line with recent

evidence and guidelines, the policy was to administer red cell transfusion at a

haemoglobin threshold of 7-8g/dL, or as guided by the clinician in severe

haemorrhage.24,25

Data collection

At each centre, data on consecutive, unselected patients presenting to the

hospital with UGIB were collected over a twelve-month period between March

Page 10 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

10

2014 and March 2015. Data were collected at each site by a dedicated research

nurse, doctor, or medical student. The data collected included patient

characteristics and haemodynamic and laboratory parameters at presentation

necessary to calculate the full Rockall (FRS), ARS, GBS, AIMS65 and PNED scores

(see supplementary Table 1). Endoscopic findings were recorded.

Interventions including blood transfusion, endoscopic therapy, interventional

radiology and surgery were recorded, as was rebleeding within 7 days as

previously defined (see supplementary Table 2),26 30-day mortality and length of

hospital stay. The collected data are shown in supplementary Table 3.

Patient involvement

There was no direct patient involvement in the development, design or conduct

of the study.

Statistical analysis

We compared each scores’ ability to predict the predetermined outcomes using

calculation of area under the receiver operator characteristics curves (AUROCs)

and 95% confidence intervals (CIs). AUROCs were compared using chi-square

tests. The optimal score thresholds to predict very low risk patients potentially

suitable for outpatient management were identified based on a sensitivity of

≥95%. For the other outcomes, optimal thresholds were identified using the

Youden index.27 In addition, the performance of the scores was assessed by

calculation of sensitivity, specificity, positive predictive values (PPV), negative

predictive values (NPV), and proportion classified as low- or high-risk patients.

Page 11 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

11

We calculated rates of transfusion, endoscopic therapy, and mortality for those

classified as low-risk. Pearsons chi-square test and Fischer’s exact test were used

to compare proportions. The Mann-Whitney U test and the Kruskal-Wallis

equality-of-populations rank test were used to compare medians.

We aimed for an adequate sample size to compare all the outcomes described

above. Existing data indicated that highest power was needed for comparative

analyses on mortality. Therefore assuming mortality AUROC of 0·86 for

AIMS658,20 and 0·78 for FRS10,21 with an alpha of 5%, power of 90%, rank

correlation of 0·5 and mortality of 5%,13 the required sample size was 2814

patients. A two-tailed significance level of 5% was used. Data analysis was

undertaken using STATA 11·0 (StataCorp, College Station, Texas, USA).

Data and publication

Only the local principal investigator and dedicated nurse, junior doctor or

medical student had access to the identifiable data from their centre, with the

anonymised data sent to SBL and AJS for central analysis. Data was collected by

RA, LZ, ST, KW, CJLK, SBL, PN, JF, AH, AF, JS, NM, EP, HSO, YKC, DSK and TS.

The paper was written by AJS and SBL with input from all co-authors who

approved the final manuscript. AJS is the study guarantor. Our report follows

STROBE guidelines.

Page 12 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

12

Results

Patient characteristics and baseline scores:

A total of 3012 patients were included in the study with 2868 (95·2%) followed

up for 30 days. Median age was 65 years and 58% were male. Patient

characteristics, endoscopic findings, interventions and outcome are shown in

Table 1. A total of 1348 (45%) patients needed hospital-based intervention or

died within 30 days. Endoscopic therapy was performed in 574 (19%), 37

(1·2%) required interventional radiological therapy or surgery and 144 (5%)

rebled within 7 days. Median length of stay was 3 days (95% CI: 0-16), and 30-

day mortality 6·9%. Overall mean GBS was 6·6 (95% CI: 0-14), mean AIMS65

was 1 (0-3), mean ARS 2·7 (0-5), mean FRS 3·9 (1-7) and mean PNED 3·0 (0-8).

These data by centre are shown in supplementary Table 4.

Page 13 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

13

Comparison of scores ability to predict outcomes:

Intervention or mortality

GBS had the highest discriminative ability (AUROC 0·89) at predicting need for

intervention or death when compared to PNED (AUROC 0·71; p<0·0001),

AIMS65 (AUROC 0·70; p<0·0001), FRS (AUROC 0·69; p<0·0001) and ARS

(AUROC 0·69; p<0·0001) (Table 2 and Fig. 1). GBS performed consistently well in

all centres with AUROCs ranging between 0·85-0·91 (supplementary Table 5).

However, performance of AIMS65 and FRS varied by centre, with AUROCs

ranging from 0·54-0·75 and 0·57-0·79 respectively. AUROCs for ARS among

centres were 0·59-0·72, and for PNED 0·70-0·76 (supplementary Table 5)

The ability of the three pre-endoscopic scores to predict very low risk patients

who did not require intervention and survived at optimal cut-offs is shown in

Table 3. GBS≤1 had a sensitivity of 98·6%, specificity 34·6%, PPV 96.6% and

NPV 56.0% for this combined end-point. Intervention or death was recorded in

3·4% of the 564 patients with GBS≤1, compared with 14% of the 436 patients

with ARS=0 and 25% of the 865 patients with AIMS65=0. The most common

intervention seen in those with GBS≤1 was blood transfusion, with endoscopic

therapy performed in 1·4%, and overall mortality 0·4%. Compared with GBS ≤1,

patients with AIMS65=0 and ARS=0 had higher rates of both endoscopic

intervention and transfusion (p<0·001; Table 3). The proportion of low risk

patients defined by GBS≤1 varied by site, from 9% in Dunedin to 31% in

Glasgow.

Page 14 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

14

Endoscopic therapy

In predicting need for endoscopic therapy, GBS (AUROC 0·75) performed better

than the other pre-endoscopic scores: AIMS65 (AUROC 0·63; p<0·0001) and ARS

(AUROC 0·61; p<0·0001; see Fig. 2). The GBS performed equally well in

predicting this endpoint in all centres (Supplementary Table 5). A threshold of

GBS ≥7 was best at predicting endoscopic therapy with sensitivity of 80·4%,

specificity of 57·4%, PPV of 31·3% and NPV of 92·4% (Table 4).

Mortality

PNED (AUROC 0·79) and AIMS65 (AUROC 0·78) had similar discriminative

abilities for predicting 30-day mortality (Table 2 and Figure 3). PNED was better

at predicting mortality than ARS (AUROC 0·76; p=0·03), FRS (AUROC 0·72;

p=0·03) and GBS (AUROC 0·69; p<0·0001). AIMS65 was better at predicting

mortality than GBS (p<0·0001), but the difference in AUROC compared with ARS

and FRS was not statistically significant. PNED ≥4, AIMS65 ≥2, ARS ≥4, FRS ≥5

and GBS ≥5 were the best score thresholds at predicting 30-day mortality. The

predictive ability of all five scores to identify high risk patients are shown in

Table 5.

Rebleeding and length of stay

PNED, which includes rebleeding in the score, was best (AUROC 0·85) at

predicting rebleeding, compared with GBS (AUROC 0·70; p<0·0001), FRS

(AUROC 0·63; p<0·0001), AIMS65 (AUROC 0·62; p<0·0001), and ARS (AUROC

0·62; p<0·0001). The GBS was better at predicting rebleeding compared with

AIMS65 (p=0·0003) and ARS (p=0·0002), but not FRS. All risk scores had

Page 15 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

15

relatively poor discriminative abilities for predicting length of stay >3 days, with

PNED AUROC=0·70, GBS and ARS AUROCs=0·69, FRS AUROC=0·65 and AIMS65

AUROC=0·64.

Missing values:

There were missing values for GBS (n=80), AIMS65 (n=511), Admission Rockall

score (n=43), Full Rockall score (n=1000), PNED (n=178), comorbidity (n=1),

systolic blood pressure (n=41), pulse (n=38), haemoglobin (n=28), findings at

endoscopy (n=2), true low-risk status (n=28), total units of blood transfused

(n=23), performance of endoscopic therapy (n=20), performance of surgery or

embolisation (n=5), rebleeding (n=51), and mortality (n=1).

Page 16 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

16

Discussion

This international multicentre study shows that following UGIB, GBS is a very

accurate risk score at all sites for predicting need for clinical intervention or

death. Although AIMS65 and PNED are the best scores for predicting mortality

and GBS best for predicting endoscopic therapy, accuracy and clinical utility is

relatively low for these endpoints.

Strengths of our study include its international multicentre design and large

number of consecutive patients studied. By assessing the five pre- and post-

endoscopy scores which appeared most promising for clinical use, we could

investigate the optimum way to risk assess patients with UGIB, both very early

after presentation to potentially direct to outpatient management, and also after

endoscopic diagnosis and therapy. Weaknesses of our study include the fact that

we excluded patients who were already inpatients when they developed UGIB.

These patients are already known to be at high risk of poor outcome. In addition,

31% of patients did not undergo endoscopy. In three centres, there was a

guideline to not admit patients with either GBS≤1 or ≤2 and age <70 years, and

many did not attend for outpatient endoscopy. In addition, some clinicians

decided an endoscopy was not required following an apparent trivial bleed, or

was deemed inappropriate for some very ill patients with multiple

comorbidities. Interestingly the UK national UGIB audit reported that 26% of

patients notified to gastroenterologists did not have an inpatient endoscopy.11

We do not believe this had a major effect on our study findings, as 95% patients

were followed up for 30 days after presentation.

Page 17 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

17

Published guidelines suggest outpatient management for patients with GBS of

zero.15-18 However some authors have suggested this low-risk threshold could be

extended to ≤1 or ≤2.13,14 We identified GBS ≤1 as the optimum score threshold

to identify patients who would not require intervention or die with sensitivity of

98·6%. There was a consistent high accuracy of this threshold across

international sites, which comprised 19% of all patients in our study, compared

with the 8·6% patients who had GBS of zero. Of the patients with GBS≤1, all

cause mortality was only 0·4%. This compares favourably with accepted low risk

thresholds of commonly used scores for other medical conditions including

CURB65 for chest infection and the Pulmonary Embolism Severity Index.28,29

Other studies have reported more accurate prediction of death, rebleeding or

length of stay using these scores, but our study has the advantages of a

multicentre design and large numbers and suggests these outcomes are not well

predicted by the studied scores.

The median length of hospital admission for a patient presenting with UGIB in

the UK is 4-5 days,3,12 and a recent study showed the mean in-hospital cost to be

£2458 per patient.30 Use of an accurate score within Emergency Departments or

Acute Assessment units, to identify very low risk patients with UGIB not

requiring admission has clear benefits. Interestingly, the proportion of very low-

risk patients defined by GBS≤1 varied by site. The differences are probably due

to different models of healthcare provision, including extent of primary care

services. At the other end of the spectrum of severity of UGIB, early identification

of high-risk patients who may benefit from urgent endoscopy or higher-level

care could help guide management. An observational study suggested that

Page 18 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

18

patients with GBS>12 have decreased mortality if endoscopy is undertaken <13

hours after presentation.19 Our data suggest that GBS≥7 and AIMS65 have the

highest combination of sensitivity and specificity for predicting endoscopic

therapy or 30-day mortality respectively, but the PPV for both is very low.

Therefore the clinical utility of these scores to direct care in high-risk patients

seems very limited.

In conclusion, GBS is accurate at predicting need for clinical intervention or

death in patients with UGIB in all countries studied. GBS≤1 is the optimum score

threshold for identifying very low risk patients suitable for outpatient

management. GBS≥7 is best at predicting need for endoscopic therapy and

PNED≥4 and AIMS65≥2 best at predicting mortality, although accuracy in

predicting these end-points is relatively low. No score appears accurate at

predicting rebleeding or length of stay. This information can help direct

management of very low risk patients with UGIB, but further studies using these,

or new scores, are required to clarify their role in directing management of

higher risk patients.

Page 19 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

19

References

1. Rockall TA, Logan RFA, Devlin HB, Northfield TC. Incidence of and mortality from

acute upper gastrointestinal haemorrhage in the UK. BMJ 1995; 311: 222-26.

2. Blatchford O, Davidson LA, Murray WR., Blatchford M, Pell J. Acute upper

gastrointestinal haemorrhage in west of Scotland: case ascertainment study. BMJ

1997; 315: 510-14.

3. Hearnshaw, SA, Logan RFA, Lowe D, Travis SP, Murphy MF, Palmer KR. Acute

upper gastrointestinal bleeding in the UK: patient characteristics, diagnoses and

outcomes in the 2007 UK audit. Gut 2011;60:1327-1335.

4. Rockall TA, Logan RFA, Devlin HB, Northfield TC. Risk assessment after acute

upper gastrointestinal haemorrhage. Gut 1996;38:316-321 (PMID:8675081)

5. Marmo R, Koch M, Cipoletta L, et al. For the Italian registry on upper

gastrointestinal bleeding. Predicting mortality in non-variceal upper

gastrointestinal bleeders: validation of the Italian PNED score and prospective

comparison with the Rockall score. Am J Gastroenterol 2010;105:1284-91

(PMID: 20051943)

6. Saeed ZA, Winchester CB, Carolyn B, Michaeletz PA, Woods KL, Graham DY. A

scoring system to predict rebleeding after endoscopic therapy of non-variceal

upper gastrointestinal hemorrhage. Am J Gastroenterol 1993;88:1842-9 (PMID:

8237930)

7. Blatchford O, Murray WR, Blatchford M. A risk score to predict need for

treatment for upper gastrointestinal haemorrhage. Lancet 2000;356:1318-21

(PMID 11073-21)

8. Saltzman JR, Tabak YP, Hyett BH, Sun X, Travis AC, Johannes RS. A simple risk

score accurately predicts in hospital mortality, length of stay and cost in acute

upper GI bleeding. Gastrointest Endosc. 2011;74:1215-24.

Page 20 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

20

9. Das A, Ben-Menachem T, Farooq FT,et al. Artificial neural network as a

predictive instrument in patients with acute non-variceal upper gastrointestinal

hemorrhage. Gastroenterol 2008;134:65-74 (PMID: 18061180)

10. Laursen SB, Hansen JM, De Muckadell OBS. The Glasgow Blatchford score is the

most accurate assessment of patients with upper gastrointestinal hemorrhage.

Clin Gastroenterol Hepatol 2012:10:1130-1135

11. Hearnshaw SA, Logan RFA, Lowe D, Travis SPL, Murphy MF, Palmer KR. Use of

endoscopy for management of acute upper gastrointestinal bleeding in the UK:

results of a nationwide audit. Gut 2010;59:1022-1029.

12. Stanley AJ, Ashley D, Dalton HR, et al. Outpatient management of patients with

low-risk upper-gastrointestinal haemorrhage: multicentre validation and

prospective evaluation. Lancet 2009;373:42-47.

13. Laursen SB, Dalton HR, Murray IA, et al. Performance of new thresholds of the

Glasgow Blatchford score in managing patients with upper gastrointestinal

bleeding. Clin Gastroenterol Hepatol 2015;13:115-121.

14. McLaughlin C, Vine L, Chapman L, et al. The management of low-risk primary

upper gastrointestinal haemorrhage in the community: a 5-year observational

study. Eur J Gastroenterol Hepatol 2012;24:288-293.

15. Barkun AN, Bardou M, Kuipers EJ, et al. International consensus

recommendations on the management of patients with nonvariceal upper

gastrointestinal bleeding. Ann Intern Med. 2010;152:101-113.

16. National Institute for Health and Clinical Excellence 2012 Acute Upper

Gastrointestinal Bleeding: management. Clinical Guideline 141. London: NICE.

17. Gralnek IM, Dumoncaeu JM, Kuipers, Lanas A, et al. Non-variceal upper

gastrointestinal haemorrhage: ESGE Guidelines. Endoscopy 2015:47:1-46

18. Laine L, Jensen DM. Management of patients with ulcer bleeding, ACG Practice

guidelines. Am J Gastroenterol 2012;107:345-60

Page 21 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

21

19. Lim, LG, Ho KY, Chan YH, et al. Urgent endoscopy is associated with lower

mortality in high-risk but not low-risk nonvariceal upper gastrointestinal

bleeding. Endoscopy 2011;43:300-306.

20. Hyett BH, Abougergi MS, Charpentier JP, et al. The AIMS65 score compared with

the Glasgow-Blatchford score in predicting outcomes in upper GI

bleeding. Gastrointest Endosc 2013;77:551-557.

21. Stanley AJ, Dalton HR, Blatchford O, et al. Multicentre comparison of the Glasgow

Blatchford and Rockall scores in the prediction of clinical end-points after upper

gastrointestinal haemorrhage. Aliment Pharmacol Therapeut 2011;34:470-475.

22. Pang SH, Ching YJL, Lau JYW, Sung JJY, Graham DY, Chan FKL. Comparing the

Blatchford and pre-endoscopic Rockall score in predicting the need for

endoscopic therapy in patients with upper GI hemorrhage. Gastrointest

Endosc 2010;71:1134-1140.

23. Yaka E, Yilmaz S, Ozgur DN, Pekdemir M. Comparison of the Glasgow-Blatchford

and AIMS65 Scoring Systems for Risk Stratification in Upper Gastrointestinal

Bleeding in the Emergency Department. Acad Emerg Medicine 2015;22:22-30.

24. De Franchis R. Revising consensus on portal hypertension; Report of the Baveno

V consensus workshop on methodology of diagnosis and therapy of portal

hypertension. J Hepatol 2010; 53: 762-768

25. Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper

gastrointestinal bleeding. N Engl J Med 2013;368:11-21

26. Laine L, Spiegel B, Rostom A, et al. Methodology for randomized trials of patients

with nonvariceal upper gastrointestinal bleeding: recommendations from an

international consensus conference. Am J Gastroenterol 2010;105;540-50.

27. Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32-5

Page 22 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

22

28. Lim W, van der Eeren MM, Laing Boersma W, et al. Defining community acquired

pneumonia severity on presentation to hospital: an international derivation and

validation study. Thorax 2003;58:377-382

29. Aujesky D, Roy PM, Vershuren F et al. Outpatient versus inpatient treatment for

patients with acute pulmonary embolism: an international, open-label,

randomised, non-inferiority trial. Lancet 2011;178:41-48

30. Campbell HE, Stokes EA, Bargo D, et al. Costs and quality of life associated with

acute upper gastrointestinal bleeding in the UK: cohort analysis of patients in a

cluster randomised trial. BMJ open 2015;5.4:e007230.

Page 23 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

23

Table 1. Characteristics, treatment and outcome of patients (n=3012) Age (years, median, [95% CI]) 65 [24-90]

Sex (male) 1750 (58)

Comorbidity

- Ischemic heart disease

- Liver disease

- Renal failure

- Any malignancy

580 (19)

453 (15)

266 (9)

430 (14)

Systolic blood pressure (mmHg, median, [95% CI]) 125 [90-170]

Pulse (median [95% CI]) 89 [61-126]

Haemoglobin (median [95% CI]) 112 [58-162]

Findings at endoscopy:

- Normal findings

- Gastric/duodenal ulcer

- Variceal bleeding

- Upper GI-cancer

Not endoscoped

297 (14)

572 (28)

143 (7)

70 (3)

937 (31)

Treatment

- Number of transfusions (mean, [95% CI])

- Endoscopic treatment

- Surgery/embolisation

1.3 [0-6]

574 (19)

37 (1.2)

Outcome

- No need for intervention or death

- Rebleeding

- Mortality

1636 (55)

144 (5)

2087(7)

Score (mean, [95% CI])

- GBS

- AIMS65

- Adm. Rockall score

- Full Rockall score

- PNED

6.6 [0-14]

1.0 [0-3]

2.7 [0-5]

3.9 [1-7]

3.0 [0-8]

Data are number of patients (%), unless otherwise stated.

CI: confidence intervals; GI: Gastrointestinal

Page 24 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

24

Table 2. Discriminative ability of evaluated scoring systems

Outcome AUROC [95% CI]

Intervention or death

- GBS

- AIMS65

- ARS

- FRS

- PNED

0.89 [0.87-0.90]

0.70 [0.68-0.72]

0.69 [0.67-0.71]

0.69 [0.67-0.71]

0.71 [0.70-0.73]

Need for endoscopic therapy

- GBS

- AIMS65

- ARS

0.75 [0.73-0.77]

0.63 [0.60-0.65]

0.61 [0.59-0.64]

Rebleeding

- GBS

- AIMS65

- ARS

- FRS

- PNED

0.70 (0.66-0.74)

0.62 (0.57-0.66)

0.62 (0.57-0.66)

0.63 (0.58-0.68)

0.85 (0.83-0.88)

Mortality

- GBS

- AIMS65

- ARS

- FRS

- PNED

0.69 [0.66-0.72]

0.78 [0.75-0.81]

0.76 [0.73-0.79]

0.72 [0.68-0.77]

0.79 [0.76-0.82]

AUROC: Area under receiver operator characteristics curves; CI: Confidence Interval

Page 25 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Confidential: For Review Only 25

Table 3. Outcomes at optimal thresholds of GBS, AIMS65 and ARS in prediction of need for intervention or death

System Cut-off

Low-risk

n(%)a

Sens.

(%)

Spec.

(%)

PPV

(%)

NPV

(%)

Transfusion

n(%)

Endo-therapy n(%)

Surgery/

Int. radiol

n(%)b

Mortality

n(%)c

GBS ≤1 564 (19.2) 98.6 34.6 96.6 56.0 10 (1.8) 8 (1.4) 2 (0.4) 2 (0.4)

AIMS65 =0 865 (34.6) 81.5 49.9 74.7 59.9 165 (19) 107 (12) 9 (1.0) 6 (0.7)

ARS =0 436 (14.7) 95.6 23.4 86.5 50.9 41 (9.4) 29 (6.7) 0 (0) 1 (0.2)

a Number (%) of classified low-risk patients according to risk-scoring system

b Number (%) of patients needing surgery, or arterial embolisation among patients classified as low-risk

c Number (%) of patients dying during admission among patients classified as low-risk

PPV: Positive predictive value; NPV: Negative predictive value

Page 26 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

26

Table 4. Need for endoscopic therapy at optimal score thresholds of each of the

pre-endoscopic scores: GBS, AIMS65 and ARS

System Cut-off

Number of classified

high-risk patients (%)

Sensitivity

(%)

Specificity

(%)

PPV

(%)

NPV

(%)

GBS ≥7 1456 (50) 80.4 57.4 31.3 92.4

AIMS65 ≥1 1619 (65) 79.7 38.7 25.9 87.6

ARS ≥3 1686 (57) 69.8 45.9 23.5 86.5


Page 27 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

27

Table 5. 30-day mortality at optimal score thresholds of GBS, AIMS65, PNED and

FRS

System

Cut-off

High-risk

n(%)a

Sensitivity

(%)

Specificity

(%)

PPV

(%)

NPV

(%)

AIMS65 ≥2 672 (26.9) 65.8 76.2 18.0 96.6

PNED ≥4 1065 (37.6) 77.3 65.3 14.1 97.5

ARS ≥4 1130 (38.1) 78.6 65.0 14.3 97.6

FRS ≥5 815 (40.5) 74.0 61.7 11.1 97.3

GBS ≥5 1812 (61.8) 88.7 40.2 9.9 97.9

a Number (%) of classified high-risk patients according to risk-scoring system


Page 28 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

28

Figure 1.

Comparisons of scores in prediction of need for any intervention (transfusion, endoscopic therapy, interventional-radiology or surgery) or 30-day mortality

Page 29 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

29

Figure 2.

Comparisons of pre-endoscopic scores in prediction of need for endoscopic therapy

Page 30 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

30

Figure 3.

Comparisons of scores in prediction of 30-day mortality

Page 31 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

31

Supplementary Table 1.

Risk scores evaluated in study.

1 a) Full Rockall score (FRS; ref 4)

1 b) Admission Rockall score (ARS) excludes parameters marked with *

0 1 2 3

Age (yrs) <60 60-79 ≥80 -

Shock No shock

(SBP≥100mmHg

and pulse

<100bpm)

Pulse

>100bpm

SBP

<100mmHg

-

Co-morbidity No major

comorbidity

- Cardiac

failure,

Ischaemic

heart disease,

Any major

comorbidity

Renal failure,

Liver failure,

Metastatic

cancer

Diagnosis * Mallory Weiss

tear, no lesion,

no stigmata

All other

diagnoses

Malignancy of

upper GI tract

-

Major

Stigmata *

None or dark

spot only

- Blood in

upper GI

tract,

adherent clot,

visible or

spurting

vessel

-

SBP: Systolic blood pressure

Page 32 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

32

1 c) Glasgow Blatchford Score (GBS; ref 7)

Admission Criteria Score

Haemoglobin (g/dl)

Male

12 - <13

10 - <12

<10

Female

10 - <12

<10

1

3

6

1

6

Urea (mmol/L)

6.5 - <8

8.0 – 10

10 - <25

≥ 25

2

3

4

6

Systolic Blood pressure (mmHg)

100 – 109

90 -99

<90

1

2

3

Pulse (bpm)

≥ 100 bpm

1

Melaena 1

Syncope 2

Cardiac failure 2

Liver failure 2

Page 33 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

33

1 d) AIMS65 (ref 8)

Admission Criteria Score

Albumin <30g/l

1

*INR >1.5

1

Mental State: +GCS <14

1

Systolic Blood Pressure <90mmHg

1

Age >65yrs

1

*International Normalised Ratio

+Glasgow Coma Scale

Page 34 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

34

1 e) Progetto Nazionale Emorragia Digestive score (PNED; ref 5)

Score: 1 2 3 4

Risk factors: *ASA 3,

Time to

admission <8hrs

Hb ≤7g/dL,

Age ≥80yrs,

Renal failure

Rebleeding,

ASA 4,

Neoplasia,

Liver cirrhosis

Failure of

endoscopic

therapy

*ASA: American Society of Anesthesiologists score

Page 35 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

35


Criteria for rebleeding (ref. 26)

Any one of:

• Haematemesis (or bloody nasogastric aspirate) > 6 hours post endoscopy

• Melaena after normalisation of stool colour

• Haematochezia after normalisation of stool colour or after melaena

• Tachycardia (pulse≥110/minute) or hypotension (systolic blood pressure

≤90mmHg) after≥1 hour of haemodynamic stability (ie no tachycardia or

hypotension) in the absence of an alternative explanation such as sepsis,

cardiogenic shock or medications

• Hb drop of ≥2g/dl after two consecutive stable Hb values (<0.5g/dl decrease ≥3

hours apart

• Tachycardia or hypotension that does not resolve within 8 hours after index

endoscopy despite appropriate resuscitation (in the absence of an alternative

explanation) associated with persistent melaena or hematochezia

• Persistent dropping Hb of >3g/dl in 24 hours associated with persistent melaena

or hematochezia

Page 36 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

36

Supplementary Table 3

Registered data

Characteristics of patients - Age - Sex

Comorbidity - ASA-score - Ischaemic heart disease - Cardiac failure - Renal failure - Liver disease - Liver cirrhosis - Liver failure - Any malignancy - Disseminated malignancy - Any major comorbidity - Previous surgery for peptic ulcer

Medication use - NSAIDs - Aspirin - ADP-receptor inhibitors - Anticoagulants

Symptoms - Coffee ground vomiting - Haematemesis - Melaena - Haematochezia - Syncope - Altered mental status

Blood tests - Haemoglobin - Urea - Creatinine - INR - Albumin

Circulatory status at hospital admission - Systolic blood pressure - Pulse

Endoscopy - Time from admission to performance of endoscopy - Findings at endoscopy - Performance of endoscopic therapy - Bleeding uncontrolled by endoscopic therapy - Performance of endoscopic therapy impossible

Other treatment - Number of received blood transfusions - Performance of surgery or interventional radiological intervention

Rebleeding - Development of rebleeding

Page 37 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

37

- Treatment of rebleeding Mortality

- Bleeding-related mortality - Non-bleeding-related mortality

Other data (related to hospitalisation and follow-up) - Time from development of symptoms to hospital admission - Admission overnight/weekend - Length of ICU admission - Patient discharged without performance of endoscopy - Endoscopy performed on an outpatient basis - Time of follow-up

ADP-receptor inhibitors : Adenosine diphosphate receptor inhibitors

ASA: American Society of Anesthesiologists

ICU: Intensive care unit

INR: International normalised ratio

NSAIDs: Nonsteroidal anti-inflammatory drugs

Page 38 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



Characteristics, treatment and outcome of patients by centre

Site Yale Glasgow Truro Odense Singapore Dunedin Overall

n 799 600 542 539 433 99 3012

Age (years, median, [95% CI]) 63 [24-90] 55 [22-87] 69 [24-91] 72 [31-93] 66 [27-87] 71 [22-88] 65 [24-90]

Sex (male) 432 (54) 371 (62) 314 (58) 301 (56) 274 (63) 58 (59) 1750 (58)

Comorbidity

- Ischemic heart disease

- Liver disease

- Renal failure

- Any malignancy

181 (23)

151 (19)

56 (7)

157 (20)

95 (16)

127 (21)

77 (13)

75 (13)

85 (16)

60 (11)

40 (7)

75 (14)

117 (22)

51 (9)

25 (5)

47 (9)

81 (19)

48 (11)

62 (14)

61 (14)

21 (21)

16 (16)

6 (6)

15 (15)

580 (19)

453 (15)

266 (9)

430 (14)

Systolic blood pressure (mmHg) median, [95% CI]) 125 [92-170] 127 [92-163] 129 [90-173] 125 [86-169] 120 [83-164] 127 [90-171] 125 [90-170]

Pulse (median [95% CI]) 87 [60-125] 92 [62-133] 90 [63-126] 90 [62-124] 91 [62-125] 86 [56-115] 89 [61-126]

Haemoglobin (median [95% CI]) 107 [60-161] 128 [66-168] 124 [64-167] 102 [52-153] 101 [54-156] 97 [54-146] 112 [58-162]

Findings at endoscopy

- Normal findings

- Erosive diseasea

- Mallory-Weiss tear

47 (10)

92 (20)

14 (3)

45 (13)

91 (27)

19 (6)

62 (16)

177 (45)

8 (2)

91 (18)

133 (27)

13 (3)

38 (12)

83 (26)

6 (2)

14 (23)

7 (11)

3 (5)

297 (14)

583 (28)

63 (3)

Page 39 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


- Gastric/duodenal ulcer

- Variceal bleeding

- Upper GI-cancer

Not endoscoped

150 (32)

30 (6)

16 (3)

335 (42)

71 (21)

37 (11)

12 (4)

269 (45)

53 (13)

19 (5)

6 (2)

147 (27)

150 (30)

33 (7)

14 (3)

41 (8)

128 (40)

18 (6)

21 (6)

109 (25)

20 (32)

7 (11)

1 (2)

36 (37)

572 (28)

143 (7)

70 (3)

937 (31)

Treatment

- Number of transfusions (mean, [95% CI])

- Endoscopic treatment

- Surgery/embolisation

1.1 [0-5]

128 (16)

22 (2.8)

0.7 [0-4]

89 (15)

1 (0.2)

0.9 [0-4]

85 (16)

8 (1.5)

2.3 [0-8]

166 (32)

1 (0.2)

1.7 [0-6]

88 (20)

5 (1)

1.2 [0-6]

18 (18)

0 (0)

1.3 [0-6]

574 (19)

37 (1.2)

Outcome

- Actual low-risk patientsb

- Rebleeding

- Mortality

440 (55)

42 (5)

48 (6)

420 (73)

12 (2)

38 (6)

331 (61)

22 (4)

48 (9)

216 (40)

35 (7)

46 (9)

178 (41)

29 (7)

21 (5)

51 (52)

4 (4)

6 (6)

1636 (55)

144 (5)

207 (7)

Score (mean, [95% CI])

- GBS

- AIMS65

- Adm. Rockall score

- Full Rockall score

- PNED

7.0 [0-14]

1.0 [0-3]

3.2 [0-5]

4.5 [2-7]

3.9 [0-10]

4.8 [0-13]

0.9 [0-3]

2.4 [0-5]

3.8 [1-6]

2.7 [0-7]

5.7 [0-14]

0.9 [0-2]

2.6 [0-5]

3.6 [1-6]

2.7 [0-7]

7.8 [0-14]

1.2 [0-3]

2.7 [0-5]

3.8 [1-7]

2.7 [0-7]

7.5 [1-14]

0.9 [0-3]

2.6 [0-5]

3.5 [1-7]

2.5 [0-7]

8.1 [0-14]

1.2 [0-2]

2.8 [0-5]

3.8 [1-6]

2.8 [0-8]

6.6 [0-14]

1.0 [0-3]

2.7 [0-5]

3.9 [1-7]

3.0 [0-8]

Data are number of patients (%), unless otherwise stated

Page 40 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


aoesophagitis, gastritis or duodenitis

bNumber of patients without need of intervention (transfusion, endoscopic treatment, surgery, or embolization) who survive

CI: Confidence Interval

GI: Gastrointestinal

Page 41 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



Discriminative ability of evaluated scoring systems by centre

Site Yale Glasgow Truro Odense Singapore Dunedin Overall P*

Intervention or death

(AUROC [95% CI])

- GBS

- AIMS65

- ARS

- FRS

- PNED

0.89 [0.86-0.91]

0.70 [0.66-0.74]

0.71 [0.68-0.74]

0.65 [0.60-0.70]

0.72 [0.69-0.76]

0.86 [0.82-0.89]

0.71 [0.67-0.76]

0.66 [0.61-0.71]

0.57 [0.50-0.64]

0.74 [0.70-0.79]

0.86 [0.83-0.89]

0.67 [0.63-0.72]

0.72 [0.68-0.76]

0.73 [0.68-0.78]

0.75 [0.70-0.79]

0.91 [0.88-0.93]

0.75 [0.71-0.79]

0.72 [0.67-0.76]

0.79 [0.75-0.83]

0.74 [0.70-0.79]

0.89 [0.86-0.92]

0.67 [0.61-0.72]

0.68 [0.63-0.73]

0.67 [0.61-0.74]

0.70 [0.65-0.75]

0.85 [0.77-0.92]

0.54 [0.42-0.66]

0.59 [0.48-0.70]

0.59 [0.44-0.73]

0.76 [0.67-0.86]

0.89 [0.87-0.90]

0.70 [0.68-0.72]

0.69 [0.67-0.71]

0.69 [0.67-0.71]

0.71 [0.70-0.73]

NS

0.007

NS

<0.001

NS

Need for endo-therapy

(AUROC [95% CI])

- GBS

- AIMS65

- ARS

0.73 [0.69-0.77]

0.62 [0.56-0.67]

0.67 [0.62-0.72]

0.79 [0.74-0.84]

0.67 [0.61-0.73]

0.60 [0.54-0.67]

0.76 [0.71-0.81]

0.60 [0.54-0.67]

0.62 [0.56-0.68]

0.75 [0.70-0.79]

0.64 [0.59-0.69]

0.61 [0.56-0.66]

0.67 [0.61-0.73]

0.53 [0.46-0.60]

0.58 [0.52-0.64]

0.76 [0.64-0.87]

0.47 [0.31-0.63]

0.64 [0.49-0.80]

0.75 [0.73-0.77]

0.62 [0.60-0.65]

0.61 [0.59-0.64]

NS

0.02

NS

Mortality

(AUROC [95% CI])

- GBS

0.74 [0.67-0.81]

0.76 [0.69-0.84]

0.71 [0.65-0.77]

0.59 [0.51-0.67]

0.69 [0.60-0.79]

0.60 [0.33-0.86]

0.69 [0.66-0.72]

0.048

Page 42 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


- AIMS65

- ARS

- FRS

- PNED

0.79 [0.73-0.85]

0.78 [0.73-0.84]

0.73 [0.62-0.83]

0.82 [0.77-0.87]

0.85 [0.80-0.89]

0.74 [0.67-0.81]

0.66 [0.56-0.75]

0.81 [0.73-0.89]

0.73 [0.66-0.80]

0.79 [0.73-0.85]

0.76 [0.68-0.85]

0.79 [0.73-0.85]

0.75 [0.68-0.82]

0.72 [0.65-0.78]

0.72 [0.64-0.79]

0.77 [0.72-0.83]

0.86 [0.79-0.94]

0.83 [0.77-0.89]

0.84 [0.75-0.94]

0.82 [0.73-0.91]

0.63 [0.38-0.88]

0.62 [0.33-0.90]

0.94 [0.85-1.00]

0.79 [0.56-1.00]

0.78 [0.75-0.81]

0.76 [0.73-0.79]

0.72 [0.68-0.77]

0.79 [0.76-0.82]

0.014

NS

0.003

NS

AUROC: area under the receiver operator characteristics curves; CI: Confidence Interval; * Difference in discriminative performance between centres

Page 43 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

43

Supplemental file:

Study protocol

International multicentre study comparing risk scoring systems for

patients presenting with upper gastrointestinal bleeding.

Study Protocol

1. General information

1.1 Running Title

“International study on scoring systems for upper GI bleeding”

1.2 Study details

Protocol version 1.4

Date: 20.01.14

REC number: 145837/544932/1/939

Sponsor’s protocol (R&D) reference number: GN13GA571

Sponsor: NHS Greater Glasgow & Clyde

Funding: There is no specific funding for this clinically designed and led study.

Each centre will access support from existing local audit/research nurses and/or

a junior doctor to collect data.

This study will be performed according to the Research Governance Framework

for Health and Community Care (second edition, 2006) and World Medical

Association Declaration of Helsinki Ethical Principles for Medical Research

Involving Human Subjects 1964 (as amended)

Page 44 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

44

1.3 Contacts

1.3.1 Chief Investigator:

Dr Adrian J Stanley MD, Department of Gastroenterology,

Glasgow Royal Infirmary, Glasgow, G4 OSF, UK

Tel: 0141 211 4073

Fax: 0141 211 5131

Email: [email protected]

1.3.2 Sponsor’s representative:

Ms Maureen Travers

NHS Greater Glasgow & Clyde health Board (GG&CHB)

Research and Development Management Office

Tennant Institute,

38 Church St,

Western Infirmary,

Glasgow G11 6NT

Tel: 0141 211 6389

Email: [email protected]

1.4 Investigators

1.4.1 Chief Investigator

AJ Stanley, MD, Department of Gastroenterology, Glasgow Royal Infirmary,

Glasgow, UK

1.4.2 Principle Investigators

1. L Laine, MD, Internal Medicine, Yale School of Medicine, New Haven,

Connecticut, USA

2. SB Laursen, MD, PhD, Department of Medical Gastroenterology, Odense

University Hospital, Odense, Denmark

3. HR Dalton, MD, DPhil, Gastrointestinal Unit, Royal Cornwall Hospital, Truro,

Cornwall, UK

Page 45 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

45

4. IA Murray, MD, Gastrointestinal Unit, Royal Cornwall Hospital, Truro,

Cornwall, UK

5. M Schultz, MD, PhD, Department of Medicine, Dunedin School of Medicine,

University of Otago, Dunedin, New Zealand

6. JH Ngu, MB ChB, PhD, Department of Gastroenterology, Singapore General

Hospital, Singapore

7. CJL Khor, MBBS, Department of Gastroenterology, Singapore General Hospital,

Singapore

8. Rozeta Abazi, MD, Department of Medical Gastroenterology, Odense University

Hospital, Odense, Denmark

1.5 Study start date

March 2014

1.6 Estimated primary completion date

March 2015

2. Background

Several recent international guidelines have recommended the use of early risk

assessment in patients who present with upper gastrointestinal bleeding (UGIB)

(1-3). Several risk scoring tools have been developed for use in this situation.

Many of these incorporate findings from the patient’s endoscopy including the

widely used Rockall score from the UK (4), the American Baylor and Cedars Sinai

scores, the Spanish Almela and Italian PNED scores (5-8). However in many

centres, endoscopy is often delayed unless for emergency reasons. Therefore

several scores have been developed which can be calculated very soon after

presentation, prior to endoscopy, which are particularly useful in helping

identify a cohort of “low risk” patients who may be able to avoid admission and

have out-patient management. These scores can also direct higher risk patients

Page 46 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

46

to appropriate management and interventional strategies. These pre-endoscopy

scores include the “Clinical” Rockall (omitting the endoscopic components of the

full Rockall score), the Glasgow Blatchford (GBS) and Cambridge scores from the

UK, the American AIM65 score and several artificial neural networks (ANNs) (9-

13).

There have been numerous studies comparing various combinations of risk

scores for UGIB although relatively few are multicentre. In addition there are

very limited data on international comparison of risk assessment scores. Current

evidence suggests the scores with the highest accuracy for identifying poor

outcome as defined by mortality, rebleeding or clinical intervention (need for

transfusion, endoscopic, surgical or radiological intervention) are the post-

endoscopic full Rockall score and the pre-endoscopic GBS or AIM65. However

several of the newer scores have not been externally validated or compared with

other risk assessment tools. Therefore the optimum score for use in clinical

practice remains uncertain.

3. Aim

The aim of this study is to compare the clinical utility of the Clinical Rockall, GBS,

AIM65, full Rockall and PNED scores in a prospective, multicentre, international

cohort of patients presenting with UGIB. All collected data will be anonymised,

with patients only identifiable to the local PI or research nurse. The data will also

be analysed to assess whether an improved new score (pre-endoscopy or post-

endoscopy) can improve the prediction of outcome for patients with UGIB.

4. Methods

The study is designed as a non-interventional, prospective, multicentre,

international cohort study. Consecutive patients presenting with UGIB and

fulfilling the criteria for inclusion (see point 4.2) will be included at the

participating centres (see point 4.4).

All centres will collect the same data. Registered data will include patients’

characteristics, blood tests, findings at endoscopy (if undertaken), and

interventions in the form of blood transfusion, endoscopic treatment, surgery, or

Page 47 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

47

radiological intervention (see point 4.6 for details). Persistent bleeding,

rebleeding, in-hospital death and 30-day mortality will be recorded. Data will be

collected by medical doctors, medical students, or research nurses depending on

the site. All data will be anonymised on the database and only identifiable to the

local PI or research nurse.

The study is non-interventional with no randomisation and all patient

management will be as per standard of care for patients with UGIB. The

registered anonymised data will be used to compare the predictive abilities of

the clinical and full Rockall score, GBS, AIM65, and PNED (see point 4.9 for

details). Furthermore, the data will be interrogated to try and develop a new risk

score and compare the performance of this with the existing risk score systems.

4.1 Study population

Consecutive patients presenting with symptoms of UGIB at the participating

centres are included. UGIB is defined as history of haematemesis, coffee-ground

vomit, or melaena. Patients with the combined presence of hematochezia and

circulatory insufficiency will also be included (see 4.2).

4.2 Inclusion criteria

- Patients presenting with haematemesis, coffee-ground vomit, melaena, or:

- Patients presenting with haematochezia and circulatory insufficiency (heart

rate >100bpm and/or systolic blood pressure <100mmHg)

4.3 Exclusion criteria

- Patients presenting with UGIB while already in-patients for another reason

4.4 Participating centres

- Glasgow Royal Infirmary, Glasgow, UK

- Royal Cornwall Hospital, Truro, UK

- Odense University Hospital, Odense, Denmark

- Dunedin Public Hospital, Dunedin, New Zealand

- Yale University Hospital, New Haven, Connecticut, USA

- Singapore General Hospital, Singapore

Page 48 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

48

(Humanitas Hospital, Milan and St Johns Hospital, London Ontario – if can obtain

resource to undertake study)

4.5 Intervention

There are no interventions. Patients will be treated according to the standard of

care for patients presenting with UGIB.

4.6 Registered data

4.6.1 Patient characteristics

Age, sex, comorbidity (ischaemic heart disease, cardiac failure, renal failure, liver

disease, liver cirrhosis, liver failure, COPD, Diabetes, cerebrovascular disease,

any malignancy, disseminated malignancy), altered mental status, previous

peptic ulcer, use of NSAID’s, antiplatelet or anticoagulant medication, American

Society of anesthesiologists (ASA) score and blood tests (serum urea, creatinine,

albumin, haemoglobin and INR)

4.6.2 Presentation

Symptoms (coffee ground vomiting, haematemesis, melaena, haematochezia,

syncope), and time from development of symptoms to presentation to hospital.

4.6.3 Bleeding source and severity of bleeding

Pulse and systolic blood pressure at presentation and findings at upper

endoscopy (and blood tests as above)

4.6.4 Outcomes

Treatment with blood transfusion, performance of endoscopic therapy,

performance of surgical/radiological haemostasis, persistent bleeding,

development of rebleeding, time to development of, and severity of rebleeding

(need for transfusion/repeat endoscopy/surgery/radiological intervention), and

30-day mortality

4.6.5 Other data regarding hospital admission/discharge from the

emergency department

Page 49 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

49

Admission overnight or at weekends, time from presentation to performance of

endoscopy, length of hospitalisation, and discharge from emergency department

without hospitalisation.

4.6.6 Study management & Data handling

Each patient will have a study number. Each centre will collect and add the local

anonymised data to their own local database. Patient data will only be

identifiable to the local research nurse and PI, with the identifiable link being

held as a paper copy in a local locked cabinet. Presentation data will be entered

anonymously onto the local database by the research nurse, with the 30-day

follow-up data added subsequently from electronic records. On completion of the

study, local anonymised data from each centre will be sent to Drs Laursen and

Stanley for final analysis.

4.7 Primary outcome

- Need for hospital based intervention defined as need for blood transfusion,

haemostatic intervention (endoscopic/surgical/radiological haemostasis), or 30-

day mortality

4.8 Secondary outcomes

- Rebleeding

- Length of hospital stay

4.9 Statistical analyses

The predictive abilities of the examined risk scoring systems will be evaluated by

calculation of sensitivity, specificity, positive predictive values (PPV), negative

predictive values (NPV), and receiver operating characteristics (ROC) curves.

Potential geographic variation will be analysed. The outcome among patients

classified as low-risk (not requiring transfusion, haemostatic intervention and

alive at 30-days) will be compared between systems.

A new risk score for identifying patients in need for hospital based intervention

will be developed using regression analysis. The predictive ability of this new

score will be described by sensitivity, specificity, positive predictive values

Page 50 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

50

(PPV), negative predictive values (NPV), and ROC curves. The predictive ability

will be compared to the performance of the clinical and full Rockall score, GBS,

AIM65, and PNED.

5. Funding

There is no specific funding for this study. Each centre will involve existing local

research or audit nurses +/- a junior doctor to collect data.

6. End of study definition

The end of the study will be 30 days after inclusion of the last patient (estimated

March 2015).

7. Publication

The results of the study will be presented at international conferences in

gastroenterology and will be published in an international pear reviewed

journal.

Page 51 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960


nly

51

Declaration of competing interests

All authors have completed the Unified Competing Interest form at

www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author)

and declare that (1) no authors have support from any company for the submitted work;

(2) no authors have relationships with any company that might have an interest in the

submitted work (3) no author, their spouses, partners, or children have financial

relationships that may be relevant to the submitted work and (4) no authors have any

non-financial interests that may be relevant to the submitted work.

Transparency Declaration

AJS (study guarantor) affirms that the manuscript is an honest, accurate and transparent

account of the study being reported. No important aspects of the study have been

omitted and no discrepancies identified.

Data sharing statement

Patient level data and/or full dataset available from the corresponding author AJS.

Consent was not obtained but the presented data are anonymised and risk of

identification is low.

Exclusive License

Dr Stanley (as corresponding author) on behalf of all authors grants a worldwide license

to the publishers and its licensees in perpetuity in all forms to publish this contribution.

Page 52 of 51


BMJ

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

BMJ · the Rockall score and recently described Progetto Nazionale Emorragia Digestive (PNED) score require endoscopy before calculation.4,5 The Rockall score has been widely reported

Documents