Digestive and Liver Disease 36 (2004) 806–810

Alimentary Tract

Measuring dyspepsia: a new severity index validated in Bologna

L. De Lucaa , R.M. Zagaria , P. Pozzatoa , T. Fiorinib ,
L. Ricciardielloa , C. Martuzzia , E. Rodaa ,
F. Bazzolia,∗ , S.J.O. Veldhuyzen van Zantenc
a Department of Internal Medicine and Gastroenterology, University of Bologna, Policlinico S. Orsola,
Via Massarenti n. 9, 40138 Bologna, Italy
b Department of Statistical Science, University of Bologna, Bologna, Italy
c Division of Gastroenterology, Dalhousie University, Halifax, NS, Canada

Received 23 March 2004; accepted 15 July 2004

Available online 25 September 2004

Abstract

Background. Measurement of the severity of dyspepsia symptoms before and after treatment and determining what is a significant change
is a major problem in designing dyspepsia treatment studies.
Objectives. To assess the reproducibility, validity and responsiveness to treatment of a dyspepsia questionnaire to be used in clinical and
population-based studies.
Methods. Seventy-three dyspeptic patients (35 male, 38 female; mean age 52 years) and 75 healthy volunteers (32 male, 43 female; mean
age 52 years) were included. Subjects were interviewed for the presence/absence and severity/frequency of 19 gastrointestinal symptoms.
Severity was measured on a 5-point scale. Frequency was also recorded on a 5-point scale. A global symptom index (severity × frequency)
was calculated for the eight most severe symptoms; a mean global symptom index (8-MGSI) was considered for the evaluation of the
instrument. To evaluate intra-observer variation, one author interviewed subjects (T0) and then repeated the interview 1 week later (T1). For
inter-observer variation, two authors interviewed patients. Validity was measured by comparing 8-MGSI of the dyspepsia patients to those of
healthy volunteers. Responsiveness was assessed by comparing mean global symptom index before and 1 month after appropriate therapy.
Results. Reproducibility: The mean 8-MGSI was 4.5 at T0 and 3.7 at T1 with a correlation coefficient of 0.62. As for inter-observer
variation, the average 8-MGSI was 4.8 by the first author and 3.9 by the second with a correlation coefficient of 0.60. Validity: The mean
8-MGSI was, respectively, 1.4 in healthy volunteers and 4.8 in dyspeptic patients (p = 0.001). Responsiveness: After treatment, a significant
improvement in 8-MGSI was detected (p = 0.001).
Conclusions. This questionnaire is a reliable, valid and responsive instrument for measuring the presence, severity and frequency of
dyspepsia.
© 2004 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.

Keywords: Dyspepsia; Gastrointestinal symptoms; Helicobacter pylori; Mean global symptom index

1. Introduction important that questionnaires are able to separate patients

Dyspepsia is a very common disorder with a prevalence of
25–50% in western countries [1,2], accounting for 10–20%
of GP consultations [3]. In population-based studies, it is
∗ Corresponding author. Tel.: +39 051 6364106; fax: +39 051 343926.
E-mail address: bazzoli@alma.unibo.it (F. Bazzoli).
with significant upper gastrointestinal (GI) symptoms from
normal individuals. Furthermore, in studies of subjects with
functional dyspepsia it is important that a dyspepsia instru-
ment is able to document the change in symptom severity
after treatment, assuming that a change in health status oc-
curred [4]. The latter is also important in patients with organic
disease, as treatment should both heal the lesion and improve

1590-8658/$30 © 2004 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.dld.2004.07.010
 L. De Luca et al. / Digestive and Liver Disease 36 (2004) 806–810 807

related symptoms. Therefore, it is necessary to have valid Table 1

tools for measuring the severity of dyspeptic symptoms in Symptoms scale for assessment of dyspeptic patients
patients with both functional and organic dyspepsia. Severitya Frequency
The most important problem in study design of dyspepsia 1 None 1 –
trials is the lack of consensus on how to measure outcome and 2 Mild 2 <2 time/week
to define the efficacy of an intervention. As a result, there is 3 Moderate 3 >3 time/week, not daily
marked variation in the outcome measures that have been 4 Severe 4 Daily, intermittent
5 Very severe 5 Daily, almost continues
used [5]. However, even if validated questionnaires are used
a As measured on the Likert scale: (1) No problem; (2) Mild problem, can
in clinical trials, they are not always suitable for different
be ignored when you do not think about it; (3) Moderate problem, cannot be
ethnic groups, countries or types of patients. ignored but does not influence daily activities; (4) Severe problem, influences
Loiano and Monghidoro are two villages in the vicinity your concentration on daily activities; and (5) Very severe problem, markedly
of Bologna where a population-based intervention study is influences your daily activities and/or requires rest.
ongoing for H. pylori infection and related pathologies [6].
The aim of the present study was to assess the reproducibil- for 4 weeks. Treatment was given open label, and patients
ity, validity and responsiveness to treatment of a disease- were not randomised.
specific dyspepsia questionnaire that can be used in clinical
trials and population-based studies. 3. Symptom assessment

The severity of symptoms and their frequency was as-

2. Materials and methods sessed in a patient interviewed by a gastroenterologist. Dur-
ing the interview the patient was asked to rate the severity
2.1. Subjects and the frequency of each symptom. The physician recorded
the patients’ score in the record. Nineteen symptoms were
Seventy-three consecutive patients (35 male, 38 female; recorded and evaluated in terms of presence/absence, sever-
mean age 52 years) referred for dyspeptic symptoms to the ity and frequency. The patient was also asked to define the
outpatient clinic of the Gastroenterology and Endoscopy Unit most bothersome symptom. Presence of alarm symptoms was
of S. Orsola Hospital, University of Bologna, and 75 healthy also recorded.
volunteers (32 male, 43 female; mean age 52 years), matched Severity of GI symptoms was scored using a previously
for age and sex, were included in the present study. Healthy validated 5-point Likert scale [11]; a 5-point scale was also
volunteers were subjects who never had a medical consulta- used to score symptom frequency (Table 1). We assessed first
tion for dyspepsia or a previous upper GI diagnosis. Patients the importance of each of the 19 symptoms. The scores were
needed to fulfil the Rome definition of dyspepsia, that they assessed for each symptom and were added for all 73 patients
had to have epigastric pain or discomfort that was present giving a minimum possible score of 73 and a maximum
either continuously or frequently recurring. Patients were al- possible score of 365 (Table 2). The eight highest scoring
lowed to have heartburn and regurgitation symptoms as has
been done in other studies [7,8]. Table 2
Exclusion criteria were the following: previous diagnosis Ranked importance of symptoms in 73 patients
of upper GI diseases such as peptic ulcer disease (PUD) or oe- Symptom Scorea
sophagitis, a history of gastric surgery, use of a proton pump 1 Epigastric pain 172
inhibitor (PPI), H2 -antagonist, antibiotics during last month, 2 Upper abdominal fullness 141
previous anti-H. pylori treatment and ongoing treatment with 3 Heartburn 136
4 Upper abdominal discomfort 135
NSAIDs or ASA. Smoking habits and consumption of alco- 5 Acid regurgitation 135
hol were recorded. 6 Pain often relieved by food or antiacid 134
Of the 73 dyspeptic patients included in the study, 40 un- 7 Abdominal discomfort or pain 128
derwent a validated 13 C-urea breath test (UBT) [6] and upper 8 Pain often before meals or when hungry 117
GI endoscopy featuring three biopsies in the gastric antrum 9 Bloating or feeling of abdominal distension 116
10 Early satiety 112
and three biopsies in the corpus for rapid urease test and his- 11 Night pain (waking the subjects) 111
tology. Subjects with at least two of the three tests resulting 12 Nausea 110
positive were considered H. pylori-positive. The remaining 13 Abdominal discomfort/pain relieved with defecation 108
33 dyspeptic patients were tested for H. pylori by 13 C-UBT. 14 Abnormal stool frequency (>3 per day and <3 per week) 96
H. pylori-positive subjects received 1-week triple therapy 15 Chest pain 91
16 Dysphagia 90
with Omeprazole 20 mg daily, Clarithromycin 250 mg twice 17 Urgency or feeling of incomplete evacuation 88
daily and Tinidazole 500 mg twice daily (OCT) [9,10] and an 18 Passage of mucus 79
additional 3-week Omeprazole 20 mg daily regimen when 19 Odynophagia 74
PUD or reflux oesophagitis was diagnosed. The H. pylori- a Cumulative score: each symptom was scored in each patient by severity,

negative subjects were treated with Omeprazole (20 mg daily) according to Table 1.
808 L. De Luca et al. / Digestive and Liver Disease 36 (2004) 806–810

Table 3
Comparison of the eight-symptom mean measurement between dyspeptic patients and healthy subjects
Symptoms Dyspeptic patients Healthy volunteers Mann–Whitney
(n = 73) (MGSI) (n = 75) (MGSI) U-test
Epigastric pain 7.2 1.3 0.001
Upper abdominal fullness 6.1 1.7 0.001
Heartburn 3.7 1.1 0.001
Upper abdominal discomfort 5.0 1.6 0.002
Acid regurgitation 4.9 1.4 0.001
Pain often relieved by food/antiacid 4.4 1.0 0.001
Abdominal discomfort or pain 3.9 1.8 0.010
Pain often before meals or when hungry 3.5 1.0 0.118
Average 4.8 1.4 0.001
MGSI: see text.

symptoms were arbitrary chosen and subsequently further 3.1.2. Reproducibility

evaluated by multiplying their frequency × severity [12].
Apart from an overall severity score of the eight symptoms
(range 8–40), a mean global symptom index (MGSI) was also
calculated for each one of the eight symptoms considered in
both patient

n and control  groups.
 The following formula was
used: i=1 S i F i /n, where is summa, n the total number
of subjects, i = 1 the single symptom considered, Si the sever-
ity of the single symptom considered and Fi the frequency of
the single symptom considered. The average of the combined
eight symptoms MGSI was also calculated (8-MGSI).
3.1. Evaluation of the dyspepsia instrument
Reproducibility of the questionnaire was assessed by re-
peating the patient interview after 1 week by either the same
or a different physician. For validity of the information ques-
tionnaire (i.e. the questionnaire measuring what it is supposed
to measure), results in dyspeptic patients were compared to
those obtained in healthy volunteers. For assessment of re-
sponsiveness, the questionnaire was readministered 4 weeks
after completion of therapy [4,5,11,13–15].
We assessed two aspects of reproducibility:
- For intra-observer variation (OV) and assessment of symp-
tom stability, one author (L.D.L.) interviewed all 73
patients with a variety of dyspeptic symptoms and 25
of the 75 healthy controls (T0). The interview was re-
peated 1 week after the first interview (T1) and prior
to any diagnostic or therapeutic intervention. The inter-
viewer was not shown the symptom scores of the first
interview.
- For inter-OV, the 73 dyspepsia patients were interviewed
by one author and the interview was repeated on a separate
occasion within 24 h by another author, who was blinded
to the first interview scores.
3.1.3. Responsiveness
The responsiveness, or the ability to detect change, of the
questionnaire was assessed by comparing the summary score
and 8-MGSI score immediately before and 1 month after
receiving therapy.
3.2. Statistical analysis

3.1.1. Validity The severity and frequency of GI symptoms were anal-

Validity was measured comparing the summary scores and ysed using non-parametric statistics (Spearman rank corre-
8-MGSI of the 73 dyspeptic patients to those of 75 healthy lation, Wilcoxon test and Mann–Whitney U-test). Statisti-
volunteers who had never had a medical consultation for dys- cal significance was accepted at the 95% confidence level
pepsia. (p < 0.05).
Table 4
Intra-observer and inter-observer variability (reproducibility)
Symptoms Intra-observer (n = 98) Inter-observer (n = 73)

T0 (MGSI) T1 (MGSI) Spearman coefficient First (MGSI) Second (MGSI) Spearman coefficient
Epigastric pain 6.9 5.2 0.56 7.2 5.9 0.49
Upper abdominal fullness 6.0 4.6 0.60 6.1 5.4 0.61
Heartburn 4.6 3.8 0.55 3.7 2.9 0.52
Upper abdominal discomfort 4.3 4.0 0.68 5.0 3.8 0.51
Acid regurgitation 4.2 2.9 0.64 4.9 3.6 0.53
Pain often relieved by food/antiacid 3.1 2.5 0.59 4.4 3.7 0.63
Abdominal discomfort or pain 3.6 2.9 0.66 3.9 3.0 0.65
Pain often before meals or when hungry 3.5 3.3 0.75 3.5 3.2 0.67
Average 4.5 3.7 0.62 4.8 3.9 0.60
MGSI: see text.
 L. De Luca et al. / Digestive and Liver Disease 36 (2004) 806–810 809

Table 5
Comparison of the eight-symptom measurement before and after treatment (responsiveness)
Symptoms GERD (n = 18); PUDs (n = 7) Functional dyspepsia (n = 15); uninvestigated
dyspepsia (n = 33)
Before After p-value Before After (MGSI) p-value
(MGSI) (MGSI) (MGSI)
Epigastric pain 7.4 2.2 0.001 7.0 1.7 0.001
Upper abdominal fullness 2.7 2.2 0.590 7.9 2.8 0.004
Pain often relieved by food/antiacid 4.8 1.1 0.030 4.2 1.1 0.001
Heartburn 5.6 1.5 0.001 2.7 1.2 0.033
Upper abdominal discomfort 3.4 2.0 0.491 5.8 2.6 0.032
Acid regurgitation 7.2 2.5 0.002 3.7 1.4 0.001
Abdominal discomfort or pain 2.3 2.4 0.918 4.7 2.6 0.001
Pain often before meals or when hungry 3.5 1.1 0.011 3.5 1.3 0.005
Average 4.6 1.9 0.001 4.9 1.9 0.001
MGSI: see text.

4. Results 4.3. Responsiveness

4.1. Validity A significant improvement of the average 8-MGSI after

Out of 73 subjects included in the study, 33 were
tested for H. pylori by UBT and treated with 1-week OCT
when positive (n = 17) or with Omeprazole 20 mg/day
for 4 weeks when negative (n = 16). Forty subjects un-
derwent upper GI endoscopy: seven were found to have
PUD and H. pylori infection and were treated with 1-week
OCT therapy plus three additional weeks with Omepra-
zole 20 mg/day; 18 were found to have GERD (6 symp-
toms plus oesophagitis; 12 symptoms only): of these 13 were
H. pylori-negative and treated with Omeprazole 20 mg/day
for 4 weeks and five were positive and treated with 1-
week OCT plus three additional weeks with Omeprazole
20 mg/day; finally in 15 subjects no relevant findings were
observed and were treated with 1-week OCT when H. pylori-
positive (n = 11) or with Omeprazole 20 mg/day for 4 weeks
(n = 4).
The average 8-MGSI was 1.4 (range 1–1.8) in healthy
volunteers and 4.8 (range 3.5–7.2) in patients with dyspeptic
symptoms (p = 0.001) (Table 3) demonstrating that indeed
the questionnaire is able to distinguish dyspeptic from non-
dyspeptic patients.
4.2. Reproducibility
- Intra-observer variability: the average 8-MGSI was 4.5
(range 3.1–6.9) on day 1 (T0) and 3.7 (range 2.5–5.2) on
day 2 (T1) with a correlation coefficient of 0.62. Individual
MGSI scores only showed minor variation during the 1-
week interval (Table 4).
- Inter-observer variability: the average 8-MGSI was 4.8
(range 3.3–7.2) by the first author and 3.9 (range 2.9–5.9)
by the other author with a correlation coefficient of 0.60
(Table 4).
treatment was detected in subjects with GERD and PUD
(summary score was 36.8 versus 15.2; p = 0.001), and also
in those with functional and uninvestigated dyspepsia (sum-
mary score was 39.2 versus 15.2; p = 0.001) (Table 5).
5. Discussion
Dyspepsia is a major health problem, and studies
evaluating prevalence, associated factors and diseases as
well as response to treatment are still needed. The Bologna
dyspepsia severity index was developed for use in the
ongoing Loiano–Monghidoro project, in which prevalence
of endoscopic lesions, H. pylori infection, and dyspeptic
symptoms are evaluated. In this project, H. pylori-positive
subjects, but without endoscopic lesions, are randomised
in a double-blind trial to either anti-H. pylori treatment or
placebo to evaluate the long-term effect of H. pylori eradi-
cation on atrophy and intestinal metaplasia and on dyspeptic
symptoms.
In this study, following the Rome recommendations, pa-
tients were asked to rate the severity and frequency of
their symptoms that was recorded in a physician interview.
Interview-based recording of symptoms may be preferable to
self-recorded symptoms by patients, especially when com-
pared to diary cards. Although dairy cards are less time-
consuming and cheaper than an interviewer-administered
questionnaire [16] and can overcome the problem of patient
recall [5], there are problems with self-recorded forms, which
include validity, logistics and accuracy. Cultural aspects and
level of education may be important impediments to the use
of dairy cards. However, our questionnaire is very simple
to use and could easily be modified further to become self-
administered.
810 L. De Luca et al. / Digestive and Liver Disease 36 (2004) 806–810
The Rome II working team discussed the use of 4-, 5-,
7- or linear-point scales [5]. The severity of GI symptoms
was scored on a previously validated 5-point ordinal (Likert
scale) [11]. The severity of each symptom was multiplied by
symptom frequency and used to assess change in symptoms
after treatment. Another way to assess symptom severity is
a Visual Analog Scale (VAS). For the VAS, patients mark
their symptom severity on a line of fixed length between two
extremes (e.g. 0–100). They have also been found to be re-
producible and sensitive to change [17]. An advantage of a
Likert scale is that it is easier to understand for patients and a
change in the score is easier to interpret by physicians [18].
The reproducibility of the questionnaire was confirmed by
a good correlation between 8-MGSI by the same interviewer
on two different occasions and when tested by different ob-
servers.
Responsiveness was examined by assessing the MGSI in
subjects with uninvestigated and investigated dyspepsia be-
fore treatment and 1 month after completion of the therapy.
The MGSI improved after treatment in GERD, PUD and
functional dyspepsia patients indicating its ability to detect
clinically important changes for each symptom. Apart from
reporting average symptom severity such as mean 8-MGSI,
the ROME Design of Clinical Trials Working Party also rec-
ommends that in treatment trials, investigators decide ‘a pri-
ori’ how much improvement is required for a patient to be
considered a responder. Using the MGSI it is certainly pos-
sible to define treatment responders according to the ROME
recommendation. Based on our results it might be suggested
that a mean score decrease by 2.5 is useful to define treatment
responders, but this will require further validation.
In summary, we conclude that our Bologna questionnaire
is a reliable, valid and responsive instrument that can be
used in clinical and epidemiological studies to measure the
presence, severity and frequency of dyspepsia. Our results
also suggest that GI symptoms scored in terms of severity
and frequency by the 5-point Likert scale fulfils the criteria
for its use as an outcome measurement in clinical treatment
trials of both investigated and uninvestigated dyspepsia.
Conflict of interest statement
The study was not supported by pharmaceutical compa-
nies and there are no conflicts of interest to be declared.
References
[1] Jones RH, Lydeard SE, Hobbs FD, Kenkre JE, Williams EI, Jones
SJ, et al. Dyspepsia in England and Scotland. Gut 1990;31:401–5.
[2] Bazzoli F, De Luca L, Pozzato P, Fossi S, Ricciardiello L, Nicolini G,
et al. Helicobacter pylori and functional dyspepsia—review of pre-
vious studies and commentary on the new data. Gut 2002;50:iv33–5.
[3] Healtly RV, Rathbone BJ. Dyspepsia: a dilemma for doctors? Lancet
1987;i:779–82.
[4] El-Omar EM, Banerjee S, Wirz A, McColl KEL. The Glasgow Dys-
pepsia Severity Score—a tool for the global measurement of dys-
pepsia. Eur J Gastroenterol Hepatol 1996;8:967–71.
[5] Veldhuyzen van Zanten SJO, Talley NJ, Bytzer P, Klein KB, Whor-
well PJ, Zinsmeister AR. Design of treatment trials for functional
gastrointestinal disorders. Gut 1999;45:II69–77.
[6] Bazzoli F, Palli D, Zagari RM, Festi D, Pozzato P, Nicolini G, et
al. The Loiano–Monghidoro population-based study of Helicobacter
pylori infection: prevalence by 13 C-urea breath test and associated
factors. Aliment Pharmacol Ther 2001;15:1001–7.
[7] Chiba N, Veldhuyzen van Zanten SJO, Sinclair P, Ferguson RA, Es-
cobedo S, Grace E. Treating Helicobacter pylori infection in primary
care patients with uninvestigated dyspepsia: the Canadian adult dys-
pepsia empiric treatment—Helicobacter pylori positive (CADET-Hp)
randomized controlled trial. Br Med J 2002;324:1012–20.
[8] Moayyedi P, Feltbower R, Brown J, Mason S, Mason J, Nathan J,
et al. Effect of population screening and treatment for Helicobacter
pylori on dyspepsia and quality of life in community: a random-
ized controlled trial. Leeds HELP Study Group. Lancet 2000;355:
1665–9.
[9] Bazzoli F, Zagari RM, Fossi S, Pozzato P, Alampi G, Simoni P, et al.
Short-term low-dose triple therapy for the eradication of H. pylori.
Eur J Gastroenterol Hepatol 1994;6:773–7.
[10] Malfertheiner P, Megraud F, O’Morain C, Hungin AP, Jones R, Axon
A, et al., European Helicobacter pylori Study Group (EHPSG). Cur-
rent concepts in the management of Helicobacter pylori infection.
The Maastricht Consensus Report 2-2000. Aliment Pharmacol Ther
2002;16:167–80.
[11] Veldhuyzen van Zanten SJO, Tytgat KMAJ, Pollak PT, Goldie J,
Goodacre RL, Riddell RH, et al. Can severity of symptoms be used
as an outcome measure in trials of non-ulcer dyspepsia and Heli-
cobacter pylori associated gastritis. J Clin Epidemiol 1993;46:273–9.
[12] Mearin F, de Ribot X, Balboa A, Salas A, Varas MJ, Cucala M, et
al. Does Helicobacter pylori infection increase gastric sensitivity in
functional dyspepsia? Gut 1995;37:47–51.
[13] Mégraud F, et al., Working Party of the European Helicobacter pylori
Study Group. Guidelines for clinical trials in Helicobacter pylori
infection. Gut 1997;41:S1–9.
[14] Talley NJ, Phillips SF, Joseph Melton III L, Wiltgen C, Zinsmeister
AR. A patient questionnaire to identify bowel disease. Ann Int Med
1989;111:671–4.
[15] Talley NJ, Haque M, Wyeth JW, Stace NH, Tytgat GNJ, Stanghellini
V, et al. Development of a new dyspepsia impact scale: the Nepean
Dyspepsia Index. Aliment Pharmacol Ther 1999;13:225–35.
[16] Moayyedi P, Duffett S, Braunholtz D, Mason S, Richards ID, Dowell
AC, et al. The Leeds Dyspepsia Questionnaire: a valid tool for mea-
suring the presence and severity of dyspepsia. Aliment Pharmacol
Ther 1998;12:1257–62.
[17] Talley NJ, Nyren O, Drossman DA, Heaton KW, Veldhuyzen van
Zanten SJO, Koch MM, et al. The irritable bowel syndrome. To-
ward optimal design of controlled treatment trials. Gastroenterol Int
1993;6:189–211.
[18] Guyatt GH, Townsend M, Berman LB, Keller JL. A comparison of
Likert and visual analogue scales for measuring change in function.
J Chronic Dis 1987;40:1129–33.