DlGE!iTLlUER 06 2001;33:606-92
A GISC* nation-wide
and causes of death in Italian patients
GI Unit, Department of Clinical Sciences,
University “La Sapienza” of Rome;
I Department of Statistics, University of
Milan - “Bicocca”; * GI Unit, University of
Padua; 3 GI Unit, IRCCS “Casa Sollievo
de/la Sofferenza’l San Giovanni Rotondo;
4 GI Units Department of Internal
Medicine, University of L Xquila; 5 GI
Unit, University “Federico II” of Naples;
6 GI Unit, “San Filippo Neri” Hospital,
Rome; 7 GI Unit, University “Federico II”
of Naples, Italy.
The following GISC investigators took
part in this study: P. Paoluzi, L.
M.C. Di Paolo, V. Stiglieno,
P. Fracasso 0?omel; G. Latella
(L’Aquilal; M. Cottone IPalermoI; R.
D’lncti, V. Di Leo [Padual; C. Del
Vecchio Blanc0 [NaplesI; G. Ciancio
(Florancel; M. lngrosso IBaril; V. Ponti
[Turin); C. Mansi (Genoa); D. Valpiani
(Forlil: G.P. Rigo (Modenal; G.P.
Bresci (Pisal; A. Tragnone IBolognal.
Dr. A. Viscido, Dipartimento di Science
Cliniche, Cattedra di Gastroenterologia,
Universith “La Sapienze” di Roma,
Policlinico “Umberto I”, viale del
Policlinico, 155, 00161 Roma, Italy.
Submitted February 19, 2001.
Accepted after revision July 5, 2001.
over last few decades.
deaths may depend on causes not strictly related
To evaluate overall and cause-specific
group of Italian patients
A total of 2,066
consecutively diagnosed in twenty Italian Gastroenterology
1964 and 1995 were followed-up from diagnosis until 1997.
ised Mortality Ratios and Relative Survival Ratios were calculated.
Overall mortality of patients
parable to that in general population
92. I expected [standardises mortality
terval, 0.8- 1.2). Significantly higher mortality
under 30 years of age at diagnosis
95% confidence interval, 1.3-4.91,
1974 [standardised mortality
I. I-5.71. Proctocolitis and complications
tioned in 1 I and 5 certificates,
deaths was observed for colorectal
tality ratio, 3.0; 95% confidence
ised mortality ratio, 4.4; 95%
95% confidence interval, I. O-6. I], in particular
non-Hodgkin lymphoma. A significant
cancer of the respiratory system
95% confidence interval, 0. I- I. 01.
This study confirms
tients with ulcerative colitis is comparable
Only 12% of deaths were due to ulcerative
of deaths were attributed to colorectal
cancer occurred, on average, 9 years after diagnosis
itis, suggesting that the risk of cancer
long-standing colitis. As to mortality
there was an excess of deaths due to malignancies
Death rate for patients with ulcerative
to that in general
colitis has changed
to colonic disease.
mortality rate in a large
colitis patients aged > 7 8 years
is comparable and that
with 93 deaths
colitis was com-
I. 0; 95% confidence
and in those diagnosed
ratio, 2.7; 95% confidence
respectively A significant
interval, I. O-6.9;
deficit of deaths was observed
I .2- I I. 3), and
that, also in ItaIM mortality
to that in general population.
colitis itself, whereas
cancer. Deaths from colorectal
of ulcerative col-
with is not limited to patients
for causes unrelated to colitis,
of the haemolym-
Digest Liver Dis 2001;33:686-92
Key words: ulcerative colitis mortality
colitis (UC) is an idiopathic
by colonic mucosal inflammation
inflammatory bowel disease (IBD)
and a chronic relapsing
A. lliscido et al.
course. Usually, it shows mild activity, but it can also
be life-threatening during
colonic and systemic complications,
course, due to the development
Prognosis in patients with UC has greatly changed over
the last few decades 2-4. Whereas a cumulative
mortality of 40% was reported in 1963 5, recent studies
indicate that cumulative long-term
creased compared with the general
However, although most recent studies indicate a good
survival of patients with UC, some reports from Scan-
dinavian countries and North America still indicate an
excess of deaths in those populations ‘3-‘x. Data on mor-
tality for UC in Italy are only available for the area of
Florence and a series of hospital patients in Rome. No
excess mortality was observed in either study ” I’.
Life expectancy of patients with UC may also depends
on causes not strictly related to colonic disease. An in-
creased mortality due to respiratory diseases, non-alco-
holic-related liver diseases, bile
leukaemia, bone tumours,
cancer, connective tissue cancers, brain cancers, and sui-
cides has been reported in patients with UC I2 I6 ” ‘9-24.
Aim of the present multicentre
evaluate the overall and cause-specific
in a large Italian group of patients with UC distributed
throughout Northern, Central, and Southern Italy. Pa-
tients were consecutively
troenterology Units co-operating
Group. Survival, changes in survival over different
time periods, and causes of death in UC patients in
comparison to the general population
the severe attacks due to
and, in the late
of colorectal cancer ‘.
mortality is not in-
cancer, breast endometrial
study was, therefore, to
diagnosed in twenty
in a National
Patients and Methods
The study population
Southern Italy that agreed to take part in the study from
the start of their activity (hospital-based
overall study population thus comprised 2,066 patients.
Of these 20 Units, only one enrolled the patients start-
ing from the Sixties (1964), 9 from the Seventies (from
1970 to 1979) and the remaining
(from 1980 to 1989). All the Units flagged the patients
diagnosed up until 1995. Seven Units contributed
the study by flagging fewer than 50 patients, another
seven with a number of patients ranging between 51
and 99, the remaining six with over 100 patients.
Diagnosis was made according
criteria based on a history of bloody diarrhoea, typical
endoscopic findings, and characteristic
consisted of all the new cases of
diagnosed in 20 University and Hos-
Units in Northern, Central and
10 from the Eighties
to routine diagnostic
All patients were followed
sis until December
up). Assessment of the vital status was carried out at
the last known municipality
concerns those patients who had died during follow-
up, a copy of the death certificate was requested at the
Local Health Service and the causes of death were
classified and coded according to the Ninth Revision
of the International Classification
up from the date of diagno-
31, 1996 (or until lost to follow-
of residence. As far as
of Diseases (ICD-9).
Overall and cause-specific
comparing the observed number of deaths in the cohort
with the corresponding expected
based on sex-, age-, period-specific
rates. The results were presented as standardised mor-
tality ratios (SMRs); the corresponding
dence intervals (CIs) were based on the assumption of
a Poisson distribution of observed deaths.
Relative survival was analysed by comparing
served cumulative survival proportion
the corresponding expected survival proportion based on
the national age, sex, and calendar year-specific life ta-
bles 26. The results were presented as relative survival ra-
tios (RSRs); the corresponding
the normal approximation of the binomial distribution.
For all hypothesis tests, two-tailed p values co.05 were
considered as significant.
mortality was analysed by
number of deaths
of the cohort to
95% CIs were based on
Table I shows the general characteristics
tality in the population
at diagnosis was 38.5 years (k16.4) with a 1:1.4 fe-
male/male ratio. The mean follow-up
years (range O-30).
A total of 93 patients died during the observation peri-
od, 1,872 were still alive at the end of follow-up,
the remaining 101 were lost during follow-up.
The observed number of deaths was of the same order
as that expected and no significant
observed with respect to sex of the patients.
There was a significantly higher rate of deaths in pa-
tients under 30 years of age at diagnosis and in those
diagnosed before 1974.
Table II shows the cause-specific mortality.
Although cancer mortality,
creased, as far as cancer sites are concerned, a signifi-
cant excess of deaths was observed for colorectal can-
cer and haemolymphopoietic
multiple myeloma and non-Hodgkin
age, at diagnosis, of the nine patients who died from
and the mor-
of 2,066 UC patients. Mean age
period was 6.1
as a whole, was not in-
neoplasms, in particular
Survival in Italian patients with UC
TsLle I. General characteristics and overall mortality in entire group of 2066 Italian patients with ulcerative colitis.
t!tlpwd N. %
ubrmll saw a% u”’
Overall 2066 100
12,211 93 92.1 1.0 0.8-I .2
Age at diagnosis [years)
Years after diagnosis
w Standardized mortality ratio lobssrvedexpect-ed number of death& and corresponding 95% confidence intewels; * pcO.05; abbreviations: see list.
We U. Causes of death in the entire gmup of 2066 ltalian patients with ulcerative colitis.
Gauws uf hatb @CD-8 cudas)
sm ‘J DIP% cl Id
Malignant neoplasms f140-2061
Digestive system (I 50-I 591
Rectum [I 541
Liver (I 551
Respiratory system [I 60-I 651
Haemolymphopoietic system (200-2081
Multiple myeloma (2031
Non-Hodgkin lymphoma (200 or 2021
All other malignancies 9 9.0 1.0 0.5-1.9
Ois. Circulatory system 1390-4591
27 36.1 0.8 0.5-1.1
Dis. Respiratory system (460~5191 1 5.5 0.2 0.0-I .I
Dis. Digestive system (520-5791
Injuries and adverse effects f600-9991
Complications from surgery (9981
All other injuries
All other causes of death
12 11.0 1.1 0.6-1.9
ial Standwdized mortality ratio lobsewetiexpacted number of deathsl. and corresponding 95% confidence intervals; peO,O5; abbreviations: see list.
It. Uiscido et al.
colorectal cancer was 42.7 years (k12.0).
curred, on average, 9.0 years (range 1-14) after diag-
nosis. Of these patients, only two were diagnosed be-
fore 1974. Mean age, at diagnosis, of the six patients
who died from malignancies
etic system was 48.4 years (k6.0) and deaths occurred,
on average, 8.3 years (range 6-12) after diagnosis. Di-
agnosis in all these patients had been made after 1974.
A significant deficit of deaths was observed for cancer
of the respiratory system.
A weak and non-significant
served for diseases of the circulatory
Proctocolitis was reported in 11 death certificates thus
explaining the excess of mortality
digestive tract. Mean age at diagnosis was 59.8 years
(k22.1) and deaths occurred,
(range 0- 18) after diagnosis. In only one of these 11
patients had the diagnosis of UC been made before
1974 and only two deaths occurred in patients aged un-
der 30 years at diagnosis.
Complications following surgery were mentioned
death certificates, entirely
mortality due to injuries and adverse effects. In none of
these patients had diagnosis of UC been made before
1974 and only one death occurred in a patient aged un-
der 30 years at diagnosis.
Among the remaining causes of death, diabetes melli-
tus was mentioned in 3 certificates (vs 2.6 expected),
ill-defined symptoms of the respiratory
certificates, and aplastic anaemia in one.
of the haemolymphopoi-
lower mortality was ob-
for diseases of the
on average, 5.4 years
explaining the excess of
system in 2
Trends in RSR of the entire group are shown in Figure
1. Observed and expected survivals did not significant-
ly differ: 15 years after diagnosis, RSR was estimated
to be 100.8% (95% CI, 97.1% - 104.7%).
Comparison of the trends in RSR of patients classified
according to the period in which the diagnosis was made
is shown in Figure 2. Two patterns were observed. The
first, referring to patients diagnosed before 1974, is char-
acterised by a continuous reduction in the RSR. The low
total number of patients in this group, however, means
that although RSR was low after 15 years, 86.0%, the
difference was not significant (95% CI, 69.2%-102.8%).
The pattern of patients diagnosed after 1975, is charac-
terised by RSR which is steadily increasing. RSRs were
estimated to be: 106.1% (95% CI, 102.2%-109.0%)
years after diagnosis in the patients entered between
1975 and 1979; 101.3% (95% CI, 98.4%-104.3%)
years after diagnosis in those entered between 1980 and
1984; 102.0% (95% CI, 100.3%-103.7%)
diagnosis in those entered between 1985 and 1989; and
100.1% (95% CI, 89.6%-101.6%)
sis in those entered after 1990.
8 years after
5 years after diagno-
12 3 15
Years after diagnosis
Fm. 1. Relative survival ratio of entire group of patients with UC. Con-
tinuous line indicates trend in relative survival ratio bf group, where-
as broken lines indicate confidence interval. 15 years after diagnosis
survival is similar to that in general population.
3 6 9 12 15
Years aftw dlagnc&
Fig. 2. Trends in relative survival ratios of patients classified accord-
ng to period in which diagnosis was made (before 1974, from 1975
;o 1979, from 1980 to 1994, from 1985 to 1989, after 19901. Pa-
;ients diagnosed after 1975 had a better survival with respect to
Ghose diagnosed earlier. The deficit of survival in patients diagnosed
oefore 1974 was not due to longer follow-up period because it was
dready evident two years after diagnosis.
Data emerging from the present study show that over-
all mortality in patients with UC in Italy is similar to
that in the general population.
1975 had a somewhat better survival with respect to
Patients diagnosed after
Survival in Italian patients with UC
those diagnosed earlier. The group studied included all
the cases consecutively diagnosed in twenty Gastroen-
terology Units distributed throughout
ing in a GISC National Study Group. The study is,
therefore, to be considered a hospital-based
though UC patients attending GI units are usually the
most severely affected, the mortality
this study was of the same order as that found in a pop-
ulation-based study recently performed
Florence, Italy (6.4 per 1,000 patients per year) l*. As-
sessment of the cause of death was carried out by death
certificates and ICD codes. This method of assessment
is the best that we can achieve although
some limitations. In fact, comparison
cates and hospital charts in a study assessing causes of
death in inflammatory bowel diseases found some dis-
cordance in assessment Ix.
Premature death in UC is considered to be due not on-
ly to the severe attacks of the disease with the associ-
ated complications and surgery-related
to the development of colorectal
course of the disease 2-5 I9 21-24. In this study, 12% of
deaths resulted from UC itself whilst those caused by
complications from surgery accounted for 5% of the
overall mortality. These data differ from earlier obser-
vations in which deaths attributed
counted for up to 50% of overall mortality 27, but are in
keeping with the current series ” ‘* ‘6-‘R. The mean age
at diagnosis of patients who died from UC was 60
years thus confirming other reports of a worse progno-
sis in patients with onset of the disease after 50 years
of age . 2 4 5 I* The finding of a marked peak in UC inci-
dence in patients over the age of 60 years in the Italian
people suggests that more attention
dressed to the management
As far as death from colorectal cancer is concerned, an
excess of deaths reaching 10% of all deaths was found.
These results are in keeping with those of other studies
in which the number of deaths attributed
cer varied between 5.5 and 14% of all deaths ” I2 ‘6-‘8.
This indicates that despite the improved
and surveillance procedures,
orectal carcinoma remains an important
Deaths from colorectal cancer occurred at a mean age
of 43 years, the mean period after the diagnosis of UC
being 9 years. This observation heeds a warning as the
current opinion is that the risk of developing
creases 10 years after disease onset 29-33. According
this finding, however, regular surveillance
considered also in UC patients in whom the disease has
been present for less than 10 years.
As far as deaths not related to colonic disease are con-
cerned, a significant excess
rate observed in
in the area of
it may have
of death certifi-
risks, but also
cancer in the late
to UC itself ac-
should be ad-
in this age of patients
to colon can-
the development of col-
of mortality from
ticular from multiple myeloma
phoma. UC is characterised
described in other chronic conditions
toid arthritis 34 35. Drug-induced
could also contribute
lymphatic tissue, even if the role of immunosuppres-
sive therapy on the risk of malignancies
clearly understood 36 37. Considering
signs of UC and haemolymphopoietic
can be very similar, early diagnosis can be masked and
appropriate treatment delayed 38.
Another finding emerging from this study was that pa-
tients with UC had a reduced risk of death from cancer
of the respiratory system, possibly related to the well-
known non-smoking habits characteristic
with UC 39 40. In fact, a trend toward a reduction
deaths from other smoking-related
diseases of the circulatory and respiratory systems was
Although the overall mortality
the patients aged under 30 years at diagnosis and those
diagnosed before 1974 showed a higher risk of death
than the general population.
not due to a specific cause. The higher mortality
tients under 30 years of age at diagnosis could derive
from the longer period of colonic and systemic inflam-
mation and need for medications
tality in patients diagnosed before 1974 was not due to
the longer follow-up period because the decrease of
survival was already evident two years after diagnosis.
It seems likely that several factors, such as better med-
ical management, earlier detection
earlier surgery and improved post-operative
account for the better survival rate after 1975 41. In fact,
the policy of prompt intensive medical treatment
led to a reduction in the overall (medical and surgical)
death rate from toxic megacolon
(1967-l 973) to 13% (after 1974) in the GI Unit of the
University of Rome 42.
In conclusion, results of the present study indicate that
overall mortality in patients with UC is similar to that
in the general Italian population.
itself occurred in those patients in whom diagnosis of
the disease had been made at an advanced age. Care of
these patients should probably
Deaths due to colorectal cancer exceeded the number
expected and occurred at a younger age i.e., an average
period of 9 years after diagnosis of UC. This finding
indicates that the surveillance
vised and that also patients with disease of short dura-
tion should be considered.
neoplasms was observed, in par-
by chronic inflammation
immune activation 2. The state of im-
could, therefore, enhance the risk for
in the lymphatic
such as rheuma-
transformation to malignant of
in UC is not
that the clinical
conditions such as
was not increased, both
In both cases, deaths were
13. The excess of mor-
dropping from 40%
Deaths caused by UC
be more intensive.
practice should be re-
Furthermore, an increased
A. Uiscido et al.
system was also observed. Considering
symptoms with UC, the possibility of a malignant lym-
phatic transformation should be borne in mind. On the
other hand, mortality from smoking-related
from neoplasms of the haemolymphopoietic
the overlap of
List of abbreviations
interval; IBD: inflammatory
disease; ICD: inter-
1996;3 &and I :417-27.
prognosis of inflammatory bowel dis-
ulcerative Arch Intern
0. Prognosis in ulcerative colitis. Med Clin North Am
SC. The course and prognosis of ulcera-
TS, Brunt PW, Mowat NAG. Non-specific
Scotland: a community Gastroenterology
SF, Zinsmeister AR.
ulcerative in a community. Gut
V. Colorectal can-
Gas- risk ulcerative colitis.
AC, Mayberry JF. Mortality
An epi- in Leicestershire,
Sci 1993;38:538-41. study. Dis
1970.1989. patients colitis:
et al. General
tis. N Engl
D, Trallori G, Saieva
risk and life
0, Edili E, D’Albasio G,
mortality of a population
bowel the Florence
Jackman RJ, Stickler
FW, Fisherman K, Jarnum
J in Crohn’s
and ulcerative Stand
S. Prior P, Dew MJ,
Saunders V, Waterhouse
JAH, Allan RN.
in ulcerative 1982;83:36-43.
CG, Zack M, Holmberg L, Adami
Stowe SP, Stormont
Am J Gastroenterol
A, Helmick C, Zack
JM, Stowe MM, Chessin
et al. The of death
colitis. extracolonic Cancer
M, Adami HO. Extracolonic
JF, Rhodes J. An international
JC, Thompson H, Waterhouse
Gut morbidity in colitis.
et al. Colorectal
G, et al. Malignancy
0, Kvist HK,
Norgaard P, Ockelmann
Stand J Gas-
Stevens A, Jewel1
study of pri-
Morson BC. Crohn’s disease
ulcerative of the large
and its distinction
DG. A note on the calculation
of liver by the survival transplant
Monsen U, Nordenwall
J, Hellers G.
in Stockholm Coun-
e de1 retto
G, Miglio F, Caprilli
per lo Studio
de1 Colon study.
Int J Epidemiol (GISC).
D, Kowdley KV,
colitis? or surveillance
Persson B, Veress
1996;3 I : 1195-204.
years’ of patients
malignant ulcerative colitis.
A. Is colonoscopic
D, Brostrom 0, Lofberg R, Persson
control A population case
A. A decision
features of cancer in
P, Feltelius N. Klareskog
J 1998; 18: 180-I
activity in patients
Dickson M, Balkwill
Survival in Italian patients with UC
3’ Farrell RJ, Ang Y, Kileen P, O’Brien
PWN, et al. Increased incidence of non-Hodgkin’s
inflammatory bowel disease patients on immunosuppressive
apy but overall risk is low. Gut 2000;47:514-9.
38 Lenzen R, Borchard F, Lubke H, Strohmeyer G. Colitis ulcerosa
complicated by malignant lymphoma: case report and analysis of
published works. Gut 1995;36:306-10.
39 Corrao G, Tragnone A, Caprilli
et al. Risk of inflammatory bowel disease attributable to smoking,
DS, Kelleher D, Keeling
R, Trallori G, Papi C, Andreoli A,
control study by GISC. Int J Epidemiol
‘” Present DH, Banks PA. Should smoking be used in the treatment
of ulcerative colitis? Inflamm Bowel Dis 1998;4:327.
” Sonnenberg A, Koch TR. Period and generation effects on mor-
tality from idiopathic inflammatory
” Torsoli A, Caprilli R. Fulminating
colon. Ital J Gastroenterol Hepatol 1978;10:245-7.
and breast-feeding in Italy: a nation-wide
bowel disease. Dig Dis Sci
colitis and impending mega-
Digestive and Liver Disease and The Italian Journal of Gastroenterology (Volumes 29-33)
are available on-line at the web site:
Possibility of on-line subscriptions.
Corso di Francis, 197
00191 Rome, Italy