American Journal of Gastroenterology


C 2007 by Am. Coll. of Gastroenterology
Published by Blackwell Publishing

ISSN 0002-9270
doi: 10.1111/j.1572-0241.2006.01000.x

Overall and Cause-Specific Mortality in Ulcerative Colitis:

Meta-analysis of Population-Based Inception
Cohort Studies
Tine Jess, M.D.,1 Michael Gamborg, M.S.,2 Pia Munkholm, M.D.,1 and Thorkild I. A. Srensen, M.D.2
1
Department of Medical Gastroenterology C, Herlev University Hospital, Copenhagen, Denmark; and 2 Danish
Epidemiology Science Center at the Institute of Preventive Medicine, Copenhagen University Hospital, Centre
for Health and Society, Copenhagen, Denmark

OBJECTIVES:

It remains debated whether patients with ulcerative colitis (UC) are at greater risk of dying and
whether a possible alteration in mortality can be attributed to specific causes of death. We aimed to
clarify this issue by conducting a meta-analysis of population-based inception cohort studies on
overall and cause-specific mortality in patients with UC.

METHODS:

The MEDLINE search engine and abstracts from international conferences were searched for relevant
literature by use of explicit search criteria. STATA meta-analysis software was used to calculate
pooled risk estimates (SMR, standardized mortality ratio, observed/expected deaths) of overall
mortality and specific causes of death and to conduct metaregression analyses of the influence of
specific variables on SMR.

RESULTS:

Ten papers fulfilled the inclusion criteria, reporting SMRs varying from 0.7 to 1.4. The overall pooled
estimate was 1.1 (95% confidence interval [CI] 0.91.2, P = 0.42). However, greater risk of dying
was observed during the first years of follow-up, in patients with extensive colitis, and in patients
from Scandinavia. Metaregression analysis showed an increase in SMR by increasing cohort size.
UC-related mortality accounted for 17% of all deaths. Mortality from gastrointestinal diseases,
nonalcoholic liver diseases, pulmonary embolisms, and respiratory diseases was increased whereas
mortality from pulmonary cancer was reduced.

CONCLUSIONS: The overall risk of dying in patients with UC did not differ from that of the background population,
although subgroups of patients were at greater risk of dying. The cause-of-death distribution seemed
to differ from that of the background population.
(Am J Gastroenterol 2007;102:609617)

INTRODUCTION

patient populations, tend to report a worse prognosis (2)

than do population-based studies (6). However, populationbased studies also differ in design and may consequently report conflicting results. Studies including both incident and
prevalent cases are not always representative for the general
UC populationif patients, for instance, move to specific
areas for medical carewhereas studies based on incident
cases onlyinception cohort studiestend to give a more
valid picture of the long-term prognosis in UC. Such knowledge is of great importance for proper information of patients, for access to reasonably priced health and life insurance, for education of physicians, and for accurate health-care
planning.
Hence, the aim of the present study was to conduct a metaanalysis of overall and cause-specific mortality in populationbased inception cohorts of patients with UC, to describe UCrelated mortality, and to determine whether specific study or
patient population characteristics influenced mortality estimates.

It remains debated whether patients with ulcerative colitis

(UC) are at a greater risk of dying compared with the general
population. A number of prognostic studies are available, but
results differ markedly. Older studies tend to report a reduced
overall survival in UC patients (1, 2) whereas an improvement
in survival in patients diagnosed after year 1980 has been
observed (3).
It is also uncertain if patients with UC are at a greater
risk of dying from specific causes, in particular the disease
itself. Studies on this issue are often founded on small patient populations and even smaller numbers of deaths (35),
which makes it difficult to reveal differences in cause-specific
mortality between UC patients and the general population.
However, the greatest hindrance in determining the true
prognosis of UC is the difference in study designs represented in the available literature on the subject. Studies from
referral centers, based on selected and presumably more ill
609

610

Jess et al.

MATERIAL AND METHODS

Literature Search
In order to identify papers in English concerning overall and
cause-specific mortality in patients with UC deriving from
well-defined population-based inception cohorts, we conducted a systematic MEDLINE search from January 1965
through March 2006 using the MESH headings inflammatory
bowel disease (IBD)/epidemiology and UC/epidemiology
combined with a free text search on the words: survival OR
mortality, with the following limitations: only items with abstracts and studies of humans. In addition, recent abstracts
available on CD-ROM from the American Digestive Disease Weeks (20022005) and the United European Gastroenterology Weeks (20022005) were searched for ulcerative
colitis, and lastly, reference lists of included papers were
scrutinized to disclose additional literature on the topic in a
multiphase process.
Inclusion and Exclusion Criteria
To be included in the meta-analysis, papers should be based
on population-based inception cohorts (i.e., including all pediatric and adult incident cases in a given geographic area in
a given time period) of patients with UC diagnosed according
to well-defined criteria. In addition, papers should be generally available, contain information on total number of patients
followed, exact number of deaths occurring in the cohort
during follow-up, expected number of deaths in a matched
background population, and/or rates of observed to expected
deaths with 95% confidence intervals (CI). Studies which
only contained information on cumulative survival, studies
representing subpopulations or selected populations (e.g.,
general practitioner databases, health insurance databases,
hospitalization-based patient populations, pediatric cohorts,
and referral center populations), studies reporting combined
Crohns disease (CD)UC outcome, and reviews were excluded. In case of duplicate publications, the paper providing
the longest follow-up of patients was used.
Data Collection
A total of 295 papers was identified and 10 papers were ultimately included in the meta-analysis. Of these, eight papers
(3, 4, 611) met the inclusion criteria completely, whereas
two papers (5, 12) were close to fulfilling the criteria. The
latter two were an inception cohort study from Auckland,
New Zealand based on information from in- and outpatient
registries at all public hospitals, which apparently provided
almost all medical care in the area (thus being close to population based) (12), and a population-based study from Florence,
Italy assessing mortality in a regional cohort of IBD patients
aged >15 yr at diagnosis, with the majority of cases being
incident cases (N = 593) and a minor part being prevalent
cases (N = 96) (5, 13). The 10 studies are presented in Table
1. The papers from Copenhagen (6), Florence (5), Olmsted
(3), and Stockholm (7) were all extensions of former studies (1420). Six (38) of the 10 papers (60%) reported on

cause-specific mortality. In addition to the 10 included studies, a paper from Spain reported on mortality among 251 UC
patients diagnosed in Gijon during 19541997 (21). However, it was unclear if only hospitalized patients in the region
had been studied, and the standardized mortality ratio (SMR)
from this paper was therefore only used in a separate pooled
estimate of SMR and not in subanalyses.
The included papers were reviewed in detail in order to
record data on number of patients studied, female to male
ratio, age distribution at diagnosis, prevalence of patients
with proctitis only at diagnosis, calendar year of publication, calendar period of inclusion and observation, duration
of follow-up, number of deaths observed during follow-up
of the cohort, expected numbers in a matched background
population, and/or observed to expected mortality rates with
95% CIs (overall and stratified for clinical characteristics),
mortality rates by cause of death (if available), and classification of these (International Classification of Diseases [ICD]
number) (7, 8, 9, or 10).
Statistical Analysis
If SMRs (observed/expected number of deaths) or identical
estimates of survival were reported without 95% CIs, the
interval was calculated using observed and expected number
of deaths and assuming a Poisson distribution of observed
cases (Table 1).
Pooled SMRs with 95% CIs for overall mortality and
cause-specific mortality were calculated using the STATA
meta-analysis program (Stata Corporation, College Station
Texas; www.stata.com). According to the heterogeneity test
(significance at a 5% level) either a fixed or a random effects
model was applied.
In addition, metaregression analyses were performed in
order to evaluate whether the overall SMR was influenced
by cohort size, mean or median observation time, middle
year in the inclusion and observation period, calendar year of
publication, percentage of patients being men, percentage of
patients aged <40 yr at diagnosis, and percentage of cohort
patients with proctitis only at diagnosis.

RESULTS
Overall Mortality
The 10 included studies reported SMRs ranging from 0.7
(95% CI 0.60.9) (5) to 1.4 (95% CI 1.21.5) (8) (Table 1).
Data were pooled by use of the random effects model as
the heterogeneity test was significant ( 2 = 54, P < 0.001).
The pooled estimate revealed an overall SMR of 1.1 (95%
CI 0.91.2, P = 0.42) (Fig. 1). Exclusion of the two studies
reporting the lowest and the highest SMR did not change
the result (SMR 1.1, 95% CI 0.91.2, P = 0.24) neither did
exclusion of the close-to-fulfilling-the-criteria studies from
Italy and New Zealand (SMR 1.1, 95% CI 1.01.3, P = 0.06)
(Table 2). The additional possible-population-based study
from Spain reported a markedly higher SMR than did any

Meta-analysis of Mortality in Ulcerative Colitis

611

Table 1. Standardized Mortality Ratios (SMRs) and Patient Population Characteristics: A Meta-analysis of Population-Based Inception
Cohort Studies of Patients With Ulcerative Colitis

Author,
Country
Jess et al., United States (3)
Persson et al., Sweden (7)
Stewenius et al., Sweden (4)
Winther et al., Denmark (6)
Ekbom et al., Sweden (8)
Eason et al., New Zealand (12)
Probert et al., United Kingdom (9)
Masala et al., Italy (5)
Jacobsen et al., Denmark (10)
Hie et al., Europe (11)

Calendar
Mean or
Patients With Patients Aged Observed
Period (Publication No. of Median
Proctitis at <40 Years /Expected
Year)
Patients Follow-Up Males
Diagnosis at Diagnosis Deaths SMR (95% CI)
19402004
(2005)
19551990
(1996)
19581990
(1995)
19621997
(2003)
19651983
(1992)
19691978
(1982)
19721989
(1993)
19782001
(2004)
19782003
(2005)
19912003
(2005)

378

15 yr

56%

17%

63%

1,573

53%

26%

67%

471

14.8 yr

57%

58%

86/66.2

1.3 (1.01.5)

1,160

19 yr

47%

44%

61%

261/249

1.1 (0.91.2)

2,509

57% (30)

43%

62% (29)

342

4 yr

60%

63%

10/13

0.8 (0.41.4)

1,014

55% (31)

48%

92/98.3

0.9 (0.81.1)

689

14.8 yr

56%

1,515

1.1 (0.91.3)

792

10 yr

53%

75/67.9

1.1 (0.91.4)

of the other studies (SMR 7.1, 95% CI 1.521), however,

because of small numbers, inclusion of this study did not
alter the overall pooled SMR (SMR 1.1, 95% CI 0.91.3,
P = 0.30).
Mortality Stratified by Cohort Characteristics
SMRs stratified by specific cohort characteristics are presented in Table 2 as highest, lowest, and pooled estimates.
The five Scandinavian studies reported higher mortality
rates (pooled SMR 1.2, 95% CI 1.11.4, P = 0.001) than
did the non-Scandinavian countries (pooled SMR 0.8, 95%
CI 0.70.9, P = 0.001). Correspondingly, the EC-IBD study
reported a slightly higher mortality in northern Europe (SMR
1.2, 95% CI 0.91.5) than in southern Europe (SMR 0.9, 95%
CI 0.51.4) (11).
Four studies presented SMRs stratified by gender (3, 6, 10,
11). Neither women (pooled SMR 1.0, 95% CI 0.91.2, P =
0.75) nor men (pooled SMR 1.0, 95% CI 0.91.1, P = 0.95)
were at greater risk of dying.
Only two studies had stratified SMR for calendar year at
diagnosis (3, 6). By use of observed and expected numbers
in different calendar year intervals, we recalculated SMRs
for patients diagnosed prior to 1980, respectively, in 1980 or
thereafter. The resulting pooled estimates were similar in the
two groups of patients (SMR 1.0, 95% CI 0.91.2, P = 0.70,
respectively, SMR 0.9, 95% CI 0.71.1, P = 0.39).
Three studies presented SMRs stratified for age at diagnosis (3, 6, 9). Because two of the three SMRs for age 018/19
yr at diagnosis were 0, no pooled estimate could be calculated. Pooled SMRs disclosed no greater mortality among
patients aged 19/2029 yr (SMR 0.9, 95% CI 0.51.5, P =

62/79.2

0.8 (0.61.0)

255/186.8 1.4 (1.21.5)

505/360.7 1.4 (1.21.5)

81/115.7 0.7 (0.60.9)

0.69), 3049 yr (SMR 0.9, 95% CI 0.71.1, P = 0.33), or

50 yr (SMR 1.0, 95% CI 0.91.1, P = 0.69) at diagnosis.
Regarding extent of UC at diagnosis, four studies presented
SMRs stratified for proctitis and more extensive colitis (6)
or for proctitis, left-sided colitis, and more extensive colitis
(810). The pooled estimate in patients with proctitis was
1.0 (95% CI 0.91.1, P = 0.33). SMRs for patients with
inflammation proximal to the rectum were recalculated and
resulted in a significantly greater pooled estimate of 1.2 (95%
CI 1.01.5, P = 0.047) (Table 2).
Two studies had stratified SMR by disease duration, but
in different ways. One study reported a significantly greater
risk of dying during the first 5 yr after diagnosis of UC (SMR
1.4, 95% CI 1.11.8), but no significantly altered risk in the
time intervals 610 yr, 1120 yr, 2130 yr, and 3136 yr
after diagnosis (6). The second study reported a greater risk
of dying during the first year after diagnosis (SMR 2.2, 95%
CI 1.53.3) (10) but not thereafter.
Another study had stratified SMR by treatment with immunosuppressive drugs, i.e., 6-mercaptopurine, azathioprine,
or infliximab. Mortality was significantly lower than expected
in patients never treated with such drugs (SMR 0.8, 95%
CI 0.60.99), and somewhat but nonsignificantly higher than
expected in patients who had at some point during the disease
course received one of these medications (SMR 1.3, 95% CI
0.33.8) (3).
Lastly, one study had stratified SMR by surgery, reporting a slightly higher than expected risk of dying among patients who had undergone colectomy (SMR 1.3, 95% CI 0.7
2.2) and a slightly lower than expected risk in those who
had not undergone surgery (SMR 0.9, 95% CI 0.71.1). The

612

Jess et al.

nal infarction, myocardial infarction second to anemia, and

end-stage liver disease second to primary sclerosing cholangitis.

Figure 1. Individual and combined standardized mortality ratios

with 95% confidence intervals in ulcerative colitis: A meta-analysis
of population-based inception cohort studies. The size of the boxes
is proportional to the weight (1/SE) of each study.

difference was not statistically significant ( 2 = 1.3, NS)

(9).
Metaregression Analyses
Metaregression analysis revealed that cohort size had a significant impact on the observed risk of dying, as SMR increased
by 2.2% (95% CI 0.44.1%, P = 0.02) per additional 100 patients observed (Fig. 2). Middle year in the patient inclusion
period, middle year in the observation period, and calendar
year of publication had no significant impact on SMR. The included studies reported similar proportions of patients being
men, patients aged <40 yr at diagnosis, and patients having
proctitis only (Table 1). Consequently, metaregression analyses revealed no significant influence of these variables on
SMR.
UC-Related Mortality
Five (36, 12) of the 10 studies reported on UC-related mortality (Table 3). The mean percentage of deaths ascribed to
UC was 17% (range 11% (5) to 30% (12)). The most common causes were colorectal cancer (CRC) (mean 37%, range
24% (6) to 44% (5)) and surgical or postoperative complications (mean 44%, range 17% (3) to 100% (12)) such as
perforations, peritonitis, and cardiovascular complications.
The remaining causes were primarily related to severe disease, i.e., toxic megacolon, intestinal perforation, intesti-

Cause-Specific Mortality
Six studies (38) reported SMRs for specific causes of death
using ICD-7, ICD-8, ICD-9, and ICD-10 codes. SMRs for
comparable categories of causes of death deriving from the
four classifications were pooled in the present meta-analysis
and presented in Table 4 together with the lowest and the
highest SMRs observed in each category.
Although the overall risk of dying from cancer among patients with UC was similar to expected (SMR 1.0, 95% CI
0.71.3, P = 0.78), a borderline-significant greater risk of
dying from CRC (SMR 1.9, 95% CI 1.03.8, P = 0.07) and
a significantly reduced risk of dying from pulmonary cancer (SMR 0.3, 95% CI 0.10.9, P = 0.04) were observed.
According to metaregression analyses, the only studied variable with influence on the SMR of CRC was year of publication of the study (13% reduction in CRC-SMR/year
of publication, 95% CI 22% to 4%, P = 0.006). One
study reported on risk of dying of leukemia (SMR 2.9, 95%
CI 0.410.3) (6) and one study reported on risk of dying
from non-Hodgkins lymphoma (SMR 2.4, 95% CI 0.38.7)
(5).
Two studies had observed a significantly reduced risk of
dying from cardiovascular diseases (SMR 0.6, 95% CI 0.4
0.9 (3) and SMR 0.7, 95% CI 0.50.95(5)) whereas no studies
reported a greater risk. The pooled SMR was, however, not
significantly reduced (SMR 0.9, 95% CI 0.71.1, P = 0.25).
Only one study had estimated the risk of dying from pulmonary embolism, observing a significantly greater SMR of
4.0 (95% CI 1.58.7).
Five studies reported on SMRs for death because of respiratory diseases resulting in a significantly greater pooled
SMR of 1.6 (95% CI 1.32.0, P < 0.001). The excess mortality was partly as a result of a nonsignificantly greater risk of
dying from chronic obstructive pulmonary diseases (COPD,
i.e., bronchitis, asthma, and emphysema) (SMR 1.6, 95% CI
0.73.7, P = 0.26) and a significantly greater risk of dying
from pneumonia (SMR 3.1, 95% CI 2.04.6, P < 0.001).
Metaregression analysis revealed no significant influence of
any of the above mentioned study variables on the SMR for
death from respiratory diseases.
The risk of dying from gastrointestinal and liver diseases
was also increased (pooled SMR 2.5, 95% CI 1.93.2, P <
0.001) although this finding did not reach statistical significance when excluding deaths from UC (SMR 1.7, 95% CI
0.83.6, P = 0.18). In addition, a greater risk of dying from
nonalcoholic liver diseases was observed (SMR 4.0, 95% CI
2.56.5, P < 0.001).
The risk of dying from genitourinary tract diseases (SMR
1.2, 95% CI 0.72.2, P = 0.48), suicide (SMR 1.3, 95%
CI 0.82.0, P = 0.23), or accidents/injuries (SMR 0.7, 95%
CI 0.51.0, P = 0.06) was not significantly different from
expected.

Meta-analysis of Mortality in Ulcerative Colitis

613

Table 2. Pooled Estimates of Mortality in Patients with Ulcerative Colitis: A Meta-analysis of Population-Based Inception Cohort Studies

Overall survival
Excluding the lowest and highest SMRs
Excluding the two borderline studies (5, 12)
Nationality
Scandinavian studies
Non-Scandinavian studies
Gender
Female
Male
Calendar year at diagnosis
<1980
1980
Age at diagnosis
018/19 yr
19/2029 yr
3049 yr
50+ yr
Extent of disease
Proctitis
More extensive colitis
Disease duration
01 yr
05 yr
Immunosuppressive drugs
No treatment
Any treatment
Surgery
No surgery
Proctocolectomy

No. of
Studies

Lowest
SMR

95% CI

Highest
SMR

95% CI

Pooled
SMR

95% CI

10
8
8
9
5
4
4
4
4
2
2
2
3
3
3
3
3
4
4
4
2
1
1
1
1
1
1
1
1

0.7
0.8
0.8

0.60.9
0.41.4
0.61.0

1.4
1.4
1.4

1.21.5
1.21.5
1.21.5

1.1
1.1
1.1

0.91.2
0.91.2
1.01.3

1.1
0.7

0.91.2
0.60.9

1.4
0.9

1.21.5
0.81.1

1.2
0.8

1.11.4
0.70.9

0.7
0.8

0.51.1
0.61.2

1.4
1.1

1.01.9
0.91.3

1.0
1.0

0.91.2
0.91.1

0.9
0.6

0.61.1
0.31.1

1.1
1.0

0.91.3
0.81.3

1.0
0.9

0.91.2
0.71.1

0.0
0.4
0.6
0.9

0.03.6
0.11.2
0.31.0
0.71.3

2.3
1.1
1.5
1.1

0.94.7
0.61.8
0.63.0
0.91.2

0.9
0.9
1.0

0.51.5
0.71.1
0.91.1

0.8
0.9

0.51.2
0.71.2

1.0
1.5

0.91.2
1.41.7

1.0
1.2

0.91.1
1.01.5

2.2
1.4

1.53.3
1.11.8

0.8

0.61.0

1.3

0.33.8

0.9

0.71.1

1.3

0.72.2

95% confidence interval excluding 1.0 (P < 0.05).

DISCUSSION
The present meta-analysis, based on population-based inception cohort studies only, revealed that overall mortality in
patients with UC did not differ from that of the background
population. However, mortality was significantly increased in
patients during the first years of follow-up, in patients with extensive colitis, and in patients from Scandinavia, and significantly reduced in patients who had never been treated with immunosuppressive drugs. Metaregression showed a significant
increase in SMR with increasing cohort size. UC-associated
mortality accounted for approximately 17% of all deaths and,
accordingly, cause-specific analyses revealed a greater risk of
dying from gastrointestinal diseases. Furthermore, mortality from nonalcoholic liver diseases, pulmonary embolisms,
and respiratory diseases was significantly increased, whereas
mortality from pulmonary cancer was significantly reduced.
The primary strengths of our study are the application
of strict search and inclusion criteria and the assessment of
population-based inception cohort studies only (implicitly assuming that such studies represent all cases with the disease
in a given population). Studies from North America, Europe,
and New Zealand were available. The majority of these studies contained a fairly high quantity of data on demographic
and clinical patient characteristics, on stratification of SMR,

and on cause-specific mortality. A limitation of the present

meta-analysis is the varying use of ICD codes, which only
gave us the possibility to study main categories of causes of
deaths. Yet, if all included studies had used the same version
of ICD, the observed number of deaths would probably still
have been too small to give reason for further subanalyses. A
general drawback of studies based on death certificate information is the known risk of classification errors (22). In most
of the studies included in the present analysis the recorded
causes of death had been verified via hospital files.
Importantly, we found that overall mortality in patients
with UC did not differ from that of the background population. There was, however, a fair heterogeneity between
studies, especially considering the excess mortality observed
in Scandinavian countries versus the reduced mortality observed in Italy, New Zealand, the United Kingdom, and the
United States. This difference may reflect the fact that the
Scandinavian studies were based on some of the oldest patient populations. Accordingly, the lowest mortality (SMR
0.7, 95% CI 0.60.9) (5) was observed in a recent study from
Italy, which may, however, have been biased by the exclusion
of pediatric cases, who probably carry a higher relative mortality than adult patients (6, 23). Actually, stratification of
SMR by calendar year at diagnosis and metaregression analyses of the influence of middle year in the patient inclusion

614

Jess et al.

Figure 2. Impact of cohort size on standardized mortality ratio (SMR) in patients with ulcerative colitis: A meta-analysis of population-based
inception cohort studies. The size of the circles is proportional to the weight (1% SE) of each study.

period on SMR both failed to show a reduction in mortality

over time.
Another explanation for the observed heterogeneity between studies may be differences in cohort compositions regarding gender, age, and calendar periods of inclusion. The
estimation of SMRs from calculation of expected numbers
of deaths on the basis of age- and gender-matched mortality rates from the background populations in the same time
periods has the implicit assumption that the ratio between
observed and expected number of deaths is the same across
the entire joint distribution of sex, age, and time period,
which may not be true. However, the age and gender distribution seemed fairly similar in the included studies, and metaregression revealed no impact of calendar period of inclusion
or observation on the observed SMR.
A third explanation for the observed country-specific difference in mortality from UC may be differences in treatment regimens between countries. Unfortunately, only a few
studies reported on treatment of UC and mainly in a descriptive way, which did not leave us the opportunity to perform

direct comparison of data or perform additional metaregression analyses. Typically, patients had received maintenance
therapy with 5-aminosalicylates, corticosteroids in case of
disease flare, and had undergone proctocolecomy in lack
of response to medical treatment (3, 6, 12). Surgery rates
were either reported as crude rates (9, 12) or as cumulative probabilities (3, 12) and it was not always clear whether
they included all kinds of surgery or total colectomy only.
Comparable data were available to show that the cumulative probability of proctocolectomy after 15 yr was higher
among Danish (30%) (24) than American (19%) (25) patients with UC. Although surgery could be expected to reduce
mortality by curing UC and removing the target for CRC, it is
more likely that surgery increases mortality because of postoperative complicationswhich explained 44% of UCrelated deaths in the present study. Accordingly, an excess
mortality was observed in newly diagnosed patients and in
patients with extensive disease, in whom surgery rates are
known to be high (24). Probert et al., on the other hand, failed
to show a significant difference in mortality among patients

Table 3. Ulcerative Colitis-Related Mortality: A Meta-analysis of Population-Based Inception Cohort Studies

Author,
Country
Jess et al., United States (3)
Stewenius et al., Sweden (4)
Winther et al., Denmark (6)
Eason et al., New Zealand (12)
Masala et al., Italy (5)

Deaths Because of
Deaths Related Surgical/Postoperative
Deaths Because of
Deaths Because
No. of Observed
to UC
Complications
Colorectal Cancer
of PSC No. (% of
Patients Deaths No. (% of All) No. (% of UC Related) No. (% of UC Related)
UC Related)
378
471
1,160
342
689

62
103
261
10
81

12 (19%)
13 (13%)
33 (13%)
3 (30%)
9 (11%)

2 (17%)
3 (23%)
19 (58%)
3 (100%)
2 (22%)

5 (42%)
8 (24%)
4 (44%)

1 (8%)

Meta-analysis of Mortality in Ulcerative Colitis

615

Table 4. Cause-Specific Mortality in Patients with Ulcerative Colitis: A Meta-analysis of Population-Based Inception Cohort Studies
Causes of Deaths
Cancer
Colorectal cancer
Pulmonary cancer
Leukemia
Non-Hodgkins lymphoma
Cardiovascular diseases
Ischemic heart disease
Pulmonary embolism
Respiratory diseases
COPD
Pneumonia
Gastrointestinal and liver diseases
All exclusive ulcerative colitis
Non-alcoholic liver diseases
Genitourinary tract diseases
Suicide
Accidents/injuries

No. of Studies
5
4
2
1
1
5
2
1
5
4
2
4
3
3
4
4
5

Lowest SMR
0.7
0.9
0.3

0.6
0.9

0.9
0.6
1.3
1.6
0.5
0.9
1.0
0.8
0.5

95% CI

Highest SMR

0.50.9
0.41.8
0.11.0

0.40.9
0.81.1

0.32.0
0.12.3
0.33.7
0.82.8
0.21.1
0.14.9
0.42.3
0.31.8
0.21.2

1.5
4.4
0.4
2.9
2.4
1.1
1.0
4.0
2.0
3.4
3.4
4.0
2.8
4.8
1.6
2.1
0.8

95% CI

1.21.8
3.25.9
0.11.3
0.410.3
0.38.7
0.91.4
0.81.3
1.58.7
1.13.5
1.75.9
2.15.0
1.97.3
1.74.4
2.19.5
0.44.1
1.03.9
0.31.9

Pooled SMR

95% CI

1.0
1.9
0.3
2.9
2.4
0.9
0.9
4.0
1.6
1.6
3.1
2.5
1.7
4.0
1.2
1.3
0.7

0.71.3
1.03.8
0.10.9
0.410.3
0.38.7
0.71.1
0.81.1
1.58.7
1.32.0
0.73.7
2.04.6
1.93.2
0.83.6
2.56.5
0.72.2
0.82.0
0.51.0

95% confidence interval excluding 1.0 (P < 0.05). COPD = chronic obstructive pulmonary disease (bronchitis, emphysema, and asthma).

who had undergone surgery versus those who had not, but
CIs were fairly broad (9).
Medical treatment may also be expected to affect survival,
especially the increasing usage of immunosuppressive drugs,
which may either reduce mortality by controlling the intestinal inflammation or increase mortality by causing severe adverse events, such as opportunistic infections and lymphoma
(26). American patients who had never received immunosuppressive drugs were found to have a significantly reduced
mortality (3), probably because this subgroup of patients had
less severe disease and still had been followed closely by
physicians.
A fourth theoretical explanation for the observed heterogeneity between studies is that SMRs from different geographic areas may not be directly comparable. SMRs (i.e.,
observations in the clinical material divided by observations
in the background population) are calculated under the assumption that characteristics of the latter are reflected in the
patient material. However, one could hypothesize that the absolute mortality rate among patients with UC was the same
in different regions, whereas SMRs differed because of different background mortality rates, i.e., if the prevalence of
smokers (smoking being a determinant for death) differed
between regions but was fairly identical among UC patients,
this would consequently result in the observed heterogeneity
in SMRs. Likewise, if all UC patients were followed closely
by physicians and received early treatment for other disorders, such as cardiovascular disease, whereas the frequency
of physician visits and early detection and treatment varied
in the background population according to geography, then
SMRs might differ despite similar absolute mortality rates
among patients with UC.
Metaregression analysis revealed an increase in mortality
by increasing cohort size, contradicting the more expected
finding of greater mortality in smaller and potentially more
selected cohorts. However, if the above mentioned factors

(differences in cohort age, cohort compositions, and/or treatment policies) explained the higher mortality observed in
Scandinavian studies, then the metaregression result could
also be secondary to the fact that the Scandinavian studies
were based on the largest cohorts. On the basis of available
data, we were not able to determine whether cohort size,
geography, and/or treatment regimens influenced SMR independently. However, the fair homogeneity in cohort characteristics, such as gender, age, and disease extent, somehow
decreased the likelihood that SMR was influenced by cohort
size (with the inborn risk of selection bias) or by geography
(expecting variations in environmental or genetic factors to
influence phenotypes). The most likely explanation remains
that patients have been treated differentlydespite similar
descriptions of treatment policies.
Concerning cause-specific mortality, pooled estimates revealed a greater risk of dying from gastrointestinal causes
including the disease itself and from nonalcoholic liver diseases and a borderline-significant risk of death from CRC.
This was in accordance with our finding of a 17% UC-related
mortality, covering death from severe disease, surgical interventions because of severe disease, end-stage liver disease
because of primary sclerosing cholangitis, and CRC. It is,
however, questionable whether it is correct to categorize death
from CRC as an UC-related complication if the overall
risk of this malignancy is not increased in population-based
settings of UC patients (25, 27). Furthermore, the varying
results on CRC mortality may be a result of populationbased differences in access to colonoscopy or other putative cancer preventive factors on which information was not
available.
We also observed a greater risk of dying from pulmonary
embolisms and respiratory diseases. UC patients are known
to be at a greater risk of thromboembolic events such as deep
vein thrombosis and pulmonary embolism (28). The excess
mortality from respiratory diseases was mainly a result of

616

Jess et al.

pneumoniaan expected complication in both chronically

ill patients and postoperative patients. However, we also observed a slightly greater mortality from COPD, which stands
in contrast both to the reduced mortality from pulmonary
cancer observed in the present study and to the fact that UC
patients often are nonsmokers. Furthermore, the difference
in smoking habits between UC patients and the background
population may per se confound the reported SMRs, as discussed above. Because the primary papers did not adjust for
smoking habits, we could not determine whether UC diagnosis was independently associated with a greater risk of death
from pulmonary embolisms and respiratory diseases.
It would have been of clinical interest to clarify whether
UC patients are at greater risk of dying from lymphoma, but
only a single risk estimate was available and CIs were too
wide to draw any reliable conclusions.
Considering the nonincreased overall mortality among UC
patients, one would intuitively expect the above mentioned
findings to be counterbalanced by a reduced mortality from
other causes. As mentioned before, it is likely that patients
followed closely by physicians carry a reduced risk of death
from certain causes because of early detection and treatment. The reduced risk of death from pulmonary cancer may
have counterweighted the risk of death from CRC, but not
the greater risk of dying from nonmalignant causes, as the
overall risk of death from cancer was close to unity. Some
studies have suggested that UC patients are at reduced risk
of dying from cardiovascular diseases (2, 3, 5) because of
the low prevalence of cigarette smokers among UC patients
and the low blood pressure in patients with extensive disease
second to sodium and water depletion (29). In our pooled
analysis we failed to confirm this association. Another possible explanation for the missing reduced mortality could be
the use of varying ICD codes in the included studies, which
limited our analysis to main categories of diseases, and the
fact that most studies used incomparable categories, such as
other causes or remaining causes, which may have hidden interesting information.
In conclusion, the present meta-analysis of populationbased inception cohort studies revealed an overall nonincreased mortality in patients with UC, although a subgroup
of patients with newly diagnosed and extensive disease carried a greater risk of dying from the disease itself and from
surgical complications to the disease. This excess mortality
seemed to be counterbalanced by a reduced mortality from
pulmonary cancer, possibly from cardiovascular diseases, or
from other yet unidentified causes.

STUDY HIGHLIGHTS
What Is Current Knowledge

r
r

Long-term prognosis in ulcerative colitis remains debated.

Results are often based on selected patient populations.

What Is New Here

r
r
r

Overall mortality is not increased among patients with

ulcerative colitis derived from population-based inception cohorts.
Subgroups of patients with newly diagnosed and extensive disease are at a greater risk of dying.
The cause-of-death distribution differs significantly
from that of the background population.

Reprint requests and correspondence: Tine Jess, M.D., Department of Medical Gastroenterology C, Herlev Hospital, University
of Copenhagen, 75 Herlev Ringvej, DK-2730 Herlev, Denmark.
Received July 28, 2006; accepted September 18, 2006.

REFERENCES
1. Edwards FC, Truelove SC. The course and prognosis of
ulcerative colitis. Gut 1963;41:299315.
2. Gyde S, Prior P, Dew MJ, et al. Mortality in ulcerative colitis.
Gastroenterology 1982;83:3643.
3. Jess T, Loftus Jr EV, Harmsen WS, et al. Survival and causespecific mortality in patients with inflammatory bowel disease: A long-term outcome study in Olmsted County, Minnesota, 19402004. Gut 2006;55:12485.
4. Stewenius J, Adnerhill I, Anderson H, et al. Incidence of
colorectal cancer and all cause mortality in non-selected
patients with ulcerative colitis and indeterminate colitis
in Malmo, Sweden. Int J Colorectal Dis 1995;10:117
22.
5. Masala G, Bagnoli S, Ceroti M, et al. Divergent patterns of
total and cancer mortality in ulcerative colitis and Crohns
disease patients: The Florence IBD study 19782001. Gut
2004;53:130913.
6. Winther KV, Jess T, Langholz E, et al. Survival and
cause-specific mortality in ulcerative colitis: Follow-up of a
population-based cohort in Copenhagen County. Gastroenterology 2003;125:157682.
7. Persson PG, Bernell O, Leijonmarck CE, et al. Survival
and cause-specific mortality in inflammatory bowel disease: A population-based cohort study. Gastroenterology
1996;110:133945.
8. Ekbom A, Helmick CG, Zack M, et al. Survival and causes
of death in patients with inflammatory bowel disease: A
population-based study. Gastroenterology 1992;103:954
60.
9. Probert CS, Jayanthi V, Wicks AC, et al. Mortality in
patients with ulcerative colitis in Leicestershire, 1972
1989. An epidemiological study. Dig Dis Sci 1993;38:538
41.
10. Jacobsen BA, Puho E, Fallingborg J, et al. Mortality in ulcerative colitis in the North Jutland County. A population based
26 years follow-up study [abstract]. Gut 2005;54(suppl
VII):A6.
11. Hie O, Schouten L, Wolters FL, et al. No increased mortality 10 years after diagnosis in a Europe-wide population
based cohort of ulcerative colitis patients (EC-IBD study
group) [abstract]. Gut 2005;54(suppl VII):A6.
12. Eason RJ, Lee SP, Tasman-Jones C. Inflammatory bowel
disease in Auckland, New Zealand. Aust N Z J Med
1982;12:12531.

Meta-analysis of Mortality in Ulcerative Colitis

13. Trallori G, Palli D, Saieva C, et al. A population-based study

of inflammatory bowel disease in Florence over 15 years
(197892). Scand J Gastroenterol 1996;31:8929.
14. Bonnevie O, Binder V, Anthonisen P, et al. The prognosis
of ulcerative colitis. Scand J Gastroenterol 1974;9:8191.
15. Hendriksen C, Kreiner S, Binder V. Long term prognosis in
ulcerative colitisbased on results from a regional patient
group from the county of Copenhagen. Gut 1985;26:158
63.
16. Langholz E, Munkholm P, Davidsen M, et al. Colorectal
cancer risk and mortality in patients with ulcerative colitis.
Gastroenterology 1992;103:144451.
17. Palli D, Trallori G, Saieva C, et al. General and cancer
specific mortality of a population based cohort of patients
with inflammatory bowel disease: The Florence Study. Gut
1998;42:1759.
18. Stonnington CM, Phillips SF, Zinsmeister AR, et al. Prognosis of chronic ulcerative colitis in a community. Gut
1987;28:12616.
19. Loftus EV Jr, Silverstein MD, Sandborn WJ, et al. Ulcerative colitis in Olmsted County, Minnesota, 19401993:
Incidence, prevalence, and survival. Gut 2000;46:336
43.
20. Brostrom O, Monsen U, Nordenwall B, et al. Prognosis and
mortality of ulcerative colitis in Stockholm County, 1955
1979. Scand J Gastroenterol 1987;22:90713.
21. Gismera CS, Fernandez ML, Fernandez GA, et al. Mortality
and causes of death in patients with chronic inflammatory
bowel disease in Gijon, Asturias (Spain). Rev Esp Enferm
Dig 1999;91:199208.
22. Nordenholtz KE, Stowe SP, Stormont JM, et al. The cause of
death in inflammatory bowel disease: A comparison of death
certificates and hospital charts in Rochester, New York. Am
J Gastroenterol 1995;90:92732.
23. Devroede GJ, Taylor WF, Sauer WG, et al. Cancer risk and

24.

25.

26.
27.

28.
29.
30.
31.

617

life expectancy of children with ulcerative colitis. N Engl J

Med 1971;285:1721.
Langholz E. Ulcerative colitis. An epidemiological study
based on a regional inception cohort, with special reference to disease course and prognosis. Dan Med Bull
1999;46:40015.
Jess T, Loftus EV Jr, Velayos FS, et al. Risk of intestinal
cancer in inflammatory bowel disease: A population-based
study from Olmsted County, Minnesota, 19402001. Gastroenterology 2006;130:103946.
Ljung T, Karlen P, Schmidt D, et al. Infliximab in inflammatory bowel disease: Clinical outcome in a population based
cohort from Stockholm County. Gut 2004;53:84953.
Winther KV, Jess T, Langholz E, et al. Long-term risk
of cancer in ulcerative colitis: A population-based cohort
study from Copenhagen County. Clin Gastroenterol Hepatol 2004;2:108895.
Koutroubakis IE. Therapy insight: Vascular complications
in patients with inflammatory bowel disease. Nat Clin Pract
Cardiovasc Med 2005;2:26672.
Gyde SN, Prior P, Alexander F, et al. Ulcerative colitis:
Why is the mortality from cardiovascular disease reduced?
Q J Med 1984;53:3517.
Ekbom A, Helmick C, Zack M, et al. The epidemiology
of inflammatory bowel disease: A large, population-based
study in Sweden. Gastroenterology 1991;100:3508.
Probert CS, Jayanthi V, Pinder D, et al. Epidemiological
study of ulcerative proctocolitis in Indian migrants and the
indigenous population of Leicestershire. Gut 1992;33:687
93.

CONFLICT OF INTEREST
The authors declare no potential conflicts of interest.