Article Text

Download PDFPDF

Original article
Mortality in adult-onset and elderly-onset IBD: a nationwide register-based cohort study 1964–2014
  1. Ola Olén1,2,3,
  2. Johan Askling1,
  3. Michael C Sachs1,
  4. Martin Neovius1,
  5. Karin E Smedby1,
  6. Anders Ekbom1,
  7. Jonas F Ludvigsson4,5,6,7
  1. 1Division of Clinical Epidemiology, Department of Medicine Solna, Karolinska Institutet, Stockholm, Sweden
  2. 2Department of Clinical Science and Education Södersjukhuset, Karolinska Institutet, Stockholm, Sweden
  3. 3Sachs’ Children and Youth Hospital, Stockholm South General Hospital, Stockholm, Sweden
  4. 4Department of Pediatrics, Örebro University Hospital, Örebro, Sweden
  5. 5Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
  6. 6Division of Epidemiology and Public Health, School of Medicine, University of Nottingham, Nottingham, UK
  7. 7Department of Medicine, Columbia University College of Physicians and Surgeons, New York, USA
  1. Correspondence to Dr Ola Olén, Division of Clinical Epidemiology, Department of Medicine Solna, Karolinska Institutet, Stockholm 17176, Sweden; ola.olen{at}


Objectives To examine all-cause and cause-specific mortality in adult-onset and elderly-onset IBD and to describe time trends in mortality over the past 50 years.

Design Swedish nationwide register-based cohort study 1964–2014, comparing mortality in 82 718 incident IBD cases (inpatient and non-primary outpatient care) with 10 times as many matched general population reference individuals (n=801 180) using multivariable Cox regression to estimate HRs. Among patients with IBD, the number of participants with elderly-onset (≥60 years) IBD was 17 873.

Results During 984 330 person-years of follow-up, 15 698/82 718 (19%) of all patients with IBD died (15.9/1000 person-years) compared with 121 095/801 180 (15.1%) of reference individuals, corresponding to an HR of 1.5 for IBD (95% CI=1.5 to 1.5 (HR=1.5; 95% CI=1.5 to 1.5 in elderly-onset IBD)) or one extra death each year per 263 patients. Mortality was increased specifically for UC (HR=1.4; 95% CI=1.4 to 1.5), Crohn’s disease (HR=1.6; 95% CI=1.6 to 1.7) and IBD-unclasssified (HR=1.6; 95% CI=1.5 to 1.8). IBD was linked to increased rates of multiple causes of death, including cardiovascular disease (HR=1.3; 1.3 to 1.3), malignancy (HR=1.4; 1.4 to 1.5) and digestive disease (HR=5.2; 95% CI=4.9 to 5.5). Relative mortality during the first 5 years of follow-up decreased significantly over time. Incident cases of 2002–2014 had 2.3 years shorter mean estimated life span than matched comparators.

Conclusions Adult-onset and elderly-onset patients with UC, Crohn’s disease and IBD-unclassified were all at increased risk of death. The increased mortality remained also after the introduction of biological therapies but has decreased over time.

  • IBD
  • Crohn’s disease
  • Ulcerative colitis
  • IBD unclassified
  • Indeterminate IBD
  • mortality
  • death
  • cause-specific
  • death cause
  • myocardial infarction
View Full Text

Statistics from

Significance of this study

What is already known on this subject?

  • IBD has been linked to an increased mortality. Whether this increased mortality is present also in patients with contemporary treatment, and whether it varies across age and IBD subtypes, remains unknown.

  • Studies on all-cause and cause-specific mortality (especially cardiovascular mortality) in IBD subtypes (UC, Crohn’s disease and IBD-unclassified) are contradictory or lacking.

What are the new findings?

  • Although excess mortality in IBD has decreased over calendar time, the mortality in IBD, across all ages and disease subtypes, remains higher than in the general population.

How might it impact on clinical practice in the foreseeable future?

  • Given the rise in IBD prevalence globally we need to optimise treatment but potentially also screening for cancer and preventive measures against cardiovascular disease to facilitate an efficient and cost-effective care of IBD.


IBD encompasses Crohn’s disease (CD), UC and IBD-unclassified (IBD-U). Although newer treatments have increased chances of remission, death in IBD is not infrequent, especially at onset1 and after emergency surgery.2

So far, there are more than 100 studies on mortality in IBD. When restricting the literature to studies in which rates and absolute numbers are presented, Bewtra et al were able to identify 35 studies.3 Since then, at least four large mortality studies have been published4–7 (online supplementary 1, eTable 1). Both the paper by Bewtra et al as well as the two meta-analyses preceding it (Jess et al8 and Duricova et al9) indicate a substantial study heterogeneity, potentially due to different study designs (single-centre vs multicentre inclusion) or period-dependent differences (older studies tend to report the highest mortality, especially in UC).3 Indeed, and somewhat disturbingly in light of the changes in the treatment paradigm for IBD during the last 20 years, there are only little data on mortality in patients diagnosed with either UC or CD after 2000.3 Furthermore, there are no studies on mortality in patients with IBD-U and only inconsistent data regarding cardiovascular and several other causes of death in IBD.3 7 10

There are few studies on elderly-onset IBD and mortality.11 Recently, Nguyen et al demonstrated a higher absolute mortality from IBD in elderly-onset compared with adult-onset IBD but that may merely reflect higher mortality rates from any cause in an older population. In addition, there are few studies with enough follow-up and power to explore any potential change in mortality since the introduction of modern IBD drugs (eg, biologics).4 6

In this paper, we examined whether mortality is higher in patients diagnosed with IBD as adults or elderly compared with matched general population reference individuals. In particular, we evaluated all-cause and cause-specific mortality in patients diagnosed with IBD in 2002 or later and in UC, CD and IBD-U separately. We also analysed whether mortality has decreased since the introduction of biologics.


Study design

In a nationwide register-based and population-based cohort study, we compared the mortality between individuals with adult-onset (≥18 years) and elderly-onset (≥60 years) IBD and matched general population reference individuals.


In Sweden, gastroenterologists and surgeons treat patients with IBD. We were able to link data from administrative and clinical national registers on demographics,12 medication,13 morbidity14 15 and mortality16 because of the unique personal identity number issued to all Swedish residents.17

Definition of IBD

IBD was defined as having at least two relevant International Classification of Diseases (ICD) codes in the Swedish Patient Registers (ICD codes used and time between the first and second code across all calendar periods are listed in online supplementary 1, eTable 2a,b). This register began in 1964, became nationwide in 1987, includes day surgery since 1997 and all hospital-based outpatient data since 2001.14. In an earlier validation, we found that 93% of patients with two or more IBD diagnostic listings were confirmed to have IBD according to patient chart data.18 At the start of the follow-up, UC, CD and IBD-U were defined as per the first two diagnostic codes only. All analyses of mortality in UC, CD and IBD-U were based on this definition.

Since the IBD subtype diagnosis may change over time, we also described characteristics of the study population at the end of follow-up. We then used a different definition of UC, CD and IBD-U based on diagnostic and surgery codes assigned during follow-up to reflect severity of the disease course (online supplementary 1, eTable 3a,b). Patients with IBD were further classified using ICD-10 codes divided according to the Montreal classification19 (online supplementary 1, eTable 4). We also categorised patients into those with and without primary sclerosing cholangitis or other extraintestinal manifestations (online supplementary 1, eTable 5).

Matched reference individuals

For each individual with IBD, the government agency Statistics Sweden identified up to 10 reference individuals through the Total Population Register,12 which covers the complete Swedish population. Reference individuals were matched with the patients with IBD for age, sex, place of residence and year of birth.


IBD-related surgery included colectomies, other bowel surgery and perianal surgery. Dates of surgery were available from inpatient care since January 1, 1964 and from outpatient specialist care since January 1, 1997. Surgical procedures were coded using the Nordic Medical Statistics Committee (NOMESCO) Classification of Surgical Procedures (online supplementary 1, eTable 6).


The Swedish Prescribed Drug Register started on July 1, 2005.13 The register covers all drugs except over-the-counter medication (which is not covered at all) and medications administered at hospitals (which is only covered to some extent in the Prescribed Drug Register and completely covered through the National Patient Register in some counties). All codes representing IBD drugs are summarised in online supplementary 1, eTable 7. The coverage of intravenous biologics and the potential effect on results are reported and discussed in online supplementary 1, eTable 8a,b.

Other covariates

Through the government agency Statistics Sweden, we obtained information on education as a proxy for socioeconomic status (compulsory school ≤9 years, upper secondary school 2 years, upper secondary school 3 years, university <3 years, university ≥3 years).20 Finally, for each individual (with or without IBD) we defined sex, age, date of birth and death and emigration from the Swedish Total Population Register.12

First-degree relatives

Through the Swedish Multi-Generation Register we linked first-degree relatives to the Patient and Cancer Registers, and so assessed the potential effect modification of family history of IBD or cancer. The Swedish Multi-Generation Register includes individuals born in 1932 and later and who are registered as residents in Sweden after 1961. The register contains information on the individuals and their biological parents, enabling linkage among parents, children and their siblings.21


Information on date of death was obtained from the Total Population Register.12 Some 93% of all deaths are reported to this register within 10 days after death and another 7% (total≈100%) within 30 days.12 We examined both all-cause and cause-specific mortality (for ICD codes, see online supplementary 1, eTable 8).

Time periods

Because some earlier studies have found a small but significantly increased risk of death in IBD based on patient materials from the 1980s and 1990s, and because both phenotype and treatment of IBD have changed over time, we a priori restricted sensitivity analyses to 2002 and later: until the year 2000, all patients with IBD were identified by inpatient care only. When the Swedish National Patient Register started to record also hospital-based outpatient care in 2001, some prevalent IBD cases would appear to be incident during 2001. From 2002 and later, all incident IBD cases can be captured in both inpatient and outpatient care. A description of the reasons behind looking back 1 year to ensure true incident cases since 2002 and proportion of IBD cases likely identified during the years before the start of the outpatient register are presented and discussed in online supplementary 1, eFigure 1.


To avoid immortal time bias, patients were considered at risk from the date of the second diagnosis of IBD (and corresponding date in matched reference individuals) out of the two diagnoses needed to fulfil the exposure definition. Follow-up ended with death, emigration, IBD status change (in the few instances when a reference individual developed IBD during follow-up) or December 31, 2014, whichever occurred first.

HRs for all-cause and cause-specific death in IBD (and separately for UC, CD and IBD-U) were calculated using multivariable Cox regression, adjusted for the matching variables (sex, age, birth year and place of residence). We also calculated the absolute risk of death per 1000 person-years. All analyses were performed for both all years (1964–2014) and restricted to 2002 and onwards.

We examined the rate of death by duration of follow-up (<1 year of follow-up, 1 to <5 years, ≥5 years of follow-up). We further examined mortality based on sex and age at first IBD diagnosis: 18–39 years, 40–59 years, 60–69 years and ≥70 years). We further calculated HRs for death according to primary sclerosing cholangitis and other extraintestinal manifestations as well as for IBD surgery (in those analyses follow-up started at the date of registration of the diagnostic code in question).

Given that the Prescribed Drug Register started in July 2005, we restricted our IBD mortality studies in which medications were involved to incident disease after July 1, 2005. In the analyses regarding IBD-related medication, mortality of patients with IBD was compared with their matched reference individuals and the risk time of patients with IBD was divided into four categories of time at risk (ie, a patient and his/her reference individuals could contribute risk time to more than one category): no immunomodulators or anti-tumour necrosis factor (TNF), immunomodulators only, anti-TNF only and both immunomodulators and anti-TNF. In the analyses of drug exposures follow-up started at the date of the first dispensing of that particular drug.

We also estimated time trends in relative mortality by plotting HRs for death during different intervals of follow-up in adult-onset IBD compared with the matched general population reference individuals by age group and year of disease onset (1964–2014). Finally, using the time scale of age rather than the time since index date, we also estimated the restricted mean survival age in adult-onset IBD and the reference group at the fixed time horizon of 100 years (stratified by year of IBD onset) by integrating the Kaplan–Meier survival curves from ages 0 to 100 years. This approach allowed assessment of survival differences between the groups without making the proportional hazards assumption.22

Statistical analyses were performed using R statistical software (version 3.3.1; R Foundation for Statistical Computing, Vienna, Austria) and the survival package (version 2.38; Therneau, T (2015), A p value <0.05 was considered statistically significant. CIs were computed by inversion of the likelihood ratio test statistic.

This study was approved by the Ethics Review Board in Stockholm. Informed consent was not deemed necessary because the study was strictly register based.23


Characteristics of the patients with IBD

We identified 82 718 individuals with adult-onset IBD (UC: n=50 684, CD: n=25 582 and IBD-U: n=6452) and 801 180 matched reference individuals (table 1 and online supplementary 2, eFigure 2). 49% of patients were women, the median age at first diagnosis was 41 years (IQR: 29–57 years) and 21% (n=17 873) of patients were categorised as elderly-onset IBD. Half of the patients were diagnosed between 2002 and 2014 (table 1).

Table 1

Baseline characteristics of adult-onset (≥18 years) IBD diagnosed in Sweden 1964–2014

The age distribution at first diagnosis of IBD and at the end of the follow-up of our cohort is reported in online supplementary 2, eFigure 3.

Some 57% of the patients were followed for at least 10 years. The subtypes of IBD, phenotypes and occurrence of complications at the end of the follow-up are reported in online supplementary 2, eTable 10. One-third of the patients with an incident diagnosis of IBD after the start of the Prescribed Drug Register (July 1, 2005) had a registered treatment with immunomodulators. One in 10 patients had a first-degree relative with IBD and 46% had a first-degree relative with cancer (online supplementary 2, eTable 10).

All-cause mortality

During 984 330 person-years of follow-up, 15 698 (19%) individuals with IBD died (15.9/1000 person-years). This exceeded the death rate in our reference population (121 095 deaths during 10 034 411 person-years (equal to 12.1/1000 person-years), figure 1; online supplementary 3, eTable 11a,b, eFigure 4), corresponding to an HR of 1.5 for IBD (95% CI=1.5 to 1.5) (figure 1; online supplementary 4, eTable 12a,b) and one extra death per 263 person-years. The restricted mean survival age during the first 100 years of follow-up (when follow-up starts at birth) was lower in adult-onset IBD than in matched reference individuals (patients with an IBD onset in 1964–2001: 3.1 years shorter mean survival, and patients with an IBD onset in 2002–2014: 2.3 years shorter mean survival; online supplementary 3, eFigure 5).

Figure 1

Kaplan-Meier cumulative incidence curves; proportion of adult-onset (≥18 years) patients with IBD at risk of death who died, by calendar period of start of follow-up. The numbers below the plots show the number at risk (top), and the number of events since the previous data point (bottom) by IBD status.

Although incidence rates and cumulative incidence of death naturally differed across age at IBD onset (online supplementary 3, eTable 11a,b, eFigure 4), HRs were similar in adult-onset (HR=1.5; 95% CI=1.5 to 1.5) and elderly-onset IBD (1.5; 95% CI=1.5 to 1.5). However, the highest HR was seen in patients with the youngest onset (≥18–<40 years: HR=1.7; 95% CI=1.7 to 1.8). Mortality was increased also specifically for UC (HR=1.4; 95% CI=1.4 to 1.4), CD (HR=1.6; 95% CI=1.6 to 1.7) and IBD-U (HR=1.6; 95% CI=1.5 to 1.7) (figure 2; online supplementary 4, eTable 12a,b). The HR for IBD was highest in the first year of follow-up but remained elevated also more than 5 years after start of follow-up (figure 2). The incidence rate was also higher in the first year of follow-up (29/1000 person-years) than after 5 years of follow-up (15/1000 person-years; online supplementary 3, eTable 11a,b). The absolute mortality was higher in men while HR for death was somewhat higher in women with IBD than in men with IBD (1.6 vs 1.4; figure 2; supplementary 4, eTables 11a,b, 12a,b).

Figure 2

Adjusted# HRs for all-cause mortality during all available follow-up in adult-onset (≥18 years) patients with IBD and matched general population reference individuals in 1964–2014. #Adjusted for sex, age at diagnosis, birth year and county. Numbers and forest plots in figure represent: N deaths in patients with IBD, HR (95% CI). CD, Crohn’s disease; UC, ulcerative colitis.

HRs for mortality from IBD were similar before and after 2002, even though absolute mortality was lower for later years (partly explained by the younger age at the end of the follow-up for more recent calendar periods (online supplementary 2, eFigure 3, supplementary 3, eTable 11a,b). In contrast to patients with IBD with low education, patients with onset since 2002 and with a university education ≥3 years did not have an increased mortality (figure 2, online supplementary 4, eTable 12b). The only IBD phenotype with a distinctly higher overall relative mortality was patients with concomitant primary sclerosing cholangitis (HR=3.3; 95% CI=3.0 to 3.6), with similar relative risks in UC, CD and IBD-U (figure 2, online supplementary 4, eTable 12a,b).

Medication data were limited to patients first diagnosed after July 2005. Most patients did not have a record of either immunomodulators or anti-TNF drugs: for these patients, the HR for death was 1.8 (95% CI=1.7 to 1.9) with a somewhat lower HR in patients treated with immunomodulators only (HR=1.3; 95% CI=1.1 to 1.5). Patients on anti-TNF were few and CIs were wide (see online supplementary 1 and 4, eTables 8a,b and 12b).

Cause-specific mortality

The most common causes of death in IBD were cardiovascular disease, malignancy, digestive disease and respiratory disease (in that order and all contributing >1000 deaths in patients with IBD) (figure 3). Death from these four causes was also more common in patients with IBD than in reference individuals, as were all of the categories of death causes, except that patients with IBD were at a lower risk of death from psychiatric disorders (HR=0.8; 95% CI=0.7 to 0.9). Patients with IBD were at a fivefold increased risk of death from digestive diseases (eg, severe colitis and liver failure, HR=5.2; 95% CI=4.9 to 5.5), 1.4-fold from malignancy (95% CI=1.4 to 1.5) and 1.3-fold from cardiovascular diseases (95% CI=1.3 to 1.3). HRs for digestive diseases and malignancies as causes of death were somewhat higher for deaths that occurred before than after 60 years of age and HRs for different causes of death were similar for UC, CD and IBD-U (online supplementary 5, eTable 13a–d). Mortality during the first year of follow-up was mainly driven by death from digestive diseases, cardiovascular death and malignancy (online supplementary 5, eTable 13i). In a post hoc analysis, we censored patients with IBD at the date of primary sclerosing cholangitis diagnosis which resulted in almost identical cause-specific mortality as in analyses without such censoring (data not shown).

Figure 3

Adjusted# HR of all-cause and cause-specific mortality in the cohort of patients with adult-onset (≥18 years) IBD during all available follow-up, compared with matched general population reference individuals (1964–2014). #Adjusted for sex, age at diagnosis, year of diagnosis and place of residence.

Elderly-onset IBD

Restricting our study population to the more than 17 000 individuals with IBD onset at ≥60 years of age, we still found an increased all-cause mortality (HR=1.5; 95% CI=1.4 to 1.5), with almost identical risk estimates in elderly-onset IBD diagnosed since 2002 (HR=1.5; 95% CI=1.5 to 1.6). HRs for different causes of death were similar for elderly-onset IBD and overall adult-onset IBD (online supplementary 5, eTable 13a–h).

Time trends

Since 1964, mortality has decreased for both patients with IBD and matched reference individuals from the general population (figure 4). Mortality has decreased more for patients with IBD and consequently, relative mortality during the first 5 years of follow-up in adult-onset IBD compared with matched reference individuals has decreased significantly in all age groups of IBD onset, particularly in the youngest age group. By contrast, relative mortality >5 years after the start of follow-up (62% of all deaths) has gone roughly unchanged (figure 5).

Figure 4

Kaplan–Meier cumulative incidence curves; proportion of adult-onset (≥18 years) patients with IBD and matched general population reference individuals at risk of death who died by age group of IBD onset and year of disease onset.

Figure 5

Adjusted# HRs and 95% CIs of death in the cohort of patients with adult-onset (≥18 years) IBD compared with matched general population reference individuals (1964–2014) by age group and year of disease onset. #Adjusted for sex, age at index date and place of residence. *Patients with a first diagnostic listing of IBD during the last years of the study period have not had the chance of being followed for as long as earlier cases.


Principal findings

In this nationwide population-based cohort study, adult-onset and elderly-onset patients with UC, CD and IBD-U all remained at an increased rate of death, also after more than 5 years of follow-up. This study confirms reports of increased mortality in IBD but adds to our knowledge by examining incident IBD since the introduction of biologics and reporting time trends in IBD mortality for the past 50 years. In addition, the study has the statistical power to calculate precise risk estimates for different causes of death according to age of IBD onset, age of death, and different subtypes and phenotypes of IBD.

Comparison with earlier literature

A notable finding in this study is the increased mortality in all IBD subtypes (UC, CD and IBD-U) and that this excess risk persisted in patients with incident IBD since 2002. Earlier reports have suggested a trend toward lower IBD mortality over time,4 especially in UC.1 3 Furthermore, with a shift to new medications (including biologics), it has been suggested that this will eventually eliminate any excess mortality in IBD.24 Meanwhile, there has been a fear that biological treatment per se could trigger an increase in cancer mortality, but this concern has not been realised so far.25 We could confirm that relative mortality in IBD during the first years of follow-up and for younger patients has indeed decreased. Yet, incident cases of IBD since 2002 (after the introduction of biologicals) still had a higher overall mortality than the general population: UC (+40%), CD (+70%) and IBD-U (+90%). These findings are in line with reports showing that relative mortality is higher in CD than in UC.3–6 The absolute mortality differences, however, were low, with only one extra death each year per 263 patients with IBD.

We found a 5.2-fold increased risk of death from digestive disease. It was beyond the scope of this study to explore individual digestive disease entities; however, except for IBD,11 liver disease was the most common condition that contributed to this figure (data not shown). Among the strata with the highest mortality in our paper were patients with primary sclerosing cholangitis. Both UC (HR=3.4) and CD (HR=3.1) patients with primary sclerosing cholangitis were at increased risk of death, confirming earlier data from Denmark.26

Our findings of an increased risk of cardiovascular death contrasts with previous data,3 although consistent with observations of an excess risk in incident cardiovascular disease in patients with IBD.10 Explanations for the discrepancy regarding cardiovascular mortality between our study and for example, the meta-analysis of Bewtra et al3 include residual confounding and potential geographical differences. We confirm studies reporting an increased mortality because of cancer and respiratory disease in IBD, but as for respiratory disease, our HRs were higher (about 2.0) than in most earlier studies (often about 1.2–1.6).1 3 6 26

We found that overall HRs were identical in adult-onset IBD and elderly-onset IBD in Sweden (HR=1.5), which is in accord with previous findings of Chu et al.7 Data on whether elderly-onset IBD is milder than younger-onset IBD are conflicting.11 27–29 Nguyen et al11 showed that IBD-specific absolute mortality was higher in elderly-onset CD (but not UC) than in younger age groups. The larger number of elderly patients with IBD in our study (17 873 as opposed to 1749 in the Canadian study11) also enabled us to examine elderly according to disease phenotype.

Ours is the first study to report extensive data on mortality in IBD-U. In our study, IBD-U (representing about 8% of all IBD cases in our cohort) was associated with a 64% increased risk of death.

Strengths and limitations

This paper has several strengths. First, the study was based on a population-based cohort of patients with IBD from throughout Sweden. That the Swedish National Patient Register not only records inpatient data but also hospital-based outpatient data since 2001 implies that selection bias for IBD is highly unlikely. Despite this, our HRs were higher than in most studies. We required two or more diagnostic listings of IBD18 when identifying cases, and this definition has been used also by others.30 We have shown that this definition has a positive predictive value of approximately 93%.18 In our previous validation study, we did not investigate sensitivity but in Denmark (where the healthcare system is almost identical to that in Sweden) the Patient Register detects 94% of all patients with IBD.31 The use of nationwide healthcare registers means that loss of follow-up before emigration or death is virtually non-existent.

More than 15 000 patients with IBD died during follow-up. The large numbers allowed us to calculate HRs for death. For instance, we calculated HRs in patients with IBD with concomitant primary sclerosing cholangitis, heredity for IBD and cancer, and in patients who had undergone IBD surgery. We used ICD codes to characterise patients. Still, we acknowledge that the presence of extraintestinal manifestations and surgery are only rough measures of disease activity. We did not have access to data on any disease activity index; nor did we have access to endoscopy scores, faecal calprotectin or C reactive protein. Hence, we cannot confirm our suspicion that the 49% increased mortality seen in our study is even higher in patients with severe disease activity. Time-dependent analyses otherwise found the highest mortality in the first year of follow-up (when disease activity is often high and also observed in other studies1 3 5). The increased mortality during first year of follow-up might in part be due to detection bias (severe comorbidity resulting in work-up and diagnosis of IBD) but was mainly driven by IBD itself and by cardiovascular mortality (which is more likely during active inflammation), malignancy (likely because of a degree of misclassification) and all other causes of death (it is more likely to die from any cause when you are ill and in need of medical care than when you are not). More than half of our patients had a follow-up ≥10 years. A longer follow-up is important since studies with a short follow-up tend to present overinflated risk estimates for death because of the dominance of the first year of follow-up. In our study, the HRs were consistently approaching 1.3 after the first year of follow-up. The absence of high relative risks (upper 95% CI never exceeded 1.42 after the first year of follow-up) should be comforting for patients.

Our cohort included more than 17 000 patients with elderly-onset IBD. The study also comprised more than 40 000 patients diagnosed since 2002 (ie, after biologics was approved for IBD treatment in Sweden). The introduction of biologics does not seem to have influenced the risk estimates more than marginally (HR for death in incident IBD was 1.55 in 2002–2014). We only had data on prescription since 2005 and incomplete coverage (estimated at 77%) of biologics. Hence, we had limited power to calculate mortality rates and HRs based on treatment with biologics, but also for other drug exposures. Thus, we urge caution when interpreting those results. Notably, however, we found that patients without a record of either immunomodulators or biologics were at a 79% increased risk of death. Importantly, a recent study from the United Kingdom found that thiopurine use (>1 year) in elderly-onset UC was at a lower risk of colectomy,32 but it was beyond this study to examine long-term thiopurine use and mortality.


Patients with IBD remain at an almost 50% increased risk of death despite modern treatment. Given the rise in IBD prevalence globally33 (in Sweden, 1 in 160 individuals have IBD),34 we need to optimise treatment but potentially also screening for cancer and preventive measures against cardiovascular disease to facilitate an efficient and cost-effective care of IBD.


This study found that even though relative mortality has decreased during the past 50 years, patients diagnosed with IBD since 2002 were still at an increased risk of death with similar relative risks in adult-onset and elderly-onset IBD. All subtypes of IBD (UC, CD and IBD-U) were associated with an increased risk of death. Some of the most common causes of death in the age group under study (cardiovascular diseases, malignancy, respiratory diseases and digestive diseases) were all associated with an increased risk of death in IBD. The excess mortality remained after more than 5 years of follow-up.


View Abstract


  • Contributors Guarantor: OO had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: OO, JFL, JA, MCS. Acquisition of data: OO. Analysis: MCS, OO, JFL. Interpretation of data: OO, JA, MCS, MN, KES, AE, JFL. Drafting of the manuscript: JFL, OO, MCS. Critical revision of the manuscript for important intellectual content and approval of final version: OO, JA, MCS, MN, KES, AE, JFL.

  • Funding OO was supported by grants from the Swedish Medical Society (Project grants SLS-789611)), Swedish Foundation for Strategic Research, and Regional Agreement on Medical Training and Clinical Research between Stockholm County Council and Karolinska Institutet (ALF, 20170720, 20170424). JA was supported by the Swedish Cancer Society, the Stockholm County Council (ALF), the Swedish Research Council and the Swedish Foundation for Strategic Research during the conduct of the study. JFL was supported by the FORTE Foundation (83278) and the Swedish Cancer Foundation. None of the funding organisations had any role in the design and conduct of the study; in the collection, management, and analysis of the data; or in the preparation, review and approval of the manuscript.

  • Competing interests OO has been principal investigator (PI) on projects at Karolinska Institutet partly financed by investigator-initiated grants from Janssen, Ferring, Takeda and Pfizer. None of those studies have any relation to the present study. Karolinska Institutet has received fees for lectures and participation on advisory boards from Janssen, Ferring, Takeda and Pfizer regarding topics not related to the present study. JA reports grants from AbbVie, Bristol-Myers Squibb, Lilly, Merck, Pfizer, Roche, Samsung Bioepis and UCB, mainly in the context of a national safety monitoring programme for immunomodulators in rheumatology (ARTIS). JFL coordinates a study unrelated to the present study on behalf of the Swedish IBD Quality Register (SWIBREG). That study has received funding from Janssen. KES has been PI for research projects partly financed by unrestricted grants from Janssen pharmaceutical to Karolinska Institutet. MN has been PI for research projects unrelated to the current paper and partly financed by investigator-initiated grants from Pfizer and Astra Zeneca to Karolinska Institutet, and has received fees for participation in advisory boards (related to rheumatology). Other authors disclosed no conflicts of interest.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional data available due to Swedish regulation.

  • Author note Several of the researchers care for patients with IBD in their clinical work. The questions addressed in this paper are based on what patients usually want to know. However, in this register-based study no patients were directly involved in the design, recruitment or conduct of the study.

  • Patient consent for publication Not required.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.