- Top of page
- Conflict of interest statement
- Author contributions
Coeliac disease (CD) is an immune-mediated enteropathy that affects about 1% of the population of Western countries. This complex chronic inflammatory disorder is associated with several autoimmune disorders, including type 1 diabetes (T1D) . The prevalence of CD in individuals with T1D varies between 3% and 12% [2, 3].
Although treatment and management of T1D has improved over the years, patients with this disease are still at increased risk of death [4-7]. Two of the main causes of death in these patients are vascular disease and cancer . An increased mortality risk in patients with CD has been reported by us  and others , in whom cardiovascular disease and cancer were the most frequent causes of death.
Concomitant CD may have a negative effect on some complications of diabetes in patients with T1D, including diabetic retinopathy [11, 12]. However, it is still unknown whether CD influences the mortality rate in patients with T1D. The aim of this population-based study was to examine the risk of death in 960 patients with both T1D and biopsy-proved CD compared with 4608 matched individuals with T1D alone (reference group). We hypothesized that CD confers an increased risk of death in patients with T1D.
- Top of page
- Conflict of interest statement
- Author contributions
To our knowledge, this is the first population-based cohort study to examine both relative and absolute risks of death in patients with T1D and CD. The findings of this study suggest that long-term CD substantially increases mortality risk in patients with T1D.
This study has several strengths, including its population-based design and the independent ascertainment of cases from national health registers. The Swedish National Patient Register has been validated repeatedly, and the majority of diagnoses have a high positive predictive value (85–95%) . Sensitivity for T1D is likely to approach 100% because hospital admission is mandatory at onset of T1D; in addition, a negative predictive value of diabetes of 100% was observed in patients admitted to hospital (n = 872/872) in a Swedish validation study . Data on pregnancy and antidiabetic medication enabled us to perform important sensitivity analyses (Table 5) to increase the specificity of T1D and to reduce the risk of misclassification.
Another strength is the ability to accurately define CD. Using nationwide data from small intestinal biopsies showing VA reduces the risk of selection bias because 96% of gastroenterologists and 100% of paediatricians in Sweden include a small intestinal biopsy in their diagnostic work-up for CD. In addition, VA is rarely explained by diagnoses other than CD (0.3% of individuals with VA had inflammatory bowel disease) . Although positive CD serology was not included within the definition of CD, it has been demonstrated that 95% of individuals with VA have CD, and that 88% of those with available CD serology data have positive antibodies at the time of first biopsy .
Furthermore, the large number of study participants led to a high statistical power and enabled the calculation of important subanalyses, including stratified analyses according to sex, age and calendar period of CD diagnosis. The increased risk of death in T1D was found in both male and female patients with a long duration of CD, and there was no interaction between sex and CD with regards to T1D mortality. However, a short follow-up precluded the estimation of aHRs in the last calendar period. Nevertheless, the absolute risks decreased in the last calendar period (Table 3). This finding is to be expected, given that the majority of study participants were children and those who entered the study after 2000 did not reach adulthood (when mortality was highest) before the end of the follow-up period. Moreover, the median age of death in our study participants might seem low, but this is also to be expected as most of the participants were children at study entry.
A final study strength is that data from the Swedish Cause of Death Register allowed for the assessment of cause-specific mortality.
Some limitations of this study should also be considered. It has been reported that the prevalence of CD in patients with T1D is between 3% and 12% [1, 2, 25, 26]. At present, Swedish patients with T1D are screened for CD, but because such screening did not occur in the early part of our study period, we cannot rule out the possibility that some individuals classified as having T1D alone may also have undiagnosed CD. However, because individuals with T1D and undiagnosed CD would not make up more than 1–2% of our reference group , such misclassification would not affect our risk estimates more than marginally, and in any case would bias the aHR values towards 1. Another limitation of this study is the lack of data on smoking habits, haemoglobin A1c, body mass index, insulin therapy and gluten-free diet (GFD). However, smoking and obesity [28, 29] are associated with higher cardiovascular disease risk in T1D, whereas CD is inversely associated with both obesity  and smoking . Thus, these factors are unlikely to explain our findings.
Individuals with T1D have an almost sevenfold higher risk of death compared with the general population . During the first 10 years after T1D diagnosis, the major cause of death is acute diabetic complications (ketoacidosis and hypo/hyperglycaemia), whereas 10–20 years after T1D diagnosis, death is primarily due to chronic complications (e.g. cardiovascular disease, renal disease and infections) . The findings of previous studies indicate that CD has an adverse effect on complications of T1D, including an increased risk of diabetic retinopathy [11, 12]. CD is associated with an increased risk of death (HR1.39) . The most common cause of death in a large Swedish cohort was cardiovascular disease followed by cancer . The underlying reason for the excess mortality seen in patients with CD has not yet been fully determined but may include malnutrition and chronic inflammation.
The lower risk of death observed during the first 5 years after CD diagnosis may be explained by a better adherence to a GFD or surveillance bias (frequent checkups soon after the diagnosis of CD). Another reason for the lower risk of death might be that patients with newly diagnosed CD have lower cholesterol levels compared with the general population .
Despite rapid clinical improvement in most patients with CD after starting a GFD, findings indicate that there is persistent low-grade inflammatory activity in the gut years after initiation of the diet [33, 34]. The explanation for the excess mortality demonstrated in this study in subjects with both T1D and CD could be persistent low-grade inflammation due to disease activity per se or poor adherence to a GFD. Adherence to a GFD is generally regarded as high in patients with CD and perhaps more so in individuals with a CD diagnosis early in life . Although we have no data on the use of a GFD at the individual level in this study, adherence to such a diet was observed in 83% of a subset of our biopsy-verified patients with CD . However, adherence to both a GFD and insulin therapy may differ . If patients with T1D and CD find it difficult to follow a GFD, this lack of adherence may explain the excess mortality seen over time in those with both T1D and CD.
It has been shown that there is an increased risk of heart disease in CD , and low-grade inflammation has been suggested to trigger atherosclerosis . An association between inflammation and atherosclerosis has further been supported by the finding in an Italian study of greater carotid intima–media thickness in patients with T1D and CD than in those with T1D alone, suggesting a higher degree of atherosclerosis progression in patients with both conditions .
We can only speculate about nonvascular and nonmalignant causes of death in T1D. Although there were more than 40 deaths in patients with T1D and CD, we did not have the statistical power to examine excess mortality for specific disorders. For example, CD is associated with a higher risk of end-stage renal disease (HR 2.87) , whereas T1D is the most common cause of this condition .
Recent findings from studies of the role of CD in metabolic control in patients with T1D are conflicting, but it seems that patients with both T1D and CD have lower height and weight gain compared with those with T1D alone [42, 43]. Poor nutritional status in patients with T1D and CD may also have contributed to the excess mortality observed in this study. However, the lack of data on metabolic control and the degree of adherence to a GFD are limitations of this study that should be addressed.
In conclusion, we have demonstrated an increased risk of death in patients with T1D and CD compared with patients with T1D alone. Based on these findings, we recommend close monitoring over time of patients with this combination of conditions (i.e. T1D and CD).