Coeliac disease is increasingly easily diagnosed now that tissue transglutaminase immunoglobulin A (IgA ) antibody assays are widely available and their use understood. As many patients have relatively mild or non-specific symptoms, and the population prevalence is about 1%, it is important to have a clear understanding of the frequency of the complications of the condition. This allows the impact of the diagnosis to be explained, and appropriate therapies in addition to a gluten-free diet to be given.

The long-term risks in coeliacs have been clarified recently with large population studies, using the UK GP research database and the Swedish national inpatient register. The relative and absolute risks are now much clearer for the complications that have been recognised for a long time, such as malignancy, lymphoma and fracture.1^–4 Other risks such as liver disease and pancreatitis are now apparent from large population studies. In this issue, Ludvigsson and colleagues5 report on the risk of sepsis in >15 000 coeliacs included in the Swedish inpatient register between 1964 and 2003 (page 1074).

When compared with an inpatient reference population, a hazard ratio of 1.6 (95% CI 1.2 to 1.9, p<0.001) was found for any diagnosis of sepsis in coeliacs. Pneumococcal sepsis had a higher risk, with a hazard ratio of 2.5 (95% CI 1.2 to 5.1, p<0.014). Comparisons with a general population gave similar results, with a hazard ratio of 2.6 for overall sepsis and 3.9 for pneumococcal sepsis. The risk was increased when the coeliac disease was diagnosed in adults using both comparison groups, but only when compared with the general population if diagnosed in childhood. Most subjects entered the study as children, but the median age for the sepsis was 62 years. Adjustment for the association with diabetes decreased the estimates of risk, but these still remained significant. The risk of sepsis from staphylococcal causes was also significantly increased, but not those from Gram-negative bacteria, other streptococci or meningococci.

Some idea of the absolute risk of sepsis in coeliacs can be obtained (see table 1) and this can be more informative than relative risk. About 200 out of 15 325 patients had sepsis over the 40-year course of the study, whereas only 44 out of 11 019 patients developed lymphoma over a shorter 30-year period.1 In comparison, over 40 years, out of about 15 000 subjects, about 400 had hip fractures and >1000 had fractures at any site.4 After 30 years of follow-up, the cumulative risk of sepsis in coeliacs was about 0.03 (3%) whereas that of hip fracture was about 0.18 (18%) and of any fracture about 0.36 (36%).

The risk of tuberculosis (TB) has also been found to be increased in the Swedish inpatient population of coeliacs.6 The hazard ratio for TB was 3.7, with 24 cases in the coeliacs. When TB was the first diagnosis, the risk of a subsequent diagnosis of coeliac disease was also significantly increased (odds ratio 2.5, p<0.001). The association of TB and coeliac disease may be explained by the effects of vitamin D deficiency, common in coeliacs, and known to reduce bactericidal activity against mycobacteria through impaired cathelicidin production.

Septicaemia due to Pneumococcus has been previously described in coeliacs in case reports, but this is the first systematic study. What can account for this association between pneumococcal sepsis and coeliac disease?

Hyposplenism is known to occur in coeliacs, and was the subject of a study performed by Di Sabatino and colleagues in 2006.7 Splenic hypofunction was assessed by looking for red cell abnormalities (pitted red cells) on blood film and counting IgM memory B cells using cell sorting. The highest prevalence of hyposplenism (80%) was found in complicated coeliacs (lymphoma, refractory sprue or ulcerative jejuno-ileitis), and in those with other autoimmune disorders (type 1 diabetes, thyroid disease or Addison disease) the prevalence was 59%. In uncomplicated coeliacs, hyposplenism was found in 19%. The degree of hyposplenism was not related to the duration of treatment with a gluten-free diet. Fewer IgM-producing memory B lymphocytes were found in the hyposplenic patients, and it is these cells that are key in the defence against encapsulated bacteria such as Pneumococcus.8 In splenectomised patients, overwhelming infection occurs at an estimated incidence of 0.23–0.42% per year, with a lifetime risk of 5%, with pneumococci being the single most frequent cause. The risk from other microbial causes including protozoa, while less easy to define, is also real.9

Increased intestinal permeability could possibly also have a role, as there is evidence that enteropathy is associated with reduced barrier function which leads to bacterial translocation and endotoxaemia.

Given the rather modest increase in absolute risk, other infrequent associations of coeliac disease and diseases masquerading as coeliac disease also have to be considered. Intestinal lesions of common variable immunodeficiency may mimic coeliac disease, and this is a condition also characterised by loss of IgM-producing memory B lymphocytes. Invariant NKT lymphocytes can sustain specific B lymphocyte responses and memory, and these cells are reduced in number in refractory coeliac disease. Additionally, type I diabetes mellitus associated with coeliac disease may also increase susceptibility to infections.

Immunisation against encapsulated bacteria is recommended in patients who are splenectomised or have impaired splenic function. This includes meningococci and Haemophilus influenzae type B, but is particularly relevant for pneumococci. Annual influenza vaccination is also appropriate as this reduces the frequency of secondary bacterial infection. Coeliacs have been included in the guidance for pneumococcal vaccination, yet it is not clear what proportion have actually received this. The polysaccharide capsulate pneumococcal vaccine elicits appropriate antibody responses in coeliac patients.10 However, protection against capsulate bacteria is dependent on the generation of protective responses by IgM memory B cells, which are impaired in hyposplenism, and lower antibody titres are achieved following meningococcal vaccination in splenectomised patients compared with controls.11 A conjugate pneumococcal vaccine may be more effective, as it depends on the presence of T rather than B lymphocytes.12

The current paper5 confirms that pneumococcal infection is a real risk in coeliacs, and it is likely that other causes of sepsis are also increased. This adds to the increasing literature quantifying the extent and magnitude of the hazards of being diagnosed with coeliac disease. In absolute terms, sepsis appears at least twice as likely to develop as lymphoma, suggesting prevention of sepsis should be a higher priority than it is at present. Those who treat coeliacs need to ensure that their patients receive the appropriate vaccinations.