Nephrol Dial Transplant (2006) 21: 1809–1815 doi:10.1093/ndt/gfl117 Advance Access publication 30 March 2006 Original Article Coeliac disease and risk of renal disease—a general population cohort study Jonas F. Ludvigsson1,3, Scott M. Montgomery2,3, Ola Olén3,4, Anders Ekbom3,5, Johnny Ludvigsson6 and Michael Fored3 1 Department of Paediatrics, 2Clinical Research Centre, Örebro University Hospital, Clinical Epidemiology Unit, Department of Medicine, Karolinska University Hospital/Institute and 4 Department of Paediatrics, Stockholm Söder Hospital, Sweden, 5Harvard Medical School, Boston, Massachusetts, USA and 6Division of Paediatrics and Diabetes Mellitus Research Centre, Department of Molecular and Clinical Medicine, Faculty of Health Sciences, Linköping University, Sweden 3 Keywords: auto-immune; coeliac; cohort study; diabetes mellitus; kidney; renal disease Correspondence and offprint requests to: Jonas F Ludvigsson, Department of Paediatrics, Örebro University Hospital, Sweden. Email: jonasludvigsson@yahoo.com Introduction Coeliac disease (CD) affects up to 1% of the population in the Western world [1–3]. CD mostly occurs in individuals with human leucocyte antigen (HLA) DR3, DQ2 and results in a characteristic T-cell-mediated inflammation in the small bowel [4]. It is associated with a large number of autoimmune diseases [5]; among them type 1 diabetes mellitus (DM) [6]. A gluten-free diet in patients with both CD and type 1 DM is not only associated with an improvement in gastrointestinal symptoms but sometimes also with lower levels of haemoglobin (Hb)A1c [7]. The incidence of end-stage renal disease with a need for renal replacement therapy such as dialysis or kidney transplantation (KT) is increasing internationally [8]. The underlying cause of this increase is largely unknown [8]. In 2002, Collin et al. [9] showed an increased prevalence of CD in patients with IgA nephropathy. Patients with primary glomerulonephritis (GN) often have an activated mucosal immune system [10] with increased numbers of intra-epithelial T-cells in the mucosa [11] and increased gut permeability [11]. Several studies have demonstrated increased levels of CD auto-antibodies in patients with renal disease [12,13]; and certain renal disease will improve on a low-antigenic diet lacking in gluten [14]. Little is known, however, about the risk of severe renal disease such as renal failure in individuals with CD. In this study, we used data from a general population-based register, the Swedish Hospital Discharge Registry, to study the risk of renal disease in individuals with CD. We examined associations of CD with GN, chronic glomerulonephritis (CGN) and renal failure assessed by the occurrence of renal dialysis or KT. ß The Author [2006]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 Abstract Background. Coeliac disease (CD) may be a risk factor for renal disease. Methods. We investigated the risk of any form of glomerulonephritis (GN) (acute, chronic and nonspecified), chronic glomerulonephritis (CGN) and renal replacement therapy including dialysis treatment and kidney transplantation (KT) in patients with CD in a general population-based cohort study. We used Cox regression to assess the risk of renal disease in 14 336 patients who had received a diagnosis of CD (1964–2003) and 69 875 reference individuals matched for age, calendar year, sex and county. Patients were identified using the Swedish Hospital Discharge Registry. Follow-up began 1 year after study entry. Results. CD was associated with an increased risk of any form of GN (hazard ratio (HR) ¼ 1.64; 95% confidence intervals (CI) ¼ 1.01–2.66; P ¼ 0.046; 89 events), CGN (HR ¼ 2.65; 95% CI ¼ 1.34–5.24; P ¼ 0.005; 39 events), dialysis (HR ¼ 3.48; 95% CI ¼ 2.26–5.37; P<0.001; 102 positive events) and KT (HR ¼ 3.15; 95% CI ¼ 1.29–7.71; P ¼ 0.012; 22 events). Conclusion. We suggest that immune characteristics associated with CD increase the risk of chronic renal disease. Individuals with CD may also be at a moderately increased risk of any form of GN. 1810 Materials and methods Definition of CD, renal disease and DM Every person who had been hospitalized from 1964–2003 under any of the following International Classification of Disease (ICD) codes was defined as having CD: ICD-7: 286.00; ICD-8: 269.00, 269.98; ICD-9: 579A; ICD-10: K90.0. We defined any form of GN, which in Sweden includes IgA nephropathy, as follows (ICD codes signifying chronic GN have been italicized): ICD-7: 590, 592, 593.00; ICD-8: 580, 582; ICD-9: 580, 582; ICD-10: N00, N01, N03, N05. Dialysis was defined according to the following surgical codes: 9200, 9206, 9207, 9211, 9212, 9213, 9214, 9223 or any of the following ICD codes: ICD-9: V45B, V56; ICD-10: Z49.1, Z49.2, Z99.2. Temporary dialysis (e.g. during cardiopulmonary surgery) was not used in the definition of dialysis in this study. We defined KT according to the following surgical codes: 6070, KAS00, KAS10 and KAS20. Finally, we defined DM as follows: ICD-7: 260; ICD-8: 250; ICD-9: 250; ICD-10: E10-E14. The HDR does not allow any distinction between type 1 or type 2 DM. Inclusion criteria The Swedish National Board of Health and Welfare identified 15 533 individuals with CD diagnosed before the end of follow-up. We excluded individuals with a shorter follow-up than 1 year in order to minimize the risk of detection bias. We also excluded individuals with any of the four renal outcome measures occurring before study entry or less than 1 year after study entry (n ¼ 1103). Hence, both the CD-patient cohort and the reference cohort consisted of individuals without renal disease at the start of follow-up. Another 94 individuals with CD were excluded due to data irregularities, such as date of death preceding that of the first recorded CD diagnosis. The analyses of the current study were therefore based on 14 336 individuals with CD and 69 875 reference individuals who had never had a diagnosis of CD. Their characteristics are given in Table 1. It is important to note, that an individual free of renal disease at, for example, 15 years of age and entering the study at that time may later have had diagnoses of all four renal disorders studied in this article, e.g. GN (aged 25 years with a follow-up of 10 years), CGN (aged 27 years; follow-up P ¼ 12 years), dialysis (aged 35; follow-up P ¼ 20 years) and KT (aged 42 years; follow-up P ¼ 27 years). Socio-economic status In a subset of individuals (n ¼ 45 367), we were able to obtain data on socio-economic status (SES) from Statistics Sweden (Table 1). The SES was based on a three-category occupational classification from 1968 [17]. Among individuals with data on SES, some 6500 children born after 1990 were assigned a socio-economic code on the basis of the occupation of the mother. Statistical methods and analyses We used Cox regression to calculate hazard ratios (HRs) for renal disease in individuals with CD. Follow-up time began 1 year after study entry and ended on the date of first discharge diagnosis of renal disease, date of emigration, death or the end of the study period (31 December 2003), whichever happened first. The Cox Table 1 Characteristics of participants (n (%))a Characteristics No CD (%) 69 875 Totala Age at first recorded diagnosis of coeliac disease 0–15 years –
16 years – Any form of glomerulonephritis 66 (0.1) Chronic glomerulonephritis 25 (<0.1) Renal dialysis 64 (0.1) Kidney transplantation 13 (<0.1) Sex Males 28 691 (41.1) Females 41 184 (58.9) No diabetes mellitus 67 875 (97.1) Diabetes mellitus 2000 (2.9) Socio-economic statusb I 7250 (10.4) II 9244 (13.2) III 20 081 (28.7) Missing data 33 300 (47.6) Calendar period 1964–73 2426 (3.5) 1974–83 19 205 (27.5) 1984–93 30 731 (44.0) 1994–2003 17 513 (25.1) CD (%) 14 336 9364 4972 23 14 38 9 (65.3) (34.7) (0.2) (0.1) (0.3) (0.1) 5906 8430 13 392 944 (41.2) (58.8) (93.4) (6.6) 1501 2149 5142 5544 (10.5) (15.0) (35.9) (38.7) 499 3928 6283 3626 (3.5) (27.4) (43.8) (25.3) a Participants with a follow-up >1 year. None of the participants had glomerulonephritis, renal dialysis or KT at the start of follow-up. See also text. b ‘I’ denotes the highest social class. See also text. For reference individuals we have given the number of individuals who constituted the basis for the Cox’s regression. We actually had data on socio-economic status in another 6519 reference individuals but these individuals were not part of the internally stratified calculations due to missing values on socio-economic status in the matched individual with CD. Adding the 6519 reference individuals to those presented above, the proportion of missing values was similar among individuals with CD and among those without CD. Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 Individuals with a hospital-based discharge diagnosis of CD between 1964–2003 were identified through the Swedish Hospital Discharge Registry (HDR). The HDR was initially established in some regions of Sweden in 1964, and since 1987 it has covered all of Sweden. It contains individual data and is maintained by the Swedish National Board of Health and Welfare. The HDR was also used to identify individuals with GN, CGN, renal dialysis treatment, KT or type 1 or type 2 DM in study participants. For each individual in the CD-patient cohort, Statistics Sweden (the government agency for population statistics), identified up to five reference individuals (matched for age, calendar year, sex and county) through the Swedish total population registry. Where more reference individuals were available, five were chosen at random. The total population registry [15] includes information on area of residence, vital status and dates of immigration or emigration. All the individuals in the HDR and the population registry are identified through their personal identity number. The personal identity number is a unique number assigned to over 99.9% of all Swedish residents at birth or immigration [16]. J. F. Ludvigsson et al. Coeliac disease and renal disease 1811 Ethics This study was approved by the Research Ethics Committee of the Karolinska Institute. None of the participants were contacted. Patient information was anonymized prior to the analyses. Results The median age at study entry (corresponds to date of CD diagnosis in individuals with CD) was 3 years (range: 0–94). The median age at first recorded diagnosis of GN in individuals with CD was 28 years (range: 4–86). Corresponding age at diagnosis of CGN was 47.5 years (4–86); dialysis ¼ 61.5 (28–87) and KT ¼ 34 (25–58). The median duration between diagnosis of CD and first recorded diagnosis of GN was 6 years (range: 1–22)(CGN ¼ 4 (1–22); dialysis ¼ 6.5 (1–27) and KT ¼ 12 (4–25)). Any form of glomerulonephritis A diagnosis of GN was more often recorded among patients with CD than among reference individuals and this association was statistically significant (Table 2). This risk increase was only seen in CD diagnosed in adulthood (HR ¼ 2.62; 95% CI ¼ 1.24–5.52; P ¼ 0.011), with the risk estimate in children not attaining statistical significance (HR ¼ 1.20; 95% CI ¼ 0.62–2.32; P ¼ 0.585). There was no notable difference between the risk estimates in males and females (data not shown). In the subset of individuals with data on SES, the crude HR for the association of CD with GN was 1.74 (95% CI ¼ 0.91–3.32; P ¼ 0.095) and the adjusted HR was 1.62 (95% CI ¼ 0.84–3.12; P ¼ 0.151). The risk estimate for any form of GN did not change when individuals with a diagnosis of DM were excluded (Table 2) or when our outcome measure was restricted to at least two discharge diagnoses of GN (HR ¼ 1.81; 95% CI ¼ 0.80–4.09; P ¼ 0.154) (based on nine positive events in 14 336 individuals with CD and 22 positive events in 69 875 reference individuals). The risk estimate for having received a diagnosis of GN in departments of paediatrics, internal medicine or renal medicine was 1.53 (95% CI ¼ 0.85–2.73; P ¼ 0.155). The risk estimate for GN was similar when we included the first year after study entry in the follow-up time (HR ¼ 1.69; 95% CI ¼ 1.07–2.66; P ¼ 0.025). When we included the ICD-10 code N02 (recurrent or persistent haematuria) among outcome measures the HR did not attain statistical significance [HR of 1.48 (95% CI ¼ 0.89–2.55; P ¼ 0.130)]. Table 2 CD and risk of later renal disease Type of renal disease HRa, 95% CI P-value AHRb, 95% CI P-value HRc, 95% CI P-value No CD Any glomerulonephritis Chronic glomerulonephritis Renal dialysis Kidney transplantation 1.00 1.64; 2.65; 3.48; 3.15; 0.046 0.005 <0.001 0.012 1.00 1.59; 2.65; 2.72; 1.83; 0.063 0.005 <0.001 0.290 1.00 1.70; 2.86; 2.24; 1.71; 0.042 0.005 0.008 0.430 a 1.01–2.66 1.34–5.24 2.26–5.37 1.29–7.71 HR, hazard ratio. Estimates derived from Cox’s regression. Adjusted HR for the presence of DM (type 1 or type 2). c Individuals with a diagnosis of DM excluded from the analyses. b 0.98–2.60 1.34–5.26 1.69–4.36 0.60–5.60 1.02–2.85 1.38–5.91 1.24–4.06 0.45–6.47 Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 model was internally stratified. This means that the analysis was divided by risk-set and then a summarized risk estimate was produced. Each risk-set consisted of one individual with CD and his/her age- and sex-matched reference individuals. All individuals in one risk-set originated from the same county and entered the study in the same calendar year. In sub-analyses, we stratified for sex and age at study entry (15 vs
16 years). In order to test the statistical significance of seemingly different HRs for renal dialysis in males and females, we carried out an additional regression analysis with regards to renal dialysis in which we included CD and sex but also a multiplicative term (interaction term) consisting of sex and CD. In separate analyses, we adjusted for DM and excluded individuals with a diagnosis of DM. We tested one model in which DM was dichotomized (no DM vs DM) and a second model in which individuals with DM diagnosed at an age of 30 years were assumed to have predominantly type 1 DM and those with a later diagnosis were assumed to have a higher proportion of type 2 DM. Both models yielded the same risk estimates for the association of renal disease with CD. In this article, we have consistently presented the data from the statistical model where DM was coded as a dichotomous variable, not taking age into consideration at the first recorded diagnosis of DM. In order to increase the specificity of GN, CGN and dialysis treatment, we calculated the risk of having at least two hospital discharge diagnoses of these outcome measures (e.g. a diagnosis of CGN during hospitalization both in 1987 and in 1989). The occurrence of repeated diagnoses minimizes the risk of findings being based on misclassifications of outcome measures. We also calculated the risk of having received a hospital discharge diagnosis of GN, CGN or dialysis in a department of renal disease, internal medicine or paediatrics in order to increase the specificity of renal disease. Misclassification is less likely in departments where a majority of individuals with renal disorders are cared for. We did not carry out the corresponding analyses for KT as this procedure only takes place in specialized settings. We also tested the relationship between CD and renal disease including events in the first year after study entry. In a separate analysis, we included the ICD-10 code N02 (recurrent or persistent haematuria) when calculating the risk of any GNF. Earlier ICD versions do not distinguish between recurrent/persistent haematuria and other haematuria. The 95% confidence intervals (CI) for HRs not including 1.00 were considered statistically significant. Statistics were calculated using SPSS 11.0 (2002. Chicago, Illinois). 1812 Chronic glomerulonephritis Table 3 Medical diagnoses of individuals with CD and renal replacement therapy Diagnosis Renal dialysis (n ¼ 38) KT (n ¼ 9) Diabetes mellitusa Renal failure, acute Renal failure, chronic Renal failure, non-specified Systemic lupus erythematosus Pyelonephritis Acute glomerulonephritis Chronic glomerulonephritis Renal atherosclerosis Cancer (small bowel, colon, Hodgkin’s) Hypertensive disease Toxic liver disease with chronic hepatitis Severe body trauma 20 2 5 1 1 2 2 2 1 4 1 3 1 1 1 1 1 1 One individual with both chronic glomerulonephritis and diabetes mellitus has been listed under both headings (renal dialysis). One individual with both systemic lupus erythematosus and chronic glomerulonephritis has been listed under both headings (renal dialysis and KT). Several of the patients also had a diagnosis of septicaemia. a Includes both type 1 and type 2 diabetes mellitus. All patients with diabetes mellitus were assigned diabetes mellitus as the underlying diagnosis. Dialysis CD was associated with a statistically significant 3-fold increased risk of dialysis (Table 2). Due to lack of positive events in individuals with CD diagnosed in childhood we were not able to calculate a meaningful risk estimate in this age group (HR ¼ 0.00; 95% CI ¼ 0 to above 100). In individuals with CD diagnosed in adulthood, the risk estimate was close to four (HR ¼ 3.71; 95% CI ¼ 2.40–5.76; P<0.001). The HR for subsequent dialysis in females with CD was 5.40 compared with 2.43 in males. A formal interaction test found that this difference (between the sexes) was not statistically significant (P ¼ 0.156). In a subset of individuals with socio-economic data, adjustment for SES had only a marginal effect on the risk estimate (crude HR ¼ 4.39; 95% CI ¼ 2.48–7.77; P<0.001) (adjusted HR ¼ 4.19; 95% CI ¼ 2.35–7.48; P<0.001). The CD remained statistically, significantly associated with dialysis after adjustment for DM or exclusion of individuals who had ever had a diagnosis of DM (Table 2). There was also a risk increase also when we restricted our outcome measure to at least two discharge diagnoses of dialysis (HR ¼ 3.97; 95% CI ¼ 2.26–6.96; P<0.001) (24 and 34 positive events), or to a diagnosis of dialysis in the departments of paediatrics, internal medicine or renal medicine (HR ¼ 3.42; 95% CI ¼ 2.18–5.35; P<0.001). Including the first year after study entry, the HR for dialysis was 3.56 (95% CI ¼ 2.41–5.27; P<0.001). Kidney transplantation KT was more common in individuals with CD than among reference individuals (Table 2). The risk increase for KT was restricted to individuals diagnosed with CD in adulthood (HR ¼ 5.45; 95% CI ¼ 1.83–16.24; P ¼ 0.002) (CD diagnosed in childhood: HR ¼ 0.80; 95% CI ¼ 0.10–6.70; P ¼ 0.841). There was no notable difference between the risk estimates in males and females (data not shown). Adjustment for SES in a subset of individuals with data on SES did not affect our risk estimates (crude HR ¼ 7.73; 95% CI ¼ 2.26–26.50; P ¼ 0.001) (adjusted HR ¼ 8.67; 95% CI ¼ 2.77–27.10; P<0.001). Excluding individuals with DM before the end of follow-up, the HR for KT was 1.71 and failed to reach statistical significance (Table 2). Including the first year after study entry in the follow-up time, the risk of KT increased slightly (HR ¼ 3.84; 95% CI ¼ 1.79–8.20; P ¼ 0.001). The underlying diagnoses of patients with dialysis or KT are given in Table 3. Discussion The current study showed an association between CD and GN, CGN and dialysis. The CD was also associated with KT, although, unlike for the other diseases, the risk increase was not statistically significant when a diagnosis of DM was taken into account. Most previous reports on renal disease and CD have concerned nephritis and IgA nephropathy [9, 12, 13, 18–20] although other renal or urinary diseases [21–25] may also be increased in individuals with CD. Collin et al. [9] reported a CD prevalence of 3–4% among patients with IgA nephropathy; Fornasieri et al. [18] found evidence of CD in 2/121 patients with IgA nephropathy. In contrast, the purpose of our cohort study was to look at the risk of renal disease in individuals with CD and we cannot therefore estimate the risk of CD in patients with renal disease. Our study results are in line with those of Peters et al. [24]. Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 CD was associated with a more than 2-fold increased risk of CGN; this association was statistically significant (Table 2). This risk increase was most pronounced in individuals with CD diagnosed in adulthood (HR ¼ 3.96; 95% CI ¼ 1.52–10.30; P ¼ 0.005) than those with CD diagnosed in childhood (HR ¼ 1.76; 95% CI ¼ 0.64–4.90; P ¼ 0.276). There was no notable difference between the risk estimates in males and females (data not shown). In a subset of individuals with socio-economic data, adjustment for SES did not affect the risk estimate (crude HR ¼ 2.93; 95% CI ¼ 1.04–8.30; P ¼ 0.043) (adjusted HR ¼ 2.77; 95% CI ¼ 0.93–8.23; P ¼ 0.067); the reduction in statistical significance compared with the main analysis is due in part to the smaller number of individuals with SES data available for analysis. The risk estimates for CGN remained statistically significant also when we excluded individuals with DM (Table 2); restricted our outcome measure to at least two discharge diagnoses (HR ¼ 2.74; 95% CI ¼ 1.08–6.96; P ¼ 0.035) or restricted our outcome measure to patients diagnosed in departments of paediatrics, internal medicine or renal medicine (HR ¼ 2.48; 95% CI ¼ 1.15–5.34; P ¼ 0.021). Inclusion of the first year after study entry, did not affect the risk estimate for CGN (HR ¼ 2.71; 95% CI ¼ 1.46–5.02; P ¼ 0.001). J. F. Ludvigsson et al. Coeliac disease and renal disease 3 years, would only have been 24-years old at the end of follow-up. None of the previous studies on CD and renal disease [9,12,13,18–20,24,35] were adjusted for SES. In our study, we had data on SES in a subset of individuals. Although adjustment for SES did not affect our risk estimates, the smaller number of participants diminishes our study power. Adjustment for SES is nevertheless important since SES is associated with use of health services [36] and lower status is linked with an increased risk of some renal diagnoses [37]. It is also closely linked to smoking pattern in Sweden [38] and as the HDR contains no information on smoking, adjustment for SES provides some adjustment for such behavioural factors. Smoking may be negatively associated with CD [39,40], but seems to be a positive risk factor for chronic renal failure [41]. Neither SES nor smoking is therefore likely to explain our results. Given these associations, it is likely that inclusion of smoking could have resulted in higher rather than lower risk estimates for the association of CD with renal disease. Due to the high CD prevalence in the Nordic countries [1,42], most Swedish doctors are well aware of the necessity to confirm CD with a small-bowel biopsy [43]. The diagnostic use of small-bowel biopsy is likely to have increased the sensitivity of our study. Especially in the earlier part of our study period hospitalization (and as a consequence of this inclusion in the HDR) was often a prerequisite for small-bowel biopsy. This is important since positive serology does not always correspond with true CD [44]. In a recent study, Smedby et al. [45] validated the diagnosis of CD in the HDR and found a specificity of above 85% in patients with lymphoma. The specificity of chronic diseases in the HDR is generally regarded as high. We chose diagnoses for dialysis or KT as proxies for renal failure since the specificity for such procedures are likely to be high. Although, the proportion of individuals with CD and subsequent dialysis/KT was small, the large study size resulted in sufficient events for stratifications. We also believe that the specificity for GN and CGN is high even though we are aware that CGN in patients with long-standing type 1 DM may be diagnosed as DM nephropathy. We cannot exclude that some individuals have been misclassified. This should, however, only affect our risk estimates marginally; otherwise the risk of misclassification differs between individuals with CD and their reference individuals. We find such differential misclassification unlikely. Instead there is a risk that misclassification may attenuate true relationships between CD and renal disease. Also the sensitivity with regards to dialysis/KT ought to be high. We do not expect any patient with endstage renal disease and dialysis/KT to lack a hospital discharge diagnosis of dialysis/KT. However, there is a risk that not all patients with CD or CGN/GN are identified through a hospital-based registry. Despite this, our study had considerable power to detect Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 That study did, however, focus on mortality in individuals with CD and was based on death certificates and not incident disease. Peters et al. [24] reported increased standardized mortality rates (SMR) for urinary diseases (SMR ¼ 2.7) and nephritis (SMR ¼ 5.4) in patients with CD. In contrast with previous research [9,12,13,18–20,24] we have tried to minimize the impact of DM, potentially a confounding factor [6,26]. The CD and type 1 DM share HLA DR3, DQ2 and DR4, DQ8 [27–29]. In addition, both type 1 DM [30,31] and type 2 DM [32,33] are major risk factors for renal failure. Type 1 DM may also be linked to IgA nephropathy [34]. The HDR does not distinguish between type 1 and type 2 DM. In this study, adjustment for DM and exclusion of individuals with type 1 or type 2 DM resulted in similar HRs. We also carried out analyses assigning different proxy values for individuals aged 30 with first recorded DM diagnosis and for those receiving a DM diagnosis at
31 years of age, strongly indicating probable DM type. This approach produced identical risk estimates to the model where we adjusted for DM. This suggests that the increase in renal disease in individuals with CD is not solely mediated through DM. After excluding individuals with type 1 or type 2 DM, CD was statistically, significantly associated with a raised risk for any form of GN, CGN and dialysis, but not with KT. As the associations with glomerulonephritis are independent of DM, presence of DM may be another marker of auto-immunity that is intermediate in the casual pathway, so that adjustment or exclusion for DM may have produced conservative estimates of the true association in our analyses. Therefore, our exclusion of patients with DM may have masked an association between CD and KT as those with the more severe autoimmunity could have been excluded from the analyses of relatively few events. In contrast with other outcome measures of this study, KT constitutes an active decision by health professionals. It may be that Swedish doctors are more inclined to transplant a non-functioning kidney in individuals with type 1 or type 2 DM, due to such factors as patient age and prognosis, than in individuals with other reasons for renal failure. This could explain the lower HR for KT in CD individuals when taking DM into account. The risk of KT when excluding individuals with type 1 or type 2 DM was instead similar to that of GN, although events were fewer and the confidence intervals therefore wider. When the number of events is small the risk estimates are inherently less stable; for instance, so were both crude and adjusted HRs for KT in a restricted sample of CD individuals with available data on SES above seven, as compared with 3–4 when estimating the risk including individuals with missing data on SES. In this study, the risk increase for renal disease was seen in individuals with an adult diagnosis of CD. This is most likely due to insufficient follow-up time in individuals with a diagnosis of CD in childhood. A child with CD diagnosed in 1982 at the age of 1813 1814 Conclusion In conclusion, our study suggests that CD is associated with an increased risk of CGN and renal failure. Acknowledgements. J.F.L. was supported by grants from the Swedish Research Council and the Örebro University Hospital while writing this article. This project was supported by The Swedish Society of Medicine, the Karolinska Institute Funds, the Swedish Research Council, the Majblomman Foundation and the Swedish Coeliac society. Conflict of interest statement. None declared. References 1. Maki M, Mustalahti K, Kokkonen J et al. Prevalence of Celiac disease among children in Finland. N Engl J Med 2003; 348: 2517–2524 2. Fasano A, Berti I, Gerarduzzi T et al. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Arch Intern Med 2003; 163: 286–292 3. Dube C, Rostom A, Sy R et al. The prevalence of celiac disease in average-risk and at-risk Western European populations: a systematic review. Gastroenterology 2005; 128 [4 Suppl 1]: S57–S67 4. Green PH, Jabri B. Coeliac disease. Lancet 2003; 362: 383–391 5. Collin P, Reunala T, Pukkala E, Laippala P, Keyrilainen O, Pasternack A. Coeliac disease-associated disorders and survival. Gut 1994; 35: 1215–1218 6. Maki M, Hallstrom O, Huupponen T, Vesikari T, Visakorpi JK. Increased prevalence of coeliac disease in diabetes. Arch Dis Child 1984; 59: 739–742 7. Amin R, Murphy N, Edge J, Ahmed ML, Acerini CL, Dunger DB. A longitudinal study of the effects of a glutenfree diet on glycemic control and weight gain in subjects with type 1 diabetes and celiac disease. Diabetes Care 2002; 25: 1117–1122 8. Hsu CY, Vittinghoff E, Lin F, Shlipak MG. The incidence of end-stage renal disease is increasing faster than the prevalence of chronic renal insufficiency. Ann Intern Med 2004; 141: 95–101 9. Collin P, Syrjanen J, Partanen J, Pasternack A, Kaukinen K, Mustonen J. Celiac disease and HLA DQ in patients with IgA nephropathy. Am J Gastroenterol 2002; 97: 2572–2576 10. Rostoker G, Terzidis H, Petit-Phar M et al. Secretory IgA are elevated in both saliva and serum of patients with various types of primary glomerulonephritis. Clin Exp Immunol 1992; 90: 305–311 11. Rostoker G, Delchier JC, Chaumette MT. Increased intestinal intra-epithelial T lymphocytes in primary glomerulonephritis: a role of oral tolerance breakdown in the pathophysiology of human primary glomerulonephritides? Nephrol Dial Transplant 2001; 16: 513–517 12. Pierucci A, Fofi C, Bartoli B et al. Antiendomysial antibodies in Berger’s disease. Am J Kidney Dis 2002; 39: 1176–1182 13. Ots M, Uibo O, Metskula K, Uibo R, Salupere V. IgA-antigliadin antibodies in patients with IgA nephropathy: the secondary phenomenon? Am J Nephrol 1999; 19: 453–458 14. Ferri C, Puccini R, Longombardo G et al. Low-antigen-content diet in the treatment of patients with IgA nephropathy. Nephrol Dial Transplant 1993; 8: 1193–1198 15. Johannesson I. The Total Population Register of Statistics Sweden. New Possibilities and Better Quality, Örebro: Sweden 2005 16. Lunde AS, Lundeborg S, Lettenstrom GS, Thygesen L, Huebner J. The person-number systems of Sweden, Norway, Denmark, and Israel. Vital Health Stat 2 1980; 2: 1–59 17. Guteland G. Socioekonomisk indelning (SEI). Swedish socioeconomic classification. 1982:4. Reports on Statistical Co-ordination. Stockholm: Statistics Sweden (SCB—Statistiska centralbyrån): 1982 18. Fornasieri A, Sinico RA, Maldifassi P, Bernasconi P, Vegni M, D’Amico G. IgA-antigliadin antibodies in IgA mesangial nephropathy (Berger’s disease). Br Med J (Clin Res Ed) 1987; 295: 78–80 19. Rostoker G, Laurent J, Andre C, Cholin S, Lagrue G. High levels of IgA antigliadin antibodies in patients who have IgA mesangial glomerulonephritis but not coeliac disease. Lancet 1988; 1: 356–357 20. Sategna-Guidetti C, Ferfoglia G, Bruno M et al. Do IgA antigliadin and IgA antiendomysium antibodies show there is latent coeliac disease in primary IgA nephropathy? Gut 1992; 33: 476–478 Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 associations, as there were 89 patients with a diagnosis of GN among our 14 000 patients with CD and some 70 000 reference individuals. This study also has some potential weaknesses. We cannot exclude the possibility that there may be false-negative reference individuals with CD. These are, however, unlikely to affect our risk estimates since individuals without CD dominate the reference population. There is also a risk that individuals with CD will receive a diagnosis of renal disease due to regular checkups for their CD. Control visits and medical investigations are usually most common just after diagnosis. In this study, the risk estimates for renal diseases were independent of events in the initial year after the first recorded CD diagnosis. For that reason, we do not think that the surveillance bias has had a large impact on the association between CD and renal disease. Furthermore, we cannot rule out that individuals with CD identified through hospitalization suffer from more severe CD than the average individual with CD, although risk estimates for CGN and dialysis both remained above 2.6 and were statistically significant when we adjusted for one marker of CD severity (DM). Patients with IgA nephropathy seem to have the same prevalence of HLA-DQ2 or DQ8 as individuals without IgA nephropathy [9]. So HLA is unlikely to explain the association between CD and GN, since IgA nephropathy is the most common form of CGN in the Western world. Instead, we speculate that many patients with an initial CD diagnosis in adulthood had active, yet undiagnosed, CD earlier in life and in some individuals there was also other auto-immune disease activity. An adverse effect on renal function in CD could be mediated through several mechanisms such as exposure to nephrotoxic substances [9,46–48], and high nitric oxide production (nitric oxide is also a pro-inflammatory mediator in renal disease [49] and nitric oxide synthase inhibitors may improve renal function [50]). Also greater activation of auto-reactive peripheral T-cells [51,52] could contribute to the increased risk of renal disease in individuals with CD. CD has a negative effect on blood glucose regulation in individuals with type 1 DM [7]. In parallel, it may worsen the prognosis of a number of renal diseases, thereby increasing the need for renal replacement therapy. J. F. Ludvigsson et al. Coeliac disease and renal disease 38. Lindstrom M, Moghaddassi M, Bolin K, Lindgren B, Merlo J. Social participation, social capital and daily tobacco smoking: a population-based multilevel analysis in Malmo, Sweden. Scand J Public Health 2003; 31: 444–450 39. Snook JA, Dwyer L, Lee-Elliott C, Khan S, Wheeler DW, Nicholas DS. Adult coeliac disease and cigarette smoking (see comments). Gut 1996; 39: 60–62 40. Suman S, Williams EJ, Thomas PW, Surgenor SL, Snook JA. Is the risk of adult coeliac disease causally related to cigarette exposure? Eur J Gastroenterol Hepatol 2003; 15: 995–1000 41. Ejerblad E, Fored CM, Lindblad P et al. Association between smoking and chronic renal failure in a nationwide populationbased case-control study. J Am Soc Nephrol 2004; 15: 2178–2185 42. Ivarsson A, Persson LA, Juto P, Peltonen M, Suhr O, Hernell O. High prevalence of undiagnosed coeliac disease in adults: a Swedish population-based study. J Intern Med 1999; 245: 63–68 43. Danielsson L, Stenhammar L, Ascher H et al. Proposed criteria for diagnosis of celiac disease in children. Lakartidningen 1998; 95: 2342–2343 44. Rostami K, Kerckhaert J, Tiemessen R, von Blomberg BM, Meijer JW, Mulder CJ. Sensitivity of antiendomysium and antigliadin antibodies in untreated celiac disease: disappointing in clinical practice (see comments). Am J Gastroenterol 1999; 94: 888–894 45. Smedby KE, Akerman M, Hildebrand H, Glimelius B, Ekbom A, Askling J. Malignant lymphomas in coeliac disease: evidence of increased risks for lymphoma types other than enteropathy-type T cell lymphoma. Gut 2005; 54: 54–59 46. Kuitunen M, Savilahti E. Gut permeability to human alpha-lactalbumin, beta-lactoglobulin, mannitol, and lactulose in celiac disease. J Pediatr Gastroenterol Nutr 1996; 22: 197–204 47. Kovacs T, Kun L, Schmelczer M, Wagner L, Davin JC, Nagy J. Do intestinal hyperpermeability and the related food antigens play a role in the progression of IgA nephropathy? I. Study of intestinal permeability. Am J Nephrol 1996; 16: 500–505 48. Biagi F, Corazza GR. Gene and gliadin/gut and kidney. Am J Gastroenterol 2002; 97: 2486–2488 49. Trachtman H. Nitric oxide and glomerulonephritis. Semin Nephrol 2004; 24: 324–332 50. Ishizuka S, Cunard R, Poucell-Hatton S et al. Agmatine inhibits cell proliferation and improves renal function in anti-thy-1 glomerulonephritis. J Am Soc Nephrol 2000; 11: 2256–2264 51. Di Sabatino A, Bertrandi E, Casadei Maldini M, Pennese F, Proietti F, Corazza GR. Phenotyping of peripheral blood lymphocytes in adult coeliac disease. Immunology 1998; 95: 572–576 52. Di Sabatino A, D’Alo S, Millimaggi D et al. Apoptosis and peripheral blood lymphocyte depletion in coeliac disease. Immunology 2001; 103: 435–440 Received for publication: 15.2.06 Accepted in revised form: 23.2.06 Downloaded from http://ndt.oxfordjournals.org/ by guest on July 14, 2015 21. Whitehead EM, Daly JG, Hayes JR. Renal tubular acidosis in association with Sjogren’s syndrome, primary biliary cirrhosis and coeliac disease. Ir J Med Sci 1987; 156: 124–125 22. Saalman R, Fallstrom SP. High incidence of urinary tract infection in patients with coeliac disease. Arch Dis Child 1996; 74: 170–171 23. Fanos V, Verlato G, Matti P, Pizzini C, Maffeis C. Increased incidence of urinary tract infections in patients with coeliac disease. Pediatr Nephrol 2002; 17: 570–571 24. Peters U, Askling J, Gridley G, Ekbom A, Linet M. Causes of death in patients with celiac disease in a population-based Swedish cohort. Arch Intern Med 2003; 163: 1566–1572 25. Gimenez Llort A, Vila Cots J, Camacho Diaz JA, Vila Santandreu A, Concheiro Guisan A, Garcia Garcia L. Nephrotic syndrome associated with Celiac disease. A report of five cases. Nephron 2002; 92: 950 26. Barera G, Bonfanti R, Viscardi M et al. Occurrence of celiac disease after onset of type 1 diabetes: a 6-year prospective longitudinal study. Pediatrics 2002; 109: 833–838 27. Sollid LM, Markussen G, Ek J, Gjerde H, Vartdal F, Thorsby E. Evidence for a primary association of celiac disease to a particular HLA-DQ alpha/beta heterodimer. J Exp Med 1989; 169: 345–350 28. Caillat-Zucman S, Garchon HJ, Timsit J et al. Age-dependent HLA genetic heterogeneity of type 1 insulin-dependent diabetes mellitus. J Clin Invest 1992; 90: 2242–2250 29. Lie BA, Sollid LM, Ascher H et al. A gene telomeric of the HLA class I region is involved in predisposition to both type 1 diabetes and coeliac disease. Tissue Antigens 1999; 54: 162–168 30. Mogensen CE, Christensen CK, Vittinghus E. The stages in diabetic renal disease. With emphasis on the stage of incipient diabetic nephropathy. Diabetes 1983; 32 [Suppl 2]: 64–78 31. Mathiesen ER, Oxenboll B, Johansen K, Svendsen PA, Deckert T. Incipient nephropathy in type 1 (insulin-dependent) diabetes. Diabetologia 1984; 26: 406–410 32. Ritz E, Stefanski A. Diabetic nephropathy in type II diabetes. Am J Kidney Dis 1996; 27: 167–194 33. Kramer HJ, Nguyen QD, Curhan G, Hsu CY. Renal insufficiency in the absence of albuminuria and retinopathy among adults with type 2 diabetes mellitus. JAMA 2003; 289: 3273–3277 34. Gans RO, Ueda Y, Ito S et al. The occurrence of IgAnephropathy in patients with diabetes mellitus may not be coincidental: a report of five cases. Am J Kidney Dis 1992; 20: 255–260 35. Reunala T, Collin P. Diseases associated with dermatitis herpetiformis. Br J Dermatol 1997; 136: 315–318 36. Adamson J, Ben-Shlomo Y, Chaturvedi N, Donovan J. Ethnicity, socio-economic position and gender – do they affect reported health-care seeking behaviour? Soc Sci Med 2003; 57: 895–904 37. Fored CM, Ejerblad E, Fryzek JP et al. Socio-economic status and chronic renal failure: a population-based case-control study in Sweden. Nephrol Dial Transplant 2003; 18: 82–88 1815