In a previous study (1), we found that the diagnosis of maternal-child blood group incompatibility appeared as a risk factor for type 1 diabetes, but we were not able to disentangle possible treatment effects from that of diagnosis. A European population-based multicenter study confirmed the association of type 1 diabetes with jaundice and blood group immunization (2). Since our previous study, the number of infants recorded in the Swedish Childhood Diabetes Registry has almost doubled. We therefore repeated our study with special stress on the possible risk associated with phototherapy.
The Childhood Diabetes Registry, containing information on 7,343 children with diabetes and born during the period 1973–1997, was linked with the Medical Birth Registry to get individual and diabetes-independent information on neonatal events. No linkage occurred for 327 (4.5%) diabetic children. We excluded twins, infants born of diabetic mothers, and infants whose county of birth was unknown, leaving 6,487 cases for analysis and 2.8 million births for comparison. We used Mantel-Haenszel’s weighted estimates of odds ratios (ORs), stratifying for year of birth. A logistic multiple regression analysis was also made with further adjustments for putative confounders.
A diagnosis of jaundice gave an OR of 1.13 (95% CI 1.01–1.26). The OR for therapy, irrespective of diagnosis, was 3.79 (3.13–4.59). Therapeutic traditions and coding habits may vary by county and over time; the OR was relatively constant over time but varied between counties. We selected three counties with more than five treated infants who developed type 1 diabetes and added the other 21 counties into one group (0–5 treated cases per county, 54 total). The OR for therapy and type 1 diabetes was 2.41 (1.36–4.17) in the first county, 6.03 (3.29–10.8) in the second county, 4.80 (2.64–8.53) in the third county, and 2.62 in the remaining counties. The heterogeneity was significant between the groups (P = 0.015). The ORs for a jaundice diagnosis for the two high-risk counties was 1.30 (0.79–2.16) and thus not different from the rest (1.01 [0.57–1.53]). We analyzed the type 1 diabetes risk in the two high-risk counties with enough recorded events and in the remaining counties in logistic multiple regression analyses, entering year of birth, preterm birth, respiratory symptoms in the newborn, blood group immunization, and phototherapy as explanatory variables. The phototherapy OR was 1.95 (1.19–3.20) in the two selected counties and 1.06 (0.78–1.43) in the other counties.
Our findings indicate that previous observations that neonatal jaundice and/or blood group incompatibility syndromes are associated with type 1 diabetes risk may be due to phototherapy treatment independent of diagnosis. The difference in OR between the counties indicates that actual treatment regimes could be of importance, and a medical record study could pinpoint such differences. The mechanism behind the association is unknown; however, effects on the neonatal gut and gut immune response are possible, and the frequent use of this treatment combined with the increasing incidence of childhood type 1 diabetes requires further study.
Article Information
This study was supported by grants from the Swedish Medical Research Council (project no. 07531), the Swedish Diabetes Association, and the Nordic Insulin Foundation.
The authors thank all members of the Swedish Childhood Diabetes Study Group who contributed to the Swedish Childhood Diabetes Registry and the National Board of Health and Welfare for providing access to the Medical Birth Registry.
References
Address correspondence to Prof. Gisela Dahlquist, Umeå University, Department of Clinical Sciences—Paediatrics, S-901 85 Umeå, Sweden. E-mail: [email protected].
