Autoantibody testing in children with newly diagnosed type 1 diabetes mellitus

Objectives To determine the role of autoantibody tests for autoimmune diseases in children with newly diagnosed type 1 diabetes mellitus. Data sources MEDLINE, EMBASE and the Cochrane Library. Citation lists of included studies were scanned and relevant professional and patient websites reviewed. Laboratories and manufacturers were contacted to identify ongoing or unpublished research. Review methods Following scoping searches on thyroid and coeliac autoantibodies, a systematic review of autoantibody tests for diagnosis of coeliac disease was carried out. Studies were included where cohorts of untreated patients with unknown disease status were included, all patients had undergone the reference test (biopsy) and antibody tests, and sensitivity and specificity were reported or calculable. Selected studies were then evaluated against a quality checklist. Summary statistics of diagnostic accuracy, i.e. sensitivity, specificity, positive and negative likelihood ratios and diagnostic odds ratios, were calculated for all studies. A decision analytic model was developed to evaluate the cost utility of screening for coeliac disease at diagnosis of diabetes. Results All antibody tests for diagnosis of coeliac disease showed reasonably good diagnostic test accuracy. Studies reported variable measures of test accuracy, which may be due to aspects of study quality, differences in the tests and their execution in the laboratories, different populations and reference standards. The decision analytic model indicated screening for coeliac disease at diagnosis of diabetes was cost-effective. Sensitivity analyses exploring variations in the cost and disutility of gluten-free diet, the utilities attached to treated and untreated coeliac disease and the decrease in life expectancy associated with treated and untreated coeliac disease did substantially affect the cost-effectiveness of the screening strategies considered. Conclusions In terms of test accuracy in testing for coeliac disease, immunoglobulin A (IgA) anti-endomysium is the most accurate test. If an enzyme-linked immunoassay test was required, which may be more suitable for screening purposes as it can be semi-automated, testing for IgA tissue transglutaminase is likely to be most accurate. The decision analytic model shows that the most accurate tests combined with confirmatory biopsy are the most cost-effective, whilst combinations of tests add little or no further value. There is limited information regarding test accuracy in screening populations with diabetes, and there is some uncertainty over whether the test characteristics would remain the same. Further research is required regarding the role of screening in silent coeliac disease and regarding long-term outcomes and complications of untreated coeliac disease.