The browser you are using is not supported by this website. All versions of Internet Explorer are no longer supported, either by us or Microsoft (read more here:

Please use a modern browser to fully experience our website, such as the newest versions of Edge, Chrome, Firefox or Safari etc.

Default user image.

Carina Törn


Default user image.

Predicting Islet Cell Autoimmunity and Type 1 Diabetes : An 8-Year TEDDY Study Progress Report


  • Jeffrey P Krischer
  • Xiang Liu
  • Kendra Vehik
  • Beena Akolkar
  • William A Hagopian
  • Marian J Rewers
  • Jin-Xiong She
  • Jorma Toppari
  • Anette-G Ziegler
  • Åke Lernmark

Other contributions

  • Daniel Agardh
  • Carin Andrén Aronsson
  • Maria Ask
  • Jenny Bremer
  • Emelie Ericson-Hallström
  • Annika Björne Fors
  • Lina Fransson
  • Thomas Gard
  • Rasmus Bennet
  • Susanne Hyberg
  • Hanna Jisser
  • Fredrik Johansen
  • Berglind Jónsdóttir
  • Helena Elding Larsson
  • Marielle Lindström
  • Markus Lundgren
  • Maria Månsson Martinez
  • Maria Markan
  • Marie Jessica Melin
  • Zeliha Mestan
  • Caroline N Nilsson
  • Karin Ottosson
  • Kobra Rahmati
  • Anita Ramelius
  • Falastin Salami
  • Anette Sjöberg
  • Birgitta Sjöberg
  • Carina Törn
  • Anne Wallin
  • Åsa Wimar
  • Sofie Åberg

Summary, in English

OBJECTIVE: Assessment of the predictive power of The Environmental Determinants of Diabetes in the Young (TEDDY)-identified risk factors for islet autoimmunity (IA), the type of autoantibody appearing first, and type 1 diabetes (T1D).

RESEARCH DESIGN AND METHODS: A total of 7,777 children were followed from birth to a median of 9.1 years of age for the development of islet autoantibodies and progression to T1D. Time-dependent sensitivity, specificity, and receiver operating characteristic (ROC) curves were calculated to provide estimates of their individual and collective ability to predict IA and T1D.

RESULTS: HLA genotype (DR3/4 vs. others) was the best predictor for IA (Youden's index J = 0.117) and single nucleotide polymorphism rs2476601, in PTPN22, was the best predictor for insulin autoantibodies (IAA) appearing first (IAA-first) (J = 0.123). For GAD autoantibodies (GADA)-first, weight at 1 year was the best predictor (J = 0.114). In a multivariate model, the area under the ROC curve (AUC) was 0.678 (95% CI 0.655, 0.701), 0.707 (95% CI 0.676, 0.739), and 0.686 (95% CI 0.651, 0.722) for IA, IAA-first, and GADA-first, respectively, at 6 years. The AUC of the prediction model for T1D at 3 years after the appearance of multiple autoantibodies reached 0.706 (95% CI 0.649, 0.762).

CONCLUSIONS: Prediction modeling statistics are valuable tools, when applied in a time-until-event setting, to evaluate the ability of risk factors to discriminate between those who will and those who will not get disease. Although significantly associated with IA and T1D, the TEDDY risk factors individually contribute little to prediction. However, in combination, these factors increased IA and T1D prediction substantially.


  • Celiac Disease and Diabetes Unit
  • EXODIAB: Excellence of Diabetes Research in Sweden
  • Transport and Roads
  • Paediatric Endocrinology

Publishing year







Diabetes Care





Document type

Journal article


American Diabetes Association


  • Endocrinology and Diabetes



Research group

  • Celiac Disease and Diabetes Unit
  • Paediatric Endocrinology


  • ISSN: 1935-5548