Little is known about the prevalence of diabetes autoantibodies (DAA) among young adults newly diagnosed with diabetes (DM). Our objective was to identify patient characteristics associated with presence of DAA among 20-45-year-olds with new-onset DM from 2.3 million Kaiser Permanente California members.

Patients underwent research testing for GADA, ZnT8A, IA2A and IAA. IAA results were not counted for patients with prior insulin use. Demographic and clinical data were from electronic health records. Models predicting any DAA positivity were trained on 70% of cases and validated on the remaining 30%. The best predictive logistic regression model was based on area under the receiver operating characteristic curve (AUC), parsimony and interpretability.

Of 7,862 patients with incident DM (mean age 36.8±6.0 years, BMI 37.1 ±8.7), 1,669 (mean age 36.9 ±6.0 years, BMI 37.2 ±9.2) were tested for DAA ∼9.5 (±3.6) months post-diagnosis; 130 (7.8%) were positive (5.4% single DAA, 2.4% multiple DAA). Prevalence of GADA was 6.1%, IA2A 2.0%, IAA 1.9%, and ZnT8A 1.7%. The best model (Table) yielded AUCs of 0.74 and 0.68 in training and validation datasets, respectively. Association with DAA was positive for younger age, lower BMI, and glucagon prescription and negative for Asian race.

These findings provide insights into patient characteristics associated with antibody positivity in young adults.


J.M. Lawrence: None. J. Slezak: None. X. Li: None. L. Yu: None. M. Rewers: None. H.S. Takhar: None. S.B. Sridhar: None. L.I. Vega-Daily: None. J.G. Alexander: None. J. Ritchie: None. S. Saydah: None. G. Imperatore: None. A. Ferrara: None.


Centers for Disease Control and Prevention (1U18DP006289); JDRF (2-SRA-2018-533-S-B)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at