Skip to main content
. Author manuscript; available in PMC: 2023 Apr 20.
Published in final edited form as: Diabetes Res Clin Pract. 2020 Dec 15;171:108624. doi: 10.1016/j.diabres.2020.108624

Table 5 -.

Demographic and Clinical Variables associated with Autoimmune Type 1 Diabetes in the Training and Validation Cohorts: the DiYA Study.

Variables Individuals in the Training Cohort (66.7%)
Individuals in the Validation Cohort (33.3%)
DAA negative DAA positive Adjusted OR
(95% CI) Model 1
Adjusted OR
(95% CI) Model 2
DAA negative DAA positive Adjusted OR
(95% CI) Model 1*
Adjusted OR
(95% CI) Model 2
92.3% 7.7% 91.6% 8.4%

DEMOGRAPHICS
Age at Dx (years)
20–24 82.2% 17.8% Reference Reference 82.3% 17.7% Reference Reference
25–29 90.7% 9.3% 0.58 (0.28,1.18) 0.68 (0.31,1.47) 88.2% 11.8% 0.75 (0.30,1.86) 0.76 (0.27,2.13)
30–39 92.2% 7.8% 0.51 (0.28,0.92) 0.69 (0.35,1.35) 91.6% 8.4% 0.60 (0.27,1.33) 0.81 (0.33,2.00)
40–45 93.5% 6.5% 0.39 (0.20,0.78) 0.57 (0.26,1.25) 93.2% 6.8% 0.46 (0.19,1.14) 0.67 (0.24,1.86)
Sex
Male 92.7% 7.3% 0.70 (0.47,1.05) 0.70 (0.47,1.06) 92.1% 7.9% 0.71 (0.44,1.16) 0.71 (0.43,1.18)
Female 91.8% 8.2% Reference Reference 91.0% 9.0% Reference Reference
Race/ethnicity
Hispanic 92.7% 7.3% 0.70 (0.41,1.18) 0.79 (0.46,1.37) 92.3% 7.7% 0.67 (0.39,1.17) 0.82 (0.46,1.48)
Non-Hispanic White 89.8% 10.2% Reference Reference 87.7% 12.3% Reference Reference
Asian 95.1% 4.9% 0.35 (0.16,0.75) 0.43 (0.2,0.95) 94.8% 5.2% 0.33 (0.14,0.80) 0.43 (0.17,1.06)
Non-Hispanic Black 92.2% 7.8% 0.66 (0.30,1.45) 0.7 (0.31,1.59) 93.0% 7.0% 0.52 (0.20,1.38) 0.66 (0.24,1.78)
Other/missing race 87.8% 12.2% 1.19 (0.52,2.77) 1.38 (0.59,3.21) 86.4% 13.6% 1.17 (0.39,3.51) 1.51 (0.49,4.65)
CLINICAL
BMI Category at Dx (kg/m2)
<20 50.1% 49.9% 2.66 (0.92,7.64) 2.15 (0.63,7.3) 41.6% 58.4% 2.38 (0.42,13.47) 1.19 (0.16,8.67)
20–24 80.4% 19.6% Reference Reference 71.1% 28.9% Reference Reference
25–29 89.7% 10.3% 0.51 (0.31,0.84) 0.67 (0.39,1.16) 88.3% 11.7% 0.35 (0.18,0.70) 0.48 (0.22,1.04)
30–39 93.5% 6.5% 0.27 (0.16,0.45) 0.38 (0.21,0.70) 93.8% 6.2% 0.16 (0.08,0.33) 0.25 (0.11,0.58)
≥40 94.4% 5.6% 0.19 (0.11,0.35) 0.29 (0.15,0.56) 94.5% 5.5% 0.12 (0.05,0.29) 0.20 (0.08,0.52)
HbA1c % mean (SD) 8.4 (2.3) 9.2 (2.7) 1.11 (1.02,1.22) 1.08 (0.98,1.18) 8.5 (2.3) 9.5 (2.7) 1.15 (1.04,1.27) 1.11 (0.99,1.23)
HbA1c (mmol/mol), mean, (SD) 68 (2) 77 (6) 69 (2) 80 (6)
Total Cholesterol 194.3 (54.1) 185.9 (48.6) 0.995 (0.991,1.000) 0.995 (0.991,1.000) 196.3 (55.7) 189.9 (50.4) 0.995(0.990,1.000) 0.995(0.990,1.000)
Any Insulin use (≤6 mos.) 77.5% 22.5% 1.86 (1.06,3.27) 73.1% 26.9% 1.90 (0.98,3.68)
% type 1 diabetes ICD10 codes 0.6 (6.8) 8.7 (25.9) 1.02 (1.01,1.03) 0.9 (8.51) 14.6 (32.45) 1.02 (1.01,1.03)
AUC (95% CI)§ 0.678 (0.674–0.682) 0.706 (0.700–0.711) 0.686 (0.682–0.690) 0.719 (0.714–0.724)

Note: All of the variables shown in these models improved the AUC, even if they do not have significant CIs.

*

Model 1 includes: demographics (younger age, female sex, NHW race/ethnicity) and clinical characteristics at time of diagnosis (higher hemoglobin HbA1c, lower BMI, and lower total cholesterol).

Model 2 includes: variables in model 1 plus insulin use within 6 months of diabetes diagnosis and percent of ICD-10 codes that were for type 1 diabetes out of all diabetes diagnosis codes in the first 6 months after diagnosis.

Out of all diabetes codes in the 6 months after diagnosis.

§

AUC: Area under the curve. AUC > 0.5 indicates that the model predicts diabetes autoantibody status greater than chance alone.