Table 2. Effect estimate of year of diagnosis, age, gender, family history of type 1 diabetes (T1D) and residence on risk of diabetic ketoacidosis at diagnosis of T1D.

Variables	pH < 7.3 vs pH ≥ 7.3			pH < 7.1 vs pH ≥ 7.3
	OR	95% CI	p	OR	95% CI	p
Year of diagnosis	1.02	1.00–1.03	0.060	1.03	1.003–1.05	0.028
Age group (0–4 vs 5–9 years)	1.85	1.65–2.07	<0.001	2.09	1.75–2.49	<0.001
Age groups (0–4 vs 10–14 years)	1.60	1.43–1.79	<0.001	1.75	1.47–2.08	<0.001
Gender (Male vs Female)	0.96	0.88–1.04	0.305	0.87	0.76–0.998	0.047
Familiarity for T1DM (Yes vs No)	0.32	0.26–0.40	<0.001	0.29	0.19–0.42	<0.001
Residence area (North vs Centre)	1.12	0.99–1.28	0.082	1.07	0.87–1.32	0.534
Residence area (South/Sicily vs Centre)	2.19	1.92–2.49	<0.001	1.85	1.51–2.28	<0.001
Residence area (Sardinia vs Centre)	0.69	0.56–0.85	<0.001	0.68	0.48–0.95	0.026

Results of the multiple logistic regression analyses. Odds ratios (ORs) for the age groups are calculated by reversing the OR, with age group 0–4 years as reference category. First model (pH < 7.3 vs pH ≥ 7.3): Hosmer and Lemeshow test:χ² = 9.775, df = 8, p-value 0.281; LR test: χ² = 546, df = 8, p-value < 0.001. Second model (pH < 7.1 vs pH ≥ 7.3): Hosmer and Lemeshow test: χ² = 8.411, df = 8, p-value 0.394; LR test: χ² = 212, df = 8, p-value < 0.001.