ObjectiveTo explore the relationship between family history of diabetes and frequency of diabetic ketoacidosis (DKA) at diagnosis and to analyze the possible association between HLA genotypes and DKA.Design and methodsWe recruited 510 children and adolescents aged <17 years with type 1 diabetes (T1D) and collected information on first-degree relative (FDR) history of T1D. DKA and severe DKA were defined as blood pH <7.30 and <7.10 at diabetes onset respectively. Risk categories for developing T1D were determined according to various HLA DQA1-DQB1 haplotype combination genotypes.ResultsThe frequency of DKA and severe DKA at diagnosis was 34.7 and 7.2% respectively. DKA was more frequent in younger patients (<2 years (60.0%; P<0.001)) and occurred less in children with at least one FDR affected by T1D (13.0 vs 37.4%, P<0.001). The logistic regression showed that age at diagnosis (<2 years) and increased HLA-associated risk genotypes were independent predictors of DKA (P<0.01, odds ratio (OR)=1.068 (95% confidence interval (CI) 1.021–1.117); P<0.05, OR=1.606 (95% CI 1.034–2.475)). Introducing the presence of T1D in at least one FDR in the logistic model, a significant association between DKA and age at diagnosis (<2 years; P<0.01, OR=1.072 (95% CI 1.024–1.123)) and absence of FDRs with T1D (P=0.001, OR=4.287 (95% CI 1.770–10.383)) was found, but no more with increased HLA-associated risk genotype (P=0.06, OR=1.550 (95% CI 0.992–2.423)).ConclusionsHLA-associated high-risk genotypes are associated with a high chance of presenting DKA at diabetes onset. However, having at least one FDR with T1D reduced the risk of DKA regardless of HLA genotype.