Diabetic Ketoacidosis at the Time of Diagnosis of Type 1 Diabetes in Children

Abstract

This cohort study assesses the risk of diabetic ketoacidosis at the diagnosis of type 1 diabetes among children who participated in the Trial to Reduce Insulin-Dependent Diabetes Mellitus in the Genetically at Risk (TRIGR).

Type 1 diabetes (T1D) is one of the most common chronic diseases of childhood. If left untreated, diabetic ketoacidosis (DKA), a largely preventable life-threatening complication, will occur. Currently, 19% of Canadian children and 40% of US children will present with DKA at the time of diagnosis of T1D.^1,2 Because symptoms of T1D exist before the onset of DKA, 1 risk factor for DKA is a delay in the diagnosis and treatment of T1D. Other risk factors include younger age (<5 years) and lower socioeconomic status (SES).¹

Several prospective studies, including the Trial to Reduce Insulin Dependent Diabetes Mellitus in the Genetically at Risk (TRIGR), have followed up children at increased genetic risk of T1D from birth.^3,4 Within these research settings, there was an increased awareness of diabetes risk and symptoms provided by the research teams, likely leading to earlier diagnosis and a decreased DKA risk at diagnosis. In a longitudinal study of children at increased genetic risk of T1D (the Environmental Determinants of Diabetes in the Young [TEDDY] study), participants experienced significantly less DKA at the time of diagnosis when compared with patients with new-onset T1D reported in national diabetes registries.⁴

The objectives of this cohort study were to assess (1) the risk of DKA at T1D diagnosis in children followed up in the prospective TRIGR study and (2) whether participation in TRIGR, in which parents are aware of their child’s increased risk of developing T1D, is associated with decreased SES and age disparities in DKA risk at the time of diagnosis.

Methods

The TRIGR study protocol and results of the main study have been previously published.^3,5 Written informed consent was obtained from the families before enrollment. The TRIGR study was approved by the ethics committees of all participating centers (eAppendix in the Supplement) and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Diabetic ketoacidosis was identified by treating physicians and defined according to centers’ respective national diabetes guidelines (ketonemia combined with a pH < 7.30).⁶ We used t tests to compare continuous variables and the Pearson χ² test for categorical variables. Two-sided P < .05 was considered statistically significant. All analyses were performed using SAS statistical software, version 9.4 (SAS Institute Inc).

Results

A total of 173 children were diagnosed with T1D, among whom 8 patients (4.6%; 95% CI, 2.0%-8.9%) presented in DKA at diagnosis. Compared with DKA rates in the general population (19%-40%; P < .001) and rates observed in the TEDDY study (13.1%; P < .001), TRIGR particpants had lower rates of DKA at T1D diagnosis. Half of the DKA presentations were in the US. Among the 32 US children diagnosed with T1D, 4 (12.5%) presented in DKA, compared with only 1 of 46 (2.2%) Canadian children. Compared with those without DKA, patients with DKA were more likely to have fewer antibodies (mean [SD] number of positive antibodies, 2.6 [1.7] vs 4.1 [1.2]; P < .001) and more likely to have reported weight loss (4 [50.0%] vs 27 [16.4%]; P = .03) and fatigue (5 [62.5%] vs 46 [27.9%]; P = .049) at diagnosis (Table). Age at diagnosis, SES (as measured by parental educational level), and HLA antigen risk category were not associated with DKA risk (Table).

Table. Study Population Characteristics in Children With or Without DKA at Time of Diagnosis.

	No. (%)			P valuea
	Study population	Participants presenting
		With DKA	Without DKA
Population size	173 (100)	8 (4.6)	165 (95.4)
Age at diagnosis, mean (SD), y	6.1 (3.5)	4.5 (4.7)	6.2 (3.5)	.18
Male	91 (52.6)	3 (37.5)	88 (53.3)	.38
Regionb
Finland	34 (19.7);	2 (5.9)	32 (94.1)	.09
Canada	46 (26.6)	1 (2.2)	45 (97.8)
United States	32 (18.5)	4 (12.5)	28 (87.5)
Other	61 (35.3)	1 (1.6)	60 (98.4)
Symptomatic at diagnosis	131 (75.7)	8 (100)	123 (74.5)	.10
Weight loss at diagnosis	32 (18.5)	4 (50.0)	27 (16.4)	.03
No. of positive antibodies (ICA, IAA, GADA, IA-2A, ZnT8A), mean (SD)	4.0 (1.2)	2.6 (1.7)	4.1 (1.2)	<.001
HbA1c at diagnosis, mean (SD), %c	8.2 (1.8)	9.0 (2.1)	8.2 (1.8)	.31
HbA1c at diagnosis, mean (SD), mmol/mmolc	66.4 (20.4)	79.5 (26.6)	66.0 (20.1)	.14
Maternal age, y
<25	30 (17.3)	4 (50.0)	26 (15.8)	.05
25-29	49 (28.3)	0	49 (29.7)
30-34	58 (33.5)	2 (25.0)	56 (33.9)
≥35	36 (20.8)	2 (25.0)	34 (20.6)
IGT on OGTT resultsd	20 (28.2)	0	20 (28.6)	.53
Level of education, mean (SD), y
Maternal	14.7 (3.0)	14.9 (3.4)	14.7 (3.0)	.85
Paternale	14.5 (3.4)	15.3 (3.8)	14.4 (3.4)	.52

Abbreviations: DKA, diabetic ketoacidosis; GADA, glutamic acid decarboxylase antibodies; HbA_1c, hemoglobin A_1c; IAA, insulin autoantibodies; IA-2A, tyrosine phosphatase–related insulinoma-associated 2-molecule autoantibody; ICA, islet cell cytoplasmic autoantibodies; IGT, impaired glucose tolerance; OGTT, oral glucose tolerance test; ZnT8A, zinc transporter 8 autoantibodies.

SI conversion factor: To convert HbA_1c to a proportion of total hemoglobin, multiply by 0.01.

^aP < .05 (2-sided) was considered statistically significant.

^bNumber of cases divided by the regional sample population (denominator).

^cData on HbA_1c were missing for 18 patients (3 with DKA and 15 without DKA).

^dMean (SD) time from IGTT results to type 1 diabetes diagnosis was 1.3 (1.2) years.

^eData on mean level of paternal education were missing for 13 patients (2 with DKA and 11 without DKA).

Discussion

TRIGR participants had a reduced risk of DKA at T1D diagnosis compared with DKA rates in the general population (19%-40%; P < .001) and rates observed in the TEDDY study (13.1%; P < .001).⁴ All TRIGR participants had a family member with T1D, whereas the TEDDY participants were primarily from the general population and fewer had a first-degree relative with T1D.⁴ Socioeconomic status was not associated with DKA risk, which may be attributable to our small sample size or suggest that increased awareness lessened the SES disparities generally observed with DKA risk at diagnosis.

With these small numbers, our results suggest regional disparities in DKA risk, necessitating future exploration of whether differences in health care availability and access are factors that underly our findings. As observed in the present study, younger age at diagnosis is known to be associated with DKA risk at T1D diagnosis. However, our results did not reach statistical significance in our small sample size. Our study has some limitations, including a small sample size; as such, statistically significant SES and age disparities in DKA risk were not observed. Although our DKA rates were low, presumably due to increased T1D awareness, further exploration is needed regarding why DKA episodes were not completely preventable in T1D longitudinal follow-up studies.

Supplement.

eAppendix. TRIGR Study Centers