Which factors predict glycemic control in children diagnosed with type 1 diabetes before 6.5 years of age?
- First Online:
- Cite this article as:
- Shalitin, S. & Phillip, M. Acta Diabetol (2012) 49: 355. doi:10.1007/s00592-011-0321-x
- 247 Views
Predictors of long-term glycemic control and growth patterns in children diagnosed with type 1 diabetes (T1D) before 6.5 years of age were evaluated. One hundred seventy-three children (84 boys) with a mean diabetes duration of 4.9 ± 2.8 years participated in this observational study. Medical charts were reviewed for background, disease- and treatment-related parameters, and growth parameters. Study endpoints were HbA1c value, rates of severe hypoglycemia and diabetic ketoacidosis events, and growth patterns. Mean HbA1c for the total duration of diabetes (HbA1c–TDD) was 7.9 ± 0.8%. Comparison of the study variables between patients with HbA1c–TDD <7.5% (n = 53) or ≥7.5% yielded a significantly shorter duration of diabetes (P = 0.01) and lower rate of diabetic ketoacidosis (P = 0.02) in those with HbA1C–TDD <7.5%, without differences between these groups in age at diabetes onset, insulin regimens, daily glucose measurements, and rate of severe hypoglycemia. Factors significantly predicting achievement of the mean target HbA1c–TDD <7.5% were lower HbA1c at 0.5 years and 1 year after diabetes diagnosis (P = 0.002 and P < 0.001, respectively). Patients followed for at least 5 years (n = 48) showed a significant decrease in height-SDS (P < 0.001) and a significant increase in weight-SDS (P = 0.004) from diabetes diagnosis to the last follow-up visit, without a significant change in weight-SDS from 0.5 years after diagnosis to the last follow-up visit. Our results suggest that in patients with T1D diagnosed during the preschool-age, mean HbA1c level in the first year is a strong predictor of achieving target HbA1c level in the subsequent years, regardless the type of insulin regimen. This “metabolic tracking” emphasizes the importance of achieving early optimal control even in younger children.
KeywordsType 1 diabetesGlycemic controlGrowth
The significant increased incidence rate of type 1 diabetes (T1D) in children [1, 2] shows a secular trend of an epidemic with a conversion from countries with the lower incidence rates in Europe to the countries with the highest incidence rates . Some reports suggest that children living in higher socioeconomic levels may be at higher risk of developing T1D, supporting the hypothesis of an etiological role of environmental factors in the onset of T1D . The increased incidence of T1D has been particularly significant in the younger age group, specifically those less than 5 years old [3, 4].
Achieving glycemic targets in this population poses a difficult challenge. Preschool children are highly susceptible to both hyperglycemia and hypoglycemia as a result of their wide fluctuations in physical activity from day to day, unpredictable eating habits, marked sensitivity to insulin, and limited ability to communicate . Furthermore, evidence indicating that recurrent hypoglycemia may predispose infants and toddlers to neurocognitive developmental defects has prompted health care providers to set higher target glycemic ranges in this age group . At the same time, inadequate glycemic control in the first 5 years of diabetes may shorten the time for the occurrence of microvascular complications . Thus, tight glycemic control is mandatory, even among pediatric patients.
The Diabetes Control and Complication Trial (DCCT) documented the importance of intensive diabetes therapy in lowering HbA1c levels . Insulin therapy can be intensified by either multiple daily injections (MDI) or continuous subcutaneous insulin infusion (CSII). Insulin regimens are believed to affect metabolic outcome, but this relationship is complex, and the optimal insulin regimen seems to be the one which works for the individual being treated.
Available data on growth in children with T1D are conflicting and are derived mainly from cross-sectional studies. Some investigators reported that longitudinal growth is impaired irrespective of the degree of metabolic control, whereas others showed that in patients with good metabolic control, growth patterns are similar to those in normal children [9, 10].
The aim of this observational retrospective study was to assess the predictors of glycemic control in children diagnosed with T1D before 6.5 years of age. Furthermore, rates of severe hypoglycemic and diabetic ketoacidosis (DKA) events, and growth patterns were evaluated.
Patients and methods
Between January 1999 and May 2009, 218 patients younger than 6.5 years (out of 1,150 patients) were diagnosed with T1D at the National Center for Childhood Diabetes, Schneider Children’s Medical Center of Israel, a major tertiary hospital. We included in the study only children who were younger than 6.5 years at diagnosis and younger than 17 years at their last visit and had been followed for at least 1 year after diagnosis at our clinic, with regular clinic attendance. Patients diagnosed before 0.5 years of age with negative diabetes antibodies were excluded due to the possibility of neonatal diabetes. Patients older than 0.5 years, who presented at onset with DKA and without evidence of genetic defects of beta-cell function (MODY syndromes, mitochondrial DNA mutations, and Wolfram syndrome), were included even if they were negative for diabetes antibodies. The medical charts were reviewed for background, disease- and treatment-related data, and growth parameters.
The study was approved by the local Institutional Review Board.
General patient management
All patients are instructed about carbohydrate counting with insulin adjustments and are asked to perform self-blood glucose measurements (SBGM) at least 6 times/day. Compliance with performance of SBGM is assessed every 2–3 months during the follow-up visit.
Several different insulin regimens were used: Type 1: 1–3 injections/day of neutral protamine Hagedorn (NPH) + 2–3 injections/day of human regular insulin; Type 2: 2–3 injections/day of NPH + 2–3 injections/day of rapid-acting insulin analogues (lispro/aspart); Type 3: 1 injection/day of long-acting analogue (glargine/detemir) + 2–3 injections/day of human regular insulin; Type 4: 1–2 injections/day of long-acting analogue + 3 injections/day of rapid-acting insulin analogues; Type 5: 3 injections/day of NPH only; and Type 6: basal-bolus with continuous subcutaneous insulin infusion (CSII).
For statistical analysis, the mode of insulin therapy was evaluated 0.5 years after diabetes onset and once a year thereafter.
The decision about assignment of patients to a specific treatment regimen (different types of MDI or CSII) is based on the recommendation of the multidisciplinary diabetes team. Criteria for changing the insulin regimen include failure to achieve the target HbA1c level, recurrent hypoglycemic episodes, impaired quality of life/need for flexibility in daily activities and meals, and recurrent DKA.
The episodes of severe hypoglycemia and DKA are documented in the patient’s medical file at each visit in the clinic, and after each hospitalization or referral to the emergency room due to these emergencies.
Children are seen routinely at 2- to 3-month intervals, and insulin dosage is adjusted as required at each follow-up visit. The multidisciplinary diabetes team is available 24 h a day for calls and faxes.
Clinical and laboratory measurements
At each clinic visit, standing height is measured using a commercial Harpenden-Holtain stadiometer in children older than 2 years; in younger patients, recumbent measurements of length are made. Weight is measured in light clothing using standard calibrated scales.
Body mass index (BMI) is calculated as weight (kg)/height2 (m). To compare height/length, weight, and BMI values across different ages and by gender, the height/length, weight, and BMI standard deviation scores (SDSs) are calculated using the growth chart percentiles of the Centers for Disease Control and Prevention . BMI-SDS calculations are limited to children older than 2 years, as data are available only for 2–20-year age groups . The physical examination also includes assessment of pubertal stage. Onset of puberty is defined as breast budding in girls and testicular volume ≥4 ml in boys (according to Marshall and Tanner criteria).
Capillary HbA1c is measured every 3 months by an automated immunochemical technique (DCA 2000; Bayer Diagnostics Inc, Tarrytown, NY; reference range: 4.3–5.8%). Mean annual HbA1c was calculated for every year after diagnosis on the basis of the 3-month measurements; these values are then used to calculate mean HbA1c level for the total diabetes duration (HbA1c–TDD). HbA1c at onset of diabetes was excluded from the calculation.
Diabetes antibodies are checked at diagnosis, including insulin antibodies (IAA), islet cell antibodies (ICA512), and glutamic acid decarboxylase (GAD) II antibodies.
The presence of other autoimmune diseases (celiac disease and Hashimoto thyroiditis) is checked from diabetes onset and once yearly thereafter by measuring anti-transglutaminase or anti-endomysial antibodies and free thyroxine (FT4), thyroid-stimulating hormone (TSH) with anti-peroxidase and anti-thyroglobulin antibodies. Patients with positive titers of anti-transglutaminase or anti-endomysial antibodies are referred for intestinal biopsy to confirm celiac disease.
Routine screening for microvascular complications is generally initiated in pre-pubertal children 5 years after onset of T1D with subsequent annual assessment. The screening includes a retinal examination consisted of direct and indirect ophthalmoscopy, testing of urine for microalbumin secretion (based on a 24-h urine collection or the microalbumin-to-creatinine ratio in a first morning voiding urine specimen) and bedside neurological assessment to screen for distal symmetric polyneuropathy.
Good glycemic control was defined as HbA1c <7.5% (target), according to the International Society for Pediatric and Adolescent Diabetes (ISPAD) recommendation . Severe hypoglycemic episodes were defined as coma or seizures or need for glucagon injections or intravenous glucose infusion. DKA events were defined as blood pH <7.3 with bicarbonate <15 mEq/l and need for intravenous fluid and insulin infusion. The frequency of severe hypoglycemic episodes and DKA events was evaluated per 100 patient-years.
Changes in growth parameters including height, weight, and BMI of the study population were evaluated throughout the study period.
Data were analyzed using BMDP software . Pearson’s chi-square test or Fisher’s exact test was used, as appropriate, for analysis of between-group differences in discrete variables; analysis of variance (ANOVA) was used for continuous variables with a normal distribution with Bonferroni correction for multiple comparisons, and Kruskal–Wallis and Mann–Whitney tests for variables with a non-Gaussian distribution. Correlations between continuous variables were analyzed with Spearman’s test. Longitudinal continuous data were compared across groups using ANOVA with repeated measures. Stepwise logistic regression analysis was used to define predictors for achieving the mean target HbA1c–TDD. P < 0.05 was considered statistically significant. Odds ratios were used to quantify the predictors of good glycemic control (target HbA1c). A receiver operating characteristics (ROC) curve was plotted to determine the sensitivity/specificity of the variables in predicting achievement of the target HbA1c–TDD.
A total of 173 patients (84 boys) met the study criteria. Mean age at diagnosis was 3.8 ± 1.6 years (median 4, range 0.7–6.4 years). Mean patient age at the last visit was 8.7 ± 3.2 years (median 8.4, range 2.6–16.9 years). Mean duration of diabetes was 4.9 ± 2.8 years (median 4.3, range 1–11 years).
The ethnic origins of the patients were 146 (84.4%) Jews, 21 (12.1%) Arabs, and 6 (3.5%) Ethiopian Jews. Marital status of biological parents: separated or divorced in 18 families (10.4%) and the others (89.6%) were living together. Low parental literacy (defined as difficulties in writing and reading) was documented in three families.
At diabetes onset, 68 (39.3%) patients presented with DKA, and 139 (80.3%) had positive diabetes antibodies, without a significant difference by gender.
No significant association was found between HbA1c level at onset and DKA at diagnosis.
The difference in HbA1c levels between the various MDI insulin regimens was not statistically significant; therefore, we compared only MDI (≥3 injections, therapy regimens type 1–5) with CSII.
Additional autoimmune diseases were diagnosed during follow-up in 26 patients (17.3%). Celiac disease was diagnosed in 21 patients (12.1%), at a mean age of 5.6 ± 2.0 years (range 2.7–9.8 years). Patients with celiac disease were instructed to maintain a gluten-free diet.
Hashimoto thyroiditis was diagnosed in five patients (2.9%), at a mean age of 9.2 ± 3.4 years (range 3.8–12.1 years). Two of the five patients had subclinical hypothyroidism and were treated with thyroxine. Laboratory control measurements of anti-transglutaminase or anti-endomysial antibodies were performed in patients with celiac disease, and FT4 and TSH in patients with subclinical hypothyroidism every 6 months to ensure compliance with treatment.
None of the patients in this cohort had evidence of microvascular diabetes complications during the time frame assessed.
Mean HbA1c–TDD was 7.9 ± 0.8% (median, 7.9%), with no significant difference by gender.
Mean number of daily SBGM over the duration of diabetes was 5.2 ± 1.5; in 71.1% of the patients, measurements were done 5 times/day or more. There was a negative correlation between mean number of SBGM and HbA1c–TDD, regardless of the insulin regimen used (MDI or CSII), although it was not statistically significant (r = −0.12, P = 0.12).
Significant correlations were found between HbA1c level at 0.5, 1, and 2 years after diabetes onset and the mean HbA1c–TDD (r = 0.68, P < 0.001; r = 0.78, P < 0.001; r = 0.83, P < 0.001, respectively).
Patients with celiac disease (n = 21) compared with those without celiac (n = 152) had a significantly lower mean HbA1c–TDD (7.5 ± 0.8% vs. 8.0 ± 0.8%, P = 0.01). No significant difference in glycemic control was found between patients with Hashimoto thyroiditis and those without.
Children whose parents were married compared with those with separated or divorced parents had a lower mean HbA1c–TDD, although not statistically significant (7.9 ± 0.8% vs. 8.3 ± 1.1%, NS).
No significant differences were found in HbA1c at each year of diabetes and in mean HbA1c–TDD between patients with positive or negative diabetes antibodies. There was also no significant difference in HbA1c levels for each year by type of regimen (MDI or CSII).
Achievement of target HbA1c
Comparison between patients with mean HbA1c–TDD at or above target
n = 53 (30.6%)
Mean ± SD
n = 120 (69.4%)
Mean ± SD
Age at data collection (years)
8.29 ± 2.67
8.89 ± 3.34
Age at diabetes onset (years)
4.12 ± 1.49
3.65 ± 1.68
Diabetes duration (years)
4.18 ± 2.20
5.24 ± 3.03
Patients with DKA at onset (%)
Patients with positive diabetes antibodies at onset (%)
Insulin dose at onset (units/kg/day)
0.85 ± 0.41
0.93 ± 1.29
Basala/total daily insulin dose at onset (%)
0.48 ± 0.14
0.56 ± 0.16
HbA1c at onset (%)
9.5 ± 2.1
10.2 ± 1.8
HbA1c at 0.5 years after onset (%)
6.8 ± 0.9
8.3 ± 1.2
HbA1c at 1 year after onset (%)
7.0 ± 0.6
8.4 ± 0.9
HbA1c at 2 years after onset (%)
7.1 ± 0.5 (n = 42)
8.3 ± 0.8 (n = 98)
HbA1c at 3 years after onset (%)
7.1 ± 0.5 (n = 37)
8.4 ± 0.8 (n = 79)
HbA1c at 4 years after onset (%)
7.2 ± 0.6 (n = 26)
8.4 ± 0.8 (n = 68)
HbA1c at 5 years after onset (%)
6.8 ± 0.5 (n = 14)
8.4 ± 0.9 (n = 57)
HbA1c at 6 years after onset (%)
6.8 ± 0.3 (n = 11)
8.4 ± 0.9 (n = 47)
HbA1c at 7 years after onset (%)
6.9 ± 0.3 (n = 4)
8.3 ± 1.0 (n = 42)
HbA1c at last visit (%)
7.3 ± 0.7
8.4 ± 1.0
Severe hypoglycemic episodesb
5 ± 12
9 ± 20
1 ± 4
3 ± 12
Mean no. of daily SBGM
5.36 ± 1.53
5.08 ± 1.41
Ht-SDS at onset
0.15 ± 0.96
0.18 ± 0.96
Ht-SDS last visit
0.06 ± 0.89
−0.22 ± 1.03
BMI-SDS at onset
−0.6 ± 1.44 (n = 47)
−0.36 ± 1.2 (n = 86)
BMI-SDS last visit
0.38 ± 0.97
0.46 ± 0.86
Severe hypoglycemic episodes and DKA events
Mean number of severe hypoglycemic episodes in the study sample was 7.6 ± 18 per 100 patient-years, and mean number of DKA events was 2.4 ± 10 per 100 patient-years. The rates of severe hypoglycemic events and DKA episodes were significantly positively correlated with diabetes duration (r = 0.33, P < 0.001 and r = 0.24, P < 0.05, respectively) and significantly negatively correlated with mean number of SBGM (r = −0.26, P < 0.01 and r = −0.24, P < 0.05, respectively). The rate of DKA episodes was significantly lower in patients who achieved the mean target HbA1c–TDD than in those who did not (1 ± 4 vs. 3 ± 12 per 100 patient-years, respectively, P = 0.02). However, there was no significant difference between these groups in rate of severe hypoglycemic episodes.
No significant difference was found in the rates of severe hypoglycemic events and DKA episodes between patients treated with MDI and those treated with CSII at each year of diabetes.
In the subgroup of 48 patients (27.7% of the total cohort) who were followed for at least 5 years, there was a significant decrease in height-SDS (0.18 ± 0.84 vs. −0.32 ± 0.98, P < 0.001) and a significant increase in weight- SDS (−0.15 ± 0.87 vs. 0.14 ± 0.84, P = 0.004) from diabetes diagnosis to the last follow-up visit without a significant change in weight-SDS from 0.5 years after diagnosis to the last follow-up visit (0.23 ± 0.84 vs. 0.14 ± 0.84).
No significant correlation was found between mean HbA1c–TDD and height-SDS, weight-SDS, or BMI-SDS at last visit. There was also no significant difference in height-SDS, weight-SDS, or BMI-SDS between the patients who achieved the mean target HbA1c–TDD at the last visit and those who did not.
At the last visit, 125 patients (72.3%) were still pre-pubertal, with a male predominance (70 boys vs. 55 girls, P = 0.012). Mean age of onset of puberty was 11.6 ± 1.1 years in boys (n = 14) and 10.2 ± 1.2 years in girls (n = 32).
Data on the relationship among components of T1D treatment in children are still limited, particularly regarding the insulin regimen and glycemic control goals.
In our cohort of children diagnosed with T1D during the preschool years, the mean target HbA1c level of <7.5% was achieved by 30.6% of patients within a mean follow-up time of 4.9 ± 2.8 years. Shorter duration of diabetes and HbA1c levels at the first year after diabetes onset were significantly related to the achievement of the target HbA1c level in the subsequent years.
Several studies have reported about better glycemic control compared to that in our cohort, especially in very young patients with T1D [14–16]. However, it is noteworthy that during most of the follow-up period (until 2007), the HbA1c target for the individual patients in our clinic was determined according to the recommendations of the American Diabetes Association (ADA) . The ADA sets goals by age: ≤8.5% for patients aged <6 years, ≤8% for ages 6–12 years, and <7.5% for 12–19 years. Thus, the mean HbA1c–TDD (7.9 ± 0.8%) of our cohort was within the ADA recommended target range for children younger than 12 years.
Previous studies in children and adolescents [18, 19] reported the association between better HbA1c level and a higher frequency of daily SBGM, which may represent how well parents/patients are making insulin adjustments in addition to overall compliance with treatment.
The lack of a significant correlation between mean HbA1c–TDD <7.5% and the daily frequency of SBGM in our study may be explained by the relatively frequent measurements performed by the majority of our patients. During the time frame assessed in this study no patient used the continuous glucose-monitoring device.
As reported previously , we found that a shorter duration of diabetes was associated with the achievement of the mean target HbA1c–TDD. Possible explanations to this association are rest of endogenous C-peptide production with shorter disease duration; onset of puberty with higher insulin resistance in part of our patients with longer disease duration; and the need of continued diabetes re-education of patients with long-standing disease (at the point at which the child begins to be partially responsible for his/her own treatment).
Ludvigsson et al.  reported that patients who had received more vigorous treatment immediately at disease onset had both a higher incidence of post-initial remission and better diabetes control. These results suggest that rapid glycemic normalization after diagnosis of diabetes increases the possibility of preserving some endogenous insulin production.
The phenomenon of individual HbA1c “tracking” recently described  suggests that the achievement of good metabolic control soon after diagnosis may lead to better control in the long-term. There are several assumptions for its occurrence, such as residual insulin production and individual biological variation in insulin sensitivity, that may facilitate achievement of better glycemic control for a longer period, as well as diabetes-management-related factors, such as familial social, behavioral, psychological, and emotional factors, which can persist over time.
In our cohort, the percent of children who had DKA at onset was slightly higher in the group with mean HbA1c–TDD ≥7.5% compared with the group with mean HbA1c–TDD <7.5%, which may point to less preserved endogenous insulin secretion from the beginning. However, HbA1c did not converge later in both groups when the remission phase was expected to be over. Therefore, the strong association we found between achieving the target HbA1c during follow-up and the HbA1c levels in the first year of diabetes emphasizes the importance of attaining the target as early as possible, which may facilitate diabetes control also beyond the remission. It may also reflect that establishment of good habits early may predict long-term compliance and adherence with therapy, especially in young patients, when parents are usually the principal caregivers.
It was shown that socio-demographic family factors as single-parenthood is associated with greater risk for poor diabetes control . We found that children with married parents had a marginal better metabolic control compared with children with separated parents.
Published randomized comparisons of MDI and CSII have reported controversial results in terms of metabolic control in children [24–26]. In observational studies, CSII use was associated with lower HbA1c levels and fewer episodes of severe hypoglycemia [15, 27], while in longitudinal observational studies, the efficacy of one particular insulin regimen over others remains unclear . In our cohort, the percent of patients treated with CSII increased during follow-up (from 20.8% after the first year of diabetes to 50.6% after 5 years). Nevertheless, no significant differences were found in glycemic control or in the rate of DKA and severe hypoglycemic events between patients treated with CSII or MDI. However, CSII therapy may still have an advantage in decreasing glucose excursions and the duration and frequency of mild hypoglycemia , but we did not evaluate these parameters.
The lack of a significant association between the rate of severe hypoglycemic episodes and achieving the mean HbA1c targets suggests that better metabolic control does not necessarily increase the rate of hypoglycemia, even in the young age group. Taken together with above-mentioned data suggesting a better chance of preserving endogenous insulin production with rapid glycemic normalization after diabetes onset , it may indicate a need to set lower HbA1c targets for young children as recommended by ISPAD .
Consistent setting of glycaemic targets by diabetes teams is strongly associated with a better HbA1c outcome in adolescents and plays a significant role in explaining the differences in metabolic outcomes between centers . Thus, it seems that setting of lower glycemic targets by the diabetes team also in the younger ages may help to improve the glycemic control of our patients in the subsequent years.
In a subgroup of patients that consisted 27.7% of our total cohort, a significant decrease in height-SDS was found without a significant change in weight-SDS (restorative weight gain after 6 months following initiation of insulin therapy) during follow-up and without a significant correlation with glycemic control. Previous longitudinal studies also have reported the loss of height-SDS in diabetic children of both genders . However, most studies [31, 32] found that the diabetic children reached a final height within the normal range for the reference population and similar or greater height-SDS at the end of the development period than their target heights . However, 125 (72.3%) of our patients were pre-pubertal at their last visit; therefore, we cannot provide data of final height of this cohort. Onset of puberty in our patients occurred at normal age in both genders.
The strengths of our study are the large number of patients and long follow-up (to 7 years). Our study has several limitations: (1) Its retrospective design and lack of a randomized controlled trial to compare the findings by the different insulin regimens; (2) Lack of C-peptide levels across time for the indication of the residual endogenous insulin secretion, which plays an important role in diabetes control; (3) We have no way to guarantee that we have included 100% of the severe hypoglycemic episodes. However, it is one of the limitations of every study whether it is retrospective or prospective, if it is based on patient’s report; and (4) Lack of data on family socioeconomic factors, which may influence glycemic outcome, especially in young children [33, 34], although the health insurance in our country provides equal health facilities for all socioeconomic groups.
In summary, this study shows that HbA1c levels in the first year after diabetes onset are the strong predictor of achieving target HbA1c levels in the subsequent years, regardless the type of insulin regimen (MDI vs. CSII). The HbA1c levels in individuals tend to “track” through the years. Our results emphasize the importance of aiming for lower targets of HbA1c levels within the first year of diabetes by intensive insulin treatment, even in the very young age group, in order to achieve the long-term good glycemic control.
The authors wish to thank Pearl Lilos for the statistical analysis and Gloria Ginzach for her editorial assistance.
Conflict of interest
All authors have nothing to declare.