|Home | About | Journals | Submit | Contact Us | Français|
Starting subcutaneous insulin doses in children with newly diagnosed type 1 diabetes vary widely from 0.2 units/kg/day to 0.8 units/kg/day.
To determine whether there are correlations between starting insulin dose and diabetes-related outcomes.
By reviewing the charts of children newly diagnosed with type 1 diabetes, the prevalence of hypoglycemia in the first 48 h was compared between those who received low (0.5 units/kg/day or less) and those who received high (greater than 0.5 units/kg/day) starting insulin doses.
Forty-two children were initially prescribed a low dose of insulin, and 55 children were given a high dose. Approximately one-third of children (36.4%) younger than six years of age who received a high starting dose of insulin had mild hypoglycemia within 48 h of subcutaneous insulin initiation, compared with 16.0% of children six to 10 years of age and 5.3% of children older than 10 years of age.
Hypoglycemia was not more frequent among children given high-insulin starting doses. However, children younger than six years of age remained at increased risk for hypoglycemia.
Les doses d’insuline de départ par voie sous-cutanée chez les enfants à qui on vient de diagnostiquer un diabète de type 1 varient énormément, passant de 0,2 unité/kg/jour à 0,8 unité/kg/jour.
Déterminer s’il y a des corrélations entre la dose d’insuline de départ et les issues liées au diabète.
Par l’examen du dossier d’enfants venant de recevoir un diagnostic de diabète de type 1, les chercheurs ont étudié la prévalence d’hypoglycémie au cours des 48 premières heures chez ceux qui recevaient une faible dose d’insuline de départ (0,5 unité/kg/jour ou moins) et ceux qui en recevaient une forte dose (plus de 0,5 unité/kg/jour).
Quarante-deux enfants ont d’abord reçu une faible dose d’insuline, et 55, une forte dose. Environ le tiers des enfants (36,4 %) de moins de six ans qui avaient reçu une forte dose d’insuline de départ avaient une légère hypoglycémie dans les 48 heures suivant l’amorce de l’insuline sous-cutanée, par rapport à 16,0 % des enfants de six à dix ans et à 5,3 % des enfants de plus de dix ans.
L’hypoglycémie n’était pas plus courante chez les enfants qui recevaient une forte dose d’insuline de départ. Cependant, les enfants de moins de six ans demeuraient plus vulnérables à l’hypoglycémie.
Type 1 diabetes affects one in 300 to one in 500 children younger than 18 years of age, and has an annual incidence in North American populations of approximately 15 per 100,000 (1). After initial stabilization, all individuals newly diagnosed with type 1 diabetes are started on subcutaneous insulin therapy. Ideally, insulin doses are chosen to achieve target blood glucose levels of 4.0 mmol/L to 7.0 mmol/L in the fasting or preprandial state, and 5.0 mmol/L to 10.0 mmol/L in the postprandial state, although these targets are less stringent in younger age groups (2).
In practice, these targets result in a wide range of recommended starting doses, from 0.2 units/kg/day to 0.8 units/kg/day (3,4), divided between short- and long-acting insulin formulations. There is little reported evidence to direct practitioners regarding the dose they should use within this dosing range to initiate insulin management. Theoretically, the sooner hyperglycemia is corrected, the lower the burden that is placed on the already stressed beta-cells of the pancreas, thus promoting a longer clinical remission or ‘honeymoon’ period (3). In fact, several small randomized, controlled trials, published in the 1980s by Madsbad (5) and Madsbad et al (6), compared intensive treatment with what was then considered conventional treatment. These studies found increased C-peptide secretion – a cleavage protein of human proinsulin that is commonly used as a surrogate for islet cell function – in the intensive treatment groups two weeks postdiagnosis. The Diabetes Control and Complications Trial (DCCT) (7) also showed that intensive insulin therapy was associated with maintenance of higher C-peptide levels. Furthermore, the subgroup of intensively treated subjects with higher C-peptide levels at the end of the study, not only had a decreased incidence of complications (such as retinopathy and microalbuminuria), but also had a decreased frequency of severe hypoglycemia (associated with neuroglycopenic symptoms and requiring assistance to treat) (7). These results suggest that there is some value in initiating treatment for patients with type 1 diabetes at higher insulin doses to preserve endogenous islet cell function.
Interestingly, there is a paucity of literature comparing different starting insulin doses with short-term clinical outcomes and, more specifically, hypoglycemic events and metabolic control. This issue is particularly important given the current trend toward outpatient stabilization programs for individuals newly diagnosed with type 1 diabetes (8). Ambulatory programs have reported initiation of insulin in the lower ranges of typical recommendations, ie, 0.3 units/kg/day to 0.5 units/kg/day (9,10). Conceivably, this course of action is related to concerns about hypoglycemic events occurring at home. The question arises as to whether these more conservative insulin doses, which likely prolong the duration of hyperglycemia, are negatively affecting the prognosis of an individual newly diagnosed with type 1 diabetes by delaying recovery of islet cell function. To our knowledge, no study has compared lower starting doses (0.5 units/kg/day or less) with higher doses (greater than 0.5 units/kg/day) for rates of hypoglycemia within the first days after initiation, nor their effect on metabolic control three to six months after diagnosis.
The present retrospective study was conducted to define subcutaneous insulin prescribing practices at the Alberta Children’s Hospital (Calgary, Alberta) in children and adolescents with newly diagnosed type 1 diabetes, and to determine whether there is a correlation between starting insulin doses and clinical outcomes such as hypoglycemic events, and achievement of target blood glucose levels and glycated hemoglobin (HbA1c) levels (a marker of glycemic control).
After obtaining approval from the local ethics review board, a retrospective chart review (Figure 1) of 135 children 0 to 18 years of age with first presentation of type 1 diabetes, who were treated at the Alberta Children’s Hospital diabetes clinic or emergency department between January 1, 2003, and December 31, 2004, was conducted. Patients were identified from the Alberta Children’s Hospital diabetes clinic files. Once selected for the study, the subjects were identified by chart number only. Eleven charts were excluded for the following reasons: one patient was diagnosed with type 2 diabetes, six patient records were incomplete because of care provision in other centres before or after diagnosis, two patients had only transient hyperglycemia that resolved fully and two patients’ clinic charts were missing. Data collected from the charts included the following: age; sex; weight; height; degree of illness at presentation (blood glucose level, presence or absence of ketones, and presence of ketoacidosis [defined as a blood pH of lower than 7.25 and CO2 concentration of less than 15 mEq/L]); intended starting subcutaneous total daily dose (TDD) of insulin in units per kilogram per day (units/kg/day) as recorded in physician notes; the delivered starting subcutaneous insulin dose, including corrections given for hyperglycemia in the first 24 h postdiagnosis (received TDD in units/kg/day) as recorded in nursing notes; number of hypoglycemic events (blood glucose level less than 4.0 mmol/L by point of care testing using validated glucometers) in the first 48 h after diagnosis; and HbA1c (normal, less than 6.1%) four to six months post-diagnosis. An attempt was made to collect additional data (average blood glucose level, average TDD of insulin in units/kg/day at two weeks postdiagnosis and number of days required to reach stable insulin doses); however, gathering this information was not reliable in the majority of charts, given the constraints of retrospective study designs.
Initial analyses focused on descriptive statistics, including means, SDs and frequencies. Group comparisons of demographic and clinical variables were conducted using ANOVA for continuous variables and χ2 tests of association for categorical variables (with the Fisher’s exact test used as required). For the first set of group comparisons, children were divided into two groups according to intended TDD. In contrast, children were divided according to TDD actually received for the second set of group comparisons. Most patients are started on either 0.5 units/kg/day or 0.7 units/kg/day, depending on the physician’s clinical judgement. However, all physicians also administer additional doses to patients with hyperglycemia; consequently, the dose actually received is often higher. Therefore, for both the intended and the actually received doses, the groups were divided into those who received 0.5 units/kg/day or less, and those who received greater than 0.5 units/kg/day. Correlations were used to examine the relationship between TDD given and HbA1c levels at four to six months after diagnosis. A regression analysis was used to explore factors associated with the occurrence of hypoglycemia.
The results obtained in the present study pertain to the 97 children for whom data regarding the starting subcutaneous insulin dose (units/kg/day) were available. The characteristics of these patients did not differ from the 27 excluded subjects (whose data on intended insulin dose and dose actually received were missing) with respect to age, sex or presenting with diabetic ketoacidosis (DKA). Within this group, the average (± SD) age was 8.2±3.6 years (range 1.0 to 14.4 years); 50 male and 47 female subjects were identified. A total of 15 children (18.1%) presented with DKA. The lowest intended TDD prescribed was 0.3 units/kg/day, which was prescribed to only one patient (1.0%). The highest intended TDD was 0.75 units/kg/day, which was prescribed to 16 patients (16.5%). The most commonly prescribed starting TDDs were 0.5 units/kg/day (42.3%) and 0.7 units/kg/day (39.2%).
For the next portion of the analysis, two groups were identified (Table 1). The lower intended TDD group consisted of 42 patients (43.3%) who were prescribed 0.5 units/kg/day or less. The higher TDD group consisted of 55 patients (56.7%) who were prescribed greater than 0.5 units/kg/day. Both groups had similar sex distributions (52.4% male subjects in the lower TDD group versus 50.9% male subjects in the higher TDD group). The mean age of the lower TDD group was significantly lower than that of the higher TDD group (7.1±3.6 years of age versus 9.0±3.5 years of age) (P=0.013). In each of these two groups, the percentage of short-acting insulin prescribed was similar (42.1% in the lower TDD group versus 41.4% in the higher TDD group). The presenting blood glucose level was equivalent between the groups (mean of 31.1±14.0 mmol/L for the lower TDD group and 30.1±13.7 mmol/L for the higher TDD group). The prevalence of DKA was not significantly different between the lower and higher TDD groups (12.5% versus 21.6%, respectively). The groups did not differ with respect to the presence of comorbidities including allergies, eczema and asthma (52.4% for the lower TDD group versus 56.4% for the higher TDD group). The percentage of children experiencing at least one hypoglycemic episode in the first 48 h was also similar (14.3% for the lower TDD group versus 12.7% for the higher TDD group).
Overall, there was a statistically significant difference between the intended TDD (units/kg/day), as recorded in physician notes, and what was actually administered to the patient based on nursing notes, with a significantly higher dose actually given compared with what was intended (0.71±0.19 units/kg/day versus 0.62±0.11 units/kg/day; P<0.001), reflecting a need to correct for hyperglycemia. When the children were separated into groups according to dose actually given, only 11 children (11.3%) received a TDD of 0.5 units/kg/day or less, while 86 children (88.7%) received greater than 0.5 units/kg/day. When divided into age groups (younger than six years of age [n=29], six to 10 years of age [n=46], older than 10 years of age [n=22]), there was no significant association between age group at presentation and dose actually received (ie, whether the child actually received less than 0.5 units/kg/day, 0.5 units/kg/day to 0.7 units/kg/day, or greater than 0.7 units/kg/day). For children younger than six years of age, 37.9% received an insulin dose of greater than 0.7 units/kg/day; for children six to 10 years of age, 54.3% received an insulin dose of greater than 0.7 units/kg/day; and for children older than 10 years of age, 86.4% received an insulin dose of greater than 0.7 units/kg/day.
Thirteen children (14.3%) experienced at least one hypoglycemic event in the 48 h following initiation of subcutaneous insulin. Of these 13 children, one child had four hypoglycemic events in the first 48 h period; another child had three hypoglycemic events; six children had two events; and the remaining five children had one event over the first 48 h period. All were mild events without neuroglycopenic symptoms, such as confusion, seizure or coma, and all were treated with oral glucose. There was no significant difference between the lower, middle and higher delivered TDD groups with respect to the number of hypoglycemic events on the first or second day; no hypoglycemic events occurred in those in the lower intended TDD group, versus no hypoglycemic events in 87.5% of those in the middle TDD group and 83.6% of those in the higher TDD group. None of the children in the lower delivered TDD group had a hypoglycemic episode, but the sample size (n=11) was considerably smaller than the higher TDD group (n=86). Beyond 48 h, after the start of subcutaneous insulin, the data were sparse because most patients were discharged from the hospital by that point.
When divided by age (Figure 2), there was a trend for an association between age group and hypoglycemia (P=0.099). Of the 29 children who were younger than six years of age, seven (24.1%) experienced at least one hypoglycemic event in the first 48 h of subcutaneous insulin initiation. This was a significantly higher proportion than that of children six to 10 years of age (10.9%), or for children older than 10 years of age (4.5%). When age group, dosage received and incidence of at least one hypoglycemic event in the first 48 h were considered, there was a trend for an association between age group and experiencing at least one hypoglycemic event for the higher TDD group (P=0.085). Approximately one-third of children (36.4%) younger than six years of age who received a high dose had at least one episode compared with 16.0% of children six to 10 years of age and 5.3% of children older than 10 years of age. A logistic regression conducted to determine which factors were predictive of whether hypoglycemia had occurred within the first 48 h was significant overall (P=0.009). Results showed that being in the younger age group (P=0.012) was predictive of experiencing hypoglycemia and that having received the higher TDD (P=0.052) tended to be predictive of experiencing hypoglycemia.
Finally, the TDD given to a subject (as documented in nursing notes) was compared with their HbA1c as measured in a follow-up clinic visit between four and six months post-diagnosis. There was no correlation between the TDD given and HbA1c after four to six months for the overall sample of 97 children. As well, no significant correlations emerged between TDD given and HbA1c at four to six months, when each age group was considered separately. A trend was seen with higher HbA1c in children who had at least one episode of hypoglycemia in the first 48 h after initiation of subcutaneous insulin (8.7±1.7% versus 7.9±1.3%; P=0.062).
Appropriate insulin dosing is critical to long-term management of type 1 diabetes. Randomized, controlled trials, such as the DCCT, have shown that intensive insulin intervention improves residual beta-cell function (7,11) and reduces diabetes-related complications (12). Residual pancreatic beta-cell function is believed to decrease glucose fluctuation and thus improves glycemic control, as measured by HbA1c (13). However, it is less clear whether there is a clinical benefit to starting intensive intervention immediately after diagnosis and stabilization, or whether it is equally efficacious to ease a patient into therapy. Without an answer to this question, recommended starting doses for subcutaneous insulin fall within a wide range, ie, between 0.2 units/kg/day and 0.8 units/kg/day. In the present study, prescribed doses ranged from 0.3 units/kg/day to 0.75 units/kg/day, with the most commonly prescribed doses being 0.5 units/kg/day and 0.7 units/kg/day. When the data were divided into a lower dose insulin group (prescribed 0.5 units/kg/day or less) and a higher dose insulin group (prescribed greater than 0.5 units/kg/day), there was no difference between the groups with respect to the presence of DKA or comorbidities. Thus, the degree of illness did not appear to affect the choice of insulin starting doses. Age was the only factor found to influence intended starting insulin doses, with younger children being prescribed lower doses.
Interestingly, there was a significant difference between the intended TDD as recorded in physician notes and what was actually administered to the patient based on nursing records, with a significantly higher dose actually given compared with what was intended. This reflects the dynamic nature of diabetes management, in which insulin doses are adjusted according to the presence of hyper- or hypoglycemia. Based on the significant upward trend, it is likely that children prescribed a lower TDD of insulin experienced ongoing hyperglycemia, indicating a higher insulin need than predicted.
Within the broad range of prescribed insulin doses, we sought to compare insulin doses with frequency of hypoglycemic events. In the present study, hypoglycemia was defined as a blood glucose level of less than 4.0 mmol/L as measured by a glucometer (or point of care testing), which, according to the Canadian Diabetes Association Clinical Practice Guidelines (2), requires a treatment. All events were confirmed by a point of care glucose test using either a hospital glucose meter or the patient’s meter once it was found to be accurate when compared against the laboratory value or both. All blood glucose tests were also either performed or supervised by a nurse; consequently, patient error in performing the test would be unlikely to have influenced the result. The question of frequency of hypoglycemia in the first few days after insulin initiation becomes especially important, given the transition within many paediatric centres away from long inpatient stays toward early diabetes care provision in the home, with support from medical day-treatment units. A descriptive study by Schneider (9) that examined outpatient management of newly diagnosed paediatric diabetes patients found that an average initial dose of regular insulin of 0.3 units/kg/day yielded no hypoglycemic events in the first 24 h of treatment. In this study, the average total insulin dose one week postdiagnosis had increased to a range of 0.6 units/kg/day to 1.0 units/kg/day, and the authors stated that “no premium was attached to rapid restoration of metabolic control”. Another recent study (14), using data from the Paediatric Quality Initiative, reviewed rates of hypoglycemia in the first two weeks after diagnosis in 1680 children with type 1 diabetes. They found that rates of hypoglycemia rose constantly from 4.8% to a maximum of 11.2% between day 2 and day 5, and then remained stable. The average daily insulin dose on day 2 was 0.79 units/kg/day. The highest average daily insulin dose was 0.91 units/kg/day on day 5, coinciding with the highest rate of hypoglycemia. Younger age, more severe disease at onset (higher HbA1c, higher blood glucose and lower pH) and higher daily insulin doses were associated with a higher risk of hypoglycemia from day 2 to day 14. This study differed from the present study in that it did not provide rates of hypoglycemia within the first 48 h. Furthermore, although they found a correlation between higher daily doses from day 2 to day 14 and hypoglycemia, they did not report on the actual starting doses or on the impact on HbA1c three to six months after diagnosis.
Our study revealed no difference between the high and low delivered TDD insulin groups with respect to hypoglycemic events within 48 h of initiation of subcutaneous insulin. In both groups, more than 85% of patients had no documented hypoglycemia. All events were mild and treated with oral glucose as per the Canadian Diabetes Association Clinical Practice Guidelines (2). However, there was a trend for an increase in the number of hypoglycemic events occurring in the youngest age bracket. Children younger than six years of age were nearly two times more likely than those six to 10 years of age (24.1% versus 10.9%, respectively) and five times more likely than those older than 10 years of age (24.1% versus 4.5%, respectively) to experience a hypoglycemic event. This could be due to reduced endogenous insulin production in younger patients with diabetes. Studies (15) have shown associations between age and beta-cell function, with older children demonstrating a longer course of preserved beta-cell function. Bonfanti et al (16) demonstrated that C-peptide secretion was significantly lower in children with diabetes onset before five years of age. This could be contributing to more labile disease in this age group. The results of our study suggest that prescribing any insulin dose within the recommended range is safe, and that prescribing doses at the higher end of the range does not result in an increase in hypoglycemia. The exception to this is in children younger than six years of age, in which case it may be prudent to consider choosing a lower starting TDD, given the increased risk of hypoglycemia in that age group, combined with the fact that younger children are likely to have more trouble recognizing, communicating and self-managing a hypoglycemic event.
We attempted to describe the long-term outcomes of different starting insulin doses to determine whether higher starting doses within the first 48 h lead to better control four to six months after diagnosis. Theoretically, earlier control of blood glucose may confer less stress on the diabetic pancreas and, thus, has the potential to preserve endogenous islet cell function, leading to more stable diabetes management. As mentioned in the introduction, studies (5,6) supporting this theory have been published. However, it is difficult to generalize these results to today’s patients with diabetes, because the standard of care has changed markedly over the past two decades. When the studies by Madsbad (5) and Madsbad et al (6) were conducted in the early 1980s, blood glucose monitoring was performed infrequently, and conventional dosing yielded high blood glucose levels compared with current standards. Today, standard insulin therapy for the control of blood glucose more closely resembles the results of Madsbad’s intensive treatment group (6). Therefore, the question remains as to whether early, followed by ongoing, intensive blood glucose control confers a benefit to patients with respect to long-term outcomes within today’s standard of care. When we compared the lower delivered TDD group to the higher delivered TDD group, there was no difference in HbA1c at four to six months. This suggests that there is no appreciable value to using higher starting doses, especially if it leads to more frequent hypoglycemia. However, it is well recognized that, over time, intensive glycemic management mitigates the complications of diabetes, including micro- and macrovascular damage; therefore, the need to reach optimal blood glucose control remains.
The limitations of the present study include a relatively small sample size, and a retrospective design that led to exclusion of subjects whose data were incomplete. We were unable to evaluate how long it took patients to reach stable blood glucose levels, and whether this period was prolonged in those started on lower TDD insulin (which would be anticipated). There is potential for poorer quality of life in those who require weeks versus days to achieve normoglycemia, with respect to the patient experiencing hyperglycemic symptoms, and family and caregivers experiencing distress over ‘failing’ to meet target blood glucose levels. Areas for further research include the examination of potential psychological and economic effects on slower progression to normoglycemia.
The results of our study suggests that while current practice in children with newly diagnosed type 1 diabetes allows for a range of starting insulin doses from 0.2 units/kg/day to 0.8 units/kg/day, there is no higher incidence of hypoglycemia with higher range insulin doses (greater than 0.5 units/kg/day), aside from in children younger than six years of age. Thus, higher doses are likely to be safe for both inpatient and outpatient management. Our results suggest exhibiting caution when prescribing insulin for children younger than six years of age, because this group was at increased risk for hypoglycemic events. While we identified no short-term value with respect to improved HbA1c levels in initiating treatment at a higher starting TDD, the improved long-term outcomes of strict glycemic control remain an important consideration.
The authors thank Ms Claire Gougeon and Ms Aditi Amin for help with the conduct of the study, and Mrs Fiona Hendry for help with the preparation of the manuscript. All three authors had full access to all of the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis. This project was not sponsored or supported by any external agency. The authors have no conflicts of interest to disclose.