Intensive management from diagnosis improves HbA1c at 12 months post-diagnosis: results from a prospective cohort study in children with newly diagnosed type 1 diabetes

Type 1 diabetes (T1D) is a demanding, life-long journey for affected people and their caregivers. Management is aimed at reducing long-term complications while avoiding episodes of hypoglycaemia. The landmark Diabetes Control and Complications Trial (DCCT) has shown that an intensive insulin regimen with multidisciplinary team support is the most effective way of achieving this for both microvascular and macrovascular complications of T1D.¹ As set out in the International Society for Paediatric and Adolescent Diabetes (ISPAD) clinical practice guidelines from 2022, the target HbA^1c for young people with diabetes is <53 mmol/mol (<7.0%).² However, two recent Australasian studies have demonstrated widespread and persistent sub-optimal glycaemic control, with only 27% of children and 12.3% of adolescents achieving the recommended HbA_1c levels.^3,4

There is evidence that it is important to establish glycaemic control early and that there is a window of opportunity at the time of diagnosis to optimise this control. This is because the trajectory of patient HbA_1c typically decreases over the first 5 to 6 months post-diagnosis, and then rises to a steady state around 12–18 months.^5,6 Subsequently, an individual’s long-term HbA_1c trend rarely alters beyond 5 years post-diagnosis.⁷ It is possible, therefore, that intensive management in the first 6 months following diagnosis could have a long-term impact on glycaemic outcomes.

Despite this finding, and the well-established efficacy of intensive management in T1D, there is still some variation in the approach from diagnosis. A survey of 100 clinicians based in Australia (69%) and Aotearoa New Zealand, which examined current clinical practice with regard to insulin regimen for children newly diagnosed with T1D, demonstrated a lack of consensus regarding starting regimen and dosing.⁸ It was found that the implementation of an intensive regimen from diagnosis was less commonly opted for in Aotearoa New Zealand.

In July 2018, Christchurch Hospital implemented a protocol for intensive management of newly diagnosed T1D children <16 years. Prior to this no protocol existed, which resulted in differing approaches for insulin regimen, and the majority of those newly diagnosed were discharged on a twice daily insulin regimen. This protocol included carbohydrate counting from first meal, limiting snacks, flexible insulin dosing from first meal on subcutaneous insulin, avoidance of twice daily insulin regimen, targeted blood glucose levels from 4–8mmol/L irrespective of time of day, and corrections of blood glucose >12mmol/L while on injections between meals. The use of continuous glucose monitoring (CGM) technology was promoted. This study aimed to report the impact of the intensive management protocol at diagnosis on glycaemic control and management strategies 12 months after diagnosis compared to a historical cohort.

Methods

All newly diagnosed patients living in the Christchurch and West Coast Districts aged under 16 years old with T1D were treated in accordance with the new intensive-management protocol from 1 July 2018. T1D was defined as per the International Society of Pediatric and Adolescent Diabetes consensus statement.⁹ The protocol consisted of a) carbohydrate counting from the first meal post-diagnosis, b) flexible subcutaneous insulin dosing from first meal, c) multi-daily injection (MDI) regimen d) targeted blood glucose levels from 4–8mmol/L irrespective of time of day, and e) corrections of blood glucose >12mmol/L (see Appendix 1).

Prior to and immediately after implementing the protocol, in-services for the ward staff were regularly given at nursing handovers to ensure the change in management was widely known and understood. This was to ensure that all healthcare professionals involved with newly diagnosed children and adolescents with diabetes delivered a consistent message.

This retrospective analysis from a prospectively recorded database analysed the first 70 consecutive young people diagnosed post-protocol implementation (enrolled from 1 July 2018 to 8 November 2020) and compared their HbA_1c at 12 months with the equivalent data from the last 70 consecutive people diagnosed immediately prior to the introduction of the new intensive-management protocol (diagnosed between 21 December 2015 and June 30 2018). Data that were collected for both groups included demographic data (age at diagnosis and prioritised ethnicity), presence and severity of diabetic ketoacidosis (DKA) at diagnosis, initial hospital stay duration, insulin regimen at 12 months post-diagnosis, use of CGM at 12 months and HbA_1c at 12 months post-diagnosis.

The MDI regimen was made up of a regular daily basal dose of long-acting insulin (glargine) together with multiple daily injections of rapid-acting insulin (insulin aspart) calculated from the patient’s carbohydrate intake. The starting basal long-acting insulin dose was calculated using a starting value of 0.5x0.75–1.0U/kg. The rapid-acting doses were calculated in accordance with the practice of “carbohydrate counting”, which combines standard carbohydrate to insulin ratios (CHO ratios) (calculated initially using the “500 rule”, i.e., divide 500 by the total daily insulin dose to find the amount of carbohydrates in grams that 1 unit of rapid-acting insulin will cover). The CHO ratio calculation was adjusted for toddlers (<5 years) using 250 as the numerator rather than 500.^10–12 Insulin sensitivity factor (ISF) was defined using the 100 rule (divide 100 by the total daily insulin dose [0.75–1.0U/kg]). All calculations were reviewed daily by the inpatient care team, and then daily following discharge until glucose stability as per discretion of the diabetes educators.

In order for dosing ratios and carbohydrate counting to be implemented for inpatients, a new fully carbohydrate-counted menu was developed through the in-house catering service at Christchurch Hospital. This menu featured standard hospital breakfast and hot dinner options, while lunch was modelled on a “lunch box” with options for sandwiches, fruit, yoghurt and snacks. Between meals, snacks were reduced from 3 times daily to 2 times daily, eliminating a supper snack. Morning and afternoon tea snacks were further limited to <15g carbohydrates. For all food provided, the total amount of carbohydrates was calculated and declared on the menu to allow families to accurately begin carbohydrate counting. An important consideration was to advise families against providing extra food between these times.

Education took place over the inpatient admission, with a number of modules delivered by diabetes nurse educators and dieticians. As well as education regarding insulin administration, carbohydrate counting and carbohydrate-free foods, these sessions promoted CGM and the benefits of its being initiated prior to discharge. Because CGM is not funded in Aotearoa New Zealand, social workers were closely involved and an application for the Child Disability Allowance was completed for each patient, which partially offset the cost of accessing CGM.

Following discharge, apart from daily phone contact, newly diagnosed families were seen in-clinic 2 weeks after diagnosis, and again 1 month later before entering into the routine 3-monthly follow-up.

In comparison, prior to the implementation of the protocol, there were inconsistent approaches at diagnosis. Specifically, insulin regimen was chosen ad hoc, there was no carbohydrate counting, hospital-provided meals were not carbohydrate counted, messaging on glucose targets was variable and promotion of CGM was inconsistent. However, all other educational modules were unchanged, as was the follow-up frequency after discharge.

The audit activity of this study was covered by “Clinical benchmarking utilising data from New Zealand Diabetes Centre Patient Management Systems”; Ethics Committee reference number HD18/098. Patient data are collected under a waiver of consent. Data collection was supported from a research grant provided by the Canterbury Medical Research Foundation.

A total of 70 in each cohort provided over 80% power to detect a moderate effect size of 0.5 with a two-sided alpha of 0.05. Assuming a HbA_1c standard deviation of 20mmol/mol³ R, this would be a difference in HbA_1c of 10 mmol/mol between the two cohorts. Cohort characteristics were summarised by treatment group as counts (percentages) for categorical variables and as means and standard deviations (SD) or medians and interquartile ranges (IQR) for continuous normally or skewed variables respectively. Differences between groups were initially assessed using unadjusted tests (Student unpaired t-Test for HbA_1c and Pearson’s Chi-squared test for categorical outcomes). Next, linear regression was used to estimate group differences in HbA_1c while firstly adjusting for the potential baseline confounders of non-European ethnicity and DKA at diagnosis, and secondly investigating the relative importance of use of CGM and insulin regimen at 12 months as effect modifiers.

Results

Table 1 describes the demographics of the two consecutive cohorts. The two cohorts were similarly matched, except for the post-intervention group being slightly older, a higher proportion having Māori ethnicity and a higher proportion presenting with DKA. All of the second cohort were educated with the intensive-management protocol, without exception.

View Table 1–2.

Table 2 shows the 12-month data post-diagnosis of the two cohorts; the post-intensive-management cohort had an improved mean HbA_1c of 58.2±15.3mmol/L at 12 months, compared to 63.7±10.7mmol/mol in the historical group (p=0.014). As expected, there were notable differences in management modalities at 12 months post-diagnosis between the two cohorts, with near elimination of the twice daily insulin regimen—this being replaced by multi-daily injections—and an increased uptake of CGM (75% in cohort 2 vs 57% in cohort 1).

In order to assess which variables contributed to this improvement in HbA_1c, further analyses were undertaken. Firstly, we adjusted for baseline characteristics (non-European ethnicity and DKA at diagnoses were assumed as a predictor for higher HbA_1c). This showed the mean (95% confidence interval [CI]) difference between the two groups was now 7.3mmol/mol (95% CI 3.2–11, p<0.001), favouring the second cohort, which suggests that the changes in management more than overcame the predictive association of a poorer HbA_1c by ethnicity and DKA at diagnosis. We then adjusted for CGM use. This showed the mean (95% CI) difference between cohorts was 5.6 mmol/L (95% CI 1.5–9.6, p=0.007), favouring the second cohort, suggesting that the increased proportion of CGM use in cohort 2 had some impact on the overall difference, but did not explain all the difference. Similarly, adjusting for insulin regimen, the mean (95% CI) difference was 6.6 mmol/mol (95% CI 1.5–12, p=0.012) favouring cohort 2. These sequential analyses controlling for variables expected to predict outcome suggest the differences observed between two cohorts is multifactorial and not principally explained by either increased CGM use or insulin regime alone.

Discussion

Implementation of an intensive-management protocol from diagnosis in the management of children with T1D resulted in improved HbA_1c levels. This finding is not unexpected, with similar results observed at the John Hunter Children’s Hospital in Australia.¹³ Further, international best-practice guidelines endorse intensive management from diagnosis.² Our experience demonstrates that translating this evidence-based approach is possible and effective.

Central to the change in practice was consistent messaging for the families coming from the whole team of healthcare professionals. Previous research has highlighted this is an important factor in influencing the success of management for adolescents.^14–16 For example, Swift et al. showed that adolescents tend to achieve lower HbA_1c targets at centres where there is a greater degree of agreement between health professionals in regard to these targets.¹³ With the implementation of this protocol, a concerted effort was made to ensure we had a coordinated multidisciplinary team. Multiple ward in-services occurred both before and after implementation of the protocol in order to embed the change in practice. Consistent education was a key element of this messaging and, under this protocol, took place over the patients’ initial admission at diagnosis. This inpatient model of education and information dissemination capitalises on the opportunity presented while patients and whānau (family) are present, engaged and have time to take information on board. Families were provided information about glucose targets, insulin dosing (insulin action where rapid-acting insulin is calculated according to carbohydrate intake and correction factors, importance of 15 minute pre-bolus), carbohydrate counting and the practicalities of administering insulin (for example, injection technique) prior to discharge. The multidisciplinary team (diabetes educator, endocrinologist and dietitian) are available at all subsequent outpatient appointments, and therefore there exists a system for ongoing education and reinforcement of management goals.

It is important to note that diabetes management is currently going through rapid evolution. With higher use of real-time CGM,¹⁵ and with modern automated insulin delivery systems becoming increasingly prevalent within 12 months of diagnosis, there is great potential for even further improvements to be seen in the future. For example, Prahalad et al. showed that CGM from diagnosis results in sustained improved HbA_1c.¹⁷ While our cohort had quite high rates of CGM use, most were using intermittently scanned rather than real-time CGM due to cost. It remains important for diabetes clinics in Aotearoa New Zealand to prepare for future improved access to these technologies and rapidly translate this to routine care as soon as possible in the patient journey from diagnosis.

Limitations of this study were that it was an audit, with a retrospective control arm, as opposed to a randomised control trial comparing the two interventions. Thus, is it possible that there are additional factors interacting to influence the results, especially as we were not able to delineate between real-time and intermittently scanned CGM, or the proportion in either cohort that had CGM applied at initial diagnoses (but can safely be assumed was higher in the second cohort) or total daily dose. The improved HbA_1c demonstrated is likely to reflect multiple factors, broadly reflecting adjusted education, insulin injection regimen, improving diabetes technology and consistent messaging from the clinical service. The second cohort was affected by COVID-19 lockdowns and some clinic appointments were made by video or telephone, and the impact of this has not been analysed. It should be highlighted that the cohort in the study was predominantly European, and results may not be generalisable to centres with different ethnic makeups. A strength of this study was the ability to collect a full dataset for each patient involved in the study thanks to the routine collection of data at admission and follow-up of patients who are diagnosed with T1D in Christchurch.

In conclusion, this study provides evidence to support the efficacy of intensive management from diagnosis for children with T1D and could be used as a model for other centres in Aotearoa New Zealand who are yet to deploy this evidence-based practice.

View Appendix.