UK's Association of British Clinical Diabetologist's Diabetes Technology Network (ABCD-DTN): Best practice guide for hybrid closed-loop therapy

1.1 Why are HCL systems needed?

Type 1 diabetes mellitus (T1DM) is a lifelong condition associated with risks of multiple complications.⁴ Burden from treatment negatively impacts quality of life.⁵ In England, approximately one third of pwT1D successfully achieve the National Diabetes Audit target of HbA_1c target of 58 mmol/mol.⁶ HCL systems have become increasingly available over recent years and have demonstrated improvements in glycaemic outcomes and reductions in diabetes distress in clinical studies and real-world use, including ABCD-DTN's audit of the NHS England closed-loop pilot.^{2, 3, 7-14}

1.2 Indications for use

With the launch of the amended NICE T1DM guidelines,¹⁵ we hope many more people will be able to access HCL therapy. However, access to CSII and real-time CGM, as well as to teams with experience and skills in managing HCL therapy, may remain limiting factors. At the time of writing, NICE multiple-technology appraisal evaluating the cost-effectiveness of HCL therapy is undergoing consultation.¹⁶ Suggested indications for HCL are displayed in Table 1. HCL systems may not suitable for all people with diabetes¹⁷ (Table 2). HCL use will vary from country to country. Timing of HCL initiation after diagnosis should be considered. There is emerging evidence that starting CGM within the first year of diagnosis of T1DM may lead to sustained improved glycaemic outcomes compared with people who do not start CGM or who start CGM >3 years after diagnosis.¹⁸ Furthermore, the major impact of glucose control in the early years of diabetes on long-term outcomes is well established.¹⁹ Although evidence to support early initiation of HCL therapy is not yet available, extrapolating data from studies of early CGM initiation and early glycaemic control and long-term outcomes, HCL therapy may result in sustained improved glycaemic outcomes if initiated early. Further study in this area is warranted.

In addition to the above, there are a cohort of individuals with extremely elevated HbA_1c levels who struggle with the daily management of diabetes that may have been considered unsafe in the past and in whom conventional pump therapy may have been contra-indicated due to the risk of diabetic ketoacidosis. As demonstrated in the NHS England Hybrid Closed-Loop (HCL) Pilot, these high-risk individuals may also stand to benefit from HCL if intensive input is available from the supporting diabetes team and risk mitigation strategies are employed.

2.1 Setting the right expectations

It is vital to set realistic expectations of the work required and the outcomes that can be achieved with HCL. Some pwT1D may have misconceptions that HCL is a ‘plug in and go’ system that does not require user involvement or have unrealistic outcomes with ‘flat lines’ on CGM. It is important to discuss the timing and delivery of mealtime insulin in order to reduce postprandial hyperglycaemia (in the case of delayed or omitted insulin) or postprandial hypoglycaemia (when the correct dose mealtime insulin is delivered late and the algorithm has already intervened to reduce the postprandial hyperglycaemia). It can take several weeks of HCL use to achieve optimal performance.

Basic knowledge of insulin pump and CGM systems is an essential component of HCL therapy. The user must be able to use the pump in manual mode. Clear guidance should be offered on when to come out of closed-loop mode, for example potentially during periods of illnesses or in the hours after an infusion set failure. Furthermore, additional training is required for HCL-specific topics (see special situations; Table 8). Taking a pre-meal bolus remains important for achieving optimal glycaemia. To get the best results from a HCL system, the user still needs to bolus at least 15 min before meals if using rapid acting insulins such as Novorapid or Humalog. However, even in those who are not able to do this, or who are advised not to do so due to concomitant conditions such as gastroparesis or hyperemesis gravidarum etc, HCL can still support improved glucose levels.²⁰ In this first phase, user effort and commitment are necessary to take onboard new information, establish new routines and learn to trust the system. The timescale for this process is highly variable and depends on individual differences. In the second phase, users often report reduced burden along with improved glycaemic outcomes. Some users find it easier to break down goals and expectations into realistic targets to support their journey onto HCL. Documenting and periodically re-visiting these with the diabetes team can highlight achievements and progression as well as areas where a user needs further support to achieve therapeutic goals.

2.2 Choosing the right HCL system—what do HCPs need to know about the system?

As more HCL systems become available, there is an increasing need to understand how these systems compare and implications for their practical use. Clinicians and educators may not need to understand subtle technical and algorithm differences, but they will need to understand essential system characteristics for supporting users of HCLs. The CARES paradigm was published in 2019 as a framework for clinicians to understand HCL systems²¹ (Tables 3 and 4). The CARES paradigm as applied to systems available in the UK is summarised in Table 5. In Table S1, The CARES paradigm for HCL systems expected to become available in the UK soon is outlined. To help discussion with pwT1D or their careers, other factors that need consideration are listed in Table 6.²²

TABLE 4. CARES paradigm for advanced diabetes devices.

C: Calculate	How does the algorithm calculate insulin delivery? Which components of insulin delivery are automated (e.g. basal suspensions, basal modulation, high glucose corrections, food boluses, etc.)?
A: Adjust	How can the user adjust insulin delivery? Which parameters can be adjusted to influence insulin delivery during automation (e.g. carbohydrate ratios, insulin action time, basal rates, sensitivity factors)? Which parameters are fixed?
R: Revert	When should the user choose to revert to open loop/no automation? When will the system default to open loop/no automation?
E: Educate	What are the key education points for the advanced diabetes device (e.g. essential training, tips and tricks, best practices, etc.)? How does the user optimise time using the automated features? Where can users and clinicians find additional education?
S: Sensor/Share	What are relevant sensor characteristics for each device (e.g. calibration and therapeutic blood glucose requirements, duration of sensor wear, etc.)? • What are the system capabilities for remote monitoring and cloud-based data sharing?

• Note: Reproduced with and adapted with permission from Laurel Messer.²¹

We recommend the use of independent resources such as the ABCD DTN-UK website (https://abcd.care/dtn-education/expert-views-on-devices), that has a range of materials available to help support decision making by pwT1D or carers.^{23, 24} Supporting the user to choose a system that is best suited to their requirements is important for user satisfaction. Published data on outcomes and user satisfaction can provide valuable reassurance when choosing devices.^25-27

2.3 HCL concepts for pwT1D

For open-loop insulin pump therapy, HCPs and pwT1D are familiar with the concepts of bolus (i.e. insulin that is typically delivered for meals or as a correction) and basal insulin (i.e. insulin that is constantly delivered in the background). In HCL systems, the concepts of user-initiated boluses and algorithm-modulated insulin are more appropriate.² User-initiated boluses are instigated by the pwT1D in advance of meals or as correction doses. Algorithm-modulated insulin delivery is instigated by the HCL system in response to glucose level data. As a rough indication, about 40% of insulin is usually user-initiated bolus depending on the individual's carbohydrate intake.² A lower proportion of user-initiated insulin delivery may indicate that boluses are being missed.²

2.4 Calculate

2.5 Adjust

To support users safely, HCPs need to have a full understanding of the settings that can be adjusted on a given HCL system and their impact. For each HCL system, only specific settings will affect insulin delivery while in closed loop. These will vary depending on the system and are summarised in Table 5. For example, glucose targets can be adjusted in CamAPS FX across a range of 4.4–11 mmol/L at different times; Medtronic MiniMed™ 780G has three target glucose options (5.5, 6.1 and 6.7 mmol/L, unless temporary targets (fixed setting of 8.3 mmol/L) are enabled); while the Tandem Control-IQ system offers a target range of 6.25–8.9 mmol/L (unless sleep or exercise activities are enabled). The aggressiveness of how the system reacts to raised glucose levels can be modified by changing the active insulin time on the Medtronic MiniMed™ 780G between 2 and 8 h, on the CamAPS FX system by using either the ‘boost’ or ‘ease-off’ function and on the Tandem Control-IQ system by weakening or strengthening the correction factor. Both the Medtronic and the Tandem systems have exercise activities that raise the glucose target or target range while CamAPS Fx uses the ‘ease-off’ function. Users can adjust programmed basal profiles which will alter AID in Control-IQ, but not in other systems. Insulin to carbohydrate ratio can be altered in all systems.

2.6 Revert

Certain situations, discussed below in Section 3: Important considerations and clinic review, may require the user to come out of closed loop and use their system in manual mode or revert to injections. There are also situations where the system may ‘kick out’ the user from closed-loop mode. Reasons for system-initiated exits from closed-loop mode will differ by device (Table 5). Once automated modes are turned off, ‘open-loop’ or ‘manual’ pump use will continue. Data from the CGM will continue to be available, provided issues with the sensor did not cause the closed-loop exit. In certain systems, such as Medtronic MiniMed™ 780G, predictive low glucose suspend that stops insulin delivery if hypoglycaemia is predicted, can still be utilised in manual mode if desired. At present, users of the Tandem Control-IQ HCL system or CamAPS cannot revert to a predictive low glucose suspend system in manual mode (important to note that Tandem Control-IQ does not revert to Tandem Basal-IQ).

2.7 Educate

Clinical experience has highlighted that user education regarding HCL systems is a key factor in determining success. All manufacturers provide system-specific educational resources, including online training platforms for users (Table 5). Education needs to ensure users understand the basics of how the system works, what settings can be adjusted and how, how to manage special situations including hypoglycaemia, exercise, sick days, alcohol use and suspected infusion site failure. Users must also be empowered with contact details for clinical and technical support and signposted to additional support (e.g. peer support forums).

2.8 Sensor/Share

Differences among sensors, such as calibration requirements, customisation of alerts and duration of wear, are important for the user experience and also maintaining full functionality of a closed-loop system (Table 5). All systems now offer the ability to share glucose data to cloud-based systems remotely and offer access to designated people (i.e. ‘followers’ who could be carers or relatives, as well as clinics). Some systems also allow insulin data to be shared automatically (Table 5). This has benefits in reducing the user requirement and burden to manually upload device data for HCP advice and appointments.

2.9 Peer support

Many users find shared experiences of other pwT1D helpful. Users' own experiences can give HCL candidates a realistic impression of what it is like to live with diabetes technology. Experience sharing can take place via one-to-one support/coaching, websites, social media, videos or peer support groups. DTN-UK strongly believes that every T1DM service should have a peer support group; successful examples include that from Derby where a Facebook group (now with over 780 members) was set up by a couple of enthusiastic service users with support from clinic leads. Groups like this drive good care, and keep the service accountable and up to date. Principles of good peer support for pwT1D are outlined elsewhere.²⁸

3.1 Remote starts

The DTN-UK has produced a standard operating procedure for starting insulin pump therapy remotely, with checklists and pathways.²⁹ For remote starts involving industry partners, HCPs should carry out a pre-assessment of the pwT1D, complete consent forms, collect contact details and agree the predetermined pump and HCL settings for each individual attending the group. These should be provided to the industry representative who will be delivering the technical training. A technical checklist needs to be completed confirming that the HCP is happy for the pwT1D to proceed with onboarding of HCL (Table S2). Due to the complexity of the topics covered, users may wish to have family or friends present at initiation. This should be facilitated where possible to provide the individual with additional support.

It is important to confirm that the pwT1D feels comfortable to use the device following training. This allows any training issues or concerns to be raised which could be followed up outside of the group training. The HCP team should provide support on a regular basis (see Section 4: post-initiation follow-up) until confidence and therapy aims have been achieved. This model promotes equitable access to diabetes technologies while supporting and releasing the clinical team to provide more intensive face-to-face support for those requiring individualised support.

3.2 Moving from MDI to HCL

Existing service pathways can be used when moving from MDI to pump therapy. It is recommended that the user receives training for open-loop pump therapy and uses this until competence is achieved on that particular pump and CGM system prior to switching to HCL. The prospective HCL user needs to have these basic insulin pump and CGM skills, before entering closed loop. The person needs to be able to operate the insulin pump in open loop and understand the use of CGM technologies. Many centres suggest a few days in open loop before transitioning to HCL, although the decisions around the speed of transition and learning needs should be individualised. On average, we anticipate this may take 7–14 days.

3.3 Moving from CSII or open-loop to HCL

If users are transitioning from another pump system (pump renewal), it is recommended to use their current pump settings if glycaemic management is deemed reasonable and safe. We recommend using the DTN-UK guide to pump settings to re-calculate basal rates, insulin to carbohydrate ratios and insulin correction factors prior to starting HCL.³⁰ Users who are new to CGM may have the option to go straight into closed loop with the sensor start, if the system being used can operate based on TDD/weight alone (some systems require a few days with sensor and pump before HCL can be initiated; Medtronic MiniMed™ 780G). Some users may switch to a different CGM system in order to use the HCL they have chosen and will need time to adapt. Therefore, a period of learning and experience on open loop if using a new pump device/CGM system is valuable.

3.4 System-specific considerations

3.5 Tandem control IQ

This system does not require a run-in period: the user can switch on closed-loop mode by providing an up to date weight (kilograms) and TDD (average of the previous 7 days). This system uses the programmed basal settings as a baseline and so for this system it is particularly important to optimise the settings in open loop before moving to closed loop. We recommend re-calculating the settings based on the DTN guidance. Close support is needed in the first few weeks to optimise settings. Exercise activity can be used for >2 week period for people with high baseline HbA_1c, unstable retinopathy or fear of hypoglycaemia. Up to two sleep activity (glucose target range 6.25–6.7 mmoL/L) schedules can be pre-programmed to turn on and off at pre-specified times (e.g. one schedule for weekdays and one for weekends). In sleep activity the system will use a different target range but does not give automatic boluses.

3.6 CamAPS Fx

This system does not require a run-in period. The user can switch on the closed-loop mode by providing an up to date weight (kilogram) and TDD (average of the previous 7 days). The currently used pump settings are entered as back up but only the carbohydrate ratio is used by the bolus calculator in closed loop. The correction factor and bolus calculator target only impact if the user chooses to give additional manual corrections. For Dana pumps, the extended bolus feature must be switched on for the system to initiate closed-loop mode. The personal glucose target for this system can be set between 4.4 and 11.0 mmol/L, with a default glucose target of 5.8 mmol/L. Higher personalised glucose targets can be used for people with high baseline HbA_1c, unstable retinopathy or fear of hypoglycaemia (Table 7A). The system has ‘ease-off’ and ‘boost’ modes which modify the responsiveness of the algorithm. This is the only system that is licensed for use in pregnancy. An important factor for a person to consider is the ability to manage insulin delivery using an app or an android phone.

Examples of suggested onboarding pathways for new and upgrading users, with suggested follow-up support per HCL system, are detailed in Figures S1–S6. Each HCL system is different so the importance of HCL system-specific training groups and follow-up are important. Individuals who are starting the same HCL system may be trained together in a group, but group starts are not appropriate if individuals are starting on different systems.

4.1 Important considerations

Important considerations for using HCL systems are summarised in Table 8. Table 9 outlines when to consider discontinuation of closed loop. Three very important aspects based on our experiences are detailed further below.

4.1.2 Unexplained hyperglycaemia/avoidance of diabetic ketoacidosis

Hyperglycaemia will cause an increase in algorithm-modulated insulin delivery. If glucose levels are sustained above 15 mmoL/L for more than 2 h canula failure should be considered (Figure 2). High amounts of insulin on board in such situations may reduce further corrective insulin advice and delivery. Undetected cannula failure can lead to diabetic ketoacidosis within a matter of hours. Any episodes of unexplained hyperglycaemia should be managed by changing the infusion set, checking for ketones, considering a manual correction by pen or syringe and reverting to open loop for at least 4 h after the last pen injection (Figure S7).

4.2 Checklist for reviewing in clinic

A checklist for reviewing metrics in clinic is provided in Table S4. This may provide a useful step-by-step approach to review patients in clinic. Top tips for pwT1D are provided in Table S5.²²

5.1 Workforce requirements

As with to all T1DM services, the cornerstone of the HCL service is the multidisciplinary team (MDT) (Table 13).

5.2 Training and competence

The MDT should have formal certified training on core competencies of HCL therapy (insulin pump therapy, CGM and HCL therapy; e.g. Glooko Academy or company specific training). Each HCL system has a different user interface and adjustable parameters. The team should have the knowledge to support the individual to make the appropriate choice of system for them. Time should be allocated to upskill on core competencies and new pumps, HCL systems and CGMs as they become available. Competency frameworks for diabetes technology can be a useful tool (see Table S6).

While at least one consultant, DSN and dietitian in the wider team should be proficient in HCL therapy, the wider team (including trainees) should also have awareness of the different devices and systems. As a minimum they should be trained in data review and optimisation, the management of specific issues such as set failures, pump failures and connectivity issues. Furthermore, they should be able to sign post to other services as appropriate. In our opinion, HCL is becoming the standard of care for pwT1D; it is prudent for teams to start planning to ensure that ALL staff who manage T1DM are upskilled in anticipation of an expanding need.

It is important that the HCL service should not depend on individual team members—pathways and processes need to be robust enough to operate safely in effectively in the context of unplanned/planned staff absences. A team ethos and shared vision and consistent messaging within the MDT and wider service has been shown in studies to improve outcomes.³² Clinical services should consider out of service resources such as expertise available within clinical networks and from industry partners.

5.3 Pathway to HCL therapy

5.3.1 Choosing and Starting HCL (consensus minimum 1 h with time for questions)

Enough time should be allowed to support a person to choose the most appropriate system for them taking in to account the differences covered earlier. This can be done via virtual or face-to-face group demonstrations with the ability to provide 1:1 consultation if required. The DTN-UK HCL video is an excellent resource (Figure 3^{24, 33}). A selection of e-learning platforms, videos and pump simulators may be beneficial.