Using technology to support diabetes care in hospital: Guidelines from the Joint British Diabetes Societies for Inpatient Care (JBDS-IP) group and Diabetes Technology Network (DTN) UK

2.1.1 Glycaemic targets in hospital

The Advanced Technologies & Treatments for Diabetes (ATTD) consensus guidance recommends a target percentage time in range (3.9–10 mmol/L) in the outpatient setting of 70%, ¹⁸ but this is reduced to 50% for those at higher risk of hypoglycaemia.

The JBDS‐IP view is that during acute illness and inpatient stays, avoiding hypoglycaemia is the priority (aim for no hypoglycaemia in hospital), with the secondary aim of avoiding significant hyperglycaemia (Figure 2). The usual JBDS‐IP glycaemic targets in hospital are 6–10 mmol/L, which apply for the acutely unwell person or 6–12 mmol/L in the elderly/frail.

FIGURE 2

Glycaemic parameters and targets for people with diabetes in hospital.

If glucose is between 4 and 6 mmol/L and CGM arrow(s) are trending down, this is indicative of looming hypoglycaemia. A small carbohydrate snack (4–8 g) can arrest the fall to hypoglycaemia. Other interventions may include adding glucose to an intravenous infusion or simply rechecking blood glucose levels sooner than planned.

2.1.2 Alarms and alerts in hospital

Within the hospital setting, alarms may be challenging, particularly when healthcare professionals are unfamiliar with the technology. Currently, alarms are not integrated within electronic health records or the nursing station, and as such, the device is primarily utilised as a self‐management tool. If there is a CGM alarm, adults in hospital should be instructed to notify nursing staff immediately (either in person or via their nurse‐call button), prompting a CBG check. In order to avoid potential delays and lack of context with call buzzers, nursing staff should be aware to prioritise CGM alarms and follow local clinical guidelines for intervention based on CBG results.

If the person is able and willing to self‐manage, alarm settings may be kept as in the outpatient setting. To avoid the burden of high alarm frequency, JBDS‐IP suggest alarms to be used as safety nets to require action, rather than efficacy.

If CGM is being used to alert the medical team, the settings can be programmed as:

• HIGH ALERT: set at 15–18 mmol/L—consider extra insulin.
• LOW ALERT: set at 4–5 mmol/L—consider treating to prevent hypoglycaemia (especially if the downward arrow is shown on the CGM).

As clinical teams gain more experience, and it becomes possible to link these systems with hospital electronic health records, this can enable deliver improvements in inpatient to on out‐of‐range glucose readings. At present, the alarms should be used to trigger a CBG reading and consideration of intervention by the ward nursing staff according to clinical advice.

2.1.3 Capillary glucose monitoring whilst the individual wears CGM

As CGM data are yet not integrated within electronic health records, and sensor inaccuracies may occur during acute illness, CGM can supplement but not replace CBG monitoring in hospitals. Thus, JBDS‐IP recommends capillary blood glucose should be checked at least twice daily for people using CGM in hospital, irrespective of whether the device needs calibration. It should be explained to the person with diabetes, that regular CBG monitoring is necessary for safety reasons, and for alerting staff to out of range results. Nursing staff should also be aware to perform additional CBG testing in case of any concerns of discrepancy with symptoms.

2.1.4 Discrepancies between CGM and CBG readings

Potential sensor inaccuracies or discrepancies between CGM and CBG readings may be observed. JBDS‐IP define an acceptable difference based on the definition for the reference standard for integrated CGM (iCGM) devices. ¹⁹ This is defined as %20/20 (i.e. if difference between readings is ±20% for CBG >5.6 mmol/L or within ±1.1 mmol/L (±20 mg/dL) if ≤5.6 mmol/L).

If discrepancies are observed, more frequent CBG monitoring is advised is for next few hours (depending on clinical need).

For CGM devices that can be calibrated (e.g. Dexcom G6®) consider calibration with point‐of‐care glucose using a blood glucose meter and use if accurate. If the discrepancy persists, sensor should be removed, and replaced.

Table 3 outlines situations in which people using CGM in hospital should be advised to check capillary blood glucose before treatment decisions (rather than using sensor glucose). Reasons for ensuring CBG checks are performed in hospital include that these are quality assured, and that point‐of‐care glucose results may be linked to safety systems and visible to the diabetes team.

2.1.5 Documentation of glucose readings in hospital

Currently, CGM data are not integrated with most hospital electronic health records, and therefore, healthcare professionals need to be mindful of how data are documented. Due to the influx of glucose data available from CGM devices, JBDS‐IP advocate a minimum standard of documentation including fasting, pre‐meal and bedtime readings. Instances of hypoglycaemia and hyperglycaemia, flagged by alarms, should also be recorded, along with the methodology employed for obtaining glucose values.

Diabetes teams receiving automated alerts from CBG monitoring should be mindful that until CGM systems are fully integrated with electronic health records, they will not receive alerts on out‐of‐range glucose values as with POCT glucose systems.

Historic CGM data can be viewed through appropriate sensor online platforms (Supplementary Material S1).

2.1.6 The person with diabetes wearing CGM in hospital

Unless incapacitated or acutely unwell (i.e. haemodynamically unstable or septic), most people using CGM in a medical or surgical ward are safe to remain on CGM if admitted to hospital. ²⁰ Figure 3 outlines the recommendations for use of CGM in hospital. Capillary blood glucose should be checked at least twice daily for people using CGM in hospital, irrespective of whether the device needs calibration.

FIGURE 3

Recommendations for use of CGM in the hospital setting.

The CGM sensor device can be left in (unless the duration of illness is longer than the manufacturer's guidance on sensor wear). Table 4 summarises use of CGM use in special situations, with Figure 4 summarising wearable technology during elective procedures. Key points include:

• Ensure that the CGM device is not inserted into area of generalised oedema or cellulitis.
• CGM glucose readings should not be used whilst on variable rate intravenous insulin infusion (VRIII)/fixed rate intravenous insulin infusion (FRIII).
• If the patient is due for a procedure or operation, ensure the sensor is on a different area of the body (contralateral side) so that it is not affected.

FIGURE 4

Wearable technology during elective procedures.

For any episodes of hypoglycaemia, the person with diabetes in hospital should inform nursing staff. Point‐of‐care capillary blood glucose should be performed to corroborate the result and treated as per local guidance.

Although there are increasing data supporting use of CGM in the intensive care unit (ITU), ²¹ , ²² , ²³ , ²⁴ concerns include lack of accuracy in critically ill patients with impaired tissue perfusion, ²⁵ hypotension, hypothermia, hypoxia, vasopressor use, and potential substance interference. ²⁶ Currently within the UK, the major barrier for incorporating CGM systems within the critical care setting is implementation. The lack of a 24/7 diabetes team and infrastructure to support in majority of UK hospitals ¹¹ remains a challenge to effective implementation within critical care.

2.1.7 Stopping and restarting CGM

As many non‐diabetes healthcare professionals may be unfamiliar with the technology devices, the person with diabetes is ideally best placed to remove the sensor (ideally with the phone app or receiver to be informed of sensor session ending). If this is not possible, the sensor may be removed and safely stored.

As some sensors have the transmitter attached, any devices removed, should be labelled, and stored in a safe place and documented. This should ensure that the relatively expensive transmitter is not discarded. Even though the implantable sensor (Eversense®) is not available in many parts of the world, including the UK, the surface transmitter can be removed and reapplied at a later date.

Similarly, for restarting CGM, the person with diabetes is ideally best placed to restart this as they will have received training and will be experienced in this process.

2.1.8 Insulin pumps/continuous subcutaneous insulin infusion therapy

Overall, there are limited randomised studies, although an observational study indicating significantly fewer episodes of severe hyperglycaemia (>16.7 mmol/L) and hypoglycaemia (<2.2 mmol/L) in people in hospital using a pump. ²⁷ No difference was observed in mean glucose between those continuing continuous subcutaneous insulin infusion therapy (CSII) and those not. ²⁷ , ²⁸ However, giving individuals autonomy to self‐manage glycaemia where possible, may also reduce the risk of insulin administration errors. People often request to continue CSII use and report high satisfaction (86%) when allowed to do so. ²⁹

On acute admission to hospital, a considered decision, involving the person with diabetes, should be made as to whether the person can continue to safely use their pump and whether they wish to do so (Figure 5). The key criteria for insulin pump continuation include:

1. The individual is medically stable, willing, and capable of self‐management.
2. The treating clinician's familiarity with the insulin pump (including ward team), appropriate hospital policies/guidance on insulin pump use and inpatient diabetes management teams to support.

FIGURE 5

Recommendations for use of insulin pumps in hospital.

If a decision is made to continue with insulin pump therapy in hospital, this should be formally documented. A full list of prerequisites for safe use and contraindications of insulin pump can be found in Tables 5 and 6. Insulin pumps are expensive and any devices removed, should be labelled, stored in a safe place and documented.

TABLE 6

Contraindications to insulin pump therapy in hospital. Other situations include where there is limited safety data to allow continued use of insulin pumps, for example during major surgery and as part of hybrid closed‐loop systems in hospital. Table adapted from Umpierrez et al. ¹⁷ and Yeh et al. ³⁶

Shortly after admission, inpatient diabetes teams and/or the outpatient diabetes service should be involved in helping with insulin adjustment and pump settings. Due to potential changes in an individual's condition during their hospital stay, the clinical team should reassess the person's ability to manage their diabetes device. If the individual is unable to do so, staff should assume responsibility and switch to alternative insulin delivery methods.

Errors in insulin pump management can potentially result in severe hypoglycaemia, hyperglycaemia or DKA that may not be caught by typical hospital safeguards, such as pharmacy review or scheduled point‐of‐care testing. Therefore, continued use of insulin pumps in the inpatient setting needs to be carefully considered. As ketosis may develop rapidly if insulin delivery from the pump is interrupted, it is essential that capillary ketone levels are checked if the CBG is elevated and corrective doses of rapidly acting insulin are given using an insulin pen. In the event of ketosis, the pump and delivery device should be checked for malfunction and the diabetes team immediately informed so that continued pump use be reconsidered and if necessary, a temporary intermittent subcutaneous insulin regimen be employed.

Specialist support is not usually available out of hours and may not be available at weekends–this lack of a 24/7 diabetes team and infrastructure to support in majority of UK hospitals remains a challenge to effective implementation within usual inpatient care. Thus, hospitals should introduce a hospital insulin pump policy, in addition to self‐management policies already in place, ³⁰ and should be provided with a detailed information sheet. An example used in the US has been published by Galindo et al. ³¹

The concern with people on insulin pump therapy is they do not usually take any long‐acting insulin so if there is any interruption to insulin delivery (e.g. if the cannula is blocked or dislodged), hyperglycaemia and then ketoacidosis can develop very quickly. Errors in insulin pump management can potentially result in severe hypo‐ or hyperglycaemia that may not be caught by typical hospital safeguards, such as pharmacy review or scheduled point‐of‐care testing.

In these situations, the problem must be identified and rectified, for example by re‐siting the cannula, changing the tubing, or starting alternative insulin such as an intravenous infusion. Technical problems with the pumps can occur; the pump manufacturing companies offer round‐the‐clock telephone support and are typically able to provide a replacement pump within 24 h if required. All individuals using pumps are advised to retain a supply of their pre‐pump insulin pens for use in an emergency situation, for example, in case of pump failure or damage (Table 7).

TABLE 7

Starting basal insulin from insulin pump therapy (e.g. people who are unable to self‐manage their pump).

2.1.9 Practical recommendations: Insulin pump in the person able to self‐manage

Unless incapacitated, most people admitted to hospital using insulin pumps, who are physically and mentally able to continue to use their pumps, are safe to remain on insulin pump therapy. However, a significant proportion of medical staff will be unfamiliar with this technology. Specialist diabetes staff are required to be available to give advice and guidance if required.

If the individual is acutely unwell, unable or unwilling to manage the pump, unable to safely demonstrate pump settings, and/or no specialist advice is immediately available, the pump should be removed and a conventional intravenous insulin infusion or a subcutaneous basal‐bolus insulin regimen started.

It may not be possible for hospitals to have all the necessary insulin pump supplies, since each device may have different components that a hospital would not typically have on formulary. Therefore, the individual or family should be advised to provide the necessary pump supplies, such as infusion sets and insulin reservoirs. If supplies are not available, then insulin pump therapy will need to be discontinued whilst in hospital.

On admission, rapid‐acting insulin used in the pump reservoir should be prescribed for use in an insulin pump (either on the parenteral section of the drug chart or on the electronic record). For electronic insulin pump order sets, hospitals should alert the medical team on how to deal with pump discontinuation.

Individuals should be made aware of differing glucose targets in hospital, as it likely differs from outpatient goals.

Insulin pumps are expensive and steps should be taken to ensure they are not lost when a patient is admitted to hospital. The pump should only be adjusted by its owner (who has received extensive training) or a member of the diabetes team in possession of the correct knowledge and skills.

Table 4 summarises use of insulin pump use in special situations.

2.1.10 Discontinuing/restarting insulin pumps

If the decision is made to discontinue the insulin pump on admission, individuals should be transitioned to a subcutaneous insulin injection regimen consisting of basal and bolus insulin, or continuous IV insulin infusion. Similarly, if an individual's medical condition changes during the hospital admission, and any concerns or a contraindication arises, insulin pump therapy should be discontinued. Examples include administered medications or analgesia, which may impair consciousness or cause confusion or development of renal injury or hepatic failure resulting in uraemia or encephalopathy.

Tables 7 and 8 summarise guidance for transferring individuals on CSII to/from subcutaneous insulin. When restarting insulin pump therapy after hospital admission with an acute illness, it is important to consider insulin needs during hospitalisation may greatly vary from pre‐admission levels and could also change after discharge.

TABLE 8

Restarting insulin pump therapy from basal IV/SC insulin. When restarting insulin pump therapy after hospital admission with acute illness, it is important to consider insulin needs during hospitalisation may greatly vary from pre‐admission levels and could also change after discharge.

2.1.11 Hypoglycaemia whilst on an insulin pump

2.2.1 In‐hospital use of hybrid closed‐loop systems

There are limited data and guidance on real‐world use and safety of hybrid closed‐loop systems in the hospital setting. Within a research setting, improved percentage time in range, and lower mean glucose levels without increased hypoglycaemia ³² , ³³ have been demonstrated in small, randomised trials with people with type 2 diabetes. There are also some case reports suggesting safe use in hospital. ³⁴ , ³⁵ Despite this evidence, the effect of hybrid closed‐loop systems on clinical outcomes, the best application of these devices, and cost‐effectiveness are yet to be determined. ³⁶ Furthermore, the safety profile for risk of hypoglycaemia/ketonaemia may be different. Finally, non‐diabetes specialists will be unfamiliar with these systems.

In someone who is well and may be in hospital for a short elective procedure or investigation, it may be appropriate to let the hybrid closed loop continue to control their glucose.

In someone who is unwell, insulin requirements can change rapidly from day to day and so JBDS‐IP recommend closed‐loop algorithms are ‘turned off’ and systems are switched to ‘manual’ mode in the hospital setting. This allows the individual with support from their diabetes team to adjust insulin pump settings, including glucose target range, insulin sensitivity factor, and basal rates.

Further challenging scenarios for hybrid close loop systems in hospital; include medications such as glucocorticoids, which may cause severe insulin resistance and uncontrolled hyperglycaemia, presenting a challenge for hybrid close loop algorithms. Other challenging scenarios include nausea and vomiting in people on hybrid close loop systems, and during periods of parental or enteral nutrition through nasogastric/percutaneous endoscopic gastrostomy (PEG) feed. In these scenarios, JBDS‐IP recommend these systems be changed to ‘manual mode’.

Individuals would need to meet the self‐management criteria as for normal continuation of their insulin pump in hospital. It is important to recognise that individuals may need adjustment in their basal rates when switched to standard insulin pump mode or basal insulin, and a review of glycaemia, pre‐admission basal rates, and number of automated insulin suspensions should be made. It is worth noting that some hybrid close loop systems, such as Omnipod 5®, may not provide specific details on basal patterns, but report the total amount of insulin given in the day. In such situations, conversion to multiple daily injections can be done through estimation from total daily insulin or weight calculations (Table 7).

For inpatients meeting the criteria to continue insulin pump and CGM therapy, continuing in closed‐loop mode is only recommended under specific guidance from the diabetes team.

After discontinuation of auto‐mode within the hybrid closed loop, the system may be used individually (as CGM only or insulin pump only). Continued use of these systems should be determined on a case‐by‐case basis, with shared decision‐making with individuals.

2.2.2 Hybrid closed loop in surgery

Overall data on the safety or maximum safe duration of hybrid close loop control during anaesthesia is limited. ³¹ Hybrid closed loops should only be used with the anaesthetist working in close collaboration with the diabetes inpatient teams, with close monitoring and a clear plan and knowledge of using these systems and what to do if glucose is out of range.