To the untrained eye, the utilization of Continuous Glucose Monitoring (CGM) in the hospital setting would appear to be an easy transition from point of care glucose testing to the use of a technology that is well established in the outpatient setting.

For those of us that attended the 2023 Virtual Hospital Diabetes Meeting (HDM) sponsored by the Diabetes Technology Society, we learned that the adoption of this technology in the acute care setting has a myriad of challenges and unanswered questions.

What is CGM and can you use it in the hospital?

CGM provides real-time glucose data that enables providers to make proactive and timely clinical decisions with regards to diabetes management. It also has the potential to limit the frequency of point of care glucose monitoring, particularly in the critical care setting when patients are on a continuous insulin infusion.

Some of the challenges to address the adoption of CGM in the hospital include:

• financial
• regulatory
• clinical goals of implementation
• staff training
• data collection and display
• patient selection
• protocol development (including alarm management, confirmatory point of care testing and calibration)

Goals of CGM Implementation

1. The predominant question here is ‘Why utilize CGM?” For some institutions the goal is to improve nursing workflow and reduce point of care tests in the intensive care unit; this was the rationale for the FDA granting emergency use authorization during Covid.
2. Patient preference is often a driving factor for inpatient CGM use as more and more patients are arriving in the hospital wearing a CGM device and they are resistant to multiple finger sticks.
3. Finally, another goal may be to monitor certain high-risk patients for hypoglycemia, particularly at night.

Who is the ideal inpatient for CGM?

Currently, not every hospitalized patient requiring insulin can be monitored using CGM.

Ideally, multiple patients could be monitored in a manner similar to telemetry, but this practice is not widely used.

As a result, many institutions have developed criteria for non-critical care use. These criteria can include:

• pre-admission use of CGM in the outpatient setting
• type 1 diabetics
• chronic kidney disease
• transplant patients
• prior history of severe hypoglycemia

The use of CGM in the perioperative period is another consideration.

What training requirements do hospitals need to consider?

There was a general consensus amongst participants at this meeting that the training of nurses to use CGM is relatively simple and generally well accepted by the nursing staff.

Adoption in the critical care units was particularly high given the improvements in workflow with CGM and a high nurse to patient ratio.

In the non-critical care environment, training and adoption is more difficult. The challenges include responding to CGM alarms, determining when confirmatory POC testing is needed, and at what CGM value hypoglycemia treatment is warranted. This last challenge is especially important as CGM readings are often less accurate at low glucose values.

Often overlooked is the need for provider training, particularly in the non-critical care environment. Hospitalists may lack the knowledge to review a prior days CGM tracing and make appropriate adjustments in insulin therapy.

CGM Protocol Development:

Of course, goals of implementation, patient selection and staff training are important aspects in protocol development but there are other factors to consider. For example

• how frequently is confirmatory POC testing performed?
• what level of discrepancy between CGM readings and POC testing is considered acceptable?
• when should additional calibration of the CGM device occur?
• are CGM values utilized to treat insulin or is confirmatory POC testing required?
• when are CGM devices removed (particularly for radiologic or surgical procedures)?
• how to respond to the trending information from CGM devices?

Data Collection and Integration with the Electronic Medical Record

Currently, there is no integration between CGM and the Electronic Medical Record (EMR). With few exceptions, nursing must enter a CGM value manually into the EMR; this practice is most common in the critical care units.

Another important consideration is what data should be integrated into the EMR. In the critical care unit perhaps just the hourly CGM readings are needed, but what about the 24 hour trend? Should that also be included?

Another perplexing thought is that in the future, will all of this data be used for CMS electronic clinical quality measures (ecqm) related to hypo and hyperglycemia?

Currently only lab and point of care glucose values are utilized, but as additional CGM data is ultimately found in the EMR, will this data be subject to scrutiny? For example, will nocturnal hypoglycemia that routinely is undetected as well as post prandial hyperglycemia be considered in the calculation of these quality measures.

What are the Financial and Regulatory Considerations?

The cost of CGM devices is certainly a consideration.

Using a device for several days in the critical care setting does result in a cost savings as the the number of POC tests is reduced. However, this saving may not be transferrable to the general inpatient setting as there will likely not be a similar reduction in POC testing.

Some institutions do not use CGM in the inpatient setting except to place a device on the patient prior to discharge.

This practice has the potential to reduce readmissions. Also, recall that FDA approval for the use of CGM devices in the inpatient setting was issued under the Corona Virus Public Health Emergency.

This emergency period terminated on May 11. Although there has been no official statement yet from the FDA regarding inpatient CGM use, many institutions may choose to refrain from continuing or expanding inpatient CGM use.

However, the consensus from participants at the DST HDM meeting was to continue CGM use as these devices are safe, they improve patient care and that their use is similar to many drugs and therapies that are often prescribed “off label.”

Is CGM reliable enough for the hospital setting?

There continues to be much deliberation about the accuracy of CGM.

FDA and ISO guidelines for CGM accuracy are outlined in the table below. These guidelines are based on comparison with confirmatory laboratory testing.

However, instead of comparison to confirmatory laboratory testing, accuracy of CGM is often compared to POC testing with a calculation of MARD or Mean Absolute Relative Difference.

MARD is calculated by averaging the absolute values of relative differences between CGM/BGM system measurement results and corresponding comparison method results. There can also be significant interpatient variability in CGM vs POC glucose values, especially at low glucose values.

Concerns about accuracy must be addressed when developing protocols for CGM and when to require confirmatory POC testing for insulin dosing decisions.

However, it was raised repeatedly that what is more important than the accuracy compared to POC testing is the continuous trending data that CGM provides.

It was continually stressed during the conference that with CGM’s ability to provide more frequent glucose values, in conjunction with trending data, can achieve better results than infrequent, albeit slightly more accurate, POC testing.

So, is the hospital ready for CGM?

In summary, one may have left the Hospital Diabetes Medicine conference with more questions than answers.

For me the takeaway was that CGM is a technology that must be embraced in the inpatient setting. It will not eliminate the need for point of care testing. At this time, a hybrid type of model with both CGM and confirmatory/adjuctive POC testing appears to be most beneficial.

Much work is needed regarding EMR integration and data collection, integration with computerized insulin dosing software, patient selection and protocol development.

Inpatient CGM use is not just a technology to be managed within the purview of endocrinology, but to be a central facet of diabetic care by nursing, hospitalists, and patients.