Despite the increasing numbers of persons with diabetes utilizing wearable diabetes technology, patients wearing an insulin pump remains a low-volume high-risk event in the hospital.
Type 1 diabetes (T1D) is increasing in both adults and children. The rate of T1D has grown 3% annually since 2000 according to the T1D Index. Currently, 1 out of every 30 families in the United States are impacted by T1D.1 Post-COVID studies suggest an increased risk for newly diagnosed diabetes, both type 1 and type 2, after infection.2,3 Thus, there may be more people living with T1D in the future than currently projected.
Persons with T1D are increasingly choosing wearable diabetes technology to manage their blood glucose. Data from the T1D Exchange Clinic Registry reveals more than 60% of individuals with T1D in the United States utilize an insulin pump.4 The complexity of insulin pump systems has increased over the past 6 years, since the first insulin pump to operate in automated insulin delivery (AID) mode became commercially available. AID is a hybrid closed-loop system where continuous glucose monitor (CGM) values are wirelessly transmitted to the insulin pump, driving an algorithm to make automatic insulin dosing adjustments. Persons with T1D that utilize insulin pump systems have improved glycemic and patient-reported outcomes.4,5
As of February 2022, all 3 manufacturers of insulin pumps sold in the United States have US Food and Drug Administration (FDA) approved devices that operate in AID mode. These systems are currently only FDA approved for outpatient management. However, professional societies, including the American Diabetes Association, the Endocrine Society, the Diabetes Technology Society, and the Association of Diabetes Care and Education Specialists, recommend the continuation of insulin pump systems during hospitalization for appropriate patients with proper supervision.6-9 Despite the increasing numbers of persons with diabetes utilizing wearable diabetes technology, patients wearing an insulin pump remains a low-volume high-risk event in the hospital.
In June 2021, the Joint Commission (TJC) published “Quick Safety 59: Safe Patient Use of Insulin Pumps and CGM Devices During Hospitalization.”10 This safety brief addressed 7 areas of practice that health care organizations should implement: (1) policies and procedures; (2) home medication usage; (3) medication self-administration; (4) orders; (5) medical record documentation; (6) education of staff, patient, and family; and (7) risk assessment. A specific recommendation stated that hospitals should implement a process to validate the accuracy of a patient’s personal glucose testing device against the hospital-approved glucose meter.
National guidelines and consensus statements on the continuation of diabetes technology in the hospital contain no recommendations on the appropriate interval for validating CGM accuracy for patients who continue use with an insulin pump during hospitalization.
Currently, only 1 of the 3 CGMs that integrate with insulin pumps sold in the United States requires routine calibration in the ambulatory setting. National guidelines and consensus statements on the continuation of diabetes technology in the hospital contain no recommendations on the appropriate interval for validating CGM accuracy for patients who continue use with an insulin pump during hospitalization.6-8 Establishing best practice for CGM validation becomes increasingly important as the numbers of patients utilizing insulin pump systems grows and patients express their preference for continuing AID systems during hospitalization.
The practice site conducted a gap analysis and convened an interprofessional workgroup after review of the TJC safety brief. The group determined that for patients continuing an insulin pump during hospitalization, CGM validation should occur at admission and once every 24 hours or per the manufacturer’s instructions for use, whichever was more frequent. Due to the lack of regulatory approval for CGM use in the hospital, glucose monitoring with point-of-care (POC) testing continued to be required with all insulin dosing decisions, defined as with meals and at bedtime (ACHS). To facilitate nursing documentation of CGM validation, flowsheet rows in the EHR built for a research project (NCT04714216) were modified to autopopulate the POC section of the vital-signs flowsheet when a CGM device was documented on the Avatar within the EHR. Between June 2021 and January 2022, the new CGM validation recommendation was incorporated into the practice site’s clinical practice guideline (CPG), computer-based learning module, and nursing documentation. In May 2022, CGM validation occurred 17% of the time per CPG, indicating an opportunity for improvement.
This quality improvement project aimed to improve nursing documentation of CGM validation for patients continuing their insulin pump during hospitalization.
The use of clinical decision support (CDS) improves quality and safety of care.11 CDS is defined as providing clinicians with knowledge and person-specific information, intelligently filtered or presented at appropriate times within the EHR, to enhance care.11 Evidence shows that CDS embedded within the EHR that provides actionable information can improve nursing process outcomes.12-16 Thus, a redesign of the nursing documentation for CGM validation was undertaken.
This quality improvement project aimed to improve nursing documentation of CGM validation for patients continuing their insulin pump during hospitalization. Actionable CDS embedded within the EHR was integrated into standard nursing workflow. The results of this quality improvement project are being reported utilizing the SQUIRE 2.0 guidelines.17
The practice site is an urban, 671-bed academic medical center. In 2022, 159 adult and pediatric patient encounters continued an insulin pump during hospitalization, and 103 utilized AID mode. Admissions occurred across 26 patient care areas.
The Model for Improvement – Plan Do Study Act (PDSA) framework was utilized to address the gap between institutional CPG and clinical practice. The PLAN for the test of change was to transform the CGM validation documentation from a multistep process, which relied on nursing staff to remember to add the CGM to the Avatar so that the CGM validation rows populated the flowsheet and then navigate to the vital signs flowsheet to document the CGM value and POC glucose value into the same time stamp, into one interactive nursing task. The new interactive nursing task is displayed in Figure 1.
DO in the PDSA cycle represents implementation. The new interactive nursing task went live in June 2022. The task populates the nursing workspace within the EHR when the insulin pump order set is initiated.
The STUDY phase was conducted from June 2022 through December 2022. To determine if the project outcomes could be contributed to the intervention, no other changes to practice regarding insulin pump patients admitted to the hospital were made during this period. Education regarding the new nursing task was disseminated via normal organizational methods used to communicate changes to the EHR and CPGs.
ACT is the final phase of the PDSA cycle, during which results are reviewed and recommendations for improving or sustaining the change are determined.
Due to the medical complexity of the patient population, which can result in multiple admissions for an individual, patient encounters were utilized as opposed to unique patients. To obtain compliance rates of CGM validation, manual chart audits were conducted for 6 months before and after implementation. The rate of CGM validation was determined based on whether the CGM validation occurred per CPG. Admission was defined as the time the patient arrived at an inpatient unit until the first 7 AM, and daily was defined as once every 24 hours occurring between 7 AM and 7 AM.
The discrepancy between the rapid advancement in diabetes technology and professional societies’ recommendations compared to the current state of regulatory approval during inpatient admission creates ethical dilemmas for clinicians trying to navigate doing what is right for the patient and doing what is required. The regulatory need to perform POC tests for all insulin-dosing decisions in the hospital can create confusion for patients because this is not required at home, uncertainty for clinicians, and conflict between patients and clinicians. Patients can exercise their autonomous right to decline POC testing. Diabetes clinicians, understanding the limitations of CGM technology in the acute care setting, are held to the principle of nonmaleficence, which dictates that they ensure no harm. Ultimately, clinicians must balance the benefits, risks, and costs to come to the best decision for each individual patient.
In an analysis of the ethical issues associated with AID systems in 2019 by Quintal et al,18 5 domains of ethically problematic situations were identified. These include (1) confidentiality and safety, (2) coverage, (3) patient selection, (4) patient coaching and support, and (5) personal identity and agency. When evaluating ethical dilemmas associated with safety, the authors discuss information transmission, data access, and complexity but do not include the potential conflict between patient autonomy and patient safety in the acute care setting. At the practice site, when a patient asserts their autonomy and declines POC testing, every effort is made by the clinicians to reach a consensus with the patient, but ultimately, patient autonomy is respected, and the patient’s desire to refrain from fingerstick glucose checks is detailed in the documentation.
In the preimplementation phase, CGM validation occurred 25% of the time and included 53 patient encounters. In the postimplementation phase, CGM validation occurred 40% of the time, thus a 60% improvement.
This project was submitted to the institutional review board of the practice site and was classified as a quality-improvement project without the need for institutional review board oversight.
There were 99 patient encounters that continued an insulin pump during hospitalization in the postimplementation phase. Of those, 8 patient encounters were excluded because the patient was not wearing a CGM with the insulin pump, and 4 patient encounters were excluded because the insulin pump was discontinued shortly after admission or initiated right before discharge. This resulted in 87 patient encounters for postimplementation analysis.
In the preimplementation phase, CGM validation occurred 25% of the time and included 53 patient encounters. In the postimplementation phase, CGM validation occurred 40% of the time, thus a 60% improvement. Pre- and postimplementation data are shown in Figure 2.
Results of this project indicate that CDS targeted at nursing processes can improve compliance with CPGs. This is particularly important for low-volume, high-risk events in which relying on clinicians to remember tasks is likely not adequate to ensure safe care. However, many factors within EHRs must be considered when designing CDS. Contextual elements, such as different modules within an EHR for different practice settings and nursing preference for different documentation workflows, can impact the effectiveness of CDS.
Although the rate of CGM validation improved, this initial PDSA cycle did not achieve 100% compliance with the organizational CPG. As shown in Figure 3, the largest number of insulin pump patients were admitted to the obstetrics unit, which has a different nursing template that does not utilize the same workspace as the standard RN template. This represents a target for a subsequent PDSA cycle, tailoring the interactive nursing task to fit the obstetrics EHR template. A second opportunity for a targeted PDSA cycle exists around patient admissions; as shown in Figure 4, the rate of validation at admission was lower than the rate of validation daily.
Limitations of this project include a low volume of patients spread across a large number of patient care areas. Staffing destabilization during the COVID-19 pandemic resulted in large numbers of travel nursing staff within the organization. Education specific to this EHR change was disseminated in June for all staff and during orientation for new staff. Just-in-time education with individual staff on the importance of validation of CGM for this low-volume, high-risk population could have increased compliance with CPG.
As the prevalence of T1D increases, wearable diabetes technology becomes more common, and patients express preference for continuing self-management during hospitalization, Diabetes Care and Education Specialists are integral to the safe integration of diabetes technology into hospital-based care through providing staff education and patient support and by ensuring policies are updated to reflect current practice.
As the prevalence of T1D increases, wearable diabetes technology becomes more common, and patients express preference for continuing self-management during hospitalization, Diabetes Care and Education Specialists are integral to the safe integration of diabetes technology into hospital-based care through providing staff education and patient support and by ensuring policies are updated to reflect current practice. TJC’s10 “Quick Safety 59” brief provides clear recommendations for processes that support safe utilization of diabetes technology in the hospital. As more patients embrace insulin pump systems that operate in AID mode, validation of CGM compared to the hospital-approved glucose meter supports patient self-management by ensuring the physiologic conditions encountered during acute illness are not interfering with CGM accuracy, which might require an alteration in the plan of care.
Kimberly D. Miller, DNP, APRN, AGCNS-BC, CDCES, BC-ADM; Clareen Wiencek, PhD, APRN, ACNP-BC, FAAN; and Richard Ridge, PhD, MBA, RN, NEA-BC, are with the University of Virginia School of Nursing in Charlottesville, VA.
This work is based on the Doctor of Nursing Practice Scholarly Practice Project of Kimberly D. Miller. I would like to thank Laurie Brock, nurse informaticist with UVA Health for her mentorship and facilitation of the implementation of nursing clinical decision support. I would also like to thank my advisor, Dr Clareen Wiencek, and the faculty of the Doctor of Nursing Practice Program at the University of Virginia.
The authors declare having no professional or financial association or interest in an entity, product, or service related to the content or development of this article.
The authors declare having received no specific grant from a funding agency in the public, commercial, or not-for-profit sectors related to the content or development of this article.
Kimberly D. Miller https://orcid.org/0009-0000-4847-5716
