Improving the efficiency of virtual insulin teaching for patients admitted to hospital through the COVID-19 pandemic: a quality improvement initiative

Problem description

Hyperglycaemia is extremely common in patients admitted to hospital.1 In a study of over 2000 adult patients admitted to a community hospital, hyperglycaemia occurred in 38% of patients, and just under one-third had no history of diabetes.2 Hyperglycaemia has also been associated with a higher risk of in-hospital mortality and longer hospital stays.2

Most clinical practice guidelines recommend insulin as first-line treatment for the management of inpatient hyperglycaemia.3 For those newly initiated on insulin or requiring a switch to a different insulin regimen (eg, basal insulin to multiple daily injections), related teaching is necessary. In hospitals and community settings, certified diabetes educators (CDEs) play an essential role in insulin education. CDEs are skilled in not only teaching patients the mechanics of insulin injection, but they can educate patients about risks and side effects of therapy along with diabetes-related complications.4

Prior to the COVID-19 pandemic, insulin teaching at our centre was provided through an in-person CDE at the patient’s bedside. With the first wave of the pandemic, our hospital shifted to the provision of virtual models of care wherever possible. Virtual care via phone or video conference allowed us to minimise points of contact with patients, and the potential spread of the virus.

Shifting to a virtual insulin teaching model, however, created many new challenges for our diabetes education team. Patients required assistance with the necessary technology to conduct virtual visits, hospital wards needed insulin supplies (prefilled pens and needles) on site (educators could no longer bring supplies to the bedside), and patients needed to be able to show correct insulin administration for safety (ie, ‘teach-back’).5 There were also many administrative tasks required to set up the virtual consultation, new processes with data protection and privacy, and the changing of existing systems and workflows that are currently optimised for in-person patient care.6

In this quality improvement (QI) project, we aimed to improve the efficiency of successful virtual inpatient insulin teaching by 0.5 days during the COVID-19 pandemic.

Patient and public involvement

Patients or the public were not formally involved in the design, conduct, reporting or dissemination plans for our research.

Setting

We conducted this project in two large tertiary care hospitals in Canada (University Hospital and Victoria Hospital, London Health Sciences Centre). Before the pandemic, insulin teaching and education was provided to admitted patients in person, and could be coordinated and completed at the patient’s bedside within 1–2 days of referral. Our CDEs followed a teach-back approach; following insulin teaching, patients were asked to demonstrate how much information they learnt during their session.5

In March 2020, when the first wave of the pandemic hit our region, our team followed hospital and public health recommendations to urgently institute virtual care. The provision of virtual insulin teaching allowed us to limit patient contact and promote the conservation of personal protective equipment. Multiple services within the hospital made the same transition from in-person to alternative delivery services during this time.

Transitioning to virtual care, however, required a substantial change to existing workflow that was complex and inefficient (online supplemental appendix figure 1). For example, patients who met criteria for insulin teaching had to first be contacted during their admission, in order to determine the appropriateness of video conferencing (eg, access to technology, appropriate level of cognition to participate). Web links required circulation to patients by email, which they accessed through their personal devices using hospital internet (eg, phones and iPads). Personal phone numbers also had to be collected where video conferencing was not possible.

Moreover, admitted patients had to acquire insulin pens for physical teaching. This was necessary to demonstrate the patient’s injection technique, and to show that they have sufficient understanding of the education provided.5 Prior to the pandemic, pens were delivered to the bedside by the in-person CDE. The virtual model, however, required the most responsible inpatient physician to order the necessary insulin pens on the electronic medical record system. These pens then required delivery to the ward from pharmacy in advance of the virtual education session. Demonstration of injection technique then had to be shared with the CDE by video conference, or to the bedside nurse in the case of phone visits.

Collection of data

We assessed the ‘baseline efficiency’ of our virtual model as soon as it was implemented in April 2020 and we continued to collect data until June 2020. As we pivoted back to in-person insulin teaching between June 2020 and December 2020, we also captured efficiency information about our standard in-person model.

Aim

Our primary aim was to reduce the mean time between CDE referral to successful virtual insulin teach by 0.5 days.

Root cause analysis

We used a cause-and-effect diagram to identify the root causes for inefficiencies with our baseline virtual model (online supplemental appendix figure 2). This analysis was conducted alongside key project stakeholders including endocrinologists, diabetes educators, patient care facilitators (PCFs), pharmacists and support staff.

Intervention

After reviewing our root cause analysis, stakeholders felt that major contributors to virtual inefficiencies were the need for multiple phone calls to set up a virtual teaching appointment(eg, ensuring access to smart phone or a touchscreen tablet and collection of email and phone numbers), as well as difficulty completing the teach due to patient readiness (cognitive barriers, lack of family member presence, etc). There were also notable delays in having insulin pens sent from pharmacy to the wards to be used for injection demonstration.

We decided to advance three tests of change over the study period using Plan-Do-Study-Act (PDSA) cycles. First, we worked with inpatient pharmacists and pharmacy leaders to develop a method to improve the efficiency of insulin pen delivery to the ward for teaching. This included marking the delivery as a ‘high priority’ in the electronic record so that the pen could reach the floor as quickly as possible (PDSA-1). Second, we created a new electronic order set that included mandatory fields including consult priority (ie, anticipated discharge timeline), need for a non-English language interpreter and/or family member for the teaching, insulin types to be taught and used, and any other special considerations (PDSA-2). The intent of this new order was to ensure patient readiness for virtual teaching and to minimise time spent gathering information to conduct the teach. Lastly, near the end of the project we started to mobilise PCFs to help patients set up the technology needed for virtual teaching (PDSA-3). PCFs in our centre function as a liaison between the medical team and patients and their supports. They promote collaboration, communication and assist with discharge planning.8 We anticipated that the inclusion of care facilitators would reduce the burden of administrative tasks on the CDE and help us anticipate discharge timelines.

Family of measures

We captured all measures continuously over the study period. Our main outcome measure was the mean number of days from CDE referral to safe and successful insulin teach (ie, safe insulin ‘teach-back’ demonstrated by patient). Process measures included the percentage of successful insulin pen deliveries to the ward in advance of the teach. Our balance measures included length of hospital stay, percentage of patients with successful insulin teaching and 30-day readmissions for acute diabetes complications (hyperglycaemia and hypoglycaemia).

Analysis

We used descriptive statistics (means, medians, numbers, percentages) to summarise patient characteristics over the study period. To examine the impact of our intervention on outcome measures, we compared the difference in means between the two virtual periods (preintervention and postintervention). We also used run charts to descriptively present data over time using QI Macros.

Implementation of intervention

Over the study period, just over half of the consultations were done virtually with video conferencing being used 29.3%, and phone used in 22.1%.

We executed our first test of change using our virtual platform in December 2020 (PDSA-1), streamlining of insulin pen deliveries. This was followed shortly by the launch of our updated electronic order set (PDSA-2) and involvement of PCFs (PDSA-3) (January 2021).

Outcome measures

Our tests of change improved the efficacy of virtual insulin teaching by 0.27 days on average (figure 1).

Request permission

Process measures

Although there was some month-to-month variability, there was a positive trend in the timely delivery of insulin pens to the ward (online supplemental appendix figure 3).

Balance measures

We did not identify any adverse consequences. There was no difference in the length of hospital stay of included patients (online supplemental appendix figure 4). A similar percentage had a successful insulin teaches whether virtual or in-person (figure 2). While we observed a lower percentage with a successful teach in October 2020, we feel this was likely an outlier. On review of these cases, we note that there were many patient-related issues that impacted successful teach at that time, including a lack of corrective lenses, acute confusion and dexterity issues. Readmissions for hyperglycaemia and hypoglycaemia did not change over the study period.

Request permission

Main findings

Virtual care can support patients with diabetes. Indeed previous studies have investigated its utility in outpatient insulin pump training,9 nurse case management10 and in the provision of outpatient diabetes consultations,11 including during pandemic times.12 To our knowledge, ours is the first QI project focused on improving the efficiency of virtual insulin teaching for patients admitted to hospital during the COVID-19 pandemic.

Our virtual care pathway and tests of change improved the efficiency of safe and effective virtual teaching. However, did not meet our goal of reducing teaching time by 0.5 days. We suspect that this is because of the ongoing administrative burden of virtual teaching that was difficult to address at the time of our study. We did not, for example, have access to an administrative assistant to help with scheduling. Moreover, many of our virtual tasks still required the support of inpatient care providers (eg, bedside nursing to facilitated insulin teach-back), and during the height of the pandemic, it was technically difficult to acquire more support for this. Many staff members were redeployed to other roles, and attention was focused on other pandemic-related matters (ie, covering sick time, infection control). We did find that our in-person teaching model remained the most efficient way to provide diabetes education, and since September 2021, we have fully transitioned back to this model.

Strengths and weaknesses

Throughout the height of the pandemic, we launched a rigorous QI study and engaged key stakeholders in its planning and delivery. In terms of limitations, it would have been ideal to capture more baseline data prior to the switch to virtual models in April 2020, but given we transitioned our processes very quickly, we were unable to do so. Also, given we shifted back to in-person care delivery in September 2021, we were not able to fully examine the utility of involving PCFs in the process (PDSA-3), or examine the utility of other tests of change that may have improved the efficiency of our virtual programme. For example, the engagement of bedside nurses, clinical administrators or pharmacists may have been helpful to streamline processes. Moreover, while we collaborated with patients with diabetes throughout this experience, we unfortunately did not have the bandwidth to formally survey them about their teaching experience during this busy time. We also were not able to examine if there was a difference in outcomes between the types of virtual platforms used (eg, video conferencing vs phone). We suspect using video conferencing required more administrative preparation, but may have been more helpful to the CDE directly observe and interact with patients.

Regardless, our study illustrates our clinical resilience to the pandemic and our ability to quickly pivot to and modify, virtual care under highly uncertain circumstances. We safely supported people with diabetes requiring insulin, and successfully provided virtual education in a timeline manner without unintended consequences. With the wide introduction of virtual tools in medicine and diabetes,12 our project highlights the importance of QI methods in healthcare delivery. It also provides a starting point for a virtual teaching framework that could be modified and tested in other healthcare settings.