A Clinical Training Program for Hybrid Closed Loop Therapy in a Pediatric Diabetes Clinic

Abstract

Background:

Hybrid closed loop (HCL) therapy is now available in clinical practice for treatment of type 1 diabetes; however, there is limited research on how to educate patients on this new therapy. The purpose of this quality improvement project was to optimize a HCL education program for pediatric patients with type 1 diabetes (T1D).

Methods:

Our multidisciplinary team developed a novel HCL clinical training program for current insulin pump users, using a quality improvement process called the Plan-Do-Study-Act model. Seventy-two patients participated in the HCL training program, which included (1) an in-person group class to reinforce conventional insulin pump and CGM use on the new system, (2) a live video conference class to teach HCL use, and (3) three follow-up phone calls in the first 4 weeks after HCL training to assess system use, make insulin adjustments, and provide targeted reeducation. Diabetes educators collected data during follow-up calls, and patients completed a training satisfaction survey.

Results:

The quality improvement process resulted in a training program that emphasized education on HCL exits, CGM use, and optimizing insulin to carbohydrate ratio settings. Patients successfully sustained time in HCL in the initial weeks of use and rated the trainings and follow-up calls highly.

Conclusions:

Ongoing educational support is vital in the early weeks of HCL use. This quality improvement project is the first to examine strategies for implementation of HCL therapy into a large pediatric diabetes center, and may inform best practices for implementation of new diabetes technologies into other diabetes clinics.

Achieving the hemoglobin A1c (HbA1c) goal of <7.5% recommended by the American Diabetes Association for pediatric patients with type 1 diabetes (T1D) is a challenge, with only one-fifth of youth meeting this target.¹ Throughout the last several decades, there have been vast technological advancements in diabetes care, including insulin pump therapy, continuous glucose monitoring (CGM), and, most recently, automated insulin delivery technologies (also referred to as artificial pancreas or closed loop technology). The goal of automated insulin delivery is to reduce the burden of diabetes management and improve glycemic control by incorporating algorithms into insulin pumps that automatically increase and decrease insulin delivery in real-time based on CGM sensor glucose trends.²

Treatment of T1D entered a new era with the FDA approval of the first automated insulin delivery system, the Minimed 670G System (Medtronic Inc, Northridge, CA), which was commercially released in mid-2017. The Minimed 670G system includes an insulin pump, the Guardian 3 CGM, and a control algorithm that automates insulin delivery. The Guardian 3 CGM consists of a glucose sensor that measures interstitial glucose levels and a transmitter that sends the glucose information to the insulin pump every 5 minutes, which is used by the pump’s control algorithm to calculate insulin doses.^3,4 This system is called a “hybrid closed loop” (HCL) because it automates basal insulin delivery but still requires user-input to deliver bolus doses of insulin for meals or high blood glucose correction. Many studies have demonstrated the safety of HCL systems, with no reported episodes of severe hypoglycemia or diabetic ketoacidosis in outpatient clinical trials.^3,5,6 In addition, several studies have shown improvements in glucose control for patients with T1D using HCL systems including improved average glucose levels, increased percentage time with glucose values in target range (eg, 70-180 mg/dL), reduced percentage time with glucose values in the hypoglycemic range (eg, <70 mg/dL), improved overnight glucose control, and improved HbA1c values.^3,5,7,8

With the commercial launch of the 670G system, HCL therapy moved from the research setting to clinical practice for the first time. However, little research has been done in the clinical setting on how to prepare patients for HCL therapy^4,5,9,10, and there are no immediate descriptions of training programs utilized at clinical centers around the world. Therefore, the purpose of this paper is to describe a HCL clinical training program developed at a large pediatric diabetes center and lessons learned in optimizing patient training and follow-up on this new technology in clinical practice.

Methods

Our pediatric diabetes center created a specialized quality improvement (QI) team to design and evaluate a clinical program to transition current insulin pump users to HCL therapy. The team consisted of nine clinical staff (3 pediatric endocrinologists and 6 diabetes educators) and two program coordinators. The program coordinators took notes during meetings, collected and analyzed the data and generated summary reports for use during the QI process. The clinical team had extensive experience with the HCL device from clinical trials and used their experience to develop and refine the training program. The HCL training program is described below, as well as the quality improvement process used to refine the program.

Hybrid Closed Loop Clinical Training Program

The HCL training program included three phases, described below: (1) an in-person group class to transition to the HCL system and reinforce conventional insulin pump and CGM therapy, (2) a live videoconference group training class to start HCL mode, and (3) three follow-up phone calls in the first 4 weeks of HCL use to provide additional support (Figure 1). The HCL program was the first at our center to train patients and families on HCL use, and coincided with the commercial roll-out of the new device.

Figure 1.

Phases of hybrid closed loop training program.

Pediatric patients and their families attended the in-person, introductory group class after receiving the HCL system from the manufacturer, which included an insulin pump and CGM. The HCL system can operate in either open loop mode (standard sensor augmented pump therapy) or HCL mode. The purpose of this class was to initiate traditional open loop insulin pump therapy using the new device, but not to start HCL mode. One of the diabetes educators on the QI team (referred to as “educators”) taught the class, which lasted 2-3 hours. Patients and families started the class by navigating through all pump menus to become familiar with the new pump. Then, patients programmed the new pump with their current insulin dose settings, and paired the blood glucose meter and CGM transmitter to the pump (both devices pair to the pump via radio frequency signal). Next, patients were instructed on the insertion of the CGM sensor and steps to initialize the CGM sensor session on the pump. Educators emphasized the importance of self-management behaviors necessary for success with HCL therapy, such as routine meal bolusing, and best practices for sensor calibration (ie, before meals, and at bedtime), in preparation for starting HCL at the next training session. A physician reviewed glucose trends and made adjustments to pump settings as needed. Finally, the educator introduced the basic concepts of HCL therapy, with emphasis on the difference in insulin delivery between standard pump mode and HCL mode. Patients were instructed to continue in open loop for 5-7 days before completing the next training session and transitioning to HCL use.

Approximately 1 week after the in-person class, patients completed a second training class, consisting of up to four families, to transition from standard open loop mode to HCL. This class was conducted remotely using a live, web-based videoconferencing software, Vidyo (Hackensack, NJ). Vidyo is a HIPAA compliant, web-based software that patients can install and use from a computer, tablet, or smartphone with a speaker and a camera. Patients and families completed this session from home or any location of their choosing. The educator used a training slide deck to facilitate the learning session. The slide deck was visible to patients through Vidyo’s screen sharing feature. The following topics were included in the HCL training session: (1) basic concepts of HCL therapy (ie, how the system calculates basal insulin delivery) and benefits of HCL therapy (ie, reduction in hypoglycemia and increased time in target range), (2) steps to activate HCL, (3) how to manage diabetes while in HCL mode (including how to bolus for meals, how to treat hypoglycemia, and how to give high BG corrections), (4) an explanation of the different system alerts and how to respond to system alerts, and (5) the reasons for system exits from HCL and how to troubleshoot system exits from HCL. The HCL system exits from HCL mode (“Auto Mode”) to standard pump mode (“Manual Mode) for prolonged hyperglycemia (i.e. sensor glucose > 250 mg/dl for 3 hours or >300 mg/dl for 1 hour). Additionally, the system may revert to a “Safe Basal” mode if the pump has been delivering a minimum or maximum insulin rate for 1-3 hours, or if sensor glucose data is missing or deemed inaccurate by the system. After 90 minutes of Safe Basal, the system will exit the user from HCL to standard pump mode. In many of these situations, the user will receive a ‘BG required’ alert and can enter a BG value into the pump to return to HCL.

During this session, patients activated the HCL functionality on their devices and verified successful initiation with the educator who visually confirmed the HCL “blue shield” on the screen of the insulin pump. The session ended with a question and answer period and explanation of the follow-up phase of training.

The final step in the training program was the completion of three follow-up phone calls between the educator and the patient/family in the first 4 weeks following HCL training (week 1, week 2, and week 4). The purpose of these calls was to assess system use, make adjustments to pump settings as needed, and provide personalized reeducation to maximize time spent in HCL and increase sensor glucose time in target range (70-180 mg/dL). Patients uploaded their HCL system to the manufacturer-specific online device software prior to the scheduled phone call. Educators reviewed the data contained in the device software reports and provided reeducation and adjustments to pump settings to increase time in HCL mode and sensor time in target range.

Quality Improvement Process

Starting at the onset of the HCL training program, the multidisciplinary QI team met monthly for 4 months to review data and refine the HCL training program using a Plan-Do-Study-Act (PDSA) framework. This framework applies a trial-and-learning methodology to develop effective changes that will result in the improvement of a process.^11,12 The Plan phase includes developing the project objective, identifying questions and predictions (why?), creating the plan for how the first cycle will be carried out (who, what, where, when), and preparing the data collection to answer the questions. The Do phase carries out the Plan, documents problems and unexpected observations, and begins analysis of the data. The Study phase completes the data analysis, compares the data to original predictions, and summarizes what was learned. The Act phase identifies how to change the process based on what was learned from the test. Purposeful improvements to a complex system will usually require one or more PDSA cycles. The team completed four separate PDSA cycles in the course of the improvement period.

Data Collection

To evaluate the effectiveness of the training program, the improvement team collected data during the follow-up phase of the program (first 4 weeks following HCL initiation). The program coordinators recorded data from the patient’s pump uploads, including percentage of time using HCL, percentage of sensor glucose values in target range (70-180 mg/dl), percentage of time wearing the CGM sensor, and the top 3 reasons for system exits from HCL. In addition, the educators recorded the educational topics covered during the follow-up phone calls, any adjustments made to the insulin pump settings, and the amount of time spent on follow-up calls and charting for each call. The QI team developed a list of education topics based on experience with previous insulin pumps, CGMs, and extensive experience with the HCL during clinical trials and commercial trainings. The educators used this list to document each topic covered during follow-up encounters. Additional topics were added to the list throughout the QI process. All data were entered into REDCap, an electronic data capture tool proving an interface for validated data entry, audit trails, automated data export, and procedures for importing data from external sources.¹³ The QI team reviewed the data during each monthly meeting and used the data to identify ways to refine the training program to improve outcomes.

To assess patient experience in starting HCL, patients/families completed a survey via email or at routine clinic visits, three to six months after the last follow-up phone call with the educator. In this survey, patients described the challenges they experienced with HCL therapy in free text. Patients then rated (on a scale of 1-10, 1 as the lowest rating and 10 as the highest rating) how well their initial training prepared them to use the system successfully. If rated less than a ten, patients provided suggestions on how the training program could be improved.

Results

Participant Characteristics and Program Completion

Seventy-two patients (age 14 ± 4.3 years; 46% female) with T1D (HbA1c 8.7 ±1.4%) were trained to start HCL therapy over a four-month period. All patients completed both the in–person introductory class and the videoconference class to initiate HCL. Ninety-two percent of patients completed at least one follow-up phone call, 81% of patients completed at least two follow-up phone calls, and 53% of patients completed all three follow-up phone calls.

Quality Improvement: PDSA Cycle Results

A description of the quality improvement PDSA cycles are shown in Table 1. PDSA Cycle 1 reviewed the successes and challenges with the initial implementation of the HCL training program, and resulted in limiting the number of families in the initial classroom training and clarified data points to collect. During PDSA Cycle 2, the QI group reviewed data and the changes from Cycle 1. It was noted that many individuals were exiting HCL mode due to hyperglycemia. Therefore, the team modified the follow-up education to include more emphasis on each individual’s top reasons for HCL exits and steps to take to prevent exits and reenter HCL mode after exits. PDSA Cycle 3 focused on the frequency of insulin to carbohydrate (I:C) ratio changes made during follow-up phone calls, and resulted in having physicians review insulin doses and making insulin dose adjustments prior to initiation of HCL to reduce the number of HCL exits due to hyperglycemia. Finally, PDSA Cycle 4 resulted in the development of a patient satisfaction survey to evaluate the overall value of the HCL program and gather feedback from patients on how improve the program.

Table1.

Results of the Quality Improvement Project PDSA (Plan-Do-Study-Act) Cycles Initiated Over the First Four Months of the HCL Training Program.

	Cycle 1 (month 1) N=26*	Cycle 2 (month 2) N=44*	Cycle 3 (month 3) N=53*	Cycle 4 (month 4) N=72*
PLAN ↓	• Plan data collection methods, discuss current successes and challenges	• Review data collected during cycle 1	• Review data collected during cycle 2 • Review common reeducation/topics during follow-up calls	• Review data collected in cycle 3
DO ↓	• Record glucose data and data on patient use of HCL from pump downloads during follow-up calls • Monitor topics discussed during follow-up call • Observe phase 1 in-person class timing and efficiency	• Record reasons for HCL exit • Track reeducation topics covered during follow-up calls	• Record I:C ratio changes made by educators during follow-up calls	• Develop survey for patients/families to provide feedback on training program and device use
STUDY ↓	• Evaluate data not recorded correctly, how to improve? • Group training sessions crowded, should we limit number? • Training Sessions too long, how can we consolidate information?	• Observed high frequency of “high blood glucose” HCL exit • Observed high frequency of reeducation regarding importance of consistent meal bolusing • Observed frequent reeducation on troubleshooting HCL exits	• Observed frequent I:C ratio adjustments at weeks 1, 2, and 4 follow-up calls • Consider making insulin dosing adjustments at initial training before HCL mode start	• Observed that adjusting I:C ratios proactively prior to HCL start was beneficial
ACT ↓	• Modify data collection fields to make data collection simpler (ie, create checklist of educational topics) • Have program assistant record data from pump downloads instead of educators • Limit max # of families for in-person training class to 5 families • Instruct patients to link the BG meter and transmitter to the pump before first class	• Modify training to emphasize importance of mealtime bolus with HCL • Modify training and follow-up education to emphasize common reasons for HCL exits and troubleshooting tips for HCL exits	• Physician will adjust I:C ratios and other pump settings as needed at in-person session (prior to starting HCL) to reduce frequency of hyperglycemia exits after starting HCL • Educators adjust I:C ratios at HCL video conference session as needed	• Administer patient surveys 3-6 months after completing training program

^*N refers to the cumulative # of patients participating in the HCL training program in each cycle.

Participant HCL Characteristics and Education

During the first 4 weeks after initiating HCL mode, mean sensor wear time was 83%, and time spent in HCL was 72%. Sensor time in target range (70-180 mg/dL) increased from 49% to 62% between the introductory group class (prior to HCL start) and the week 4 follow-up after HCL start (P < .001) (Table 2). Sensor wear and HCL time did not significantly change between weeks. The average number of HCL exits per week was 5.8. The top three reasons for HCL system exits were “high sensor glucose” (1.98/week), “maximum insulin delivery” (0.79/week), and “blood glucose test required to continue HCL” (0.63/week).

Table 2.

Comparison of HCL Use, Sensor Wear, and Glucose Values for the Week After the Introduction Class (Open Loop) and Weeks 1, 2, and 4 After HCL Initiation for the 72 Individuals Who Participated in the HCL Training Program.

	Open loop (week after introduction class) N=72	Week 1 post–HCL class N=66	Week 2 post–HCL class N=52	Week 4 post–HCL class N=49	P valuea
% HCL	N/A	73.6 (21.6)	74.2 (22.9)	69.5 (24.9)	.37
% sensor wear	77.7 (20.1)	84.8 (18.2)	82.2 (19.4)	80.5 (16.4)	.21
% sensor 70-180 mg/dl	49.1 (16.5)	59.3 (15.1)	61.6 (16.3)	62.2 (13.6)a	<.001
% sensor <70 mg/dl	1.8 (2.5)	1.9 (2.6)	1.5 (1.4)	1.6 (1.6)	.45

Summary values are percentage mean (SD) unless otherwise stated.

^aOpen loop compared to week 4 post HCL using t-test.

During the follow-up phone calls, educators strengthened insulin to carbohydrate ratios to increase mealtime insulin doses in 75% of patients, with average percentage change of +24% (Table 3). Forty-four percent of patients had changes to active insulin time, with 80% of changes decreasing duration. The top three diabetes self-management education topics reviewed during follow-up phone calls included counseling on premeal bolusing (65%), instruction on correction boluses for high blood glucose (48%), and responding to system alerts to prevent HCL exits (45%). Reeducation on calibration practices and hypoglycemia treatment were also frequent. On average, follow-up phone calls lasted 16 minutes (range 5-60 minutes), followed by 11 minutes (range 0-45 minutes) of charting. Time for both phone calls and charting did not significantly decrease from weeks 1 to 4.

Table 3.

Mean Insulin to Carbohydrate Ratio Dose Adjustments Made During Follow-Up Phone Calls for 72 Participants of a HCL Training Program.

	% change	# (%) individuals who made changes
Breakfast	+22%	62 (86%)
Lunch	+18%	54 (75%)
Afternoon	+25%	52 (72%)
Dinner	+22%	63 (84%)
Bedtime	+32%	49 (68%)

Patient Feedback Survey

The response rate for the patient feedback survey was 51%. Patients rated their program satisfaction as 9.3 out of 10. Suggested improvements included adding more description on HCL exits, and extending the follow-up period from 1 month to 3 months with more time in-between phone calls. The top challenges patients reported related to device use were: CGM sensor calibration issues (system not accepting the calibration, or requiring calibration too often or at inconvenient times) (32%); frequency of blood glucose checking (27%), and system alerts (too often, repeated) (20%); and general sensor frustrations (adhesive not sticking, length of warm-up, accuracy) (20%).

Discussion

The patients who completed this unique clinical training program achieved success with the HCL system, with an average time in HCL of 72%, similar to the adolescents in the pivotal trial of the 670G system who spent a median of 75% time in HCL mode.³ Two in-depth training sessions along with close clinical follow-up resulted in increased time in range (70-180 mg/dL) and targeted reeducation helped patients sustain CGM sensor wear and time in HCL mode in the initial weeks of system use. Our program included a novel, remote training session using live videoconferencing for HCL training and initiation. High reported patient satisfaction with the training coupled with success in using HCL suggests that remote training using web-based technologies may be a promising avenue for clinic staff to meet the increased demand for training with new technologies while also reducing the burden of travel to clinic for patients and their families.

Even with in-depth training and support, patients still experienced challenges in using HCL. The Plan-Do-Study-Act model helped identify ways to refine the training program and optimize patient transition to this new technology. Throughout the QI process, we learned that one of the main barriers to time in HCL mode was high sensor glucose values. Prolonged high sensor glucose values cause system exits from HCL mode to standard pump mode and was the most common reason for HCL exits for our patients. Insulin to carb (I:C) ratios were strengthened by 24%, on average during the first 4 weeks of HCL use, which is slightly higher than previous reports of adolescents needing a 10-15% increase with HCL use.¹³ Patients may benefit from more aggressive I:C ratios in HCL because the algorithm may substantially decrease basal insulin following a large bolus dose due to the insulin feedback mechanisms involved in many automated insulin delivery algorithms. The focus on analyzing I:C ratio settings during training and follow-up coupled with reeducation on the importance of consistent meal boluses for all meals and snacks is an important point of emphasis for pediatric patients and their families in any HCL training program. Finally, even though patients were taught how to troubleshoot system exits from HCL mode during the training sessions, reeducation on how to respond to system alerts to maintain time in HCL was common during follow-up calls. This suggests that close follow-up with patients in the first weeks following HCL start is very important to provide reeducation and reinforce key behaviors introduced during training.

There are limitations to consider when interpreting the results of this QI training project. First, this training program focuses on the Medtronic 670G HCL system specifically, as it is the only HCL system commercially available to date. Therefore, portions of this training program and some of the lessons learned from the QI process may not be generalizable to other HCL systems. For example, other systems may not exit users from HCL in response to hyperglycemia and strengthening I:C ratios may not be necessary with future HCL systems. Additionally, all 72 patients in our clinical sample were previous insulin pump users, and thus the training program was designed specifically for patients experienced with standard insulin pump therapy. Patients transitioning from multiple daily injections (MDI) to HCL therapy may require a different training approach than described here. For example, MDI patients may require more training on standard pump therapy than we completed in this training program before transitioning to HCL therapy. Loss to follow-up was also a barrier, with only 50% of patients completing all 3 follow-up phone calls. Educators had difficulty reaching some patients by phone and found that scheduling phone calls ahead of time increased success. One of the suggestions for improvement was to extend the follow-up period and increase the time in between the follow-up phone calls. Therefore, it’s possible many families felt they did not need three follow-up phone calls in the first month and may have benefited from a follow-up call after more time had passed (ie, 2-3 months after HCL start). Nonetheless, flexibility and clear communication with families is necessary to ensure sufficient follow-up and support. Finally, the goal of this training program was to optimize patient transition to HCL therapy and thus focused on the first four weeks of HCL use only. Future research should investigate how much clinical support is needed to help patients maintain HCL use long term.

Conclusion

As new diabetes technology comes to market, clinics must develop time and cost effective strategies for patient training. Most published studies focus on the benefits of the technology but do not provide guidance on how to implement new technologies in clinical settings. Quality improvement projects, such as the one described here, provide valuable information to help clinicians translate research findings to clinical practice, and prepare patients for success on new diabetes technologies in the real world.