A Virtual Training Program for the Tandem t:slim X2 Insulin Pump: Implementation and Outcomes

Abstract

Insulin pump training has traditionally been performed in-person. The coronavirus disease 2019 (COVID-19) pandemic necessitated vast increases in the number of virtual pump trainings for Tandem t:slim X2 insulin pump starts. A customized structured pump training curriculum specifically tailored to virtual learning was deployed in early 2020, and included (1) preparation for training with use of the t:simulator app, (2) use of the teach-back method during video training, and (3) automating data uploads for follow-up. Retrospective analysis from >23,000 pump training sessions performed from January 1, 2020 to July 28, 2020 showed sensor time-in-range for up to 6 months after training was 72% (60%–81%) for virtual training versus 67% (54%–78%) for in-person training. Higher user satisfaction (4.78 ± 0.52 vs. 4.64 ± 0.68; P < 0.01) and higher user confidence (4.61 ± 0.75 vs. 4.47 ± 0.0.85; P < 0.01) were reported after the virtual sessions. Virtual pump training was well received and proved safe and effective with the new virtual training curriculum.

Introduction

A cornerstone of insulin pump therapy has been training individuals on a multitude of skills, collectively referred to as pump training. These skills range from infusion set care to delivering boluses, as well as settings optimizations. The addition of automated insulin dosing features to insulin pumps has further complicated the training paradigm, adding educational requirements such as understanding the impact of late insulin bolus timing and not overtreating hypoglycemia after suspension or attenuation of insulin delivery. ^{1 –3} Yet many individuals with diabetes need some assistance in understanding key concepts that are core to insulin pump use, such as carbohydrate ratios and correction factors. Low numeracy skills (ability to understand and use numbers in daily life) have been adversely associated with diabetes control in certain studies. ⁴ For these reasons, insulin pump training sessions have traditionally been performed by a certified diabetes specialist in-person most of the time, with regular follow-up as part of the process.

In early 2020, the coronavirus disease 2019 (COVID-19) pandemic necessitated quarantines, limiting in-person patient contact. Before this, Tandem Diabetes Care offered remote training as an option for patients transitioning from other pump systems and trained clinical staff in these protocols. Tandem refined this process, and in March 2020 implemented a structured virtual pump training program specifically designed for remote trainings for all Tandem t:slim X2 insulin pump starts (including those who were transitioning from multiple daily injections [MDI]) to include training on use of Control-IQ or Basal-IQ technology. To determine the long-term feasibility of using this newly designed virtual training curriculum, we assessed glycemic and safety outcomes, as well as user experience after pump start. Since we made significant design decisions to optimize virtual training, we hypothesized that irrespective of training method (i.e., in-person or virtual) individuals onboarding to t:slim X2 insulin pump with Control-IQ technology would continue to experience real-world glycemic and safety benefits. ⁵

Methods

Data from all insulin pump trainings recorded in the Tandem Customer Relations Management database from January 1, 2020 to July 28, 2020 were reviewed. Owing to the COVID-19 pandemic, Clinical Diabetes Specialists at Tandem Diabetes Care started performing all pump trainings virtually on March 19, 2020. Additional trainings, both in-person and virtually, were performed by health care providers, contracted offices, and independent certified Tandem pump trainers according to patient need and local pandemic-related lockdown regulations. The virtual training program was designed specifically for remote sessions and involved more than performing standard pump training over a video call. Structured training specifically tailored to remote learning included (1) practice of pump-related features while showing competency related to specific tasks with use of the t:simulator™ smartphone app, (2) standardized instruction and feedback using the teach-back method, ^{6 –8} and (3) follow-up review with the pump trainer and health care providers. Key aspects of the newly designed virtual training program, including preparation, in-session, and follow-up visits, are shown in Table 1. Duration of training was also not the same as standard pump trainings, with repeated virtual sessions as needed to ensure patient competency, satisfaction, and confidence in using their system. In cases where the trainer felt that the user was struggling to comprehend or was finding the remote training session challenging, transitioning to in-person training was recommended.

Table 1.

Summary of Specific Changes to the Tandem Pump Training Curriculum for Virtual Sessions

Pretraining	Includes creating a user account for the t:connect® web application.
	Reading from the Tandem training workbook.
	Using the t:simulator™ app to simulate tasks on the pump interface.
Training sessions	Tasks are translated to the actual insulin pump.
	Pump trainers use the teach-back method with audio and a video camera to explain, ask for recall, and assess for comprehension and proper technique.
	Training sessions for MDI users were scheduled over two sessions.
Follow-up	Trainers always conduct a follow-up call within 3–5 days.
	Tandem also released the t:connect mobile app, which uploads to the t:connect cloud where data can be reviewed in the t:connect HCP web application.

MDI, multiple daily injections.

We collected up to 6 months of post-training glycemic data to compare sensor time-in-range (70–180 mg/dL) and time in hypoglycemia (<70 mg/dL) ⁹ for patients receiving virtual versus in-person trainings. Self-reported safety and adverse events were also compared. User reported confidence (not confident at all [1] to extremely confident [5]) in using the pump and satisfaction with the training (not satisfied at all [1] to extremely satisfied [5]) were assessed with a 5-point Likert scale after pump training sessions.

Regression models were used to assess the effect of training method, age, previous therapy, trainer type, and baseline hemoglobin A1c on outcomes such as sensor time in range, sensor time in hypoglycemia, number of adverse events, user satisfaction, and user confidence. The type of regression used was selected based on the distribution of the dependent variable. Poisson regression was used for modeling adverse events. Beta regression was used to model time-in-range and time in hypoglycemia, and linear regression was used to model confidence and satisfaction. The Mann–Whitney U test was used to compare distributions. The Bonferroni method was used to adjust for multiple comparisons.

Participants consented to the use of their data for research purposes as part of their onboarding to Tandem and initiating their t:connect^® account, from which all glycemic data were obtained. No IRB approval was sought for this retrospective analysis.

This analysis did not include existing Tandem t:slim X2 pump users who performed online training before completing software feature updates to their existing Tandem pump, ¹⁰ or new to Tandem users who declined to receive training (in-person or remote).

Results

In all, 23,268 pump training sessions were performed between January 1, 2020 and July 28, 2020, with 8,984 (36%) performed virtually. From January to March 2020, most training sessions were in-person. After March, the majority were virtual due to COVID-19 lockdowns, then by June 2020 a mix of in-person and remote training sessions was observed. Contracted offices, health care providers, and contracted pump trainers performed the majority of their visits in-person (89%, 80% and 87%, respectively), whereas Tandem Clinical Diabetes Specialists performed most of their trainings virtually (70%). Those trained virtually were more likely to be previous pump users (59%) than those trained in-person (46% previous pump users).

When analyzed by age, younger individuals (6–12.9 years of age) and older individuals (≥65 years of age) had the highest percentage of in-person training sessions (65% and 69%, respectively), whereas MDI users had the highest percentage of in-person training sessions across all age groups.

Glycemic metrics for up to 6 months after training are shown in Table 2. Mean number of days with glycemic data available for follow-up was 75 ± 57 days. Percentage time-in-range was higher for those were trained virtually, and for those who were trained by health care providers or by Tandem Clinical Diabetes Specialists. There was a positive interaction between training method and prior therapy, with prior MDI users who were trained virtually showing higher time-in-range than those trained in-person (P < 0.001). Our regression model showed the relationship between outcomes and training modality was not causal, as there were many contributing factors.

Table 2.

Glycemic Metrics and Adverse Events Up to 6 Months Out After Virtual and In-Person Training (n = 23,268 Pump Training Sessions)

Outcome	Virtual training	In-person training	P
Time in range 70–180 mg/dL	72% (60–81)	67% (54–78)	<0.01
Time <70 mg/dL	0.9% (0.4–1.9)	0.9% (0.3–1.9)	0.13
User satisfaction	4.78 ± 0.52	4.64 ± 0.68	<0.01
User confidence	4.61 ± 0.75	4.47 ± 0.85	<0.01
Adverse events	0.03 ± 0.20	0.04 ± 0.24	0.003

Survey results for user confidence and user satisfaction were obtained at the end of training and were scored on a 5-point Likert scale. Glycemic metrics are reported as median (IQR), whereas user satisfaction, user confidence, and adverse events are reported as mean ± SD.

IQR, interquartile range; SD, standard deviation.

In terms of patient-reported outcomes, satisfaction with training and confidence in using the system as intended was higher after the virtual training sessions (Table 2). Although user confidence decreased with age (P < 0.01), there was a positive interaction between age and training method, with confidence decreasing less with age in the virtual training group than in the in-person group (P = 0.01).

Conclusions

Before the COVID-19 pandemic, in-person insulin pump training sessions were the norm, although some individuals had completed these visits virtually. Virtual insulin pump training for existing pump users who were switching to the Tandem t:slim X2 insulin pump had been an option for the past few years, and is currently an option for those performing remote software updates to add features to their existing t:slim X2 insulin pump. ¹⁰ Most users performing software updates completed only a self-training module and then proceeded to use their pump with updated features. The benefits of virtual training include scheduling flexibility, access to individuals who live in more remote locations, and reaching individuals who experience challenges traveling to appointments. This allows for the broadest possible inclusion of all individuals to use advanced insulin pump technology in their diabetes care. ¹¹

Recent pivotal trials of insulin pumps, such as the MiniMed 670G™ hybrid closed-loop system, as well as the Tandem t:slim X2 insulin pump with Basal-IQ or Control-IQ technologies, have required in-person device training. ^{12 –16} Yet there has been a tremendous increase in telemedicine visits since the start of the COVID-19 pandemic. Others have reported success of performing virtual pump trainings, with comparable glycemic outcomes to in-person trainings, but these trainings were based on existing curriculum with the same training duration. ¹⁷ The Tandem s:slim X2 pump training program, redesigned specifically for virtual training, uses technology in a number of specific ways to further enhance the training process, from pretraining with a simulator app to enhancing the follow-up process with automatic pump uploads through a smartphone app (Table 1).

Users have faced challenges uploading important pump data to their health care providers and pump trainers for follow-up visits. Traditionally, users had to login to their diabetes management application and plug their insulin pump into a computer to upload pump data for remote review. In fact, it has been previously reported that older individuals who are not as comfortable with technology may have trouble engaging with their device in this manner. ¹⁷ Tandem initially addressed this by using pump serial numbers to automatically select the correct t:connect account, rather than requiring usernames and passwords. More recently, Tandem released the t:connect mobile app to further reduce barriers to accessing pump data for provider and trainer review. This smartphone app, available for both iOS and Android, pairs through Bluetooth^® wireless technology to the Tandem t:slim X2 insulin pump and automatically transmits insulin delivery and continuous glucose monitoring (CGM) data to the t:connect web application, obviating the need for traditional wired pump uploads. The app also acts as a secondary display of the pump.

Follow-up training, whether virtual or in-person, is increasingly complex as technologies are being developed to automate tasks previously requiring patient and/or their provider engagement. Some artificial pancreas systems have automated learning algorithms that self-adjust insulin delivery settings, although these changes are not always visible to the user. ¹³ Decision support systems can now provide automated analysis of CGM and insulin delivery data, and then provide recommendations for insulin dose adjustments to providers. ¹⁸ Exactly how much each individual needs to know about how these systems work, and how best to interact with these systems is a dynamic ongoing discussion.

In this retrospective study comparing in-person and virtual training metrics, including glycemic and user reported outcomes, we found time-in-range after training was improved in the virtual training group compared with in-person training. Although it was not possible to determine a causal relationship between glycemic outcomes and training modality, it is clear the device can be used safely and effectively after virtual training. Our data suggest that there is a selection bias, as individuals more experienced with pumps and those in the age range 12–64 years were more likely to receive virtual training. Virtual training also required setting up a t:connect account for data uploads in advance and practicing assigned tasks with the t:simulator app. Therefore, those who completed virtual training may have been more engaged with the overall process. It is also likely that those who completed only virtual training were less likely to need intensive training than those who opted for in-person training or who converted to in-person training.

In conclusion, we found use of Tandem's new virtual pump training program to be safe and effective. It is likely many of the teaching methods employed as part of the virtual training program will gradually be integrated into the in-person pump training curriculum as well. Therefore, we expect there to be continued use of the virtual training program, in coexistence with in-person training, even after the COVID-19 pandemic is over.