MiniMed 670G virtual training during COVID-19 has comparable glycemic results and better patient satisfaction and efficiency compared to in-person – July 27, 2020

Paper in DT&T; 68% Time in Range, 96% time in Auto Mode; Net Promoter Score increased from 78 to 84 in first month using virtual 670G training

Diabetes Technology and Therapeutics recently published a retrospective study showing that compared to pre-COVID, in-person trainings, Medtronic’s MiniMed 670G virtual trainings have resulted in improved trainee satisfaction and efficiency and comparable Time in Range and time in Auto Mode. The paper, published by Medtronic researchers Robert Vigersky, Kevin Velado, Alex Zhong, Pratik Agrawal, and Toni Cordero found that mean time in range dropped only slightly from 70% with in-person training to 68% among those trained virtually during COVID-19 (a difference of ~29 minutes/day – but most patients will take that! Another 3.5 hours a week …). With virtual training, time in Auto Mode actually increased slightly to 95.5% from an already impressive 94.9% time in Auto Mode for in-person trainees. In terms of user satisfaction, 98% of trainees found the Zoom video conferencing platform “good,” and Net Promoter Score for the training increased from 78 to 84. Training efficiency was also improved with a reduced time between pump shipment and first and last training. Notably, with virtual training, researchers saw a decline in technical support calls for educational assistance, but a 187% increase in requests for login and software installation support.

As background, the study compares virtual MiniMed 670G training during the 33-day period between March 20 to April 22, 2020 to pre-pandemic, in-person training (January 20 to February 22, 2020). Both virtual and in-person training consisted of three sessions on the pump (90 minutes), CGM (60 minutes), and SmartGuard tech (60 minutes) with supplementary teaching materials. Trainees included those who previously used MDI with BGM or CGM, as well as pump and sensor users upgrading their pump. The study’s size was not disclosed, but the authors did note the “large data set.” See the bullets below for all the key findings and takeaways. In the past, some pump trainers and HCPs have had concerns about “the feasibility of therapy training in a virtual environment.” This study’s findings respond these concerns, showing how effective and efficient virtual training can be with the right technical support for users. Importantly, the mean age of subjects in the study was 47 years, mitigating some fears that virtual training might only work for younger, more tech-savvy users. Amid the pandemic and telehealth boom, it is exciting to see yet another way that telehealth and technology can be leveraged to support patients in new and creative ways. We look forward to seeing how studies like this result in adjustments to remote training and telehealth so that we can further improve patient health outcomes and quality of life. See the full paper here and read more below.

  • Time in range was generally comparable between the in-person and virtual training cohorts. Those who had been trained in the pre-COVID, in-person program had an average of 70% time in range while the intra-COVID training cohort spent 68% of time in range. Although very similar glycemic results, the in-person cohort had marginally better time in range, perhaps, as the researchers suggests, because they had longer to get used to their AID system before the data for analysis was obtained in May and June. Dr. Vigersky and colleagues intend to reassess the CGM data in three to six months, allowing for differences in user acclimation to washout.

  • Time in Auto Mode was 94.9% and 95.5% in the in-person and intra-COVID cohorts, respectively. Notably, both of these numbers are considerably higher than previously shared real-world data on 670G. Data from the Barbara Davis Center showed ~77% time in Auto Mode, while another paper from Medtronic in 2018 showed 87% time in Auto Mode. The authors of the new paper speculated the ~95% time in Auto Mode may have been driven by more enthusiasm for the system right after initiation and also a patch for the Guardian Sensor 3 transmitter that reduced number of alerts and Auto Mode exits.

  • Satisfaction with the virtual interface was particularly high, and satisfaction with the virtual training program was higher than that of the in-person program. Almost all of the trainees (98%) were satisfied with the Zoom video conferencing platform. Of the 2% “not good” ratings, 41% were due to poor video quality and 31% were because “they could not hear us.” In terms of patient satisfaction with the program, the Net Promoter Score, rose from 78 for the in-person group to 84 for the virtual group. As a reminder, Net Promoter Scores ≥70 are considered “world-class.” Patient satisfaction NPS was determined by asking: (i) “Based solely upon your recent training experience, how likely might you recommend Medtronic to another person who themselves are insulin-dependent (if they were considering pump therapy)?” and (ii) “Based upon all of your product and service experiences to-date, how likely might you recommend Medtronic to another person who themselves are insulin-dependent (if they were considering pump therapy)?”.

  • In addition to achieving comparable glycemic results and better satisfaction, virtual training was more efficient than was in-person training. Used as a proxy for training efficiency, time between pump shipment and first and last training was significantly reduced from 14±7 to 11±5 (p<0.001) and from 19±7 to 15±15 (p<0.01), respectively. We’d imagine this improvement comes from the ability of virtual trainings to be more “on-demand” and offer more flexibility in terms of timing. There was no change in length of training: for both cohorts, the difference between the first and final training was four to five days.

  • Virtual training had mixed impacts on calls to the Medtronic 24-hour Technical Support team, which served as a proxy for training effectiveness.  Compared to in-person, pre-COVID training, virtual training came with a substantial increase (187%) in requests for login and software installation support. However, virtual training also resulted in a 19% drop in calls for SmartGuard technical support and a 13% drop in calls for system feature inquiries. This suggests to us that the trainings around MiniMed 670G were comparable for in-person vs. virtual (or perhaps more effective for virtual training); however, with more patients needing to upload their data at home (instead of the clinic), requests for support around CareLink software increased.

  • Unrelated to the trainings, CareLink saw a 92% increase in unique product page views and a 37% YOY decrease in uploads to CareLink software during the COVID-19 period. From March to April, there were 27,000 more product page views in two weeks than there were in two weeks during the pre-COVID period. From March to April, there was a 37% YOY decrease in uploads to CareLink software. Four-fifths (81%) of the drop in uploads can be attributed to reduced uploads to CareLink Professional, which is likely due to fewer in-person visits during which providers upload the data. This was partially offset by a 13% increase in Care Link Personal uploads, which partly explains the 187% increase in request for software support.


by Katie Mahoney, Albert Cai, and Kelly Close