April 21-22, 2017; Houston, TX; Highlights – Draft

Executive Highlights

Our team traveled down to Houston, Texas earlier this month for the 13th annual Clinical Diabetes Technology Meeting, an intimate conference that always offers an on-the-ground look at the latest in blood glucose monitoring, CGM, insulin delivery, and software tools. We heard particularly thoughtful commentary this year on hybrid closed loop, with experts sharing clinical pearls when it comes to Medtronic’s MiniMed 670G (currently in a customer training phase), how to educate patients, and how to set reasonable expectations. Below, we include our view of the top eight highlights from this two-day meeting. Enjoy!

Top Eight Highlights

1. We heard tons of tips, tricks, and real-world considerations for using the MiniMed 670G, particularly from Barbara Davis’ Ms. Laurel Messer. Speakers touched on patient selection (“type A+s” and “type Zs” may not have the best experience); expectations management (this is “not a system you put on and forget about your diabetes”); top reasons for 670G auto mode “exits” (missed CGM calibration; prolonged high glucose; auto mode max delivery timeout; sensor algorithm under-read; auto mode min delivery timeout); and optional setup and use.

2. A number of presenters shared case studies illustrating the potential for the MiniMed 670G to make a big difference in users’ glycemic control and quality of life – especially overnight and in compensating for incorrect mealtime boluses. Still, there were mentions of some patients that have found the system burdensome, yet another reminder that this will not be perfect for everyone at this early point.

3. In the hands-on sessions, we heard lots of enthusiasm for Abbott’s FreeStyle Libre Pro, heard mention of a possible Tandem Bluetooth BGM integration, and tested G5 mobile.

4. Meeting co-organizer Dr. David Klonoff (Mills-Peninsula Medical Center, San Mateo, CA)  asserted that, although CGM is useful and improving, BGM will be widely used for the next 10 years because of its cost, accuracy, and availability. He also made a few comments on the Novartis-Verily contact lens, which he helped do feasibility studies on (no timing update).  

5. OHSU’s Dr. Jessica Castle kicked the meeting off with an all-encompassing overview of CGM, including commentary on the Edelman-Pettus dosing method, desires for future alarms, and her CGM recipe for success.

6. UCLA’s Dr. David Ahn led a broad-scoped discussion on mobile apps that addressed questions of provider burden, patient engagement, and app assessment. Quotable quotes included below!

7. In a rapid-fire overview of the diabetes software, Dr. Ahn covered insulin bolus calculators like Lilly’s under-the-radar Go Dose app for Humalog (FDA cleared, but not yet launched), subscription-based coaching from mySugr and One Drop, and cloud-based portals for data sharing like Dexcom’s Clarity platform.

8. In a talk on insulin delivery devices, local endocrinologist Dr. Archana Sadhu discussed the latest and greatest in syringes (BD U500), shared notable excitement for upcoming connected pens, and highlighted HCP concerns with Afrezza’s impact on the lungs.

Top Eight Highlights

1. Practical Tips and Tricks for 670G Use

With reps as well as 670G pivotal trial administrators present, there was no shortage of tips, tricks, and considerations to keep in mind when guiding a patient who is using the new hybrid closed loop system. Barbara Davis’ Ms. Laurel Messer’s talk was particularly rich in practical insights she’s picked up in her three years overseeing use of the system.

Communication/Expectation Management

• We’ve heard over and over that 670G isn’t for everyone – Ms. Messer suggested that “type A+s,” “type Zs,” and those with difficulty wearing CGM may not have the best experience. She added, however, that everyone in between – Bs to Ys (most people!) – could be good candidates for hybrid closed loop therapy. A+s may have anxious obsessive tendencies with diabetes care or expect perfect diabetes control, which the 670G is not going to deliver. For best results (apart from bolusing), users should, as Dr. Bergenstal urged at ENDO 2017, “let it work, let it work, let it work.” On the other side of the coin, “type Zs” may struggle because of more elementary things, like missed CGM calibrations or skipped bolus insulin doses. Getting the best results out of this system will certainly require a mindset shift from just about everyone who starts it, and it is crucial that patients have the right expectations and support as they proceed.
  
  • “Type A patients will need to be taught to let go and trust the system, except for bolusing.” – Dr. Sherr
  • Ms. Messer emphasized that the number one job of the provider is to set reasonable expectations. “This is not a panacea,” Ms. Messer underscored. “It’s not a system you put on and forget about your diabetes.” Ms. Messer described the ideal candidate for hybrid closed loop as a patient willing to troubleshoot for the first few months before deciding to discontinue, and as someone who can articulate back that this system will not cure their diabetes but will help them achieve better glycemic control.
• “It’s important to talk to the patients. Tell them that “bolus” in Latin really means “best guess.” – Dr. Sherr. Of course, this was said with tongue in cheek, but the point is that patients shouldn’t be afraid to bolus for fear of miscalculating the dose – an estimate is better than nothing. A Medtronic rep simply added, “Please – just bolus.” We are delighted that the importance of bolusing is not being understated, in terms of expectation management.
• Ms. Messer suggests setting the expectation that there will be a multi-week learning curve with the system. “Glucose control is not magically better overnight – it will really take a few weeks of intensive tweaking.” During this learning phase, patients will become accustomed to higher frequency of alerts and should expect to endure less than optimal blood glucose control while optimizing dose settings. At Yale, Dr. Sherr noted that on-boarding for the pivotal study (n=11, including herself) took ~two weeks.
• A CareLink analysis revealed the top five reasons for 670G auto mode “exits”: (i) Missed calibration (15% of exits); (ii) Prolonged high sensor glucose (15% of exits); (iii) Auto mode max delivery timeout (10% of exits); (iv) Sensor algorithm under-read – the difference between the BGM reading and the sensor glucose is ≥35% (10% of exits); and (v) Auto mode min delivery timeout (9% of exits). The reps emphasized that exits should be viewed as learning opportunities rather than failures, and often times this is self-enforced – in their experiences, patients strived to avoid exits to avoid the hassle of getting back into Auto Mode. (Overall, we thought this put a positive spin on what could be considered negative by some – leaving closed-loop reverting back to manual mode). Minimizing these transitions, in our view, is likely to result in the best outcomes. And if that is true, the most effective closed loop systems should ultimately be those that minimize the number of open loop/closed-loop tradeoffs – there is plenty of time to develop this.

Practical Tips

• The Medtronic rep running the hands-on session recommended using the pump for six or more days in manual mode before entering auto mode. The system gets to know patients better, and among Medtronic employees currently on the system, those who waited a week to enter auto mode did better than those who waited just the minimum required two days. To illustrate how the two-day period could provide incomplete information, the rep posed a scenario: A woman starts in manual mode on Saturday. She is very active on the weekend, leading to numerous insulin suspensions that result in 25% of her typical daily basal dose being lopped off. She then goes to work on Monday, where she sits in front of a computer all day. Because the system adjusts insulin delivery based on TDD of the past six days, it will likely not be aggressive enough on Monday. 
• “Close follow-up is key. We would recommend at least weekly contact for three weeks.” – Dr. Sherr. The first iteration of the customer rollout phase at Yale followed a specific protocol: In week one, patients were trained on the Guardian 3 sensor and 670G pump, and then sent home in manual mode for a week; In week two, auto mode was turned on (this could be done remotely with review of CareLink data prior); Week three is a follow-up to troubleshoot issues and tweak settings as needed. We assume the ongoing customer training phase will help optimize the onboarding process, but still, we’ve heard anecdotally that the 670G is not the simplest of products to learn and set up.
• “Basal adjustments are an intellectual problem. How do you keep basals set correctly when sensor data is mostly representing auto mode?” – Ms. Messer. Ms. Messer recommended checking the total auto basal dose against the pump open loop basal settings to make sure they are comparable, making adjustments so the basal total is ~10% less than the auto basal total, and, if concerned, running manual mode for 3-5 days to optimize the basal rate.
• The insulin to carb ratio and the insulin action time are the only two parameters that can be adjusted. Barbara Davis’ Ms. Laurel Messer claims that she and her colleagues, who have now used the system for three years, didn’t grasp this point initially. When they first began, they adjusted basal rates for a year before Medtronic clarified that doesn’t do anything to the system’s aggressiveness. At Yale, however, Dr. Sherr commented that this was thoroughly reviewed during the site initiation visits.
  
  • Between two and three hours seems to be the sweet spot for the insulin action time, at least in clinical trials. Ms. Messer recommends starting higher and slowly working down in 15-30 minute jumps to the two to three hour setting to make the system more aggressive. For background, the accepted action time setting on a pump is in the neighborhood of four hours, but this hasn’t proved aggressive enough with the 670G’s PID algorithm.
• The BG entry “recommended” alert does not mean calibrate – rather, Ms. Messer likened it to “a needy doggy that needs reassurance.” A fingerstick is still requested, but it does not adjust the sensor’s readings, it simply reassures the system that it is reading well. We weren’t clear on when such an alert would pop up, and if it can be ignored. This will be important for the perceived burden of using the system, as the Guardian Sensor 3 requires a minimum of two calibrations per day, but 3-4 are “recommended.”
• Patients should always suspend the pump when it is not being worn. With a typical pump, lost insulin is the only concern, but with hybrid closed loop, this lost insulin will be considered in the TDD calculation, which could make the system overly aggressive in the subsequent days.

2. 670G Case Studies and User Experience (Feat. Dr. Sherr’s Own Story)

A number of presenters shared case studies illustrating the potential for the MiniMed 670G to make a big difference in users’ glycemic control and quality of life. Most of these stories come from Yale’s Dr. Jennifer Sherr (many from her own experience wearing the system)!

• Dr. Sherr’s 670G story likely reflects the best possible scenario, and we wonder what percentage of patients will be as successful as she is. Upon receiving the system, she immediately had a “Price is Right” moment – she was so excited that she captured a photo of her CGM trace indicating that she was at 115 mg/dl with two up arrows. Engaging the system felt to her like getting in a self-driving car – where’s the brake and where’s the steering wheel? “I’m type A, how do I develop trust?” She eventually did learn to trust the system, and her overnight control was likely one big reason why: “Parents in my clinic report that they get no sleep. I had no idea how poorly I slept. Now, I can just sleep, and that’s pretty darn nice. I didn’t realize how chronically fatigued diabetes makes you. This will help parents, kids, and really all relationships.” In fact, her family can already feel improvements – her 13-year-old son even told her that she’s “a heck of a lot nicer” since she started wearing 670G.
  
  • With regards to food, Dr. Sherr is less burdened when it comes to carb counting, and has noticed marked improvements in her control in “danger zones.” The 670G allows her to be less precise with prandial insulin doses, because it will compensate for slight errors on the backend. Before 670G, she noted that all of her boluses ended in threes, sevens, and ones. Now, they all end in zeros and fives. Further, she was able to enjoy culinary experiences that she had previously avoided: (i) She had stopped eating Chinese food years ago because dosing was too complex, but thanks to the system, she was able to sit through a meal with a steady sensor glucose after a “wild guess” estimated carb count. (ii) Colleague Dr. Bill Tamborlane holds an annual Christmas celebration at an Italian restaurant, a wonderful party from which Dr. Sherr always ended up departing with high blood sugar. With the 670G this year, however, she left with an 89 mg/dl. As previously mentioned, this n=1 experience my not be typical of everyone – Dr. Sherr is tech-savvy and experienced and we’ll be interested to see how broadly applicable her experience is.
• Within a week of wearing the system, Yale’s first commercial patient (started April 4^th) was already spending more time in target, so much so that she felt physiologically terrible when she forgot to bolus for a cookie. The improvement was so abrupt in this individual that she regained autonomic responsiveness to elevated blood glucoses within a weeks’ time. Wow, talk about a positive feedback loop and behavioral modifier! 
• At Barbara Davis Center, each member of a group of young children that started on the system earlier in the week woke up with blood glucose readings between 110-121 mg/dl. BDC’s Ms. Laurel Messer claims that some gasped upon checking and realizing that they had woken up in range.
• The 670G is not for everyone – Ms. Messer related the story of a young man who had an A1c of 6.6% and endorsed many more concerns about the system than benefits. He complained of pain issues related to wearing both a pump and sensor and that he didn’t like how it made him pay more attention to his diabetes with alarms and advanced statistics. This patient claimed that it felt like the system was running his life, and he was generally more anxious. One possible culprit for his experience, said Ms. Messer, is that he was not a regular CGM user; he was not accustomed to the on-body burden, nor the constant influx of blood glucose data. He discontinued use after three months, and she emphasized that this is ok – “If it ain’t broke, don’t fix it.”
  
  • While this patient was certainly not the ideal candidate for hybrid closed loop, Ms. Messer wonders how the discontinuation could’ve been avoided. What if his expectations had been managed better? What if providers had informed him in more detail how much work the system would entail with CGM calibrations, bolusing, infusion set changes, etc.? On the other hand, we believe the design of the 670G is very conservative, and this scenario is likely to be common – the system does give a lot of alarms, and though burden will go down at night, it may not go down during the day for many users.  
• According to Dr. Sherr, the typical CGM adherence in adolescents at the end of a traditional study is near 30% (in reference to the JDRF CGM trial), compared to 76% by the end of the 670G pivotal. “You can tell teens to wear the device over and over, but until they decide it’s important, they’re not going to do it. They must’ve felt there was a benefit.”
• Dr. Sherr projected a visual depicting insulin delivery during the overnight periods and exclaimed: “Oh my god, to think I had been spending all this time thinking of variable basal rates. We’re really seeing what physiologic insulin delivery can do.” Without a doubt, the overnight period is where 670G and all hybrid closed loop systems will offer the most benefit for the majority of users, and the figures showed exactly why – in one patient over four consecutive nights of wear, the insulin delivery profiles were completely distinct. The dawn phenomenon, ambient temperature, bedtime snacking, previous day exercise, stress and so many other factors can impact a night’s glucose trace, so having automated insulin modulation is incredibly valuable. (Read Adam’s diaTribe article here for a taste of this, and much more in his upcoming book.)
• “What I’ve figured out so far is that the algorithm is just much smarter than I am” – a Medtronic representative who has been wearing the system. This recalls our team’s assertion over a long period of time – why should patients do these calculations manually when a computer can do them?
• Comments on the new Guardian 3 (the sensor component of the 670G system):
  
  • Dr. Sherr finds the sensor’s on-board diagnostic chip to be a huge improvement. As a CGM wearer, she commented that whenever something goes wrong with the signal, she automatically thinks “what did I do wrong?” With the chip, however, she feels that the onus is off the user and onto the system, and is more than happy to change a sensor when it fails knowing that it usually wasn’t because of something she did. Again, we think this could be annoying for some patients not used to wearing CGM, as frequent alarms and prompts may not reduce the burden of managing diabetes.
  • BDC’s Ms. Laurel Messer stated that her patients are comparing the accuracy of the Guardian to that of Dexcom’s sensors. Dexcom holds a slight advantage over the Guardian Sensor 3 in the MARD department (9.0% on two calibrations per day vs. just over 10% on the Guardian), but even to be compared favorably is a huge departure from how the Enlite sensor is generally viewed.

3. Hands on Sessions: Abbott Shows Libre Pro, Dexcom Shows G5, and TanDem Talks Pipeline

• At the Abbott hands-on session, enthusiasm ran high for the FreeStyle Libre Pro. As providers handled the sensor and reader, many shared their positive experiences with the device. One woman thanked the rep – “I love it, it’s so cheap. Thank you for developing something so affordable and low-maintenance.” Following the same theme, another regretted that her practice had begun a contract with a different professional CGM company in September (a month before Libre Pro rolled out in the US): “My patients would be so much happier with this, it looks so much less painful and we would save so much money.” According to one Houston-based educator, the round-the-clock monitoring helped her diagnose one of her patients with a sleep disorder – she found that the patient’s blood glucose spiked to 400 every night, and the patient had no idea and no idea why. She was referred to a sleep study, where she was diagnosed with sleep apnea. The one common question – one we hear frequently – is when will this be real-time? On January’s 4Q16 call, management projected a 2H17 launch, though FDA is hard to predict, so we wouldn’t be surprised if this took longer. There was no timing update on the 1Q17 call; the Abbott CEO expressed eagerness to see this approved as part of his commentary.
• Tandem’s Mr. Aymeric Lecanu-Fayet led this hands-on session, where he briefly mentioned integration with Dexcom’s G5 as a priority (summer 2017 launch expected), and in new news, alluded to a possible integration with a Bluetooth BGM. The latter has never been shared on a Tandem quarterly call and we’ll be interested to see if something comes of that – presumably this is needed for Tandem’s automated insulin delivery products, paralleling Medtronic’s MiniMed 630G/670G with the paired Ascensia Contour Next Link BGM. We might guess Tandem’s BGM partner is either Ascensia, J&J, Roche, or AgaMatrix (all of whom have Bluetooth BGMs). He also emphasized the patient-centric design of the company’s insulin pumps as audience members viewed the products up close. Tandem’s pumps are designed to be small, discreet, and green, with a rechargeable battery (apt, as Mr. Lecanu-Fayet reminded everyone that it was Earth Day). These features improve quality of life and also reduce out-of-pocket costs for patients, since disposable batteries can get expensive. He described Tandem’s “t:slim in the wild” campaign, where patients send in photos of themselves being active and adventurous while wearing a Tandem pump, which reflects the underlying notion that patients should be empowered with technology to live their lives as close as possible to how they would without diabetes. (We love this!) He also mentioned in passing that Tandem is working on a hybrid closed loop system with Dexcom’s G6 sensor that will improve on Medtronic’s MiniMed 670G (launch expected by the end of 2018), and that the next big Tandem project will be the t:slim X2 with predictive low glucose suspend (launch expected in early 2018). We’re so impressed with all that Tandem has going.
• Dexcom’s Ms. Christy Pospisil led the G5 hands-on session, in which audience members could practice inserting a sensor (on a piece of foam) and demo the G5 app on their personal phones. She confirmed that Dexcom’s next-generation, one-button insertion system with a 50% smaller transmitter (G5x) is coming soon, though no more specific details were shared. Dexcom received questions from the FDA in February; at JPM, this was positioned as a 2017 launch. Ms. Pospisil alluded to an upcoming Android version of the G5 app, slated for a mid-2017 launch, per the 4Q16 call. She told the story of how after many long discussions, the FDA finally approved Dexcom’s G5 CGM for non-adjunctive insulin dosing in December. The company is now in a “mad rush” to establish the necessary educational resources, she explained, which includes updating the G5 app and disseminating all the right information to patients and providers so people can make safe treatment decisions. The G5 app was updated in March with some education, and dexcom.com/fingersticks has some useful tips too.

4. Dr. David Klonoff Predicts BGM Will Be Widely Used for 10 More Years, Talks Emerging BGM Technology

Meeting co-organizer Dr. David Klonoff asserted that, although CGM is useful and improving, BGM will be widely used for the next 10 years because of its cost, accuracy, and availability. That said, pooled Big 3 (Roche, J&J, Abbott) sales have now fallen for five consecutive years, and revenue dropped below $5.0 billion in 2016 for just the second time since we began tracking over a decade ago. Competitive and pricing pressures are stressing the traditional players, and we seeing more focus on monitoring + service (insulin dose titration, coaching, real-time support). Dr. Klonoff highlighted a number of these, including insulin dosing apps (“there are currently three approved apps, Voluntis, Eli Lilly, and Sanofi. We’ll see more of these. If they’re FDA cleared, that means they’re pretty safe), dosing incorporated into meters (Sanofi’s MyStar Dose Coach, Roche’s Aviva Expert), pattern analysis (Ascensia Contour Next One meter), and data aggregation/presentation platforms (Glooko/Tidepool). The BGM field will continue to offer more value to patients, and we wonder if Dr. Klonoff’s 10-year estimate will be the tipping point at which CGM becomes inexpensive, accessible, and small enough to be widely adopted – current CGM penetration in type 1s in the US is estimated to be just ~10% to ~15% (depending on the denominator used). We anticipate that the impending commercialization of FreeStyle Libre, Medtronic’s standalone Guardian Connect CGM, and Dexcom/Verily’s first-gen product will play a big part in boosting that number.

• Dr. Klonoff also touched on Verily’s glucose-sensing contact lens project and the Apple non-invasive CGM rumors, though he didn’t make predictions as to whether they would succeed. He actually performed feasibility studies of the contact lenses at his hospital (presumably Mills-Peninsula in San Mateo), and he hopes that it works out and is developed. He acknowledged that Verily is still working on it but in a silent period. He similarly said little about the Apple rumors, only acknowledging that he knows someone working on the project but “if he told us what he was doing he’d have to kill us.”
• A protocol for the SMBG Study – a one-year RCT (n=450) investigating the efficacy of SMBG in non-insulin treated type 2s – was published in BMC Endocrine Disorders in January 2017. Dr. Klonoff believes that the study, performed by Dr. David Owens’ UK group will show a benefit vs. the control group, presumably due to behavioral modification – i.e., “If I check and see my blood glucose spikes to 180 after a bowl of cereal, maybe I don’t eat that cereal anymore.”
• Dr. Klonoff and Weill Cornell’s Ms. Jane Seley hold opposing views as to which is better – handwritten or digital logbooks. Dr. Klonoff jokingly began by saying “logbooks are useful,” before adding “if you’re from the state of Alaska Fish and Game, but not for diabetes. This is the 21^st century – this should be digital.” Ms. Seley finds digital logbooks easier to fake (she can tell if manual logbooks are authentic if they have blood on them, different inks, and/or messy) and harder to interpret. We agree with Dr. Klonoff – handwritten is just not the way to go anymore.

5. Dr. Jessica Castle Comments on Dosing off CGM, Desires for Future Alarms, and Recipe for CGM Success

OHSU’s Dr. Jessica Castle kicked CDTM 2017 off with an all-encompassing overview of CGM, including commentary on the Edelman-Pettus dosing method, desires for future alarms, and her CGM recipe for success. She noted there are a number of approaches to dosing off of CGM readings and trend arrows. DirectNet and Mr. Gary Scheiner have shared similar strategies, and respected Drs. Steve Edelman and Jeremy Pettus published a method last August in JDST. Dr. Castle didn’t endorse one over others, but she did point to one practical drawback of the Edelman-Pettus method: Because it is based on an anticipated glucose value 30 minutes in the future, it can be confusing for a provider to look at the pump download later on. Then again, if it results in the best glycemic control for the individual, he/she should continue to use it. Dr. Castle characterized high/low alerts as both a friend and an enemy – while they can minimize the duration and magnitude of glucose excursions, patients will turn them off if they become too big of a nuisance. (And on the other hand, patients can easily overreact to CGM highs with a “rage bolus” and stack insulin, or overreact to lows by eating too much [especially since CGM lags when recovering from hypoglycemia]). With today’s technology, she recommends setting the alerts to maximize the benefit:nuisance ratio for patients, but in the future, hopes that alarms will be “smarter” and take into account factors such as when the patient last ate or last took insulin (i.e., if someone just took a bolus but it hasn’t onset yet, it would be quite annoying to get multiple alerts that blood glucose is elevated). To conclude, Dr. Castle gave her recipe for CGM success: (i) Calibrate properly – garbage in, garbage out; (ii) Wear it full time – think of the CGM like a toothbrush; having it in your drawer doesn’t reduce dental decay, using it does; (iii) Look at the monitor approximately once every hour; (iv) Don’t overreact to the data – take context into account; and (v) Be mindful with the alert settings.

• Dr. Castle was relieved when Medicare opted to cover therapeutic CGM (Dexcom’s G5): “They changed how they are covered, which is very important. It has been a fantastic advancement – previously, patients had been using CGM for years, then turn 65, and Medicare wasn’t paying for devices. That was a huge problem. Especially for someone with hypoglycemia unawareness, dependent on CGM, and then not covered. Thankfully that has changed.” Anecdotally, the rollout hasn’t been without hiccups, as we’re hearing that some patient populations (e.g. Medicare Advantage) are having a hard time getting coverage. We hope to hear more on Dexcom’s upcoming call! Of course, this has been an enormous win for the field and for patients, and it’s unrealistic to expect the 800-lb CMS gorilla to have everything up and running quickly.

6. Mobile App Panel: How to Engage Patients?

UCLA’s Dr. David Ahn led a broad-scoped discussion on mobile apps that addressed questions of provider burden, patient engagement, and app assessment. University of Missouri’s Dr. Mark Clements made a very poignant point early on – even though some clinicians believe that apps will add burden for the care team, “the burden is already there. We all field calls in-between visits. It’s unmeasured work.” Some apps not only offer means to track and monitor behavior and physiology and communicate, but they could also present a way of measuring or even reducing this between-visit work. The majority of the panel explored ways to keep patients engaged in their care with apps. But before engagement can be addressed, patients have to use smart devices; UMass’s Dr. Michael Thompson believes we are in the middle of a spike in connected device usage, but at this point, Dr. Ahn said that <5% of patients in his clinic use connected meters, and Dr. Thompson added that 15% of people in his neighborhood still uses flip phones. Once patients have transitioned to app-enabled devices, Dr. Clements advocated for diligent and iterative human factors work to explore what makes apps sticky. Baylor’s Dr. Bijan Najafi believes that the way to go is to help people take charge of their own health, but we frequently hear that patients want to spend less time thinking about their diabetes. Dr. Clements suggested that a slot machine approach, in which patients are entered into a lottery to win a prize if they complete certain tasks, might work. (We disagree.) At the end of the day, it will take experimentation to determine what methods engage individual patients. But as Dr. Clements pointed out, RCTs are difficult to conduct in this domain because apps have different modules and can therefore be used in completely different manners from patient to patient (not to mention that the RCT process is long, and the software will likely be out of date by the time the trial is completed). One solution, proposed by Dr. Clements, is to conduct RCTs with smart adaptive designs – i.e., segmenting patients depending on their usage of different features within the app. We’ve included some of our favorite quotable quotes below:

• “On the worst extremes, I’ve seen people in our Livongo trial who we shipped the meter to and have never opened the box, even after consenting and signing to participate.” – Dr. Thompson
• “When we start talking about apps, our peers are immediately critical – how are you going to deal with the burden on care team? The burden is already there. How can we take that model, turn it on its head, and make it work for our patients? It needs to be studied, from an economic perspective.” – Dr. Clements
• “How can we make mobile apps sticky, so that people keep coming back to them? We have to think about behavioral components and what incentivizes people to return to an app over and over and over again. Getting feedback from your diabetes care team is one thing that brings people back, but it’s probably not enough to sustain the level of patient engagement we’re seeking. Companies need to continue diligent human factors work. They need to keep coming up with new designs and asking physicians and patients for input – that’s the only way sticky apps are going to happen.” – Dr. Clements
• “As a consumer observing the marketplace, the cellphone has become a pretty crowded space, I think I have well over 100 apps, and probably about 10 are sticky, and the ones that are the stickiest are the ones that send a message on the banner bar. The watch is still a wide open space that’s not crowded for most people, where they’re doing ~100 things with the phone.” – Dr. Clements [Editor’s Note: We disagree on the value of bombarding patients with push notifications. This is not a good strategy for behavior change in our view, as people with diabetes – almost universally – want to spend less time on diabetes.]
• “What percent of the clinic is using these telehealth tools? I’m excited about all of these things, but I’d say it’s marginal; less than 5% of my practice is using a wireless meter.” – Dr. David Ahn
  
  • “In our area, at least 15% or more of the population still uses a flip phone.” – Dr. Thompson
• “In a value-based healthcare system, we’ll see more opportunity to incentivize patient engagement. We’re moving from a fee-for-service model to a bundled care model, and that’s going to create an opportunity to leverage behavioral economics in a unique way. Right now, outside of philanthropy, we don’t have a good way to create consumer-based incentives.” – Dr. Clements
• “A small observational study with Glooko showed that while A1c doesn’t drop in a short period of time, you do get meaningfully and significantly increased engagement with blood glucose monitoring. But knowing that individuals who engage with the mobile app also adhere more to other aspects of self-care – this leads me to believe that we might be able to demonstrate significant A1c-lowering over time.” – Dr. Clements
•  “Apps that empower patients are the most important – apps that motivate people to take charge of their own health and emphasize people’s self-efficacy.” – Dr. Najafi
• “As we talk about studying these mobile apps as an intervention and other functionalities, a randomized controlled trial falls short, because a mobile app is not one thing. There are modules – patients can engage with one aspect and not others, RCTs with smart adaptive designs will be much more important.” – Dr. Clements
  
  • “Since it’s so costly to look at long-term outcomes in trials, I would suggest just looking at rate of readmission to the hospital. That would give you an idea of the clinical and economic benefit” – Dr. Klonoff
• “I liken technology acceptance to online banking – it really shot up from 2000-2013. I think we’re in that same rate of growth with connected devices.” – Dr. Thompson
• “In one behavioral reinforcement paradigm, you first show a participant what they can win, what they can lose, but the other way is to incentivize in a way that’s more like a sweepstakes, the slot machine approach. You show people if they engaged whether they won, but also if they had engaged what they would’ve won – you were selected, but you didn’t win. We have a lot to learn from fields like HIV/AIDS treatment, addiction treatment, and others. – Dr. Clements
• “In diabetic foot ulcer, if a patient is coming frequently to the hospital, then it creates a back-log. We’re trying to create a model with an app and some modeling to look at wound classification, creating a triage – some of them can see local practitioners, others need referral to a specialist, so we don’t create a huge backlog on seeing the patient. Triage, personalized care, how to schedule/educate patients – this could be a very exciting direction for engagement and personalized care. – Dr. Najafi

7. Dr. David Ahn on Software Available to People Living with Diabetes

UCLA’s Dr. David Ahn provided a rapid-fire overview of the software available to support people living with diabetes, covering insulin bolus calculators like Lilly’s Go Dose app for Humalog (FDA cleared, but not yet launched), subscription-based coaching from mySugr and One Drop, and cloud-based portals for data sharing like Dexcom’s Clarity platform. Dr. Ahn shared that he presents to UCLA medical fellows on software tools for diabetes every so often, and that it surprises him how frequently he needs to update his slides. He described diabetes technology as rapidly evolving – new software allows us to collect more, better data, which only necessitates further advances in software. The sheer number and variety of apps for diabetes can be overwhelming, but Dr. Ahn pointed to a distinct advantage of variety: more personalized tools to an individual patient. (We agree on one hand, though the low barriers to entry – and high barriers to making something great – mean the app landscape is cluttered with many diabetes logging apps that add burden, rather than subtract it. In turn, this can give digital health a bad rep.) Aside from Lilly’s Go Dose, Dr. Ahn also mentioned Sanofi’s recent FDA clearance for My Dose Coach, a basal insulin titration app for type 2 diabetes. We’re elated to see the insulin companies committing to digital health (Novo Nordisk also has the Glooko and IBM Watson partnerships), and we believe the next few years will see these products reach significant scale and hopefully new business models.

Questions and Answers

Dr. Klonoff: Can you comment on concerns related to security and privacy with this technology? A lot of information is sent to these systems, and we may have no idea what they’re doing with it.

Dr. Ahn: I think that’s kind of the elephant in the room. I was at a technology meeting a couple years ago, and there was a patient advocate who had a screen with the CGM tracings of everyone else in the room. There was no identifying information and of course it was a diabetes conference, but I still found that kind of disturbing. If you’re in a classroom and you’re the only person with type 1, people can identify what your sugars are. So yes, there are a lot of issues that need to be worked out. On the other hand, our culture is changing. People are now, generally-speaking, much more open with their data. People throw caution to the wind with the amount of information they share about themselves on social media. I’m not saying that’s a good thing, but I don’t think patients are actively concerned about privacy right now. Maybe they should be, but they’re not right now. That’s how I’d sum it up.

[Editor’s Note: Overall, we think concerns over privacy and security are completely overblown at diabetes technology meetings, especially those run by DTS. Ultimately, locked down, über secure diabetes devices are more likely to hinder innovation and increase patient burden in our view. Security and privacy are clear musts, but we also see these a check-the-box – the technology exists to do this now. Mobile payments and banking were perceived as “security risks” years ago, but now, none of us think twice. Of all the things we have to worry about in diabetes technology, this falls pretty low on the list in our view. We thought the second half of Dr. Ahn’s comment nailed it.]

Q: I’m having issues getting some of these software tools hospital-approved. The hospital’s hesitation with cloud-based download programs is that we can’t figure out how to protect ourselves if patient information was accessed by a hacker.

Dr. Ahn: After a security breach in 2015, UCLA clamped down on third party glucose monitoring software, which led to our use of Diasend (after signing a BAA). Hospitals are clamping down, and sometimes I think their concerns are overreaching. For instance, we were initially instructed to not use the Dexcom Clarity website, but many providers still do because it’s just a website and does not require hardware/software installation. My personal perspective on Clarity is that you’re not doing anything with your patient that they couldn’t do themselves in the room - you are just logging in together to their Clarity account. If it’s all directly through the cloud and web, it’s hard for hospitals to enforce rules on that, but I do share your frustration that hospitals won’t install certain software on computers.

Q: Who owns the data when these cloud systems are used by patients and diabetes centers?

Dr. Klonoff: That’s a very good question. We talked about this at the Digital Diabetes Congress last month in San Francisco. I’d say the companies own the data, unless you have evidence to the contrary. It also leads to a question of who’s responsible for storing data – if you download onto the hospital server, is the hospital now responsible?

Comment: In Kansas City, we’ve signed business associate agreements asking companies to follow HIPPA and maintain security between institutions. Our philosophy is that it’s the patient’s personal health record, and we interact with it to prescribe care. It’s not our data to lose. I think that outlook has really helped. We should also be talking about end-to-end encryption.

8. Dr. Archana Sadhu Discusses BD U500 Syringe, Connected Insulin Pens, and Afrezza

Local endocrinologist Dr. Archana Sadhu reviewed insulin delivery devices, focusing on the latest and greatest in syringes, pens, and inhaled insulin. She discussed BD’s dedicated syringe for Lilly’s Humulin U500, launched in November 2016, and emphasized that while the product is priced on par with the U100 syringe, it shouldn’t be a snap decision to make the switch. Rather, providers should take concerted efforts to educate patients before switching them over. That said, she acknowledged the benefit to quality of life that comes with the newer, U500 syringe, especially since it doesn’t require calculations and thus greatly reduces room for error. (We agree.) Turning to insulin pens, Dr. Sadhu expressed enthusiasm for Bluetooth connectivity, though this feature isn’t yet available for any pen marketed in the US. Companion Medical’s InPen is on its way to being the first, slated for a 2017 launch with Lilly’s Humalog and Novo Nordisk’s NovoLog following FDA clearance in August 2016. Dr. Sadhu characterized these smart pens as “approaching what an insulin pump would do,” and suggested them as an alternative to pumps if patients face obstacles to reimbursement or are opposed to wearing something on the body. Of course, questions of patient access to the InPen and smart insulin pens remain – we’ll have to wait and see how these new products are priced and reimbursed by different payers. Novo Nordisk’s connected Echo pen is also expected to pilot in Europe this year (per the Glooko partnership announced in January).

• On MannKind’s inhaled insulin Afrezza, Dr. Sadhu shared her personal concerns with the product, which largely reflect Afrezza’s major obstacles to uptake. As an experienced endocrinologist, Dr. Sadhu remembers well Pfizer’s first-generation Exubera – this mealtime insulin came with a cumbersome, impractical inhalation device, sparked lung cancer-related safety concerns within the diabetes community, and ended up a commercial failure. With its Technosphere technology, MannKind has created a safer, more convenient inhaled insulin options for patients with type 1 or type 2 diabetes. While the majority of patient feedback on Afrezza has been positive, providers in the field have shown more reluctance, and we imagine this is at least partly due to safety concerns and baggage from Exubera. To MannKind’s credit, the company has invested in a commercial strategy to distinguish its product from Pfizer’s former product – focusing on the “ultra-fast” onset/offset of Afrezza rather than the inhaled aspect – and has distributed spirometers to doctor’s offices in order to facilitate safety tests. Still, Dr. Sadhu articulated what many diabetes providers likely have on their minds: “The biggest hurdle for me is the unknown about long-term effects on lung function. The lung is a huge organ, with lots of absorption, and insulin is a growth factor so we don’t really know what’s going to happen over time.” MannKind may need to address safety even more explicitly to boost volume and sales of Afrezza, which we are rooting for – inhaled insulin avoids needles, expands patient choice for mealtime boluses, lowers risk for hypoglycemia due to its very rapid onset/offset, and could improve glycemic control and quality of life for many people with diabetes.

--by Payal Marathe, Brian Levine, Adam Brown, and Kelly Close