Open menu

Off Label Pump Programming or Looping

Hybrid closed loop (HCL) insulin pumps are integrated with CGM devices. These systems use algorithms to make decisions about automated insulin dosing (stopping, increasing, decreasing basal and/or micro boluses to help prevent hypo and hyperglycemia). Off-Label HCL or "looping" uses free algorithms developed by online communities.  Systems require an external piece of hardware (a Link or a Rig) to communicate decisions between the app and pump. Some systems require older pump models to work. Carb counting is still necessary. Patients building and using these systems do so at their own risk.

Consideration: If A1c prior to the start of using a hybrid closed loop (HCL) could drop quickly by more than 2% over 3 months (particularly for those with A1c in double digits), consider discussions with the specialist physican and patient about higher temporary glucose targets. Rapid A1c improvements have been associated with changes in retinopathy and rarely, painful treatment induced neuropathy of diabetes (TIND) or insulin edema.

 

Content Links

  1. Role of Educators in Off-Label Pump Programming
  2. Key Similarities & Differences between HCL Systems (on and off label) 
  3. Types of off-label "Looping"
  4. Loop specific elements that users can adjust
  5. Resources for Off-Label Hybrid Closed Loop (HCL)

 

Role of Educators in Off-Label Pump Programming


  • It is important to still review the basics of diabetes control in pump therapy, especially safety and DKA prevention. As of April 2021 these off-label apps still require basal rates programmed by the user. Rates that are not appropriate for the user are likely to pose problems with glycemic control.  
  • Since these are "off label" usages, we advise diabetes educators to not suggest the systems but rather to support those who are on them.
  • Alhtough off-Label HCL pumping can be complex, much of the analysis of data is similar e.g. focus on problem areas first (lows, then highs). Solutions to those problems may be similar or unique compared to normal pumping or to other algorithms. Most patients who loop are savvy in how to address most of their concerns and are safe on the system. If you feel they are not safe and cannot troubleshoot to stay safe, please consider discussions with the patient and endocrinologist to return to normal pumping (or as an alternate “open the loop”) for a time. This gives patients time to refer to their algorithm’s extensive documentation to become better informed, talk with their key looping resources (endos, some trainers) and gives other clinicians time to seek guidance if available. 
  •  "Opening their loop"  usually returns the user to most (but not neccesarily all) of the features of manual pumping while still using the Looping App with current programmed basal rates, ICR and ISF but not using automated insulin modulations.
  • Patients returning to "normal" pumping (e.g. stop looping altogether) need to know their TDD at a minimum, but ideally their basal rates, ISF, ICR and insulin action time. Patients can provide clinicians their TDD according to the history in their Apple Health App if using "Loop" app, or their profile report in Nightscout (settings > reports > profile). 
  • The electronic boxes that facilitate communication between off-label apps and pumps can lose connections. Usually after 20-30 minutes of lost connection the app resumes the reguarlly programmed basal rate in the pump. 
  • Educators can learn from patients.  Avoid generalizing what patients tell you, as versions and types of algorithms differ and regularly change. Ask for explanations when you hear information that is unknown to you or sounds contrary to your knowledge. 
  • In an ideal world, clinicians would be just as informed of these algorithms as are savvy users. However, numerous pragmatics prevent this. Resources for learning about algorithms are below for clinicians, with the repeated caveat that algorithms can be updated at any time with different features and tweaks that are not Health Canada approved.  

 

Key Similarities & Differences between HCL Systems (on & off label)

Key Similarities
  1. For adjustments, focus first on patterns of problem areas (lows & highs) that are not addressed by the HCL system. 
  2. Require CGM & carbohydrate counting 
  3. System toggles basal off, on or up or may also use microboluses in response to CGM and algortihm 
  4. Have different algorithms which determine when to toggle basal/microboluses
  5. Have varying features that may or may not be adjustable while loop is closed
  6. Subtract IOB from bolus calcuator result if the BG, IOB and algorithm determine it is required for safety. 
  7. Can "open the loop" (turn off automated insulin adjustments)
  8. Have different glucose target(s) for use e.g. to address exercise
Key Differences
  1. Algorithms vary. E.g.  The extent of correction given varies. Loop may also give correction solely as temp basals or if chosen, may give it as microboluses (AB= Automatic Boluses) with same amount over 30 minutes as temp basals, but more given up front. (Note, correction bg range is different than target bg range). Suspension of basals in Loop occurs if the 6 hour prediction is to go below target at any point; Tandem suspends basals if predicts low up to 30 min in advance; Medtronic will not suspend if BG is > 4 mmol greater than target whereas Loop will. 
  2. Bolusing varies. E.g. Loop requires the user input absorption time of carbohyrdrate; this can be corrected after the fact and the algorithm will respond appropriately. Users often input one meal as two factions: part of the meal they think will get in faster and the rest they think will get in slower. Other HCL require only the carbohydrate be inputted. A 2 hr max extended bolus is available only in Tandem CIQ; Medtronic AutoMode does not permit extended bolusing.
  3. IOB calculations vary. E.g. Loop IOB is all bolus and temp basal above the programmed basal rate. Other on label HCL usually report IOB as past boluses only.  
  4. Overrides vary: E.g. Loop can have multiple individualized "overrides" preprogrammed or programmed on the fly to address different glucose targets and % basal (they do not use the term "Temp" basal for this. Medtronic and Tandem have 1-2 glucose target overrides (termed differently).  
  5. Features vary: E.g. Loop allows users to adjust all aspects of insulin dosing (ISF, ICR, basals, insulin action, targets, accept or change suggestions from bolus calculator). Medtronic 770G permits ICR and Insulin action only along with one preset option for higher glucosse target. 
  6. Viewing reports may vary. e.g almost all can view CGM reports in their respective software (Clarity, Libreview, Carelink). However viewing insulin for off-label loopers usually involves using  Nightscout or Tidepool. See the heading below called "Resources...". 

  

Types of Off-Label "Looping" 

Many off-label systems are commonly referred to as “looping” but have different set ups and use different algorithms. Generally speaking, there are two major types of algorithms, below. However, each may have different "branches", meaning slightly different versions available where the code and features have been tweaked.  All off-label HCL systems share the similarities of stopping, increasing or decreasing insulin delivery based on CGM, and using the patients manually programmed basal rates, ISF and ICR. Although, how they do that varies according to the algorithms. Some algorithms may give micro boluses instead of increased temp basal rates when increased insulin is needed. They each handle bolusing a bit different. 

“Loop” based algorithms.

The algorithm lives in an off-label  iPhone app. Communications from pump to app occur using a small radio device (RileyLink, OrangeLink etc) 

  • “Loop.” This is its formal name. This 9 minute video "Bolusing for Challenging Meals" nicely represents bolusing options available in “Loop” app. Loop bolusing asks patients to choose an absorption time after they input grams of carb. E.g. 2 hr, 3 hr etc. Some of the Loop’s capabilities are listed in the heading further down this page.
  • FreeAPS: This is a branch of Loop and runs on iphone. This branch is no longer supported e.g .no bug fixes or improvements coming but people can still use it. The video above still applies. There are more individualization options

OpenAPS algorithms using Oref 0 and Oref 1:

The algorithm lives in a "rig" (separate tiny computer) or in a smart phone app. The communications from pump to app occur using a small radio link device (RileyLink, OrangeLink etc) or the rig, if being used. 

  • OpenAPS and can work with Android or iPhone if the patient has a “rig” (a tiny boxed computer to run the algorithm). Generally speaking, the patient inputs their total grams of carb into the pump, gives a partial bolus e.g. 40-60%, and lets the algorithm top up as it sees fit. Some folks just input the carbs with no bolus, and the let system do its thing. This has varying rates of success. There are features to tweak the performance of the algorithm.
  • Android APS is a branch of OpenAPS, and runs the algorithm in an app on Android phone. Communications between pump and app occur via a radio link. 
  • FreeAPS X (note the X) is the newest branch of OpenAPS and runs the algorithm in an app on iPhone only. Communications between pump and app occur via a radio link.

 

"Loop" specific elements that the user can adjust. 

This is for "Loop" and may not apply to Open APS (oref) systems.The following options can be adjusted by the user in the Master Version of Loop App as of now (April 13, 2021, tomorrow may be different.). This means, any of the following can be adjusted by the user to address glycemic control. Sorry for any ommissions! 

  1. Basal rates
  2. Glucose targets for any time(s) of day can be preset and of course, changed on the fly (see Overrides below)
  3. ICR
  4. ISF
  5. Any units that are suggested for meal bolus or correction can be overriden with the units preferred by the user.
  6. Duration of Insulin Action (Settings >Insulin Model > Walsh)
  7. Overrides. These can be preset options or programmed on the fly. These can change glucose target and/or insulin dosing for any duration from 15 minutes up to indefinite. Note: 110% override of insulin means that all basal, and manual boluses and corrections will be increased by 10% while that override is on.
  8. Add Carbs Button and Absorption Time.
    • Absorption time: For each addition of carb the person must choose an absorption time from 30 minutes to max 8 hours (although the dial offers 24 hours, it will roll back to 8 hr max). In other words, every food bolus is an extended bolus of some fashion. In this Master Branch, you cannot give all the food bolus at 0 minutes when entering the insulin through the "Add Carb Entry" button. Instead, 30 minutes is the lowest option, but most of the bolus is given at the time of manually programming, with the rest given as increased basal over the 30-minute period if that is the absorption period chosen. Pre bolusing is important for many people. Less important for those on very low carb diets.  There are different options/methods for bolus delivery in the AB and FreeAPS branches of Loop (see below under Resources for URLs).
    • People can input the meal in sections with different absorption times eg 20 g CHO @ 30 minutes plus 40g CHO @ 4 hours absorption. They can do as many of these as they wish (versus normal pumps that usually only allow for one extended bolus).
    • Backwards adjustments to grams of carb eaten and duration of absorption are possible. e.g. The user can program in 30 g carb with suspected absorption over 2 hr. At 1 hour pc they are trending towards a low and realize they didn’t eat it all. They can go back and edit that they ate only 20 g. The algorithm will reduce the basal to adjust. They could also go backwards in time to change the absorption time to 1 hr.
    • Any CHO entered can be "Saved without Bolusing." The system will interpret this to mean the user ate the carb and didn't bolus. The system will attempt to increase the basal over time to address the carb eaten for the absorption duration chosen...although it will still modulate basal according to glucose readings, so may not give all the insulin a user expected.  This is not perfect. It can still over-deliver and result in hypoglycemia or under-deliver and result in hyperglyemia. Carbs saved without bolusing that weren’t actually eaten require the user to view the history, swipe left and delete, elsewise the current Master algorithm will still give insulin for it as per above.
    • For any CHO entered, the user can look at the progress of the CHO eaten and bolus delivered. e.g. 30 g CHO programmed for 3 hr absorption and currently 16 g has been absorbed with a system-estimated 4 hr 15 min time for full absorption of the 30 grams. 
    • Many users will be surprised to go back through their day's meals to learn how much "estimated" CHO was truly absorbed. e.g. more CHO absorbed than what they ate. In other words, many users may find it beneficial to bolus for protein and fat according to whatever method they prefer or works for them. 
    • Again, these algorithms are tweaked regularly so the above items many not be accurate at the time you are viewing this. Please see the resources below for more up to date documents. 

 

Resources for Off-Label Hybrid Closed Loop (HCL)

Please note that any off-label hybrid closed loop algorithm may be updated and revised at any time, eg. tweaks, new features etc. 

Looping Resources

OpenAPS Resources

  • At https://openaps.org/
  • #openaps and #WeAreNotWaiting on Twitter
  • YouTube Dana Lewis and Open APS
  • A "Rig" or "Edison" or sometimes even a Link (Riley etc) could be used. These are all "boxes" that faciliate communications between pump and off-label looping apps.  
  • FreeAPS X is a new (end April, 2021) looping system for iOS based on OpenAPS  algorithms. FreeAPS X  (the X is the critical part to distinguish from an earlier FreeAPS based on loop code) is not a branch or fork of DIY Loop. The iPhone and radio-link will take the place of the "rig" and entering information into Nightscout. This may still be in Beta testing as of May 2021.

Report Generating Software