Comparing Tandem’s Basal-IQ to Medtronic’s SmartGuard. Which Predictive Low Glucose Suspend is best?

Comparing Tandem’s Basal-IQ to Medtronic’s SmartGuard. Which Predictive Low Glucose Suspend is best?
Comparing Tandem’s Basal-IQ to Medtronic’s SmartGuard. Which Predictive Low Glucose Suspend is best?

If you’ve been looking at Diabetes technology recently, it’s hard not to have seen the recent news from Tandem about rolling out their Predictive Low Glucose Suspend (PLGS) functionality, known as Basal-IQ.

What’s really important to note is that this is the first PLGS-only system to be made available in the US, because Medtronic never bothered with FDA approval of the 640G with SmartGuard, opting instead for the 670G with full closed loop capability (as well as what they classed as “Low Management, which is the PLGS system from the 640G).

As a previous user of the SmartGuard functionality, and one who was impressed with it, I thought it would be worthwhile taking a look at the two systems and seeing if we could determine whether there’s really much difference between the two offerings.

Based on the user manuals, I’ll briefly outline the two systems and then look at which I think offers the best functionality.

Medtronic SmartGuard (640G, 670G)

The Medtronic Smartguard functionality first came with the 640G pump and provided the first PLGS system. It is also implemented in the 670G.

The components required for this system are:

  • Medtronic 640G or 670G
  • Medtronic Enlite sensors
  • Medtronic Guardian Transmitter

The functionality of the SmartGuard system is described by Medtronic as:

Suspend before low conditions
When a Suspend before low event occurs, all insulin delivery is suspended. For a Suspend before low event to occur, both of these must happen:

  • Your Sensor Glucose (SG) value is at or within 3.9 mmol/L above your low limit. 
  • Your SG is predicted to reach or fall below a level that is 1.1 mmol/L above your low limit within approximately 30 minutes.
After the minimum 30-minute suspend time, basal insulin delivery will automatically resume if the following conditions are met, or after 2 hours of suspension: 
  • Your SG is at least 1.1 mmol/L above your low limit. 
  • Your SG is estimated to be more than 2.2 mmol/L above your low limit within 30 minutes.

Complete with these useful graphs to help to understand what they mean:

And when a suspend before low would occur:

As I described in this post, you can use this quite effectively to set up some specific suspend thresholds that really make a big difference in using a pump. 

Obviously, there’s a downside to this, and that’s the Enlite system, which requires multiple daily calibrations, has a short sensor life compared to Dexcom, and, if you’re not covered by insurance, is very expensive. This is why it has had little coverage outside of Europe, even though the PLGS system (if you get on with Enlites) works very well.

Tandem T:Slim X2 with Basal-IQ

Basal-IQ is the first automated control algorithm for Tandem’s T:Slim X2.

The components required for this system are:

  • Tandem T:Slim X2 pump
  • Dexcom G6 sensor
  • Dexcom G6 Transmitter

The functionality of the system as described by Tandem is:

The Basal-IQ Technology utilizes CGM sensor readings to stop and resume insulin based on the current sensor value and a 30 minute future predicted value along with the following four rules:

  1. Insulin delivery is suspended if the current CGM sensor reading is less than 70 mg/dL.
  2. Insulin delivery is suspended if the 30 minute predicted glucose value is less than 80 mg/dL.
  3. Basal Insulin delivery is resumed once the current CGM sensor reading increases
    compared to the previous reading; also known as the nadir.
  4. Basal insulin delivery is resumed if insulin delivery has been suspended for 2 hours in a 2.5 hour window. 

For example: If insulin is suspended for 2 hours, it will resume for at least 30 minutes. If after 30 minutes have passed, rule 1 or 2 above is true, insulin will be suspended again for another 30 minutes.

Along with these graphs showing what happens in rules 1 & 2:

and rule 3:

The main benefit of this system is that it uses the Dexcom G6, which, with the 10-day sensor life and no need to calibrate in those ten days, makes the whole upkeep of the system much easier to deal with than that of the Medtronic. Especially if you’re dealing with a child using the system.

Looking at the two side-by-side

Both sets of algorithms produce similar results, but differ in the way they do this. It’s also worth noting that Basal-IQ uses three data points in its predictions, whereas SmartGuard appears to use only two. 

Suspend Algorithms

The basis of both suspend algorithms is noticeably different, with Basal-IQ setting absolute limits of 80mg/dl (4.4mmol/l) and 70mg/dl (3.9mmol/l) as it’s threshold levels, whereas Medtronic took a relative limit approach, using your own selected low level as it’s datum and then suspending relative to that. 

The key difference lies in the flexibility that this gives you with SmartGuard, allowing for, for example, a higher low limit overnight, and variation throughout the day, whereas you are penned into a single set of limits at all times on Basal-IQ.

Given the issues that people have raised over locked in targets on the 670G, it’s interesting to see that Basal-IQ is getting some great initial reviews, and this is even more of a point when you consider that the PLGS system is available as part of the standard set-up in the 670G, so the tens of thousands of 670Gs that have been shipped could be used in this way if people have issues with “Auto Mode”. 

On the other hand, optimising the 640G/670G SmartGuard does require a bit of consideration and work, and this can easily put people off, resulting in the use of default settings that may not be optimal. In this respect, Basal-IQ may be easier to use. 

Resumption Algorithms

This is where the two algorithms differ quite dramatically, with SmartGuard being significantly more conservative than the Basal-IQ equivalent.

Basal-IQ resumes insulin once the “nadir” or low point, has been moved away from and it detects an upturn in glucose levels, whereas SmartGuard waits until glucose levels are at least 1.1mmol/l (20mg/dl) above the low threshold and you are estimated to be hitting double that level within 30 mins. I can’t find anything in the documentation that suggests that Basal-IQ won’t resume if you’re still below 70mg/dl, so I assume it resumes anyway, as long as you’re ticking up. I’m happy to be corrected if this is wrong.

In practice what this means is that you’ll be getting basal insulin back sooner with Basal-IQ than you will with SmartGuard, which should mean that post suspension glucose increases are limited. Indeed, one of the complaints about SmartGuard has been that it suspends for too long, resulting in “rebound highs”, and this can be an especially tough one when your uptick results in you ending up between 1.1mmol/l and 2.2mmol/l above your low level, in a perfectly safe zone, but with flat sensor readings. This has been known to result in full two hours of suspension, if manual intervention isn’t given.

Other Factors

Aside from the variances between the two algorithms, and people will be able to choose which they prefer, the other key differences are physical and interaction driven. 

Neither pump offers native upload of data for sharing, but at least with the T:Slim, you could use the Dexcom app alongside the pump to provide data to share, where you’d have to go to WeAreNotWaiting for the 600 Uploader to share the 640G or 670G data.

Firstly, the Dexcom G6 sensor technology far outstrips the Enlite/Guardian 2 and 3. Sensor life and removal of calibrations make it much, much easier to live with and change how you interact with a system dramatically. That’s a big win for the Tandem system.

Secondly the phyiscal form factor of the pumps. The T:Slim X2 is significantly smaller than the 640G/670G, and that’s a big deal for a lot of people. It’s also, dare I say it, more aesthetically pleasing, and definitely less industrial looking. Comparing the sizes and weights:

T:Slim – 3.13” x 2.0” x 0.6” (L x W x H) – (7.95 cm x 5.08 cm x 1.52 cm); Mass: 112 grams full

6x0G – 3.78″ x 2.1″ x 0.96″ – (9.6 cm x 5.3 cm x 2.44 cm);  Mass: 95.7 grams (assumed unfilled)

Thirdly, the X2 is an upgradeable pump, so that as and when the next phase of their offering becomes available (Control-IQ, which is full hybrid closed loop), the pump doesn’t need replacing. You just upgrade the software.

Finally, Tandem have made a lot of the user interface on their pumps, and the touch screen that they provide is heralded by many users. I can’t comment on it as I have no experience of using it, but based on the feedback from others, and the often clunky way you have to interact with the Medtronic, Tandem probably takes this one. 

Can you draw any conclusions?

The question was, which one’s best, and for most people, that’s the one you can get hold of, use effectively and turns out to be the most cost-competitive for you.

In the UK, if I wasn’t using a closed loop, and was offered both on the NHS, with the NHS offering funding for the CGM system, the Tandem appears to be the better option, in spite of there being less ability to manage your suspend options.

It’s not just the algorithm, but also the aspects of living with the pump itself that play an important part, and the CGM is significantly better. Similarly, the future-proofing delivered by software upgrades is key.

With the Tandem T:Slim X2 due for arrival in the UK shortly, and hopefully, with Basal-IQ in tow, we’ll need to see access to the Dexcom G6, which is expensive in itself, improving to see notable uptake. That’s one for Dexcom, Partha Kar, Tandem and the NHS beancounters to resolve.


  1. Great read Tim. Thanks for digging into this. Disappointed that the suspend values are hard-coded in the T:Slim – I suspect this would lead me to have more low level hypos using that system as suspends on my MM640G often only act early enough when they happen at more like 90-100mg/dl (5-5.5mmol/L). Great news on the more rapid resumption of basal though. T:slim is very much on my watch list for next time around, so this is a really interesting read.

  2. I’ve used Guardian and Enlite with VERY good HbA1C reductions and no massive hypos. BUT it’s very VERY unreliable with 60% failure at 3 day point.
    It’s just unreliable. It complains about losing SG and lots of other errors.
    I spend more time waiting for replacement sensors than using it .
    £250 for 30 day cover expensive.
    £100 for 28 day cover with Libre.
    And not to mention my flipping sore fingers, waking several times a night to check BG and 10% failure rather than 60% yes I understand that it doesn’t alarm YET but I’m sure that’ll be sorted in R&D.

  3. It is true that Basal IQ resumes insulin on an uptick of BG even when the reading is below 70. And it suspends insulin whenever it projects below 80 in the next 30 minutes. So it certainly suspends insulin if BGs are headed down and the current value is below 70! And if headed up, or even becomes steady, and below 70, it still resumes insulin. So I can’t yet find any answer to how the “suspend below 70” rule does anything extra or exceptional to the other rules. It seems to be irrelevant given the rule priorities. Can anyone discern a situation where not having the “below 70” rule would cause the pump to behave differently than having it?

  4. I have experience with both the 670G and the X2…I can attest that the care and feeding associated with the 670G is excessive and obtrusive to everyday life-so much so that after three months I switched to the X2. The 670G does indeed bias toward higher glucose levels and does adaptively learn using the past 7 days of data as an adjustment point. With this in mind, one can see that it could take 6-9mos for the 670G to “learn” and fine tune the individual user. It does require about two weeks of diligent careful tuning to get a good Auto Mode starting point..then proceeds in Auto Mode without using the two weeks of work one put in. The game of adjusting ISF and correction factors all starts anew. After the initial two weeks of basal, ISF and carb ratio adjustments…I settled on two simple basal profiles and ISF=1:65 Carb ratio=1:12 and 1:15 at night and IOB duration at 3.5hrs. After a month on Auto Mode ISF=1:30 Carb ratio=1:8 all the time and IOB duration was at the minimum limit of 2hrs…these are almost T2D settings. But, this is where I had to go to get the 670G to react aggressively enough to keep me in range 80+% of the time. All this with the constant calibration of the CGM sometimes every 12 hours as designed and every 6 hours the first day a new sensor was inserted. And yet other times it would just ask for a fingerstick but no calibration…most of these happening between 1am-430am. On the days a new sensor was inserted, I was guaranteed at most 2hrs of sporadic sleep. Then, the horrible accuracy of the Medtronic CGM…90 points off reading 137 with a fingerstick of 48!! Or a CGM of 177 and a fingerstick of 254…glucose readings that demand very different actions in both cases. The accuracy never really improved to my liking even with the constant calibration requirements at very inopportune times. Then there was the insertion of the 670G CGM in the back of the arm…not entirely intuitive and impossible to do without a mirror…no changing the sensor at a backcountry campsite if needed!!

    I can confirm as well that the X2 will resume basal insulin with BG below 70! This is very concerning. The Basal IQ algorithm looks at 3 out of 4 of the readings to make a suspend/resume decision. In my opinion, this is much too simple. It makes no effort to predict the slope of rise for the resume while it will use the fall rate to suspend early to try and prevent hypoglycemia but will then resume at the first uptick then suspend again with two down ticks. It does not take into account the IOB for the resume function either. This seems rather elementary in my opinion. IOB plays a major role on where the BG is headed in the next hour or more. Incorporation of the IOB into the resume function would have prevented several <50mg/dl hypos for me thus far. These are particularly difficult to recover from when the X2 will just not stop the suspend/resume cycle. On a positive note, the high and low alert points, as well as the target BG, are user defined…this is a plus. The accuracy of the Dexcom G6 is very good. My current sensor has a total of 17days and is less than 5% different than the occasional fingerstick spot check. Wearing the X2 versus the 670G is night and day. The X2 is barely noticeable on the belt and under clothing. Tandem offers a 43" infusion set…a must for having enough tubing to have the pump in hand and bolus or check settings. Medtronic longest infusion set at 32" is severely lacking clothed or unclothed. I was constantly pulling the pump off the dresser just trying to get into a favorable position to tuck my shirt and pull up my pants at the same time without stressing and pulling the infusion site excessively.

    I hope this helps to answer and clarify some of the questions and operations between the two pumps. I am anxiously awaiting Control-IQ software update for my X2…the 670G is mothballed in the box with 3 months of sensors wasting away.

  5. A great read I’m on Medicare and waiting out my warranty 2021. I use the 670G as a stand alone with a Dexcom G5. I was involved in the animas buy out forced to use the 670G I do not like it. I’m not experiencing the miss haps as one who uses the Medtronic sensors. It’s just I’m experiencing so many lows during the night and have had my delivery settings changed several times.. I could good on and on but I can’t wait to get rid of it. I’m having to eat two times during the night and NEVER had this problem with my animas pump. Do different pumps have major differences when dispensing insulin? I mean a unit of insulin shouldn’t be any different. Thanks for your help and the break down of bothpumps TSlim is for me. Oh yea the 670 is too big too heavy and the cartridge sticks up above the top of the pump and digs into your skin yuk. The clip that attaches to your belt or waist band is huge and sticks out like a huge close pin. A terrible design. I threw it away I’m using the one off my Animas Ping pump.

  6. To clarify the MiniMed 670G system actually uses the Guardian Sensor 3 (different from the Enlite sensor which was compatible with the MiniMed 530G system.

    • That’s true, however most people’s experience with Smart guard was 640G, which was very much enlite driven.

Leave a Reply

Your email address will not be published.