Dexcom trend arrows

Your eyeballs are better than the trend arrows. Sometimes it’s obvious to me that the curve is leveling off, even though the trend arrow is continuing to show a rise/decrease–if I wait a little longer, the trend arrow will change too. I never dose off of trend arrows alone, I use them as a nudge to examine the curve visually, and that’s what I make dosing decisions off of.

1 Like

Jay6’s post has really valuable info. When I started using a CGM (Dexcom G5), I did not immediately appreciate two of Jay’s key points: Finger Stick (FS) tests with your meter are measuring glucose in the blood, while the CGM system measures glucose in the interstitial liquids. When BG levels are changing, it takes awhile for the CGM-measured interstitial liquids’ glucose level to “catch up” to your actual BG level.

The implication of that is try to avoid FS tests when BG levels are changing. This time for BG to “move” into interstitial liquids is probably why Jay6 mentions only +/-20% accuracy – because people do compare FS and CGM readings while BG levels are changing.

When I am able to do a FS test only when BG levels are flat, then I personally expect accuracy closer to +/-10%, and often see somewhat closer readings.

By the way, I tend to be suspicious of CGM readings during the first 12-24 hours of a new sensor (though my G6 sensors are an improvement over G5). It likely comes down to the very minor trauma of inserting the wire under the skin. So most of my FS tests now are to make sure the new sensor is correctly calibrated, and after that first day of a new sensor, I rarely feel the need to do a FS. But I am willing to follow Dexcom’s advice – if the CGM reading doesn’t “feel right,” then do a FS test. Then add in Jay6’s advice: try to avoid making important treatment decisions if you can, while levels are changing.

Sally7 also has key advice: “I have learned to just watch and wait.” My Endo has been advising that for years. And we all know the yo-yo ups and downs come from over-correcting one direction, then over-correcting the other. I know being patient is difficult, but it is very good advice.


@Jim10 You and Sally are right, patience is a virtue. :slightly_smiling_face: I am a somewhat patient person in general, but not as much in this case. I have been known to “rage bolus” even when I know that I probably shouldn’t.

Same here. If I’m ever aware of the trend arrow, it’s just in a passing “That’s nice” way. Much more interested in what the line for the previous half-hour or hour shows.

Absolutely, and I think you get about 90% the information you need about that from the graph and only a marginal 10% from the arrows. If the graph is trending consistently down or up over enough time to indicate a real trend and not just a transient blip—say 20 minutes—then you’ve got something you can consider responding to, especially if the intervals are growing, indicating the trend is accelerating. The arrows are a kind of shorthand representation abstracted from that information, but for that reason can be more of a hindrance then a help. They give an impression of cybernetic certainty that they don’t actually possess. This is one of the reasons I use the 3rd-party widget, Sugarmate, which replaces arrows with the numerical change since your previous value atop the graph line. Feels more like it’s saying “Here’s what I’m seeing right now” instead of presuming to tell you where you’re going. If that number is big enough, you can look at the graph and draw your own conclusions.

Sugarmate piggybacks on the Share stream, and if you wear a Versa2 Fitbit or higher you can install a clock face that incorporates it as well. This screenshot is from my laptop. Very handy little widget (Mac and IOS only, alas).

But just to emphasize, yes, the HUGE difference CGM makes is that it’s giving you a reading every 5 minutes, so you CAN see trends, and it’s the trends that tell you what’s happening and what to do about it (if anything). If a finger stick is a snapshot, CGM is the movie. It’s just that you have to use a certain amount of judgment, and take those arrows with a big grain of salt.


I agree with this analysis. I also agree with @Robyn_H’s observation that the arrows are not predictions.

@DrBB, I can’t help noting that your three-hour low 80s line that you posted is a thing of beauty! It almost makes you want to skip a meal and avoid the risk of adding a food and insulin disruption to this oasis of metabolic tranquility. Nice work!

1 Like

Hah! Thanks. It’s actually been dipping a notch or two into the 70s all morning and annoying me with “low” alerts (threshold is 80). The fingerstick I did a little while ago indicates I’m actually around 90, which is a bit too close to calibrate on, so I’m just putting up with it. I also credit Jardiance, which I’ve been taking off-label for a couple of years now and which has a remarkable effect on my inter-meal stability.

1 Like

RIght. I’m so glad that I posted this topic, though I wish that I had done so a few months ago. All of these comments are great. Knowing that the arrows are not predictions, and, in fact, add little value over the graph is immensely helpful. I have been trying to retrain my brain to read the Dexcom this way over the last few days. I think I might look into the sugarmate app.

I agree! That is a graph to aspire to.

1 Like

Amen. . .This is exactly how I treat it most of the time.

It seems that, despite misinterpreting the trend arrows, the Dexcom is helping me with control. Yesterday I got my first A1C after starting Dexcom/Tandem - 6.0!!

While I know that 6.0 is average, or even a bit high for many people on this forum (and also not necessarily the best indicator of overall control), this is an accomplishment for me. My A1C is usually in the 6.5 - 7.0 range. So… woo!! I’ll take it. :slight_smile:


The trend arrows are only a part of the picture with a CGM. I also rely on the graph to complete the picture which way my BG is going. I have on many occasions seen the trend arrow going up but by looking at the graph I see it peaking after a meal / bolus so no need for any correction. If I see both the graph and the trend arrow both on a steep increase and heading out of range then I will do a correction.

Comparing one person’s A1c number to another person’s is not a good comparison. There are just too many variables that confound that comparison. Comparing your earlier A1c’s to your current one is a good way to garner feedback.

Your drop from the 6.5-7.0% range to 6.0% is truly impressive! You deserve much credit for this.

1 Like

The meter confirmation is exactly what I do also, especially when I see a vertical up or down trend. This has served me well enough.

As has been noted many times, a graph of the first 12-24 hours of a new sensor can look wacky, and no one should take significant actions based on the CGM reading or trend arrow alone. The BG reading may be spot on with the meter, other times it is 20-40 points off. It’s frustrating when the crazy readings happen during the night.

Welcome to the club of worry worts. none of us type1 or type 2 have all the answers. We glean a little info from every post that hits this forum.
I have been a type2 for two many years, and have been using Dexcom on my phone sense 2017. In fact I got an Iphone just so I could use a Dexcom. For some reason Flip Phones don’t work.
Besides being a type 2, I also have chronic Kidney impairment but not on Dialysis YET.
The Dexcom has saved a lot of finger pricks over the years, but still refer back to the Glucometer when I fell something is not right with the trend in the graphic. Usually I use the meter, if I see a sudden rise in glucose level that I can’t explain.
Now this is what works for me and maybe not others. I don’t full trust the Dexcom glucose levels so I will uses my Glucometer to check. If I ( again person presence) see more then 40mg/dL difference , I will recalibrate.
Also over the years prior to and during Dexcom I have noted I can have more then 15mg/dL from finger to finger , hand to hand . In which case (again personal choice) I will average the readings and calibrate the Dexcom.
I too use a third party app on my Fitbit Versa. In fact I use three different source for readings. We all know Bluetooth and be a fickled Mistress. I use the the Dexcom meter, my phone and my watch.
I have always had the problem of my BS dropping suddenly while I sleep . yes the alarms wake me. But over the last couple of years I will wake up several time a night just to look at the meters. This is where recently my watch comes into play, I don’t have to fumble around my night stand to find one of the meters.
Here’s hopping you will eventually come upon a system that will work for you.
Oh you may also note that as you come close to the 10 day window for changing a sensor that may or may not give you sensor loss messages. Usually for me they only las an hour or so. I understand stand the G7 has corrected this.
Again good luck on your in devours

@Jim10 ,

Thank you for the kind words. There are some places needing illumination:
[1] The overall accuracy of FS meters is only ±15% in 95% of meters compared to labs and ±20% in 99% of comparisons of meter to laboratory findings. So our meters are not the sharpest knives in the drawers.
[2] Research published in 2020 reports the G6 accuracy is within 16% of laboratory findings over a variety of clinical conditions including HIIT.
[3] My personal experience during hospitalization for a CABG, demonstrated a consistent 15mg/dL separation comparing the G6 with hospital lab glucose over all phases of the admission. It is of note, my anesthesiologist mandated the G6 during my hospitalization.
[4] With my multiple prescriber’s advice and consent, quarterly and other labs where a glucose is run have demonstrated to my prescriber’s approval that calibrations are unnecessary. I have not done a finger stick for more than two years and am amazed at reports about deviations in the first 24 hours of new sensors or similar issues in the last 24 hours. I must be the exception and have great sensors.

@Sally7 Sally’s comment “I have learned to just watch and wait.” merits what I have shared on another site to new users of the Tandem CIQ enabled pumps with CGM integration. My thoughts there are centered around PATIENCE and TRUST. Here is an excerpt from those concepts.

[1] Start with ‘patience’, knowing several pieces of information or facts.
[1a] insulin does not act instantly if given by pen or syringe. The same is true for a pump. Insulin takes time to work.
[1b] CIQ takes time to adjust insulin according to the ‘software’ in the pump. Pick a set of ‘Emergency Action Points’ or EAPs. These are points where safety intersects waiting. My points vary depending on my perception of near future events related to activity and range from 150 to 200mg/dL for HIGH and 59mg/dL LOW. For example, if you are going low, watch what CIQ and CGM are doing in the Tandem t:Connect app. You may not need to act as CIQ will throttle back basal insulin allowing the body to bring BG up on its own without intervention. Be careful. Because of the way CIQ cuts back basal, any carbs may cause BG reported by the CGM in the next hour to rocket upwards.
[1c] work with your local Dexcom rep and pump team to learn the display and interpretation of CGM data and the indicator arrows. Case in point, what does a real COMPRESSION LOW look like? How will the CGM respond? What are the signs the CGM has recovered? What action do you need to take during the recovery? Patience is part of the ‘knowing what to do’ and sharpened with ‘knowing when you should act.’
[1d] master the differences between the various physiological parameters in the game.
What is the difference between a finger stick, medical lab glucose, and the CGM? What is the difference between the Dexcom GMI, a Sugarmate A1C projection & a real lab A1C? What do you need to remember when juggling all of this information? Here patience is ‘knowing what the numbers mean’ and make the ‘best practices’ decision because you are ‘knowing how to use the CGM tool & these numbers to your advantage.’

[2] The second lesson is TRUST. Just like all of the thoughts above. You will need to learn to trust your your tools, your CGM & pump, and your gut so everything works to keep you in a good position.
[2a] Because I had been taught by my pump team (endo, office nurse CDE pump trainer, Tandem nurse CDE pump trainer, and Dexcom nurse CDE) when a COMPRESSION LOW woke me up, I looked to Tandem’s t:Connect. I saw the red line and knew CIQ had turned off the basal insulin.
[2b] Trust the pump programming is correct as reflected in the combined CGM & CIQ information. Tandem’s t:Connect displays integrated data from both the insulin delivery system (the pump) and the CGM. The CIQ software integrates the data and adjusts the insulin delivery. Observing this interaction and consistent interaction is a place of trust.
[2c] When the CGM dots disappear, the CGM is checking the sensor and interstitial fluid glucose and may not restart until it was sure of itself. It is wise to be patient with the system. Patience is derived from trust.

@beacher, you and @cardamom are correct. The arrows are not for dosing. However, in the USA, the G6 is approved by the approval bodies and Medicare for insulin dosing.

@DrBB , @Brianna1 , @Terry4 , your comments are gems.

Hope this helps.

My point isn’t that you shouldn’t use the CGM for dosing. I do all the time (I almost never test anymore). I just don’t use the arrows–I look at the line itself which is far more informative since it has curves and other things not captured by the arrows so I can usually guess if it’s leveling off or about to go up even steeper or if the arrow might be swayed by a recent errant number.

1 Like