Is Time-In-Range (T-I-R) a good/the best metric for assessing BG Control?

There have been many comments recently that Time-In-Range (TIR) is becoming the preferred metric of determining management of blood glucose (bg).

I have found that there is no one metric that is particularly effective. Later I will attempt to explain which of the two metrics combined are the most effective (in my opinion). Do you remember “My Cousin Vinny” where the judge (Ted Cassidy) made Joe Pesce start every sentence with “In my opinion”? Assume all my statements begin with that phrase.

As an illustration let’s assume that for illustrative purposes the ideal situation is that BG remains at 115 or reasonably close all the time.

T-I-R. If you use the “standard” T-I-R, the range would be 70-180. If your BG was 180 all the time your T-I-R would be 100% which is as far from ideal as possible while still being in range. Luckily we don’t need to use the standard range and can set our own ranges which could make T-I-R a much improved metric.

Average BG. Let’s assume your BG was 70, 50% of the time and 160, 50% of the time. Your average BG would be 115 but you would be close to low 50% of the time and close to high 50% of the time. Not a great metric.

Standard Deviation If you were 70 or 180 or 150 all the time your standard deviation would be 1 but that would not represent a meaningful measure of “good management”.

T-I-R and Average BG Combined Same explanation as “Average BG”. Your average BG would be 115 AND you T-I-R would be 100%.
but you could be close to low 50% of the time and could be close to high 50% of the time.

T-I-R and Standard Deviation Same explanation as “Standard Deviation”. This is the most misrepresentative of "good control. If you were 180 all the time your T-I-R would be 100% and your standard deviation would be 1. Perfect control. Hardly.

Average BG and Standard Deviation. This I believe is the most meaningful combination. If your average BG is close to you ideal BG and your standard deviation is low that you have good BG control because the average is good and your variation (standard deviation) are narrow. If the average is due to an average of mostly highs and mostly lows then you standard deviation would be high. So it requires both metrics to be close to ideal.

In my opinion, a Low Average BG and Low Standard Deviation is the combination we should shoot for and T-I-R (unless it is a fairly narrow range) is not worthy of the high praise it is currently getting.

Other opinions?


You have some very good points. I believe TIR combined with BG/A1c are both important. I think if you have tight control with a TIR range the SD follows. I have never tried to do anything with my SD, but the narrower and tighter control I have over my TIR range, the SD then stays lower. The average BG we can use at home, (but by a sensor that can be off,) the A1c is a lab measurement over time, so it will usually be more accurate number. BG level at home is what I use to judge my A1C to see how I have done over time. Sure my TIR can be good, but it doesn’t guarantee the levels I am at on average. So I always look at my BG average.

Your point about if your BG average is close to normal, and your SD is low, that you then have good control is very very valid. But I think it’s harder to aim at a lower SD? Sure you can prebolus and I am a firm believer in the timing of dosing, but how do you judge if you went up to 170 for 5-10 minutes that how is that influencing your SD? . I can’t tell what I did wrong looking at SD as easily as I can look at TIR and go I spent 3% of the time over 160 and usually it’s 1%, I better work on my higher numbers. I just think looking at TIR is easier to tell where you maybe could improve and what you can do about it.

We all can set our TIR at what we want to tighten control. The doctors use a range they are hoping that most of the people can do or will aim for. We all know a huge percentage don’t.

And then let’s not even get into if your tech goes haywire your “graphs” go haywire too. (3 sensor failures in a row here)


I use Tandem CIQ so I can see all three everyday or over any period of time I want. If I keep my BG low and try not to vary it too much my TIR will almost always be 100%. My goal is an average of 100 and standard deviation of 15% but I haven’t achieved that yet (except for daily).

1 Like


If I remember correctly, Judge Chamberlain Haller was played by Fred Gwynne AKA Herman Munster …

However, while TIR is not perfect, it’s moving in the right direction.

Now, I probably never seen a day of 100% TIR … I’m more of a 80-85% guy. Nonetheless, clearly if one day, I spend 6 hrs at 185 mg/dL which results in a TIR of 75% and the next day also spend 6 hrs above 180, but get all the way up to 350 mg/dL, I expect we would all agree that the first day was better than the second.

As statisticians have known for some time, mean and standard deviation ARE a better way of describing “level of control” … but are less easily conveyed than something like Time In Range. After all, many endos continue to focus on A1C … despite deficiencies in that number.

I continue to be amazed by the level of control demonstrated by many of you on this forum. Bravo!!!


Hi John,

Thanks for the correction on Fred Gwynne.

Re: your example. I agree the first day was better but you could tell that by looking at either Ave BG or STD deviation as both would be higher on the second day but T-I-R might be the same. Same as if you were out of range 100%. They would both be 0% T-I-R in range but you couldn’t tell which day was better or worse looking at T-I-R.

I spend more attn to Std Dev because if my BG ave was the same both days and I was 100% T-I-R both days, then STD Dev tells me I had tighter control if the number was lower one day versus the other.

I would say the Std Dev may be preferred by those who have tight control and T-I-R may be preferred by those whose control is not as tight despite their best efforts. I just wanted people to know that T-I-R has its limitations.

The Standard Deviation of a constant (a value that does not change) is equal to zero (the deviations of its spread is zero).

Hence if your blood glucose was always the same you’d have a SD of zero.

Okay, thanks

The International Consensus on TIR allows for customization. For TIR, I use 65/140. I abuse target range in my pump at 90-95 depending on time of day, using it as the aim point for bolus insulin. Alerts at 75 and 120 as guard rails.

The Consensus TIR is less the 4% of the time low and less 1% of the time below 54 (inclusive of 54?). It is definitely not never low, something that the endo clinic I go to has a difficult time accepting.

I set tir at 65 to 150. I could tighten it up but I rarely use it.


I certainly agree with you that SD is an exceedingly important number and captures a lot in a single number. However, @William7 makes an excellent point: In many respects the most useful part of TIR is looking at the details of Time NOT In Range. In particular too much time below 70 mg/dL is not a good thing. Moreover, that is information that SD can’t provide. (Yes, I know that if one ASSUMES a normal distribution, SD allows you to calculate time below X … but I’d wager than none of us have a normal distribution).

In general, I would wager that all would agree that a lower mean is better, a tighter distribution (lower SD) is better, but too much low time is not good.

Have a good day.


Lucky for me I have a dexcom and I wear it all the time, this way my doctor and I can look at the specific data. The highs, lows the TIR. I think concentrating on any single indicator will miss something.
Doctors like to have that a1c or TIR so they can quickly approve scripts.
It’s a double edged sword, my doc sees me in the mid 60s and tells me I need to fix that, but I don’t mind being there sometimes. It’s gotta be a collaboration

1 Like

I’m starting to see a new acronym in use - TITR. It references blood glucose levels between 70-140. My endo’s office is using the term and I saw a short Diatribe article about it. 'Bout time.


I still like TIR better for me and I still think it’s just an easier number to judge for a lot of us. I have great control. But here are my stats.

I am generally 96-100% between 70-160 and 92-94% TIR between 70-140. I have maintained an A1C of around 5.1%. I aim for a TIR of at least 96% between 70-160. My SD on the 100% days is usually 16-22 and the days I have 96-98% TIR it’s usually 20-25. I only look at my SD as an after fact. What I look at is when I’m not in TIR. What caused those highs and lows. (almost always 0% very low) As long as I’m not having an over abundance of out of TIR, I’m fine with my splurges and my failures.

I am heavily reliant on my CGM so I can catch and react to my trends. And all my graphs are a mess now lol… I had 3 new sensor failures in a row that were hugely erratic. I deleted the ap and just used the receiver and I think that’s solved it. I am going to re download the ap today. Maybe… But my SD said I was at 21 during that time, and my TIR says I sucked with 5% lows. That would be the night that it said I was under 40 the whole night, I ended up turning my phone off because it wouldn’t take a calibration and I wanted to sleep. And it says I was under 50 a lot the next few days of sensor failures. And almost all of it is on my graph. Sigh…



Control is the effect of management. No single number can ever indicate quality of management. Meaningful metrics are the basis of good management, which is about making decisions. Time in Range can’t be used to make management decisions. A good management decision has a metric that is known to directly and sensitively respond to that decision. If you can’t measure what you are doing you are not managing.

Management decisions require multiple statistics, like standard deviation of a process or product from a desired value, reject/defect rates, incremental cost of output or improvement, and net profit.

A1C has the sole advantage of being a single lab test number. It is a cheap and simple test but is useless for making high quality effective decisions. It’s analogous to body temperature because it doesn’t tell you what is causing it to be abnormal. So all you can do is guess and treat the gross symptom.

TIR is only a slightly better metric than A1C because it shows statistically how much BG is “in control” rather than being a simple average. But TIR has a weakness similar to A1C. It doesn’t indicate variability and by itself can’t be used to make decisions. Like a grade in a subject, it doesn’t indicate strengths and weaknesses. High TIR means little if it is accompanied by Trips to ER.

imo BG vs time in the form of statistics provides the best metrics.

The best set we have today is the Ambulatory Glucose Profile, but which numbers you focus on depends on what’s most important to you, how good your quality of management is.

I ranked metrics by priority. "Lows can kill me quickly, highs will eventually make me less healthy. " I also know the limitations of Control IQ, so my priority is to stay in range where it can handle things reliably and well. I’ve been making systematic lifestyle changes based on accomplishing that.

So the number that I first considered most important was TBR2, (percentage) time below ~58mg/dL. TBR2 by itself is a metric of danger and a gauge of probable stress. If it isn’t zero, then the person is in danger or the measurement is defective.

But I didn’t stop with that number. I looked at the supporting data to see when those times were. Time of day can be used to evaluate how likely the person is to be able to react to that danger. I correlated the pattern of lows with likely causes and addressed those causes.

After that I looked at TBR1, time below 70 mg/dL and did the same thing.
Then I did the same thing with TAR2 and TAR1.

Now I’m tightening my ranges. (BTW, My A1C for this period was 5.2, not even close to that estimated from my mean or median BG. )


@rcarli I like where you are going with this. Appreciate the effort that went into your analysis.

IMO :slight_smile: many diabetics do not even take the time to actually look at the metrics to this level of detail. This, and forums like it, probably represent the most informed and comprehensive people on the planet as it relates to the topic.

Personally, I watch the numbers to include TIR, EaG, A1c and all the rest, but I try to maintain the perspective that each of us is dealing with a similar but somewhat unique situation on a daily basis. I like to say I am building my own little private database that applies to my particular case. I am still new to the game but I also factor in how I feel, sleep quality, water consumption, overt symptoms and my general mental disposition. I use all of that data to create the control profile then look for indications that I am trending inside of it, if that makes any sense at all.


A terrible disease but I’ve learned to live with it. Complications from not being in tight control and stress from work. Started much improved control when I retired. I liked and appreciated your comments.

1 Like

Same here @rcarli, retirement let me improve my control. I am amazed I had seemingly decent control when working. I was in the low 6’s for A1c. I am convinced that I probably had a lot more highs and lows in that number as I did not use a CGM back then.

Every different way of measuring and looking at problems has costs and benefits.

I want to emphasize that HOW individuals trouble shoot system behavior changes over time. For example, you might be traveling overseas in a month and you might be traveling alone. You might have some really specific concerns about that. That might impact your risk analysis and how you want the system to behave. Your goals might change for system behavior.

Whatever analysis you perform should be flexible enough to allow you to learn and achieve your goals…whatever they are. I don’t like to “tie down” how I do my analysis. I like to leave the world full of open doors to look at problems from different perspectives because I think that has helped me over a long period of time in different situations. It helps me not “tie down” my physical body to one specific algorithm that determines what me and my body are and are not capable of.


Every one has made some good valid points, and the discussion is very educational for me. I try to be a perfectionist, but never seem to achieve it.
TIR is valid for a vast range of people - picture a bell curve. The center of the curve are those who have difficulty or lack the knowledge that is exhibited by those here. The right side of the curve are those who have extreme difficulty trying to maintain BG. The left tail of the curve are us, and try to understand and do better. Now, the endo needs a standard to advise and guide patients that is uniform in nature. They can’t/don’t have the time to get into the daily curves; they instead focus on when lows occur (because for most it is dangerous), and how to adjust basels or long acting insulin doses. They don’t want patients to die or have trouble.
So, TIR is useful for the vast majority in maintaining health, but not probably as useful for the few. Adjust/tailor the TIR to benefit you.
Oh, btw, the SD can be fitted to a regression curve to gauge control; hence a SD of 1 is perfect on a straight line curve.

Hoping to live long and prosper.


Coefficient of Variation (%CV) – standard deviation divided by mean glucose (times 100 to produce a percentage) – captures some of what is discussed above and can produce a pretty good single indicator if that’s what you’re looking for. In general, the lower the %CV, the better. Most of the literature I’ve seen suggests that a %CV below 33% is considered a marker of “stable” glucose control. For what it’s worth.