Your CGM report: time in range, GMI and the AGP graph

A continuous glucose monitor, or CGM, is a small sensor worn on the back of your upper arm that measures your sugar every minute of the day and night for about two weeks, and sends the readings to your phone. At the end of it you are handed a printed summary called an AGP report — the ambulatory glucose profile — which is several pages of percentages, a coloured bar, and a grey ribbon of a graph that almost nobody explains.

Like the lipid profile and blood pressure, this is not a report about the heart itself. If you have read how your heart works, sugar belongs with the things that are wearing the plumbing out — running high year after year damages the lining of every artery in the body, including the three pipes that feed the heart muscle, which is why a cardiologist takes as much interest in it as your physician does.

This page explains what a CGM actually measures, what it can and cannot tell you, and what every line on the AGP report means.

Being your own dietitian

The reason a cardiologist reaches for a sensor has very little to do with technology.

A person with diabetes whose sugars are not coming down is usually doing their best. They have been given a diet sheet — the same generic diabetic diet that everyone is given — they are taking their tablets, and they believe, sincerely, that they are eating carefully. And their HbA1c is still 8.5%. Somewhere between what they believe and what is happening, something is not adding up, and no amount of asking in the clinic will find it.

A CGM finds it, because it lets a person become their own dietitian.

Not in the sense of adjusting medicines — that is not yours to do, and this page comes back to it. In the sense that matters day to day: which foods, cooked in your own kitchen, in the portions your own household serves, send your sugar up — and which do not.

That question has never had a general answer, and this is the part that surprises people most. A large study fed identical meals to hundreds of people while measuring their sugar continuously, and found that the response to the same food varied enormously from one person to the next — while each individual’s own response to it stayed steady and repeatable. One person’s sugar climbs after a food that leaves the next person’s completely flat.

So it is entirely possible — and common enough in clinic to be unremarkable — for your sugar to jump after two chapatis and stay level after a bowl of rice, when everything you have ever been told says the opposite should happen. Your neighbour may well be the other way round. Both of you are right about yourselves and wrong about each other, and until somebody measured it, neither of you could have known.

That is what the fortnight buys: not a diet sheet, but your own list. The dishes that do not move your sugar, which you can go on eating without anxiety. And the ones that do, which are now worth thinking about — swapping, reducing, moving to a different time of day, or pairing with something else.

What the sensor is, and what it is actually measuring

The sensor is a disc a little smaller than a two-rupee coin, stuck to the back of the upper arm, with a filament finer than a hair sitting just under the skin. It is applied with a spring-loaded applicator in a second, and the great majority of people report it as painless. It stays on for up to 15 days, through bathing and ordinary work, and reads your glucose every minute, sending the number to a smartphone app or to a small separate reader.

One point of mechanism is worth understanding, because it explains almost every confusion people have with these devices. The sensor is not measuring your blood. It sits in the fluid between the cells just under the skin — the interstitial fluid — and glucose takes a few minutes to move from the bloodstream into that fluid. So when your sugar is changing quickly, the sensor is reading where your blood was a few minutes ago.

That single fact accounts for the two most common complaints: that the sensor disagrees with the finger-prick meter, and that it seems slow to show a rise after a meal. When sugar is steady the two agree closely. When it is moving fast — climbing after a meal, or falling during a hypo — the sensor legitimately runs a few minutes behind. Neither device is broken. When it matters urgently, and particularly when someone feels low, the finger prick is the one to trust.

The version most used in India today also carries alarms, which can be set to sound when the sugar goes above or below chosen levels. This is not a small feature. An alarm is what turns a night-time low from something discovered a fortnight later on a printout into something that wakes you at the time.

The two weeks are only worth having if you eat normally

Almost every other test asks you to prepare — fast overnight, avoid exertion, come with an empty stomach. This one asks the exact opposite, and the instruction is easy to get wrong precisely because it sounds too relaxed to be important.

Eat exactly what you normally eat. Your usual household food, cooked the usual way, in the usual portions, at the usual times. Your usual tea, with the usual sugar in it if that is what you take. Do not eat carefully because you are being watched.

The whole purpose is to learn how your daily routine kitchen food behaves in your body. A patient who spends the fortnight on boiled vegetables and one chapati produces a beautiful report — flat lines, excellent time in range — that describes a diet they will not be eating next month, and teaches neither of us anything at all. Two weeks of that is ₹4,500 spent on finding out that eating almost nothing keeps your sugar down.

Three other things make the report worth reading:

• Note what you ate and when. The app allows this, or a piece of paper will do. A spike at 9 p.m. is only useful if somebody can say what was eaten at 8:30.
• Keep the phone with you if you are using the app, so the readings are captured rather than lost, and keep the alarms switched on.
• Wear it through an ordinary fortnight — normal working days, not a holiday or a week of travel and hotel food, unless that is genuinely how you live.

Reading the AGP report, line by line

The AGP report is standardised, so it looks broadly the same whichever sensor was used. The first page carries a block of percentages, then a coloured bar, then a large grey graph, and then a page of small daily charts. Here is what each of them means, what the target is, and — which is the part that is never explained — what each one is really telling you indirectly.

The targets below are the international consensus figures for most adults with diabetes.

Two caveats before you read your own figures against that table. These are the targets for most adults with diabetes; for older people, and for anyone in whom a low sugar would be especially dangerous, the targets are deliberately relaxed — often more than 50% in range — with an even tighter limit on time spent low. And a target is a direction of travel. Moving time in range from 40% to 55% is a real and worthwhile improvement, even though 55% is still below the line.

Time sensor active

This is the first number on the page and the one nobody looks at, and if it is low, nothing else on the report can be trusted. It is simply the proportion of the fortnight during which the sensor was actually recording — data can be lost if the sensor falls off, if it fails, or if you are using a device that must be scanned and you did not scan it often enough.

The rule is at least 14 days of wear with 70% or more of the data captured before drawing conclusions. Below that, the percentages on the rest of the page are describing a sample too small and too patchy to represent your usual life.

What it is indirectly telling you: whether this report is a description of your fortnight, or of the bits of it the sensor happened to catch.

Average glucose and GMI

The average glucose is exactly what it sounds like — every reading over the fortnight, averaged. The GMI, or glucose management indicator, is that average converted into a percentage that looks like an HbA1c, so it can be compared with the number you are used to.

The confusion worth killing here is that the GMI is not your HbA1c, and it is not supposed to match it exactly. They measure different things over different periods. GMI comes from two weeks of sensor readings. HbA1c is a blood test measuring how much sugar has attached to your red blood cells over roughly three months, so it is also affected by anything that alters the lifespan of a red cell — anaemia, iron deficiency, kidney disease, a recent transfusion. A gap of a few tenths of a per cent is expected and normal.

What they are indirectly telling you: your overall sugar burden — and where GMI and HbA1c disagree by a lot, that disagreement is itself a finding worth taking to your doctor.

Time in range

This is the number that matters most, and if you read only one figure on the report, read this one. It is the percentage of the day your sugar spent between 70 and 180 mg/dL — and the target is more than 70%, which is about 17 hours out of every 24.

Its value is that it says something an average cannot. Two people can share the same average glucose and the same HbA1c while living completely different days: one steady and level almost all the time, the other swinging from 60 to 300 and averaging out in the middle. The averages are identical. The days are not remotely the same, and neither is the risk.

What it is indirectly telling you: how much of your life your body is actually spending in a safe zone — the closest single number there is to “how is the diabetes really going”.

Time above range

The share of the day spent above 180 mg/dL, usually split to show separately how much was above 250. Target: under 25% above 180, and under 5% above 250.

What it is indirectly telling you: where the excess sugar is coming from, once you look at when it happens rather than how much. Highs concentrated in the two hours after meals point at food and at the medicines that cover meals. Highs that are already present on waking, before anything has been eaten, point somewhere else entirely — at the overnight and background control. Those two patterns lead to different treatment changes, and separating them is one of the main reasons this test gets ordered.

Time below range — the lows, and the false ones

Here is the half of the report that patients never expect, and the half that most often changes a prescription.

The report gives the share of the day spent below 70 mg/dL — target under 4%, which is under about an hour — and separately below 54 mg/dL, target under 1%. These are small numbers because a low sugar is immediately dangerous in a way a high one is not.

Most of the lows a CGM finds are ones nobody knew about, and the commonest are at night. A sugar that falls at three in the morning is almost invisible to ordinary monitoring, because nobody is awake to prick a finger, and people frequently sleep through it — or half-wake sweating and put it down to a bad dream. Some lows are also silent: after many years of diabetes the warning symptoms can fade, so a person genuinely does not feel a sugar of 55.

When a report shows lows, the treatment usually comes down, not up. That is the opposite of what most patients expect from a test they went into thinking their sugars were too high. It tells us we are over-treating at some point in the day — a sulfonylurea doing too much overnight, an insulin dose too large, a gap between meals too long — and it is often the reason sugars have been erratic rather than simply high.

The false low. A sensor that is lain on for a long stretch — sleeping on the arm the sensor is fitted to — can read falsely low, an effect known as pressure-induced sensor attenuation. Pressure temporarily pushes fluid away from the filament, so the sensor reports a fall that is not happening in your blood. The clue is the shape rather than the number: a compression low drops abruptly while you lie still and recovers within minutes of turning over, whereas a real low has a gentler slope and takes time to come back. This is worth knowing, because otherwise a frightening 3 a.m. reading gets treated as real. Show the graph to your doctor rather than reporting the number, and if you see it repeatedly, try sleeping on the other side.

What this section is indirectly telling you: whether the current doses are right — and it is the single thing a CGM sees that finger-prick testing structurally cannot.

Coefficient of variation

The CV is a measure of how much your sugar swings around its own average, given as a percentage. The target is 36% or less.

It is the difference between a day that stays broadly level and a day that climbs and crashes repeatedly while arriving at the same average. High variability is uncomfortable to live with, it makes every other number harder to interpret, and it goes hand in hand with a higher chance of hypoglycaemia — which is really what the figure is there to warn about.

What it is indirectly telling you: how predictable your diabetes is. Erratic sugars usually have a findable cause — irregular meal timings, a sulfonylurea, missed or doubled doses, or food that is very different from one day to the next.

The AGP graph

The large grey graph is the cleverest part of the report and takes a moment to read, because it is not one day. It is every day of the fortnight laid one on top of another and compressed into a single 24-hour picture, from midnight to midnight.

The dark line running through the middle is your median — the middle of all your readings at that hour of the day. The shaded bands around it show the spread: how widely your sugar varied at that hour across the two weeks. A narrow band means you do much the same thing every day at that time. A wide band means that hour is unpredictable.

Read it by shape rather than by number. A hump in the late morning is breakfast. A band that sits high all night is a background control problem. A climb in the small hours before waking — the dawn phenomenon, driven by the body’s own morning hormones — explains a high fasting sugar in someone who ate nothing after dinner, and it is a common source of blame that patients have been laying on themselves for years.

What it is indirectly telling you: which hours of your day are the problem — which is a far more actionable answer than which foods, because it tells your doctor where in the day to place or move a medicine.

The daily profiles — where your kitchen shows up

The last pages are a calendar of small charts, one for each day. This is where the AGP graph’s neat average pulls back apart into real days, and it is where the dietitian work actually happens.

Take your food notes and lay them against these charts. The chapati at 8:30 with a peak behind it. The rice at lunch with no peak behind it. The tea and two biscuits at 5 p.m. that turn out to matter more than the meal did. The evening walk that flattens the whole post-dinner curve.

This is the part of the report to sit with, and the part that is yours rather than your doctor’s. What comes out of it is not a diet sheet. It is a list of the specific dishes in your own house that your own body does not handle well — and, just as usefully, the list of ones it handles perfectly well and that you can stop worrying about.

What a CGM cannot do

• It cannot diagnose diabetes. A sensor is not a diagnostic test, and the American Diabetes Association says explicitly that there is not enough evidence to use one for screening or diagnosis. Diabetes is diagnosed on a laboratory fasting glucose, an HbA1c or a glucose tolerance test. This matters increasingly, because sensors are now marketed to people without diabetes: a healthy person’s sugar rises after a meal, sometimes past 140 or 160, and seeing that happen is normal physiology, not a disease.
• It is not measuring your blood, and it lags when sugar is moving fast — so when someone feels low, the finger prick decides.
• It cannot tell you why. It shows a spike; it does not know whether the cause was the food, the missed tablet, the illness, the bad night’s sleep or the steroid injection last week. That interpretation needs the notes you kept and the doctor reading them with you.
• It is a fortnight, not a life. If those two weeks were unusual — travel, a festival, an illness, a wedding — the report describes those two weeks faithfully and your ordinary life not at all.

What changes because of it

A finger-prick test tells you what your sugar is at that moment. A CGM is ordered to answer the questions that follow from it — the decisions, not the description:

• Are we under-treating, or over-treating? The lows answer this, and they are the finding that most often reduces a dose rather than increasing one.
• Is the problem the fasting sugar or the meals? Highs before breakfast and highs after meals call for different medicines. Seeing which pattern you have is how the treatment gets distributed correctly across the day rather than simply increased.
• Which foods, specifically, in this house? The part you take away and act on yourself.
• Is the current regime working at all? When someone is genuinely taking their medicines and genuinely eating carefully and the HbA1c still will not move, the sensor is what finds the gap between what is believed and what is happening.

After a stent or a bypass

This is where a sensor stops being a diabetes tool and becomes a cardiac one, and it is the reason it comes up in a cardiology clinic at all. A person with diabetes who has just had a heart attack has two problems that make each other worse, and the sugar is no longer somebody else’s department.

When the sensor goes on differs between the two operations. Practice varies between units and this is not a rule written in any guideline, but the logic behind the timing is worth understanding, because the two operations create two different jobs.

After bypass surgery — a sensor at discharge. A bypass leaves a healing wound in the breastbone, and high sugar in the weeks that follow raises the risk of that wound becoming infected, which is a serious complication. Controlling the sugar therefore matters immediately, whatever the underlying reason for it being high. In the intensive care unit that job belongs to an insulin infusion, with a target of keeping the glucose under 180 mg/dL; nothing replaces that. But the wound goes on healing for weeks after the patient goes home, and a sensor is how that control continues once the drip has come out and the daily blood tests have stopped.

After a stent — a sensor at the first follow-up, seven to ten days later. Here there is no wound, and the job is a long-term HbA1c target rather than an urgent one. That makes the timing worth getting right, because sugars measured during and just after a heart attack are not trustworthy. The stress of the event itself pushes glucose up in people whose diabetes is mild and even in some who do not have diabetes at all. Fitting a sensor in that window would measure the heart attack rather than the diabetes. Waiting for the first follow-up, when a fasting and a post-meal sugar are available and honest, places the sensor against a real baseline.

On targets: the European guidance is for an individualised HbA1c, generally in the range 6.5–8.0%, aiming under 7.0% wherever it can be achieved safely. After a stent that commonly works out at a target of about 6.5–7% — but the word individualised is doing real work in that sentence. An older patient, or one who has had severe hypoglycaemia, is deliberately given a gentler target, because in them a low sugar is the greater danger.

Who should not spend the money

A sensor costs around ₹4,500 and lasts about a fortnight, and it is not a purchase everyone with diabetes needs to make.

If your diabetes is controlled on diet alone and your HbA1c is already at target, a CGM has almost nothing to teach you. Your two weeks will show what you already know. Keep the money.

Where it earns its cost is narrower and more specific:

• The HbA1c is above target despite genuinely taking the medicines and genuinely eating carefully — the situation this page opened with.
• Hypoglycaemia is suspected, or has happened, or the sugars swing unpredictably.
• Your doctor needs to know whether the fasting sugar or the post-meal rise is the problem before changing treatment.
• You have just had a stent or bypass surgery and the sugar now has a second job to do.

For most people it is one sensor, not a subscription. The point is to learn how your own food and your own routine behave, and that lesson does not need repeating every month.

What happens next

You will come back with the report, and most of it will be unremarkable. Your doctor will draw your eye to three or four lines — the time in range, the time below range, the shape of the graph overnight, and the hours where the band is widest — and the rest is context.

From there, two conversations follow, and they are different conversations. One is about medicines: which dose, at what time, and whether anything needs to come down because of lows nobody knew about. That one belongs to your doctor. The other is about food, and that one is largely yours — which is the whole point of putting the sensor on. It is worth having with whoever cooks in your house present, because in practice that is who any change actually depends on.

Common questions

What is a good time in range on a CGM report?

For most adults with diabetes the target is more than 70% of the day spent between 70 and 180 mg/dL — that is about 17 hours out of every 24. Alongside it, less than 4% of the day should be below 70 mg/dL and less than 1% below 54 mg/dL. Older people, and anyone at high risk from a low, are given a gentler target — often more than 50% in range, with an even stricter limit on time spent low. Your own target is set by your doctor, not by the printout.

Why is my GMI different from my lab HbA1c?

Because they are two different measurements of two different things, and a gap between them is normal. GMI is calculated purely from your average sensor glucose over the past two weeks. HbA1c is a blood test measuring how much sugar has stuck to your red cells over roughly three months. So they cover different periods — and HbA1c is also affected by anything that changes the life of a red blood cell, such as anaemia, iron deficiency, kidney disease or a recent transfusion. A difference of a few tenths is expected. A large gap is itself useful information and worth showing your doctor.

Can a CGM tell me if I have diabetes?

No. A continuous glucose monitor is not a diagnostic test, and the American Diabetes Association states plainly that there is not enough evidence to use one to screen for or diagnose diabetes. Diabetes is diagnosed on a fasting blood sugar, an HbA1c, or a glucose tolerance test done in a laboratory. This matters because sensors are now sold to people who do not have diabetes at all, and a perfectly healthy person will see their sugar rise after a meal — sometimes past 140 or 160 — because that is what normal physiology does. Seeing that number is not a diagnosis of anything.

My CGM showed a very low sugar at 3 a.m. but I felt nothing. Is it real?

It may be real, and it may be an artefact — and both possibilities matter. A genuine low overnight is one of the most valuable things a CGM finds, precisely because nobody is awake to prick a finger, and it usually means a medicine dose needs reducing. But a sensor lain on for hours also reads falsely low, an effect called pressure-induced sensor attenuation, because pressure temporarily squeezes the fluid away from the sensor filament. The clue is the shape: a compression low drops abruptly while you are lying still and recovers the moment you turn over. Show your doctor the graph rather than the number.

Does rice always spike sugar more than chapati?

Not necessarily, and this is one of the most useful surprises a CGM delivers. Research that fed the same food to hundreds of people found that the sugar response to an identical meal varied enormously from person to person, while each individual’s own response stayed consistent and repeatable. So it is entirely possible for your sugar to climb after two chapatis and stay steady after a bowl of rice, while your neighbour is the other way round. General advice about which staple is worse is built on averages; a CGM shows you your own answer.

How long do I wear a CGM sensor, and how much does it cost?

The sensor used most commonly in India is worn on the back of the upper arm for up to 15 days and costs in the region of ₹4,500. It is applied at home with a small applicator, is water-resistant enough for a bath, and sends readings to a smartphone app or to a separate reader. For most people one sensor is enough — the aim is to learn how your own food and your own routine behave, and that lesson does not need repeating every month.

Can I change my insulin or diabetes tablets based on what the CGM shows?

No — and this is the one firm limit on the whole exercise. What a CGM lets you decide for yourself is food: which dishes in your own kitchen send your sugar up and which do not. Changing the dose of insulin or of a sulfonylurea on the basis of sensor readings is a medical decision, and getting it wrong in the direction of too much causes hypoglycaemia, which is dangerous. Bring the report to your doctor and let the doses be changed there.

Is a CGM worth it if my sugars are already well controlled?

Usually not. Someone controlling their diabetes on diet alone with an HbA1c already in target has very little to learn from a sensor, and the money is better kept. A CGM earns its cost when the HbA1c is above target despite a person genuinely taking their medicines and eating what they believe is a careful diet, when hypoglycaemia is suspected, or when a doctor needs to know whether the problem is the fasting sugar or the post-meal rise before changing treatment.