How Do I Compare Two Dexcom Clarity Exports to See if My Glucose Improved

Take two Dexcom Clarity CSV exports from two date ranges, open the free Compare two CGM exports tool at HealthOS, and drop both files in. It reads them in your browser, no account, nothing uploaded. Then compare the outputs for time in range, GMI, variability, and biggest rises.

How do I compare two Dexcom Clarity exports to see if my glucose improved?

First, get two exports that cover the periods you want to compare. In Dexcom Clarity, set the first date range and export the CSV. Then set the second date range and export that CSV. Name the files so you remember which is baseline and which is follow-up.

At https://healthosx.com/glucose, open Compare two CGM exports and drop both files in. The tool runs in your browser, so no glucose readings leave your device and there is no account.

A useful order:

• Time in range. Should be higher in the follow-up export.
• GMI. Should be lower if average glucose came down.
• Variability (CV). Should be lower if the days got steadier.
• Biggest rises. Should be smaller or return to baseline faster.

The four numbers to compare: time in range, GMI, variability (CV), and biggest rises.

Read time in range before you read anything else

Time in range is the share of sensor readings inside a set band. Dexcom Clarity usually reports it against 70 to 180 mg/dL (3.9 to 10.0 mmol/L), the band set by an international consensus for people with diabetes. The consensus goal for people with diabetes is more than 70% of time in that band. If the follow-up export shows a higher percentage inside that band, more of your day is staying out of the high and low ends.

For people without diabetes, the tighter 70 to 140 mg/dL band is often more informative. A higher follow-up percentage means the curve got flatter.

GMI and variability show whether the average moved or just the noise

GMI estimates the A1C lab result that usually goes with your average sensor glucose. The formula is 3.31 plus 0.02392 times your average glucose in mg/dL. A drop in average glucose from 140 to 130 mg/dL lowers GMI from about 6.7% to 6.4%. GMI is not a lab result, and it is most trustworthy with 14 days or more. But it is useful for comparing two exports from the same sensor type.

Variability is the standard deviation of all the readings divided by the mean, shown as a percentage. Two periods can have the same average glucose and completely different days. Suppose both exports average 120 mg/dL. One stays between 100 and 140 most of the time, the other swings between 70 and 190. Same average, different stress on the system. A lower CV means the follow-up is steadier.

These two numbers answer different questions. GMI answers whether the center moved. Variability answers whether the edges moved.

Biggest rises are the number that matches what you felt

Time in range and GMI smooth out the day. The biggest rises do not. If your largest breakfast rise used to take you from 95 to 185 mg/dL and now peaks at 155 mg/dL, that is a specific, real change. Compare the largest meal responses in each export, but compare the same meals where you can. A smaller breakfast rise is not cancelled by a new giant dinner rise.

Look for two things in the follow-up export. The peak should be lower, and the return to the starting value should be sooner. A lower peak with a long, slow tail is only half an improvement.

Match the days before you call it an improvement

The most common mistake here is comparing exports of different lengths. A 30-day export will almost always look smoother than a 7-day export, because a bad Tuesday is diluted by 29 other days. Decide what counts as an improvement before you open the files.

Aim for at least 14 days in each export when you can. GMI and time in range get noisy with less than that. If one export is 21 days and the other is 8, treat the comparison as exploratory, not settled.

Also check that both files use the same glucose unit. The tool reads both mg/dL and mmol/L, but your own summary should compare like with like.

If you menstruate, note the cycle phase for each export so you can compare similar phases when possible.

When the follow-up looks worse

If the four numbers moved the wrong way, do not immediately conclude the change failed. Look at what else changed in that window. Sleep, training load, illness, a new sensor, more restaurant meals. A single week with a cold can lift average glucose and variability. The mechanism is not the food or habit you changed, it is the stress response and less movement.

Run a third export over the next 14 days with the same habit in place. You do not need more precision to start, but you do need enough days.

Open the free report at healthosx.com/glucose and drop both files in. If you do not wear a sensor yet, HealthOS Sensor is a month of continuous glucose with the app for $70, no membership, at healthosx.com/glucose/sensor.