Comparing blood test results: read the trend, not one value

Two "normal" results can still hide a real change. How to compare blood test results over time, which difference is real, and how often to retest.

6 min - BioSense Team - 2026-10-13

level:understanding
comparer deux bilans
évolution biomarqueurs
suivi analyses sang
variation biologique

Comparing blood test results over time shows what a single result cannot: two values "within range" can reflect a real change. For many tests, the reference ranges on your report are poor at detecting the changes that concern you personally. What is informative is the difference from your own previous test, read in light of each marker's natural variation.

You have lab results from last year and a recent set. Both came back "normal." You file them away and think no more about them.

Put side by side, however, your fasting glucose has gone from 86 to 97 mg/dL (4.8 to 5.4 mmol/L). It remains below the 100 mg/dL (5.6 mmol/L) threshold the American Diabetes Association uses for prediabetes. So it remains "normal." But it has moved by about 13%.

This article explains what comparing blood test results adds, how to spot a difference that genuinely matters, and the conditions under which the comparison is reliable.

Why "normal" twice in a row does not mean "stable"

The reference ranges on your report come from a group of healthy people. They describe the spread of values seen in about 95% of them. So they are information about a group.

Your body, though, regulates each marker around a set point of its own. That personal set point occupies a narrow space inside the lab's range, which is far wider. You can therefore drift well away from your usual level while still staying "within range."

Laboratory scientists measure this phenomenon. They compare the size of your personal variations with the size of the differences between individuals. When your variations are small relative to the spread across the population, the reference range becomes of little use for spotting a change in you. This is the case for many routine tests.

For many markers, the most meaningful comparison is therefore the one with your own test from last year.

Key point - Reference ranges describe a population, not your personal baseline. For many markers, they are poor at spotting an individual change. Two results "within range" can therefore reflect a real trend.

Between two blood tests, which difference is real?

Before concluding that a figure has moved, you need to know how much it moves on its own, with nothing having happened. Two causes add up.

Analytical variation

No measurement is perfectly reproducible. If the lab ran the same tube twice, it would not find exactly the same figure. For common tests this difference is small and well controlled. It is not zero all the same.

Within-person biological variation

This one weighs much more heavily. Your body oscillates naturally around its set point, from one day to the next and from one week to the next, with no disease involved.

The size of that oscillation changes enormously depending on the marker. Few people know it, and it changes everything:

A1C, which reflects your average blood sugar over recent months, is very stable. A systematic review found a median within-person variation of about 2% in healthy people, and more in people with diabetes.
TSH, the hormone that drives the thyroid, behaves the opposite way. A large European reference study measured a natural variation of about 18%. A TSH moving from 2.0 to 2.3 mIU/L has probably signaled nothing at all.
Creatinine, used to assess how the kidneys are working, sits between the two, at around 5%.

By adding the imprecision of the measurement to the natural variation, laboratory scientists calculate what they call the reference change value. It is the minimum difference between two results above which the change can be considered real rather than down to chance.

Hence a consequence that often comes as a surprise. A 15% rise in your TSH probably means nothing. Without diabetes, a 15% rise in your A1C is a strong signal. Same percentage, opposite reading.

Key point - Each marker varies naturally within an amplitude of its own. A1C moves by a few percent, TSH by almost twenty. The same percentage difference therefore does not have the same meaning from one test to the next.

Your reports already exist, scattered across PDFs, emails and patient portals. What is missing is the reading. BioSense aligns each biomarker from one test to the next, explains it in plain language and sets it against what you tell us about your situation: your history, your medications, your lifestyle, the reason for the test.

[→ Import my blood tests](/en/onboarding/upload)

The markers where the trend matters more than the threshold

Some tests read poorly in isolation and far better in series.

Fasting glucose and A1C

The clearest example. Prediabetes usually takes years to develop. Blood sugar rising steadily each year inside the normal range signals a trend well before the 100 mg/dL (5.6 mmol/L) threshold is crossed. Thresholds also differ between countries: the ADA starts prediabetes at an A1C of 5.7% (39 mmol/mol), while NICE in the UK uses 6.0% (42 mmol/mol).

Ferritin

Ferritin, which reflects your iron stores, illustrates the phenomenon well. The differences between individuals are enormous. MedlinePlus gives a normal range of 13 to 150 ng/mL (13 to 150 µg/L) for women and 30 to 400 ng/mL (30 to 400 µg/L) for men, and your usual value occupies a narrow space inside it. Ferritin that halves while staying "within range" is worth flagging to your doctor.

One useful caveat: inflammation can raise ferritin, and published estimates of its natural variation differ between studies. Any overly mechanical reading of its trend calls for caution.

Kidney function

Kidney function can decline slowly and without symptoms. For this marker, the slope over several years teaches more than a single value. The KDIGO guideline states that for people with chronic kidney disease, a change in eGFR of more than 20% on a later test exceeds expected variability and warrants evaluation.

The lipid panel

LDL cholesterol responds to changes in diet, physical activity and medication. Without a point of comparison, there is no way to know whether your efforts have had an effect.

Comparing two blood tests: four conditions for reliability

A comparison only holds if the two measurements are comparable. Four points contribute to that.

The same lab where possible. Instruments and reagents vary from one lab to another. Switching between labs can add a difference that has nothing biological about it, especially for tests such as ferritin, vitamin D or vitamin B12.
The same draw conditions. Fasting or not, according to what the test requires. The same time of day for markers with a marked daily rhythm.
A comparable context. An infection, ongoing inflammation, a new medication or heavy physical effort the day before will alter certain markers. Without that information, a passing circumstance risks being read as a real trend.
The same units. Most US labs report glucose in mg/dL, while many other countries use mmol/L. If you compare a report from abroad, convert first: reference ranges change as well.

How often to repeat a blood test to track the trend

No single rhythm suits everyone. The right frequency depends on your age, your history, your medications and what your doctor is monitoring. That decision is theirs.

A few general points do exist all the same.

Repeating a test too early adds nothing for slow markers. An A1C measured again three weeks after the previous one will say little that is new. It reflects average blood sugar over about three months and takes time to move.

Conversely, too long an interval loses the thread. Between two tests six years apart, there is no way to know whether the change happened gradually or last year.

Finally, regular monitoring beats intensive monitoring in bursts. Three tests spread over three years teach more than six tests concentrated into a single year, followed by five years of silence.

Key point - For a comparison to hold: the same lab if possible, the same draw conditions, the context recorded, identical units. With regular spacing, matched to how fast each marker moves.

The BioSense method

BioSense sets every result you import against your personal profile and official recommendations. It also aligns it with your previous tests, so that the trend of each biomarker can be read at a glance.

*ChatGPT answers you. BioSense knows you.* *BioSense, your prevention companion.* BioSense does not provide a diagnosis and does not replace your doctor. It is a prevention companion that helps you understand, ask the right questions and act day to day.

Comparing your blood test results: three questions to ask your doctor

If you have several tests, these three questions move the visit from the latest result toward your trend:

Comparing my recent tests, do any markers deserve attention even though they stay within the reference ranges?
For those markers, is the difference large enough to count, or too small to draw a conclusion from?
How often should these tests be repeated given my profile, and should I go back to the same lab?

Bringing your previous tests, and not only the latest one, changes the conversation with your doctor.

Frequently asked questions

How many blood tests do I need before a trend in a biomarker means anything?

There is no universal number. Two results are enough for a first comparison if the difference exceeds the marker's reference change value, and several spaced tests then show whether the trend holds.

What change in blood test results is significant?

It depends on the marker, because each has its own natural variation. The reference change value indicates the difference above which a change is considered real, and your doctor weighs it against your context.

Can you compare blood test results from different labs?

Yes, but with caution. Methods, units and reference ranges can change, so part of the difference may come from the lab rather than your body.

Should a result that changes but stays within range worry me?

Not necessarily, but it is worth flagging if the difference is clear or steady. The range describes a population, whereas your usual value occupies a narrow space inside it.

*This article is strictly informational and constitutes neither medical advice nor a diagnosis. For any question about your health, consult your doctor.*

The first test you import becomes your starting point. The next one will be read against it. The sooner you start, the more meaningful the comparison. Import your blood tests, even older ones, and receive a plain-language reading of each biomarker, placed in the context of your profile. Your data is hosted on servers located in France.