Read the trend, not one number: why repeat tests matter
A single blood result is a snapshot. Your values move a little from day to day, and every lab measurement carries a small margin of error. Laboratory medicine uses a tool called the reference change value to judge whether a difference between two…

En este artículo
A single blood result is a snapshot. Your values move a little from day to day, and every lab measurement carries a small margin of error. Laboratory medicine uses a tool called the reference change value to judge whether a difference between two results is likely real. Repeat tests, done under the same conditions, make that judgment possible.
Key takeaways
- Each person's values fluctuate around their own set point. This is called within-person biological variation [1][2].
- Every measurement also has analytical variation, the small imprecision of the method itself [1].
- The reference change value combines both to show how big a change must be before it is likely to be real [1].
- For many markers, the difference between people is much larger than the swing within one person, so your own past results are often a better guide than the population range [5][7].
- Repeat tests are most useful when done in the same lab, under the same conditions. Discuss any change with your doctor.
Why one number can mislead
Say your creatinine was 0.90 mg/dL last year and 1.00 mg/dL this year. That is an 11% rise. Is kidney function changing, or is this normal movement plus measurement error? One number cannot tell you. Two numbers, read with the right tool, often can.
Most people compare a result with the reference range on the report. That range is commonly defined as the central 95% of results from a healthy reference population [6]. By design, some healthy people will fall outside it. The same source notes that if you run 10 independent tests, the chance that all of them fall inside that central 95% interval is only about 60% [6]. A flag on a report is a prompt to look closer, not a verdict.
Reference ranges also vary by laboratory, method, sex and age. The range on one report does not apply to every lab.
Two kinds of noise
Biological variation
Your body keeps most values close to a personal set point, a bit like a thermostat. Results drift randomly around that point from one day to the next [1][2]. Researchers measure this as the within-subject coefficient of variation, written CVI. Some markers also follow daily, monthly or seasonal rhythms [2].
Between-subject variation (CVG) describes how far apart different people's set points are. For sodium, Fraser gives the example of four stable patients with set points of 142, 137, 144 and 139 mmol/L [1].
Analytical variation
No method measures with perfect precision. If the same sample runs twice, the results differ slightly. This is the analytical coefficient of variation, CVA. Good laboratory practice aims to keep it below half of the within-person variation, so the method does not hide real biological change [1].
The reference change value
The reference change value (RCV) puts both sources of noise together. The classic formula is [1]:
RCV = 1.414 × Z × √(CVA² + CVI²)
Z reflects how confident you want to be. A value of 1.96 corresponds to 95% confidence and 2.58 to 99% [1]. The 1.414 (the square root of 2) is there because you are comparing two results, each with its own noise.
Fraser's worked example uses potassium. With an analytical imprecision of 2.1% and a within-person variation of 4.8%, the RCV at 95% confidence is 14.5%. Only changes larger than that are likely to be significant [1].
The Spanish Society of Laboratory Medicine (SEQCML) agrees: reference ranges alone cannot interpret a change between two results, because for most markers within-person variation is smaller than between-person variation [7].
How much do common markers move?
The table below uses within-person (CVI) and between-person (CVG) estimates from the 2014 biological variation database compiled by Ricós and colleagues [4]. The RCV column is an illustration only. We calculated it at 95% confidence, assuming a method imprecision of half the CVI, which is the desirable target described above [1][4]. A real lab's RCV depends on its own measured imprecision.
| Marker | CVI (within you) | CVG (between people) | Illustrative RCV (95%) | Notes |
|---|---|---|---|---|
| HbA1c | 1.9% | 5.7% | about 6% | Very stable; small changes can matter |
| Glucose (serum) | 5.6% | 7.5% | about 17% | Depends on fasting state |
| Creatinine | 5.95% | 14.7% | about 18% | Muscle mass shifts the set point |
| Total cholesterol | 5.95% | 15.3% | about 18% | |
| Triglycerides | 19.9% | 32.7% | about 62% | Large swings are common |
| TSH | 19.3% | 24.6% | about 60% | Newer studies report lower CVI [2] |
| Free T4 | 5.7% | 12.1% | about 18% | |
| ALT | 19.4% | 41.6% | about 60% | |
| Calcium | 2.1% | 2.5% | about 7% | Tightly regulated |
| PTH | 25.9% | 23.8% | about 80% | One of the most variable |
Source for CVI and CVG: Ricós et al., 2014 database [4]. The EFLM Biological Variation Database now replaces it for the markers it covers [2][3].
The lesson: a 20% change in HbA1c would be well beyond normal noise, while a 20% change in triglycerides or TSH may sit within it.
Your baseline beats the population range
Researchers compare within-person and between-person variation with the index of individuality. When it is below 0.6, population reference ranges are of little use for spotting a meaningful change in one person [5]. Using the figures above, creatinine (5.95 ÷ 14.7) and total cholesterol (5.95 ÷ 15.3) both come out at about 0.4.
So your creatinine could rise meaningfully for you and still sit inside the printed range. You could also sit just outside the range and be at your own long-standing normal. Your history of results gives the context a single number lacks. Ask your doctor to look at the sequence, not only the latest value.
Making repeat tests comparable
Biological variation studies work hard to remove pre-analytical variation, such as how and when samples are taken [2]. You can borrow the same habits:
- Test in the same lab where possible, since methods and ranges differ between labs.
- Keep the time of day and fasting state the same when the test calls for it.
- Note anything unusual before the test, such as illness, new medication or intense exercise, and tell your doctor.
How often to repeat a test depends on your situation. That is a decision to make with your doctor.
What you can measure at Santé Labs
At Santé Labs in Barcelona, you can book these profiles today. They cover the markers discussed above:
- Glycaemic Control Profile: glucose, HbA1c, lipids, creatinine.
- Basic Kidney Profile: creatinine, eGFR, electrolytes, calcium.
- Complete Thyroid Profile: TSH, free T4, free T3.
- Complete Liver Profile: ALT, AST, GGT.
- Bone Health: calcium, vitamin D, intact PTH.
The Standard Lipid Profile is coming soon. We measure and explain. Your doctor interprets results in the context of your health.
FAQ
My result changed. Should I worry?
Not necessarily. Every marker moves a little on its own, and every method has some imprecision [1]. Whether a change matters depends on its size compared with the reference change value for that marker. Discuss it with your doctor.
Why is my result flagged if I feel well?
Reference ranges usually cover the central 95% of healthy people, so some healthy people fall outside them by design [6]. A flag is a reason to review the result with your doctor, not a diagnosis.
Does the reference change value apply to everyone?
It is an estimate based mostly on healthy adults. Spanish laboratory literature notes it may not suit people whose condition is changing [7]. Your doctor will judge how it applies to you.
Is a result inside the range always fine?
Not always. For markers with a low index of individuality, a meaningful change for you can still sit inside the range [5][7]. That is why your doctor may look at the trend.
References
- Fraser CG. Biological variation: what's it all about? Acute Care Testing, 2004 – https://acutecaretesting.org/en/articles/biological-variation--whats-it-all-about
- Fraser CG. Biological variation: a still maturing aspect of laboratory medicine. Adv Lab Med. 2020;1(3):20200003 – https://pmc.ncbi.nlm.nih.gov/articles/PMC10197770/
- EFLM Biological Variation Database – https://biologicalvariation.eu/
- Ricós C, et al. Desirable biological variation database specifications (2014 update), hosted by Westgard QC – https://www.westgard.com/clia-a-quality/quality-requirements/biodatabase1.html
- Li D. Pearls of Laboratory Medicine: Biological variation. ADLM (AACC), 2012 – https://myadlm.org/science-and-research/clinical-chemistry/clinical-chemistry-trainee-council/pearls-of-laboratory-medicine-in-english/2012/biological-variation
- Ekelund S. Reference intervals and percentiles: implications for the healthy patient. Acute Care Testing, 2018 – https://acutecaretesting.org/en/articles/reference-intervals-and-percentiles-implications-for-the-healthy-patient
- Rigo Bonnin R, et al. (SEQCML). Procedimiento para la interpretación de un cambio entre dos valores consecutivos de una magnitud biológica. Rev Lab Clin. 2019;12(2):93-97 – https://www.elsevier.es/es-revista-revista-del-laboratorio-clinico-282-pdf-download-S1888400818300047


