Why Do My Blood Test Results Fluctuate When Nothing Has Changed

Your blood test results fluctuate because a blood draw is a single sample of a moving biological system, read by an instrument that has its own built-in error. Most changes between two draws, especially small ones, are biological and analytical noise, not a true change in your health. The useful question is not why the number moved, but whether it moved more than the reference change value for that marker.

The three layers of variation

A single result carries three layers of variation: analytical variation from the lab instrument, within-person biological variation from your own rhythms, and preanalytical variation from sleep, food, exercise, posture and time of day. When a result moves, all three layers move inside that number. The reference change value, or RCV, combines the first two into a threshold that a change must clear before it probably means something.

The bars show approximate within-person biological variation from the EFLM biological variation database (Ricós et al., Scand J Clin Lab Invest 1999). Analytical variation adds another 1 to 5 percentage points on top for most routine chemistry markers, which is why the formula uses both.

A blood test result is a sample from a moving river, not a photograph of a fixed state. Treat a single small change as noise until it clears the reference change value.

How much change is a real change?

The threshold is not the reference range on your report. It is the reference change value. For a 95% confidence level:

RCV = 2.77 × √(CVa² + CVi²)

CVa is analytical variation, CVi is within-person biological variation (Harris & Yasaka, Clin Chem 1983). If the percentage change between two results is smaller than the RCV, the change is not distinguishable from noise.

Approximate RCVs for common markers, using EFLM biological variation estimates:

• Fasting glucose: about 16%. A rise from 90 to 104 mg/dL is right at the edge; 90 to 120 is a change.
• Total cholesterol: roughly 15 to 18 percent. A move from 200 to 215 is probably noise; 200 to 236 is a change.
• HDL cholesterol: about 20%.
• Triglycerides: 50 to 60 percent. A shift from 150 to 220 can still be noise; 150 to 260 is more likely real.
• ALT: 50 to 60 percent. A jump from 30 to 45 U/L is borderline and may reflect a recent workout, a passing infection, or random variation, not necessarily liver disease.

hs-CRP is the extreme case. Its within-person variation is so large that a single small rise usually carries no information unless it is part of a trend or a clear illness.

A change that clears the RCV and repeats on a confirmatory draw, or any change with symptoms, is worth taking to a doctor.

These percentages assume the same lab and method. A different lab introduces bias on top of the noise, so the threshold is higher in practice.

One bad night is enough

Most "nothing has changed" claims collapse in the details. Fasting means 8 to 12 hours with water only. A single short night can blunt insulin sensitivity. A hard lifting session can raise creatine kinase for up to a week, and can nudge ALT, AST and creatinine. Dehydration concentrates creatinine, urea and haematocrit. A viral infection you barely noticed can raise hs-CRP, ferritin and white cells. Alcohol in the previous 24 to 48 hours can raise triglycerides and GGT. Posture matters too: moving from lying to standing can concentrate blood proteins by roughly 10 percent. Time of day moves cortisol, iron and testosterone.

Practical rules before a draw:

• Fast for the same number of hours each time.
• Avoid heavy training for 48 hours beforehand.
• Draw at the same time of morning.
• Note any infection, vaccination, period, poor sleep or alcohol on the report.

Does "normal range" mean "no change"?

No. A reference interval is built from a population, usually the middle 95 percent of healthy people. You can move from the top of that range to the bottom and stay inside it, while the change is still a genuine change for you. This is the common mistake: people treat the population range as a personal target. The RCV is personal; the reference range is population-based. For markers with high within-person variation, such as ALT and triglycerides, two different people can have very different stable points that are both normal.

On a graph, noise looks like a squiggle around a flat line. A real change looks like a step that keeps moving and clears the RCV band. That is why three or more draws are more informative than a single pair: a pair gives you a difference, a series gives you a trajectory.

What about a different lab or different units?

Units do not create variation by themselves if the conversion is correct, but different labs use different analysers, reagents and calibrators. A jump after changing labs can be a method difference, not biology. If you must change lab, compare the percentage change and look for a one-time offset across many markers at once. That pattern suggests the measurement system changed, not you.

The next time a single number moves, ask two questions: is the change larger than the reference change value, and is it part of a trend? If your earlier reports are already in your browser, upload your newest report at https://healthosx.com/blood and the free tool shows how each marker moved since your last report, across labs and units.