Reference Intervals Are Population Statistics, Not Personal Targets

Your lab result is 2.50 mmol/L for serum calcium, and the reference interval printed next to it says 2.15-2.55 mmol/L. You are inside the range, so your calcium is normal. That conclusion is wrong more often than most clinicians realise, because reference intervals are population statistics, not personal targets. The interval tells you how 95 percent of a healthy reference population scored; it does not tell you whether your value is optimal, pathological, or even accurate for your physiology. A reference interval is a statistical construction, not a medical verdict, and treating it as a personal target leads to missed hyperparathyroidism, untreated ionised calcium deficiency, and overconfidence in a number that was never designed to carry that weight.

Take the corrected calcium calculation. A patient with chronic kidney disease has a total calcium of 2.30 mmol/L and an albumin of 28 g/L. The Payne equation adds 0.02 × (40 − 28), or 0.24 mmol/L, producing a corrected calcium of 2.54 mmol/L. The reference interval printed on the report is 2.15-2.55 mmol/L. The result falls inside it. The clinician sees a normal corrected calcium and moves on.

What the interval does not say is that this patient has a measured ionised calcium of 1.05 mmol/L, well below the ionised calcium reference interval of 1.15-1.30 mmol/L. The corrected calcium is falsely elevated because the formula overcorrects for low albumin in patients with abnormal protein profiles. The reference interval, derived from a healthy population with normal albumin, has no way to capture this mismatch. The example shows exactly what the interval says, this patient is statistically unremarkable, and what it does not say, that the ionised calcium is dangerously low.

The calculation of a reference interval follows a fixed procedure defined by CLSI guideline C28. For a Gaussian distribution, the interval equals the mean plus or minus 1.96 times the standard deviation. For serum calcium, a lab determines its own mean and standard deviation from a reference population of at least 120 healthy individuals. If the lab's mean is 2.35 mmol/L and the standard deviation is 0.10 mmol/L, the reference interval is 2.35 ± 1.96 × 0.10, or 2.15-2.55 mmol/L.

When you need to assess whether a specific result falls inside or outside that interval, you do not need a statistics calculator. The arithmetic is straightforward. A result of 2.50 mmol/L is inside the interval. A result of 2.60 mmol/L is outside it. What the interval does not provide is the probability that this result is pathological. A patient with 2.60 mmol/L may have primary hyperparathyroidism, a transient calcium spike from a meal, or preanalytical error from a clotted sample. The interval cannot distinguish among these. The calculation produces a boundary, not a diagnosis.

The statistical logic behind a reference interval is straightforward, but the interpretation trap is deep. The interval is the central 95 percent of a healthy reference population. Five percent of healthy people will fall outside it by definition. A result outside the interval is statistically unusual, nothing more. A result inside the interval is statistically typical, nothing less.

For corrected calcium in particular, the interval fails in the sickest patients. The original Payne equation was derived from a single UK hospital population. Its correction factor works best when albumin is between 30 and 50 g/L. Outside that range, error increases. In a patient with an albumin of 22 g/L, the correction adds 0.36 mmol/L, often pushing a truly low ionised calcium into the normal range. The reference interval, which was built from healthy individuals with normal albumin, does not account for this distortion.

KDIGO 2017 explicitly advises against using albumin-adjusted calcium as the primary measure in chronic kidney disease. The KDOQI US commentary reinforces the caution. Despite these guidelines, many clinicians continue to rely on the corrected value because it is free and familiar. The reference interval on the report provides context without offering medical interpretation. A low corrected calcium in a patient with hypoalbuminaemia may be falsely reassuring. A normal corrected calcium in an ICU patient with acidosis may mask a true hypocalcaemia. The interval says the number is statistically typical. It does not say the number is clinically normal.