Vitamin D3 for Life

Live Enlightened

Article

How Vitamin D Population Reference Ranges Are Derived and Statistically Constructed: Full Study

Previous related pageFree vs Total Vitamin D ExplainedNext related pageHow Vitamin D Population Reference Ranges Are Derived and Statistically Constructed: Study Summary
Study at a glance

What is a vitamin D reference range?

A reference range describes blood results found in a selected group of people. It often includes the middle 95% of results. For vitamin D, the measurement is usually 25-hydroxyvitamin D. Being inside this range does not automatically mean the level is ideal, and being outside it does not automatically mean disease.

How are vitamin D reference ranges calculated?

Researchers can recruit people who meet chosen health rules and measure their blood levels. They can also analyse results already held by laboratories. The people included, season, laboratory method and calculations all affect the range. A large collection of results does not by itself ensure that the group represents everyone.

Why did two ways of selecting blood results give different vitamin D ranges?

A Spanish study compared 312 healthy volunteers with 7,222 existing laboratory results. The volunteer range was about 9.0 to 41.4 ng/mL; the range from existing results was about 5.6 to 57.2 ng/mL. Both described the same vitamin D form, but different ways of selecting and analysing results produced different ranges. Seasonal ranges also differed.

Does a vitamin D reference range set a treatment target?

A population range answers what values were found in a particular group under particular conditions. It does not automatically set a treatment target or identify where deficiency or toxicity begins. Those decisions require different evidence. A range from one population, season or laboratory method cannot be assumed to apply unchanged everywhere.

Read the full article below ↓

A population reference range for serum 25-hydroxyvitamin D, or 25(OH)D, is not a biologically predetermined zone waiting to be discovered. It is a statistical description derived from measurements obtained from a defined reference population under specified sampling and analytical conditions. Who is included, when samples are collected, which assay is used, how abnormal observations are handled and which statistical model is chosen can all influence the resulting interval. Understanding those steps is essential before a laboratory reference interval is treated as evidence about vitamin D sufficiency or deficiency. See Study Summary: How Vitamin D Population Reference Ranges Are Derived and Statistically Constructed: Study Summary.

What a population reference interval actually represents

In laboratory medicine, a conventional population reference interval describes the central distribution of results found in a selected reference population. A commonly used construction is the central 95 percent of observed values, bounded by the 2.5th and 97.5th percentiles. That definition is statistical. It does not mean that values inside the interval are necessarily optimal or that values outside it automatically represent disease.

This distinction is particularly important for vitamin D because reference distributions are frequently confused with biological targets or clinical decision limits. Vitamin D Thresholds vs Targets addresses that distinction directly. A population interval asks what values occur in a selected population. A decision threshold asks whether evidence supports taking a particular clinical action at a particular concentration. Those are different questions.

Reference-population selection comes before the statistics

Before percentiles can be calculated, investigators must decide whose blood results count as reference observations. Classical direct approaches deliberately recruit people who meet predefined inclusion and exclusion criteria. The purpose is to create a group judged suitable for describing the expected distribution of the analyte under the conditions being investigated.

This decision is not neutral. If a reference group differs in age, latitude, ultraviolet exposure, diet, supplementation, body composition, ethnicity, health status or season of sampling, its 25(OH)D distribution may differ from that of another apparently healthy group. A reference range is therefore inseparable from the population from which it was derived.

That is one reason apparently different vitamin D ranges need not imply that one study is simply wrong. The broader problem of apparently discordant measurements and populations is discussed in Why Vitamin D Results Differ.

The direct method

In the direct method, individuals are intentionally selected as reference participants. Eligibility can be determined before sampling, known as an a priori approach, or exclusions can be applied after collection, known as an a posteriori approach. The laboratory then measures the analyte in the resulting group and estimates the reference limits from those observations.

A widely used non-parametric approach ranks the measured concentrations from lowest to highest and estimates the central 95 percent directly from the ordered observations. With a sufficiently large sample, this has an important advantage: investigators do not have to assume that 25(OH)D follows a normal Gaussian distribution.

Traditional guidance commonly uses at least 120 reference individuals for a non-parametric reference interval because that number permits estimation of the 2.5th and 97.5th percentiles together with confidence intervals around the reference limits. Larger populations can improve precision and permit meaningful partitioning when subgroups genuinely differ.

Why 2.5th and 97.5th percentiles do not define physiology

The central 95 percent convention necessarily places approximately five percent of a suitably sampled reference population outside the interval, even when the entire population is being used descriptively. That fact alone demonstrates why a reference limit cannot automatically be interpreted as the point where healthy physiology ends and disease begins.

The same issue applies in the opposite direction. A value lying comfortably inside a reference population distribution is not proof that the concentration is sufficient for every physiological purpose in every individual. The conceptual distinction is developed further in Vitamin D Sufficiency Explained.

Indirect reference intervals use existing laboratory data

Direct recruitment is scientifically attractive but expensive, slow and sometimes difficult. Indirect methods instead analyse large collections of results already stored in laboratory information systems. Statistical procedures attempt to identify the distribution most likely to represent the reference population while reducing the influence of pathological or otherwise unsuitable values.

The advantage is scale. An indirect study can potentially use tens of thousands or hundreds of thousands of observations. The disadvantage is that the individuals were not recruited specifically as healthy reference volunteers. The statistical method therefore has to separate the presumed reference distribution from contamination by disease, treatment, repeated testing or other selection effects.

Neither direct nor indirect methods are inherently correct in all circumstances. They answer the same descriptive problem through different sampling architectures, and both depend on the quality of the population and analytical data entering the calculation.

Vitamin D demonstrates how much the method can matter

A 2024 Spanish investigation compared direct and indirect 25(OH)D reference intervals within the same geographical setting and analytical platform. The indirect dataset contained 7,222 observations, while the direct study recruited 312 healthy volunteers.

The indirect interval was approximately 5.6 to 57.2 ng/mL. The directly derived interval was approximately 9.0 to 41.4 ng/mL. Those are materially different distributions despite both being labelled population reference intervals for the same analyte.

The difference demonstrates that a range inherits the architecture used to construct it. Inclusion criteria, the source population and the statistical treatment of results can materially change the endpoints.

Season changes the distribution before statistics are applied

Vitamin D adds another complication because the reference population itself moves over the year. In the Spanish direct study, the lower reference limit was about 7.4 ng/mL in winter and 13.3 ng/mL in summer. The upper estimates also shifted.

Those seasonal differences arise before the percentile calculation. The statistics are describing a biological population whose exposure environment changes with season. They are not generating the seasonality.

This is one reason the wider site distinguishes population statistics from Biological Noise in Vitamin D Testing. A mathematical reference interval can only describe the biological and environmental variation present in the observations supplied to it.

Analytical methods also shape the observed distribution

A population interval is calculated from measured values, not from inaccessible true concentrations. If the assay has systematic bias, the whole observed distribution can shift. This is particularly important for 25(OH)D because historical comparisons have documented substantial disagreement among assays and laboratories.

The Vitamin D Standardization Program was developed to improve comparability by linking measurements to recognised reference measurement procedures. National survey studies demonstrate why this matters. In Canadian Health Measures Survey data, standardisation changed the estimated proportion of participants below 50 nmol/L from 29.0 percent using the original data to 36.8 percent after standardisation.

An Irish national survey showed the same principle. The estimated prevalence below 30 nmol/L changed from 6.5 percent using the original ELISA measurements to 11.4 percent following standardisation, with LC-MS/MS reanalysis producing 11.2 percent.

The people did not change during those calculations. The analytical calibration changed. This illustrates why Variability in Vitamin D Measurements is relevant even when the final objective is a population statistic.

Reference intervals can be partitioned

A single interval may be inappropriate when scientifically meaningful subgroups have different distributions. Laboratories can consider partitioning by age, sex, season or another justified variable where evidence shows that combining groups obscures a material difference.

Partitioning is not automatically beneficial. Every additional subgroup requires enough observations to estimate its limits reliably, and indiscriminate partitioning can create unstable ranges. The scientific question is whether the variable causes a sufficiently meaningful distributional difference to justify separate interpretation.

Large datasets do not remove the need for biological judgement

Modern indirect methods can process very large laboratory datasets. A recent Barcelona analysis used more than 130,000 serum 25(OH)D measurements and an indirect statistical method designed to recover a reference distribution from real-world laboratory data.

Scale improves statistical precision, but it does not magically solve selection. A very large dataset can estimate the distribution represented in the data extremely precisely while still leaving the biological question of whom that distribution represents.

This is a general statistical lesson relevant to Misuse of Vitamin D Statistics: numerical precision and biological validity are different properties.

Reference intervals are not the same as clinical decision limits

A clinical decision limit is normally established because evidence links a concentration with risk, diagnosis, treatment eligibility or another decision. A population reference limit is obtained from the statistical distribution of a reference group.

The two values can sometimes be numerically similar, but they arise from different evidence. Treating one as though it automatically proves the other is a category error.

This distinction becomes especially important when different organisations use different vitamin D thresholds. Those disagreements may involve different clinical objectives, evidence frameworks and interpretations rather than disagreement about how the 2.5th percentile is calculated. That wider problem contributes to why Why Vitamin D Studies Disagree cannot be reduced to one statistical explanation.

What the evidence establishes

The evidence establishes that vitamin D population reference intervals are constructed objects whose numerical limits depend on the reference population, sampling method, season, analytical method and statistical procedure. Direct studies deliberately recruit reference individuals. Indirect studies derive distributions from existing laboratory data. Non-parametric methods commonly estimate the central 95 percent using the 2.5th and 97.5th percentiles.

Vitamin D provides a particularly clear example of why those design choices matter because its population distribution changes seasonally and its measured distribution can change after assay standardisation.

What the evidence does not establish

The evidence does not establish that the central 95 percent of a population is biologically optimal. It does not establish that a lower reference limit is automatically a deficiency threshold. It does not establish that an upper reference limit is automatically a toxicity threshold. It does not establish that a range derived in one population, season or analytical system transfers unchanged to another.

These boundaries are why Limitations of Vitamin D Blood Tests must include not only assay performance but also the interpretive framework placed around the number.

Final interpretation

A vitamin D population reference interval is best understood as a statistically estimated description of a defined population measured under defined conditions. Its endpoints are informative about that distribution, but they do not acquire physiological or clinical authority merely because they appear beside a laboratory result. The scientific meaning of a range depends on how the population was selected, when it was sampled, how 25(OH)D was measured and what question the range is being asked to answer.

Primary sources and publication details

Horowitz GL. Reference intervals: practical aspects. eJIFCC. 2008. Read the full source.

Jones GRD, Haeckel R, Loh TP, et al. Indirect reference intervals: harnessing the power of stored laboratory data. Clinical Chemistry and Laboratory Medicine. 2019. Read the full source.

Direct and indirect reference intervals of 25-hydroxyvitamin D: it is not a real vitamin D deficiency pandemic. 2024. PMID: 38882584. Read the full study.

Sarafin K, Durazo-Arvizu R, Tian L, et al. Standardizing 25-hydroxyvitamin D values from the Canadian Health Measures Survey. American Journal of Clinical Nutrition. 2015. Read the full study.

Cashman KD, Kiely M, Kinsella M, et al. Evaluation of Vitamin D Standardization Program protocols for standardizing serum 25-hydroxyvitamin D data. American Journal of Clinical Nutrition. 2013. Read the full study.

Related reading

Original sources

Source

Read the full source

Horowitz GL. Reference intervals: practical aspects.

Publication details

Horowitz GL. Reference intervals: practical aspects. eJIFCC. 2008. Read the full source.

Read more →
Source

Read the full source

Jones GRD, Haeckel R, Loh TP, et al.

Publication details

Jones GRD, Haeckel R, Loh TP, et al. Indirect reference intervals: harnessing the power of stored laboratory data. Clinical Chemistry and Laboratory Medicine. 2019. Read the full source.

Read more →
Source

Read the full study

Direct and indirect reference intervals of 25-hydroxyvitamin D: it is not a real vitamin D deficiency pandemic.

Publication details

Direct and indirect reference intervals of 25-hydroxyvitamin D: it is not a real vitamin D deficiency pandemic. 2024. PMID: 38882584. Read the full study.

Read more →
Source

Read the full study

Sarafin K, Durazo-Arvizu R, Tian L, et al.

Publication details

Sarafin K, Durazo-Arvizu R, Tian L, et al. Standardizing 25-hydroxyvitamin D values from the Canadian Health Measures Survey. American Journal of Clinical Nutrition. 2015. Read the full study.

Read more →
Source

Read the full study

Cashman KD, Kiely M, Kinsella M, et al.

Publication details

Cashman KD, Kiely M, Kinsella M, et al. Evaluation of Vitamin D Standardization Program protocols for standardizing serum 25-hydroxyvitamin D data. American Journal of Clinical Nutrition. 2013. Read the full study.

Read more →