Reference intervals and reference ranges per CLSI EP28-A3C Establish, partition, verify, and transfer reference intervals using parametric, non-parametric, robust, bootstrap, and Harrell-Davis quantile methods — with a full range of transformations for non-normal distributions.
Try free for 15 days — no credit card required
Trusted by 75,000 scientists at most of the top-10 IVD manufacturers, and at thousands of ISO 15189, ISO/IEC 17025, and CLIA-regulated laboratories worldwide for over 30 years.
Every method EP28-A3C requires
Parametric, non-parametric, robust bi-weight, bootstrap, and Harrell-Davis quantile methods — all in one package. With Box-Cox, Manly, and 2-stage exponential transformations for non-normal distributions. Choose the method that fits your data and sample size, not the one your software happens to support.
Validated against CLSI EP28-A3C data
Every calculation runs in Analyse-it's own engine — no Excel formulas, no third-party functions. Results are validated against the example datasets published in the CLSI EP28-A3C guidelines before every release. Defensible in a 510(k), CE-IVD technical file, CAP inspection, or ISO 15189 audit.
Data stays in your facility
Analyse-it runs entirely within Microsoft Excel on your PC. No cloud processing, no data transmission — your patient sample data stays within your facility under your data governance controls.
Establishing a reference interval requires more decisions than most tools accommodate. Real patient populations are rarely normally distributed, which rules out simple parametric methods for many analytes. Clinically meaningful differences between sex, age groups, or ethnicity mean a single interval is often inappropriate — but partitioning reduces the sample size in each subgroup, which puts pressure on your choice of statistical method. And once intervals are established, confirming that they apply to a different laboratory, a different patient population, or a new measurement procedure adds another layer of statistical complexity that most implementations simply don't address.
Analyse-it covers the full EP28-A3C protocol. Parametric methods when the normality assumption holds. Non-parametric quantile methods with three computation approaches — (N+1)p, Np+1/2, and (N+1/3)p+1/3 — to match your protocol requirements. Robust bi-weight methods for symmetric and skewed distributions when sample sizes are smaller than EP28-A3C ideally requires. Bootstrap and Harrell-Davis quantile methods for additional flexibility. A full range of transformations — log, square root, reciprocal, Box-Cox, Manly exponential, and 2-stage exponential/modulus — to bring non-normal data into line before applying parametric approaches. Shapiro-Wilk and Anderson-Darling normality tests and normal Q-Q plots to assess the distribution before and after transformation. Confidence intervals on all reference limits.
Reference interval studies sit within a broader validation workflow. IVD manufacturers need reference ranges for product labelling and regulatory submissions. Clinical laboratories need to verify that manufacturer-provided intervals are applicable to their patient population, or establish their own when the population differs. Analyse-it includes transfer and verification procedures — the binomial test for proportion inside the reference interval, and transfer to a new measurement procedure using method comparison regression — so these common scenarios are handled within the same tool, not approximated in spreadsheets.
Scientists at Abbott, Roche, Thermo Fisher, and most of the world's leading IVD manufacturers have relied on Analyse-it for reference interval studies for over 30 years — as have clinical laboratories establishing population-specific reference ranges for their patient communities.
Seen in the field
What's included
-
Choose the right method for your data and sample size
Parametric normal quantile methods when normality holds. Non-parametric quantile methods with three computation approaches. Robust bi-weight methods for smaller samples with symmetric or skewed distributions. Bootstrap and Harrell-Davis quantile methods for additional flexibility. Confidence intervals on all reference limits.
-
Handle non-normal distributions with the right transformation
Log, square root, reciprocal, Box-Cox, Manly exponential, and 2-stage exponential/modulus transformations. Shapiro-Wilk and Anderson-Darling normality tests and normal Q-Q plots to assess the distribution before and after transformation.
-
Partition reference ranges by clinically relevant factors
Derive separate intervals for sex, age group, ethnicity, or any combination of factors where biological differences make a single reference interval clinically inappropriate. Partition by any number of subgroups within a single analysis.
-
Screen the reference population for outliers
Tukey outlier box plots and formal outlier tests to identify values that should be excluded from the reference sample before limits are established — following the EP28-A3C approach to reference sample screening.
-
Verify and transfer existing reference intervals
Verify whether published or manufacturer-provided reference intervals are transferable to your laboratory and patient population using the binomial test for proportion inside the reference interval. Transfer existing intervals to a new measurement procedure using method comparison regression, without repeating a full reference population study.
-
Validated engine you can defend at inspection
Every calculation runs in Analyse-it's own engine — no Excel formulas or third-party functions. Validated against the CLSI EP28-A3C published example datasets before every release. See the development and validation process.
-
Part of the Method Validation edition
Reference intervals are included in the Method Validation edition, which covers the full CLSI method validation workflow — measurement system analysis (EP05-A3, EP06-A, EP15-A3, EP17-A2), method comparison (EP09-A3, EP21-A), diagnostic performance (EP24-A2, EP12-A2), and total analytical error (EP21-A). One tool for everything your laboratory or IVD submission requires.
Trusted by 75,000 scientists at most of the top-10 IVD manufacturers, and at thousands of ISO 15189, ISO/IEC 17025, and CLIA-regulated laboratories worldwide for over 30 years.
Try free for 15 days — no credit card required
See all Method Validation features
Pricing and licensing
Reference lab case study
