STATISTICS
value resulted from a simple transcription error, then correction or exclusion is justified. If review shows that the specimen arrived late because of an identified processing delay, then the value is genuine and belongs to the service process, even if it is unusual. In that situation it may be reasonable to present the full dataset as the primary analysis and a secondary sensitivity analysis showing what the summary looks like without the delay. That is very different from silently deleting the value because it appears awkward. The example therefore shows the interaction between all three themes of this article: the distribution determines which summaries are most informative, the interval estimate determines how confidently the summary can be interpreted, and the unusual value demands investigation rather than reflex removal.
Common mistakes, limitations, and points of caution A recurring mistake in laboratory writing is to treat descriptive statistics as routine decoration rather than as a genuine analytical choice. Mean and standard deviation are often reported by default because they are the first values produced by software, not because they are the best descriptions of the data. In visibly skewed datasets this can be misleading. A second mistake is to treat normality
testing as a mechanical gatekeeper. Statistical tests of normality can contribute to the assessment, but they are not substitutes for looking at the data. A third mistake is to confuse types of intervals. Confidence intervals, reference intervals, and decision limits answer different questions and should not be
used or described interchangeably. A fourth mistake is to present percentiles or medians as exact quantities with no uncertainty around them. Perhaps the most consequential
mistake is casual outlier removal. Outlier detection procedures may be useful, but they do not reveal cause, and they are not interchangeable. Different procedures may identify different values and lead to different summaries. This is not a minor technicality; it can materially change the final result. The prudent position is therefore straightforward: outliers should be investigated, exclusions should be justified, and reporting should be transparent.
This article has also deliberately
stayed within limits. It does not cover formal inferential comparisons, regression, advanced robust methods, or full reference interval establishment procedures. Those topics require separate discussion. The purpose here is to improve the quality of the first statistical encounter with a dataset.
When decisions about transformation, exclusion, or grouping are made, record them clearly enough that another reader could understand and challenge them.
Suggested further reading Cerioti F, Hinzmann R, Panteghini M.
PPi
Practical take-home points Good statistical practice in laboratory medicine begins with description, not hypothesis testing. Inspect the data before summarising them. Do not assume normality by habit. Match the summary statistic to the shape of the data and the question being asked. Report uncertainty around important estimates, especially in small datasets. Treat outlier detection as the beginning of a review process, not an automatic reason for deletion.
Additional resources to support this series
To spice things up a bit I thought I would try to develop some useful resources to supplement the material in the articles. For the first article I have developed a teaching tool for very basic understanding of probability distributions. The app is shown below, and you can access it for free, of course, at: htps://
pathologyuncertainty.com/educational-support-materials/
Reference intervals: the way forward. Ann Clin Biochem. 2009;46(Pt 1):8–17. Clinical and Laboratory Standards Institute. EP28: Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory. Wayne (PA): CLSI; current standard product page. Hickman PE, Koerbin G, Poter JM et al. Choice of statistical tools for outlier removal causes substantial changes in analyte reference intervals in healthy populations. Clin Chem. 2020;66(12):1558–1561. Horn PS, Pesce AJ. Reference intervals: an update. Clin Chim Acta. 2003;334(1-2):5–23. Horn PS, Pesce AJ, Copeland BE. A robust approach to reference interval estimation and evaluation. Clin Chem. 1998;44(3):622– 631.
Horn PS, Feng L, Li Y, Pesce AJ. Effect of outliers and nonhealthy individuals on reference interval estimation. Clin Chem. 2001;47(12):2137–2145. Ialongo C. Confidence interval for quantiles and percentiles. Biochem Med (Zagreb). 2019;29(1):010101. Ichihara K, Boyd JC; IFCC Commitee on Reference Intervals and Decision Limits (C-RIDL). An appraisal of statistical procedures used in derivation of reference intervals. Clin Chem Lab Med. 2010;48(11):1537–1551. Ozarda Y, Ichihara K, Jones G, Streichert T, Ahmadian R. Reference intervals: current status, recent developments and future considerations. Biochem Med (Zagreb). 2016;26(1):5–16. Ozarda Y, Sikaris K, Streichert T, Macri J; IFCC Commitee on Reference Intervals and Decision Limits (C-RIDL). Distinguishing reference intervals and clinical decision limits: a review by the IFCC Commitee on Reference Intervals and Decision Limits. Crit Rev Clin Lab Sci. 2018;55(6):420–431. Solberg HE, Lahti A. Detection of outliers in reference distributions: performance of Horn’s algorithm. Clin Chem. 2005; 51(12):2326–2332. Solberg HE. The IFCC recommendation on estimation of reference intervals. The RefVal program. Clin Chem Lab Med. 2004;42(7):710– 714.
Dr Stephen MacDonald is Consultant Clinical Scientist, The Specialist Haemostasis Unit, Cambridge University Hospitals NHS Foundation Trust, Cambridge Biomedical Campus, Hills Road, Cambridge CB2 0QQ.
+44 (0)1223 216746 26
WWW.PATHOLOGYINPRACTICE.COM May 2026
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
