Clinical Biochemistry 43 (2010) 933–934

Contents lists available at ScienceDirect

Clinical Biochemistry
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / c l i n b i o c h e m

Letter to the Editor

Limitations of the Hoffman approach to determine pediatric The testosterone data show interesting differences between the
reference intervals for two steroids two studies. First, for younger boys, 7–12 years old, and for girls of all
ages, the lower reference limit is higher with the Hoffmann approach
compared to those we found using samples from healthy children [5].
Keywords: For boys 13 years and older, the lower reference limits are higher for
Hoffmann approach healthy children compared to the Hoffmann approach. For girls of all
17-Hydroxyprogesterone ages the upper reference limits are remarkably similar for both
Pediatric reference intervals
methods. For boys 7–9 years of age, the upper reference limit for
Testosterone
healthy individuals is lower than that determined using the Hoffman
approach. For older boys the upper reference limits are similar.
A number of factors may account for the differences seen. First,
To the Editor: there may be racial and ethnic differences. The healthy children in our
studies were mostly Caucasians. The race and ethnicity of the subjects
We read with interest the article by Soldin et al. detailing pediatric in the Soldin study were not given. Second, there could be analytical
reference intervals for several steroid hormones obtained using mass differences. Results generated by our testosterone method have been
spectrometry [1]. Reference intervals were provided from birth to compared to a reference measurement procedure and the agreement
adulthood. This is the first time, to our knowledge, that information was good [7]. We are not aware of any comparative data for the Soldin
for mass spectrometric methods has been available for children method. Third, there could be geographic differences. Salt Lake City is
younger than 7 years. However, we have concerns about the subjects at an elevation of 4200 ft above sea level while parts of Washington, D.
being considered healthy. They are first described as outpatients who C. are at sea level. We think this is an unlikely explanation. Finally, the
had blood drawn for testing of one or more steroids. In the statistical lack of adequate discrimination between healthy and sick children
analysis section, the population is described as a cohort of healthy using the Hoffman approach could explain the differences. Testoster-
pediatric outpatients to whose laboratory results the Hoffmann one testing is usually ordered in children when precocious or delayed
approach was applied. Given that blood apparently was drawn for puberty is suspected based on a physical examination. In younger
analysis of at least one steroid based on clinical need, we question children, precocious puberty is more likely. If this were the case, then
whether these subjects can all truly be classified as healthy. it would explain the higher values for the lower reference limits that
Presumably, steroid analysis was ordered to support a diagnosis were seen in younger boys by Soldin et al. Conversely, in older boys,
with some expectation that the result might be abnormal. delayed puberty is a more likely concern. Lower values were seen for
A Clinical and Laboratory Standards Institute document indicates
that reference intervals determined in this manner should be
considered rough estimates at best, since the datasets are contami-
nated with an unknown percentage of results from individuals who Table 1
are not truly healthy [2]. Furthermore, the Hoffmann approach Reference intervals for 17-hydroxyprogesterone and testosterone.
assumes a normal distribution of healthy results [3]. More recent
Age and Hoffmann approach Healthy volunteers
approaches to reference interval analysis using patient data employ gender (reference 1) (references 5 and 6)
more sophisticated statistical methods that do not assume a Gaussian
n 2.5th 97.5th n 2.5th 97.5th
distribution of normal results and can separate healthy and disease percentile percentile percentile percentile
result distributions mathematically [4].
17-Hydroxyprogesterone (ng/dL)
Due to concerns about whether the report by Soldin et al. would
Boys
accurately delineate reference intervals for healthy children, we 6–9 years 153 7 100 125 b5 63
compared reference intervals for 17-hydroxyprogesterone and tes- 10–17 years 7 100 330 b5 149
tosterone determined by Soldin et al. to those we previously Girls
6–9 years 182 6 62 125 b5 71
published for healthy children 7–17 years old [5,6]. We repartitioned
10–17 years 144 15 137 330 b5 177
our data to conform to the age partitioning used by Soldin et al.
(Table 1). Non-parametric analysis was used for all data. For 17- Testosterone (ng/dL)
hydroxyprogesterone, the lower reference limit determined using the Boys
Hoffmann approach was higher across all groups than those we 7–9 years 148 4 25 104 b1 9
10–12 years 188 5 418 103 2 413
obtained using healthy children, especially in girls 10 years and older.
13–14 years 245 6 647 67 11 525
The upper reference limits we determined were also different [6]. 15–16 years 209 42 880 70 99 623
They were substantially lower for 6–9 year old boys and substantially Girls
higher for 10–17 year old boys and girls compared to the Soldin values 6–9 years 171 5 13 102 b1 15
[1]. It also appears that partitioning of boys at 10 years of age is 10–14 years 190 14 50 172 2 52
15–16 years 175 12 53 70 8 58
required, due to a higher upper reference limit for older boys.

0009-9120/$ – see front matter © 2010 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.clinbiochem.2010.04.064
934 Letter to the Editor

lower reference limits by Soldin et al. compared to what we saw for PM. State-of-the-art of serum testosterone measurement by isotope dilution-
liquid chromatography–tandem mass spectrometry. Clin Chem 2008;54:
healthy children in this age group. 1290–7.
These data for testosterone may illustrate the limitations of using
the Hoffmann approach to establish reference intervals. Healthy
children, or perhaps sick children who do not have a disease that
would significantly alter the analyte of interest, should be used to
William L. Roberts⁎
establish pediatric reference intervals wherever possible. It would be
Alan L. Rockwood
interesting to reanalyze the data from Soldin et al. using the statistical
Department of Pathology, University of Utah,
approach described by Concordet et al. [4].
Salt Lake City, UT, USA
⁎Corresponding author. ARUP Laboratories, 500 Chipeta Way,
References
Salt Lake City, UT 84108, USA.
[1] Soldin SP, Sharma H, Husted L, Soldin SJ. Pediatric reference intervals for aldosterone, E-mail address: william.roberts@aruplab.com (W.L. Roberts).
17alpha-hydroxyprogesterone, dehydroepiandrosterone, testosterone and 25-hy-
droxy vitamin D3 using tandem mass spectrometry. Clin Biochem 2009;42:823–7.
Ashley M. Bunker
[2] CLSI. How to define and determine reference intervals in the clinical laboratory;
approved guideline. second edition. Wayne, PA: CLSI; 2000. CLSI document C28-2A. Mark M. Kushnir
[3] Hoffmann RG. Statistics in the practice of medicine. JAMA 1963;185:150–9. ARUP Laboratories, Salt Lake City, UT, USA
[4] Concordet D, Geffré A, Braun JP, Trumel C. A new approach for the determination of
reference intervals from hospital-based data. Clin Chim Acta 2009;405:43–8.
[5] Kushnir MM, Rockwood AL, Roberts WL, Pattison E, Bunker A, Meikle AW. A. Wayne Meikle
Performance characteristics of a novel tandem mass spectrometric assay for serum Department of Pathology, University of Utah,
testosterone. Clin Chem 2006;52:120–8. Salt Lake City, UT, USA
[6] Kushnir MM, Rockwood AL, Roberts WL, Pattison EG, Owen WE, Bunker AM, Meikle
AW. Development and performance evaluation of a novel tandem mass Department of Internal Medicine, University of Utah,
spectrometry assay for four adrenal steroids. Clin Chem 2006;52:1559–67. Salt Lake City, UT, USA
[7] Thienpont LM, Van Uytfanghe K, Blinkco S, Ramsay CS, Xie H, Doss RC, Kevil BG,
Owen LJ, Rockwood AL, Kushnir MM, Chun KY, Chandler DW, Field HP, Sluss 16 July 2009