original article

Reference intervals for common biochemistry

laboratory tests in the Saudi population by a
direct a priori method
Xuejiao L. Hu, Huda Hassan, Fouad Hassan Al-Dayel

From the Department of Pathology & Laboratory Medicine, King Faisal Specialist Hospital and Research Center, Riyadh 11211, Saudi Arabia

Correspondence: Dr. Fouad Hassan Al-Dayel · King Faisal Specialist Hospital and Research Center, Pathology & Laboratory Medicine,
Riyadh, 11211, Saudi Arabia · T: 966-11-442-7224 F: 966-11-442-4280 · dayelf@kfshrc.edu.sa

Ann Saudi Med 2017; 37(1): 16-20

DOI: 10.5144/0256-4947.2017.16

BACKGROUND: Reference intervals (RI) for biochemistry laboratory tests are now based on Caucasian
rather than Saudi populations. Test parameters may vary because of race, lifestyle, population structure and
geographic location.
OBJECTIVES: To establish reference intervals for common clinical chemistry laboratory tests for the Saudi
population.
DESIGN: Direct a priori method.
SETTING: Tertiary care hospital.
MATERIALS AND METHODS: Blood samples were taken from 625 individuals aged from 2 to 87 years
from different geographic areas for 93 biochemistry tests. RIs were established following the International
Federation of Clinical Chemistry guideline.
MAIN OUTCOME MEASURE(S): Reference values for common biochemistry lab tests.
RESULTS: Ninety-three age- or gender-stratified reference intervals (RIs) based on the Saudi population
were established. There were 72 non-partitioned tests. Most of the tests were similar to RIs from manufac-
turer’s inserts. For some sex hormones (estrogen, luteinizing hormone, follicle-stimulating hormone, proges-
terone and 17α-Hydroxyprogesterone) only male RIs were established as there were not enough samples to
stratify for females based on physiologic status.
CONCLUSION: The RIs are reliable and applicable to a general Saudi population.
LIMITATIONS: Due to the sample size, RIs were not generated for some sex hormones for females.

R
eference intervals are critical for clinical decision lation groups. Some well-designed reference interval
making. Due to differences in age, sex, ethnic studies have been completed in Western societies,1-3
group, life styles, geographic location and popu- but in regions such as the Middle East, few laboratories
lation structures in the region served by the laboratory, have established their own reference values, especially
we hoped to establish complete reference intervals comprehensive ones.4 To fill this gap, we decided to es-
based on representative population tested by the medi- tablish reference values for some commonly used bio-
cal laboratory. However, because of the high cost, strin- chemistry laboratory tests following the proposed 2008
gent requirements, the time required and lack of man- Clinical and Laboratory Standards Institute/ International
power, establishing reference intervals (RIs) is a daunting Federation of Clinical Chemistry (CLSI/IFCC) document
task. Most Saudi clinical labs, including our laboratory at EP28-A3 guideline.5 There are direct (a priori and a pos-
KFSHRC rely solely on manufacturer inserts or published teriori) and indirect methods to establish RIs. The direct
peer review data, which are mainly based on Caucasian method is favored due to controversy over the indirect
populations. The drawback is that these test results may method.6 Arguably, it is not necessary to establish RIs
not be representative of the Saudi population or may for all tests, especially for the tests in which RIs have
be based on old data that do not reflect current popu- been replaced by decision limits by international con-

16 ANN SAUDI MED 2017  JANUARY-FEBRUARY   WWW.ANNSAUDIMED.NET

RIs IN SAUDI POPULATION original article
sensus, such as for cholesterol, troponin T or glycated Testing instruments
hemoglobin (HbA1c), and others. However, the data ob- Roche MODULAR ANALYTICS, E170 module, an elec-
tained could be used to compare the accuracy of the trochemiluminescence based assay was used to per-
RIs; therefore, reference ranges for these parameters form most tests. The Dade Behring BNII nephelometer
were still calculated. The data collected for other tests was used for apolipoprotein B, apolipoprotein A1,
were used for establishing our own RIs. The tests were transferrin, haptoglobin, homocysteine, IgG, IgM, IgA,
grouped as electrolytes, liver function, renal profile, lipid C3, C4, beta 2 microglobulin, a1 antitrypsin, cerulo-
profile, thyroid function, cardiac panel, anemia profile, plasmin and prealbumin; DiaSorin Liaison chemilumi-
immune function, cancer markers, metabolic panel, and nescent immunoassay for calcitonin and somatomedin
sex hormones for the convenience of application. There C (IGF-1); DPC immulite for insulin-like growth factor
were 72 non-partitioning RIs, 21 for either age or gender binding protein 3; Gamma counter radioimmunoassay
partitioning RIs, including 6 male-only RIs. for aldosterone and Vitamin 1,25 D; Radial immunodif-
fusion test for IgD; Cobas Mira for angiotensin con-
MATERIALS AND METHODS verting enzyme, Hb plasma and glucose-6-phosphate
dehydrogenase. Routine calibration and controls on all
Reference population testing instruments were performed following the inter-
The reference population were 625 healthy individuals nal procedure protocol.
from the north, west, and east of Saudi Arabia and the
Riyadh area. Adults were recruited as well as children Statistics
under 17 after receiving permission from their parents. Reference ranges were determined by the test values
Some children were middle school students and others that fell within 2 standard deviations or 95% of a nor-
were recruited from daycare centers. In this way, the mal distribution of the sampled reference population.
geographic limitations of the reference population were The lower 0.025 and upper 0.975 percentiles of the test
minimized and the age scope was broadened. This ref- results were excluded. This distribution is expressed by
erence population consisted of 316 (50.6%) females and equation: RI min=a* 2.5 + b; RI max = a* 97.5 +b. Final
309 (49.4%) males, ranging from 2 to 87 years of age. sample size of each test for the RI was determined by
There were 201 children younger than 17 years of age elimination of outliers using the modified Thompson
(32.2%) including 105 males from 3 to 17 years of age Tau method7 or due to variation in the number of tests
and 96 females from 2 to 17 years of age. Participation performed on each sample. Each test parameter was
was voluntary. plotted to a histogram to view its distribution and out-
A questionnaire similar to the document C28-A3 liers. RIs were determined either by nonparametric
(Defining, Establishing, and Verifying Reference Intervals parameters or using the log transformed data if the
in the Clinical Laboratory; Approved Guideline—Third distribution was skewed. Occasionally parametric val-
Edition)5 was used in selecting individuals. The individu- ues were adopted if a test values were close to a per-
als should have no of signs of disease specifically relat- fect normal distribution with a better confidence ratio
ed to the biochemical laboratory measure. A history of (CR). To stratify tests for gender, a Harris-Boyd parti-
diabetes, hypertension, chronic kidney disease, cancer, tioning model built in EP Evaluator (Data Innovations
taking prescription drugs and other chronic illness was LLC, South Burlington, Vermont) was used to calculate
grounds for exclusion. After selection, test samples were a critical z value and SD ratio. If the SD ratio >1.5 or
drawn after 10 hours of fasting. Samples were taken us- z max >critical z, partitioning would be done or the
ing standardized test tubes by routine phlebotomy. manufacturer’s instructions would be followed for the
Blood samples were allowed to clot at room tempera- purpose of comparison. Age partitioning was based on
ture, and serum was obtained after brief centrifugation. a t test correlation by dividing the test results into ≥18
Blood in anticoagulated tubes was also centrifuged to adult group and ≤17 of children group using IBM SPSS
collect the plasma except for a few that needed whole Statistics 20 software or it was based on published test
blood for tests. All specimens were placed into a −70ºC data. When there was a statistically significant differ-
freezer until analysis. Due to variations in the number of ence between RIs in two different age groups (P<.05),
available tests, not all the same number of individuals RIs would be established separately. The data were
tested varied from test to test. The study was funded analyzed using the EP Evaluator software. Based on
by King Faisal Specialist Hospital and Research Centre the IFCC requirement, 90% confidence intervals (CIs) of
(KFSHRC) and approved by the Ethics Committee of the mean for each test parameter were calculated and
KFSHRC. listed along with reference values. The 90% CIs were

ANN SAUDI MED 2017  JANUARY-FEBRUARY   WWW.ANNSAUDIMED.NET 17

 original article RIs IN SAUDI POPULATION

calculated using the equation: x±zα/2 [σ/√n], where x of parametric and nonparametric distributions on a (A)
represents sample mean, plus and minus margin error, histogram and (B) probability plot for original and trans-
where z stands for z-scores, α is the significance level formed data, respectively.
of the test, 0.10 for 90% CI and σ equals the standard For 108 confidence ratios calculated for 93 tests, the
deviation. The confidence intervals mean there is a 90% average was 0.097 (desirable <0.1). To further verify the
chance that the true population mean would fall within sensitivity and usefulness of the RIs we established, the
the confidence interval, which is a reliable indicator of index of individuality was calculated (data not shown).
the uncertainty of RIs. The confidence ratio, expressed The index of individuality was below 0.6 for only 4 tests,
as 0.5*(URLU-URLL+LRLU-LRLL)/(URL-LRL), is the ratio above 1.4 for 43, and 27 were in between. According to
of the average confidence interval width to the refer- Harris15 if the ratio of variation within (CVi) or between
ence interval width, which is related to the sample size subjects (CVi/CVg ratio) is .1.4 the RIs will be sensitive
(0.1 or less is desirable and less than 0.3 is acceptable). and useful, whereas if the ratio is 0.6 the utility of the RI
is low. We calculated the CVg (between subject varia-
RESULTS tion) for 74 tests for which data was available.16
The total of 93 biochemistry tests were divided into 11
groups for the convenience of indexing. There were 72 DISCUSSION
non-partitioned tests with an age span of 2 to 87 years Because many factors impact reference ranges, such
so that these reference values would be applicable for as selection of the reference population, test methods,
both males and females for this age group. Test sam- sample size, statistical methods, partitioning, and oth-
ples for aldosterone were obtained in a sitting position, ers, we paid close attention to pre-analytical, analytical
not in a supine or standing position so partitioning for and the post-analytical phases to ensure the reliability
position was unnecessary. Cortisol was not partitioned and accuracy of the reference values. We adopted a
as the majority of samples were taken in the morning, standardized questionnaire for the reference popula-
so the test results were taken as random and no gender tion, choosing adequate samples sizes, using traceable
stratification was indicated by the partitioning calcula- test methods, partitioning as necessary, and applying
tion. Among 21 partitioned tests, 5 were sex hormones suitable statistical methods, strictly following the inter-
(estrogen, luteinizing hormone, follicle-stimulating hor- nal procedure protocols for handling of all specimens. As
mone, progesterone and 17α-Hydroxyprogesterone) there are no universally accepted criteria to define outli-
and only for males as there were not enough samples ers, we based our calculations on different equations,
to stratify for females based on physiologic status. Our including the Dixon Q test, X(n)−X(n−1)>X(n)−X(1)/3
samples were mixed with narrower reference ranges to reject the largest value and X(2)−X(1)<X(n )−X (1)/3
than the insert; therefore, the test would be applica- to reject the smallest value. Since the calculations were
ble for both population groups. Detailed results are not all in accordance, we took comprehensive measures
listed in Appendix 1. Figures 1 and 2 are examples to review the results, including histograms, examining

A B

Figure 1. Example of normally distributed histogram of ceruloplasmin values and probability plot after removing
outliers. The central 95% of the data is the reference interval. SDI: standard deviation index.

18 ANN SAUDI MED 2017  JANUARY-FEBRUARY  WWW.ANNSAUDIMED.NET

RIs IN SAUDI POPULATION original article

A B

Figure 2. Example of histogram of skewed distribution of CA19.9 values that was logarithmically transformed to a linear
probability plot and then 95% of reference values were selected. SDI: standard deviation index.

parameters before and after eliminating outliers, and the lipid profile. This may reflect the prevalence of hy-
confidence ratios (CR) to extrapolate the best fit RIs. perlipidemia in the Saudi population. The same was
We undertook these measures because skewed distri- true of glucose and HbA1c, which apparently overlap
butions were present. Even after elimination of outliers, with the decision limits for the diagnosis of diabetes.
the CR was unacceptable, even with nonparametric This may reflect that in our “healthy” reference popula-
calculations. Conversely, log-transformed parametric tion, there is a less than “healthy” group with impaired
values provided much better CRs and reference ranges glucose tolerance. Similar observations have been re-
(e.g., anti-TPO, glucose). For tests that had Gaussian ported in other countries and ethnic groups,12,13 which
distributions, parametric methods were adopted (e.g., is compatible with the growing global diabetes epi-
C3, IgG and TT3). If CRs were all the same, we choose demic.14 It is important that RIs are not confused with
nonparametric values (e.g., IGFBP3 and TSAT). All the clinical decision limits (CDLs). The RIs are calculated
data used to determine RIs were then verified using specifically for health whereas CDLs indicate sensitivity
a function for verification of RIs by the EP Evaluator; of disease. In general, CDLs are determined by con-
all passed. Although the CLSI C28-A3 recommends sensus. They are the thresholds above or below which
the nonparametric method, the RIs calculated by the a specific medical decision is recommended and are
parametric and nonparametric methods were com- derived from receiver operating characteristic curves
pared in a recent IFCC, C-RIDL study, 10 which con- and predictive values.10 Specifically, CDLs are based on
cluded that the results of the two methods were very the diagnostic question and are obtained from clinical
close, concluding that parametric methods can also trials designed to define the probability of the pres-
be used as a first choice. In theory, parametric meth- ence of a certain disease. These limits lead to decisions
ods will produce more accurate and precise estimates about how individuals with values above or below the
than non-parametric methods if assumptions are met. CDLs should be treated. Therefore, RIs of those few
Parametric methods may also have an advantage over tests we calculated should not be used for clinical de-
non-parametric methods in allowing identification and cision making in place of following CDLs. Some of our
exclusion of extreme values when computing RIs.11 RIs had much higher or lower values than that of the
In comparison with package inserts, many RIs we package inserts; LDH, for example, is known for high
established were very close to the manufacturer’s RIs. biological variation. Serum folate RIs may reflect the
For instance, electrolytes, which are known for having diet of Saudi population, which involves less consump-
low biological variation, and for liver enzymes and tests tion of green leafy vegetables in addition to there be-
in immune function panels, some RIs and inserts are ing biological variation. Also, some tests may need fur-
identical (e.g., homocysteine, troponin T, direct bili- ther stratification or a different stratification, because
rubin, female gamma-glutamyl transferase and male results of partitioning varied, depending on method.
human chorionic gonadotropin). However, several RIs We adopted the partitioning specified in manufactur-
were much higher than values in inserts, especially er’s inserts in some cases even though no or a different
where the manufacturer used decision limits, such as partitioning was indicated.

ANN SAUDI MED 2017  JANUARY-FEBRUARY   WWW.ANNSAUDIMED.NET 19

 original article RIs IN SAUDI POPULATION

Limitations of the study were that all biochemistry ther investigation.

laboratory tests were not included. To reduce cost Though the index of individuality was below 0.6
and produce more homogeneous RIs, multicenter col- for only 4 tests, some studies17,18 have shown that if a
laboration may be preferable for establishing RIs in single sample is taken, the index of individuality has no
the future. For females, we were unable to generate influence on the usefulness of the RIs. Therefore, these
progesterone and 17αHydroxyprogesterone. Human results are reliable and applicable to a general Saudi
chorionic gonadotropin was only generated for non- population.
pregnant women; some tests lacking partitioning by
time (morning or afternoon) , position (standing or su- Conflict of interest
pine) or lifestyle (smoker or non-smoker), deserve fur- The authors declared no conflict of interest.

REFERENCES
1. The REALAB Project: A New Method for 7. Home Page for Professor John M. Cimbala cal Practice Guidelines Expert Committee.
the Formulation of Reference Intervals Based of the Department of Mechanical and Nucle- Targets for glycemic control clinical practice
on Current Data E. Grossi, R. Colombo, ar Engineering of The Pennsylvania State Uni- guidelines for WHO, Global report on dia-
S. Cavuto, C. Franzini Clinical Chemistry. versity. ME 345: Instrumentation, Measure- betes www.who.int/diabetes/global-report
2005;51(7):1232-1240. ments, and Statistics; Outliers (http://www. April 7, 2016; 1-88.
2. NORDIC Reference Interval Project. P, Rus- mne.psu.edu/cimbala/me345/Lectures/ 14. Harris EK. Effects of intra- and inter-in-
tad. 2002. http://www.furst.no/norip/, 2003. Outliers.pdf). dividual variation on the appropriate use of
3. Suggested Reference Intervals For 8 Serum 8. Thomas L. Blut glucose. In: Thomas L, normal ranges. Clin Chem 1974; 20:1535 –
Enzymes Based On Data From NORIP Data- ed. Labor und Diagnose, 6th ed. Frankfurt/ 42.
base Stromme J. et.al.: http://www.ifcc.org/ Main:TH-Books,2005;193–199. 15. C. Ricós, V. Alvarez, F. Cava, J. V. Gar-
ifcc-communications-publications-division- 9. Klein G. Berger A, Bertholf R, et al. Ab- cía-Lario, A. Hernández, C. V. Jiménez, J.
(cpd)/ifcc-publications/ejifcc-(journal)/e-jour- stract: multicentre evaluation of liquid re- Minchinela, C. Perich, M. Simón. Current
nal-volumes/ejifcc-2008-vol-19/vol-19-n°-2/ agents for CK, CK-MB and LDH with deter- databases on biological variation: pros,
accurate-reference-intervals-are-required- mination of reference intervals on Hitachi cons and progress. Scandinavian Journal
for-accurate-diagnosis-and-monitoring-of- systems. Clin Chem 2001; 47: suppl. A30. of Clinical and Laboratory Investigation
patients/ 10. Ozarda Y. Reference intervals: current 2009;59:491-500 (http://www.tandfonline.
4. Borai A, Ichihara K, Al Masaud A, et al. Es- status, recent developments and future con- com/doi/abs/10.1080/00365519950185229)
tablishment of reference intervals of clinical siderations Biochem Med (Zagreb). 2016 15; 16. Westgard J. 2009. https://www.westgard.
chemistry analytes for the adult population 26(1): 5–1. com/biodatabase1.htm. Accessed on Feb 2
in Saudi Arabia: a study conducted as a part 11. Ichihara K, Boyd JC IFCC committee An 2017/
of the IFCC global study on reference values. appraisal of statistical procedures used in 17. Petersen PH, Fraser CG, Sandberg S,
Clin Chem Lab Med. 2016 May;54(5):843-55. derivation of reference intervals Clin Chem et al. The index of individuality is often a
5. CLSI and IFCC. C28-A3 document; De- Lab Med. 2010 Nov;48(11):1537-51. misinterpreted quantity characteristic. Clin
fining, establishing and verifying reference 12. Bimenya G S, Byarugaba W, Kalungi S, et Chem Lab Med 1999; 37:655–61.
intervals in the clinical laboratory: approved al.Initial attempt to establish population ref- 18. Petersen PH, Sandberg S, Fraser CG
guideline-third edition, 2008; 28:1-76. erence values for blood glucose and lipids in et al. Influence of index of individuality on
6. Ferruccio C, Hinzmann R and Panteghini Makerere university undergraduate students false positives in repeated sampling from
M Reference intervals: the way forward. Ann Afr Health Sci. 2006; 6(4): 247–253. healthy individuals. Clin Chem Lab Med.
Clin Biochem 2009; 46: 8–17. 13. Canadian Diabetes Association Clini- 2001; 39(2):160-5.

20 ANN SAUDI MED 2017  JANUARY-FEBRUARY  WWW.ANNSAUDIMED.NET