See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326634570

Distinguishing reference intervals and clinical decision limits – A review by

the IFCC Committee on Reference Intervals and Decision Limits

Article  in  Critical Reviews in Clinical Laboratory Sciences · July 2018

DOI: 10.1080/10408363.2018.1482256

CITATIONS READS

51 4,292

4 authors:

Yesim Ozarda Ilcol Kenneth Andrew Sikaris

Istanbul Health and Technology University University of Melbourne
94 PUBLICATIONS   2,213 CITATIONS    177 PUBLICATIONS   4,583 CITATIONS   

SEE PROFILE SEE PROFILE

Thomas Streichert Joseph Macri

University of Cologne Hamilton Health Sciences
181 PUBLICATIONS   3,869 CITATIONS    65 PUBLICATIONS   3,356 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Gestational diabetes View project

MXF / periodontitis research View project

All content following this page was uploaded by Yesim Ozarda Ilcol on 31 July 2018.

The user has requested enhancement of the downloaded file.

 Critical Reviews in Clinical Laboratory Sciences

ISSN: 1040-8363 (Print) 1549-781X (Online) Journal homepage: http://www.tandfonline.com/loi/ilab20

Distinguishing reference intervals and clinical

decision limits – A review by the IFCC Committee
on Reference Intervals and Decision Limits

Yesim Ozarda, Ken Sikaris, Thomas Streichert, Joseph Macri & on behalf of
IFCC Committee on Reference intervals and Decision Limits (C-RIDL)

To cite this article: Yesim Ozarda, Ken Sikaris, Thomas Streichert, Joseph Macri & on behalf
of IFCC Committee on Reference intervals and Decision Limits (C-RIDL) (2018): Distinguishing
reference intervals and clinical decision limits – A review by the IFCC Committee on Reference
Intervals and Decision Limits, Critical Reviews in Clinical Laboratory Sciences

To link to this article: https://doi.org/10.1080/10408363.2018.1482256

Published online: 26 Jul 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=ilab20
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES
https://doi.org/10.1080/10408363.2018.1482256

REVIEW ARTICLE

Distinguishing reference intervals and clinical decision limits – A review by

the IFCC Committee on Reference Intervals and Decision Limits
Yesim Ozardaa, Ken Sikarisb, Thomas Streichertc and Joseph Macrid; on behalf of IFCC Committee on
Reference intervals and Decision Limits (C-RIDL)
a
Department of Medical Biochemistry, Uludag University School of Medicine, Bursa, Turkey; bDepartment of Pathology, Melbourne
University, Parkville, Melbourne, Australia; cInstitute for Laboratory Medicine, Klinikum Bremen-Mitte, Bremen, Germany; dHamilton
General Hospital, Hamilton, Ontario, Canada

ABSTRACT ARTICLE HISTORY

Reference Intervals (RIs) and clinical decision limits (CDLs) are a vital part of the information sup- Received 28 February 2018
plied by laboratories to support the interpretation of numerical clinical pathology results. RIs Revised 3 May 2018
describe the typical distribution of results seen in a healthy reference population while CDLs are Accepted 25 May 2018
associated with a significantly higher risk of adverse clinical outcomes or are diagnostic for the Published online 23 July
2018
presence of a specific disease. However, as the two concepts are sometimes confused, there is a
need to clarify the differences between these terms and to ensure they are easily distinguished, KEYWORDS
especially because CDLs have a clinical association with specific diseases and risks, thereby imply- Reference intervals; clinical
ing that effective clinical interventions are available. It is important to note that, because popula- decision limits; C-RIDL;
tion-based RIs are derived from the range of values expected in a typical community population, clinical laboratories
laboratory results that fall outside a RI do not necessarily indicate a disease but rather that add-
itional medical follow-up and/or treatment may be warranted. In contrast, CDLs are associated
with a risk of specific adverse outcomes, and are commonly used to interpret laboratory test
results, including lipid parameters, glucose, hemoglobin A1c (HbA1c), and tumor markers, to
determine risk of disease, to diagnose or to treat. In recent years, the International Federation of
Clinical Chemistry and Laboratory Medicine (IFCC) Committee on Reference Intervals and Decision
Limits (C-RIDL) has focused primarily on RIs and has performed multicenter studies to obtain
common RIs. However, the broader responsibility of the Committee, from its name, includes
“decision limits”. C-RIDL now aims to emphasize the importance of the correct use of both RIs
and CDLs and to encourage laboratories to specify the appropriate information to clinicians as
needed. This review discusses RIs and CDLs in detail, describes the similarities and the differences
between these two important tools in laboratory medicine, and clearly explains the processes
used to define them. C-RIDL encourages the involvement of laboratory professionals in the estab-
lishment of both RIs and CDLs.

Abbreviations: ADA: American Diabetes Association; ALT: alanine aminotransferase; Apo A: apoli-
poprotein A; Apo B: apolipoprotein B; BMI: body mass index; BNP: B-type natriuretic peptide;
CA125: carbohydrate antigen 125; CDLs: clinical decision limits; CLSI: Clinical Laboratory Standards
Institute; C-RIDL: Committee on Reference Intervals and Decision Limits; CRP: C-reactive protein;
cTn: cardiac troponin; CVD: cardiovascular disease; CVi: within-individual CV; DM: diabetes melli-
tus; EFLM: European Federation of Laboratory Medicine; ERSPC: The European Randomized Study
of Screening for Prostate Cancer; GDM: gestational diabetes mellitus; GGT: gamma glutamyl-trans-
ferase; HAPO: Hyperglycemia and Adverse Pregnancy Outcome; HbA1c: hemoglobin A1c; HDL-C:
high-density lipoprotein cholesterol; IADPSG: International Association of Diabetes and Pregnancy
Study Groups; IFCC: International Federation of Clinical Chemistry and Laboratory Medicine; ISO:
International Organization for Standardization; LDL-C: low-density lipoprotein cholesterol; non-
HDL-C: non-HDL-cholesterol; PLCO: Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial;
PSA: prostate specific antigen; RIs: reference intervals; ROC: receiver operator characteristic

Introduction
Every laboratory request has a purpose, with specific
questions such as “Is the patient healthy?” “Is the
patient at risk of developing a disease?” “Is the patient
diseased?” and “Is the patient worsening?”. The value of
laboratory test results is in answering these questions
so that effective interventions can be applied to
improve clinical outcomes.

CONTACT Yesim Ozarda yesim@uludag.edu.tr Department of Medical Biochemistry, Uludag University School of Medicine, Bursa, Turkey
ß 2018 Informa UK Limited, trading as Taylor & Francis Group
2 Y. OZARDA ET AL.

The first question (“Is the patient healthy or not may be used. The most obvious example is fasting glu-
healthy?”) relates to reference intervals (RIs) that cose, where several decision limits are defined: a level
describe the typical distribution of results seen in an
3.9 mmol/L (
70 mg/dL) for the diagnosis of hypogly-
apparently healthy reference population [1]. In the trad- cemia [
2.2 mmol/L (
40 mg/dL) for life-threatening
itional approach to establishing RIs, termed the “direct” hypoglycemia] [20]; 5.6–6.9 mmol/L (100–124 mg/dL) for
approach, the RI is typically defined as the interval an increased risk of diabetes or prediabetes; and
between the two reference limits (2.5th and 97.5th per- 7.0 mmol/L (125 mg/dL) for diabetes mellitus
centiles) derived from the distribution of results (DM) [21].
obtained from a sample of the reference population [2]. RIs are focused on optimizing specificity (typically to
Historically such ranges have been called “normal 95%) while CDLs are also focused on optimizing sensi-
ranges” [3], but this term has long been considered to tivity for the disease. In other clinical circumstances that
be ambiguous [4] and subject to misuse [5] and is now place importance on both sensitivity and specificity,
formally avoided by the International Federation of optimal limits may be derived from receiver operator
Clinical Chemistry and Laboratory Medicine (IFCC) [2] characteristic (ROC) curves, which balance sensitivity
and International Organization for Standardization (ISO) and specificity [22] (Table 1). Ideally, a Bayesian
[6]. About 30 years ago, the IFCC published a series of approach is also required to balance pretest probability
six articles to clarify this concept [7–12], and over the or prevalence against sensitivity and specificity [23].
last decade, the IFCC’s Committee on Reference “Optimal limits” derived from ROC curves, an intermedi-
Intervals and Decision Limits (C-RIDL) has focused on ate category of threshold, are a compromise between
direct common RIs; it has developed guidelines on con- specificity and sensitivity [24].
ducting such studies and has conducted multicenter This review describes the differences between RIs
studies to derive common RIs at a national level and CDLs, gives definitions and provides examples to
[13–15]. An alternative approach is to perform data- help readers to avoid confusion. We also consider the
mining of results that have been generated as part of preanalytical, analytical, and postanalytical factors influ-
routine clinical pathology testing and then to use encing them and apply the principles of analytical qual-
appropriate statistical techniques to determine the ity specifications to the quality of RIs and CDLs. Finally,
underlying RIs. This is considered the “indirect” we discuss the importance of clearly specifying the use
approach [16]. of either RIs and/or CDLs on reports to improve post-
According to the directive on in vitro diagnostic
analytical interpretation.
medical devices of the European Union, diagnostic
manufacturers are now required to supply their clients
with appropriate RIs for use with their assay platforms RI definitions
and reagents [17], and the ISO 15189 standard for clin- RIs are derived from the reference population value dis-
ical laboratory accreditation states that each laboratory tribution, usually the central 95% interval, and describe
should periodically reevaluate its RIs [6]. Despite these a specific population using a minimum sample size of
efforts and requirements, the process of producing the
120, as recommended by the Clinical Laboratory
RIs is complex and the implementation of RIs in most
Standards Institute (CLSI) guideline EP28-A3c [2]. The
clinical laboratories is still incomplete [18]. Furthermore,
reference individuals form the reference sample group
obtaining RIs for pediatric, geriatric, and pregnant pop-
ulations is particularly challenging because of the tech- Table 1. Comparison of CDLs and RIs.
nical difficulty of enrolling a sufficient number of RIs CDLs
reference individuals [19]. Number of values Statistical limits Clinical threshold
The other questions (“Is the patient at risk of a devel- (two values) (one value)
Derivation A biological characteristic A decision regarding a
oping a disease, or is the patient diseased, or worsen- of the unaffected clinical condition
ing?”) are related to clinical decision limits (CDLs), population
Population General population Clinical population
where values above or below the threshold are associ- Based on 95% central interval of Clinical outcome studies,
ated with a significantly higher risk of adverse clinical the reference guidelines and consen-
outcomes or are defined as diagnostic for the presence distribution sus values, ROC curves,
predictive values
of a specific disease [6]. In contrast to the RIs, where Defined by experts Laboratory experts Clinicians and labora-
there are two limits (upper and lower), there is only one tory experts
Most promin- Laboratory experts Clinicians
CDL, which is usually an upper limit. However, accord- ent experts
ing to the likelihood of various clinical situations or dif- Consensus Well-defined [2] Still to be developed
standard
ferent clinical questions, multiple low and high CDLs

for measurement of the values from the reference
population. Through statistical analysis of the distribu-
tion of the obtained values, the reference limits are cal-
culated, and these limits then define the RIs. However,
at other times, such as for cardiac troponin (cTn) [25,26]
and carbohydrate antigen 125 (CA125) [27,28], the 99th
percentile is used as the appropriate upper reference
limit. The rationale for the use of the 99th centile for
cTn or CA125 is that a false positive rate of 1.0% would
cause 60% less undue clinical concern compared to a
false positive rate of 2.5%.
While concentrations above the 97.5th (or 99th) per-
centiles of the apparently healthy population are associ-
ated with an absolute increase in risk for clinical
outcomes, it could mistakenly be assumed that patients
with results within the RI are healthy and have no risk.
However, results within the RI do not all necessarily
imply a minimal risk of adverse clinical outcomes, and
risk still varies within the RI, for example, across the ter-
tiles of C-reactive protein [29], across the quartiles of
cTn [30], or below and above the median level of pros-
tate specific antigen (PSA) [31].
It should be noted that RIs have diminishing utility
when the within-individual biological variability is sig-
nificantly tighter than the between-individual variability,
or in other words, when a patient can have a clinically
significant change but still remain within the wider
population RI [32].
Direct approach
With this approach, individuals from a healthy popula-
tion (the reference population) are selected for sam-
pling based on defined criteria. Specimens are then
collected from these individuals and analyzed for the
selected measurand. The designation of good health
and determination of normalcy for a candidate refer-
ence individual may involve a variety of examinations
such as history and physical examination and/or certain
clinical laboratory tests [2]. Therefore, the identification
of a healthy status can be self-declared or confirmed by
the absence of disease. Confirmation of the absence of
disease can be by clinical examination that includes
specific extra testing; examples include conducting
semen analysis to confirm testicular health when esti-
mating testosterone RIs in men [33] or conducting B-
type natriuretic peptide (BNP) testing to define healthy
cardiac function when defining cTn RIs [34]. The CLSI
guideline EP28-A3c [2] also describes the parametric or
nonparametric statistics that should be applied to the
reference values.
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 3
Indirect approach
Although the direct approach to establishing RIs has
been recommended by CLSI guideline EP28-A3c [2], in
recent years there has been increasing interest in indir-
ect methods [35], particularly in situations where it is
difficult to recruit healthy patients (e.g. pediatrics)
[36,37]. The rationale for this approach is that routine
clinical pathology databases, which often contain many
thousands or millions of results from many hundreds or
thousands of patients, also include results from patients
without disease, and that by using clinical exclusion
[38] and statistical approaches [39,40], the underlying
reference distribution can be distinguished. The advan-
tages of the indirect approach compared to the direct
approach include that it is faster and less expensive and
that it more easily overcomes ethical problems such as
obtaining informed consent, issues that relate particu-
larly to the elderly or children, especially newborns. The
disadvantage of the approach lies in the statistical and
clinical assumptions made. However, another assump-
tion is that there is no apparent disease affecting the
distribution of that analyte, although the observed ref-
erence distribution may not be from “healthy” individu-
als. The increased interest in data mining is gradually
improving our understanding of the many factors to be
considered, including preanalytical conditions, analytical
stability, defined mechanisms of outlier removal, statis-
tical assumptions, minimal sample sizes, partitioning,
and ethical considerations in the use of patient data.
The C-RIDL has recently published a review of the
strengths and weaknesses of the indirect approach [41].
Common RIs (including harmonized RIs)
There are two types of common RIs. The first are object-
ive RIs, which have many prerequisites [42] and are
defined by well-conducted multicenter studies [43–45],
and the second are subjective conventional RIs, defined
by survey(s) and guidance from a group of experts
using the harmonization approach [46–48]. The most
important question is “What is a significant difference in
reference limits?” International guidelines indicate that
if analytical methods and reference populations are the
same, a subjective approach can be used to adopt com-
mon limits [2]. The ISO 15189 quality standard also
encourages the use of interpretive comments at a local
or national level [6]. When comparing candidate com-
mon RIs, a pragmatic consideration is, what is a clinic-
ally acceptable flag rate variation across analytical
platforms and clinical populations [40]. The flag rate
represents the balance between false positives, true
4 Y. OZARDA ET AL.
positives and the understanding of the expected preva-
lence of disease in that particular population.
Significant figures used in RIs
The number of significant figures in the reported value
is defined largely by the analytical confidence [49]. The
number of significant figures used in a reference limit
should also be agreed upon when defining common RIs
[50]. For example, is a serum albumin value of 45.4 g/L
greater than an upper reference limit of 45 g/L?
Conversely, if a research study has defined the 97.5th
percentile for the serum albumin reference sample to
be 45.4 g/L, should this be rounded down to 45 g/L
(slightly increasing the false positive rate) or should it
be rounded up to 46 g/L (ensuring the false positive
rate does not rise above 2.5%)?
CDLs definitions
As previously mentioned, RIs relate to studies based on
apparently healthy individuals while CDLs are based
mainly on clinical outcome studies (e.g. prospective
cohort studies, meta-analysis), guidelines and consensus
values. It is important that these two approaches are
clearly distinguished [51,52].
The approaches to identifying CDLs can be catego-
rized [1]:
1. The Bayesian approach is probably the most evi-
dence-based approach to modifying the manage-
ment of the patient. Following these criteria, a
value resulting from a diagnostic test that serves
to distinguish between two clinical subgroups is
based on stated assumptions regarding (i) the clin-
ical sensitivity of the diagnostic test; (ii) the clinical
specificity of the diagnostic test; (iii) the relative
distribution of individuals between the two sub-
groups; and (iv) the clinical costs of misclassifica-
tion [4]. This concept has been applied to
troponins [23,34], where the balance among these
factors had to be considered to maintain the spe-
cificity and sensitivity required for patients pre-
senting with chest pain.
2. The epidemiological approach for defining CDLs is
based on clinical outcome derived from popula-
tion-based studies and is typically applied to lipid
parameters (total cholesterol, low density lipopro-
tein cholesterol [LDL-C], high density lipoprotein
cholesterol [HDL-C], etc.) [53] as well as to inter-
vention-based studies such as diabetes interven-
tion [54].
3. The physiopathological approach involves the use
of “critical values” that represent a pathophysio-
logical state with such variance from normal as to
be life-threatening unless prompt action is taken.
While many clinical endpoints can be difficult to
define, the endpoint of mortality is clear and,
because it defines the risk of dying or of major
patient harm (e.g. neonatal hypoglycemia), it
defines a particular set of high risk CDLs often
called “critical values” [55]. Prognostic CDLs for
tumor markers may also belong to this category.
Regardless of the approach used and the clinical
endpoints defined, CDLs are defined according to an
arbitrary risk level.
Diagnostic and staging cutoff values
A diagnostic cutoff is defined by a question such as,
“When does a patient have diabetes or prediabetes?”
For hemoglobin A1c (HbA1c), the definitions of the
American Diabetes Association are <39 mmol/mol
(<5.7%) for healthy, 39–46 mmol/mol (5.7–6.4%) for
prediabetic, and 48 mmol/mol (6.5%) for diabetic
status [56]. Prediabetes, defined as impaired glucose
tolerance, impaired fasting glucose, or intermediate
HbA1c, is of particular importance as a diagnostic cat-
egory because it has been associated with an increased
risk of cardiovascular disease (CVD) [57], and cardiovas-
cular risk has also been shown to be increased in peo-
ple with a fasting glucose concentration as low as
5.6 mmol/L or HbA1c of 39 mmol/mol [58].
While risk may begin at a certain threshold, the def-
inition of what constitutes a significantly increased risk
is usually defined arbitrarily by an expert consensus
group. An example is the definition of gestational dia-
betes mellitus (GDM) from the International Association
of Diabetes and Pregnancy Study Groups (IADPSG) [59],
which was established using the results of the
Hyperglycemia and Adverse Pregnancy Outcome
(HAPO) study [60] and which used an odds ratio of
1.75 in selecting the likelihood of complications such
as macrosomia. The prevalence of GDM would have
been altered if an odds ratio of 2.0 had been chosen
instead [61].
Tumor markers represent another group where cut-
off values are used instead of RIs. These markers are
often present in nondiseased people at concentrations
that overlap significantly with those in people with
malignancies. This typically impedes the general use
of tumor markers as a screening tool. For staging
purposes, the International Germ Cell Consensus
Classification represents a tumor marker-based staging

system in which three biomarkers contribute to the
staging of men suffering from nonseminomatous germ
cell tumors; they are classified into three prognostic
groups, depending on the levels of the markers: good
(alpha fetoprotein [AFP] < 1000 ng/mL, human chorionic
gonadotrophin [(hCG] < 5000 IU/L, lactate dehydrogen-
ase [LDH] < 1.5 upper RL); intermediate (AFP
1000–10000 ng/mL, hCG 5000–50000 IU/L, LDH 1.5–10
upper RL) and poor (AFP >1000 ng/mL, hCG >50000 IU/
L, LDH >10 upper RL) [62]. This help in the selection
of the appropriate therapy [63]. Interestingly, this classi-
fication combines two strategies of defining decision
limits: for AFP and hCG, absolute cutoffs are used,
whereas for LDH, multiples of the upper reference
limit(s) are used.
Critical values
The term “critical value,” also known as critical result,
panic value, or alert value, represents a pathophysio-
logical state different from normal that poses a risk to a
patient’s life unless immediate action is taken [64].
Currently, the use of the term, panic value, is discour-
aged, because it suggests emotional stress, and because
it is contrary to the process of communicating informa-
tion clearly [55].
There is an overall risk of mortality in any tertiary set-
ting compared to the general community, but the
increase in that risk that is associated with laboratory
results can be further used as a flag for clinical action.
As results deviate from normality, they represent a loss
of homeostasis and the risk of decompensation and,
ultimately, of death. Any result that is outside the RI is
associated with a relative increase in the risk of mortal-
ity [65], but the odds ratio for death varies from analyte
to analyte [66]. The perceived risk of death should be
understood, for example, a 25 versus 50% mortality risk
in patients who have been admitted in a hospital, and it
becomes a specifically stated arbitrary setting that can
be agreed upon by clinicians and laboratorians. It is
important to note that in some circumstances (e.g.
hyponatremia), the abnormality may directly reflect the
cause of death and the need to focus treatment on cor-
recting the abnormality is implied, while in others (e.g.
hypernatremia), the abnormality may be an indirect
reflection of the cause of death (e.g. dehydration and
hypovolemia) and correction of the abnormality is sec-
ondary to addressing the underlying mechanisms [67].
Reference change values
The progress of a disease or recovery from it is
often reflected by the dynamics of test results (delta
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 5
values/delta change). Reference change values,
which can be defined by absolute (± delta) or relative
(± delta%) means, can help in the interpretation of the
results of serial measurements. The example of absolute
and relative kinetic changes of cTn in patients with
acute coronary syndrome shows that serial measure-
ments may assist in diagnosis and may be used to rule
out non-ST elevation myocardial infarction [68]. The
prerequisites to calculate delta changes from serial
measurements are a well-accepted clinical algorithm
with defined time points (e.g. baseline, 3 h, 6 h for cTn)
and the knowledge of the intra-individual biological
variation of the measurand (within-individual CV [CVi]).
Although the source of biological variation data is typic-
ally from a healthy reference population, its application
to disease assumes that biological variation is the same
in chronic disease as in health [69], and this has been
adopted as a surrogate for clinically significant changes.
The choice of statistical model used for clinical monitor-
ing varies according to clinical circumstance [70].
What is an acceptable risk?
The arbitrary setting of the risk level to be flagged, pri-
marily a clinical decision, is based on the balance of
false positive alarms that are acceptable compared to
the clinical outcome benefit gained from true positive
alarms. It is important that the rationale behind any
choice of risk level be explained. Whilst they are primar-
ily clinician-defined, it is essential that laboratory profes-
sionals are involved in their definition for several
reasons. Firstly, the laboratory will need to manage the
reporting mechanisms. Secondly, some assays are not
standardized, and the risk levels could be method-
dependent. Thirdly, it is important that results outside
CDLs are not due to preanalytical or analytical artifacts,
and that this potential is managed and communicated
when necessary.
There are many other points to mention, including
that the laboratory may report to several clinical set-
tings that may have different clinical questions. Gamma
glutamyl-transferase thresholds can be set to determine
alcohol abstinence or alcohol abuse [71]. Similarly, an
alanine aminotransferase cutoff could be set to flag
everyone in the emergency department with possible
paracetamol overdose (e.g. >1000 U/L) [72], everyone in
the blood bank with possible subclinical viral hepatitis
(e.g. >20 U/L) [73], everyone in the lipid clinic with pos-
sible fatty liver (e.g. >26 U/L) [74], or everyone in a
health screen who may be possibly obese (e.g. >40 U/L)
[75] (although there are easier ways to define obesity).
6 Y. OZARDA ET AL.
Harmonization of CDLs
The implementation of shared electronic health records
has highlighted the need for standardized or harmon-
ized reporting of laboratory test results [47]. Most of the
criteria checked during the harmonization of RIs, such
as method specificity, partitioning based on age or gen-
der, and preanalytical differences, apply to the harmon-
ization of CDLs [47]. In addition, the clinical settings
could form an additional obstacle for the harmonization
of CDLs. The measurand could have a central and well-
defined role in the decision-making for a specific dis-
ease or clinical situation, but there could be several
CDLs, such as for diagnosis of anemia (e.g. hemoglobin
130 g/L in men, 120 g/L in nonpregnant women, 110 g/L
in children) or for performing a blood transfusion (e.g.
hemoglobin 70–80 g/L or according to the clinical situ-
ation; if the patient is stable, transfusion might not be
needed even at these concentration levels) [1].
Significant figures used in CDLs
The issues regarding the number of significant figures
in an analytical result are the same as discussed above
for RIs. The number of significant figures used in CDLs
should ideally be the same as that used in the clinical
study that determined that limit. Problems can arise
when reporting unit changes occur. For example, the
CDL for the diagnosis of diabetes is 6.5% using National
Glycohemoglobin Standardization Program (NGSP) units
and 48 mmol/mol using IFCC units, while an IFCC result
of 47 mmol/mol converts to an NGSP result of 6.4505%
(both nondiabetic), but the rounded NGSP value
becomes 6.5% (diabetic).
Preanalytical, analytical, and
postanalytical factors
Preanalytical factors
General awareness of preanalytical problems that may
cause false positive results is needed. These include
sample contamination, inadequate transport conditions,
incorrect timing of sample collection (e.g. for toxico-
logical testing), and delay in sample processing.
Generally, the preanalytical considerations for RIs
involve biological (i.e. sampling time in relation to bio-
logical rhythms, fasting or non-fasting and physical
activity) and methodological factors (i.e. sample collec-
tion techniques, type of additives, collection with or
without tourniquet and sampling equipment, specimen
handling, transportation, time and speed of centrifuga-
tion, and storage conditions). For reproducibility and
standardization, it is essential that the preanalytical
aspects are accurately defined and described, because
the preanalytical phase is known to have the highest
error rate in the total testing process [76].
As RI studies are generally well-designed and con-
trolled, and efforts have been made to standardize
them according to guidelines, it is accepted that prea-
nalytical problems occur less often. However, it should
be noted that this extra attention and effort may result
in narrower RIs than if the usual preanalytical processes
were to be used.
Age and gender are acknowledged as major factors
affecting RIs. The main factors that cause the difference
in RIs between the genders are muscle mass, body
mass index (BMI), different organs (uterus, ovaries,
breast, testis, and prostate), related hormonal control
systems, and higher iron demands in females related to
blood loss with menstruation, pregnancy and breast-
feeding [19].
Many studies of specific RIs for geriatric age groups
have had methodological issues, which means that
careful consideration is required before applying these
to clinical care [77].
The preanalytical effect of age, gender, race, ethni-
city, time of day and collection processes that have
been acknowledged as factors with a modifying effect
on RIs may be equally relevant to CDLs. Gender can
affect the decision limits for HDL-C (<1.0 mmol/L in
men and <1.3 mmol/L in women) in the definition of
metabolic syndrome [78]. Race may also change clinical
decision thresholds, and the typical decision thresholds
for overweight (BMI >25 and <30 kg/m2) and obesity
(> ¼30 kg/m2) may vary with race; for example, there is
evidence that Indian adults should have much lower
thresholds [79]. Diurnal effects should be considered in
setting clinical decision thresholds, and there has been
a recent shift from fasting to nonfasting lipid studies
because they have been generally found to be equiva-
lent in their ability to act as clinical decision points [80].
Collection conditions during trials that establish clinical
decision thresholds should continue to be applied in
clinical practice, although this is not always the case; for
example, in the global trial that established internation-
ally agreed-upon criteria for gestational diabetes, glu-
cose samples were stated to be collected on ice, a
practice not often observed in routine phlebotomy [81].
While RIs for pediatric bilirubin can be defined, for
early identification of biliary atresia the focus changes
from the RI to the CDL that is applied to the direct bili-
rubin measurement to distinguish biliary obstruction
from conjugation defects [82].
Creatinine RIs have long been defined by gender [83]
and are now increasingly defined by age [84]. Without
considering the expected differences in muscle mass

between individuals, it is difficult to achieve the primary
purpose of serum creatinine measurement, which is to
gauge renal function. Serum creatinine should now be
routinely translated to an estimation of eGFR using
acknowledged formulae that account for the impact of
age and gender on muscle mass. While appropriate
serum creatinine RIs could also be translated to eGFR
RIs [85], the primary purpose for eGFR is to allow appli-
cation of the profession-defined CDLs for eGFR that
describe the risk associated with the various stages of
chronic kidney disease. Creatinine reporting is therefore
associated with gender and age-related RIs, while eGFR
reporting is associated with CDLs that take gender and
age into account.
In the case of nutritional factors such as vitamin D,
Vitamin B12, and ferritin, where the reference popula-
tion has a high prevalence of deficiency, the use of cor-
rect RIs is generally problematic because of widespread
subclinical deficiency, and these are typically handled as
CDLs [86–88].
Analytical factors
In 1999, the IFCC participated in a consensus meeting
in Stockholm as part of a global effort to define analyt-
ical quality specifications [89]. The hierarchy of
approaches to determine analytical quality require-
ments was headed by clinical outcome studies, but
because these were seldom available, the second level
of the hierarchy was strategies supported by clinical
opinion and biological variability. These principles sug-
gested that CDLs based on clinical outcome were super-
ior to RIs that were based on biological variation [90].
Fifteen years later in Milan, at a European Federation of
Laboratory Medicine (EFLM) meeting, approaches based
on clinical outcome and biological variation were
endorsed [91] and specifically applied to the postanalyt-
ical phase, including CDLs and RIs [92].
Analytical quality affects the reliability of both RIs
and CDLs. The biological variability theory suggests
that the desirable bias for RI classification takes into
account intraindividual and interindividual variability
[<0.25  (CVi2 þ CVg2)1/2] and that it will prevent an
unacceptable increase in the proportion of healthy
individuals flagged as outside RIs. Analytical quality will
similarly affect the application of CDLs, although the
impact is defined not by the statistics of the reference
population distribution but by the clinical risk defini-
tions as well as the prevalence of disease [93].
Increasing measurement uncertainty generally causes
greater clinical uncertainty; similarly, the impact of
uncorrected measurement bias will lead to clinical bias.
The traceability of method calibration is vitally
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 7
important for both RIs and CDLs. Neither universal CDLs
(e.g. for lipids and HbA1c) nor common CDLs (e.g. for
routine analytes) can be clinically reliable without trace-
ability and analytical quality standards [94]. When
adopting RIs derived externally, a laboratory can per-
form a local validation (e.g. 20 healthy people) [95],
although these procedures are not available for adjust-
ing CDLs and we are reliant on other techniques to con-
firm method suitability (e.g. external quality assurance).
Postanalytical factors: reporting CDLs and RIs
The CLSI guideline EP28-A3c states that “To avoid con-
fusion, the working group encourages laboratories to
report either decision limits or RIs, but not both, with a
clear indication of which has been used” [2]. The stand-
ard also states that “When decision limits determined
by national or worldwide consensus exist, these limits,
rather than RIs, should be reported,” thus prioritizing
the preference for CDLs over RIs. Current practice is to
report CDLs for the following measurands: total choles-
terol, LDL-C, HDL-C, non-HDL-cholesterol (non-HDL-C),
triglycerides, apolipoprotein A (Apo A), apolipoprotein B
(Apo B) and HbA1c. However, in the section of the med-
ical decision limits of the CLSI guideline EP28-A3c, an
example of a report shows that the CDLs for total chol-
esterol and HDL-C are listed in the column entitled “RI”
[2]. With many routine laboratory reports, RIs and CDLs
are also listed in the “RI” column. This practice disre-
gards the basis of the terminology and confuses the dis-
tinction between RIs and CDLs. If this column of the
CLSI guideline EP28-A3c were to be updated and enti-
tled “Reference”, not “RI”, the reference comparator
could then be either a RI or a CDL. As C-RIDL, we would
like to emphasize that the distinction between RIs and
CDLs should be clear in the report; such a distinction
will have valuable effects on postanalytical quality. A
clearer distinction could also be facilitated with new
presentation formats including graphical ele-
ments [96–98].
The aim of report interpretation is to lead to clinical
decisions that benefit clinical outcomes, and the basis
for those decisions usually starts with the rationale for
flagging an important result.
The rationale for flagging elevated cholesterol
related results is based on the two main purposes for
this testing; primary and secondary prevention. In pri-
mary prevention, risk thresholds and lower treatment
goals for preventative statin therapy are usually defined
through epidemiological studies; while in secondary
prevention, clinical studies have defined even stricter
treatment goals. It is also not possible to tailor the
report to a patient’s individual clinical need if the
8 Y. OZARDA ET AL.
clinical background of the patient has not been shared
with the laboratory. Nevertheless, the issue of variable
interpretation could be covered with an interpretative
comment (including a table or link) that details the clin-
ical situation and the agreed-upon corresponding deci-
sion thresholds.
Desirable HDL-C values, according to the European
Guidelines, are 1.03 mmol/L (40 mg/dL) for males
and 1.20 mmol/L (46 mg/dL) for females [99]. HDL-C
is an example of how physiology, which is mainly a RI
concern, also impacts the risk point. In the case of HDL-
C, decision limits can be used to categorize people
at increased risk (<1.04 mmol/L) or decreased risk
(>1.55 mmol/L) of coronary artery disease based on
data from large population studies [99]. For example, it
is well-established that the Turkish population has a
high prevalence of coronary heart disease associated
with some known risk factors [100]. Turks have distinct-
ively low concentrations of HDL-C that are associated
with elevated hepatic lipase activity and fasting trigly-
ceridemia [101]. Therefore, it would be better to state
only the population-based RIs in the RI column of the
laboratory report, and to state the CDL clearly as a com-
ment, for example at the bottom of the report, when a
parameter has a well-defined CDL [102].
PSA is a good example for contrasting the applica-
tion of RIs and CDLs. The first RI for PSA, defined as
<4.0 lg/L, was based on studies of apparently healthy
men [103]. It was soon found that this upper reference
limit was not specific in older men, and possibly too
specific (and insensitive) in younger men, so age-related
PSA RIs were developed and adopted [104]. The consid-
erable debate around PSA screening was addressed in
the European Randomized Study of Screening for
Prostate Cancer (ERSPC) [105] and Prostate, Lung,
Colorectal, and Ovarian Cancer Screening Trial (PLCO)
[106] studies, and the combined results of these studies,
which are now considered to agree [107], indicated that
PSA testing could prevent death from prostate cancer
in some men. In the ERSPC trial, a uniform cutoff of
3.0 lg/L (based on World Health Organization calibra-
tion) was stated as the threshold, which led to signifi-
cant improvements in clinical outcome. Therefore,
according to a quality hierarchy prioritizing CDLs based
on clinical outcome over RIs based on healthy individu-
als, the cutoff of 3.0 lg/L must be considered at least as
equivalent to any RI. Other PSA CDLs can also be con-
sidered; Epidemiological studies have shown that the
concentration of PSA at age 60 predicts the lifetime risk
of metastasis and death from prostate cancer [108],
and, although men aged 60 with concentrations below
the median (
1.0 lg/L) may harbor prostate cancer, it is
unlikely to become life-threatening; thus, the clinical
decision to exempt these men from further screening
could be made to allow greater focus on men with
higher concentrations.
Concluding remarks
The principles for describing RIs and CDLs should be
kept separate, and this distinction should be clear in
the laboratory report. We believe that this would greatly
improve the postanalytical quality of interpretation and
facilitate the national and international adoption of
common RIs (including harmonized RIs) and CDLs.
The IFCC C-RIDL workplan includes participation in
the creation of guidelines for the definition of CDLs.
The requirements for the creation of CDLs could include
aspects relating to:
1. the process; that is, the involvement of both clini-
cians and laboratory professionals,
2. the definition; that is, the precise definition of the
clinical question (and clinical decision) to
be addressed,
3. the clinical confidence with which the question
needs to be answered, and finally,
4. the mechanisms for the broader implementation
of CDLs at the national and international level.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
[1] Ceriotti F, Henny J. Are my laboratory results normal?
Considerations to be made concerning reference
intervals and decision limits. EJIFCC. 2008;19:1–9.
[2] CLSI. Defining, Establishing, and Verifying Reference
Intervals in the Clinical Laboratory; Approved
Guideline. 3rd ed. CLSI EP28-A3c. Wayne,
Pennsylvania: Clinical and Laboratory Standards
Institute; 2010.
[3] Keyser JW. The concept of the normal range in clin-
ical chemistry. Postgrad Med J. 1965;41:443–447.
[4] Sunderman FW. Current concepts of “normal values,”
“reference values,” and “discrimination values,” in
clinical chemistry. Clin Chem. 1975;21:1873–1877.
[5] Murphy EA, Abbey H. The normal range-a common
misuse. J Chronic Dis. 1967;20:79–88.
[6] International Organization for Standardization.
Medical laboratories – Requirements for quality and
competence. ISO 15189:2012. [cited 2018 Jun 9].
Available from: https://www.iso.org/standard/56115.
html
[7] Solberg HE. International Federation of Clinical
Chemistry (IFCC). Approved recommendation (1986)
on the theory of reference values. Part 1. The

concepts of reference values. J Clin Chem Clin
Biochem. 1987;25:337–342.
[8] Petit Clerc C, Solberg HE. International Federation of
Clinical Chemistry (IFCC). Approved recommendation
(1987) on the theory of reference values. Part 2.
Selection of individual for the production of refer-
ence value. J Clin Chem Clin Biochem. 1987;25:
337–342.
[9] Solberg HE, Petit, Clerc C. International Federation of
Clinical Chemistry (IFCC). Approved recommendation
on the theory of reference values. Part 3. Preparation
of individuals and collection of the specimens for the
production of reference values. J Clin Chem Clin
Biochem. 1988;26:593–598.
[10] Solberg HE, Stamm D. International Federation of
Clinical Chemistry (IFCC). Approved recommendation
(1991) on the theory of reference values. Part 4.
Control of analytical variation in the production,
transfer and application of reference values. J Clin
Chem Clin Biochem. 1991;29:531–535.
[11] Solberg HE. International Federation of Clinical
Chemistry (IFCC). Approved recommendation on the
theory of reference values. Part 5. Statistical treat-
ment of collected reference values. Clin Chem Clin
Biochem. 1987;25:645–656.
[12] Dybkaer R, Solberg HE. International Federation of
Clinical Chemistry (IFCC). Approved recommendation
(1987) on the theory of reference values. Part 6.
Presentation of observed values related to reference
values. J Clin Chem Clin Biochem. 1987;25:657–662.
[13] Ozarda Y, Ichihara K, Barth JH, et al. Committee on
Reference Intervals and Decision Limits (C-RIDL),
IFCC. Protocol and standard operating procedures for
common use in worldwide multicenter study on ref-
erence values. Clin Chem Lab Med. 2013;51:
1027–1040.
[14] Ichihara K, Ozarda Y, Barth JH, et al. Committee on
Reference Intervals and Decision Limits (C-RIDL),
IFCC. A global multicentre study on reference values:
1. Assessment of methods for derivation and com-
parison of reference intervals. Clin Chim Acta. 2017;
467:70–82.
[15] Ichihara K, Ozarda Y, Barth JH, et al. Committee on
Reference Intervals and Decision Limits (C-RIDL),
IFCC. A global multicentre study on reference values:
2. Exploration of sources of variation across the
countries. Clin Chim Acta. 2017;467:83–97.
[16] Baadenhuijsen H, Smit JC. Indirect estimation of clin-
ical chemical reference Intervals from total hospital
patient data: application of a modified Bhattacharya
procedure. J Clin Chem Clin Biochem. 1985;23:
829–839.
[17] Directive 98/79/EC to the European Parliaments and
of the Council 1998-10-27 on in vitro Diagnostic
Medical Devices. 1998 OJ (L 331), 1. [cited 2018 Jun
15]. Available from: https://ec.europa.eu/growth/sin-
gle-market/european-standards/harmonised-stand-
ards/iv-diagnostic-medical-devices_en
[18] Jones GR. Validating common reference intervals in
routine laboratories. Clin Chim Acta. 2014;432:
119–121.
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 9
[19] Sikaris K. Physiology and its importance for reference
intervals. Clin Biochem Rev. 2014;35:3–14.
[20] EMA: Guideline on clinical investigation of medicinal
products in the treatment or prevention of diabetes
mellitus. European Medicines Agency; 2012 [cited
2018 June 15]. Available from: http://www.ema.eur-
opa.eu/docs/en_GB/document_library/Scientific_
guideline/2012/06/WC500129-256.pdf
[21] Expert Committee on the Diagnosis Classification of
Diabetes Mellitus. Report of the expert committee on
the diagnosis and classification of diabetes mellitus.
Diabetes Care. 2003;26:S5–S20.
[22] Zweig MH, Campbell G. Receiver-operating character-
istic (ROC) plots: A fundamental evaluation tool in
clinical medicine. Clin Chem. 1993;39:561–577.
[23] Kallner A. Bayes' theorem, the ROC diagram and ref-
erence values: Definition and use in clinical diagnosis.
Biochem Med (Zagreb). 2018;28:010101.
[24] Reach G. Which threshold to detect hypoglycemia?
Value of receiver-operator curve analysis to find a
compromise between sensitivity and specificity.
Diabetes Care. 2001;24:803–804.
[25] Greene DN, Tate JR. Establishing consensus-based,
assay specific 99th percentile upper reference limits
to facilitate proper utilization of cardiac troponin
measurements. Clin Chem Lab Med. 2017;55:
1675–1682.
[26] Clerico A, Zaninotto M, Ripoli M, et al.; on the behalf
of the Study Group on Cardiovascular Risk
Biomarkers of the Italian Society of Clinical
Biochemistry (SIBioC). The 99th percentile of refer-
ence population for cTnI and cTnT assay: method-
ology, pathophysiology and clinical implications. Clin
Chem Lab Med. 2017;55:1634–1651.
[27] Bast RC Jr, Klug TL, St John E, et al. A radioimmuno-
assay using a monoclonal antibody to monitor the
course of epithelial ovarian cancer. N Engl J Med.
1983;309:883–887.
[28] Vuento MH, Stenman UH, Pirhonen JP, et al.
Significance of a single CA 125 assay combined with
ultrasound in the early detection of ovarian and
endometrial cancer. Gynecol Oncol. 1997;64:141–146.
[29] Ridker PM, Glynn RJ, Hennekens CH. C-reactive pro-
tein adds to the predictive value of total and HDL
cholesterol in determining risk of first myocardial
infarction. Circulation. 1998;97:2007–2011.
[30] Olson F, Engborg J, Grønhøj MH, et al. Association
between high-sensitive troponin I and coronary
artery calcification in a Danish general population.
Atherosclerosis. 2016;245:88–93.
[31] Loeb S, Roehl KA, Antenor JA, et al. Baseline pros-
tate-specific antigen compared with median pros-
tate-specific antigen for age group as predictor of
prostate cancer risk in men younger than 60 years
old. Urology. 2006;67:316–320.
[32] Harris EK, Boyd JC. On dividing reference data into
subgroups to produce separate reference ranges.
Clin Chem. 1990;36:265–270.
[33] Sikaris K, McLachlan RI, Kazlauskas R, et al.
Reproductive hormone reference intervals for healthy
fertile young men: evaluation of automated platform
assays. J Clin Endocrinol Metab. 2005;90:5928–5936.
10 Y. OZARDA ET AL.
[34] Krintus M, Kozinski M, Boudry P, et al. Defining nor-
mality in a European multinational cohort: Critical
factors influencing the 99th percentile upper refer-
ence limit for high sensitivity cardiac troponin I. Int J
Cardiol. 2015;187:256–263.
[35] Haeckel R, Wosniok W, Arzideh F, et al. Critical com-
ments to a recent EFLM recommendation for the
review of reference intervals. Clin Chem Lab Med.
2017;55:341–347.
[36] Zierk J, Arzideh F, Haeckel R, et al. Pediatric reference
intervals for alkaline phosphatase. Clin Chem Lab
Med. 2017;55:102–110.
[37] Grecu DS, Paulescu E. Quality in post-analytical
phase: Indirect reference intervals for erythrocyte
parameters of neonates. Clin Biochem. 2013;46:
617–621.
[38] Kouri T, Kairisto V, Virtanen A, et al. Reference inter-
vals developed from data for hospitalized patients:
Computerized method based on combination of
laboratory and diagnostic data. Clin Chem.
1994;40:2209–2215.
[39] Bhattacharya CG. A simple method of resolution of a
distribution into Gaussian components. Biometrics.
1967;23:115–135.
[40] Jones G, Horowitz G. Reference intervals Data mining:
getting the right paper. Am J Clin Pathol.
2015;144:526–527.
[41] Jones G, Haeckel R, Loh TP, et al. Indirect methods
for reference interval determination – review and
recommendations. Clin Chem Lab Med. 2018. DOI:
10.1515/cclm-2018-0073
[42] Ceriotti F. Prerequisites for use of common reference
intervals. Clin Biochem Rev. 2007;28:115–121.
[43] Colantonio DA, Kyriakopoulou L, Chan MK, et al.
Closing the gaps in pediatric laboratory reference
intervals: a CALIPER database of 40 biochemical
markers in a healthy and multiethnic population of
children. Clin Chem. 2012;58:854–868.
[44] Adeli K, Raizman JE, Chen Y, et al. Complex biological
profile of hematologic markers across pediatric, adult,
and geriatric ages: Establishment of robust pediatric
and adult reference intervals on the basis of the
Canadian Health Measures Survey. Clin Chem.
2015;61:1075–1086.
[45] Hilsted L, Rustad P, Aksglaede L, et al.
Recommended Nordic paediatric reference intervals
for 21 common biochemical properties. Scand J Clin
Lab Invest. 2013;73:1. 9.
[46] Berg J. The UK Pathology Harmony initiative; The
foundation of a global model. Clin Chim Acta.
2014;432:22–26.
[47] Tate JR, Sikaris KA, Jones GR, et al. and paediatric ref-
erence intervals in Australia and New Zealand: An
evidence-based approach for establishing. A First
Panel of Chemistry Analytes. Clin Biochem Rev.
2014;35:213–235.
[48] Koerbin G, Tate JR. Harmonising adult reference
intervals in Australia and New Zealand – the continu-
ing story. Clin Biochem Rev. 2016;37:121–129.
[49] Hawkins RC, Johnson RN. The significance of signifi-
cant figures. Clin Chem. 1990;36:824.
[50] Tate JR, Koerbin G, Adeli K. Opinion paper: Deriving
harmonised reference intervals – global activities.
EJIFCC. 2016;27:48–65.
[51] Lindstedt G, Tryding N. There is difference between
decision limits and reference intervals. Reference
intervals are based on measurements in healthy indi-
viduals, decision limits on measurements in patients.
Lakartidningen. 2007;104:2076–2079.
[52] Waise A, Price HC. The upper limit of the reference
range for thyroid-stimulating hormone should not be
confused with a cut-off to define subclinical hypothy-
roidism. Ann Clin Biochem. 2009;46:93–98.
[53] Abbott RD, Wilson PW, Kannel WB, et al. High density
lipoprotein cholesterol, total cholesterol screening,
and myocardial infarction. The Framingham Study.
Arteriosclerosis 1988;8:207–211.
[54] Diabetes Control and Complications Trial Research
Group, Nathan DM, Genuth S, et al. The effect of
intensive therapy of diabetes on the development
and progression of long-term complications in
insulin-dependent diabetes mellitus. N Engl J Med.
1993;329:977–986.
[55] White GH, Campbell CA, Horvath AR. Is this a critical,
panic, alarm, urgent, or markedly abnormal result?
Clin Chem. 2014;60:1569–1570.
[56] American Diabetes Association. Diagnosis and classifi-
cation of diabetes mellitus. Diabetes Care. 2014;37:
S81–S90.
[57] Geng J, Zhang Y, Wang B, et al. Glycosylated haemo-
globin levels and clinical outcomes in nondiabetic
patients with coronary artery disease. Medicine.
2017;96:e6784–e6786.
[58] Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglo-
bin, diabetes, and cardiovascular risk in nondiabetic
adults. N Engl J Med. 2010;362:800–811.
[59] International Association of Diabetes and Pregnancy
Study Groups Consensus Panel, Metzger BE, Gabbe
SG, Persson B. International association of diabetes
and pregnancy study groups recommendations on
the diagnosis and classification of hyperglycemia in
pregnancy. Diabetes Care. 2010;33:676–682.
[60] HAPO Study Cooperative Research Group.
Hyperglycemia and Adverse Pregnancy Outcome
(HAPO) Study: associations with neonatal anthropo-
metrics. Diabetes. 2009;58:453–459.
[61] Metzger BE, Gabbe SG, Persson B, et al. International
Association of Diabetes & Pregnancy Study Groups
(IADPSG) Consensus Panel Writing Group and the
Hyperglycemia & Adverse Pregnancy Outcome (HAPO)
Study Steering Committee. The diagnosis of gesta-
tional diabetes mellitus: new paradigms or status quo?
J Matern Fetal Neonatal Med. 2012;25:2564–2569.
[62] International Germ Cell Cancer Collaborative Group.
International Germ Cell Consensus Classification: A
prognostic factor-based staging system for metastatic
germ cell cancers. J Clin Oncol. 1997;15:594–603.
[63] von Eyben FE. Laboratory markers and germ cell
tumors. Crit Rev Clin Lab Sci. 2003;40:377–427.
[64] Campbell C, Caldwell G, Coates P, et al. Consensus
statement for the management and communication
of high risk laboratory results. Clin Biochem Rev.
2015;36:97–105.

[65] Sikaris KA, Martin A, Guerin MD. Relationship
between reference intervals and mortality-based ref-
erence ranges. Clin Biochem Rev. 1991;12:S81.
[66] Asadollahi K, Hastings IM, Beeching NJ, et al.
Laboratory risk factors for hospital mortality in
acutely admitted patients. QJM. 2007;100:501–507.
[67] Whelan B, Bennett K, O'Riordan D, et al. Serum
sodium as a risk factor for in-hospital mortality in
acute unselected general medical patients. QJM
2008;102:175–182.
[68] Mueller M, Biener M, Vafaie M, et al. Absolute and
relative kinetic changes of high-sensitivity cardiac
troponin T in acute coronary syndrome and in
patients with increased troponin in the absence of
acute coronary syndrome. Clin Chem. 2012;58:
209–218.
[69] Ricos C, Iglesias N, Garcıa-Lario JV, et al. Within-sub-
ject biological variation in disease: collated data and
clinical consequences. Ann Clin Biochem. 2007;44:
343–352.
[70] Stevens RJ, Oke J, Perera R. Statistical models for the
control phase of clinical monitoring. Stat Methods
Med Res. 2010;19:394–414.
[71] Balldin J, Berggren U, Berglund K, et al. Gamma-glu-
tamyltransferase in alcohol use disorders:
Modification of decision limits in relation to treat-
ment goals? Scand J Clin Lab Invest. 2010;70:71–74.
[72] Levine M, O’Connor AD, Padilla-Jones A, et al.
Comparison of prothrombin time and aspartate ami-
notransferase in predicting hepatotoxicity after acet-
aminophen overdose. J Med Toxicol. 2016;12:
100–106.
[73] Assy N, Beniashvili Z, Djibre A, et al. Lower baseline
ALT cut-off values and HBV DNA levels better differ-
entiate HBeAg- chronic hepatitis B patients from
inactive chronic carriers. World J Gastroenterol.
2009;15:3025–3031.
[74] Chang Y, Ryu S, Sung E, et al. Higher concentrations
of alanine aminotransferase within the reference
interval predict nonalcoholic fatty liver disease. Clin
Chem. 2007;53:686–692.
[75] Koerbin G, Cavanaugh JA, Potter JM, et al. 'Aussie
normals': an a priori study to develop clinical chemis-
try reference intervals in a healthy Australian popula-
tion. Pathology 2015;47:138–144.
[76] Plebani M. Errors in clinical laboratories or errors in
laboratory medicine? Clin Chem Lab Med.
2006;44:750–759.
[77] Arseneau E, Balion CM. Statistical methods used in
the calculation of geriatric reference intervals: a sys-
tematic review. Clin Chem Lab Med. 2016;54:
377–388.
[78] Alberti KGMM, Eckel RH, Grundy SM, et al.
Harmonizing the metabolic syndrome: a joint interim
statement of the International Diabetes Federation
Task Force on Epidemiology and Prevention; National
Heart Lunch and Blood Institute; American Heart
Association; World Heart Federation; International
Atherosclerosis Society; and International Association
for the Study of Obesity. Circulation. 2009;120:
1640–1645.
CRITICAL REVIEWS IN CLINICAL LABORATORY SCIENCES 11
[79] Marwaha RK, Tandon N, Garg MK, et al. Normative
data of body fat mass and its distribution as assessed
by DXA in Indian adult population. J Clin Densitom.
2014;17:136–142.
[80] Nordestgaard BG, Langsted A, Mora S, et al.
European Atherosclerosis Society (EAS) and the
European Federation of Clinical Chemistry and
Laboratory Medicine (EFLM) joint consensus initiative.
Fasting is not routinely required for determination of
a lipid profile: clinical and laboratory implications
including flagging at desirable concentration cut-
points-a joint consensus statement from the
European Atherosclerosis Society and European
Federation of Clinical Chemistry and Laboratory
Medicine. Eur Heart J. 2016;37:1944–1958.
[81] Daly N, Turner MJ. Laboratory diagnosis of gesta-
tional diabetes mellitus. BJOG. 2016;123:1430–1433.
[82] Devgun MS, Chan MK, El-Nujumi AM, et al. Clinical
decision limits for interpretation of direct bilirubin–a
CALIPER study of healthy multiethnic children and
case report reviews. Clin Biochem. 2015;48:93–96.
[83] Rodger RS, Laker MF, Fletcher K, et al. Factors influ-
encing normal reference intervals for creatinine, urea,
and electrolytes in plasma, as measured with a
Beckman Astra 8 Analyzer. Clin Chem. 1985;31:
292–295.
[84] Huang M, Yang JJ, Yang JJ, et al. Reference intervals
for serum creatinine levels in the healthy geriatric
population. Clin Biochem. 2013;46:1419–1422.
[85] Pottel H, Delanaye P, Weekers L, et al. Age-depend-
ent reference intervals for estimated and measured
glomerular filtration rate. Clin Kidney J. 2017;10:
545–551.
[86] Nowson CA, McGrath JJ, Ebeling PR, et al. A Position
Statement. Med J Aust. 2012;199:686–687.
[87] Aparicio-Ugarriza R, Palacios G, Alder M, et al. A
review of the cut-off points for the diagnosis of vita-
min B12 deficiency in the general population. Clin
Chem Lab Med. 2015;53:1149–1159.
[88] Daru J, Colman K, Stanworth SJ, et al. Serum ferritin
as an indicator of iron status: what do we need to
know? Am J Clin Nutr. 2017;106:1634S–1639S.
[89] Fraser CG, Kallner A, Kenny D, et al. Introduction: strat-
egies to set global quality specifications in laboratory
medicine. Scand J Clin Lab Invest. 1999;59:477–478.
[90] Sikaris KA. Application of the Stockholm hierarchy to
defining the quality of reference intervals and clinical
decision limits. Clin Biochem Rev. 2012;33:141–148.
[91] Sandberg S, Fraser CG, Horvath AR, et al. Defining ana-
lytical performance specifications: Consensus Statement
from the 1st Strategic Conference of the European
Federation of Clinical Chemistry and Laboratory
Medicine. Clin Chem Lab Med. 2015;53:833–835.
[92] Sikaris K. Performance criteria of the post-analytical
phase. Clin Chem Lab Med. 2015;53:949–958.
[93] Petersen PH, Jensen EA, Brandslund I. Analytical
performance, reference values and decision limits.
A need to differentiate between reference intervals
and decision limits and to define analytical quality
specifications. Clin Chem Lab Med. 2011;50:819–831.
[94] Petersen PH, Klee GG. Influence of analytical bias and
imprecision on the number of false positive results
 12 Y. OZARDA ET AL.
using guideline-driven medical decision limits. Clin
Chim Acta. 2014;430:1–8.
[95] Ozarda Y, Higgins V, Adeli Khosrow. Verification of
reference intervals in routine clinical laboratories:
practical challenges and recommendation. Clin Chem
Lab Med. 2018. DOI: 10.1515/cclm-2018-0059
[96] Bauer DT, Guerlain S, Brown PJ. The design and
evaluation of a graphical display for laboratory data.
J Am Med Inform Assoc. 2010;17:416–424.
[97] Shirts BH, Larsen N, Jackson BR. Utilization and utility
of clinical laboratory reports with graphical elements.
J Pathol Inform. 2012;3:26.
[98] Torsvik T, Lillebo B, Mikkelsen G. Presentation of clin-
ical laboratory results: an experimental comparison
of four visualization techniques. J Am Med Inform
Assoc. 2013;20:325–331.
[99] Nordestgaard BG, Langsted A, Mora S, et al.
Fasting is not routinely required for determination
of a lipid profile: clinical and Laboratory implica-
tions including flagging at desirable concentration
cutpoints-A joint consensus statement from the
European Atherosclerosis Society and European
Federation of Clinical Chemistry and Laboratory
Medicine. Clin Chem. 2016;62:930–946.
[100] Onat A. Risk factors and cardiovascular disease in
Turkey. Atherosclerosis. 2001;156:1–10.
[101] Bersot TP, Vega GL, Grundy SM, et al. Elevated hep-
atic lipase activity and low levels of high density
View publication stats
lipoprotein in a normotriglyceridemic, nonobese
Turkish population. J Lipid Res. 1999;40:432–438.
[102] Ozarda Y. Reference intervals: current status, recent
developments and future considerations. Biochem
Med (Zagreb). 2016;26:5–16.
[103] Catalona WJ, Smith DS, Ratliff TL, et al. Measurement of
prostate-specific antigen in serum as a screening test
for prostate cancer. N Engl J Med. 1991;324:1156–1161.
[104] Oesterling JE, Jacobsen SJ, Chute CG, et al. Serum
prostate-specific antigen in a community-based
population of healthy men. Establishment of age-
specific reference ranges. JAMA. 1993;270:860–864.
[105] Schroder FH, Hugosson J, Roobol MJ, et al. Prostate-
cancer mortality at 11 years of follow-up. N Engl J
Med. 2012;366:981–990.
[106] Andriole GL, Crawford ED, Grubb RL, et al. Prostate
cancer screening in the randomized prostate, lung,
colorectal, and ovarian cancer screening trial: mortal-
ity results after 13 years of follow-up. J Natl Cancer
Inst. 2012;104:125–132.
[107] Tsodikov A, Gulati R, Heijnsdijk EAM, et al.
Reconciling the effects of screening on prostate can-
cer mortality in the ERSPC and PLCO Trials. Ann
Intern Med. 2017;167:449–455.
[108] Vickers AJ, Cronin AM, Bjork T, et al. Prostate specific
antigen concentration at age 60 and death or metas-
tasis from prostate cancer: case-control study. BMJ.
2010;341:c4521.