This document discusses reference intervals, which provide the range of normal test results for clinical decision making. It outlines the process for establishing and verifying reference intervals, including selecting a healthy reference population and statistically analyzing the results. Both nonparametric and parametric statistical methods can be used to determine the reference interval that spans the central 95% of reference values. Clinical laboratories often rely on published reference intervals from assay manufacturers or literature to avoid the cost and effort of establishing their own.

1. Dr. S. Rathore
MD (Sco)
Dept. Of Biochemistry
Pt. J. N. M. Medical College,
Raipur

2.  Reference interval:
 A pair of medical decision points that span the limits
of results expected for a given condition
 Reference interval are required by professional
accreditation and regulatory standards
 In most cases, a “normal range” is used as the
test’s “reference Interval”
 For some analytes, the reference interval is
defined as “less than” or “greater than” a
certain value
Example: total cholesterol: <200mg is desirable

3.  To make medical diagnoses
 Assess physiologic function
 Manage therapy
 When interpreting laboratory data,
clinicians compare the measured test result
from a patient with a reference interval

4.  Reference intervals are sometimes referred
to as reference ranges; the preferred term is
reference interval because range implies the
absolute maximum and minimum values

5.  Establishing a reference interval:
 A new reference interval is established when
there is no existing analyte or methodology in
the clinical or reference laboratory with which to
conduct comparative studies
 It is a costly and labor-intensive study that will
involve laboratory resources at all levels and may
require from 120 to as many as ≈700 study
individuals

6.  Verifying a reference interval (transference):
 This is done to confirm the validity of an existing
reference interval for an analyte using the same
(identical) type of analytic system (method
and/or instrument).
 These are the most common reference interval
studies performed in the clinical laboratory and
can require as few as 20 study individuals.

7.  Medical decision level:
 Value for an analyte that represents the
boundary between different therapeutic
approaches
 Normal range:
 Range of results between two medical decision
points that correspond to the central 95% of
results from a healthy patient population
 Of the results, 2.5% will be above the upper limit
and 2.5% will be below the lower limit of the
normal range

8.  Therapeutic range:
 Reference interval applied to a therapeutic drug
 Reference intervals are needed for all tests in
the clinical laboratory
 The provision of reliable reference intervals is an
important task for clinical laboratories and test
manufacturers
 The dynamic review of existing reference
intervals by the health care team (scientific
community, manufacturers, and clinical
laboratory) is crucial to meeting the challenges
of providing optimal laboratory data for patient
care.

9.  Reference interval use can be grouped into
three main categories:
 Diagnosis of a disease or condition
 Monitoring of a physiologic condition
 And therapeutic management
 Examples:

13.  Has the laboratory established or verified its
reference intervals (normal values)?
 Reference intervals are important to allow a clinician
to assess patient results against an appropriate
population
 The reference range must be established or verified
for each analyte and specimen source (e.g., blood,
urine, cerebrospinal fluid [CSF]), when appropriate
 For many analytes (e.g., therapeutic drugs and CSF
total protein), literature references or a
manufacturer’s package insert information may be
appropriate

14.  Does the laboratory evaluate the
appropriateness of its reference intervals,
and take corrective action if necessary?
 Criteria for evaluation of reference intervals
include:
 Introduction of a new analyte to the test repertoire
 Change in analytic methodology
 Change in patient population
If it is determined that the range is no longer
appropriate for the patient population, corrective
action must be taken

15. 1
• Define an appropriate list of biological
variations and analytic interferences
from medical literature
2
• Choose selection and partition (e.g.,
age or gender) criteria
3
• Complete a written consent form and
questionnaire to capture selection
criteria

16. 4
• Categorize the potential reference
individuals based on the questionnaire
findings
5
• Exclude individuals from the reference
sample group based on exclusion criteria
6
• Define the number of reference
individuals in consideration of desired
confidence limits and statistical accuracy

17. 7
• Standardize collection and analysis of reference
specimens for the measurement of a given analyte
consistent with the routine practice of patients
8
• Inspect the reference value data and prepare a
histogram to evaluate the distribution of data
9
• Identify possible data errors and/or outliers and
then analyze the reference values

18.  A set of selection criteria determines which
individuals should be included in the group of
reference individuals
 Such selection criteria include
 Statement describing the source population
 Specifications of criteria for health
 The disease of interest

19.  Often, separate reference values for each sex and for
different age groups, as well as other criteria, are
necessary
Our group of reference individuals therefore may have
to be divided into more homogeneous subgroups. For
this purpose, specific rules for the division, called
stratification of partitioning criteria, are needed
It is important to distinguish between selection and
partitioning criteria. First, selection criteria are
applied to obtain a group of reference individuals
Thereafter, this group is divided into subgroups
using partitioning criteria. Whether a specific
criterion (e.g., gender) is a selection or a partitioning
criterion depends on the purpose ofthe actual project.

23.  Nonparametric method:
 Statistical test that makes no specific assumption
about the distribution of data
 Nonparametric methods rank the reference data
in order of increasing size
 Because the majority of analytes are not
normally (gaussian) distributed , nonparametric
tests are the recommended analysis for most
reference range intervals

24.  Parametric method:
 Statistical test that assumes the observed
values, or some mathematical transformation of
those values, follows a (normal) gaussian
distribution

26.  Confidence interval:
 Range of values that include a specified
probability, usually 90% or 95%
 For example, consider a 95% confidence interval
for slope = 0.972–0.988 from a method
comparison experiment. If this same experiment
were conducted 100 times, then slope would fall
between 0.972 and 0.988 in 95 of the 100 times
 Confidence intervals serve to convey the
variability of estimates and quantify the
variability.

27.  Bias:
 Difference between the observed mean and the
reference mean
 Negative bias indicates that the test values tend
to be lower than the reference value
 Whereas positive bias indicates test values are
generally higher. Bias is a type of constant
systematic error

28.  With the development of statistical software
packages such as
 EP Evaluator
 MedCalc, GraphPad Prism
 Minitab
 JMP
 SAS/STAT
Now reference intervals are rarely determined
manually

29.  When possible, clinical laboratories rely on assay
manufacturers or on published primary literature to
determine reference intervals
 This avoids the expensive and lengthy process of
establishing a reference range interval on a minimum of
120 healthy people
 The CLSI allows less vigorous studies to verify a reference
interval with as few as 20 subject specimens
 Method verification studies can be used if the test method
and study subjects are similar to the vendor’s reference
data and package insert information
 The main assumption in using transference studies is that
the reference method is of high quality and the subject
populations are similar
 The manufacturer’s reported 95% reference limits may be
considered valid if no more than 10% of the tested subjects
fall outside the original reported limit

32.  Parameters are broadly defined as diagnostic
efficiency, which can be broken down into sensitivity,
specificity, and predictive values
 Diagnostic sensitivity:
 Ability of a test to detect a given disease or condition.
 Diagnostic specificity:
 Ability of a test to correctly identify the absence of a
given disease or condition.
 Positive predictive value:
 Chance of an individual having a given disease or
condition if the test is abnormal.
 Negative predictive value:
 Chance an individual does not have a given disease or
condition if the test is within the reference interval

35. References
Tietz Textbook of Clinical Chemistry and Molecular Diagnostics
Clinical Chemistry: Principles, Techniques, and Correlations
- Bishop, Michael L, MS.