• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services

Medical Devices

  • Print
  • Share
  • E-mail

Review Criteria for In Vitro Diagnostic Devices for the Assessment of Thyroid Autoantibodies using Indirect Immunofluorescence Assay (IFA), Indirect Hemagglutination Assay (IHA), Radioimmunoasay (RIA), and Enzyme Linked Immunosorbent Assay (ELISA)

This guidance was written prior to the February 27, 1997 implementation of FDA’s Good Guidance Practices, GGP’s. It does not create or confer rights for or on any person and does not operate to bind FDA or the public. An alternative approach may be used if such approach satisfies the requirements of the applicable statute, regulations, or both. This guidance will be updated in the next revision to include the standard elements of GGP’s.

REVIEW CRITERIA FOR IN VITRO DIAGNOSTIC DEVICES FOR THE ASSESSMENT OF THYROID
AUTOANTIBODIES USING INDIRECT IMMUNOFLUORESCENCE ASSAY (IFA), INDIRECT 
HEMAGGLUTINATION ASSAY (IHA), RADIOIMMUNOASSAY (RIA), AND ENZYME LINKED
IMMUNOSORBENT ASSAY (ELISA).

This is a flexible document representing the current concerns and suggestions
regarding thyroid autoantibodies in vitro diagnostic devices employing IFA,
IHA, RIA, and ELISA methodologies.  It is based on (1) current basic science,
(2) clinical experience, (3) the Safe Medical Devices Act of 1990 (SMDA) and
(4) FDA regulations in the Code of Federal Regulations (CFR).  As advances are
made in science and medicine, these review criteria will be re-evaluated and
revised as necessary.


PURPOSE OF THE GUIDANCE DRAFT

The purpose of this document is to provide guidance and clarification on
information to present to the Food and Drug Administration (FDA) before a
device to detect, quantitate and/or semi-quantitate thyroid autoantibodies in
clinical specimens can be cleared for marketing.

A premarket notification 510(k) submission provides evidence that the device
is accurate, safe, effective and substantially equivalent to a predicate
device legally marketed in the United States. 


DEFINITION

This generic type of device is intended for use in clinical laboratories or
physician's office laboratories* as an in vitro diagnostic test for the
qualitative, quantitative and/or semi-quantitative measurement of thyroid
autoantibodies by IFA, IHA, RIA or ELISA.

*Devices may be cleared for use in physician's office laboratories when
additional data are submitted to demonstrate performance in these settings.

PRODUCT CODES: JNL, DDC, DDJ, JZO

REGULATION NUMBER:

21 CFR 866.5870  Thyroid autoantibody immunological test system.

Identification.  A thyroid autoantibody test system is a device that consists
of the reagents used to measure, by immunochemical techniques, thyroid
autoantibodies (antibodies produced against the bodies own tissues). 
Measurement of thyroid autoantibodies may aid in the diagnosis of certain
thyroid disorders, such as Hashimoto's disease (chronic lymphocytic
thyroiditis), nontoxic goiter (enlargement of the thyroid gland), and Graves'
disease (enlargement of the thyroid gland with protrusion of the eyeballs).


CLASSIFICATION:  CLASS II (Performance Standards)

PANEL:  Immunology (82)

REVIEW REQUIRED:  Premarket notification (510(k))


I.   CLINICAL INDICATIONS/SIGNIFICANCE/INTENDED USE

A.   INTRODUCTION

The principle role of the thyroid gland is the storage and synthesis of
thyroid hormones.  The normal thyroid gland is composed of numerous follicles,
each of which consists of a single layer of epithelial cells surrounding a
central lumen containing colloid.  Thyroglobulin is the major component of the
thyroid follicular colloid.  It is produced by the thyroid epithelial cells
and is a water soluble glycoprotein with a molecular weight of 670,000
daltons.  Small amounts of thyroglobulin are present in serum.   Microsomal
antigen is a 110,000 dalton glycoprotein present in the cytoplasm and on the
apical membrane of the thyroid cell.  Reports have shown that thyroid
peroxidase (TPO) is the major component of the thyroid microsomal antigen and
is the antigenic component recognized by autoantibodies directed against the
microsomal glycoprotein. 1

Autoimmune thyroid disease is organ specific and is defined by the presence of
circulating antibodies.  The most common antibodies seen are antibodies to
thyroglobulin and microsomal or thyroid peroxidase (TPO) of the thyroid
epithelial cell lining. 

Autoantibodies to these thyroid antigens are detected in Hashimoto's disease
and its variants, Graves' disease, myxedema, nontoxic goiter, and thyroid
carcinoma.  Antithyroglobulin and antimicrosomal (TPO) antibodies are seen
most often and are in the highest titers in Hashimoto's disease.  Patients
with Graves' disease can also demonstrate thyroid antibodies of relatively
high titer.  These autoantibodies are also seen in low titers in other thyroid
diseases and in individuals without clinical evidence of disease. 2
  
Two other thyroid antibodies less commonly seen are antibodies to the second
antigen of the colloid or CA-2 and human thyroid-stimulating immunoglobulin
(TSI).  CA-2 is a minor protein of colloid distinct from thyroglobulin and
antibodies to it are of uncertain significance.  Positive CA-2 reactions are
seen in a low percentage of thyroiditis patients in the absence of other
antithyroid antibodies and in 5 to 10 per cent of patients with Graves'
disease and thyroid cancer. 2  CA-2 antibodies can be detected by IFA.  TSI,
previously referred to as long-acting thyroid stimulator (LATS), is a
polyclonal gamma globulin which appears to bind to a receptor on thyroid cells
and stimulates thyroid activity.  It is present in the sera of about 50 per
cent of those with Graves' disease and is absent or present only in a small
number of patients with nodular toxic goiter or other thyroid disorders.  TSI
can be measured by a competitive inhibition assay using 125I labeled TSH and
thyrotropin (TSH) receptors.
  
Hashimoto's thyroiditis is an inflammatory condition occurring in about 1 to 2
per cent of the population, mainly in middle-aged women, and is characterized
by gland enlargement as a result of marked lymphocytic inflammatory changes. 
The latter may consist of lymphoid follicles with active germinal centers in
which much of the antithyroglobulin antibody appears to be synthesized. 
Normal thyroid glandular structures are adversely altered, and in prominent
cases progressive disease may lead to thyroid atrophy and myxedema (dry, waxy
type of swelling, with abnormal deposits of mucin in the skin and other
tissues associated with hypothyroidism).  In thyrotoxicosis, the thyroid may
contain small areas of lymphoid infiltration as well as evidence of the
typical glandular hyperactivity.  Graves' disease is a multi-systemic
disorder, particularly in young to middle-aged females, consisting of varying
degrees of (1) hyperthyroidism with diffuse hyperplasia of the thyroid (the
most common pattern seen with diffuse toxic goiter); (2) a myopathy; and (3)
an infiltrative ophthalmopathy, frequently leading to exophthalmos (protrusion
of the eye ball). 2 

As with all the organ-oriented diseases associated with autoantibody
production, it is important to determine when the antibodies under discussion
are pathogenic or are reacting to antigens liberated as a result of tissue
damage due to non-immune causes.  Immune reactivity may not be the primary
pathogenic event but, once present, causes further tissue damage.  Evidence
against a primary pathogenic role for thyroid autoantibodies in Hashimoto's
thyroiditis and Graves' disease is (1) the lack of correlation between the
level of autoantibody and the severity of disease in individual cases, and (2)
the lack of development of thyroid disease in infants with high levels of
antithyroid antibodies because of placental transfer. 


B.   INSTRUCTIONS:

Provide a concise discussion to include the following as appropriate.  Support
the discussion with key literature citations.

    1.  Clinical indications, significance and intended use.

    2.  Background description of the thyroid disease involved including the
        type of population affected (sex, age, etc.)

    3.  Significance of a positive result (disease indication and follow up
        testing).

    4.  Significance of false positive and false negative results.

    5.  Salient concerns of the medical community including relevant medical
        issues that may impact the review process or possibly the development
        of public policy.

    6.  A brief historical summary of all test methodologies used to detect
        the antibody(ies).

    7.  Merits/advantages and limitations/disadvantages of the device
        methodology(ies) compared to other available methodologies.

    8.  Matrices.


II. DEVICE DESCRIPTION:

The determination of substantial equivalence is based on the specific intended
use (what analyte is detected and the indications for use) and the
technology/methodology utilized in the device.  Discuss the principles of the
device methodology and whether it is well-established or new and unproven. 


III. CLINICAL AND NONCLINICAL LABORATORY STUDIES: SPECIFIC PERFORMANCE
     CHARACTERISTICS:

FDA requests different types and amounts of data and statistical analyses in
pre-market notification applications to market in vitro diagnostic devices. 
The amount and type of data requested depends on the intended use,
technological characteristics of the new device, whether the test is
qualitative, quantitative or semi-quantitative and on certain claims made by
the manufacturer.  The performance of the device can be established by
comparison to any legally marketed medical device (the predicate) with the
same intended use.

Prove all claims for substantial equivalence and specific performance
characteristics for using the device.  Clearly document all protocols for in
vitro testing.  Present test data results with analyses and conclusions. 
Summarize results and include explanations for unexpected results and any
additional testing performed.  Charts (scattergrams, histograms, etc.) may be
used as part of the analyses and conclusions when appropriate.  Actual,
unprocessed laboratory data may be requested.

A.  ANALYTICAL/LABORATORY/IN VITRO STUDIES

   1.   Validation of the Cut-off

    Describe the rationale for determination of the assay cut-off(s). 
    Furnish descriptive information and laboratory data to show how the
    cut-off point (distinction between positivity and negativity or
    medical decision limit) was determined by the assay.

    a.  Define the population(s) used, including the following
        information:

        i.    Number of samples in the normal population (used to
              determine initial screening dilution) with samples
              summarized according to gender and age groups. 4

        ii.   Number of specimens included in each disease group
              summarized according to gender and age groups.

        iii.  Geographical area(s) from which the population was derived.

        iv.   Graphical (e.g., scattergrams, histograms, etc.)
              representation of population characteristics.

    b.  Define the statistical method used to determine the cut-off
        point(s).

    c.  Present a Receiver Operator Curve (ROC) analysis of cut-off point
        selection and other graphical representations as appropriate.

    d.  Define the basis for the equivocal zone (if applicable).

    2.  Reagent characterization

        a.    Give a brief description of the antigen(s) and antibody(ies) 
	      used in the assay.

        b.    If any recombinant technology was used in the preparation of the
              antigen(s), describe method used. 

    3.  Assay Specificity/Interfering Substances

        Any potentially cross reacting or interfering substances encountered
        in specific specimen types or conditions should be tested using the
        assay system, e.g., hemolysis, lipemia, microbial contamination,
        additional analytes or other autoantibodies present, and storage or
        freeze-thawing.

        a.    Verify that recommended storage conditions are compatible with the
              assay.  State the optimal conditions based on specimen storage
              stability studies.  Both false positivity and negativity should be
              evaluated (if applicable).

        b.    If the use of plasma is claimed, a study with each anticoagulant
              must be performed to show that each anticoagulant does not
              interfere with the assay.

         i.   For each anticoagulant, test 10 matched serum and plasma
              specimens which are positive at the cut-off point.

         ii.  For each anticoagulant, test 10 matched negative serum and
              plasma specimens.

         Provide an explanation if interference from the anticoagulant is
         not anticipated (e.g., high dilution factor).

    4.  Performance Characteristics

        Include the following performance characteristics:

        a.    Analytical Sensitivity (if applicable)

         The analytical sensitivity or detection limit is defined as the
         lowest quantity differentiated from zero (95% confidence intervals
         or 2 standard deviations (SD) above the mean of the Zero control
         are commonly used). 4,5  Run the Zero standard (Zero diluent) at
         least 20-25 times in the same run and calculate the mean of the
         Zero standard and 2 SD of the mean (counts, OD's, etc.).  If
         levels of the analyte are not clinically significant,
         determination of the detection limit may be irrelevant. 

        b.    Relative Sensitivity and Specificity

         The relative sensitivity and specificity as determined by
         comparison to a legally marketed device or to a reference method
         should be determined and reported in the Performance
         Characteristics section of the package insert.

        c.    Linear Range

         Validate the linear range of the assay with normal and abnormal
         specimens covering the entire reportable range of the assay. 6

        d.    Reproducibility and Repeatability Studies 4,5,6,7,8

         The National Committee for Clinical Laboratory Standards (NCCLS)
         recommends an analysis of variance experiment testing two
         clinically significant levels near medical decision limits (normal
         or elevated) of an analyte, in this case thyroid autoantibodies. 8 
         Use controls simulating patient samples or actual patient
         specimens 3 times in the same run and in two different runs each
         day for 20 days.  This permits separate estimation of between-day,
         between-run and within-day standard deviations (SDs), as well as
         within-run and total SDs.

         i.   Qualitative/Quantitative Tests:

              Calculate total, between- and within-day and between- and
              within-run means and coefficients of variation of
              imprecision for each set of values.

         ii.  Semi-quantitative Tests:

              In devices with a titration format, e.g., immunofluorescence
              assays, demonstrate that intra-run reproducibility is within
              the commonly accepted limits of plus or minus one two-fold
              dilution.

         iii. Means, SDs, and Coefficients of Variation:

              Report in the Performance Characteristics section of the
              package insert the appropriate means, SDs, and/or
              coefficients of variation with confidence levels according
              to number of times the sample is repeated.  Report the
              number of runs per day.

        e.    Prozone or High-Dose Hook Effect Studies

         Test a sample with the highest titer available, serially diluted
         and undiluted.  If prozone problems are encountered, state in the
         Performance Characteristics section of the package insert the
         titer at which prozone problems were detected and a procedure for
         the user to follow to correct the problem.  Where appropriate,
         describe the appearance of a prozone reaction for subjective
         tests.

        f.    Alternative Testing Sites

         Include reproducibility studies performed in these settings.  In-site 
	 testing for new technologies should include at least three
         independent alternative testing sites.  At each of the three
         sites, the precision and accuracy of the device should be
         evaluated.  A statistically valid number of samples should be
         tested by the site personnel and by professional laboratory
         personnel, and the results compared, to show how the device
         performs in the hands of the lesser trained user.

    5.  Comparison Studies

        Compare the new device to a legally marketed device.  Include the
        package insert for the legally marketed device.

        It is recommended that a recognized reference method (if available)
        also be employed for comparison to enable a fair evaluation of the
        proposed device's performance characteristics, particularly if there
        are broad differences in methodology/ technology between the new
        device and the legally marketed device.

        a.    Qualitative Tests:

         The studies should be performed on an adequate number of positive
         and negative specimens to support statistical significance.  (An
         appropriate number may be suggested by a statistician.)

        b.    Quantitative/Semi-quantitative Tests:

         An assay is considered quantitative only if a recognized reference
         material of known concentration is available for standardization
         of a calibrator or standard used in the assay to calculate
         results.

         If the same reference material is used in the new device as in the
         predicate device to substantiate the quantitative claim,
         comparison data should be presented to show correlation between
         the two assays when running the reference material as samples. 
         Run the serially diluted reference material by the new device and
         the predicate device.  The assays should show similar results.  

         Compare results obtained using positive thyroid autoantibody
         samples free from interfering substances from 40-100 persons
         covering the whole assay range (from low to high levels of
         antibodies). 7,9

         Perform a linear regression analysis and report the slope,
         intercept, correlation coefficient, the assay range, and the
         nature and number of samples tested.

        c.    Comparison Discrepancies:

         Equivocal results or discrepancies between the new device and the
         comparison method should be resolved using another method or
         clinical diagnosis.

    6.  Specimen Collection and Handling Conditions

        State specimen collection, storage and handling conditions in the
        package insert and provide data or appropriate literature references
        in the submission to substantiate claims.

     7. Computer Controlled Medical Devices

        For information regarding computer assisted clinical laboratory
        devices, refer to "Review Guidance for Computer Controlled Medical
        Devices Undergoing 510(k) Review" available from the Division of
        Small Manufacturers Assistance (DSMA), 1-800-638-2041.


B.  CLINICAL INVESTIGATIONS

    In certain instances it is necessary to require comparative clinical data
    to establish substantial equivalence, e.g., a new or unfamiliar methodology
    or technological feature is introduced in a device category in which
    clinical performance is claimed to be equivalent to a legally marketed
    device using "conventional" technology.

    For 510(k) submissions, perform a comparison of the device to a legally
    marketed device.  Ideally this study should be done at independent clinical
    laboratory site(s).  A minimum of two additional independent investigators
    at separate outside locations is recommended.  The investigators should be
    identified by institutional name and address.

    1.  Adequate Clinical Investigations

        a.    Prove all claims for substantial equivalence and specific
              parameters for using the device.

        b.    Describe all protocols for clinical studies and consistently
              adhere to the protocols.

        c.    Determine the sample size, prior to beginning the study, that will
              be statistically sufficient to determine whether or not the device
              is safe and effective.

        d.    Sampling Method:

         Describe sampling method used in the selection and exclusion of
         patients.

         i.   Patient selection:

              Include samples from individuals with diseases or conditions
              that may cause false positive or false negative results with
              the device.  Ideally, a prospective study is preferred. 
              However, if a retrospective study is used, include all
              eligible patients who meet the patient selection criteria as
              specified in the protocol.

         ii.  Account for all patients and samples.  Insure that data
              points are included for every sample for every patient.


    2.  Establishing Reference Ranges

        a.    Normal individuals:

         Establish a normal reference range with a statistically sufficient
         number of samples from normal persons characterized by age, sex,
         geographical location and any other factors that would influence
         the values obtained. 3,10

        b.    Patient groups:

         i.   Confirm that the new device detects the percentage of
              positives generally expected for each disease for which the
              device is intended.  Use a statistically sufficient number
              of patients characterized by age, sex, geographical
              location, any symptoms of disease, clinical presentation,
              and any other factors that would influence the values
              obtained. 3,10

         ii. False results:

              Patients positive for antinuclear antibodies (ANA) could
              give a false positive result in an IFA assay for the
              detection of thyroid autoantibodies.

              Radioisotopes administered to the patient for diagnostic or
              therapeutic purposes may interfere in some RIA assays.

              Provide reports, if any, of false positive and false
              negative results for each disease as appropriate.

        c.    Sample Types Claimed:

         Investigate all sample type(s) claimed in the intended use
         statement unless other data proves that there is no difference
         between them.

IV. LABELING CONSIDERATIONS

The following are additional details for some of the points in the statute
[502(f)(1)] and regulations [21 CFR 809.10(b)].


Package Insert

Include the package insert for the new thyroid autoantibody device.  Support
the statements throughout the document with key literature citations.

A.  INTENDED USE

    Briefly describe the intended use based on the technology/methodology used
    in the device.  Include the following information:

    1.  Whether the assay is qualitative, quantitative, or semi-quantitative. 
        In order to claim to provide a quantitative result, the calibrators
        or standards must be calibrated to a reference material of known,
        established value.

    2.  Test methodology.

    3.  Specimen type(s).

    4.  Indicate if the device is for use in clinical laboratories and/or
        alternative care sites.  The Limitations section should include any
        specific training required for test performance.

    A typical Intended Use statement would be:

    "ABC's *** test system is a device for the semi-quantitative measurement of
    anti-thyroglobulin antibodies by indirect immunofluorescence in human serum
    to aid in the diagnosis of certain thyroid disorders, such as Hashimoto's
    disease, nontoxic goiter, and Graves' disease."

B.  QUALITY CONTROL (QC)

    The package insert should recommend levels of quality control samples and
    their number, matrix type, placement and interpretation to ensure that the
    system meets its performance claims.  Include a statement that if controls
    do not behave as expected, assay results are considered invalid and should
    be repeated.

    Controls should be handled in the same manner as patient samples.  For
    example, if the patient sample is diluted or titered the control material
    should also be diluted or titered using the same diluent.

C.  RESULTS

    Give an adequate description of expected results and interpretation.
  
    1.  IFA

        a.    Give a description of the fluorescence for a positive and negative
              result.

        b.    List possible staining patterns which may be found.  Photographs
              or diagrams may be helpful.

        c.    Give pattern description and interpretation as it relates to the
              particular antibodies.

        d.    Give instructions for titering positive samples to end point.

    2.  IHA

        a.    Describe in detail the appearance of a positive and negative
              result.

        b.    Give a definition and description of the end point for a positive
              result.

        c.    Describe the appearance of a prozone reaction (antibody excess)
              and instruct the user what to do if prozoning is suspected.

        d.    Explain the procedure for repeating samples which have results
              above the linearity of the assay.

    3.  RIA

        a.    Explain the procedure for manual (if applicable) calculation of
              percent bound for each sample and include a sample calculation.

        b.    Give instructions for plotting percent bound versus concentration
              of the standard showing examples of typical results (numerical and
              graphical).

        c.    Give a brief explanation of how automated calculations are
              performed, e.g., the type of data reduction program used.
 
    4.  ELISA

        a.    Explain the procedure for calculating the value of the unknown
              including a sample calculation.

        b.    Explain the procedure for repeating samples which are above the
              linearity of the assay.  Give instructions for dilution of samples
              including the dilution factor and type of diluent to be used.

D.  LIMITATIONS OF THE PROCEDURE

    Include a statement of limitations of the procedure to include the
    following:

    1.  A statement that the test result in and of itself is not diagnostic
        for thyroid disease and should be considered in conjunction with
        iodine uptake and other standard thyroid tests and the clinical
        presentation of the patient.

    2.  IFA

         a.   Explain possible variations between different types of
              fluorescent microscopes.

         b.   Give warnings concerning distinguishing the thyroid specific
              cytoplasmic fluorescence from that obtained with
              mitochondrial antibody in primary biliary cirrhosis.  If
              mitochondrial antibody is suspected, the distinction can be
              made by running more specific tests for antinuclear
              antibodies.

         c.   Multiple antibodies may be present and complicate the
              staining interpretation.  Serially diluting the patient
              sample will often aid in distinguishing multiple patterns.

         d.   A prozone reaction can appear as a doubtful positive or
              negative because of a small amount of antigen in relation to
              the large amount of antibody present.  If prozone is
              suspected, the patient sample should be serially diluted.

    3.  IHA

        Give warnings concerning heterophile antibodies and a possible
        prozone reaction 

    4.  RIA

         a.   Give warnings concerning possible interference from             
              radioisotopes administered to the patient for diagnostic or
              therapeutic purposes.

         b.   Supply instructions for proper disposal of radioactive
              materials.

E.  EXPECTED VALUES

    1.  The expected value in the normal population is negative.  However,
        apparently healthy, asymptomatic individuals (5-10%) may test
        positive for thyroid autoantibodies.  The incidence of these
        antibodies increases with increasing age beginning in the seventh
        decade for women and the eighth decade for men.

    2.  Thyroid autoantibodies may be present in non-thyroid disorders such
        as pernicious anemia, diabetes mellitus, Addison's disease, and
        Sjogren's syndrome.

    3.  Present information showing the incidence or prevalence of each type
        of thyroid autoantibody for each disease state.


From:         Division of Clinical Laboratory Devices
              Immunology Branch

Prepared by:  Deborah M. Moore, Scientific Reviewer
              February, 1994


V.  BIBLIOGRAPHY

    1.  Czarnocka B,et al. Purification of the Human Thyroid Peroxidase and
        its Identification as the Microsomal Antigen involved in Autoimmune
        Thyroid Disease. FEBS 190:147 (1985).

    2.  Henry J, Immunology and Immunopathology.  Clinical Diagnosis and
        Management by Laboratory Methods, 18th ed 1985; WB Saunders Co.,
        Philadelphia, PA.

    3.  National Committee for Clinical Laboratory Standards.  How to define,
        determine, and utilize reference intervals in the clinical
        laboratory; proposed guideline. Villanova, PA 1991. Order code C28-P.

    4.  Vadlamudi SK, Stewart WD, Fugate KJ, Tsakeris TM. Performance
        characteristics for an immunoassay.  Scand J Clin Lab Invest
        1991;51:134-138.

    5.  Peters T, Westgard JO.  Evaluation of methods, Chapter 7 in: Tiets
        NW, editor.  Fundamentals of Clinical Chemistry, 3rd ed, 1987: 225-37
        Philadelphia, PA; WB Saunders Co.

    6.  National Committee for Clinical Laboratory Standards.  Evaluation of
        the linearity of quantitative methods; proposed guideline.  1986 
        Order code EP6-P.

    7.  Information for authors.  Clin Chem 1991; 37:1-3.

    8.  National Committee for Clinical Laboratory Standards.  Evaluation of
        precision performance of clinical chemistry devices - 2nd ed;
        tentative guideline. 1991:1-56.  Order code EP5-T2.

    9.  National Committee for Clinical Laboratory Standards.  User
        comparison of quantitative clinical laboratory methods using patient
        samples; proposed guideline.  1985; 6(1). Order code EP9-P.

    10. Ash KO.  Reference Intervals (Normal Ranges): A Challenge to
        Laboratories.  Am J. Med Tech 1980; 46:504-11.