Control Serum Preparations

James E. Logan and Rovelle H. Allen

A total of 17 different commercial control serums prepared by 6 manufacturers were

examined to determine the accuracy of labeled values for as many as 9 constituents.
All serums were analyzed by generally acceptable reference procedures. Several lots
of each product were checked at approximately 6-month intervals up to 2 years or
more to assesstheir stability on storage. It is concluded that commercial control
serumsshould not be used in place of a primary standard when the constituents to be
assayedcan be obtained in a relatively pure form and a stable solution can be prepared.

TE”HE NUMEROUS INTMlLABORATO1tY suuvis which have been conducted

since the early report of Belk and Sunderman (1) have confirmed the
distressingly wide variatioii of results that are often ol)tained among
laboratories and have emphasized the continued iieed for quality-
control practice. The method of quality control generally adopted has
been the inclusion of an aliquot of a reference serum pooi in the series
of analyses to determine whether the assay value obtained falls within
the control limits for the procedure. These may have been set at ± 2 oi’
3 standard deviations from the mean value, which was established by
repetitive assays when the serum was prepared or’ obtained.
Two types of control serum preparations* are used by laboratories:
(1) those serums collected as a pool remaining after’ specimen analyses,
divided into suitably sized ahiquots, and frozen until the day of use, and
(2) commercially available serums. During the past 10 years many
laboratories have found it more convenient and timesaving to use the
commercial serums; consequently, the reliability of these products has
assumed major importance. It is with this aspect of the problem that
our laboratory has been primarily concerned.
A number of manufacturers have placed control serum preparations
on tire market during the past few years. These have all, with 1 excep-
tion (Chemtrol-Clinton Laboratories-a pooled bovine serum), been

From the Clinical Laboratories, Laboratory of Hygiene, Department of National Health

and Welfare, Ottawa 3, Ont., Canada.
Received for pubiication Aug. 23, 1967; accepted for publication Oct. 2, 1967.
*TIIe tes-ni “reference sample” adopted by tue Standards Committee of the AACC and Board
of Editors of CLINICAL CHEMISTRY would cover these commercial preparations.

437
438 LOGAN & ALLEN Clinical Chemistry

prepared from human serum or plasma.* They are mostly distributed

in a freeze-dried form requiring reconstitution to a given volume. Dade
Reagents Inc., in addition to lyophilized serums, distribute serums in
the liquid state. Assay values are supplied for a number of constituents
in most of the preparations. On tire other hand, some of the manufac-
turers also sell unassayed serums which lowers the cost of the product.
For the assayed serums, a range of allowable variation or a range of
acceptable values is usually given for each constituent, with the
expectation that values obtained iii tire individual laboratories should
fall within this range.
In our examination of commercial control serums, we have analyzed
them for a number of constituents and compared results with the values
assigned by the manufacturer. We have also checked the effects of
storage in the refrigerator for periods up to 2 years or more on the
stability of most of these constituents. The present report summarizes
our results.
Materials and Methods
Analytic Procedures
All commercial control serums were dated on receipt from the sup-
plier and stored in the refrigerator at 2-5#{176}
until the day of analysis.
The lyophilized preparations were reconstituted with distilled water to
the volume specified by the mannfacturer and allowed to stand at least
#{189}
hr. before use. All constituents were analyzed within 5 days following
reconstitution with the exception of glucose, which analysis was carried
out within 2 hr. A number of constituents were determined using the
following procedures for analysis:
Urea nitrogen-automated diacetyl monoxime (2)
Glucose-automated ferricyanide (8)
Total protein-automated biuret reaction (4)
Cholesterol-method of Abell et al. (5)
Chloride-automated mercuric tuiiocyanate (6)
Sodium-automated flame photometry (7)
Potassium-automated flame photometry (7)
Calcium-automated cresolphtlialein complexone (8)
Uric acid-automated method of Crowley (9)

Commercial Serums Examined

The following products of the companies listed below were examined
and several lots of each were included:
Warner-C hilcott Versatol, Versatol-A, Versatol-A Alternate,
Versatol Pediatric, and Serachol
has been determined that some of the manufacturers now use defibrinated plasma as
their source material.
Vol. 14, No. 5, 1968 CONTROL SERUM PREPARATIONS 439

Iiylan(i 11lylaiid Normal, Ilylalid Abnormal, ilyland Supple-

mental, and Ilyland Elevated Cholesterol
Dade Lab-trol, Patho-trol, Choles-trol, Moni-trol I, and Moni-
trol II
Clinton Chem-trol
Michael Reese Metrix
Burroughs Weilcome “Wellcome” Chemical Control Serum

Stability
The stability of the serums was checked by assaying several lots of
each product at approximately (3-month intervals. Values were 01)-
tamed for the following constituents: urea nitrogen, glucose, total
protein, cholesterol,calcium, and chloride.
Results
Precision of Analytic Procedures
The day-to-day precision of the methods employed in our laboratory
for the determination of each of the 9 serum constituents is shown iii
Table 1. It is expressed in each case as percentage confidence limits
which are calculated as ± 3 standard deviations from the mean, divided
by the mean, and multiplied by 100. The values have been established by

Table 1. PRECISION OF METHODS FOR 9 SERUM CONSTITUENTS

Mean ± % confidence limits

Constituent Method Low or normal Intermediate High

Glucose (mg./l00 ml.) Automated 84.4 ± 4.4 206.2 ± 10.8 296.6 ± 5.2
ferricyanide
Urea nitrogen (mg./I00 nIl.) Automated 13.0 ± 14.0 25.9 ± 6.1 47.7 ± 8.4
diacetyl
monoxi me
Total protein (gm./lOO ml.) Automated .‘i.8 ± 2.6 7.1 ± 3.8 7.6 ± 3.95
biuret
Total cholesterol (mg/100 ml.) Manual, Abell 83.6 ± 9.0 189.0 ± 3.2 328.0 ± 1.8
ci al.
Chloride (mEq./L.) Automated 50.3 ± 4.9 97. 1 ± 4. S 130.6 ± 7.6
Inercummic
lIio(vatmate
Sodium (mEq./L.) Automated 119.1 ± 7.8 141.5 ± 6.8 157.5 ± 2.5
flame
Potassium (mEq./L.) Automated 4.46 ± 12.1 5.13 +4.8 5.76 ± 10.8

Calcium (mg./lOO ml.)

flame
Manual,
Clark-Collip
- 10.0 ± 12.6

Uric acid (nig./lOO ml.) Automated, 4.83 ± 8.7 - 11.3 ± 6.1

Crowley
 a

CaCa - 1’- - -
C’)

0
a
a

a CI a
-
a
a x
I -
Ca CC-

4-

a
‘ m-

a
a
w I I I I I’
a4-

cI
4a a
a
CI ‘
z a
0 C)
‘-4 CC- C)
0 a
-- Ca

0
C
z z 41
1.1
0
0 1.1 I-
a a a
0
0 C)
Ca C-I - a Ca a
z
0
z -
z
I I I I
a
C) a
z a -u
0 a
a
z C)
Ca
C)
Ca - ‘ Ca - a CaCa
I., I I
a
4-

c-I a
-a .a
4)
.a - a a

p
. I- 01 Ca C’l C-I CC-Ca
Cl - Ca Ca Ca CI - a
C)
C)
a
a
a

a
a
.0
a a
‘a
ha
-H CaH.C’E ‘a E

1:
N#{176}

t - -H -H 2 -HH -112
a

I m. a’i
0
4..
a -u i a
2 C-I C-I

-
a
a
0 a
4-
aj 2 bf
a - ‘
a
:
Vol. 14, No. 5, 196$ CONTROL SERUM PREPARATIONS 441

analysis of the same lot number of commercial serum over a 10-day

period according to the method of Bai-nett (10). In each case, 2 bottles
of the lot number w’ere used, with the contents of the first analyzed for
3 consecutive days and the contents of the second then reconstituted
for use in the second ;-dav period. Values have been obtained using
serums coiitaining normal, intermediate, and high concentrations of
the 9 constituents. it will be noted that the precision varies markedly
in some cases with tile coilcelitration of the constituent. For example,
with urea nitrogen we obtained confidence limits of 14.0% for the
normal range and 6.1% for the intermediate range. The poorer sensi-
tivity of the automated diacetvl nlOlloXiIne method in the normal range
may then account for the high proportions of disagreements we ob-
served in the urea nitrogen results with the Michael Reese serums
(Table 2), which are prepared to contain only normal levels of urea.

Accuracy of Assigned Values

The accuracy of the manufacturers’ assigned values for 9 constitu-
ents of commercial serums when compared with the values obtained in
our laboratory is shown in Tables 2 and 3. No attempt has been made
to list separately tile different products of each manufacturer. In the
case of urea nitrogen, the Dade Reagents products examined were
Lab-trol, Patho-trol, Moni-trol I and Moni-trol II; Hyland Laboratories
products were Hyland Normal and Hyland Abnormal Clinical Chem-
istry Control Serum; Warner-Chilcott products were Versatol, Versa-
tol-A, Versatol-A Alternate, and Versatol Pediatric; Clinton Labora-
tories was Chemtrol; Michael Reese Research Foundation was Metrix
Serum Standard and Control; and Burroughs Welicome & Co. was
“Welicome” Chemical Control Serum.
The methods by which the various manufacturers express their limits
for acceptable ranges differ. Some are shown as a range between 2 con-
centrations, some as ± concentration limits and one as a ± percentage
allowance. We have chosen to show all acceptable ranges in the tables
as a percentage in order to provide uniformity and to facilitate the
interpretation of the data. In some instances the range varies with the
product and more than 1 is shown, e.g., Dade Reagents allow ± 13%
for Lab-trol and ± 10% for Patho-trol. Since the Michael Reese Re-
search Foundation and Burroughs Welicome & Co. do not show an ac-
ceptable range, we have evaluated these products on a ± 5% spread
from their assigned values.
The columns to tile right in tile tables give an indication of the extent
of agreement of our results on the different products within the manu-
facturer’s acceptable range, using the same method as the manufac-
442 LOGAN & ALLEN Clinical Chemistry

turer or using a different method. For convenience ill tabulating, the

Warner-Chilcott values based on “weighed-in” amounts following
dialysisare included in tile ‘‘same method’’ column. One exception to
this occurs in the glucose results, where the manufacturer has quoted
values for glucose by different Illethods. Marginal agreement includes
lots where our results lay just within or just outside the limits of the
manufacturer’s range. Marginal values have been arbitrarily chosen
as those which differ from the borderline limits of the acceptable range
by ± / of the acceptable range. As an example, consider an acceptable
range for glucose of 90-110 mg./100 ml. Marginal values would then
include those from 87 to 93 mg./100 ml. at the lower end of the range
and 107-113 mg./100 ml. at the upper end. In Table 2 (urea nitrogen),
2 of the values obtained on 27 lots of Dade products disagreed with
those of the manufacturer. These 2 lots were early lots and contained
an ingredient which reacted nonspecifically in the diacetyl monoxime
reaction. The manufacturer was apparently aware of the problem and
was able to rectify it. With Hyland Laboratories products, our results
showed disagreement only once in the examination of 19 lots of control
serum, and with Warner-Chilcott serums there was disagreement in 1
lot of 30 examined. In each of these instances our result was just out-
side the marginal value. The Nichael Reese and Burroughs Wellcome
preparations with which we have allowed a ± 5% variation showed 3
disagreements in the case of the former and 2 in the case of the latter.
In these our values were higher in each instance than those given by
the manufacturer.
In the summary of our findings for glucose determinations (Table 2),
it is evident, for the first 2 manufacturers listed, that there were fewer
discrepancies when the same method was used than when a different
method was used. With Hyland products, agreement was satisfactory
with 20 of 22 different lots using the same method, and the 2 discrepant
results were not markedly so. We obtained low values with 2 early lots
of the Clinton product, Chemtrol. In one of these lots the glucose level
fell off over a period of a year from 293 mg./100 ml. to 152 mg./100 ml.
After a second year of storage, a value of 146 mg./100 ml. was obtained.
This occurred only with the 1 lot; with later lots purchased, our results
over a similar period of time have agreed well with the assay values of
the manufacturer. For the glucose determinations we used either a
manual Folin-Wu method (ii) or the automated ferricyanide method.
In general, we have found that the precision of the manual Folin-Wu
method i slightly better than the automated procedure.
Considering the findings for the total protein determinations, it is
interesting to note that with one of the Dade products, Lab-trol, there
Vol. 14, No. 5, 1968 CONTROL SERUM PREPARATIONS 443

were a number of discrepancies between our results and the manufac-

turer’s values (Table 2). In the case of Lab-trol, the manufacturer’s
acceptable range was more demanding than for some of tile other
products. If the acceptable range was calculated instead on the basis
of ± 5% of the assay value, there were considerably fewer discrepan-
cies, as seeii by the figures in parentheses. There was disagreement
with our results in 2 lots, compared with 8 lots on the basis of the manu-
facturer’s allow-able range. The findings with the Clinton product have
been similarly treated and show a similar tendency. For all these
analyses we used a biuret method, either manual or automated.
The chloride determinations were performed using the automated
mercuric thiocyanate method (Table 2). It will be noted that the ac-
ceptable ranges indicated by some manufacturers were quite exacting,
e.g., ± 1.6%. When the acceptable range was calculated on the basis of
± 5%, the number of discrepancies was reduced considerably as seen
by the figures ill parentheses. On a 5% basis, there are 3 disagreements
in 34 lots tested with the Warner-Chilcott serums, 2 of 22 with
Dade products and the 1 Chemical Control Serum of Burroughs
Weilcome.
The results obtained for the total cholesterol determinations using
the Abell et at. method (5) are shown in Table 3. The values obtained
in cholesterol analyses vary considerably with the method used. We
have used both a modified Pearson and a modified Zak procedure in our
laboratory (ii) and have generally found these to yield higher results
than the Abell et at. method. These have not been included in our data
since the Abell procedure is now generally accepted as a reference
method. There was good agreement between our results and the values
given for the Warner-Chilcott products with only 1 discrepancy. Five
different lots of the Dade product, Choles-trol, were examined. No
assay method or acceptable range was given, and we have used a range
of ± 5% to evaluate these. Two disagreements were recorded on this
basis. With Hyland products, a modified Zak procedure was given as
the assay method for Hyland Normal and ilyland Special Clinical
Chemistry Control Serum; a method by Searcy and associates was
indicated for tile Hyland Elevated Cholesterol preparation. rilile
Schoenheimer-Sperry and Zak procedures were specified for the Clinton
product, Chemtrol, while Abell et aL, Schoenheimer-Sperrv, and Pear-
on methods were used in the establishment of values for Metrix by
the Michael Reese Foundation. Tile discrepancies observed with these
products may well be due to tile fact that out’ results were obtained with
a different method.
The sodium determinations (Table 3) were performed on the com-
444 LOGAN & ALLEN Clinical Chemistry

Table 3. 1)ETERMINATIONS ON COMMERCIAL C ONTROL SERUMS

A greement with manufacturers

acceptable range
Lots teAed
Manufacturer & acceptable ranges (No.) Agree .lfarginal Disagree

TOTAL CHOLESTEROL

Warner-Chilcott (±5%) 17 11 3 1
Dade (±5%)t 5 1 2 2
Hyland (±5%) 14 7 5 2
Clinton (±4% to ±8%) 5 1 2 2
Michael Reese (±5%)t 3 1 2 -

SOI)IUM

1)ade (±3%) 10 6 4 -

Hyland (±1.5%; ±2%) 7 4 1 2

Clinton (±1.5%) 2 - 1 1
Warner-Chilcoft (±5%) 16 11 5
Michael Reese (±5%)t 1 1 - -

POTASSI (TM

Warner-Chilcott (±5%) 14 3 5 3
Hyland (±2.5%; ±4%) 5 4 1
Clinton (±1.7%) 1 - 1 -

1)ade (±5%; ±4%) 5 1 4

Michael Reese (±5%)t 1 - 1 -

(AmAmcM

Warner-Chilcott S 4 :i 1
Dade 4 1 2 1
Hyland 2 1 1 -

Clinton 2 1 - 1

URI( A( ID

Warner Chilcott (±5%) 6 5 1 -

Hyland (±6%; ±9%) 13 12 1 -

Dade (±5%)f :1 3 0 -

Clinton (±3%) 1 1 0 -

Michael Reese (±5%)f 3 1 2 -

* Using method of Abell et at. (5).

t Acceptable range not given. We used ±5%.
Basis of acceptable range of ±5%; Clark-Collip method used.

mercial control serums using an automated flame Photometer procedure.

In spite of some rather narrow acceptable railges proposed by some
of the manufacturers, the number of discrepancies was minimal.
Potassium determinations (Table 3) were carried out with the
AutoAnalyzer flame photometer attachment. Agreement with the manu-
facturers’ values was quite good. The 3 discrepancies, 2 with Versatol
and 1 with Versatol Pediatric, were not marked.
The Clark-Collip method (ii) was used for calcium determinations
Vol. 14, No. 5, 196$ CONTROL SERUM PREPARATIONS 445

(Table 3). Actually, most of our results were obtaillt’d using the
AutoAnalyzer method of Kessler and Wolfinan (8). With 1 t’xceptioii
(the Hyland products), there was considerable disagreenlent, the
AutoAllalyzer results tending to be higher than the di If erent maiiu-
facturers’ assay values. Aryan (12) has reported values by the Kessler
automated procedure averaging 3% lower than those by the Clark-
Collip method. We had performed the Clark-Collip niethod Ofl some
commercial products. Since at least 3 manufacturers used this method,
it seemed indicated to present only Clark-Collip results. It will be lloted
that discrepancies occurred in 3 of 16 lots exannne(l by this method.
uric acid determinations on the commercial control sei’ums were
made using tile automated method of Crowley (9), the results of which
are shown ui Table 3. The agreement was good in tile limited number
of products that we have tested for this constituent.
Stability
As mentioned above, we have studied tile stability of several con-
stituents in commercial control serums for 2 years and sometimes 3 or
4 years, when stored according to tile manufacturer’s directions. Sev-
eral bottles or vials of a given lot were purchased and a new vial opened
at each interval for analysis. For each constituent the same procedure
was used throughout. The day-to-day precision of the procedures as
shown in Table 1 was used to determine the significance of change in
coiiceiitration from the initial value. The serums were checked foi’
stability of urea nitrogen, glucose, total protein, total cholesterol,
chloride, and calcium values. Data were obtained up to 2 years on each
of tile constituents; for glucose amid urea nitrogen, some lots were
checked up to 4 years. Only in the total protein analyses were any sig-
nificant changes noted. With I lot of Chemtrol, we obtained a value of
7.73 gm./100 ml. initially and 7.2 gm./i00 ml. after 2 years. On the other
hland, 1 lot of Versatol-A Alternate increased from an initial value of
4.5 gm./i00 ml. to 4.9 gm./100 ml. after 2 years. These are not large
changes but are greater than expected on the basis of the precision of
the automated biuret method which was used for the total protein
determinations.
Discussion
Our results have shown that in analyzing a control serum for a given
constituent, the values obtained will vary frequently according to the
method or modification of the method used. This is particularly true
for glucose and cholesterol. When the same method was used as that
of the manufacturer in assigning values, tile agreement was often
better than when a different method was used. In general, when the
446 LOGAN & ALLEN Clinical Chemistry

same method was used, the value we obtained fell within tue acceptable
range recommended by the manufacturer or quite closely to it. More-
over, the various constituents we have analyzed are (1iute stable for at
least 2 years.
Our studies have led us to conclude that commercial control serums
should not be used in place of a primary standard when the constituent
to be assayed can be obtained in a relatively pure form and a stable
solution can be prepared. This opinion has been reached by a number
of other workers (13-17) and was emphasized by the participants of
the symposium on standards and reference samples at the 18th National
Meeting of the American Association of Clinical Chemists in 1966.
Perhaps more important is the fact that at least 1 manufacturer has
accepted this limitation for the use of their assayed serums. If used
as a primary standard, the assay values of the control serum iiiust be
absolutely accurate and the serum itself must be absolutely stable for
an indefinite period. Only Hyland Laboratories and Burroughs Well-
come & Co. place expiration dates on their packages of commercial
control serums. With 1 exception, the manufacturers derive the values
for their serums’ constituents from tile mean of a number of analyses,
and thus these results cannot be expected to have the accuracy required
of a primary standard. In addition, if these products are to be used as
standards the laboratory should follow exactly the same technic as that
used by the manufacturer in establishing the assay value. One other
manufacturer claims that their products caii be considered as “stand-
ards-in-serum” since they are prepared by weigiiing the constituents
into serums which have first been dialyzed either to remove them com-
pletely from the serum base or to reduce them to fixed low levels. Since
dialysis would not remove protein, it follows that values given for total
protein, total nitrogen, and protein-bound iodine are not based on a
weighing system. For those constituents which are “completely”
removed and then weighed back in, the claim as a standard may have a
slightly better foundation. However, to be considered a primary stand-
ard, the standard-in-serum must be absolutely correct as to concentra-
tion and identity. As Radin (17) points out, as long as the analyst meas-
ures constituents in the presence of interfering substances, the back-
ground absorbance of the manufacturer’s serum base would have to
be the same as that of the unknown to avoid introduction of error. The
manufacturer would seem to be assuming a difficult task to ensure that
his product can be used as a standard for the following reasons: (1)
he is working with human serum or defibrinated plasma which, because
of its inherent biologic variatiQn, can never be exactly duplicated from
batch to batch; (2) he must be certain of the absolute stability of the
Vol. 14, No. 5, 1968 CONTROL SERUM PREPARATIONS 447

material ; and (3) liemay he renloving some constituents in the dialysis

process which play a significantrole in some noiispecific reactions, and
since they are not of’ primary interest to tile laboratory, may not have
been returned to the serum. in the case of those constituents dialyzed
to a fixed level, the final value shown for a constituent must depend on
an assay which determines the portion remaining after the fixed level
is reached-thus for these constituents the case for use as a standard
is less convincing than in those where removal by dialysisis complete.
In any situation in wilich the manufacturer’s value depends 011 an assay
value, it immediately introduces the idea of a prior standard and,
thus, the assayed material becomes secondary. Klugerman and Boutwell
(13) have carefully pointed out the fallacy of using a single control
sample as both a standard and a control in a series of analyses.
A number of constituents are measured ill the clinical laboratory for
which it is difficult to obtain reference material of sufficient purity and
stability in solutiomi to serve as a primary standard. In these situations
one might then choose to use a control serum as a standard as the best
available. We agree with Klugerman amid Boutwell (13) that the control
serum of a different manufacturer then be used to determine whether
the precision of tile method is renlaunhig within the control limits
established for it.
Our data support the use of commercial control serums as quality-
control specimens in daily use in the laboratory. For those laboratories
wishing to carry this out more inex)ensively, lots of unassayed control
serum may be purchased and serve equally well once the allowable
range of values for the required level of precision hlaS been established
for the lot number. Ideally, a control serum, as well as a standard solu-
tion, should be included with each series of analyses. The control serum
goes through the same steps as time specimens being analyzed and is
subject to the same errors. For a given method the mean value and
standard deviation for tile control serum can be established. With this
information it is known when a method is out of control ; if this occurs
the whole procedure must, of course, be put under scrutiny. Commercial
serums with assay values give laboratories some guidance and assur-
ance in establishing their own mean values amId standard deviations.
Actually, as Caraway (16) has pointed out, commercial serums with
carefully estabhisiledassay values are a means of self-evaluation,
particularly if material from more than 1 manufacturer is available
for comparison.
References
1. Belk, W. P., and Sunderman, F. W., A survey of the accuracy of chemical analysis in
clinical laboratories. Am. J. Gun. Pathol. 17, 853 (1947).
448 LOGAN & ALLEN Clinical Chemistry

Urea
#{176}. nitrogen. Technicon Laboratory Method File No. N-la. Technicon IllstrU!I14’lItS
Corporation, Chauncoy, N. Y.
3. Glucose. Technicon Laboratory Method F’ile No. X-2n. Techinico,r Instruinei*ts (Torporatioli,
Chauncey, N. Y.
4. Total protein. TeeIinion Lal)oratorv M(tIio,I I’ih No. N-I 4a. Technicon Instruments
Corporation, Ghauiicey, N. Y.
5. Aboll, IL. L., Levy, B. B., Brodie, B. B., and Kendall, F. E., Simplified method for thi
estimation of total cholesterol in serum and demonstration of its specificity. .1. Biol.
Chem. 195, 357 (1952).
6. Chloride. Technicon Laboratory \lethiod File No. N-5a. Techinicoii Instruments Corpora-
tion, Chauncey, N. Y.
7. Sodium and potassiuiit. ‘l’cchii,ieoit Laboratory Method File No. N-20a. Teelinicon Tnstru-
nieiits Corporation, Chauiicey, N. V.
8. Calcium. Technicon Laboratory Method File No. N-3. Technicon Instruments Corporation,
Chauncey, N. Y.
9. Crowley, L. V., I)eterlninatioll of uric acid--all automated analysis based on a carbonate
method. Gun. Chem. 10, 838 (1964).
10. Barnett, R. N., A scheme for the comparison of quantitative methods. Am. J. GUn. Pat hol.
43, 562 (1965).
11. Manual of Clinical Chemistry (rev. ed.) Laboratory of Hygiene, Department of National
Health & Welfare, Ottawa, Canada, 1966.
12. Aryan, D. A., Observations oti an automated method for determination of serum and urine
calcium. Am. J. Gun. Pat/ioi. 45, 358 (1966).
13. Klugerman, M. R., and lloutwehl, J. H., Commercial control sera in the clinical chemistry
laboratory. GUn. Chem. 7, 185 (1961).
14. Henry, R. J., Clismhai Chemistry: Principles and Technics. Hoeber, New York, 1964.
15. Wootton, I. D. P., Micro-analysis in Medical Bio-chemistry (ed. 4). Churchill, London,
1964.
16. Caraway, W. T., “Sources of Error in Clinical Chemistry.” In Standard Metl,od. of
Clinical Chemistry (Vol. 5), Meites, S., Ed. Acad. Press, New York, 1965, p. 19.
17. Badin, N., What is a standarM GUn. Chem. 13, 55 (1967).