CLIN.CHEM.

32/3, 470-475 (1986)

GraphicalComparisonsof Interferencesin ClinicalChemistryInstrumentation

Melvin Kenneth W. Ryder, and Sheila A. Jackson
R. GIlcl#{231}

Instrument-and analyte-specificinterferencescan be pro- ences, we present a method for measuring and reporting
ducedby addingknownconcentrationsof potentiallyinterfer- interference data, and have begun assembling a “user’s
ing substancesto serum from healthyvolunteers.Analytical guide” for those who wish to compare results from various
resultsare expressedas a percentageof the original(unaf- clinical chemistry instruments. Here we describe the tech-
fected) result. Graphical displays of the transformeddata niques used to generate the graphical displays we shall refer
document the conditions under which erroneous results to as “interferographs,” and discuss the potential usefulness
would be probablefor specimenscontainingthe additive. If of the information these graphs contain (9, 10).
several analyticalsystems are available, one can make an
informedchoice of which to use for a particularanalysisby MaterIals and Methods
comparing the appropriategraphical information.Compari- Serum was obtained from freshly clotted blood drawn
sons of such displays of data from newer systems being from the authors, who were in good health and had used no
consideredfor purchase may facilitatedecisions regarding drugs, including alcohol, for at least 48 h before phlebotomy.
laboratoryinstrumentation. To evaluate the effect of in vitro hemolysis, turbdity, and
bilirubinemia on currently available clinical laboratory
AddItIonalKeyphrases:variation,source of discrete analysis instrumentation, we prepared a series of 20 supplemented
hemolysis .Iipemia turbidity . bilirubinemia
.
sera, as described below. The wide range of concentrations
economicsof laboratoryoperation ‘inferferograplis” for the added substances was chosen during pilot studies,
and by considering the variety of specimens received in our
The ubiquitous nature of interfering substances in clinical hospital laboratory. Aliquots of the supplemented aera were
chemistry assays is generally appreciated by laboratory frozen (-20 #{176}C)and were vigorously mixed after thawing
personnel and, to a lesser extent, by physicians using and immediately before use.
clinical chemistry laboratory data. Extensive lists of inter- Addition of biirubin: We suspended 6 mg of bilirubin
fering compounds (1,2) indicate the widespread effects of (reference standard; Pfanstiehl Labs, Inc., Waukegan, IL
certain drugs and serum constituents; other reports have 60085) in 0.1 mL of dimethyl sulfoxide (ACS grade; Fisher
focused on factors affecting either a specific analysis (3,4), a Scientific Co., Fair Lawn, NJ 07410), then added 0.2 mL of
particular instrumental system (5), or the interfering effect sodium carbonate solution, 0.1 mol/L. This solution was
of a selected drug or class of compounds (6). Reviews added slowly, with continuous mixing, to 9.4 mL of serum;
describing the quantitative effects of interfering substances then 0.2 mL of a 0.1 mol/L HC1 solution was added. We also
have necessarily been limited in scope (7). added proportional amounts of dimethyl sulfoxide, sodium
Technical disclaimers by the manufacturers about the carbonate, and HC1 to another aliquot of the same serum;
potential effect of an interferent are often frustratingly thus producing a “diluent serum” that was similar to the
vague or couched in such general terms as to be almost bilirubin-enriched serum in pH, ionic composition, and
meaningless-e.g., “... hemolysis should be avoided .. other characteristics. These sera were then mixed to make
“... lipemia may cause a falsely increased result ...“ specimens with added bilirubun concentrations of 9,19, 38,
Unfortunately, the unique nature of specimen collection 75,150,300, and 600 mg/L. Bilirubin concentrations were
sometimes makes it impossible to reconstruct the patient’s confirmed by analysis of the diluted serum (11).
original condition or to obtain a specimen that is free from Addition of lipid.s: The turbidity of lipemia was simulated
potential interfering substances. In these circumstances it by adding Intralipid#{174}
20% (Cutter Laboratories, Inc., Berke-
would be useful to have specific information available about ley, CA 94710) to serum. The msrimum concentration, lOg
both the qualitative and quantitative effect of an unterferent of Intralipid per liter of serum, was attained by adding one
on the analytical system, in order to obtain information volume of hntralipid to 19 volumes of the serum. To an
unavailable otherwise. The product labeling of reagents is aliquot of the same serum we added an equal proportion of
not always a reliable source of information about interfering water, to simulate the dilution from the added lipids.
substances, or at best is not in a format that is easily used in Admixture of the supplemented and diluted sera produced
the laboratory (5). Although the subtle differences among specimens containing added hntralipid in concentrations of
similar clinical chemistry analyses are not always appreci- 0.67, 1.25, 2.5, 5, and 10 g/L. The most turbid of these
ated, these differences can account for dramatically different specimens caused a triglyceride readout of 25 g/L in the
assay results when an interfering substance is present (8). Hitachi 705 analyzer.
To encourage standardization in the reporting of interfer. Addition of hemolysat#{233}: Fresh hemolysate (hemoglobin
content at least 200 g/L) was prepared by washing erythro-
cytes (from the same volunteers who were the source of
Department of Pathology, Indiana University School of Medicine, serum) four times with cold isotonic saline solution. After
Wishard Memorial Hospital, 1100 W. Tenth St., Indianapolis, IN eachresuspension of cells and eachrepacking by centrifiiga-
46202.
Presented in part at the 35th and 37th national meetings of the tion, we discarded the supernatant solution and leukocytes.
AACC in New York, NY, July 1983, and Atlanta, GA, July 1985. After the final centrifugation and removal of all possible
Received October 21, 1985; accepted December 16, 1985. saline solution, we lysed the erythrocytes by adding distilled

470 CLINICALCHEMISTRY,Vol.32, No.3, 1986

water, filtered the hemolysate through untreated glass Results
wool, and centrifuged the ifitrate (20 mm, 2000 x g) to
remove cellular debris. The concentration in the final clear The interferographs described here are the result of our
hemolysate was quantified by measuring the cyanmethemo- discussions and attempts to develop a logical, easily under-
globinderivative. We then added the hemolysate to serum stood format to present interference data obtained in the
samples, appropriately diluted with water, to produce a laboratory. Figure 1, an example of an instrument-specific
series of samples containing 6.3, 12.5,25,50, and 100 mg of interferograph, shows the effect of in vitro hemolysis on the
hemoglobin per liter. Before use, each serum was centri- test performed with the aca. Results of assays for potassium,
fuged in blood-collection tubes contiiining serum separator ammonia, and certain enzymes, which are constituents of
gel(Becton Dickinson & Co., Rutherford, NJ 07070). the “contaniinsting” erythrocytes, are omitted fromthe plot.
Instrumentation: Instruments and reagents evaluated Interferographs depicting the effect of added turbidity and
during this investigation included: aca#{174} II and III (Du Pont bilirubin on the analytical results from the aca are shown in
Instruments, Wilmington, DE 19898); Ektachem#{174}-400 and
-
Figure 2. Figure 3 shows results for a similarly treated set of
700 (Eastman Kodak Co., Rochester, NY 14650); Hitachi#{174} 20 sera assayed with a Hitachi 705, Figure 4 the results
705 and 737 chemistry analyzers (Boehringer Mannheim with an Ektachem-700 in use in our laboratory. The De-
Diagnostics, Inc., Indianapolis, IN 46250); Multistat-IlI mand AU-500 analytical system was used to generate the
(MCA’) centrifugal analyzer (Instrumentation Laboratory, data shown in Figure 5.
Inc., Lexington, MA 02137); Olympus Demand” AU-500 Inter-instrument int.erferographs. To show the extent of
(Cooper Biomedical, Diagnostic Division, Malvern, PA inter-instrument variability encountered when we used
19355); DACOS”’ chemistry analysis system (Coulter Elec- different instruments of the same model, we assayed all-
tronics, Inc., Hialeah, FL 33012); KDA#{174} and Parallel#{174} quotaof pooled sera with added interfering substances by
analytical systems (American Monitor Corp., Indianapolis, several similar instruments. Figure 6 shows the range of
IN 46268); and SMA#{174} 12/60, satc#{174},and RA-l000” analyz- glucose assay results obtained with six different aca’s andan
ers CFechnicon Instruments Corp., Tarrytown, NY 10591). equal number of Hitachi 705’s.
Unless noted otherwise in the legends to figures, all Analyte-specific interferographs. Comparisons among dif-
analyses were performed with reagents and instrumental ferent analytical instruments are facilitated when the infor-
conditions exactly as specified by the instrument manufac- mation from multiple interferographs is combined, as Fig-
turer. ure 7 shows for the effect of added turbidity on the estima-
Graphical treatment: The average assay value foreach tion of serum protein concentration by the instruments
specimen was calculated as a percentage of the original listed
(before supplementation) concentration or activity. From a
plot of these percentages vs the concentration of potential
interferent added, we constructed a smooth-fitting curve or DiscussIon
line. Table 1 shows the original mass concentrations or We recognize a distinctionbetween “contamination” and
catalytic activity concentrations determined for the serum “interference” when the effect of an uncharacterized sub-
samples we used in thesestudies. stance such as hemolysate is reported. For example, an

Table 1. AnalytIcal Results (aca) for

Unadulterated Serum
Concn or
Constituent ecty, per Ifter
Glucose 0.91 g
Urea nitrogen 0.109 0
0

Creatinine 80 mg
LDH 136U
AST 31U U,

ALT 15U
4
Bilirubin(total) 9 mg z
Billrubln(direct) 2 mg 0

Calcium 91 mg 0

Creatine kinase 129 U 4

z
Phosphorus 33 mg
Magnesium 18 mg
Uric acid 62 mg
ProteIn 74 g
AlbumIn 41 g
Upase 11U
Cholesterol 2.3 g HEMOLYSATE ADDED (HEMOGLOBIN. gil)
Triglycerides 0.62 9
Lactate 1.2 mmol Fig. 1. Effectof increasingamountsof addedhernolysateon ace results
Aniyiase 53U (as a percentage of the original “no hernolysate added” result)
Glutamyltransferase 10 U The bilirubin was assayed by the pnItrobenzenedezonkim te5a1tuorobortee
SodIum 142 mmol method (Du Pont modifIcation, 1965). Starred C) analyses differed horn the
Potassium 4.4 mmol originalvalue by leesthan3% at any addedhemolyaaleconcentrstontested, and
Chloride 105 mmol we considered to be “not affected” by added hemclysate. GGT, gamma-
BIcarbonate 26 mmol glutarnyltransferase;ALB ( or (F’, albumindeterminedwith bromcreeol green
Abbreviationsas listed in Figs. 1 and 2. or bromcresol purple; ACP,acid phosphatase; 7RIGL, Pefldes; ALI( affcaline
phoephatass; C creatine kinase

CLINICALCHEMISTRY,Vol.32, No. 3, 1986 471

 200

*AST
180 - *AMYLASE
*UREA NITROGEN

180
* CK
* LOU
* LIPASE
* CALCIUM
acaTM
GE GE
* PH OS PHORUS
0 140
0
S * URIC ACID
* ALBUMIN
* CHOLESTEROL
I-
120 - *ALT PHOSPHASE
0I A,
B B

4 4 ROTEjpj......

::
2 C
0 0
B B
0 0
4 4
C C

40

20

I I I I I I I I I
.05 .10 .15 .20 .25 .30 .35 .40 .45 .50 .55 .60
LIPEMIA NTRALIPID ADDED, giLl BILIRUBIN ADDED (gill

Fig. 2. Effect of added turbidity (left) or bilirubin (right) on clinical chenilstiy analyses with the ace
Left: billiubin assayed as described in the legend to 19g.1. C. bHritfn, direct bitrubin; ALT, alanine aminoiraneferase; AST, aspeftate aminotransferase; LDH,lactate
dehydrugenase.Other abbreviations as listed for FIg. 1

200

IHO

lb *500400

S *P0TA* HITACHI’
IOU U 705
to *40,
ID
LEETSN0L4ft****** CO LOB IDE.
ALOALNEp0 so
HO *GLuc001m
0 *CNEAI1NINE
*000IUM
So *POTAISIUM
,urncAao
*PHOEPOOHOS
HITACHI#{174}
*1DB
40 SAlT 705
* ALT
Sd
20

1111111111
1 2 3 4 5 0 7 0 5 tO
OOMOLYSATE ADDED IHEMOOLOSIN. B/LI LIPEMIA IINTOAUPID ADDED. g/Ll DUDSIO 000IDIWLI

Fig. 3. HItachi 705 readout as a function of added hemolysate (left), turbidity (center), and bilirubin (rht)
Abbreviationsas for Figs. 1 and 2. M-E, Htachi bltnln assay by the Malloy/Evelynmethod. Two differentglucose methods were used: T, Trinder’s;H, hexoidnase

200

SOLUCOIB
100 0 US IA
0 AMMONIA
*CHLO*IDU
#{149}AMYLA*E
‘SO S CALCIUM
*UBOTEIN
U * ACT
45 SALT

if ifTM “OH
‘CI
EKTACHEM#{176}
120 SIIUNUSIN 700
*CSIATININE

I
*HICASSOAATE

10 00REI$#{176}
J)sIc..*SeD-.0oIUM’
.uer
00

#{149} EKTACHEM”
700
40

20

HIMOLVEATE *0010 IGEMOOLOHIW. 9/LI

I 2 -
UPEMIA PNTSAUPID ADDED. 0/1.1
S4U5IJBIId ADOBE i/LI

Fig. 4. Ektachem 700 readout as a functionof added hemolyeate (lef, turbidity (center), and bilirubin (right)
Abbreviationsas for Figs. 1 and 2

upwardly-sloping line should result when one evaluates the showing that the added contaminant is actuaUy being
effect of added hemolysate on valuesfor potassium concen- detected. Although viewed as an interference and a contra-
trations. This (predicted) response would be reassuring, indication to the use of hemolyzed serum for potassium

472 CLINICALCHEMISTRY, Vol. 32, No. 3, 1986

 200
U

ISO
/
7
4t

if 140 if
120
,fl,_C C

MVL*UO_
if
S 00

I SO

0
20

I I I I” ‘s., 11111
2 3 4 5 0 7 0 9 10
OEUDLYU*TD ADDED IIIEMDGLDHIN. i/LI UPEMIA IINTSAUPIO ADDED. 4/LI SIUBUBOE ADDED 1,/ti

Fig.5. Readoutfrom the Demand AU-500 analyzerafter additionsof hernolysate(left), Intralipid (center), and bilirubin (fight)
GOT (La, unbianked (I.e.,no serum blank) for gamma gkitam’,l tranepeptidase resufts were lagged” with a warning code, indicating absorbance limits were
exceeded. Abbreviationsas In FIgs. 1 and 2

tion of added substance, when the effect of interfering

0.12
compounds on clinical chemistry assays is evaluated. Be-
GE
cause response to added interferents might not be linear, it
0.10
0 is necessary to challengethe system at several concentra-
tions across the entire range. Observations reported here
r5 (Figure 5) and others we have accumulated coniirm the need

I
0.05

0.00 for a series of dilutions of the interferent, such as we have

described, for one to observe any nonlinear responses to
0.44 interferences. The choice of the highest concentration of
added interferent may seem arbitrary, but should at least
0.02 cover the range of concentrations encountered in thelabora-
tory in which the interferograph will be used (12). When
drugs are added, a peak concentration of 10-fold the expect-
200 400 600
Hemoly0050 Added (AS H.moglobin. rngidL(
ed concentration with usual dosage regimens has been
suggested (13). The concentrations of hemolysate, and the
Fig. 6. Range of inter-Instrument (six aces and six HItachi 705 turbidity and biirubin concentrations we used in this study
analyzers) variation for glucoseassays as a functionof addedhemoly-
sate are within the range encountered in specimens from surviv-
ing patients at this hospital.
The format of the interferographs used here and reported
200 -
14.0 previously (9, 10) to display the results of comprehensive
ISO -
PR 0 T El N
12.5 interference studies allows inter-method comparisons unaf-
fected by the absolute value of the analytical result, analo-
160 gous to the use of the coefficient of variation (CV) to compare
GE standard deviations irrespective of the absolute value of the
a 140
0 mean. The slope of the line plotted on the interferograph
t 120 lob 0.0 allows detection of those analyses that are likely to be
U,
altered significantly if the interferent is present. Further-
4
101 * ****
more, if the concentration of interfering substance is known,
C
0
B
80 - the approximate error can be calculated. An application of
0 this is a listing of 9nterferovalues” (7) quantifying the
4
C
50 40 extent of interference by endogenous substances and cepha
Not Aff0c10d
* EKTACHEM’ 700 * OS00ow
drugs with creatinine assays.
40
* HrTACHr737 The interferographs shown here were generated by using
* HITACHr 705 I+SB)
20 * 1.5 serum from healthy 0normal” individuals as the baseline for
all calculations, and the percentage error shouldbe compa-
1 2 2 4 5 5 7 8 9 10 rable for all similar specimens. The percentage error would
LIPEMIA IINTRALIPID ADDED, giLl not necessarily be constant if the initial analyte concentra-
Fig. 7. Effect of added Intralipid on results for protein assays by the tion is increased. Because of the uncertainty involved with
IndicatedInstrumen systems pathological specimens we recommend caution in the ex-
4(98)Il refers to the use or nonuse of a sample blank abeor’oance correction In trapolation of this data. For example, the apparent 180%
the assay. oAcos (Old) or (New) resultspriorto or after softwarerevision2.0
00 increase in bilirubin concentration (increasing from 9 to
16.2 mg/L) illustrated in Figure 1 would be calculated as a
assay, this response is not unexpected. Thus potassium more modest 114% increase if the initial bilirubin concen-
would be classified as a cont.ssminfint rather than an inter- tration were 50 mg/L, and the effect of added hemolysate
ferent. were to produce a constant positive bias of 7.2 mg/L. An
Attention must also be directed to the choice of concentra- alternative approach would be to prepare other specimens

CLINICALCHEMISTRY, Vol. 32, No. 3, 1986 473

with the analyte concentration near the medically signifi- ination of up-to-date information about the instrument-
cant action level, or any other selected concentration. Quan- specific and reagent-related effects of interfering substances.
titative performance limits could be chosen based upon the Without this information, comparisons of the accuracy of
medical usefulness criteria of Barnett (14) or other appropri- analytical results to be expected after introduction of newer
ate limits. When one is evaluating interference effects, it is instrumentation will be difficult or impossible to make.
also important that a distinction be made between medical Contrary to optimistic expectations on the part of some
and statistical significance as it applies to a series of results persons, our studies of interferences convince us that some
(6). of the second- and third-generation analyzers may be more
Interferographs showing assay results for a single constit- affected by interfering substances than were the earlier
uent could have each data point plotted directly in the laboratory systems. A degradation in performance is espe-
concentration units of the instrumental readout; however, cially apparent in the transition away from a pre-analytical
the choice of scales on the ordinate will determine the slope separation step. In this case, newer and faster systems are
of the resulting line. We recommend a combined plot, such not necessarily more specific than the systems they sup-
as that shown in Figure 7. An advantage of norrnfihiCing planted.
each assay value to a percentage of the original value for the We selected the interferographs presented here to ifius-
unadulterated specimen, as the glucose-specific example in trate the kinds of information we have assembled, showing
Figure 6 shows, is the resulting ease of comparing assay the response of different instruments and reagent combina-
results from instruments that are calibrated in unrelated tions to interfering compounds. We are expanding this study
units. This relative independence from the absolute value of of interfering substances to include the newer analytical
the original readout facilitates comparison of the effects of systems as they become available. In additional investiga-
turbidity on, for example, isIki1ine phosphatase assay, even tions planned, we will add other potentially interfering
though the analytical methods and calibrations may be substances, such as hemoglobin (not hemolysate), carot-
dissimilar. enoids, drugs or metabolites, vitamins, and various sub-
By bringing diverse information together in a format that groups of proteins and lipoproteins. To be maximally useful,
invites comparison, directions for future research and devel- numerous interferographs must be made available for com-
opmental efforts may be suggested. For example, Figure 7 parison. Similar evaluations are underway for those instru-
shows that several of the systems currently available for mental systems designed to be used in satellite and physi-
protein quantification are not affected by added turbidity cian’s-office laboratories. To facilitate this effort, we invite
from lipid micelles. Further delineation is required to de- participation with us in this ongoing project.
scribe the critical components required for an “optimal”
method for protein quantification in the presence of lipemia,
but it appears that the use of appropriate serum blsinlring is We are grateful to our colleagues at various institutions for
effective for some instruments. A more direct approach to helpful suggestions and analyses of the described serum specimens,
including Drs. Robert P. Hooker and (larry L Bolinger (satsc and
counteract the light-scattering effect of lipid turbidity is Hitachi-705), Lloyd S. Rothouse (KDA and Parallel), Richard it
exemplified by the thin-film technology used in the Ekta- Swain (Multistat-m, Ektachem-700, and Hitachi 737), and K
diem analyzer (15). Similar combinations of reagents and Owen Ash (s?&Acand RA-1000). Although the studies reported in
instrumental parameters can now be described for many of this article were not so funded, current production of computer-
the constituents we have examined. Theoretically it now is assisted drawings of interferographs is made possible by a grant
possible, through judicious choice of instruments, reagents, from Eastman Kodak Co., Rochester, NY 14650.
or instrumental conditions, to identify a combination of
analytical methods that are nearly free of interferences.
Close inspection of a complete set of interferographs can
lead to other suggestions for improvements in methods, if
the requisite flexibility in instrumentation is available. References
Alternatively, one might choose to obtainthe analytical 1. Young DS, Pestaner LC, Gibberman V. Effects of drugs on
system that is most nearly free of those interferences likely clinical laboratory tests [Special issue]. Cliii Chem 1975:21-11)-
4321).
to be encountered.
2. Friedman RB, Anderson RE, Entune SM, Hirshberg SB. Effects
Figure 6 shows that generic interferographs are not of diseases on clinical laboratory tests [Special issue]. Cliii Chem
sufficient to define the response of all imi1ar instruments to 1980;26:1D-476D.
interfering substances. A comparison of the responses of the 3. Paterson Y, Lawrence EF. Factors affecting serum creatine
multiple sea’s, contrasted with the wider range of responses phoephokinase levels in normal adult females. Clin Chim Acts
from several Hitachi 705’s illustrates the need for instru- 1972;42:131.-.9.
ment-specific interferographs if assay systems are used that 4. Kroll MH, Hagengruber C, Elm RJ. Reaction of picrate with
allow for choices in reagents, wavelength, blanking, or other creatinune and cepha antibiotics. Cliii Chem 1984;30:1664-6.
such variables. Also, whenever a change is made in reagent 5. Yoeaelson-Superstine S. Drug interference with clinical tests
or other assay condition, it is imperative that the effect of performed by a 20.chsnnel computerized autoanalyzer. Am J Hoep
interfering substances be re-evaluated. During the course of Pharm 1982;39:848-9.
this study, we became aware of circumstances in which the 6. Baer DM, Jones RN, Mullooly JP, Homer W. Protocol for the
subtraction of the absorbance of an appropriate “serum study of drug interferences in laboratory tests: cefotaxime interfer-
ence in 24 clinical tests. Clin Chem 198329:1736-40.
blank” decreased the response of an analytical system to
7. Okuda T, Ito J, Nishida M. “Interferovalue” indicates the
turbidity, while at the same time increasing the instrument interference of substances with creatinine determination [Letter].
response to added hemolysate (unpublished observation). Clin Chem 1984;30:1888-9.
Rapidly changing technology, exemplified by the recent 8. Naiji AA, Whitlow KJ. Spurious increase in serum creatinine
introduction of numerous “random-access, high-throughput, associated with intravenous methyldopate therapy [Letter]. Drug
multi-wavelength” analyzers makes imperative the dissem- Intell Cliii Pharm 1984;18:896-9.

474 CLINICALCHEMISTRY,Vol.32, No. 3, 1986

9. Glick MR. Ryder KR, Hooker EP, et al. “Interferograms” de- (Fluosol-DA#{174})
on clinical chemistry tests [Abstract]. Am J Clin
signed to depict the influence of interfering substances on many Pathol 1984;81:751.
clinical chemistry instruments [Abstract]. Clin Chem 198329: 1208.
13. Anonymous. Guidelines for interference testing in clinical
10. Glick MR. Ryder KW, Jackson SA. Comparisons of clinical chemistry. National Committeefor Clinical Laboratory Standards,
chemistry systems: response to interfering substances, as shown by PA 19085, 1985;NCCLS publica-
711 E. Lancaster St., Vilanova,
analyte.specific “interferograms” [Abstract]. Clin Chem tions EP7-P.
1985;31:1015-6.
14. Barnett EM. Analytic goals in clinical chemistry the patholo-
11. Perry BW, Doumas BT, Bayse DD, et al. A candidate Reference gist’s viewpoint 1983. Pathology 1983;31:319-22.
Method for determination of bilirubin in serum. Test for transfer-
ability. Cliii Chem 1983;29:297-301. 15. Curme HG, Columbus RL, Dappen GM, et al. Multilayer film
elements for clinical analysis: general concepts. Clin Chem
12. Mulluns RE, Hutton PS, Cons RB. Effects of artificial blood 1978;24:1335-42.

CLINICALCHEMISTRY,Vol.32, No. 3, 1986 475