not distinguish fetal blood from certain hemoglobinopupper range of linearity is -2000 mgfL. Thus this
method may be performed
with mecomum samples conathies such as hereditary
persistent
HbF.
taming HbF as little as 10% of total Hb by weight, for
An average error for this method is -10% (Table 2).
An experimental
error of this magnitude
will not affect
which the original color of the specimen may be brownish-green.
Samples with relatively high Hb content may
the deduction of source of blood in the meconium. Conneed to be diluted. To obtain the proper dilution, use a
sequently, the spectrophotometric
modification of the
freshly made hemolysate containing 20 L of blood in 5
Apt test can unambiguously
distinguish the origin of
mL of water for comparison.
blood in a newborn’s discharge.
There are no commercial
quality-control
samples
References
available for assay of these specimens. However, control
1. Apt L, Downey WS Jr. “Melena” neonatorum:
the swallowed
specimens can be easily prepared by lysing fresh blood
blood syndrome. J Pediatr 1955;47:6-12.
cells in deionized water to generate a Hb solution of -10
2. Silber G. Lower gastrointestinal bleeding. Pediatr Rev 1990;12:
85-93.
mgfL. A positive blood specimen containing HbF should
3. Brinkman R, Jonxis JHP. The occurrence of several kinds of
also be used. These solutions can be aliquoted into
hemoglobin in human blood. J Physiol 1935;85:117-27.
1.5-mL fractions
and, stored frozen or refrigerated,
4. Singer K, Chernoff Al, Singer L. Studies on abnormal hemogloshould be stable for at least 6 months. Stored at 4#{176}C, bin: their determination on sickle celi anemia and other hematologic disorders by means of alkali denaturation. Blood 1952;6:413HbF decreases by -10% after 1 year.
28.
An HbF content >50% probably indicates that the
5. Henry RJ, Cannon DC, Winkelman JW, eds. Clinical chemisblood is fetal in origin. Blood with HbF <10% definitely
try: principles and technics. Hagerstown,
MD: Harper & Row,
comes from the mother. Obviously, this procedure can1974:1116-80.
CLIN.CHEM.39/11, 2329-2332 (1993)
HemoCue p-Glucose Photometer Evaluated for Use in a Neonatal Intensive
Care Unit
Elizabeth Vadasdi’ and Ellis Jacobs”
We evaluated the HemoCue p-Glucose Photometer system for use in our neonatal intensive care unit by assaying
178 heparinized whole-blood samples obtained by heel
stick. The required sample size is 5 L. Plasma glucose
was analyzed by a glucose oxidase/oxygen electrode
methodology. Across the glucose range of 1.28-21.87
mmol/L, the regression slope was 0.976 (r = 0.976, S =
0.475). For samples with hematocrit 0.30, the regression slope was 0.981 (r = 0.950, S, = 0.415); for hematocrit of 0.31-0.49, the regression slope was 0.984 (r
=
0.972, S = 0.508); and for hematocrit 0.50, the
regression slope was 0.959 (r = 0.988, S
= 0.394).
Human whole blood, bovine whole blood, and bovine
serum-basedquality-controlmaterialswere studied.Except for assays of the low-concentration
humancontrol
material, the total CV was <3.5%. The accuracy and
precision of the HemoCue system were comparable with
those of conventional laboratory instrumentation.
Ind.xlng Terms: reflectance photometiy . point-of-care testing
hematocrit . critical care medicine
pediatric chemlstiy
‘Center for Clinical Laboratories, The Mount Sinai Hospital,
and2 Department of Pathology, Mount Sinai School of Medicine,
New York, NY 10029-6574.
3Mdress correspondence to this author at: The Mount Sinai
Medical Center, STAT Laboratories,
Box 1519, One Gustave L.
Levy Place, New York, NY 10029-6574. Fax 212-876-0651.
Received November 6, 1992; accepted June 26, 1993.
For the rapid assessment of glucose homeostasis, reagent strips impregnated
with glucose oxidase have
been available since the 1960s, and portable reflectance
meters based on this principle have been marketed since
1970. By the 1980s, bedside capillary glucose monitoring had become a standard of practice even though several authors expressed concerns with the accuracy of the
measurement (1-3). New and modified solid-phase reagent strips/meter systems were developed for utilization with the latest reflectance technology (e.g., Glucometer, Ames; One Touch, Lifescan; Accu Chek II,
Boehringer Mannheim Diagnostics) and electrochemical technology
(Satellite
G, Medisense) (4-6). All these
developments improved the accuracy of measurement,
but none fully eliminated the effects of extreme hematocrit values on glucose determination in whole-blood
samples.
In clinical practice, dry-reagent strip methods are
widely used to screen for abnormal
glucose concentration, most frequently for identifying neonatal hypoglycemia (7, 8). Several reports have emphasized (7-9)
that, although dry-reagent strip technology is useful for
screening glucose in neonates, it cannot be relied upon
for monitoring glucose homeostasis. Thus, before any
therapeutic intervention, confirmation by a conventional laboratory method is necessary.
In our previous investigation of the influence of heCLINICAL CHEMISTRY, Vol. 39, No. 11, 1993 2329
matocrit
on glucose measurement
in whole blood (10),
two of the systems we studied were dry-reagent
strip!
reflectance meter systems: One Touch (Lifescan, Milpitas, CA) and Accu Chek II (Boehringer
Mannheim
Diagnostics,
Indianapolis, IN). Hematocrits
>0.55 significantly
affected the precision of both systems. Because the normal range for hematocrit
in newborn infants is 0.55-0.65,
neither system is applicable, as
indicated by previous authors, for accurate assessment
of glucose homeostasis
in neonates.
To eliminate the
effect of extreme hematocrit values on the accuracy of
glucose analysis, one may assay either serum or plasma
(11). However, use of these samples would necessitate
a
centrifugation
step, which would make the system cumbersome to use at the bedside.
Ultimately,
a system for quanti1ying
whole-blood glucose that is not significantly
affected by hematocrit
is
the ideal solution. The HemoCue a-Glucose
Photometer
system (HemoCue AB, Angeihoim, Sweden) reduces the
effect of hematocrit on the analysis by hemolyzing the
erythrocytes.
This whole-blood glucose testing system
consists of disposable 5-pL cuvettes and a photometer.
The cuvette contains
freeze-dried reagents
deposited
uniformly in the cavity. The sample, which is drawn
into the cavity by capillary
force, mixes with the reagents spontaneously.
The erythrocytes
are hemolyzed
by sapornn, releasing the intracellular
glucose. The a-Dglucose content of the specimen is transformed
to 13-Dglucose by the enzyme mutarotase
(EC 5.1.3.3). Glucose
dehydrogenase
(EC 1.1.1.47) catalyzes the oxidation of
p-D-glucose to form NADH. Via the action of diaphorase
(EC 1.8.1.4), the NADH produced reduces methylthiazolyldiphenyl tetrazolium
to a colored formazan dye.
When the chemical end point is reached, the formazan is
quantified
photometrically.
To eliminate
the effect of
turbidity, the photometer uses dual-wavelength
measurement, at 660 and 840 nm.
The purpose of this study was to ascertain the viability of HemoCue p-Glucose photometer for use in a neonatal intensive-care
unit. Only neonatal
specimens
were used in this study. Results were correlated to glucose oxidase analysis of plasma specimens. Alternative
quality-control
materials were studied.
Materials and Methods
Specimens. Whole-blood
specimens anticoagulated
with sodium heparin were selected at random from
those routinely received in the stat laboratory for blood
gas analysis. These specimens were obtained by heel
stick from premature or critically ifi infants, <12
months old, in our neonatal intensive care unit. After
their arrival in the laboratory at room temperature, the
samples were immediately assayed with the HemoCue
p-Glucose system. Plasma was obtained by centrifuging
analiquotfor2min
at 2540 xginaMicrofugel2
(Beckman
Instruments, Brea, CA) with a 13.2 fixedangle rotor. The plasma was analyzed for glucose with a
Beckman
Glucose II Analyzer. All testing was done
according to manufacturer’s recommended procedures.
Microhematocrit was determined by centrifugation for
2330 CLINICALCHEMISTRY,Vol. 39, No. 11, 1993
3.5 mm at 15225 x g in a Microfuge 12 with a microhematocrit rotor. All specimens were analyzed in duplicate,
and testing was completed within 10 mm of receipt.
Quality control. For quality control, three different materials at two concentrations each were used: a stabilized
human whole-blood matrix control, Meter Trax (Hematronix, Bemcia, CA), which requires storage at 2-8 #{176}C;
a
stabilized
bovine whole-blood matrix control,
SugarChek (Streck Labs., Omaha, NE), which was stored at
room temperature;
and a bovine serum matrix control,
Test Point (Technicon Instruments,
Tarrytown,
NY),
which was kept at 2-8 #{176}C
before and after reconstitution.
Statistics.
The National Committee
for Clinical Laboratory Standards
(NCCLS) EP5 protocol for the assessment of imprecision of an analytical
system was utilized
(12). Paired Student’s t-test and least-squares
linear
regression
analysis were used for the methodology comparison and to assess the influence of hematocrit.
Results
The within-run
imprecision
(CV) of the HemoCue system was ascertained
with the three types of qualitycontrol materials
(Table 1). The human whole-blood
control generated significant imprecision, CV
at
the lower concentration. Both bovine controls,
i.e.,
whole blood and serum, generated
acceptable precision
(CVs 1.5-3.1%). However, after reconstitution,
the serum control was stable for only 8 days at 4-8 #{176}C,
whereas the Sugar-Chek solution was stable for 30 days
after opening at room temperature.
Therefore, we used
the Sugar-Chek
material for the remainder of the study.
The effect of storage temperature
on the microcuvettes was measured
for 30 days. Microcuvettes
from
the same lot number were kept between 2 and 8#{176}C,
as
recommended
by the manufacturer;
a duplicate was
kept at room temperature.
Assay imprecision
for both
types of cuvettes was ascertained with use of the SugarChek material and the NCCLS EP5 protocol (Table 2).
5.1%,
Table 1. WithIn-Run ImprecisIon of HemoCue
p-Glucose
Photometer
Mean,mmol/L (CV, %)
Ceofral matedal
3.76
Level2
14.53 (1.6)
6.17 (2.6)
13.30(1.5)
14.90 (3.1)
Level
TestPoint
MeterTrax
Sugar-Chek
1
(1.7)
2.64 (5.1)
n = 20 each.
Table 2. Effect of Cuvette Storage Temperature
on
HemoCue /3-Glucose Photometer ImprecIsIon
Storags
mmol/L (CV,%)
temp, C
Central
Lovsl
4
Low
7.62
4
High
15.81
25
Low
8.06
25
High
16.10
SIR
S,, Intrarunstandarddeviation;STOT,
0.320
(4.20)
0.360 (4.72)
0.590
(3.73)
0.773(4.89)
0.284
(3.52)
0.318
0.467 (2.90)
0.528
totalstandarddeviation.
(3.94)
(3.28)
Table
3. Influence of Hematocrlt
range
n
0.187-0.77
178
HemoCu.
5.17
(2.16)
5.49(1.34)
5.17 (2.15)
5.21 (2.18)
40
4.84(2.54)
4.91
>0.49
(2.46)
The difference was statistically
significant
(P <0.005)
between the low control results obtained
with the Cuvettes that had been stored refrigerated
vs those stored
at room temperature,
the later giving the higher results. Additionally
there was greater imprecision of the
assay at both concentrations
when refrigerated cuvettes
were used. Storage of cuvettes at higher temperatures
reduced both the intrarun and total imprecision
of the
assayto
<4%.
On 178 paired
whole-blood and plasma samples obtained by heel stick, glucose values were determined
with both the HemoCue system and the glucose oxidase/
oxygen electrode methodology (Beckman Glucose II Analyzer). Paired Student’s t-test showed no statistically
significant difference between the mean values (Table 3)
of the two methods over a hematocrit
range of 0.1850.72 and a glucose range of 1.28-21.87
mmol/L. Furthermore, linear regression analysis demonstrated excellent
correlation of the two methodologies,
with a slope of
0.976 (r
=
Photometer
Slope
(2.16)
5.55(1.29)
23
/3-Glucose
GlucoseII
5.15
115
<0.31
0.31-0.49
on Accuracy of HemoCue
Mean (SD), mmolL
and Correlation with Enzymatic
Rsgrssalondata
Assay
y.kilarcmpl
9,,,
r
0.9760
0.1496
0.4745
0.976
0.9807
0.9835
0.0193
0.4153
0.950
0.1205
0.2740
0.5081
0.3937
0.9586
0.972
0.988
>0.65 (17). Additionally,
about 1% of newborns has
hyperviscosity caused by factors other than high hematocrits. The primary contributing factors to whole-blood
viscosity are the number, size, and shape of erythrocytee
and concentrations of plasma proteins (18). In the high
hematocrit range, deformability of erythrocytes causes
increased viscosity; as the hematocrit rises to >0.65, the
viscosity
increases progressively
(19). Furthermore,
Gross et al. (20) found the deformability
of preterm
infants’ cells to be less than that of adult cells, resulting
in an increased viscosity in the blood of infants. Because
the dry-reagent strip systems, whether based on reflec15
250
12
200
9
150
0.976).
To ascertain the influence of hematocrit on the wholeblood glucose measurements
with the HemoCue system,
we separated the data into three subgroups based on the
hematocrit (Table 3). There was no difference in either
the accuracy or the correlation
of the HemoCue system
when broken down by hematocrit range (Figure 1). Of
particular
importance
was the fact that there was no
statistically
significant
difference
between
the values
determined for specimens with hematocrits >0.49.
S
100
3
50
-J
-J
15
0
E
E
250
E
12
200
Id
In
0
U
9
150
U,
0
-j
Discussion
‘5
The purpose of this study was to determine
whether
the HemoCue /3-Glucose Photometer is accurate and
reliable for monitoring capillary blood concentrations
of
glucose in critically ill newborns. In neonates, the normal hematocrit
values of capillary blood may be as high
as 0.629 (±0.032) (13), and these high values are often
accompanied by hypoglycemia. This combination is outside the reliable performance range of the currently
used dry-reagent
strip system and thus necessitates the
use of conventional laboratory methodology for monitor-
0
ing glucose
6
100
50
0
-J
Id
-J
Id
-J
15
250
0
12
200
9
150
6
100
homeostasis.
in whole-blood from neonates is
The presence of extreme hematocrit values could itself give misleading
results
(14-16)-a
factor relevant nisiinly in small-for-date
neonates with polycythemia,
in whom a falsely low value
may be cause for an unnecessary intravenous infusion of
glucose. About 4% of the newborn population has polycythemia,
with increased blood viscosity and hematocrit
0
0
3
Glucose analysis
fraught
c
50
with pitfalls.
p
0
0
3
6
9
12
15
GLUCOSE(mmol/L)
Influence of hematocrlt on correlation of HemoCue wholePLASMA
FIg.
1.
blood glucose values to plasma values
sM testIngwas performedas decilbed
CUNICAL
Inthe text
CHEMISTRY. Vol. 39, No.
11,
1993 2331
tance meters or electrochemical
detection systems, depend on a timed diffusion of glucose from the plasma
into the test pad, an increase in sample viscosity will
slow the diffusion and result in depressed values.
The results of the current study indicate that the
analytical performance of the HemoCue system for testing whole-blood glucose in critically ifi neonates closely
approximates the values obtained with conventional
laboratory instruments. There was no statistically significant difference in the results obtained by both methods, and the correlation was excellent, yielding slope, r,
y-intercept,
and S
values of 0.976, 0.976, 0.150, and
0.475, respectively. Moreover, there was no significant
difference in the correlation studies when the data were
separated into low, normal, and high hematocrit ranges.
The influence of erythrocyte concentration, i.e., hematocrit, on the automated whole-blood glucose analysis is
decreased significantly
by hemolyzing
the sample before
initiating the chemical reaction. There is no significant
change in the plasma glucose concentration
upon hemolysis because the concentration of glucose in the water phase of both erythrocytes and plasma is identical
(21). Furthermore, because the size (mean corpuscular
volume) of neonatal erythrocytes is -30% greater than
that of adult erythrocytes (22), upon hemolysis, the dilutional effect of membrane proteins is less with neonatal than with adult specimens.
Data obtained with different quality-control materials indicate that the bovine serum chemistry control
used in routine clinical laboratories is also a suitable
control material for the HemoCue system. However,
after reconstitution, the control is stable for only 8 days
and has to be refrigerated, whereas the Sugar-Chek
controls (stabilized bovine whole blood) were stable for
30 days after opening and are routinely stored at room
temperature. The Meter Trak stabilized human wholeblood controls were also acceptable. However, the requirement of constant storage at 2-8 #{176}C
makes this material a less viable alternative in clinical practice.
The storage temperature of the microcuvettes affected
the analytical performance
of the HemoCue system.
Storage at higher temperature significantly increased
(P <0.005) the measured mean glucose value for the
low-concentration
control solution from 7.62 to 8.06
mmol/L. Additionally,
there was an increase in precision when the nonrefrigerated
rather than the refrigerated microcuvettes were used. This is probably because,
according to the manufacturer’s
procedure, the cuvettes
are used immediately
after removal from the storage
container, so that the contents of the refrigerated
cuvettes do not have sufficient time to temperature
stabilize. This, in conjunction with fluctuations
in room temperatures, probably causes a significant variation in the
final reaction
temperature.
Thus, implementation
of the
HemoCue /3-Glucose Photometer system as a point-ofcare device requires careful consideration of the effect of
temperature
on the accuracy.
2332
CLINICAL CHEMISTRY, Vol. 39, No. 11, 1993
In conclusion, whole-blood glucose analysis with the
HemoCue /3-Glucose Photometer system is simple, accurate, and precise. Accuracy and precision comparable
with those of “wet” chemistry methods were achieved by
a system with minimal
maintenance
and a sample size
of 5 L. Results were not affected by extremes in sample
hematocrits.
We thank
HemoCue AB for supplying the materials
used in this
study.
References
1. Chu SY, Parker HE. Evaluation of the reliability
of bedside
glucose testing. Lab Med 1989;20:93-6.
2. Price CP, Burrin JM, Nattrass M. Extra-laboratory
blood glucose measurement:
a policy statement. Diabetic Med 1988;5:705-9.
3. Kelly TM, Jensen RL, Robinson MK. Method for estimating
confidence levels for measurements
by glucose monitoring system.
Diabetes Care 1988;11:808-12.
4. Hay WW Jr, Osberg IM. The “Eyetone” blood glucose reflectance colorixneter for in vitro and in vivo accuracy and clinical
efficacy. Clin Chem 1983;29:558-60.
5. Brooks EK, Rawal N, Henderson RA. Laboratory assessment of
three new monitors of blood glucose: Accu Chek II, Gluconometer
II, and Glucoscan
2000. Cliii Chem 1986;32:2195-200.
6. Burritt MF, Klee GG, Hanson E, Converse JT, Foreman B..
Evaluation of the One Touch glucose meter for use by a phlebotomy team. Lab Med 1990;21:512-6.
7. Herrera AJ, Hsiang YH. Comparison
of various methods of
blood sugar screening in newborn infants. J Pediatr 1983; 102: 76972.
8. Perehnan RH, Gutcher GB., Engle MJ, MacDonald MJ. Comparative analysis of four methods for rapid glucose determination
in neonates. Am J Dis Child 1982;136:1051-3.
9. Liii HC, Maguire
C, Oh W, Cowett B.. Clinical laboratory
observations.
Accuracy and reliability
of glucose reflectance
meters in the high-risk neonate. J Pediatr 1989;115:998-1000.
10. Jacobs B, Vadasdi B, Roman S. Colman N. The influence of
hematrocrit, uremia and hemodialysis
on whole blood glucose
analysis. Lab Med 1993;24:11-6.
11. Kaplan
M, Blondheim 0, Alon I, Eylath U, Trestian S,
Eidehnan A. Screening for hypoglycemia with plasma in neonatal
bloodof high hematocrit value. Crit Care Med 1989;17:279-82.
12.
NCCLS
Tentative Guideline EP-5T. User evaluation of preci-
sion performance
of clinical
chemistry devices. Villanova,
PA.
Laboratory
Standards,
June
National Committee for Clinical
1984.
13. Gatti RA. Heniatocrit
values
infant. J Pediatr 1967;70:117-9.
of capillary blood in the newborn
14. Sidebottom RA, Williams P, Kanarek K. Glucose determinations in plasma and serum: potential error related to increased
hematocrit. Clin Chem 1982;28:190-2.
15. Barreau PB, Buttery JE. The effect of the haemathcrit
value
on the determination of glucose levels by reagent-strip methods.
Med J Aust 1987;147:286-8.
16. Wiener K. The effect of haematrocrit
on reagent strip tests for
glucose. Diabetic Med 1991;8:172-5.
17. Wirth FH, Goldberg KR, Lubchenco LO. Neonatal hyperviscosity: I. Incidence. Pediatrics 1979;63:833-6.
18. Bergquist G. Viscosity of the blood in the newborn infant. Acta
Paediatr Scand 1974;63:858-64.
19. Mackintosh TB, Walker CHM. Blood viscosity in the newborn.
Arch Dia Child 1973;48:547-53.
20. Gross GP, Hathaway
WE, McGaughey HR. Hyperviscosity
in
the neonate. J Pediatr 1973;82:1004-12.
21. Caraway WF. Carbohydrates. In: Tietz NW, ed. Textbook of
clinical chemistry,
2nd ed. Philadeiphim
WE Saunders,
1986:
784-7.
22. Sega!
CB, Oski BA. Hexnatolor of the newborn. In: Williams
WJ, Beutler B, Erslev AJ, Lichtman MA, eds. Hematolor, 4th ed.
New York:McGraw-Hill, 1990:100-li.