KRYNSKI AND LOGAN: DEXTROSTIX
Canad. Med. Ass. J.
Oc.2,1967, vol. 97
que ceux qui n'avaient aucune reaction. On a aussi
decele 23 erreurs dans l'administration des medica-
s'agissait d'erreurs provenant de changements des
prescriptions du medecin et, dans 13 cas, ces erreurs
sont survenues apres des changements des equipes
d'infirmieres. Huit de ces erreurs se sont produites
chez des malades
qui recevaient plus de cinq
medicaments ce jour-la. Au
malades ont presente des reactions defavorables a
des medicaments ou
Dans11 cas, il
total, 30% des
The authors wish to thank Dr. F. S. Brien, Professor
of Medicine, for his assistance in setting up this study
and for his continued interest.
1. BARR, D. P.: J. A. M. A., 159: 1452, 1955.
2. SCHIMMEL, E. M.: Ann. Intern. Med., 60: 100, 1964.
L.E., THORNTON, G.
J. A. M. A., 188: 976, 1964.
4. MOSER, R. H.: Diseases of medical progress, Charles
C Thomas, Publisher,
5. MACDONALD, M. G. AND MACKAY, B. R.: J. A. M. A.,
190: 1071, 1964.
6. SMITH, J. W., SEIDL, L. G. AND CLUFF, L. E.: Ann.
Intern. Med., 65: 629, 1966.
7. VERE, D. W.: Lancet, 1: 370, 1965.
8. MODELL, W.: Ann. Rev. Pharmacol., 5: 285, 1965.
F. AND SEIDL, L. G.:
Dextrostix as a Quantitative Test for Glucose in Whole Blood
I. A. KRYNSKI, M.D. and J. E. LOGAN, Ph.D., Ottawa, Ont.
pregnated with enzyme reagents that gives
approximate results for glucose concentration
with one drop of whole capillary or venous
blood. The reagent area contains glucose oxi-
dase, peroxidase and a chromogen or indicator
system, which varies in colour and intensity
from shades of grey to deepening shades of
blue-purple depending on the amount of glucose
present. A reference colour chart is supplied on
the bottle label with values at 40, 65, 90, 130,
150, 200 and 250 mg. per 100 ml. The test,
which is based upon the enzymatic reaction of
glucose with glucose oxidase,
screening and is not intended to replace the
more precise analytical procedures. The manu-
facturer's statement is as follows: "Dextrostix is
specific for glucose; no other reducing sub-
stances or sugars will react with Dextrostix.
Thus, the test will correspond to other 'true
Fluorides inactivate glucose oxidase and blood
specimens containing fluorides will not give reli-
able results with Dextrostix. Oxalate, citrate or
heparin does not affect the test when used in
blood specimens in recommended amounts."
Several investigators have compared results
obtained with Dextrostix with glucose values
obtained by various methods.'-18 Most of them
reported discrepancies between the Dextrostix
test and the selected reference methods but did
is a firm cellulose strip im-
is designed for
From the Clinical Laboratories, Laboratory of Hygiene.
Department of National Health and Welfare. and the
Department of Medical Education and Research, Ottawa
Civic Hospital, Ottawa, Ont.
*Ames Co. of Canada, Toronto, Ontario.
Hygiene, Department of National Health and Welfare,
Ottawa 3. Ont.
Logan, Laboratory of
not always agree concerning the size and direc-
tion of the discrepancies. In part, this may be
due to the varying specificity of these methods
among the technicians comparing the strips with
the colour blocks. In our studies, the data were
compared with the automated ferricyanide pro-
cedure over the whole range and, in addition,
with an automated glucose oxidase method over
the lower part of the concentration range. All
Dextrostix readings were made by one of us
(I.A.K.) and spot confirmation made by the
other (J.E.L.). Both authors scored "superior" on
the Farnsworth-Munsell colour discrimination
Our observations with Dextrostix are discussed
in the light of the manufacturer's claims.
Two hundred and twelve specimens of ven-
ous blood were obtained from diabetic patients
in the Ottawa Civic Hospital. Fasting and post-
prandial specimens were used in order to check
the whole range of the Dextrostix chart. Results
were compared with those obtained by the
AutoAnalyzer ferricyanide method for the de-
termination of glucose.19 During the collection
of the venous blood specimens, one drop of
blood was applied to the sensitive end of the
test strip. After exactly 60 seconds' contact, the
blood was washed off under a gentle stream of
tap water and the colour change compared with
a graded colour chart. Twenty
blood from the same patient were collected in a
,ul. of venous
*Becton, Dickinson & Co., Rutherford, N.J.
Canad. Med. Ass. J.
Oct. 21, 1967, vol. 97
Krynski and Logan:
sodium fluoride. The diluted specimens
was determinedbyamicro-ferricyanide proce¬
dure similar to that describedby Gray,Stowe
and Holden.20 Theanalyseswerealwayscom¬
pletedon thedayof collection.
AutoAnalyzervalues in the lower partof the
scale (< 40-90),36specimensofsheepblood
and 20specimensof human blood were ana¬
lyzed byDextrostix andbyan automatedglu¬
method of Hill and Kessler.21 Nosignificantdif¬
ferences were found between Dextrostix read¬
ingsonsheepblood and those on human blood
when glucosewas added toyield a series of
concentrationsequivalentto those used in the
colour block series. It seemed feasible then to
use sheepblood to check Dextrostixperform¬
ance in the lower concentrationranges.
Glucoserecovery experiments usingDextro¬
stix were carried outby adding dry glucoseto
Chemvarion,f an ion-free serumcontainingno
glucose,in amounts to yieldconcentrations of
50, 100, 150,200 and 250mg. per100 ml.
Dextrostix andAutoAnalyzerresults were, in
mostcases, based onduplicatedeterminations.
to the AutoAnalyzer and glucose
a modification of the
Table IcomparesDextrostix andAutoAnalyzer
arbitrary rangeof values was setupfor each
concentrationrepresentedin the manufacturer's
colour scale. Therangefor aparticularconcen¬
tration on the scale, e.g. 65, includes all values
from themidpointbetween 40 and 65 to the
midpointbetween 65 and 90. For colours ob¬
tained with Dextrostix which did not match
exactlythatgivenon thechart, e.g.40 to65,the
midpoint concentration, i.e. 52.5, was used to
calculate thearbitrary range (46to59).In our
evaluation, we designated
Dextrostix readings where the parallel Auto¬
Analyzer values fell into the correct arbitrary
readings (49.6%) were confirmedbythe Auto¬
Analyzerresults. One hundred and seven of 212
Dextrostixreadingswere not confirmedbythe
yielded readingslower than those of the Auto¬
In view of the manufacturer's claim that 95%
of Dextrostix visualreadingsfail within ±30%
of the values recordedbytheAutoAnalyzer,the
AutoAnalyzerresults in this ±30% correlation
on 212 whole blood specimens. An
as "confirmed" all
It will be noted that 105 out of 212
were compared with the
TABLE I..Comparison of Dextrostix Readings with
AutoAnalyzer Values (Macro Procedure)
(Fig. 1).The allowablerangesfor thisfigure
have been established as follows: forexample,
a Dextrostixreadingof 90 would be 30% above
anAutoAnalyzervalue of 69, and 30% below
one of 129; thus the ±30% allowable range
becomes 69 to 129. For the "in-between" Dex¬
trostixreadings,themidpointof the Dextrostix
rangewas used to make a similar calculation.
Our results show that 87.3% of all the Dextro¬
stixreadingsfail within the ± 30% limits. In
17 of the 27 resultslyingoutside these limits,
the Dextrostixreadingswere low. Best results
were obtained in the 65 to90, 90,and 90 to 130
mg. % zones.
Table II shows
Dextrostixreadingson 78specimensand values
obtained with theAutoAnalyzer microprocedure
(ferricyanide), using the B.D. Unopettes for
specimencollection. In this comparison,31 of
a comparison between the
TABLE II..Comparison of Dextrostix Readings with
AutoAnalyzer Values (Micro Procedure)
tClinton Laboratories, Los Angeles, California.
1008 Krynski and Logan: Dextrostix
Canad. Med. Ass. J.
Oct. 21, 1967, vol. 97
40 40-65 65 65-90
Fig. 1..Block diagram shows the distribution of Dextrostix readings within and outside
30% limits of the AutoAnalyzer values.
90 90-130 130 130-150 150 150-200 200 2O0-2S0 250
Canad. Med. Ass. J.
Oct. 21, 1967,vol. 97
Krynski and Logan:
of Dextrostix Readings
with AutoAnalyzer Values
the 78 Dextrostix readings (39.7%)were con¬
firmedbytheAutoAnalyzer.Nineteen of the 47
We tried to assess the efficacyof Dextrostix
Chemvarion. In these serumsamples,the colour
responsewith Dextrostix was more bluish than
that indicated on the scale for whole blood.
However, on the basis of colour intensity,we
were able to match these colourresponsesfrom
the sera with the Dextrostix scale. Goodrecovery
was obtained at all levels of the scale from 65
to 250mg. %.
Table III comparesDextrostix and AutoAna¬
lyzer glucoseoxidasereadingson 36sheep-blood
specimensand on 20 human-blood specimens.
Nineteen of the 36 readings on sheepblood
were confirmed bythe automated enzyme
sults. Of the 17 readingsthat were not con¬
higherthan the AutoAnalyzer.When the same
Dextrostixreadingsare evaluated in terms of a
± 30% AutoAnalyzer correlation, 81% of the
readingsfail within these limits. Five of the
high.With human blood, five of 20 Dextrostix
readingswere confirmed bythe AutoAnalyzer.
Of the 15readingsthat were not confirmed,14
werehigher byDextrostix. When the same Dex¬
trostix readingswere evaluated in terms of a
± 30% correlation, 75%
within these limits. All five of thereadingsout¬
side therangewere toohigh.
A careful examination of the Dextrostixdupli¬
catereadingsindicated that there was no varia¬
outside the range
of the results
tion between individual strips.These findings
indicate that the test hashigh precision.
Glucose values obtained with Dextrostix often
do notagreewell with those obtainedbythe
automatedferricyanide procedure.The Dextro¬
stixreadingis often lower than theAutoAnalyzer
value, especially in the hypoglycemic region
(< 40 to 40-65mg. per 100 ml.) and the
hyperglycemic region (200to > 250).A num¬
ber of workers46 havepreviously reportedthat
Dextrostix yields low values when compared
with the automatedferricyanidemethod. In a
smaller number ofspecimens,collected in Uno¬
pettes containingfluoride and measured by a
micro modification of the automated method,
thepercentageof lowreadingswas somewhat
less butagainthe correlation was notgood.
Because low readings by Dextrostix might
contain "normal" concentrations by the ferri¬
cyanide procedure,the low bloodglucosecon¬
method. In thiscomparison,theagreementwas
better in thehypoglycemic region (16of 29 con¬
firmed), and 11 of the 13discrepant readings
were nowhigh.In thisregionitappearsthat
Dextrostrixyields highervalues forglucosethan
the automatedglucose oxidaseprocedure, but
lower values than the automatedferricyanide
procedure.On the Dextrostixscale,areadingof
should be checkedbya more sensitiveanalytical
method. Whencomparedwith theglucoseoxi¬
daseprocedure, Dextrostix wasnearly always
higherin the normalrangeof concentration(65
to 90 to130),butagreementwith theferricya¬
nide results was much better in thisregion.
Dextrostix should be valuable inmaking
rapid diagnosisin thosepatientswho come to
theemergency departmentin coma. In ourdata,
using Dextrostix, nospecimenwhichmightbe
considered hypoglycemic (<
yieldeda valuegreaterthan 65mg. % bythe
ferricyanide procedureorgreaterthan 59mg. %
bytheglucose oxidase method. In thehyper¬
glycemic regionwith Dextrostix(i.e. 200),the
lowest value obtainedbytheferricyanide pro¬
cedure was 160mg. %. Thus Dextrostix will
It has beensuggestedthat Dextrostixmight
be used in mass screeningfor diabetes. Since
Dextrostix isdesignedto detectglucosein blood
at levels below which
urine, it should detectearlydiabetes more fre-
were checked against values ob¬
or belowmay indicate hypoglycemia and
over into the
Krynski and Logan:
Canad. Med. Ass. J.
Oct. 21, 1967,vol. 97
quentlythanurine-screening methods. Packer,
Hawkes and Ackerman22 showed that a blood-
glucosetest(Clinitron)was more than twice as
cases of diabetes. On the basis of our data,
however, the accuracy
screeningisquestionable.Rock and Gerende16
few falsely high readings with normal speci¬
mens, but more serious was thegreaternumber
of borderlinehigh specimensthat weredesig¬
nated as normalby the Dextrostix test. In a
would not besubjectedtoany confirmatorytest
and thus would be missed.
In astudyto determine theaccuracyof Dex¬
trostix in mass screening for diabetes, West,
Stein and Sanders17compared Dextrostix esti¬
mates of bloodglucosewith those of the Auto¬
Analyzer ferricyanidemethod.Theydivided the
AutoAnalyzervalues into threegroups: < 150
mg. %, negative;150 to 199mg. %, borderline;
> 199mg. %, positive.On this basistheyfound
that Dextrostixyieldedfalsepositivevalues in
three of 396 venous bloodspecimens and false
negatives in six of 107specimens. They also
classified as borderline 7% of thenegativeand
21% of thepositive AutoAnalyzerresults. When
weanalyzedour data in thisway,we found no
falsepositivevalues in 116 and one falsenega¬
tive in 58 specimens.
values were recorded in nine of the 116speci¬
mens that had beennegativeand 13 of 58 that
had been positive on the AutoAnalyzer. Our
results are similar to those of West and his col¬
leagues,and indicate that Dextrostix underesti-
mateshigher glucose values. West, Stein and
Sanders17 have recommended that Dextrostix be
used to screenspecimenstaken one hour after
the glucose load was administered and that
borderlinespecimensbe followedup byaquan¬
titative method. The criticalpointseems to be
the level ofaccuracyneeded for suchscreening.
They appear readytoaccept the loss ofap¬
proximately 6% of thepopulationwhosespeci¬
mens were "falsenegative"with Dextrostix. In
their studies, Hollister, Helmke andWright13
specimensin the 130 to 200mg. % range.
It is ouropinion, however,that whenchoosing
ascreening test,this level ofinaccuracyand the
additional uncertainty arising from individual
variations in colour discrimination isunaccept¬
able. A better choice is aprocedure,such as that
ofGray,Stowe andHolden,20in which blood is
collected in aUnopetteforanalysislaterbythe
AutoAnalyzer.a method that
precise and accurate. In ascreening program,
as urine tests inscreening for new
of this procedure
it unsatisfactory. We encountered
to 10% low Dextrostixreadings for
Kent and Leonards23 submitted allsubjectshav¬
ing capillarybloodglucose levels higherthan
120mg. per100 ml.(asdeterminedbyDextro¬
comparative study where thepaper stripand
ouslyrun on each of 1900specimens, theyfound
thatonlyonespecimenread under 120mg. %
byDextrostix and over 140mg. % bythe auto¬
matedprocedure. In our data, 2 of 57speci¬
mens, which were 150mg. % bythe AutoAna¬
lyzer,were 90 and 90-130byDextrostix.
Ourfindingsalso indicate that the Dextrostix
test should not be used in themanagement of
the diabetic. The
changesinglucose concentration, but where an
accurate measurement ofglucoselevels is needed
toregulate insulindosage,themarginof error
with Dextrostix isprobably toogreat. Almost
12% of Dextrostixreadingsin ourstudydiffered
values. These resultsagreewell with those re¬
ported by Chryssochoosand Weisenfeld.8
Thephysiciancan use Dextrostix in his office
to discover diabetes in those in whom the blood
sugar level is elevated but has not reached a
level where itspillsover into the urine. How¬
ever, anegativeresult in this situation does not
rule out thepresenceof diabetes.
to an automated glucose analysis. In
test would identify large
more than 30% from the AutoAnalyzer
Summary W^T ^ex^ostixreadillgs
whole blood specimens
pared with values determinedby automated ferri¬
cyanide and glucose
one-half of thereadingsdiffered from the automated
values. Compared to theferricyanide method, the
Dextrostixreadings were more oftenlow, especially
at the lower andhigher concentration levels of the
colour chart. When compared with the glucose
oxidase method (a morespecific determination) in
the lower concentrationrange, Dextrostixreadings
were higherin most instances. Dextrostixprovides
hyperglycemia, but it is not accurateenough to be
used for thequantitative determination ofglucose.
means ofidentifying hypoglycemia and
total avec les valeurs obtenues par les methodes
duferricyanure et duglucose oxydase apermisde
constater qu'au moins la moitie des lectures du
methodes. Parrapportaux chiffres de la methode au
ferricyanure, les lectures du Dextrostix etaientplus
souventinferieures, sp6cialement aux deux concen¬
trations inferieures etsuperieures de la charte des
couleurs. Par rapport aux chiffres donnes par
methode duglucose oxydase (methode plus specifi¬
que) dans lagammedes concentrations inferieures,
les lectures du test de Dextrostix etaientsuperieures
^acomParalsondes lectures du test de
Dextrostix sur 212 echantillons desang
Canad. Med. Ass. J.
Oct. 21, 1967, vol. 97
RESNICK AND JOUBERT: TREATMENT OF OBESITY
dans la plupart de case. Le Dextrostix constitue un
moyen rapide de deceler l'hypoglycemie et l'hyper-
analyse quantitative du glucose.
il n'est pas assez precis pour une
The authors wish to thank Dr. J. B. R. McKendry,
Mrs. B. Burpee, and the staff of the Diabetes Clinic,
Ottawa Civic Hospital, and also Dr. A. W. Jackson of
this laboratory, for their co-operation in providing blood
specimens for this study. We also are indebted to Dr.
D. A. Kavelman, Ames Co. of Canada, Ltd., for a gener-
ous supply of Dextrostix.
2. MACKAY, N., GORDON, A. AND NEILSON, J. M.: Ibid.,
2: 269. 1965.
3. MARKS, V. AND DAWSON, A.: Brit. Med.
4. ALBERTI, K. G. M. M., MIDDLETON, G. G. AND CAIRD,
F. I.: Lancet, 2: 319. 1965.
5. HERSCHBERG, S.: Ibid., 2: 902, 1965.
2: 884, 1964.
I. D. B., KEEN, H. AND SOUTHON, A.: Ibid.,
R. M. AND JEPSON, E. M.:
195: 643, 1965.
8. CHRYSSOCHOOS, T. J. AND WEISENFELD, S.: Diabetes,
14: 452, 1965 (abstract).
9. SHIPP, J.
C. AND PAWLIGER, D. F.: Lancet, 2:
10. BECKETT, A. G. AND COOK, I. J. Y.: Ibid., 2: 591, 1965.
11. SMITH, A. W. M. AND MCILWAINE,
Med. J., 1: 661, 1965.
Diabetes, 15: 691, 1966.
14. SCHERSTEN, B.: Acta Med. Scand., 178: 583, 1965.
15. TURPIN. B.
C. AND DOWNEY, P. G.: Amer.
Tech., 32: 327, 1966.
16. ROCK, J. A. AND GERENDE, L. J.: J. A. M. A., 198: 231,
Amnier. J. Public Health, 56: 2059, 1966.
18. CIAmPI, G.:
e di Laboratorio, 1: 364, 1965.
19. "Glucose" Technicon Laboratory Method File No. N-
2a, Technicon Instrument Corporation, Chauncey,
STOWE, H. W. AND HOLDEN,
Public Health Rep., 79: 1081, 1964.
21. HILL, J. B. AND KESSLER, G.: J. Lab. Clin. Med., 57:
22. PACKER, H., HAWKES, J. M. AND ACKERMAN, R. F.:
Diabetes, 10: 280, 1961.
23. KENT, G. T. AND LEONARDS, J. R.: Diabetes, 14: 295,
I. V.: Lancet, 2: 1062, 1964.
A Double-Blind Evaluation of an Anorexiant, a Placebo and
Diet Alone in Obese Subjects
MORRIS RESNICK, M.D.,* Ottawa, and
LUCIEN JOUBERT, M.D., M.Sc.,t Montreal
A NUMBER of controlled studies of the value
of various anorexiants in weight reduction
have been published.1-4 Very few, however,
include diet alone as a comparative regimen.
Some clinicians object to the use of anorexi-
genic agents on the grounds that overweight
persons would lose as much weight on a low-
calorie diet alone. This study sought to chal-
lenge this assumption and to compare a popular
anorexiant which is a combination of dextro-
amphetamine sulfate and prochlorperazine di-
maleate (Eskatrol; Smith Kline & French) with
placebo and with diet alone.
Ninety obese female students from a general
hospital school of nursing were included in the
study following their annual physical examina-
tion. Criteria for inclusion were: body weight
10% or more in excess of ideal weight, as speci-
fied in a modified Metropolitan Life Insurance
Company table (Table I); no attempt to lose
*Director, Nurses Health Service, Ottawa Civic Hospital,
Pharmacologist and Research
Reprint requests to: Dr. Morris Resnick, Suite 307, 150
Metcalfe Street, Ottawa 4, Ontario.
TABLE 1.-DESIRABLE WEIGHTS FOR WOMEN
Weights in Pounds (in indoor clothing, but without shoes)
Height (iv-thout shoes)
weight within the previous three months; and
absence of any organic disorder (besides excess
weight) or any contraindication to the drug
At the time of her examination, each subject's
height and actual weight (to the nearest quarter
pound) were measured. Excess weight was cal-
culated using the following formula:
Actual weight - Ideal weight