Reliability of bedside blood glucose estimating methods in detecting hypoglycaemia in the childrens emergency room.html

Niger J Paed 2015; 42 (1): 34 – 38

ORIGINAL

Oyenusi EE

Reliability of bedside blood

Oduwole AO

Oladipo OO

glucose estimating methods in

Njokanma OF

detecting hypoglycaemia in the

Esezobor CI

children’s emergency room

DOI:http://dx.doi.org/10.4314/njp.v42i1,8

Accepted: 17th October 2014

Abstract: Background: Hypogly-

0.05) between Accuchek Active®

caemia occurs in many disease

results and laboratory values than

Oyenusi EE

(

)

states common in the tropics. Fa-

was obtained with Betachek Vis-

P.O.Box 4337, Ikeja,

cilities and skilled manpower re-

ual®(r = 0.48, p=0.000).In detec-

Lagos, Nigeria.

quired for laboratory blood glu-

tion of hypoglycaemia, both bed-

Email: ebikike@yahoo.com

cose measurement are not always

side glucose monitors were found

Oduwole AO, Esezobor CI

available in health facilities in

to have a high specificity and high

Department of Paediatrics,

developing countries.

predictive values of a negative test

Lagos University Teaching Hospital,

Objective: The study was carried

(99.8% and 98.5% for Accuchek

Idi-Araba, Lagos, Nigeria.

out to determine the validity of

Active®and 89.4% and 97.8% for

bedside methods of blood glucose

Betachek Visual®respectively)

Oladipo OO

measurement in detecting hypo-

with moderate sensitivity (75.0%

Staten Island University Hospital,

glycaemia.

and 66.7% respectively). However,

Staten Island. NY. USA

Methods: Blood glucose was de-

the Accuchek Active® monitor has

termined by two bedside methods

a much higher predictive value of

Njokanma OF

( Accuchek Active® and Betachek

a positive test (94.1%) compared

Department of Paediatrics,

Visual®) in 430 patients aged

to the Betachek Visual® (27.1%).

Lagos State University Teaching

Hospital, Ikeja, Lagos, Nigeria.

between one month and 10years

Conclusion: The bedside glucose

and simultaneously sent for labo-

monitors are valid bedside tools

ratory spectrophotometric analysis

for detecting or ruling out hypo-

at a wavelength of 500nm using

glycemia.

the hexose kinase method. Hypo-

glycaemia was defined as plasma

Keywords: hypoglycaemia, chil-

glucose < 2.5mmol/L.

dren, bedside investigation, glu-

Results: The prevalence of hypo-

cose estimation, blood glucose.

glycaemia was 5.6%.There was a

higher correlation (r =0.84, p<

Introduction

ever, the absence of clinical symptoms does not always

indicate that the glucose concentration is normal because

hypoglycaemia may be frequently asymptomatic . Thus

Hypoglycaemia occurs in many disease states such as

severe malaria, severe malnutrition, diarrhoea amongst

it is important to investigate for hypoglycaemia in

others common in the tropics

1-4

and may also complicate

acutely ill children, more so as it complicates paediatric

treatment with some drugs like quinine used in the treat-

emergencies in Africa

9-11

ment of severe malaria . A previous study had docu-

mented a prevalence rate of 17% among children with

The gold standard technique of determining blood glu-

severe malaria seen at the children’s emergency centre

cose level involves laboratory spectrophotometric analy-

of the Lagos University Teaching Hospital (LUTH) .

sis using methods involving enzymes such as glucose

Significant morbidity and mortality can be associated

oxidase, hexokinase and glucose dehydrogenase. These

with hypoglycaemia . Survivors of prolonged hypogly-

6,7

methods require standard laboratory equipment, skilled

caemia are prone to neurological complications such as

manpower, consistent electricity and water supply.

mental retardation, recurrent seizure activity, transient

These resources required for the laboratory determina-

cognitive impairment and neurological deficits .

6,7

tion of blood glucose level are not always available in

our health facilities in the developing countries, yet early

Clinical manifestations of hypoglycaemia include sweat-

diagnosis and prompt treatment of hypoglycaemia in

ing, jitteriness, anxiety and nervousness. Others are

critically-ill children minimizes hypoglycaemia-related

organ damage .

6,7

headaches, visual disturbances, lethargy, restlessness,

irritability, convulsions and mental confusion . How-

In addition, the cost of laboratory determination of

blood was wiped from the test pads with a folded tissue

blood glucose level could be high when serial measures

paper and repeated using a clean region of the tissue

are required. Because of relatively long turn-around time

paper. It was ensured that strip was blood-free, as excess

blood left on the pads usually affects the result . After

results from laboratory determination of glucose may

not be available in a timely fashion to affect manage-

an additional 30 seconds, the test result was read by

ment of a child. To overcome some of these challenges,

comparing the reacted test pad to the B etachek Visual®

various types of reagent test strips have been developed

colour chart label. If the colour fell between two colour

blocks, the average value was taken as the test result .

for bedside usage. An added advantage of these rapid

bedside diagnostic tests is that less expertise/training is

required prior to their usage. These strips use a method

For the Accuchek Active® the test strip was held so that

in which a dye is coloured by the glucose oxidase-

the application area and arrows were facing upwards. In

peroxidase chromogenic reaction

12,13

. The resulting col-

this position the test strip was pushed into the glucome-

ter until it clicked into place . The insertion of the test

our change is either visually interpreted by comparison

with an accompanying colour chart or read with a reflec-

strip automatically turned the monitor on and put it in

test mode . When the drop symbol flashed in the dis-

tance meter (glucometer)

14-15

Therefore, it is imperative to validate these bedside

play, a drop of blood was applied to the middle of the

methods for reliability of results obtained from them.

orange coloured test pad. The test was complete after

Hence, this study was conducted to determine the sensi-

about five seconds and the result appeared in the display

tivity, specificity and predictive values of two bedside

and was automatically saved with the date and time. The

methods of glucose determination using a laboratory-

Accuchek Active®monitor measures blood glucose

based enzymatic method (the hexokinase method) as the

within a defined range (10-600 mg/dl or 0.6-33.3 mmol/

gold standard.

L). Values outside this range are displayed as “lo”

meaning the result is less than 0.6 mmol/L or “Hi”

meaning the result is greater than 33.3 mmol/L .

The blood in the fluoride oxalate-containing bottle was

Subjects and methods

sent to the Chemical Pathology Department of the hospi-

tal for plasma glucose determination. The blood sample

The study was conducted at the Olikoye Ransome-Kuti

was centrifuged for five minutes at about 1600Gy to

obtain plasma not more than one hour after collection .

Children’s Emergency Centre of the (LUTH), Lagos

The separated plasma was frozen at −80 C and stored in

Nigeria. All consecutively admitted children, aged be-

tween one month and 10 years over a period of nine

the refrigerator until analysis. Samples were analysed in

months (September 2009 to May 2010) were eligible for

batches of 25 samples using the Hitachi Automatic Ana-

inclusion in the study. Approval was obtained from the

lyser (Cobas- ROCHE). The analyser uses spectropho-

Human Research and Ethics Committee of LUTH before

tometric analysis at a wavelength of 500nm and hexose

kinase method . Hypoglycaemia was defined as whole

commencement of the study. Written informed consent

was also obtained from all parents or care –givers of

blood glucose less than 2.2 mmol/L for the bedside

participating children.

methods, ( Betachek Visual® Accuchek Active® ) and as

plasma glucose less than 2.5 mmol/L from the sample

analysed in the laboratory . This was based on the fact

Data Collection

that whole blood glucose concentration is approximately

For each subject recruited, at the point of admission into

8 to 15% less than plasma glucose because the water

the emergency room, before commencement of intrave-

content of plasma is approximately 12% higher than that

of whole blood .

nous fluids, a drop of capillary blood obtained by a fin-

ger prick was applied to each of the two reagent strips

according to the manufacturer’s instructions

16,17

and

Data management and analysis

about 1.5mls of blood was withdrawn from a convenient

peripheral vein and put in a fluoride oxalate-containing

The data was analysed using the Statistical Package for

bottle for laboratory spectrophotometric analysis using

the Social Sciences (SPSS) version 20. Measures of

the hexose kinase method.

statistical location (mean, standard deviation, median

and range) were determined. Probability (p value) less

A visually interpreted test strip by Betachek Visual and

than 0.05 was taken as statistically significant. The sen-

a reflectance meter, Accuchek Active were used for the

sitivity, specificity, positive and negative predictive val-

bedside determination of blood glucose because these

ues of blood glucose obtained from the two reagents

are readily available and commonly used in our facili-

strips were calculated using the blood glucose from the

laboratory as the gold standard . For the purposes of

ties. The test pad of the Betachek Visual

contains

0.09% glucose oxidase, 0.19% peroxidase, 1.2% TMB

this study, the values of sensitivity, specificity, positive

(3, 3', 5, 5'-Tetramethylbenzidine dihydrochloride) and

and negative predictive values of blood glucose were

0.35% DCP (2,4-dichlorophenol 6-monooxygenase) .

classified as low if between 0 to <50%, moderate if be-

A test strip was removed and the bottle was immediately

tween 50-80% and high if from >80 to 100%.

recapped. A drop of blood was pressed onto the centre

of two test pads, moving it around to ensure complete

coverage. Exactly 30 seconds after application, the

Results

diagnostic accuracy were calculated for results obtained

using the bedside blood glucose monitors (Table 2 for

Four hundred and sixty one children aged 1.5 to 120

Betachek Visual®strips and Table 3 for Accuchek Ac-

months were admitted over a six-month period. Of the

tive®meter). With respect to Betachek Visual®strips,

461 patients, 31 laboratory blood glucose samples

the sensitivity was 66.7% and the specificity was 89.4%.

obtained were not valid due to improper handling. Four

The positive predictive value and the negative predictive

hundred and thirty patients had complete data sets and

value were 27.1% and 97.8% respectively.

were analysed.

Table 2: Sensitivity and specificity of the Betachek Visual®

Blood glucose (BG) patterns of the patients

strip, Laboratory based method

Test

Positive

Negative

Total

The descriptive statistics of blood glucose results by the

Betachek Positive

True Positives = 16 False Positives = 43

various methods is shown in Table 1. Using the labora-

Visual ®

tory based method, the mean BG of all the study sub-

Negative False negatives= 8 True Negatives= 363

371

jects was 5.19 ± 2.05 mmol/L (range of 0.1-18.0 mmol/

Total

406

430

Sensitivity = TP / (TP + FN) = 16 / (16+8) ×100 = 66.7%

L, median was 4.9 mmol/L).

Specificity = TN / (TN + FP) = 363 / (363+43) ×100 = 89.4%

Positive Predictive Value = TP /(TP+FP)=16/(16+43)×100= 27.1%

Table 1: Descriptive statistics of the blood glucose results by

Negative Predictive Value = TN/(FN+TN)=363/(8+363)×100 = 97.8%

the various methods

Test

Range

Mean

Median

Total

The sensitivity, specificity and the positive and negative

Method

(mmol/L) (mmol/L) (mmol/L) (mmol/L)

predictive values of the Accuchek Active® meterstrips

Laboratory

0.1− 17.2 5.19

2.05

4.9

430

are shown in Table 3. Measures of diagnostic accuracy

Accuchek Active® 0.5 −17.1 5.69

2.04

5.6

430

were calculated for results obtained using Accuchek Ac-

Betachek Visual® 0.0 −18.0 5.54

2.83

6.0

430

tive®meterstrips. The sensitivity was 75.0% while the

specificity was 99.8%.The positive predictive value was

The scatter diagram (Fig 1) demonstrates the relation-

94.7% while the negative predictive value was

ship between laboratory blood glucose and blood glu-

98.7%.Four out of the eight false negatives had blood

cose determined by Accuchek Active® and by Betachek

glucose values ranging from 2.7 to 2.9 mmol/L.

Visual®.Visual inspection further indicates a better

positive relationship between BGs determined by labo-

Table 3: Sensitivity and specificity of the Accuchek Active®

ratory and Accuchek Active® than between laboratory

glucometer, Laboratory based method

and Betachek Visual® .

Test

Positive

Negative

Total

The blood glucose values measured by the Accuchek

Accuchek Positive True Positives =18

False Positives =1

Active®meter correlated better with the values from the

Active®

laboratory than those measured by the Betachek Visual®

Negative

False negatives =6 True Negatives = 405 411

strip ( with R, 0.84 versus 0.48) and the difference was

Total

406

430

statistically significant (p < 0.05).

Sensitivity= 18/ (18+6) ×100 = 75.0%

Specificity = 405/ (405+1) ×100 = 99.8%

Positive Predictive Value = 18/ (18+1) ×100 = 94.7%

Fig 1: Scatter diagram showing the relationship between Labo-

Negative Predictive Value = 405/ (6+405) ×100 = 98.5%

ratory blood glucose and blood glucose determined by

Accuchek Active®and by Betachek Visual®

Discussion

The blood glucose values measured by the Accuchek

Active®meter correlated better with the values from the

laboratory than those measured by the Betachek Visual®

strip. The chances are very high that an increase/

decrease in laboratory result will be reflected by an in-

crease/decrease in Accuchek Active® meter result.

Correlation of Betachek Visual with laboratory blood glucose, r =

0.48, p<0.05

The specificity and negative predictive value of the Be-

Correlation of Accuchek Active with laboratory blood glucose, r=

tachek Visual®strip were high, in the range of 90% and

0.84, p<0.05

above. Sensitivity on the other hand was moderate but

Hypoglycaemic patients

positive predictive value was low. This implies that the

use of this technique as a screening tool confers moder-

Twenty-four out of the 430 study subjects had hypogly-

ate success. However, the high specificity and a high

caemia by the laboratory method giving a prevalence

predictive value of a negative test means that a

nega-

rate of hypoglycaemia of 5.6%.

tive result can be relied upon as being truly negative.

Sensitivity and specificity of bedside glucose monitor

The high specificity and negative predictive value of the

for the detection of hypoglycaemia: Using the laboratory

Betachek Visual®strip is comparable to a specificity of

values of blood glucose as gold standard, measures of

94% and negative predictive value of 99% of a different

brand

visually-interpreted

reagent

strip

( ChemstripbG ) in a previous Cincinnati

study. How-

Active®was higher than the 63.1% of the Prestige IQ

ever, the sensitivity and the positive predictive value of

meter. Also, in comparison with findings in a Kenyan

study , the specificity of the Accuchek Active® meter in

the Betachek Visual® strip used in the current study was

lower than the reported sensitivity and positive predic-

the current study was higher than the specificity of 48%

tive value of the ChemstripbG strip which was 97% and

but the sensitivity was lower than 84% of the Refloflux-S

72% respectively . A possible explanation for the dis-

reflectance meter respectively.

parity may be the definition of hypoglycaemia as labora-

tory blood glucose value of <45 mg/dl (2.5mol/L) in the

The different bedside equipment for measuring blood

current study compared to the higher laboratory blood

glucose are associated with different degrees of success

glucose value of <60mg/dl (3.3mol/L) defined as hypo-

with identifying true positives and true negatives as well

glycaemia in the Cincinnati study . Another possible

as avoiding false positives and false negatives. How-

reason for the low sensitivity of the Betachek Visual®

ever, the advantages they offer in terms of low cost,

may be because it is semi-quantitative and measures

immediate availability of results and the option of

blood glucose to the nearest 1mmol/L(18mg/dl) com-

immediate intervention make them very attractive. There

pared to the standard laboratory method which is more

will, however, remain a need for laboratory determina-

quantitative measuring to the nearest 0.1mmol/L(1.8mg/

tion of values, especially in cases where the results are

dl). Furthermore, the visual reading with a colour chart

borderline.

used in the visually interpreted strips may introduce in-

ter-observer bias as people may interprete colour inten-

sity differently while visual acuity and colour blindness

may be sources of bias. However, from the results in the

current study, the Betachek Visual® strip may be more

Conclusion

useful as a confirmatory rather than a screening test for

hypoglycaemia.

In conclusion, the glucose meter, Accuchek Active® and

Betachek visual®are reliable in detecting or ruling out

The specificity and negative predictive value of the

hypoglycaemia by the bedside which will enable prompt

Accuchek Active®were nearly 100%. The positive

intervention while awaiting confirmatory results from

predictive value was also very high but the sensitivity

the laboratory where these are available.

was only moderate. Sensitivity was hampered by the

occurrence of false negative results in six cases but it is

Author's Contributions

noteworthy that four of the six false negative blood glu-

The research work was conceived by all the authors.

cose values were in the lower range of normal. The im-

OEE carried out the field work. All the authors contrib-

plication is that blood glucose values in the lower band

uted to the writing of the manuscript.

of normal may require confirmation using laboratory

Conflict of interest: None

techniques. The high specificity and negative predictive

Funding: None

value imply that the Accuchek Active® meter is very

useful for ruling out hypoglycaemia and a positive value

can be relied upon as being truly hypoglycaemic.

Acknowledgements

The high specificity and negative predictive value of the

Accuchek Active®meter is similar to a specificity of

The authors hereby acknowledge the children and

96.2% and a negative predictive value of 99.7%

parents who participated in this study. We also want to

obtained with a different brand of a blood glucose kit

appreciate all the registrars in the children’s emergency

also interpreted by a reflectance meter ( Prestige IQ ) as

room of the Lagos University Teaching Hospital and the

documented in a previous Nigerian study . However,

other research assistants who contributed to the work.

the sensitivity of the Accuchek Active® meter was lower

We also acknowledge the laboratory scientist, Mr Tony

than the sensitivity of 96% of the Prestige IQ meter

Ani for his assistance and guidance in the laboratory

while the positive predictive value of the Accuchek

analysis.

References

Elusiyan JBE, Adejuyigbe EA,

Nwosu SU, Lesi FEA, Mafe AG,

Phillips RE, Looareesuwan S,

Adeodu OO. Hypoglycaemia in a

Egri-Okwaji MTC. Hypoglycae-

White NJ, Chanthavanich P, Karb-

Nigerian paediatric emergency

mia in children with cerebral ma-

wang J, Supanaranond Wet al.

ward. J Trop Paediatr 2006; 25:96

laria seen at the Lagos University

Hypoglycaemia and antimalarial

-102.

Teaching Hospital: predisposing

drugs: quinidine and release of

Osier FHA, Berkley JA, Ross A,

factors. Nig Med J 2004; 45:9-13 .

insulin. BMJ 1986; 292:1319-21.

Sanderson F, Mohammed S, New-

Onyiriuka AN, Awaebe PO,

Idro R, Carter JA, Fegan G,

ton CRJC. Abnormal glucose

Kouyaté M. Hypoglycaemia at

Neville BGR, Newton CRJC. Risk

concentrations on admission to a

point of hospital admission of

factors for persisting neurological

rural Kenyan district hospital:

under-five children with acute

and cognitive impairments follow-

prevalence and outcome. Arch Dis

diarrhoea: prevalence and risk

ing cerebral malaria. Arch Dis

Child 2003; 88:621-5.

factors. Niger J Paed 2013; 40:

Child 2006; 91:142-8.

384-8.

16. Betachek Visual test strips

Menni F, de Lonlay P, Sevin C,

11. Duru C, Peterside O, Akinbami

Touati G, Peigne C, Barbier V et

F.Pattern and outcome of admis-

[product information] Sidney Aus-

al.Neurological outcomes of 90

sions as seen in the paediatric

tralia: National Diagnostic Prod-

neonates and infants with persis-

emergency ward of the Niger Delta

ucts (NBP) [www.betachek.com];

tent hyperinsulinaemichypogly-

University Teaching Hospital

2006.

17. Accuchek Active blood glucose

caemia. Pediatrics 2001; 107:476-

Bayelsa State, Nigeria. Niger J

Paed 2013; 40:232-7.

monitor [user’s manual] Mann-

Thornton TS, Finegold DN,

12. Sacks DB. Carbohydrates. In: Bur-

heim (Germany): Roche Diagnos-

Stanley CA, Sperling MA. Hypo-

tis CA, Ashwood ER (editors).

tics; 2003.

glycaemia in the infant and child.

Tietz Fundamentals of Clinical

18. Cheesebrough M. District Labora-

Chemistry. 5 edition. London:

In: Sperling MA (editor) Paediat-

tory Practice in Tropical Countries,

ric Endocrinology. 2 edn. Phila-

WB Saunders; 2001.pp.776-85.

Part 2. Cambridge: Cambridge

delphia: Saunders; 2002.pp.135-

13. Cheesebrough M. District Labora-

University Press; 2000.pp.299.

59.

tory Practice in Tropical Countries,

19. Oyediran ABOO. Screening and

Diakparomre MA, Obi JO. The

Part 2. Cambridge: Cambridge

validity in clinical research. In:

pattern of paediatric emergencies

University Press; 2000.pp.341-3.

Bankole MA (editor). Handbook of

in the University of Benin Teach-

14. Scott PA, Wolf LR, Spadafora MP.

Research Methods for Doctors and

ing Hospital. Niger J Paed 1980;

Accuracy of reagent strips in de-

Dentists. Lagos: National Post-

7:43-5.

tecting hypoglycemia in the emer-

graduate Medical College of Nige-

10. Oviawe O. Comparative study of

gency department. Ann Emerg

ria; 1989.pp.384-9.

the pattern and severity of child-

Med 1998; 32:305-9

20. English M, Wale S, Binns G,

hood diseases seen at the chil-

15. Elusiyan JBE, Adeodu OO, Ade-

Mwangi I, Sauerwein H, Marsh K.

dren’s emergency room of the

juyigbe EA. Evaluating the valid-

Hypoglycaemia on and after ad-

Lagos University Teaching Hospi-

ity of a bedside method of detect-

mission in Kenyan children with

tal in 1967 and in 1982. Niger J

ing hypoglycaemia in children.

severe malaria. Q J Med 1988;

Paed 1987; 14:41-4.

Pediatr Emerg Care 2006; 22:488-

91:191-7.

90.