Nigerian Journal of Paediatrics

Evaluation of the Bilicheck transcutaneous bilirubinometer in jaundiced Nigerian term and preterm neonates

Niger J Paediatr 2019; 46 (1):30 – 38

ORIGINAL

Akinbolagbe YO

CC – BY

Ezeaka VC

Evaluation of the Bilicheck®

Akinsulie AO

transcutaneous bilirubinometer

in jaundiced Nigerian term and

preterm neonates

DOI:http://dx.doi.org/10.4314/njp.v46i1.6

Accepted: 19th February 2018

Abstract : Background : Transcu-

diazo method.

taneous

bilirubinometry

has

Results : The mean (SD) difference

Akinbolagbe YO (

)

shown some promise as a safe,

between TcB and TSB was 1.5

Department of Paediatrics,

non-invasive method that corre-

(2.6) mg/dL (95% CI of 1.2 - 1.9

Lagos University Teaching

lates well with traditional labora-

mg/dL); p= 0.000. There was

Hospital, Idi-araba, PMB 12003.

tory methods of estimating serum

strong correlation between TcB

Lagos, Nigeria.

bilirubin. However there is a pau-

and TSB with a correlation coeffi-

Email: yesidekush@yahoo.com

city of studies done on neonates

cient (r) of 0.77 (p = 0.000). The

of African descent.

95% limits of agreement were be-

Ezeaka VC, Akinsulie AO

Aim : The aim of the study was to

tween -3.5mg/dL and 6.6mg/dL.

Department of Paediatrics,

evaluate the diagnostic perform-

There was poor correlation be-

College of Medicine, University of

ance of the Bilichek® transcuta-

tween TcB and TSB of r= 0.43 at

Lagos/ Lagos University Teaching

neous bilirubinometer in jaun-

high TSB levels >12mg/dl. The

Hospital, Idi-araba, PMB 12003.

diced Nigerian term and preterm

TcB cut-off that most accurately

Lagos, Nigeria

neonates.

predicted TSB was 11.7mg/dL.

Methods : This was a cross-

Conclusion :

Transcutaneous

sectional study involving 169

bilirubinometry is

a reliable

jaundiced preterm and term neo-

screening method for assessing

nates studied over a 4 month pe-

severity of hyperbilirubinaemia in

riod. A total of 200 transcutane-

black African neonates. However,

ous bilirubin (TcB) readings de-

due to the occurrence of wide dis-

termined using the Bilichek®

parities,

confirmatory

serum

device were obtained simultane-

bilirubin measurements should be

ously with total serum bilirubin

done for TcB values above

(TSB) readings determined by the

11.7mg/dL.

Introduction

and vitros), enzymatic methods and spectrophotometric

methods.

Because it is expensive, cumbersome and

Neonatal jaundice (NNJ) is one of the most common

highly technical, HPLC is rarely used in clinical prac-

tice.

presenting complaints in the newborn period, occurring

in 30-60% of term infants and almost all preterms. In

the United States of America, it affects about 60% of

The indirect methods are visual assessment, icterometer,

term babies and 80% of preterms. In Kenya, Africa,

and transcutaneous bilirubinometry which measure

prevalence is about 35%. In Nigeria, mostly from hos-

bilirubin in the skin as an estimate of total serum

pital based studies, the prevalence of neonatal jaundice

bilirubin. Studies have shown high correlation between

is between 16.9 and 45.6%

[4-8]

and as high as 71.2%

among preterms. Though most cases are self-limiting

the intensity of the yellow colour of the skin and serum

bilirubin levels.^14-16

Bilirubin estimation from the blood

and do not require treatment, NNJ can lead to serious

by various laboratory methods is currently the most ac-

and life-threatening complications like acute bilirubin

curate and widely used method, though fraught with its

encephalopathy (ABE), cerebral palsy and deafness.

own deficiencies.^17-19

They require specialized person-

nel and equipment which are not always readily avail-

The management of neonatal jaundice largely depends

able in resource poor countries. Repeated blood sam-

on accurate assessment of bilirubin levels. There are

pling is also often required with the attendant risk of

various methods of estimating the level of bilirubin in

infection to both patient and healthcare provider. This

the blood. The methods can be direct from invasive

can also lead to iatrogenic anaemia which is particularly

blood sampling or indirect, measured non-invasively via

an issue in preterm babies with their smaller blood vol-

the skin. The gold standard method is High Performance

umes. These limitations have stimulated a lot of research

Liquid Chromatography (HPLC), which can assay all

on transcutaneous bilirubinometry, which started gain-

forms of bilirubin, including protein-bound bilirubin.

ing some credence following the development of the

Other direct methods include coupling reactions (diazo

first transcutaneous bilirubinometer for clinical use in

Japan during the 1980s.

These earlier types such as

Materials and Methods

JM101® and JM102® though useful for screening and

Study site and source population

research purposes were found to be grossly inaccurate in

neonates with low gestational age, low birth weight,

The study location was at a tertiary health institution,

dark skin pigmentation and during phototherapy use.

15,20

Lagos University Teaching Hospital, in Lagos State.

Lagos is a metropolitan megacity in southwest Nigeria.

The Bilichek® ((Respironics inc., Murrysville, PA,

The hospital has about 761 beds, with an average of 170

USA) a non-invasive Bilirubin Analyzer,

is a multi-

babies delivered in the labour ward complex monthly.

wavelength transcutaneous bilirubinometer. It was de-

An approximate number of 30 and 33 cases of neonatal

veloped with refinement of the old technology, to coun-

jaundice are managed per month in the inborn NICU and

teract the earlier defects or inadequacies. It works by

Children’s Emergency Room/ out born NICU respec-

directing light into the skin of the neonate, analyzing the

tively with about 25% of them being preterm neonates.

spectrum of optical signals reflected from the neonate’s

Apparently healthy jaundiced neonates are also seen at

subcutaneous tissues and converting them to an electri-

the post-natal ward with their mothers or at the out-

cal signal by a photocell which is analyzed by a micro-

patient clinic on follow up after discharge from hospital.

processor to generate a serum bilirubin value. The major

Averages of 2500 babies are seen annually in these loca-

skin components which impact the spectral reflectance

tions at the hospital.

in neonates are: melanin, dermal maturity, haemoglobin

Ethical approval was obtained from the Health research

and bilirubin.

ethics committee of the Hospital. Written informed con-

sent was obtained from the parents of the participants.

The newer multi-wavelength meters such as Bilichek®,

No extra cost was incurred by the patients on account of

subtract the spectral contribution of the known skin

the study.

components and the bilirubin absorbance is thus quanti-

fied, with results displayed directly in either mg/dl or

Study design

μ mol/L.

Various studies done worldwide, have re-

ported the device to be quite accurate in estimating se-

This was a cross-sectional study among jaundiced neo-

rum bilirubin with good correlation, agreement, sensitiv-

nates involving clinical and laboratory data.

ity, specificity, and positive predictive value when com-

pared with various laboratory methods.

13,17,22,23

Its accu-

Selection and description of participants

racy has also been found not to be affected by skin col-

our, birth weight, gestational age, post-natal age

17,24,25

The study was carried out over a four-month period

and phototherapy, if measurement is taken on a patched

(August 2013 through November 2013). All jaundiced

Bhutani et al in the US, compared the

part of skin.

13,26

neonates for whom a serum bilirubin measurement was

Bilichek® device with HPLC and found it to be accurate

required were eligible for the study. Inclusion criteria

with a good correlation of r = 0.91, mean difference

were:

[HPLC-Transcutaneous bilirubin (TcB)] of 0.47mg/dl,

Postnatal age from birth to 28 days of life,

sensitivity of 100%, specificity of 88.1%, negative pre-

Gestational age of 27 to 41 weeks,

dictive value of 100%, positive predictive value of

Weight range between 1000 and 4995 grams.

32.8% and likelihood ratio of 8.43. However, this study

These criteria were selected according to the category of

had preponderance of Caucasian neonates who were

patients for which the manufacturers of the Bilichek®

mostly term and of normal birth weights.

device stated that its performance had been clinically

proven.

Exclusion criteria were:

There is a paucity of literature on studies evaluating the

utility of the Bilichek® device in Africa, with popula-

Exposure to phototherapy before recruitment (those

tions of predominantly black neonates. An earlier study

patients used for the study while receiving photo-

done in 2004 by Slusher et al in Jos, and a recent one

therapy, had an area of forehead patched before

by Olusanya et al

in Lagos, both in Nigeria found

commencement),

good correlation between the device and serum

Undergoing an exchange blood transfusion,

bilirubin. However they also both obtained wide dispari-

Skin lesion on measurement site (forehead) such as

ties between both methods, with overestimation by the

bruising, nevus, haemangioma, haematomas, birth-

Bilichek® device, compared with studies done in Cau-

marks and excessive hairiness.

casian subjects. Rylance and colleagues in Malawi also

documented similar findings of good correlation. Some

The sample size was determined using the formula for

difference of means.

[29]

of these African studies excluded patients with certain

From previous study by Leite et

characteristics like pre-terms, sick neonates, out-born

al: difference between mean transcutaneous bilirubin

neonates and those receiving phototherapy. This study

(TcB) and total serum bilirubin (TSB) = 0.72 mg/dL and

therefore, aimed to fill some gaps in knowledge about

standard deviation of the difference between mean TcB

[23]

the diagnostic performance of the Bilichek® transcuta-

and TSB = 1.57.

By substituting in the formula, using

neous bilirubinometer, among jaundiced neonates of

a power of 80%, the minimum sample size was calcu-

African descent, in a typical hospital setting.

lated to be 154.8; this was rounded off to 155 paired

readings. A total of 200 paired readings were done to

account for incomplete data, invalid measurements or

lysed blood samples. Paired reading refers to measure-

and subsequent TcB measurements for them were done

ment of different variables (TcB and TSB) on the same

at the patched area. All transcutaneous measurements

subject done simultaneously or within a short interval of

were performed by the same researcher (AY).

each other. Patient selection was consecutive until sam-

ple size of number of paired readings calculated was

Data Analysis

reached. Multiple paired readings were done on patients

who required repeat bilirubin estimations.

Data was initially entered into a Microsoft Excel spread-

sheet (2007 version) then analysis was done using the

After assessing for the inclusion and exclusion criteria,

Statistical Package for the Social Sciences (SPSS) for

paired transcutaneous bilirubin measurement (average of

windows software version 20.0. Means and standard

five measurements) and blood sample collection for

deviations were calculated for continuous variables. The

serum bilirubin estimation was done within 30 minutes

difference between means of TcB and TSB were com-

of each other.

13,23,26

pared using the paired Student’s t -test. Linear regression

equation of TcB on TSB was generated.

Blood sample collection and analysis

Pearson’s correlation co -efficient (r) and coefficient of

determination (R ) were calculated to assess the relation-

Blood sample collection for the TSB measurement was

ship between the two methods of measurement. Agree-

by venepuncture as routinely done in the hospital. The

ment between TcB and TSB was determined using the

method of Bland and Altman;

specimen bottles were wrapped with aluminum foil pa-

the laboratory method

per to protect the specimen from light and avoid photo

was used as the ‘gold standard.’ For the purpose of the

conversion of bilirubin; then transported to the depart-

study, the pre-determined acceptable limit of agreement

mental laboratory and analyzed immediately with a time

between TcB and TSB was taken as 30 µmol/L or 2 mg/

dL. Limits of agreement of the mean differences were

lag of not more than 15 minutes. Sample analysis was

carried out at the Department of Paediatrics Research

given as 95% confidence intervals (CIs).

Laboratory. This laboratory routinely processes about 60

-70% of blood samples from paediatric patients in the

The performance of the Bilichek® device at high levels

hospital for various investigations including serum

of TSB was assessed by segregating subjects into two

bilirubin. The total and conjugated bilirubin was quanti-

groups; ≤ 12 mg/dL (low) and >12 mg/dL (high ). These

fied by the Malloy and Evelyn (Diazo) Method and the

levels were based on the institutional protocol of initiat-

bilirubin values read using the SFRI BSA-3000® Chem-

ing treatment of NNJ at TSB levels of above 12 mg/dL

istry Auto Analyzer, France according to the manufac-

for term babies (depending on postnatal age and pres-

turer’s instructions. The working reagent used to obtain

ence of risk factors, this level could be lower or higher).

the diazo derivative was prepared freshly daily, as well

For pre-terms the level for treatment is based on the

as daily calibration and standardization of the spectro-

weight (10 × infant weight in kg) or the same level for

photometer. The sample analysis was done in conjunc-

term babies, whichever is lower. Correlation coefficients

tion with two dedicated laboratory scientists with regu-

and Bland-Altman plots were also generated for these

lar inter-rater comparison of results. Quality control was

two groups.

done 2-4 weekly by analysis of the same sample in two

other laboratories in the hospital (Main Chemical Pa-

The sensitivity and specificity of the TcB measurement

thology laboratory and Department of Medicine labora-

to predict accurately the TSB at a range of values was

tory). Comparison of results was then undertaken with

estimated and plotted on receiver operator characteristic

necessary adjustments and recalibration of equipment

(ROC) curves. The interpretation of the area under the

done if necessary.

curve is as follows: 0.90 - 1(excellent); 0.80 - 0.90

(good); 0.70 - 0.80 (fair); 0.60 - 0.70 (poor); 0.50-0.60

Transcutaneous bilirubin measurement

(fail). The TcB cutoff point with the best sensitivity and

specificity to predict TSB was determined. The level of

The transcutaneous measurement was done with the non

statistical significance was set at p < 0.05.

-invasive Bilichek® bilirubin analyzer (Respironics inc.,

Murrysville, PA, U.S.A.). This is a handheld device that

estimates total serum bilirubin via the skin at multi-

wavelengths. Measurement was done according to the

Results

manufacturer’s instructions.

For each measurement, a

disposable calibration tip (Bilical®) was attached to the

A total of 200 paired readings were taken from 169 ba-

fibreoptic probe. If calibration was successful, five con-

bies (multiple readings were taken from some babies).

secutive measurements were taken from the forehead,

The babies were all of Nigerian origin. The study popu-

before an average final reading was displayed either in

lation consisted of 93 (55.0%) preterm neonates and 76

mg/dL or µmol/L according to the setting fixed. For the

(45.0%) term neonates. One hundred and twenty and 80

purpose of this study, the device was set to display re-

paired TcB and TSB readings were carried out on the

sults in mg/dL. The process took an average of 30 sec-

two groups respectively. Ninety-one (53.8%) males and

onds to 1minute. For patients requiring phototherapy, a

78 (46.2%) females were studied, giving a male: female

phototherapy protective patch (BiliEclipse®) was ap-

ratio of 1.2: 1. One hundred and four (61.5%) patients

plied to the forehead before commencement of treatment

were inborn while 65 (38.5%) were out-born. Table 1

summarizes the baseline characteristics of the subjects.

The correlation coefficient (r) between TSB levels and

TcB

readings

obtained

from

preterm

neonates

Table 1: Baseline characteristics of the study population

(gestational age less than 37 weeks) was 0.80 while that

Variable

n (%)

Range

Mean(SD)

for term neonates (gestational age 37 weeks and above)

( N=169)

was 0.73. Both correlation coefficients were strongly

Gestational age (weeks)

27-41

35.0 (3.9)

positive with p values of 0.000. The correlation coeffi-

cient in preterm neonates was higher than that for term

27 - 30

30 (17.8)

neonates, however this difference was not statistically

31 - 33

32 (19.0)

34 - 36

31 (18.3)

significant (p = 0.37, Z - 0.899).

37 - 39

40 (23.6)

40 - 42

36 (21.3)

Fig 1: Linear regression of TcB on TSB for all the readings

Birth weight Groups*

1000-4990

2335 (880)

VLBW

40 (23.6)

LBW

54 (32.0)

NBW

75 (44.4)

Sex

Male

91 (53.8)

Female

78 (46.2)

M:F; 1.2:1

Postnatal age (hours)

20-456

107 (72)

Site of Birth

Outborn

65 (38.5%)

Inborn

104 (61.5%)

Phototherapy use

Phototherapy

30 (17.8%)

No phototherapy

139 (82.2%)

*VLBW= very low birth weight; LBW= low birth weight; NBW=

Figure 2 is a Bland-Altman plot of the paired readings

normal birth weight

showing the agreement between TSB and TcB. It is a

plot of the differences between TcB and TSB against the

Five neonates had TcB readings of ‘HIGH’, indicating

mean or average of the paired values. The red broken

values of above 20 mg/dL. These TcB measurements

line represents the bias (mean TcB-TSB), while the

and their corresponding TSB values were excluded from

black broken lines show the Upper (UP) and Lower

analysis. This was due to the inability of the transcutane-

(Lw) limits of agreement (95% Confidence interval of

ous bilirubinometer to generate a numeric value for the

the bias or mean difference). The adjoining table shows

TcB readings, which made it impossible to assess corre-

the figures for the bias, standard deviation (SD) of bias

lation and differences between the two measurements.

and the 95% Limits of agreement. It shows a bias of

Therefore 195 paired readings were eventually analyzed.

1.5mg/dL and 95% limit of agreement (imprecision) of

The number of paired readings per subject ranged be-

between -3.5mg/dL and 6.6mg/dL.

tween one and three taken between 24-96hr intervals,

Table 2 shows the correlation between TcB and TSB for

with 145 (85.8%) of the babies having only one paired

the ≤ 12 mg/dL TSB group and the

> 12 mg/dL

reading taken from them. TcB readings were generally

TSB group while Table 3 shows the difference in means

higher than TSB levels for most of the paired measure-

of TcB and TSB for the 2 TSB groups. The correlation

ments; 148 (76%). TcB readings were lower than TSB

coefficient (r) between TSB and TcB was 0.64 for the ≤

values in 41 (21%) of the paired measurements and were

12mg/dl TSB group and 0.43 for the ≥ 12mg/dl TSB

the same as TSB values in 6 (3%) of the paired measure-

group.

ments. Seventy-seven (52%) out of 148 TcB readings,

were higher than corresponding TSB values by more

Fig 2: Bland-Altman plot of differences between TcB and

than 2 mg/dL, while 17 (41.5%) out of 41 TcB readings,

TSB against average or mean of TcB and TSB for all subjects.

were lower than corresponding TSB values by more

Upp (Upper), Lw (Lower), Diff (difference).

than 2 mg/dL. Thirty-three readings were taken from

patients receiving phototherapy. Thirty-two of them

were preterm while only one was a term neonate. The

five excluded paired readings were all from preterm

neonates.

The mean (SD) of TcB readings was 11.3 (3.8) mg/dL.

This was higher than the mean (SD) TSB level of 9.8

(3.9) mg/dL with a difference of 1.5 mg/dL, SD of 2.6

Bias

1.521

mg/dL and 95% CI of 1.2-1.9 mg/dL. This difference

SD of bias

2.607

was statistically significant with a p value of 0.000, t =

95% Limits of Agreement

From

-3.588

8.147.

6.630

The overall correlation coefficient (r) between TSB and

TcB was 0.77 (p < 0.001). Figure 1 shows the scatter

diagram of the regression between TcB and TSB.

The agreement between TcB and TSB at the two differ-

ent TSB level groups was also determined by the

method of Bland and Altman. Figures 3 and 4 show the

Fig 4: The Bland-Altman plot of the difference between mean

Bland-Altman plots of the difference between TcB and

TcB and TSB against the mean of TcB and TSB for the > 12

TSB against the mean of TcB and TSB for the ≤ 12 mg/

mg/dL TSB group. Upp (Upper), Lw (Lower), (Diff) Differ-

dL and >12 mg/dL TSB level groups respectively. The

ence.

limits of agreement for the “low” TSB group were be-

tween - 3.0 and 6.7 mg/dL while that of the “high” TSB

group were between -8.0 and 7.8 mg/dL.

Table 2: The correlation between TcB and TSB at TSB levels

≤ 12 mg/dL and > 12 mg/dL

TSB ≤ 12 mg/dL

TSB > 12

mg/dL

Bias

-0.08958

Number of readings

152

SD of bias

4.020

95% Limits of Agreement

Correlation coefficient ( r )

0.636

0.429

From

-7.969

Coefficient of determina-

0.40

0.18

7.790

tion (R )

P value

0.000

0.004

Fig 5: Receiver Operating Characteristic (ROC) curve of plots

Regression equation

TcB = 3.45 +

TcB = 10.23

of sensitivity and 1-specificity of TcB levels in predicting TSB

0.80TSB

+ 0.35TSB

as gold standard test. The blue line represents the curve of the

plots.

The receiver operating characteristic (ROC) curve plot-

ted to assess the accuracy of TcB to predict TSB is

shown in Figure 5. The ROC curve was obtained by

plotting the sensitivity and specificity of TcB to predict

TSB (gold standard test) at different TcB cut-off levels.

In this curve, an area of 1 represents perfect sensitivity

and specificity of 100% each. The area under the curve

(AUC) for the two diagnostic modalities in this study

revealed an area of 0.73 (standard error of 0.05 and 95%

CI of 0.63 - 0.83). This showed that the accuracy of TcB

as a diagnostic method to predict TSB is “fair.” The best

TcB cutoff point which maximized sensitivity and speci-

ficity to produce this AUC was 11.65 mg/dL with sensi-

tivity of 0.87 and 1- specificity of 0.452.

Table 3: Difference in mean bilirubin readings of subjects at

different TSB levels

TSB ≤ 12 mg/dL

TSB >12 mg/dL

Number of readings

152

Mean TcB (SD) mg/dL

10.1 (3.1)

15.7 (2.3)

Mean TSB (SD) mg/dL

8.2 (2.5)

15.4 (2.8)

Mean difference (95%

1.9 (1.5 - 2.3)*

0.3 (-0.5 - 1.1)

Discussion

CI) mg/dL

This study revealed that the transcutaneous bilirubin

Key: Paired t-test comparison between TcB and TSB for TSB

(TcB) readings generally overestimated the correspond-

values ≤ 12mg/dL and > 12mg/dL; *t statistic 9.349, degree of

ing total serum bilirubin (TSB) levels. Also the TcB

freedom = 151, p = 0.000, ** t statistic 0.675, degree of free-

dom = 42, p = 0.503.

readings obtained with the Bilichek® device, had strong

positive linear correlation with TSB levels. This finding

was consistent with previous studies

13,17,22-24,32,33

Fig 3: Bland-Altman plots of difference between mean TcB

carried

and TSB and mean of TSB and TcB for the ≤ 12 mg/dL TSB

out both in Nigeria and in other regions worldwide.

group. Upp (Upper), Lw (Lower), Diff (Difference).

However, the correlation in the present study (r = 0.77)

was found to be slightly lower than that of others which

ranged from r of 0.79 to 0.98.

13,17,23,24,27,32-35

These differences could be due to variations between the

studies in relation to type of transcutaneous bilirubi-

nometer, laboratory method used for TSB analysis, and

population of neonates studied. For instance in other

studies done in Nigeria; Owa et al

used a different

transcutaneous bilirubinometer, JM 102®; Kayode-

Bias

1.871

Adedeji et al

SD of bias

2.468

used the JM103; Olusanya et al

used

95% Limits of Agreement

[24]

From

-2.966

both Bilichek® and JM 103® and Slusher et al

used

the Bilichek® device. Rylance et al

6.708

in Malawi, used

the JM 103®. Furthermore, Owa et al

studied well,

US.

Conversely Bhutani et al in the US,Rubaltelli et

al in Europe and Jangaard et al in Canada

term neonates; Kayode-Adedeji and colleagues

stud-

found an

ied only preterms; Olusanya et al

studied well, term,

underestimation of TSB by TcB. There was an overesti-

inborn neonates; Rylance et al studied both in and out-

mation of > 2mg/dl in about 52% of the babies com-

pared to the finding by Olusanya et al of 64.5%. Simi-

born, preterm and term neonates, excluding only ex-

tremely premature and very sick neonates; while the

lar findings of wide disparities between TcB and TSB of

present study and that of Slusher et al were carried out

>2-3mg/dl were also obtained from other studies by

Rylance et al in Malawi and Taylor and colleagues in

on both preterm and term neonates born in and out of

the hospital, regardless of their health status. Some of

the USA. This US study found this disparity more in

the other studies,

13,17,22

had study populations consisting

black neonates than in Caucasians.

mainly of term and well Caucasian subjects, this could

also account for the disparity in the correlation coeffi-

It was noticed that most of the studies above that found

cients observed.

an overestimation of TSB had either predominantly

There was strong linear correlation between TcB and

black or preterm subjects. This would further support

TSB for both preterm (r-0.80) and term neonates (r-

the contribution of melanin and reduced subcutaneous

0.73) with no statistically significant difference between

tissue to increased TcB measurements as proposed by

the correlation coefficients obtained for the two groups.

Yamauchi

and shows that the technology improve-

This finding was in line with previous studies done by

ments made in the Bilichek device® may not be ade-

Deluca et al in Italy and Ahmed and colleagues in the

quately eliminating the effect of these factors.

UK who both documented good correlation between

Regarding performance of the Bilichek device® at dif-

TcB and TSB in preterm neonates. Contrary results were

ferent levels of TSB, the present study found weaker

reported by Jangaard et al in Halifax, Nova Scotia, of

correlation between TcB and TSB at “high” levels of

poorer performance of the Bilichek device in preterm

TSB above 12 mg/dL. This finding was also reported by

neonates, though they stated their small sample size of

other studies,

23,34,39

but was contrary to the study by

63 readings in preterm neonates as a limitation of their

Slusher et al who found a stronger correlation at levels

study.

above 12 mg/dL. Slusher et al

however obtained an

underestimation of TSB at “high” levels of TSB which

The coefficient of determination (R ) obtained by this

was not corroborated in this study. This may be due to

study supported the linear relationship between TcB and

the small sample size of neonates with “high” TSB lev-

els of above 12 mg/dL, compared to the Slusher et al

TSB, but this was not very strong. Also, the slope of

0.74 and intercept of 4.02 obtained from the regression

study which consisted of a large population of neonates

equation generated in the present study, were consistent

with severe hyperbilirubinaemia. There was also a

with values of 0.73 and 4.5 obtained by Slusher and col-

poorer agreement between TcB readings and TSB lev-

els, at higher TSB levels similar to the Slusher et al

leagues but contrary to other studies which had higher

R , low intercepts of approximately 1 mg/dL and slopes

study.

close to one.

13,33,34

The clinical implication of overestimation of TSB by

The high intercept and low coefficient of determination

TcB would be the tendency to increase the number of

obtained by this present study indicated the presence of

neonates requiring treatment, but would ensure no case

unknown factors or variables contributing to the TcB

of severe NNJ is missed. In the present study, 152 out of

readings generated which resulted in the TcB readings

the 195 readings had TSB values 12 mg/dL and below,

generally being higher than the corresponding TSB lev-

if only transcutaneous bilirubinometry was applied on

els. Slusher and colleagues had postulated that the rea-

them, 78% of blood sampling for TSB measurement

son for this high intercept could be due to the presence

would have been avoided. On the other hand, poor

of light absorbing non bilirubin, nutrition-derived yel-

agreement obtained, especially at high TSB levels with

low pigments (carotenoids) in the skin of the infants

differences as high as 8mg/dl between TcB and TSB,

which could interfere with TcB measurements since they

would necessitate confirmation of the total serum

share similar physical and optical characteristics with

bilirubin level by another method directly from the

bilirubin.

blood.

This may also hold true in the population of the present

Due to the finding of poor agreement between TcB and

study, since babies here are from the southwest region of

TSB at high bilirubin levels, it was expedient to find a

Nigeria where their mothers consume diet rich in carote-

cut off level of TcB beyond which its accuracy or reli-

noids similar to those studied by Adelekan et al which

ability was reduced and the assessment of severity of

can be transferred to them through breastmilk. However

hyperbilirubinaemia by another method may be neces-

this has not been corroborated.

sary. This was done by generating ROC curves. A TcB

Mean TcB readings obtained by the Bilichek device®

level of 11.7 mg/dL (200 μ mol/L) was obtained with an

were found to overestimate mean TSB values, with a

area under the curve (AOC) of 0.73. The TcB cut-off

statistically significant difference. This finding was con-

levels obtained by other studies are 14 mg/dL (239

sistent with other studies done in Nigeria by Slusher et

μ mol/L) with AOC of 0.98 by Leite et al; 15.2 mg/dL

al,

and Olusanya et al;

in Europe by Ahmed et al

(260 μ mol/L) by Jangaard et al;

bilirubin index of 20

and De Luca et al;

and by Karon and colleagues in the

(TSB level of 9.9 mg/dL) by Owa et al;

10.3 mg/dL

with AOC of 0.706 by Olusanya et al; and 14.6 mg/dL

fect of melanin on the accuracy of TcB measurements.

(250 μ mol/L) by Boo and Ishak. The differences in cut

This would help in the development of transcutaneous

-off levels are likely due to population differences as

bilirubinometers with less disparity of results compared

well as the different methods of obtaining them.

with laboratory methods in neonates of African descent.

Thus, the Bilichek® Transcutaneous Bilirubinometer

The present study has demonstrated that transcutaneous

can become an indispensable tool in the management of

bilirubin measurements with the Bilichek device corre-

neonatal jaundice, aiding in early determination of treat-

late with total serum bilirubin levels in neonates of Afri-

ment and reduction of blood sampling by health person-

can descent. It has proven to be a reliable screening

nel, to achieve Nigeria’s goal of reducing the great mor-

method to determine the severity of hyperbilirubinaemia

bidity and mortality from this easily treatable condition.

in clinically jaundiced babies. However, the finding

from this study, of the possibility of wide disparities

between TcB and TSB, as well as lower sensitivity and

specificity especially at bilirubin levels >11.7 mg/dL,

Conclusion

precludes the use of the Bilichek device as a substitute

for laboratory measurements. Confirmatory serum

Transcutaneous bilirubinometry is a simple, fast and

bilirubin measurements are recommended at these lev-

reliable screening method for assessing severity of hy-

els, as well as for any neonate for which the TcB reading

perbilirubinaemia in black term and preterm neonates

suggests need for treatment such as preterm neonates,

which can reduce the need for blood sampling. However

who may require treatment at bilirubin levels below

due to the occurrence of wide discrepancies, it is recom-

12mg/dL.

mended that confirmatory serum bilirubin measurements

are done for neonates with high TcB levels of

Due to the ease of use, portability and lack of depend-

11.7mg/dL or any TcB level suggesting need for treat-

ence on electricity, its use transcends the hospital setting

ment.

and can be used even in primary health care centers in

rural communities and homes, for bilirubin monitoring,

referral and follow up of patients. Transcutaneous

Acknowledgements

bilirubinometry has been found to be a more accurate

screening method for determining severity of hyper-

We would like to acknowledge the following individuals

bilirubinaemia and need for blood sampling compared to

who contributed to the success of this work. The Resi-

visual assessment

40,41

which is the current method used

dents and the Laboratory scientists in the Department of

in most centers in Nigeria. It is currently being used in

Paediatrics of the Lagos University Teaching Hospital

developed countries like the United Kingdom and

for their assistance with blood sample collection and

United States of America, where it has helped to reduce

analysis.

Dr Remi Ogundimu for help in procurement

the number of readmissions for severe NNJ

and save

of the Bilichek® device and Mr Samuel Bodunrin for

costs, as unnecessary blood sampling and laboratory

statistical analysis of data.

measurements are avoided.

43,44

However its utility in

predominantly black neonates as a sole means of

bilirubin measurement is somewhat limited, due to the

Authors’ contributions

overestimation (with some wide differences) obtained

Akinbolagbe YO: concept and design; data acquisition;

between the instrument and laboratory methods.

analysis and interpretation of data; manuscript prepara-

Another major drawback of the device as compared to

tion, editing and review.

other transcutaneous bilirubinometers such as the JM

Ezeaka VC: study design; analysis and interpretation of

103® is the added cost of the disposable calibration

data; manuscript preparation, editing and review.

pads. This can be circumvented by subsidy of the cost

Akinsulie AO: study design; analysis and interpretation

by the manufacturers for low income countries such as

of data; manuscript preparation, editing and review.

ours, or refinement of the technology to make them re-

Conflict of Interest: None

usable. More studies should be done to evaluate the ef-

Funding: None

References

Knudsen A. Prediction of the

Simiyu DE. Morbidity and

Effiong CE, Aimaku VE,

development of neonatal jaun-

mortality of neonates admitted

Oyedeji GA, Ikpe DE. Neona-

dice by increased umbilical

in general paediatric wards at

tal Jaundice in Ibadan: Inci-

cord blood bilirubin. Acta Pae-

Kenyatta National Hospital.

dence and Etiological Factors in

diatr Scand 1989;78:217-21.

East Afr Med J 2003;80:611-

Babies bornn in Hospital . J Natl

Stoll BJ, Kliegman RM. Jaun-

Med Assoc 1975;61:208-13.

dice and hyperbilirubinaemia in

Ahmed H, Yukubu AM,

Udo JJ, Anah MU, Ochigbo SO,

the newborn. In: Behrman R,

Hendrikse RG. Neonatal jaun-

Etuk IS, Ekanem AD. Neonatal

Kliegman R, Jenson H, (editors).

dice in Zaria, Nigeria- a sec-

morbidity and mortality in Cala-

Nelson Textbook of Pediatrics,

bar, Nigeria: a hospital-based

ond prospective study. West

17th edition. New Delhi: Elsevier;

study. Niger J Clin Pract

Afr J Med 1995;14:15-23.

2004.p.592-596.

2008;11:285-9.

Owa JA, Osinaike AI. Neonatal

17. Karon BS, Teske A, Santrach

25. De Luca D, Zecca E, de Turris

Morbidity and Mortality in

PJ, Cook WJ. Evaluation of

P, Barbato G, Marras M., Ro-

Nigeria. Indian J Pediatr

the Bilichek Noninvasive

magnoli C. Using Bilichek for

1998;65:441-9.

Bilirubin Analyzer for Predic-

preterm neonates in a sub-

Onyearugha CN, Onyire BN,

tion of Serum Bilirubin and

intensive unit: diagnostic use-

Ugboma HAA. Neonatal Jaun-

Risk of Hyperbilirubinemia.

fulness and suitability. Early

dice: Prevalence and associated

Am J Clin Pathol

Hum Dev 2007;83:313-7.

factors as seen in Federal

2008;130:976-82.

26. Jangaard KA, Curtis H, Gold-

Medical Centre, Abakaliki,

18. Schreiner RL, Glick MR. In-

bloom RB. Estimation of

bilirubin using Bilicheck

Southeast Nigeria. J Clin Med

terlaboratory bilirubin vari-

, a

Res 2011;3:40-5.

ability. Pediatrics

transcutaneous bilirubin

Owa JA, Dawodu AH. Neona-

1982;69:277-81.

measurement device: Effects

tal jaundice among Nigerian

19. Lo S, Doumas B, Ashwood E.

of gestational age and use of

preterm infants. West Afr J

Bilirubin proficiency testing

phototherapy. Paediatr Child

Med 1990;9:252-7.

using specimens containing

Health 2006;11:79-83.

10. Martin CR, Cloherty JP. Neo-

unconjugated bilirubin and

27. Olusanya BO, Imosemi DO,

natal Hyperbilirubinemia. In:

human serum. Arch Pathol

Emokpae AA. Differences

Cloherty J, Eichenwald E,

Lab Med 2004;128:1219-23.

between transcutaneous and

Stark A (editors). Manual of

20. Robertson A, Kazmierczak S,

serum bilirubin measurements

Neonatal care. 6th edition.

Paul V. Improved Transcuta-

in black African neonates.

Philadelphia: Wolters Kluwer/

neous bilirubinometry: Com-

Pediatrics 2016;138:1-12.

Lippincott, Williams & Wil-

parison of SpectRx Bilichek

28. Rylance S, Yan J, Molyneux

kins; 2008.p. 181-212.

and Minolta Jaundice meter

E. Can transcutaneous

11. Lauff JJ, Kasper ME, Ambrose

JM 102 for estimating Total

bilirubinometry safely guide

RT. Seperation of bilirubin

serum bilirubin in a normal

phototherapy treatment of

species in serum and bile by

newborn population. J Perina-

neonatal jaundice in Malawi?

high-performance reversed-

tology 2002; :12- 4.

Paediatr Int Child Health

21. Respironics, inc. Bilichek

phase liquid chromatography. J

2014;34(2):101-7.

Chromatogr 1981;226:391-

Non-Invasive Bilirubin Ana-

29. Eng J. Sample Size estima-

402.

lyzer. User Instruction Manual

tion: How many individuals

12. Grohmann K, Roser M,

2003. p. 1-36. [assessed and

should be studied? Radiology

Rolinski B, Kadow I, Muller C,

cited 2011 Nov 7] available

2003;227:309-13.

Goerlach-Graw A, et al.

from: URL:

30. Simmons N. An automated

Bilirubin measurement for

www.ebookbrowse.com/

method for serum bilirubin

Neonates : comparison of 9

respironics-bilichek-

determination. J Clin Pathol

frequently used methods. Pedi-

bilirubinanalyzer-service-

1968;21:196-201.

atrics 2006;117:1174-83.

manual-pdf-d202106075.

31. Bland JM, Altman DG. Statis-

13. Rubaltelli FF, Gourley GR,

22. Bhutani VK., Gourley G R.,

tical methods for assessing

Loskamp N, Modi N, Roth-

Adler S. Noninvasive meas-

aggreement between two

Kleiner M, Sender A, et al.

urement of total serum

methods of clinical measure-

Transcutaneous Bilirubin

bilirubin in a multiracial pre-

ment. Lancet 1986;1:307-10.

Measurement: A Multicenter

discharge newborn popupation

32. Owa J A, Esimai VC, Olowu

Evaluation of a New Device.

to assess the risk of severe

WA, Jegede OA. Correlation

Pediatrics 2001;107:1264-71.

hyperbilirubinemia. Pediatrics

between readings on Icterome-

14. Kramer L.I. Advancement of

2000;106:E17.

ter, Jaundicemeter and Serum

Dermal Icterus in the Jaundiced

23. Leite MG, Granato VA, Fac-

bilirubin Concentrations in

Newborn. Amer J Dis Child

chini F, Marba S. Comparison

Newborn Infants. Niger J

1969;118:454-8.

of transcutaneous and plasma

Paed 1995;22:24-30.

15. Yamauchi Y, Yamanouchi I,

bilirubin measurement. J Pe-

33. Kayode-Adedeji BO, Owa JA,

Igarashi I. Transcutaneous

diatr (Rio J) 2007;83:283-6.

Akpede GO, Alikah SO.

bilirubinometry: preliminary

24. Slusher TM, Angyo IA, Bode-

Evaluation of Jaundice meter

studies of noninvasive transcu-

Thomas F, Akor F, Pam SD,

in the assessment of jaundice

taneous bilirubinometer in the

Adetunji AA, et al. Transcuta-

among Nigerian preterm neo-

Okayama National Hospital.

neous Bilirubin Measurements

nates. Niger J Paed 2015;42

Pediatrics 1980;65:195-202.

and Serum Total Bilirubin

(3):194-8.

16. Gosset IH. A perspex icterome-

Levels in Indigenous African

34. Boo NY, Ishak S. Prediction

ter. Lancet 1960;1:87-8.

Infants. Pediatrics

of severe hyperbilirubinaemia

2004;113:1636-41.

using the Bilichek transcuta-

neous bilbirubinometer. J

Paediatr Child Health

2007;43:297-302.

35. Bhutani V, Johnson L, Sivieri

39. Kolman KB, Mathieson KM,

42. Kuzniewicz MW., Escobar

E. Predictive ability of a pre-

Frias C. A Comparison of

GJ., Newman TB. Impact of

discharge hour-specific serum

Transcutaneous and Total

Universal Bilirubin Screening

bilirubin for subsequent signifi-

Serum Bilirubin in Newborn

on severe hyperbilirubinemia

cant hyperbilirubinemia in

Hispanic Infants at 35 or more

and phototherapy use. Pediat-

healthy term and near-term

weeks of Gestation. J Am

rics 2009;124:1031-9.

infants. Pediatrics 1999;103:6-

Board Fam Med 2007;20:266

43. Eggert LD, Wiedmeiser SE,

14.

-71.

Wilson J, Christensen RD.

36. Ahmed M, Mostafa S, Fisher

40. Mishra S., Chawla D., Agawal

The effect of instituting a pre-

G, Reynolds TM. Comparison

R., Deovari AK., Paul VK,

discharge newborn bilirubin

between transcutaneous

Bhutani VK. Transcutaneous

screening program in an 18-

bilirubinometry and total serum

bilirubinometry reduces need

hospital health system. Pediat-

bilirubin measurements in pre-

for blood sampling in neo-

rics 2006;117:855-62.

term infants <35 weeks gesta-

nates with visible jaundice.

44. Hartshorn D, Buckmaster A. '

tion. Ann Clin Biochem

Acta Paediatr 2009;98:1916-

Halving the heel pricks' :

2010;47:72-7.

evaluation of a neonatal jaun-

37. Adelekan DA, Fatusi AO, Fa-

41. Szabo P, Wolf M, Bucher H.

dice protocol incorporating the

kunle JB, Olotu CT, Olukoga

Assessment of jaundice in

use of a transcutaneous

IA, Jinadu MK. Prevalence of

preterm neonates: comparison

bilirubinometer. J Pediatr

malnutrition and vitamin A

between clinical assessment,

Child Health 2010;46:595-9.

deficiency in Nigerian pre-

two transcutaneous bilirubi-

school children subsisting on

nometers and serum bilirubin

high intakes of carotenes. Nutr

values. Acta Paediatr

Health 1997;12:17-24.

2004;93:1491-5.

38. Taylor JA, Burgos AE, Flaher-

man V, Chung EK, Simpson

EA, Goyal NK, et al. Discrep-

ancies between transcutaneous

and serum bilirubin measure-

ments. Pediatrics 2015;135

(2):224-31.