Screening for Hyperbilirubinemia in Japanese Very Low Birthweight

Infants Using Transcutaneous Bilirubinometry

Daisuke Kurokawa, MD1, Hajime Nakamura, MD, PhD1, Tomoyuki Yokota, MD, PhD1,2, Sota Iwatani, MD1,
Takeshi Morisawa, MD, PhD2, Yoshinori Katayama, MD, PhD3, Hitomi Sakai, MD4, Tomoaki Ioroi, MD, PhD5,
Kazumoto Iijima, MD, PhD1, and Ichiro Morioka, MD, PhD1

Objectives To assess the accuracy of transcutaneous bilirubin (TcB) measurements at 5 different body sites in
Japanese very low birthweight (VLBW) infants and to determine a cut-off value of TcB to detect total serum/plasma
bilirubin (TB) levels $10 mg/dL (171 mM).
Study design In a prospective multicenter study, 85 Japanese VLBW infants were enrolled from 5 neonatal inten-
sive care units during the study period. A total of 383 blood samples from infants not receiving phototherapy or
$24 hours postphototherapy were analyzed. TcB was measured at the forehead, sternum, upper back, lower
abdomen, and waist within 1 hour of blood collection. Linear regression analysis and Bland-Altman plots were
used to compare TcB values at each site with TB levels. The TcB cut-off value for detecting TB $10 mg/dL was
determined by receiver operating characteristics curve analysis.
Results TcB significantly correlated with TB, but the coefficient of determination varied among the sites (forehead:
0.5294, sternum: 0.6488, upper back: 0.6321, lower abdomen: 0.5430, waist: 0.7396). At a TcB value $8, the sensi-
tivity was 100% at the sternum and upper back, 85% at the waist, 84% at the forehead, and 64% at the lower
abdomen to detect TB $10 mg/dL.
Conclusions In Japanese VLBW infants, the accuracy of TcB measurements varies according to body site. TcB
$8 on the sternum or upper back is more reliable than that on the forehead, lower abdomen, or waist to detect TB
levels $10 mg/dL. (J Pediatr 2016;168:77-81).

See editorial, p 6

C
linical kernicterus has been observed in Japan, with an incidence of 1.8 per 1000 live births <30 weeks of gestational age
(GA).1 We previously found that total serum/plasma bilirubin (TB) levels in most very low birthweight (VLBW) infants
with kernicterus peak after 1 week of age (median age: 28 days).2 This finding indicates that VLBW infants should be
continuously monitored for hyperbilirubinemia during their neonatal intensive care unit (NICU) stay, in contrast to moni-
toring for one week after birth as currently practiced in Japan.
Transcutaneous bilirubin (TcB) measurements taken at the forehead and ster-
num are a noninvasive screening method for hyperbilirubinemia. These mea-
1
From the Department of Pediatrics, Kobe University
surements are clinically useful for identifying term or late preterm infants with Graduate School of Medicine, Kobe; Department of 2

hyperbilirubinemia because TcB values correlate with TB levels, except for in Pediatrics, Kakogawa West Municipal Hospital,
3
Kakogawa; Department of Pediatrics, Takatsuki
Black infants.3-6 Previous studies in preterm or VLBW infants have shown that General Hospital, Takatsuki; Department of 4

Neonatology, Hyogo Prefectural Kobe Childrens

TcB significantly correlates with TB levels, with a slightly weaker correlation co- 5
Hospital, Kobe; and Department of Pediatrics,
efficient than in term infants.7-13 However, TcB measurements are not routinely Japanese Red Cross Society Himeji Hospital, Himeji,
Japan
used for screening for hyperbilirubinemia in this vulnerable population Supported by JSPS KAKENHI (26461633 [to H.N. and
throughout the world. The major reasons might be that TcB in VLBW infants I.M.] and 15K19652 [to S.I.]) and Kurozumi Medical
Foundation (H26 [to S.I.]). Konica Minolta, Inc provided a
underestimates TB compared with that in term infants,10,14 and the most accu- JM-105 jaundice meter during the study period for use in
each hospital. I.M. has received grants from the Japan
rate body site has not been determined. Blood Product Organization and AbbVie LLC; received
lecture fees from Pfizer Japan, Inc, Novo Nordisk Pharma
Ltd, Eli Lilly Japan KK, Shionogi Co, Ltd, AbbVie LLC, GE
Healthcare Japan Ltd, and Atom Medical Corp; and
serves as a paid consultant for Sanofi KK. K.I. has
CV Coefficient of variation received grants from Daiichi Sankyo Co, Ltd, Japan
GA Gestational age Blood Product Organization, Miyarisan Pharmaceutical
NICU Neonatal intensive care unit Co, Ltd, AbbVie LLC, CSL Behring, JCR Pharmaceuticals
Co, Ltd, and Teijin Pharma Co, Ltd; and served as a paid
NPV Negative predictive value lecturer for MSD, ALEXION Pharmaceuticals, AstraZe-
PPV Positive predictive value neca KK, Meiji Seika Pharma Co, Ltd, Novartis Pharma
KK, Zenyaku Kogyo Co, Ltd, Chugai Pharmaceutical Co,
R2 Coefficient of determination Ltd, Astellas Pharma Inc, Daiichi Sankyo, Co, Ltd,
ROC Receiver operating characteristics Springer Japan, and Asahi Kasei Pharma Corp. The other
authors declare no conflicts of interest.
TB Total serum/plasma bilirubin
TcB Transcutaneous bilirubin 0022-3476/$ - see front matter. Copyright 2016 Elsevier Inc. All
VLBW Very low birthweight rights reserved.
http://dx.doi.org/10.1016/j.jpeds.2015.08.038

77
THE JOURNAL OF PEDIATRICS  www.jpeds.com Volume 168

We assessed the accuracy of TcB measurements at 5
different body sites in Japanese VLBW infants, and then
determined a cut-off TcB value for detecting TB levels
$10 mg/dL (171 mM), when phototherapy is initiated in
VLBW infants per Japanese guidelines.15,16
Methods
We performed a multicenter prospective study of 85 VLBW
infants who were born from April 2013 to December 2014
and admitted to the NICU at Kobe University Hospital, Ka-
kogawa West Municipal Hospital, Takatsuki General Hospi-
tal, Hyogo Prefectural Kobe Childrens Hospital, and
Japanese Red Cross Society Himeji Hospital, Japan. We
received institutional review board approval from Kobe Uni-
versity Hospital (1450) and written informed consent from
the parents.
Before enrollment of patients, the following protocol was
established. In VLBW infants not receiving phototherapy
or $24 hours postphototherapy, blood was collected for
TB measurements based on the clinical indication by neona-
tologists at each hospital, and TcB measurements were then
taken at some or all of 5 different body sites: forehead, ster-
num, upper back, lower abdomen, and/or waist. All TcB
measurements were taken within 1 hour of blood sampling.
Blood was collected by venipuncture, then shielded from
exposure to light, and measured within 1 hour after blood
sampling.
TcB was measured 3-6 times at each body site using the
JM-105 jaundice meter (Konica Minolta, Inc, Tokyo, Japan)
according to the manufacturers instructions. The median
value was designated as the TcB value at each measurement
site. TB levels were measured by spectrophotometry using a
UB-Analyzer (Arrows Co, Ltd, Osaka, Japan).17,18 TB values
that were measured using this device correlate with those
measured by high-performance liquid chromatography,
which is the gold standard method for determination of TB.4
ating characteristics (ROC) curve analyses. The sensitivity,
specificity, and positive and negative predictive values
(PPV and NPV) were calculated and determined for the
TcB cut-off value and the best measurement site. Validation
of TcB measurements by examiners was analyzed by the
Kruskal-Wallis test. All analyses were conducted using Excel
Statistics (Statcel 3; Social Survey Research Information Co,
Ltd, Tokyo, Japan). Statistical significance was determined
when P < .05.
Results
A total of 85 VLBW infants not receiving phototherapy or
$24 hours postphototherapy were enrolled because TcB
values did not correlate with TB levels during phototherapy
(n = 35, R2 = 0.1662).
Median GA and birthweight were 29 weeks (range, 22-
36 weeks) and 1154 g (range, 470-1490 g), respectively
(Table I). Forty-two percent of infants were small for GA,
which indicated that birthweight was below the 10th
percentile compared with mean values at the same GA in
Japanese infants.19 All of the infants were of Japanese
descent. Two infants were diagnosed with hemolysis with
positive Coombs tests or elevations of carboxyhemoglobin
(ABO incompatibility and unknown cause). A total of 383
blood samples from all 85 infants were analyzed (median
age: 20 days; range: 1-117 days).
Precision of TcB Measurements by Examiners
The CVs for TcB measurements at each body site taken by
each examiner are shown in Table II (available at www.
jpeds.com). Although most CV values were more than
Table I. Background characteristics in VLBW infants
(n = 85)
n %

Validation of TcB Measurements by Examiners Birth weight

<500 g 2 2.4
To determine the consistency of TcB values, 3 independent 500-999 g 28 32.9
examiners (neonatologist, resident doctor, and nurse) per- 1000-1499 g 55 64.7
formed 6 replicates of TcB measurements at 5 different GA (wk)
22-27 25 29.4
body sites in a VLBW infant and results were compared 28-33 52 61.2
among users. The mean TcB values and coefficients of varia- 34-36 8 9.4
tion (CVs) were then calculated. Male 44 51.8
Small for GA 36 42.4
Delivery mode
Data and Statistical Analyses Vaginal 6 7.1
Linear regression analysis was used to determine the correla- Cesarean 79 92.9
History of childbirth
tion between TcB at each body site and TB levels.6 The regres- Nullipara 42 49.4
sion equation and coefficient of determination (R2) were Multipara 43 50.6
calculated using the results between the TcB and TB values. Nationality
Japanese 85 100
In addition, R2 were compared between 0 and 7 days of age Hemolysis 2 2.4
and thereafter. Bland-Altman plots (mean
2 SD) were Postnatal age at blood sampling (n = 383)
used to determine the difference between TcB at each body 1-7 d 92 24.0
8-14 d 59 15.4
site and TB levels.6 TcB cut-off values for detecting TB levels 15-117 d 232 60.6
$10 mg/dL for each body site were analyzed by receiver oper-
78 Kurokawa et al
January 2016
10%, there were no significant differences in TcB
measurements taken at any body site by the 3 examiners.
Linear Regression Analysis between TB Levels and
TcB Values
Figure 1 shows the linear regression analysis between TB and
TcB values for the sternum and upper back. We found that
TcB was significantly correlated with TB at every body site
(P < .0001). R2 was highest for the waist (0.7396, n = 157),
followed by the sternum (0.6488, n = 222), upper back
(0.6321, n = 177), lower abdomen (0.5430, n = 174), and
then forehead (0.5294, n = 277).
When we compared R2 between 0 and 7 days of age and
thereafter, R2 was higher in infants $8 days of age than in-
fants at 0-7 days of age at all of the body sites, except at the
waist (Table III; available at www.jpeds.com).
Bland-Altman Plots of TB Levels and TcB Values
Figure 1 shows Bland-Altman plots for the sternum and
upper back. The mean value of the difference between TB
and TcB values (TcB-TB) ranged from 1.1 to 1.3 at the
sternum (1.1, n = 222), lower abdomen (1.2, n = 174),
upper back (1.3, n = 177), and forehead (1.3, n = 277),
but was 1.7 at the waist (n = 157). The 2 SDS were 3.0 at
the waist, 3.7 at the sternum, 3.9 at the upper back, 4.3 at
the lower abdomen, and 4.4 at the forehead.
ROC Curve Analyses of TcB Values for Detecting TB
Levels 10 mg/dL and Determination of the TcB
Cut-Off Value
ROC curve analyses at each body site for detecting TB levels
$10 mg/dL by TcB are shown in Figure 2. The area under the
curve was more than 0.95 in at the waist, upper back, and
sternum. Because the cut-off value was raised to
approximately 8 in these ROC curves, we investigated the
sensitivity, specificity, PPV, and NPV for a TcB cut-off
Figure 1. Linear regression analysis between TB and TcB values and Bland-Altman plots for the 2 best sites of the body (sternum
and upper back).
Screening for Hyperbilirubinemia in Japanese Very Low Birthweight Infants Using Transcutaneous Bilirubinometry
ORIGINAL ARTICLES
value at 8. The sensitivity and NPV were both 100% at the
upper back and sternum. When the TcB cut-off value was 7
at the lower abdomen and waist and 6 at the forehead, the
sensitivity and NPV became 100%. However, these
specificities and PPVs were lower than those at the upper
back and sternum (Table IV).
Discussion
We report TcB measurements at 5 different body sites to
screen for hyperbilirubinemia in Japanese VLBW infants.
Importantly, we found that the TcB cut-off value was $8
for detecting TB levels $10 mg/dL, which is the representa-
tive value for initiating phototherapy after 1 week of age in
VLBW infants.15,16 In addition, TcB measurements taken at
the sternum or upper back were more reliable than those at
other body sites.
Similar to previous reports for preterm infants,7-13 our
study also showed that TcB significantly correlated with TB
levels at every body site. However, R2 at the forehead or lower
abdomen was lower than that at the other sites. In agreement
with previous studies,10,14 our results clearly showed that TcB
values at any body site underestimated TB levels (the differ-
ence between the 2 values was approximately 1-2 as shown by
Bland-Altman analyses). Among the body sites, the waist had
the greatest difference between TB and TcB values, but had
the smallest SDS. More importantly, we also found that
TcB $8 is needed for detecting TB levels $10 mg/dL. Because
100% sensitivity is desirable for any screening method, TcB
should be measured at the sternum and upper back rather
than at the forehead, lower abdomen, or waist. Although
the sensitivity is 100% at the forehead, lower abdomen, and
waist when the TcB cut-off value was 6 or 7, the specificity
was low. Yaser et al12 reported that the interscapular site
was a better and safer screening site than the forehead or
79
THE JOURNAL OF PEDIATRICS  www.jpeds.com Volume 168

Figure 2. ROC curve analysis of TcB values for detecting TB levels $10 mg/dL at the 5 different body sites. The candidate cut-
off TcB value is 8. AUC, area under the curve.

Table IV. Test performance characteristics for each TcB chest. Our study showed that measurements not only at the
cut-off value for detecting TB levels $10 mg/dL at the 5 upper back, but also at the sternum were reliable. This result
body sites is different from that in term or late preterm infants who usu-
Upper Lower ally have TcB measured at the forehead.6 We speculate that
Forehead Sternum back abdomen Waist the TcB measurement technique is not stable and consistent
n = 277 n = 222 n = 177 n = 174 n = 157 at the forehead and lower abdomen because the area of the
Cut-off value 8 8 8 8 8 forehead in VLBW infants is too small and the lower
Sensitivity 84.2% 100% 100% 63.6% 84.6% abdomen is too soft. We also investigated if TcB measure-
Specificity 88.0% 85.5% 85.4% 86.5% 93.1%
PPV 34.0% 33.3% 35.1% 24.1% 52.4%
ments were different according to the examiner. We found
NPV 98.7% 100% 100% 97.2% 98.5% that TcB measurements were not user-dependent, although
Cut-off value 7 7 7 7 7 the CV was slightly high. We consider that TcB measurement
Sensitivity 89.5% 100% 100% 100% 100%
Specificity 77.5% 76.3% 78.0% 71.8% 80.6%
taken at the sternum and upper back can be applied to the
PPV 22.7% 23.4% 26.5% 19.3% 31.7% clinical setting, with the understanding that TcB underesti-
NPV 99.0% 100% 100% 100% 100% mates TB in VLBW infants.
Cut-off value 6 6 6 6 6
Sensitivity 100% 100% 100% 100% 100%
TcB measurements within 7 days after birth may not be
Specificity 65.9% 60.9% 70.1% 60.7% 72.2% necessary because blood sampling is often performed for un-
PPV 17.8% 15.6% 21.0% 14.7% 24.5% stable respiratory or circulatory conditions in VLBW infants.
NPV 100% 100% 100% 100% 100%
In our study, TcB was less reliable within 7 days of birth
Bold type indicates the best test performance at each body site. than that thereafter, probably because of immaturity of the
80 Kurokawa et al
January 2016
skin. However, TcB measurements taken at 8 days or older are
important because severe hyperbilirubinemia at an older age
during the NICU stay can occur and lead to kernicterus in
VLBW infants.2 Therefore, we need to establish noninvasive
and painless screening for hyperbilirubinemia that requires
phototherapy, especially for older infants in the NICU. This
transcutaneous measurement could enhance early detection
of hyperbilirubinemia in this vulnerable population as well
as decrease the amount and frequency of blood sampling.
This study has some limitations. We do not know if the
criteria of TcB for detecting TB levels $10 mg/dL and the
reliable regions (sternum and upper back) can be applied
to other ethnic populations. Only 2 patients who weighed
less than 500 g at birth were enrolled in this study. Whether
TcB measurements are accurate in these more premature in-
fants is unknown. In spite of these limitations, we conclude
that in Japanese VLBW infants the accuracy of TcB measure-
ments varies according to body sites, and a TcB value $8 that
is measured at the sternum or upper back may be reliable for
detecting TB levels $10 mg/dL. n
We thank Tsubasa Koda, MD, PhD, and Masahiko Yonetani, MD,
PhD, for their technical support of the TcB measurements. We also
thank the medical technologists and nurses in each hospital for their
measurements of total bilirubin and transcutaneous bilirubin levels,
respectively.
Submitted for publication Jun 15, 2015; last revision received Jul 21, 2015;
accepted Aug 13, 2015.
Reprint requests: Ichiro Morioka, MD, PhD, Department of Pediatrics, Kobe
University Graduate School of Medicine, 7-5-2 Kusunoki-cho, Chuo-ku, Kobe
6500017, Japan. E-mail: ichim@med.kobe-u.ac.jp
References
1. Morioka I, Nakamura H, Koda T, Yokota T, Okada H, Katayama Y, et al.
Current incidence of clinical kernicterus in preterm infants in Japan.
Pediatr Int 2015;57:494-7.
2. Morioka I, Nakamura H, Koda T, Sakai H, Kurokawa D, Yonetani M,
et al. Serum unbound bilirubin as a predictor for clinical kernicterus
in extremely low birth weight infants at a late age in the neonatal inten-
sive care unit. Brain Dev 2015;37:753-7.
3. American Academy of Pediatrics. Management of hyperbilirubinemia in
the newborn infant 35 or more weeks of gestation. Pediatrics 2004;114:
297-316.
Screening for Hyperbilirubinemia in Japanese Very Low Birthweight Infants Using Transcutaneous Bilirubinometry
ORIGINAL ARTICLES
4. Yasuda S, Itoh S, Isobe K, Yonetani M, Nakamura H, Nakamura M, et al.
New transcutaneous jaundice device with two optical paths. J Perinat
Med 2003;31:81-8.
5. Engle WD, Jackson GL, Engle NG. Transcutaneous bilirubinometry.
Semin Perinatol 2014;38:438-51.
6. Akahira-Azuma M, Yonemoto N, Ganzorig B, Mori R, Hosokawa S,
Matsushita T, et al. Validation of a transcutaneous bilirubin meter in
Mongolian neonates: comparison with total serum bilirubin. BMC
Pediatr 2013;13:151.
7. Karolyi L, Pohlandt F, Muche R, Franz AR, Mihatsch WA. Transcuta-
neous bilirubinometry in very low birthweight infants. Acta Paediatr
2004;93:941-4.
8. Namba F, Kitajima H. Utility of a new transcutaneous jaundice device
with two optical paths in premature infants. Pediatr Int 2007;49:
497-501.
9. Stillova L, Matasova K, Zibolen M, Stilla J, Kolarovszka H. Transcuta-
neous bilirubinometry in preterm neonates. Indian Pediatr 2009;46:
405-8.
10. Schmidt ET, Wheeler CA, Jackson GL, Engle WD. Evaluation of trans-
cutaneous bilirubinometry in preterm neonates. J Perinatol 2009;29:
564-9.
11. Ebbesen F, Vandborg PK, Trydal T. Comparison of the transcutaneous
bilirubinometers BiliCheck and Minolta JM-103 in preterm neonates.
Acta Paediatr 2012;101:1128-33.
12. Yaser A, Tooke L, Rhoda N. Interscapular site for transcutaneous bili-
rubin measurement in preterm infants: a better and safer screening
site. J Perinatol 2014;34:209-12.
13. Ahmed M, Mostafa S, Fisher G, Reynolds TM. Comparison between
transcutaneous bilirubinometry and total serum bilirubin measure-
ments in preterm infants <35 weeks gestation. Ann Clin Biochem
2010;47:72-7.
14. Engle WD, Jackson GL, Stehel EK, Sendelbach DM, Manning MD. Eval-
uation of a transcutaneous jaundice meter following hospital discharge
in term and near-term neonates. J Perinatol 2005;25:486-90.
15. Maisels MJ, Watchko JF, Bhutani VK, Stevenson DK. An approach to the
management of hyperbilirubinemia in the preterm infant less than 35
weeks of gestation. J Perinatol 2012;32:660-4.
16. Nakamura H, Yonetani M, Uetani Y, Funato M, Lee Y. Determination of
serum unbound bilirubin for prediction of kernicterus in low birth-
weight infants. Acta Paediatr Jpn 1992;34:642-7.
17. Shimabuku R, Nakamura H. Total and unbound bilirubin determina-
tion using an automated peroxidase micromethod. Kobe J Med Sci
1982;28:91-104.
18. Miwa A, Morioka I, Yokota T, Shibata A, Matsuo K, Fujioka K, et al.
Correlation and precision of serum free bilirubin concentrations deter-
mined by single and two peroxidase concentration methods in term or
late-preterm newborn infants using a FDA-approved analyzer. Clin
Lab 2012;58:507-14.
19. Itabashi K, Miura F, Uehara R, Nakamura Y. New Japanese neonatal
anthropometric charts for gestational age at birth. Pediatr Int 2014;56:
702-8.
81
THE JOURNAL OF PEDIATRICS  www.jpeds.com Volume 168

Table II. Precision of TcB measurements by examiners

Examiner
A* B Cz
Mean SD Mean SD Mean SD
Median, range Median, range Median, range
n=6 CV (%) N=6 CV (%) n=6 CV (%) P value
Forehead 4.5
0.53 11.7 4.8
0.38 7.9 4.3
0.65 15.0 .31
4.5, 3.9-5.2 4.7, 4.5-5.4 4.4, 3.2-5.2
Sternum 5.1
0.56 11.2 4.6
0.46 10.0 4.6
0.24 5.2 .07
5.2, 4.0-5.6 4.8, 3.8-5.0 4.7, 4.3-4.9
Upper back 7.0
0.54 7.7 6.5
0.69 10.6 6.8
0.79 11.7 .52
6.9, 6.3-7.7 6.2, 5.9-7.7 6.9, 5.8-7.7
Lower abdomen 5.5
0.86 15.6 6.0
0.59 9.8 5.1
1.1 20.8 .30
5.8, 4.3-6.6 6.0, 5.2-6.8 5.3, 3.3-6.3
Waist 4.9
0.73 14.7 5.0
0.83 16.5 4.4
0.16 3.8 .15
4.8, 4.2-6.3 5.0, 4.1-6.4 4.3, 4.2-4.6

Note: measurements were taken in a neonate at a GA of 30 weeks, birthweight of 1352 g, and age of 14 days.
*Neonatologist.
Resident doctor.
zNurse.

Table III. Comparisons of regression equations and R2 between TcB and TB measurements at 0-7 days of age and after
0-7 d of age 8 d of age or older
n Regression equation R2 n Regression equation R2
Forehead 77 y = 0.8021x + 0.9449 0.3848 200 y = 0.9951x  1.6844 0.5895
Sternum 64 y = 0.8314x + 0.9328 0.3566 158 y = 0.9705x  1.3133 0.7624
Upper back 31 y = 1.396x  3.9616 0.5716 146 y = 1.0832x  1.8325 0.6477
Lower abdomen 51 y = 0.9363x + 0.6808 0.3570 123 y = 0.9196x  1.3506 0.6786
Waist 24 y = 1.2276x  2.9703 0.7944 133 y = 0.9995x  1.8116 0.7248

81.e1 Kurokawa et al