You are on page 1of 7

Clinical Chemistry / Evaluation of Transcutaneous Bilirubin

Evaluation of the BiliChek Noninvasive Bilirubin


Analyzer for Prediction of Serum Bilirubin and Risk
of Hyperbilirubinemia
Brad S. Karon, MD, PhD,1 Ann Teske,2 Paula J. Santrach, MD,1 and Walter J. Cook, MD2

Key Words: Bilirubin; Serum bilirubin; Transcutaneous bilirubin

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


DOI: 10.1309/AJCPRX1E3NWCXHMZ

Abstract Jaundice is common in the neonatal period owing to the


We identified clinical and laboratory variables shortened lifespan of neonatal blood cells and other factors.1-3
affecting the relationship between transcutaneous Extremely high levels of bilirubin are toxic to the brain and
and serum bilirubin levels and determined whether may cause an irreversible bilirubin-induced encephalopathy
transcutaneous bilirubin values could be used to known as kernicterus.4 To encourage increased awareness
predict the risk of hyperbilirubinemia. Median bias and detection of severe hyperbilirubinemia, the American
between transcutaneous and diazo serum bilirubin Academy of Pediatrics (AAP) issued revised guidelines for
was 2.0 mg/dL (34.2 µmol/L), while median bias the management of hyperbilirubinemia in term or near-term
between transcutaneous and the Vitros (Ortho Clinical newborn infants in 2004.5
Diagnostics, Rochester, NY) serum bilirubin values The revised guidelines call for transcutaneous (TcB) or
was 1.3 mg/dL (22.2 µmol/L). The mother’s ethnicity, serum (TsB) bilirubin measurement in all jaundiced infants in
the gestational age, and postnatal age did not impact the first 24 hours of life and for infants with jaundice appearing
the relationship between transcutaneous and serum excessive for the infant’s age. To move toward more system-
bilirubin values. In contrast, the serum bilirubin method atic assessment of jaundice, several studies have examined
(diazo vs Vitros) and collection container (clear vs the usefulness of TcB monitors as a means to predict the TsB
amber tube) significantly impacted the relationship concentration.
between transcutaneous and serum bilirubin values. One study found that the BiliChek transcutaneous bili-
Transcutaneous bilirubin was a sensitive but not rubin monitor (Respironics, Marietta, GA) overestimated the
specific predictor of the risk of hyperbilirubinemia TsB on average by 1.1 mg/dL (18.1 µmol/L).6 In contrast,
using a conventional risk nomogram. Because 4 other studies found that the BiliChek monitor slightly
systematic differences between serum bilirubin methods underestimated the TsB level across a wide range of bilirubin
and local laboratory practices impact the relationship concentrations.7-10 Two other studies found that BiliChek
between transcutaneous and serum bilirubin values, the TcB slightly overestimated the TsB at lower serum bilirubin
effectiveness of transcutaneous prediction of the serum concentrations but underestimated the TsB at concentrations
bilirubin risk zone will vary by institution. of more than 11.7 mg/dL (200.1 µmol/L).11,12 It is not clear
what factors explain the variability in the relationship between
BiliChek TcB and TsB.
Despite the AAP recommendations to interpret all biliru-
bin measurements with respect to an infant’s postnatal age,5
most studies of BiliChek accuracy have evaluated the ability
of TcB to predict a TsB level above or below a fixed cut-
off.6,8-12 Nomograms based on TcB values have recently been

976 Am J Clin Pathol 2008;130:976-982 © American Society for Clinical Pathology


976 DOI: 10.1309/AJCPRX1E3NWCXHMZ
Clinical Chemistry / Original Article

published,13,14 but clinical experience with these nomograms described use of the device according to the manufacturer’s
is limited, and it is not clear whether nomograms based on instructions. TcB precision was assessed by repeated measure-
TcB are specific to the device and infant population studied. ment on 40 infants, with an SD of 1.1 mg/dL (18.8 µmol/L) at
In our practice, the optimal use of TcB in screening infants an average TcB value of 12.0 mg/dL (205.2 µmol/L).
would be to use the TcB value and the infant’s postnatal age to
predict the serum bilirubin risk zone based on a conventional TsB Measurement
serum bilirubin nomogram currently in use. Serum samples were obtained by capillary puncture or
One study evaluated the accuracy of the TcB value and venipuncture. For the study, 88 specimens were collected
the infant’s age in hours to predict the risk zone for hyperbili- into plain (no additive) clear serum tubes (Terumo CapiJect,
rubinemia compared with a matching serum value. BiliChek Somerset, NJ), and 89 serum samples were collected using an
TcB, used with postnatal age, could predict a high risk (>95th amber-colored gel barrier serum tube (Terumo CapiJect). The

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


percentile for age) TsB with 100% sensitivity and 66% speci- TsB is routinely measured using a modification of the diazo
ficity.7 However, this study did not examine the sensitivity or method.15 The diazo method used was the Amaresco DPD
specificity of BiliChek TcB for predicting high-intermediate reagent (Amaresco, Solon, OH) or the Roche Total Bilirubin
(>75th percentile for age) TsB levels, which limits the useful- reagent (Roche Diagnostics, Indianapolis, IN) run on a Roche
ness of these findings for many practices. Modular Analytics system. Correlation between the 2 diazo
The goals of the present study were to identify clinical methods based on 100 samples covering the reportable range of
and laboratory variables that impact the relationship between the assay was as follows: Roche T Bili = 0.9767 × Amaresco
TcB and TsB and to define the sensitivity and specificity of DPD – 0.09 mg/dL (1.54 µmol/L), with an r2 of 0.9978.
the BiliChek TcB for predicting high-intermediate (>75th Precision of the diazo method was assessed by replicate (n =
percentile for age) and/or high (>95th percentile for age) TsB 20) measurement of serum with a mean bilirubin value of 21.0
values in a population of term and near-term infants in a well- mg/dL (359.2 µmol/L), demonstrating an SD of 0.2 mg/dL
infant nursery. (3.4 µmol/L) at this level. The Vitros method (Ortho Clinical
Diagnostics, Rochester, NY) demonstrated similar precision.
The level of hemolysis (serum free hemoglobin) in each
Materials and Methods
sample was determined by converting the hemolysis index
Patient Selection measured by the Roche Modular Analytics system into a free
Infants admitted to the well-infant nursery at the Mayo hemoglobin level as described previously.16 When sample
Clinic (Rochester, MN) were eligible for prospective enroll- volume remained after routine serum bilirubin analysis (131
ment in the study if a participating physician ordered a serum of 177 samples), samples were analyzed using the Vitros
bilirubin level to assess risk for hyperbilirubinemia. The study BuBc slide method on a Vitros 250 analyzer (Ortho Clinical
period was August 2006 through July 2007. Recruitment of Diagnostics). Unlike the diazo methods, which measure
infants was not consecutive as not all physicians practicing in bilirubin by colorimetric reaction with 2,5-dichlorophenyl
the nursery were involved in the study. If parents consented diazonium tetrafluoroborate dye, the Vitros BuBc slide uses a
to involvement in the study, a TcB measurement was obtained mordant to partially separate the spectra of unconjugated and
within 30 minutes of blood collection for TsB measurement. conjugated bilirubin, allowing measurement of both species
Only the first bilirubin measurement for any infant was used by reflectance spectrophotometry on a single slide.
for purposes of the study, consistent with the study’s intent to
test the ability of the TcB to screen for risk of hyperbilirubine- Statistical Analysis
mia. Infant gestational age at birth, postnatal age at the time Median bias (TcB minus TsB) was calculated for the
of transcutaneous measurement, and the mother’s ethnicity diazo and Vitros TsB data sets, along with 95% confidence
were recorded for each patient participating. There were 146 interval (CI) of the median bias. Bias was assessed by testing
Caucasian, 19 Asian, 9 Hispanic, and 3 African American the hypothesis that the slope of the regression of TsB on TcB
infants included in the study. The study design was approved was equal to 1 (indicating that the values were identical). A
by the Mayo Clinic Institutional Review Board. significant P value of less than .05 would mean that there was
significant bias between the TcB and TsB values. Generalized
TcB Measurement estimating equations were also used to determine the impact
TcB was measured on the forehead by using 1 of 2 iden- of gestational age, postnatal age in hours, the mother’s ethnic-
tical BiliChek devices. The BiliChek devices were calibrated ity (Caucasian vs non-Caucasian), blood collection technique
with a disposable tip (BiliCal) before each measurement; the (capillary puncture vs venipuncture), hemolysis level, and col-
device displays the average of 5 readings. Nurses performed lection container (clear vs amber) on the relationship between
BiliChek measurements following a training session that TcB and TsB values. Significance of any change in median

© American Society for Clinical Pathology Am J Clin Pathol 2008;130:976-982 977


977 DOI: 10.1309/AJCPRX1E3NWCXHMZ 977
Karon et al / Evaluation of Transcutaneous Bilirubin

bias (ie, Caucasian vs non-Caucasian, capillary puncture vs Bland-Altmann plots of TcB vs diazo TsB zFigure 1z and
venipuncture) was defined as a P value less than .05. Because TcB vs Vitros TsB zFigure 2z are shown. There is a positive
most data sets were not normally distributed, continuous data bias between TcB and TsB that is relatively constant over the
are summarized as median values, with interquartile range and range of bilirubin values measured (Figures 1 and 2).
minimum and maximum values observed. Bland-Altman plots
were used to display the relationship of TcB to TsB across
the range of bilirubin values observed. CIs for sensitivity and 8
7
specificity were calculated by using the Fisher exact test. 6

TcB – Diazo TsB


5
The clinical significance of differences between TcB and 4
TsB was defined by risk level determination according to the 3
2
nomogram produced by Bhutani et al,17 which plots the biliru- 1

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


0
bin level as a function of postnatal age in hours, from 0 to 144 –1
hours of life. For each TcB and TsB measurement, the risk zone –2
–3
(low, low-intermediate, high-intermediate, high) was deter- –4
5 6 7 8 9 10 11 12 13 14 15 16 17 18
mined by using the Web site bilitool.org, which allows the user TcB and Diazo TsB Mean
to determine risk category according to Bhutani et al17 by enter-
ing the bilirubin level and the infant’s age in hours. Bilirubin zFigure 1z Bland-Altman plot of differences between
levels exceeding the 75th percentile for age in hours are consid- BiliChek transcutaneous (TcB) and laboratory serum bilirubin
measurements by diazo method (diazo TsB) vs mean of TcB
ered high-intermediate risk, and bilirubin levels exceeding the
and diazo TsB methods. The best-fit line between bias (TcB
95th percentile for age are considered high risk.
minus TsB) and mean value (TcB + TsB/2) is shown. The
best-fit line was given by y(bias) = 0.10x(mean value) + 0.99
Results mg/dL; r2 = 0.02, indicating a very weak relationship between
TcB bias and bilirubin value.
Relationship Between TcB and TsB
8
7
TcB consistently overestimated concentrations of TsB by 6
TcB – Vitros TsB

the diazo and Vitros methods. The median TcB concentration 5


4
was 12.2 mg/dL (208.6 µmol/L), while median diazo TsB 3
2
bilirubin level was 10.1 mg/dL (172.7 µmol/L) zTable 1z. For 1
the 131 samples analyzed by Vitros, the median Vitros TsB 0
–1
bilirubin level was 10.9 mg/dL (186.4 µmol/L; Table 1). The –2
–3
median bias between the TcB and diazo TsB was 2.0 mg/dL –4
5 6 7 8 9 10 11 12 13 14 15 16 17 18
(34.2 µmol/L; 95% CI, 1.9-2.3 mg/dL [32.5-39.3 µmol/L]),
TcB and Vitros TsB Mean
and the median bias between the TcB and Vitros TsB was
1.3 mg/dL (22.2 µmol/L; 95% CI, 1.0-1.6 mg/dL [17.1-27.4 zFigure 2z Bland-Altman plot of differences between
µmol/L]). Both bias measurements were statistically signifi- BiliChek transcutaneous (TcB) and laboratory serum bilirubin
cant (P < .001). Correlation between methods was calculated measurements by the Vitros method (Vitros TsB) vs mean of
as follows: y(TcB) = 0.89x (diazo TsB) + 3.3 mg/dL (56.4 TcB and Vitros TsB methods. The best-fit line between bias
µmol/L); r2 = 0.65; and y(TcB) = 0.90x (Vitros TsB) + 2.5 (TcB minus TsB) and mean value (TcB + TsB/2) is shown. The
best-fit line was given by y(bias) = 0.12x(mean value) + 0.09
mg/dL (42.8 µmol/L); r2 = 0.66.
mg/dL; r2 = 0.03, indicating a very weak relationship between
TcB bias and bilirubin value.

zTable 1z
Median, Interquartile Range, Minimum, and Maximum Values for Transcutaneous Bilirubin, Serum Bilirubin (Diazo and Vitros
Methods), Gestational Age, Postnatal Age, and Serum Free Hemoglobin

Variable Median Interquartile Range Minimum Maximum

Transcutaneous bilirubin, mg/dL (µmol/L) 12.2 (208.6) 10.7-13.9 (183.0-237.7) 6.4 (109.4) 18.9 (323.2)
Serum bilirubin (diazo), mg/dL (µmol/L) 10.1 (172.7) 8.7-11.5 (148.8-196.7) 5.3 (90.6) 16.6 (283.9)
Serum bilirubin (Vitros), mg/dL (µmol/L) 10.9 (186.4) 9.4-12.3 (160.7-210.3) 5.8 (99.2) 17.9 (306.1)
Gestational age (wk) 390/7 380/7-396/7 326/7 413/7
Postnatal age (h) 48 42-55 8 81
Serum free hemoglobin (g/L) 1.76 1.29-2.33 0.45 13.71

978 Am J Clin Pathol 2008;130:976-982 © American Society for Clinical Pathology


978 DOI: 10.1309/AJCPRX1E3NWCXHMZ
Clinical Chemistry / Original Article

Relationship of Clinical Variables to TcB Minus TsB Bias relationship between the level of free hemoglobin (hemoly-
Generalized estimating equations were used to determine sis) in the serum sample and TcB minus diazo TsB bias (P =
whether the mother’s ethnicity, the gestational age, or postna- .1151). There was a weak relationship that did not approach
tal age in hours impacted the relationship between TcB and statistical significance between free hemoglobin level and
TsB (ie, TcB minus TsB bias). For purposes of data analysis, TcB minus Vitros TsB bias (P = .0657).
the median bias (TcB minus TsB) was analyzed for Caucasian There were 88 samples collected into clear serum tubes
(n = 146) and all non-Caucasian (n = 31) participants. The and 89 samples collected into amber serum tubes, which
median TcB minus diazo TsB bias was not significantly dif- protected samples from light. All samples were analyzed by
ferent between Caucasian and non-Caucasian participants (P the diazo method, and a subset (n = 131) was analyzed on the
= .2207). For the subset having Vitros TsB measured, the Vitros. The median TcB minus diazo TsB bias was 2.2 mg/
median bias (TcB minus Vitros TsB) was not significantly dL (37.6 µmol/L) for the 88 samples collected into clear tubes

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


different between Caucasian and non-Caucasian participants and 2.0 mg/dL (34.2 µmol/L) for the 89 samples collected into
(P = .1003). Thus, the mother’s ethnicity did not contribute to amber tubes (P = .7437; no significant difference between clear
variability between observed TcB and TsB values. and amber tubes). In contrast, the median bias for TcB minus
The median gestational age was 39.0 weeks (Table 1). Vitros TsB was 1.7 mg/dL (29.1 µmol/L) for the 60 samples
There was no relationship between median bias (TcB minus collected into clear tubes and 0.9 mg/dL (15.4 µmol/L) for the
diazo TsB) and gestational age (P = .6026) or between TcB 71 samples collected into amber containers (P = .0119; signifi-
minus Vitros TsB bias and gestational age (P = .7570). The cant difference between clear and amber tubes).
median postnatal age at the time of TcB measurement was
48 hours. The median bias (TcB minus diazo or Vitros TsB) Using TcB to Predict Risk of Severe Hyperbilirubinemia
was not related to the patient’s age of life in hours (diazo, P = One of the goals of the study was to determine whether
.4461; Vitros, P = .9133). the TcB value and the infant’s age in hours could be used
to predict whether the serum bilirubin result fell into high
Relationship of Laboratory Variables to TcB Minus or high-intermediate risk zones (demanding serum bilirubin
TsB Bias measurement and follow-up) or whether the serum bilirubin
There were 126 collections by capillary heel stick, 20 result fell into low-intermediate or low risk zones (potentially
collections by venipuncture, and 31 instances in which the avoiding serum measurement). TcB risk was in a higher risk
collection method was not specified. The collection method zone in 128 (72.3%) of 177 neonates compared with the diazo
(capillary heel stick vs venipuncture) was not related to TcB TsB, while the TcB was in a higher risk zone for 68 (51.9%)
minus diazo bias (P = .6173) or to TcB minus Vitros TsB of 131 neonates compared with the Vitros TsB.
bias (P = .7382). The median free hemoglobin level in the The concordance between risk category determined by
177 serum samples was 1.76 g/L (Table 1). There was no TcB and TsB is shown in zTable 2z and zTable 3z. Sensitivity,

zTable 2z
Concordance Between Transcutaneous and Serum (Diazo) Bilirubin*

Transcutaneous Bilirubin

Serum Bilirubin (Diazo) Low or Low-Intermediate Risk High-Intermediate or High Risk Total

Low or low-intermediate risk 48 72 120


High-intermediate or high risk 1 56 57
Total 49 128 177
* All results were classified as low, low-intermediate, high-intermediate, or high risk based on infant’s age of life in hours and bilirubin level using the Bhutani nomogram.

zTable 3z
Concordance Between Transcutaneous and Serum (Vitros) Bilirubin*

Transcutaneous Bilirubin

Serum Bilirubin (Vitros) Low or Low-Intermediate Risk High-Intermediate or High Risk Total

Low or low-intermediate risk 35 29 64


High-intermediate or high risk 4 63 67
Total 39 92 131
* All results were classified as low, low-intermediate, high-intermediate, or high risk based on infant’s age of life in hours and bilirubin level using the Bhutani nomogram.

© American Society for Clinical Pathology Am J Clin Pathol 2008;130:976-982 979


979 DOI: 10.1309/AJCPRX1E3NWCXHMZ 979
Karon et al / Evaluation of Transcutaneous Bilirubin

specificity, and the 95% CI are given in the following text. The across a wide range of serum bilirubin levels.7-10 These studies
sensitivity of high or high-intermediate TcB for predicting a have used diazo methods,7,8 multiple laboratory methods and
high or high-intermediate diazo TsB was 98% (CI, 90%-100%). high-pressure liquid chromatography,9 or direct photometric
The specificity of TcB for predicting high or high-intermediate methods.10 Populations studied included healthy Hispanic
diazo TsB was only 40% (CI, 31%-49%). If high-intermediate newborns,7 healthy preterm infants,10 and full-term infants
or high TcB was used to predict only high risk zone (>95th determined to be at risk for hyperbilirubinemia.8 It is not clear
percentile for age) diazo TsB, the sensitivity went up to 100% what factors (eg, population studied, serum bilirubin method,
(CI, 75%-100%), but the specificity dropped to 30% (CI, gestational age, or postnatal age) account for variability in the
23%-37%). Assuming that infants with low-intermediate or relationship between TcB and TsB observed between studies.
low-risk TcB results would have been spared a blood draw, Two other studies found that the BiliChek TcB slightly
then 49 (27.7%) of 177 blood draws would have been avoided overestimated the TsB at lower serum bilirubin concentrations

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


by using TcB to screen for serum bilirubin risk zone. but underestimated serum bilirubin values at concentrations of
The sensitivity of high or high-intermediate TcB predicting more than 11.7 mg/dL (200.1 µmol/L).11,12 We found that the
high or high-intermediate Vitros TsB was 94% (CI, 85%-98%). TcB systematically overestimated serum bilirubin values by
The specificity of TcB for predicting high or high-intermediate the diazo and Vitros methods over a wide range of bilirubin
Vitros TsB was 55% (CI, 42%-67%). If high-intermediate or concentrations, which has not been observed in previous stud-
high risk TcB was used to predict only high risk zone (>95th ies using similar serum bilirubin methods.
percentile for age) Vitros TsB, the sensitivity went up to 100% Local differences in the performance of similar serum
(CI, 79%-100%), but the specificity fell to 34% (CI, 25%-43%). bilirubin methods are observed in proficiency testing sur-
Assuming that a low or low-intermediate TcB value would veys. In 1 study, a sample with a reference bilirubin value
have spared a blood draw, then 39 (29.8%) of 131 blood draws (high-performance liquid chromatography) of 19.44 mg/dL
would have been avoided by using TcB screening. (332.42 µmol/L) was sent to 1,743 laboratories for analysis
To determine the optimal relationship between TcB and by the local method used. The range of results reported by
TsB, 1.0 mg/dL (17.1 µmol/L) was subtracted from each laboratories using the Vitros BuBc technology ranged from
of the 177 TcB values obtained, such that the median bias 18.7 to 21.5 mg/dL (319.8-367.7 µmol/L), while values from
between TcB and diazo TsB was 1.0 mg/dL (17.1 µmol/L). laboratories using the Roche Cobas Integra ranged from 18.7
Recalculation of sensitivity and specificity as above resulted to 24.3 mg/dL (319.8-415.5 µmol/L), and differences among
in 95% sensitivity (CI, 85%-99%) and 63% specificity (CI, laboratories using the same assay technology were almost as
54%-72%) for high-intermediate or high risk TcB predicting great as systematic differences between methods.18
high-intermediate or high risk diazo TsB. The sensitivity for Blood sample transport (clear vs amber tubes) has not
detection of high risk diazo TsB was still 100% after subtrac- been studied previously. We found that for the Vitros method
tion of 1 mg/dL (17.1 µmol/L) from each TcB value. In this the bias between TcB and TsB was significantly smaller when
case, 45% of blood draws could be safely avoided by using samples were collected into amber tubes that protected them
the same assumptions as defined above. Subtraction of 1.5 from light. No relationship was found between transport condi-
mg/dL (25.7 µmol/L) or more from each TcB value decreased tion and TcB minus diazo TsB bias. In addition, the relation-
the sensitivity for the detection of high risk (>95th percentile ship between the hemolysis level and TcB minus TsB bias
for age) diazo TsB to less than 100%. Therefore, the optimal was more significant for the Vitros method compared with the
relationship between TcB and diazo TsB for risk category pre- diazo method (although neither reached statistical significance).
diction within our institution occurs when median TcB values Local practices related to blood collection and transport affect
are approximately 1.0 mg/dL (17.1 µmol/L) higher than diazo the relationship between TcB and TsB values in manners that
serum values. will also vary by bilirubin method. Together, systematic differ-
ences in serum bilirubin methods (diazo vs Vitros in our study)
and local differences in the performance of serum bilirubin
assays (eg, calibration issues, specimen collection and transport
Discussion
issues) account for much of the variability in the relationship
between TcB and TsB values observed in previous studies.
Factors Affecting the Relationship Between TcB and TsB
One previous study found that the BiliChek device over- Factors Not Affecting the Relationship Between TcB
estimated serum bilirubin as measured by a direct photometric and TsB
method,6 with a mean bias between TcB and TsB very similar One previous study found that the difference between
to that observed in our study. At least 4 studies have con- TcB and TsB as measured by high-performance liquid chro-
cluded that the BiliChek TcB slightly underestimates the TsB matography was independent of race, gestational age, and

980 Am J Clin Pathol 2008;130:976-982 © American Society for Clinical Pathology


980 DOI: 10.1309/AJCPRX1E3NWCXHMZ
Clinical Chemistry / Original Article

postnatal age.9 Our study extends these findings by observing screening by a noninvasive method at any postnatal age, with
that the difference between TcB and TsB values as measured confirmation in high risk infants and clinical decisions based
by diazo or Vitros is not affected by ethnicity, gestational age, on a measurement that can be standardized across institutions
or postnatal age. A limitation of our study and the previous to a reference method.
investigation9 was the inclusion of a majority of Caucasian
infants. One previous study also found that TcB significantly Conclusion
underestimated TsB in infants with a postnatal age of more We have shown that in our institution, BiliChek TcB
than 80 hours,11 but our study did not include enough infants in measurement is a sensitive but not specific screening method
this category to address this population. The method of speci- for hyperbilirubinemia when TcB values are used in place of
men collection (venipuncture vs heel stick) does not seem to serum values to predict risk category based on the Bhutani
influence the relationship between TcB and TsB values. nomogram. Variables associated with serum bilirubin trans-

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


port and measurement, method of serum bilirubin measure-
TcB to Predict the Risk of Severe Hyperbilirubinemia ment used in the clinical laboratory, and local performance
Only 1 previous study used the BiliChek to predict characteristics of laboratory bilirubin assays are the primary
serum bilirubin risk zone as determined by the nomogram of determinants of the relationship between TcB and TsB val-
Bhutani et al.17 This study found that the sensitivity of high- ues. Until laboratory methods for serum bilirubin are better
intermediate or high risk TcB values for predicting high risk standardized, the safety and efficacy (percentage of blood
(>95th percentile for age) TsB was 100%, with a specificity draws eliminated) of TcB screening for prediction of TsB risk
of 66%.7 These authors compared BiliChek TcB with a diazo category will likely vary within each institution as a function
TsB method and found that the TcB slightly underestimated of local practices in the measurement of serum bilirubin.
the diazo TsB by 0.6 mg/dL (10.26 µmol/L).7 In contrast, we
found that the specificity of high-intermediate or high risk From the Departments of 1Laboratory Medicine and Pathology
TcB for predicting high risk TsB was only 30% (diazo) or and 2Pediatric and Adolescent Medicine, Mayo Clinic, Rochester,
MN.
34% (Vitros) owing to systematic overestimation of the serum
bilirubin level as performed in our institution. The previous Address reprint requests to Dr Cook: Dept of Pediatric and
study7 did not address use of TcB values to predict high- Adolescent Medicine, Mayo Clinic, 200 First St SW, Rochester,
MN 55905.
intermediate or high risk (>75th percentile) serum bilirubin
values. In our institution, the optimal relationship between
TcB and TsB for this type of risk prediction occurs when TcB
values are approximately 1.0 mg/dL (17.1 µmol/L) higher References
than TsB results. This allows for 95% sensitivity and more 1. Jain S. The neonatal erythrocyte and its oxidative susceptibility.
Semin Hematol. 1989;26:286-300.
than 60% specificity for prediction of high-intermediate or
2. Askin D, Diehl-Jones W. The neonatal liver, part III:
high risk TsB values and would allow for a 45% reduction in pathophysiology of liver dysfunction. Neonatal Netw.
the number of blood draws for serum bilirubin measurement. 2003;22:5-15.
Because the relationship between TcB and TsB values will 3. Maisels M. Neonatal jaundice. Pediatr Rev. 2006;27:443-454.
vary by institution and method, this relationship would need 4. Van Praagh R. Diagnosis of kernicterus in the neonatal period.
to be established in every institution and for each serum bili- Pediatrics. 1961;28:870-876.
rubin method used for risk assessment. 5. American Academy of Pediatrics Subcommittee on
Hyperbilirubinemia. Management of hyperbilirubinemia in
the newborn infant 35 or more weeks of gestation [published
Use of TcB in the TsB Nomogram correction appears in Pediatrics. 2004;114:1138]. Pediatrics.
Several previous studies6,8-12 have used TcB to predict 2004;114:297-316.
whether serum bilirubin was above or below a fixed cutoff 6. De Luca D, Zecca E, de Turris P, et al. Using BiliCheck for
value, an approach that is limited because it does not allow for preterm neonates in a sub-intensive unit: diagnostic usefulness
and suitability. Early Hum Dev. 2007;83:313-317.
bilirubin levels to be evaluated in the context of postnatal age,
7. Kolman K, Mathieson K, Frias C. A comparison of
as recommended.5 Several institutions have created indepen- transcutaneous and total serum bilirubin in newborn Hispanic
dent nomograms specifically for TcB values.13,14 However, infants at 35 or more weeks of gestation. J Am Board Fam Med.
TcB measurement is difficult if not impossible to standardize, 2007;20:266-271.
whereas a reference method exists for TsB measurement.18 8. Samanta S, Tan M, Kissack C, et al. The value of BiliCheck as
a screening tool for neonatal jaundice in term and near-term
In addition, transcutaneous nomograms may not be as read- babies. Acta Paediatr. 2004;93:1486-1490
ily available for everyday use as is the Bhutani nomogram 9. Rubaltelli F, Gourley G, Loskamp N, et al. Transcutaneous
via tools such as bilitool.org. The use of TcB as a screening bilirubin measurement: a multicenter evaluation of a new
method for predicting the serum bilirubin risk zone will allow device. Pediatrics. 2001;107:1264-1271.

© American Society for Clinical Pathology Am J Clin Pathol 2008;130:976-982 981


981 DOI: 10.1309/AJCPRX1E3NWCXHMZ 981
Karon et al / Evaluation of Transcutaneous Bilirubin

10. Szabo P, Wolf M, Bucher H, et al. Assessment of jaundice in 15. Malloy H, Evelyn K. The determination of bilirubin with the
preterm neonates: comparison between clinical assessment, two photoelectric colorimeter. J Biol Chem. 1937;119:481-490.
transcutaneous bilirubinometers and serum bilirubin values. 16. Algeciras-Schimnich A, Cook W, Milz T, et al. Evaluation
Acta Paediatr. 2004;93:1491-1495. of hemoglobin interference in capillary heel-stick samples
11. Boo N, Ishak S. Prediction of severe hyperbilirubinemia using collected for determination of neonatal bilirubin. Clin Biochem.
the BiliCheck transcutaneous bilirubinometer. J Paediatr Child 2007;40:1311-1316.
Health. 2007;43:297-302. 17. Bhutani V, Johnson L, Sivieri E. Predictive ability of a
12. Grohmann K, Roser M, Rolinski B, et al. Bilirubin predischarge hour-specific serum bilirubin for subsequent
measurement for neonates: comparison of 9 frequently used significant hyperbilirubinemia in healthy term and near-term
methods. Pediatrics. 2006;117:1174-1183. infants. Pediatrics. 1999;103:6-14.
13. Maisels MJ, Kring E. Transcutaneous bilirubin levels in the 18. Lo S, Doumas B, Ashwood E. Bilirubin proficiency testing
first 96 hours in a normal newborn population ≥35 weeks' using specimens containing unconjugated bilirubin and human
gestation. Pediatrics. 2006;117:1169-1173. serum. Arch Pathol Lab Med. 2004;128:1219-1223.

Downloaded from https://academic.oup.com/ajcp/article-abstract/130/6/976/1760463 by guest on 01 February 2019


14. De Luca D, Romagnoli C, Tiberi E, et al. Skin bilirubin
nomogram for the first 96 h of life in a European
normal healthy newborn population, obtained with
multiwavelength transcutaneous bilirubinometry. Acta Paediatr.
2008;97:146-150.

982 Am J Clin Pathol 2008;130:976-982 © American Society for Clinical Pathology


982 DOI: 10.1309/AJCPRX1E3NWCXHMZ