Reliability of Transcutaneous Bilirubin Devices inPreterm Infants: A Systematic Review

abstractBACKGROUND AND OBJECTIVE: Transcutaneous bilirubin (TcB) devicesare widely used for the estimation of serum bilirubin levels in term andnear-term infants. Our objective was to review the diagnostic accuracyof TcB devices in preterm infants.

METHODS: Medline, Embase, Cochrane library, Cumulative Index toNursing and Allied Health Literature, and Scopus were searched (fromdatabase inception date until December 2012). Additional citationswere identified by using the bibliographies of selected articles andfrom conference proceedings. The studies were included if theycompared TcB with total serum bilirubin in preterm infants beforephototherapy and presented data as correlation coefficients or as Bland-Altman difference plots. Data were extracted by 1 reviewer and checkedfor accuracy by the second reviewer. An assessment tool (qualityassessment of diagnostic accuracy studies) was used for risk of biasassessments.

RESULTS: Twenty-two studies met the inclusion criteria; 21 studiesreported results as correlation coefficients, with pooled estimatesof r = 0.83 for each site of measurement. Pooled estimates ininfants ,32 weeks’ gestation were similar to the overall pretermpopulation (r = 0.89 [95% confidence interval: 0.82–0.93]). For the 2commonly used TcB devices (ie, JM103 and BiliCheck), the resultswere comparable at the forehead site, although the JM103 deviceexhibited better correlation at the sternum. Analysis of the Bland-Altman plots (13 studies) revealed negligible bias in measurementat the forehead or sternum site by using either the JM-103 orBiliCheck device; however, the JM-103 device exhibited betterprecision than the BiliCheck (SD for TcB – total serum bilirubindifferences: 24.3 and 31.98 mmol/L, respectively).

CONCLUSIONS: The TcB devices reliably estimated bilirubin levels inpreterm infants and could be used in clinical practice to reduce bloodsampling. Pediatrics 2013;132:871–881

AUTHORS: Gaurav Nagar, MD,a Ben Vandermeer, MSc,b

Sandra Campbell, MLS,c and Manoj Kumar, MD, MSca

aDivision of Neonatology, Department of Pediatrics, bAlbertaResearch Center for Health Evidence, and cJohn W. Scott HealthSciences Library, University of Alberta, Edmonton, Alberta,Canada

KEY WORDSBland-Altman difference plots, correlation coefficient, meta-analysis, systematic review, transcutaneous bilirubin

ABBREVIATIONSCI—confidence intervalr—correlation coefficientTcB—transcutaneous bilirubinTSB—total serum bilirubin

Dr Nagar contributed to all stages of the review and wrote thefirst draft of the manuscript; Mr Vandermeer providedstatistical support to the project and helped with data analysis;Ms Campbell was involved with the planning and conducting ofthe literature search for this study; and Dr Kumar contributed toall stages of the review and reviewed all drafts of themanuscript. All authors approved the final manuscript assubmitted.

www.pediatrics.org/cgi/doi/10.1542/peds.2013-1713

doi:10.1542/peds.2013-1713

Accepted for publication Aug 21, 2013

Address correspondence to Manoj Kumar, MD, MSc (ClinicalEpidemiology), Department of Pediatrics, Edmonton ClinicalHealth Academy, Room 3-528, 11405 87 Ave NW, Edmonton, AB,Canada T6G 1C9. E-mail: manojk@ualberta.ca

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2013 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

FUNDING: Dr Nagar was supported by a postgraduate traineegrant from the Women and Children’s Health Research Instituteat the University of Alberta, Edmonton, Alberta, Canada.

POTENTIAL CONFLICT OF INTEREST: The authors have indicatedthey have no potential conflicts of interest to disclose.

PEDIATRICS Volume 132, Number 5, November 2013 871

REVIEW ARTICLE

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

Jaundice is common in the newbornperiod, with a majority of neonatesdeveloping visible jaundice within thefirst few days of birth.1,2 In a significantproportion, this development does notpossess any risks to the newborn;however, up to 10% of term and 25% ofnear-term neonates develop significantserum bilirubin levels requiring treat-ment.3 Optimal diagnosis and manage-ment of neonatal hyperbilirubinemiarequire clinical examination and esti-mation of serum bilirubin. On clinicalexamination, the newborn jaundice isnoted to have cephalocaudal pro-gression.4 However, visual assessmentof serum bilirubin has been shown tocorrelate poorly with measured bili-rubin levels in recent studies.2,5,6 Bloodsampling for estimation of serum bili-rubin is one of the most common testsordered in the neonatal units. Theblood sampling is often done by heelprick and is painful, with potential long-term consequences.7

Transcutaneous bilirubin (TcB) devicesestimate serumbilirubin noninvasively.These devices work by directing lightinto the skin of the neonate and mea-suring the intensity of specific wave-lengths returned.8 The number ofwavelengths varies depending on theTcB device. These devices have beenshown to correlate well with serumbilirubin levels in term and near-terminfants.9–13 The American Academy ofPediatrics recommends the use of TcBdevices for the evaluation of jaundicein infants .35 weeks’ gestation.14

Hyperbilirubinemia in preterm infantsis more prevalent, more severe, and itscourse more protracted than in termneonates,15 likely from slower post-natal maturation of hepatic bilirubinuptake and conjugation mechanisms.16

In addition, delay in initiation of enteralfeedings may limit intestinal motilityand bacterial colonization, resulting inenhancement of bilirubin enterohepaticcirculation. Although the existing

guidelines allow use of TcB devices forthe evaluation of jaundice in term andnear-term neonates,14 the accuracy ofTcB devices for estimation of serumbilirubin in preterm infants remainsunclear.

The objective of the current systematicreview was to assess the diagnosticaccuracy of TcB devices compared withthe total serum bilirubin (TSB) mea-surement in preterm infants during theneonatal period.

METHODS

Search and Selection

Weexecuted a sensitive search strategyof the following databases: Medline,Embase, Cochrane library, CumulativeIndex to Nursing and Allied Health Lit-erature, and Scopus (from the date ofinception of the database to December2012) by using both Medical SubjectHeadings and key words such as: ((expInfant, Newborn/ or exp Intensive CareUnits, Neonatal/ or exp Neonatal Nurs-ing/ or nicu.mp. or exp Infant, Newborn,Diseases/ or exp Infant, Premature/ orneonat*.mp. or exp Neonatal Screening/or exp Premature Birth/ or pre-term.mp. or preterm*.mp. or post-term*.mp.) and (bilitest* or bilimed* or biliblitz*or bilicheck* or bilichek* or tcbr or

icterometer* or bilirubinometer*).mp.and ((exp Hyperbilirubinemia, Neo-natal/ or (bilirubin* or hyperbilirubin*).ti,ab. or exp Hyperbilirubinemia, He-reditary/ or exp Jaundice, Neonatal/)and transcutaneous* or non-invasive*or noninvasiv* or minolta or skin* ortissue).mp.)) and ((blood or capillar* orplasma* or prick* or “heel poke*”or serum or tsbr or tsb).mp. or expBlood/or exp Capillaries/ or exp Serum/or exp Plasma/)). Additional terminol-ogy and predefined database limitswere added to restrict the referencesto those related to infants ,1 monthof age. English language restrictionwas applied. Conference proceedingsand bibliographies of included stud-ies were searched for additionalstudies.

Studies were included in the review ifthey enrolled preterm (,37 weeks’gestation) infants and compared TcBresults with TSB estimation during theneonatal period. We excluded pilotstudies (defined a priori as those en-rolling #20 subjects), studies enroll-ing preterm infants along with terminfants if they did not provide separatepreterm data, and studies evaluatingTcB devices in subjects receiving pho-totherapy (Supplemental Information).

FIGURE 1Flow of studies through the selection process.

872 NAGAR et al

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

TABLE1

Characteristicsof

theIncluded

Studies

Author

(Reference),Year

PopulationCharacteristics;

Ethnicity

M/N

TcBSite

TcB:Device

ComparisonMethod

TSB:Method

Maximum

Interval

BetweenTests,min

Comments

Badiee

etal21(2012)

Iran;Persian

63/63

FBiliCheck

r,BA

DS10

Separatedataprovided

for#30

wk

(18infants)and.30

wk(45

infants)GA

GA(range):25–33

wk

Ahmed

etal20(2010)

UnitedKingdom;m

ixed

ethnicity

183/57

FBiliCheck

r,BA

Diazo

15–30

Results

for1assessmentpersubject

also

provided

GA(range):26–34

wk

Siuetal34(2010)

Hong

Kong;Chinese

110/30

F,S

JM-103

r,BA

DS10

110measurementsateach

site

GA(range):27–34

wk

Stillovaetal36(2009)

Slovakia;w

hite

32/32

F,S,A

JM-103

r,BA

DS10

32measurementsat

each

site

GA(range):25–31

wk

Schm

idtetal33(2009)

UnitedStates;m

ixed

ethnicity

131/90

SJM

-103

r,BA

Diazo

45Results

for1assessmentpersubject

also

provided

GA(3

subgroups):

24–28

wk

39/30

29–31

wk

43/29

32–34

wk

49/31

Karenetal26(2009)

Switzerland;m

ixed

68/51

SBilim

edr,BA

Diazo

15Studyalso

enrolledterm

infants;

show

nhere

aredatafor

preterminfants

GA(2

subgroups):

28–33

wk

21/13

34–36

wk

47/38

Namba

andKitajim

a29(2007)

Japan;Japanese

351/50

FJM

-103

rDS

60Allsubjectshadbirthweight,1500

gGA:,

34wk

Stillovaetal35(2007)

Slovakia;w

hite

44/44

F,S,A

JM-103

r,BA

DS10

44measurementsat

each

site

GA(range):32–34

wk

DeLuca

etal23(2007)

Italy;ethnicitynotmentioned

340/340

FBiliCheck

r,BA

DS10

GA(range):30–36

wk

SanpavatandNuchprayoon3

2(2007)

Thailand;ethnicitynotm

entioned

249/196

FJM

-103

r,BA

DS60

GA(range):30–35

wk

Jangaard

etal25(2006)

Canada;m

ajority

white

65/33

FBiliCheck

BADS

TcBmeasuredbefore

orsoon

afterTSB

Studyhadmultiplegroups;datashow

nhere

areforpreterminfantsnot

receivingphototherapy

GA(range):notstated

Mean6

SD:30.86

2.5wk

Nanjundasw

amyetal31(2005)

UnitedStates;ethnicitynotmentioned

70/70

FBiliCheck

rDS

30Show

nhere

areprephototherapydata

GA(range):25–35

wk

Willem

setal39(2004)

Holland;m

ajority

white

93/24

FBiliCheck

r,BA

Diazo

30Agreem

entstatistic

provided

basedon

1datasetp

ersubject

GA(range):26–29

wk

Szaboetal37(2004)

Switzerland;m

ajority

white

107/69

F,S

BiliCheck

JM-102

r,BA

DSTSBmeasured

soon

afterTcB

107measurementsateach

sitewith

Bilicheck

device.JM-102

onlytested

atsternum(107

measurements)

GA(range):34–36

wk

S

Karolyietal27

(2004)

Germ

any;ethnicity

notmentioned

212/124

SJM

-102

rDiazo

10Allsubjectshadbirth

weight,

1500

g(VLBW)

GA(range):23–33

wk

Yasuda

etal40(2003)

Japan;ethnicity

notm

entioned

75/24

FJM

-103

r,BA

DS30

Studyalso

enrolledterm

infants;data

show

nhere

areforpreterminfants

GA(range):27–36

wk

JM-102

Knüpferetal28(2001)

Germ

any;majority

white

245/135

FBiliCheck

rDiazo

60Datashow

nhere

areforpreterm

infantsnotreceiving

phototherapy

GA(range):23–36

wk

Bhutanietal9(2000)

UnitedStates;m

ixed

ethnicity

163/45

FBiliCheck

rHPLC

30Studyalso

enrolledterm

infants;data

show

nhere

areforpreterminfants

GA(range):35–36

wk

REVIEW ARTICLE

PEDIATRICS Volume 132, Number 5, November 2013 873

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

The outcome of interest was agree-ment statistic between TcB and TSBmeasurements, provided either as thecorrelation coefficient or as the meanand SDs of absolute difference plots(Bland-Altman difference plots).17

Data Extraction and Assessment ofRisk of Bias

Titles, abstracts, and citations wereindependently assessed by 2 reviewersfor inclusion based on predefined se-lection criteria. Data from includedstudies were extracted on a specificallydesigned data extraction form by 1reviewer and checked for accuracy by asecond reviewer. Risk of bias assess-ments were conducted according tothe QUADAS-2 tool.18 This tool consistsof 4 key domains: patient selection,index test, reference standard, andflow and timing. Each study is as-sessed for risk of bias in each of thedomains and for concerns regardingapplicability in the first 3 domains.Disagreements were resolved by con-sensus among the members of the re-view team.

Data Analysis

A meta-analysis was performed on theavailabledata fromboth thecorrelationcoefficients between measurements ofTcB and TSB and the Bland-Altman dif-ference plots. All correlationswerefirstconverted to Fisher z scores beforebeing pooled. The resulting pooledFisher z scores were then transformedback into standard correlation coef-ficients for ease of interpretation. ForBland-Altman difference plots, we pooledthe mean TcB – TSB differences andvariance across eligible studies for esti-mation of bias and SDs, respectively, byusing methods as described by Peytonand Chong.19

A priori subgroup analyses were plan-ned toexplore the influenceof the siteofTcB measurement and the type of TcBdevice used. Data pertaining to all TcBTA

BLE1

Continued

Author

(Reference),Year

PopulationCharacteristics;

Ethnicity

M/N

TcBSite

TcB:Device

ComparisonMethod

TSB:Method

Maximum

Interval

BetweenTests,min

Comments

DonzelliandPratesi24

(2000)

Italy;Caucasian

82/51

SJM

-102

rDS

10Show

nhere

areprephototherapy

data;study

also

provides

data

foroneassessmentp

ersubject

GA(range):24–36

wk

Bhardw

ajetal22(1989)

India;ethnicity

notm

entioned

90/30

FMinolta

Air-Shields

bilirubinmeter

rDS

TCBandTSBmeasured

atthesametim

eThestudyhad3

equalgroupsof

10subjectseach

with

3readingpersubject

GA:,

37wk(range

notstated):3

subgroupswith

meanGA

of29.2,33.2,and35.1wk

TanandMylvaganam

38(1988)

Singapore;Chinese

614/40

F,S

Minolta

Air-Shields

bilirubinmeter

rDS

TSBmeasuredsoon

aftertheTcB

614measurementsateach

site

GA:M

ean6

SD:29.96

2.9wk

Palmer

etal30(1982)

Australia;ethnicitynotm

entioned

(fair-skinnedinfants)

30/30

S,A

Minolta

bilirubinmeter

rDiazo

Notm

entioned

Studyalso

enrolledterm

infants;

datashow

nhere

areforpreterm

infantsnotreceivingphototherapy.

Thirtymeasurementsateach

site

GA:,

37wk

A,abdomen;BA,Bland-Altm

andifferenceplots;DS,directspectrophotom

etry;F,forehead;GA,gestationalage;HPLC,high-perform

ance

liquidchromatography;M,totalnumberofpaired

measurements;N,num

berofsubjects;S,sternum

;VLBW,verylow

birthweight.

874 NAGAR et al

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

devices were included for most of theanalyses; however, for comparison ofindividual TcB devices, we restrictedour analyses to the 2 commonly useddevices in current clinical practice (ie,JM-103 and BiliCheck). Additional sen-sitivity analyseswereplanned toassessthe accuracy of TcB devices in infants,32 weeks’ gestational age and forsingle measurements per enrolledsubject data.

Meta-analyses were performed by us-ing Review Manager Version 5.2 soft-ware (The Nordic Cochrane Centre, TheCochrane Collaboration, 2011; Copen-hagen, Denmark). I2 statistic was cal-culated for each analysis to quantifyheterogeneity across studies. Forestplots were created by using SPlusSoftware version 3.4 (TIBCO SoftwareInc, Palo Alto, CA).

RESULTS

We identified 22 studies9,20–40 providing3527 paired measurements of TcB and

TSB in 1628 patients who fulfilled theinclusion criteria (Fig 1). The baselinecharacteristics of the included studiesare presented in Table 1. Studies variedin terms of the gestational age of theparticipants (23–36 weeks); site of TcBmeasurement (forehead: 16 studies;sternum: 10 studies; abdomen: 3 stud-ies); TcB device used (BiliCheck: 9studies; JM-103: 7 studies; JM-102: 4studies; other devices: 4 studies);method used for serum bilirubin mea-surement (direct spectrophotometry:14 studies; Diazo method: 7 studies;high-performance liquid chromatogra-phy: 1 study); and agreement statisticused for comparison (correlation co-efficient: 21 studies; Bland-Altman dif-ference plots: 13 studies [with 12studies reporting results by bothmethods]).

Results for risk of bias assessments onthe included studies are provided inTable 2. Using the QUADAS-2 tool, themajority of the included studies wereassessed as low risk for bias with

respect to patient selection, index test,reference standard, and flow and tim-ing. Most studies conducted the TcBand TSB estimations within a short in-terval of time (within #30 minutes: 17studies [77%]; within 1 hour: 21 studies[95%]). For applicability, 3 studies22,30,38

were assessed as high risk for appli-cability concerns as an index test be-cause the TcB devices used in thosestudies are no longer used in clinicalpractice.

Meta-analysis of CorrelationCoefficients

Twenty-one studies9,20–24,26–40 providedresults for correlation coefficients, andthe pooled estimates according to thesite of measurement were as follows:forehead9,20–23,28,29,31,32,34–40 (16 stud-ies): 0.83 (95% confidence interval [CI]:0.80–0.86); sternum24,26,27,30,33–38 (10studies): 0.83 (95% CI: 0.76–0.87); andabdomen30,35,36 (3 studies): 0.83 (95%CI: 0.72–0.90) (Fig 2). There were nosubgroup differences with respect to

TABLE 2 Risk of Bias Assessments of the Included Studies

Author (Year) Risk of Bias Applicability Concerns

PatientSelection

IndexTest

ReferenceStandard

Flow andTiming

PatientSelection

IndexTest

ReferenceStandard

Badiee et al21 (2012) ✓ ✓ ✓ ✓ ✓ ✓ ✓

Ahmed et al20 (2010) ✓ ? ✓ ✓ ✓ ✓ ✓

Siu et al34 (2010) ? ? ✓ ? ✓ ✓ ✓

Stillova et al36 (2009) ✓ ? ? ✓ ✓ ✓ ✓

Schmidt et al33 (2009) ✓ ? ? ✓ ✓ ✓ ✓

Karen et al26 (2009) ✓ ? ✓ ✓ ✓ ✓ ✓

Namba and Kitajima29 (2007) ✓ ✓ ✓ ✓ ✓ ✓ ✓

Stillova et al35 (2007) ✓ ? ✓ ✓ ✓ ✓ ✓

De Luca et al23 (2007) ✓ ✓ ✓ ✓ ✓ ✓ ✓

Sanpavat and Nuchprayoon32

(2007)✓ ? ✓ ✓ ✓ ✓ ✓

Jangaard et al25 (2006) ✓ ✓ ✓ ✓ ✓ ✓ ✓

Nanjundaswamy et al31 (2005) ✓ ✓ ✓ ✓ ✓ ✓ ✓

Willems et al39 (2004) ✓ ? ✓ ✓ ✓ ✓ ✓

Szabo et al37 (2004) ✓ ? ✓ ✓ ✓ ? ✓

Karolyi et al27 (2004) ✓ ? ✓ ✓ ✓ ? ✓

Yasuda et al40 (2003) ✓ ✓ ✓ ✓ ✓ ? ✓

Knüpfer et al28 (2001) ✓ ? ✓ ✓ ✓ ✓ ✓

Bhutani et al9 (2000) ✓ ? ✓ ✓ ✓ ✓ ✓

Donzelli and Pratesi24 (2000) ? ✓ ✓ ✓ ✓ ? ✓

Bhardwaj et al22 (1989) ? ✓ ✓ ✓ ✓ X ✓

Tan and Mylvaganam38 (1988) ✓ ? ✓ ✓ ✓ X ✓

Palmer et al30 (1982) ✓ ✓ ✓ ✓ ✓ X ✓

✓, low risk; X, high risk; ?, unclear risk.

REVIEW ARTICLE

PEDIATRICS Volume 132, Number 5, November 2013 875

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

site of TcB measurement. There wassignificant heterogeneity noted in thepooled estimates from the foreheadand sternum sites. In predefined sen-sitivity analysis, when data for a singlemeasurement per enrolled subjectwere used for analysis, the pooledestimates were no more heteroge-neous (forehead site: 0.83 [95% CI:0.79–0.85], I2 = 16%; sternum site: 0.87[95% CI: 0.81–0.90], I2 = 42%) (Fig 3).

Figure 4 shows the meta-analysisresults of correlation coefficients forthe 2 most widely used devices in cur-rent clinical practice. For the BiliCheckdevice, 8 studies9,20,21,23,28,31,37,39 pro-vided data for correlation coefficients(all measured at the forehead site ex-cept for 1 study, which also provided

data for the sternum site),37 and thepooled estimate was 0.83 (95% CI: 0.77–0.88) for measurement at the foreheadcompared with 0.77 (95% CI: 0.68–0.84)at the sternum. For the JM-103 device, 7studies provided data for correlationcoefficients, and the pooled estimatefor measurement at the forehead sitewas 0.85 (6 studies29,32,34–36,40 [95% CI:0.80–0.89]) and for the sternum site, itwas 0.87 (4 studies33–36 [95% CI: 0.82–0.91]).

Figure 5 shows pooled estimates from5 studies20,21,33,36,39 that provided sep-arate data for infants ,32 weeks’gestation. The results show a trendtoward a slightly better correlationcoefficient (r = 0.89 [95% CI: 0.82–0.93])than overall estimates in the preterm

population with no significant het-erogeneity noted. Two of these stud-ies provided separate data forsubjects #28 weeks’ gestation andreported correlation coefficients of0.92 (Schmidt et al33) and 0.94 (Ahmedet al20).

Bland-Altman (TcB – TSBDifferences) Plot Analysis

Thirteen studies20,21,23,25,26,32–37,39,40

provided results as an analysis of theBland-Altman difference plots. Thestudies were published between theyears 2003 and 2012, and the majorityused either JM-103 (6 studies) or Bili-Check (6 studies) devices except forthe study by Karen et al,26 whichused Bilimed. In addition, 2 studies

FIGURE 2Pooled estimates of correlation coefficients according to the site of TcB measurement.

876 NAGAR et al

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

FIGURE 3Pooled estimates of correlation coefficients from the studies reporting data for a single reading/subject.

FIGURE 4Pooled estimates of correlation coefficients for the BiliCheck and JM-103 devices.

REVIEW ARTICLE

PEDIATRICS Volume 132, Number 5, November 2013 877

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

evaluated the JM-102 device along withthe BiliCheck and JM-103.37,40 The re-sults of TcB – TSB differences, alongwith their SDs for individual studies,are listed in Table 3.Table 4 shows pooled data for the biasestimates and precision according tothe study characteristics. The biasestimates were negligible at both the

forehead and the sternum sites, witha comparable precision noted acrossboth sites (pooled estimate of SD atforehead: 29.46 mmol/L; pooled esti-mate of SD at sternum: 26.06 mmol/L).In terms of devices used, the 2 com-monly used TcB devices showed com-parable bias; however, the JM-103 wasnoted to be more precise compared

with the BiliCheck (pooled estimate ofSD: 24.3 and 31.98mmol/L, respectively).

DISCUSSION

We identified a large number of studiesevaluating the diagnostic accuracy ofvarious TcB devices in a pretermpopulation. The results of this reviewfound that TcB measurements corre-late reasonably well with the serumbilirubin estimation in prematureinfants, particularly for the 2 widelyused TcB devices in practice (ie, Bili-Check and JM-103). The accuracy ofthese 2 devices was similar for themeasurement at the forehead site;however, JM103 exhibited better cor-relation with TSB for measurement atthe sternum (P = .02). The analysis ofabsolute TcB – TSB difference plotsrevealed minimal bias in measure-ments, irrespective of the site and thedevice used, although the JM103 de-vice showed a slightly better precisioncompared with the BiliCheck device.When the data from subjects born at#32 weeks’ gestation were analyzedseparately, the results were compa-rable to the diagnostic accuracy in theoverall preterm population.

To the best of our knowledge, this is thefirst systematic review looking at the

FIGURE 5Pooled estimates of correlation coefficients for studies reporting data for infants ,32 weeks’ gestation.

TABLE 3 Results of Bland-Altman Difference Plots for the Studies Providing Those Data

Study TcB Device Measurements/No. of Subjects

Biasa

(mmol/L)Precision(SD)

Site of measurement: foreheadBadiee et al21 2012 BiliCheck 63/63 7.4 25.7Ahmed et al20 2010 BiliCheck 183/57 28.08 31.26Siu et al34 2010 JM-103 110/30 9.3 27.1Stillova et al36 2009 JM-103 32/32 5.1 6.8Stillova et al35 2007 JM-103 44/44 222.4 4.2De Luca et al23 2007 BiliCheck 340/340 218.1 34.2Sanpavat and Nuchprayoon32 2007 JM-103 249/196 5.1 25.7Jangaard et al25 2006 BiliCheck 65/33 23.8 33.57Willems et al39 2004 BiliCheck 93/24 24.92 27.15Szabo et al37 2004 BiliCheck 107/69 28 33Yasuda et al40 2003 JM-103 75/24 7.48 27.2Yasuda et al40 2003 JM-102 75/24 17.1 35.2

Site of measurement: sternumSiu et al34 2010 JM-103 110/30 20.1 27.6Stillova et al36 2009 JM-103 32/32 12 6.8Schmidt et al33 (24–28 wk) 2009 JM-103 39/30 218.7 32.3Schmidt et al33 (29–31 wk) 2009 JM-103 43/29 213.6 22.1Schmidt et al33 (32–34 wk) 2009 JM-103 49/31 217 27.2Karen et al26 (28–33 wk) 2009 Bilimed 21/13 28 38Karen et al26 (34–36 wk) 2009 Bilimed 47/38 16 45.5Stillova et al35 2007 JM-103 44/44 1.7 5.4Szabo et al37 2004 BiliCheck 107/69 10 31Szabo et al37 2004 JM-102 107/69 56 28

a TcB – TSB difference.

878 NAGAR et al

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

diagnostic accuracy of TcB devices ina preterm population. We presentpooled data for bias and precisionestimates along withthe more com-monly usedmeasure of the correlationcoefficient, as the latter typicallydescribes the strength of a relationbetween 2 variables rather than agree-ment between them.17 Thus, the clinicalutility of correlation coefficient data islimited because they intuitively do notprovide information regarding expecteddifferences between the measurementsconducted on a given patient by 2 sep-arate tests.

The pooled estimates of bias noted herearecomparable to theaccuracyof thesedevices in a term population,41–44 inwhom the use of these devices hasbeen shown to result in a marked de-crease in blood sampling for assess-ment of neonatal jaundice.45,46 However,there is a lower threshold for the initi-ation of phototherapy for preterminfants, with certain guidelines pro-viding specific cutoffs for each gesta-tional week according to the postnatalage.47 Thus, the information from thissystematic review should be incor-porated in clinical practice, taking intoconsideration the thresholds for pho-totherapy in preterm infants. In ouropinion, based on the data presentedhere, a TcB reading$50 mmol/L belowthe phototherapy threshold foran infant

could be considered safe for not ini-tiating phototherapy in an otherwisewell preterm, without the need for TSBestimation from the laboratory. Simi-larly, a TcB reading above the photo-therapy threshold may be sufficientgrounds to initiate phototherapywithout the invasive test in most sit-uations. The latter recommendation ismade despite knowing that some ofthese infants may be classified as be-low the phototherapy threshold basedon TSB results because those infantsare still likely to be reasonably close tothe threshold.

Our review is not without limitations.First, we were unable to include a fewstudies that enrolled preterm infantsalong with the term infants. Thesestudies did not provide comparisondata for preterm populations sepa-rately in their publication and uponcontacting the principal author of thestudy. The majority of these excludedstudies enrolled near-term infants inwhom the accuracy of TcB devices isnot debated. Second, the estimatesprovided for the very preterm pop-ulation are based on limited data. Al-though several studies enrolledsubjects #32 weeks’ and #28 weeks’gestation, only a few studies providedcomparison data separately for thesesubpopulations. However, it is reas-suring that the estimates for these

subpopulations were comparable tothe overall estimates in the pretermpopulation. Third, several of the in-cluded studies also provided results ofthe relationship between TcB and TSBas a linear equation with slope andintercept. We did not pool results ofthese data because it would be difficultto interpret such information in clinicalpractice. Fourth, we did not includestudies that provided data for infantsunder phototherapy or postphototherapy.Thus, the results of our review cannotbe applied to those situations. Fifth,we did not apply a formal test tocheck for publication bias; however,the funnel plot of the included studiesdid not reveal any obvious asymmetrysuggestive of missing studies withpoor correlation coefficients. Lastly,several of the studies included in themeta-analysis provided multiple read-ings from each patient enrolled, leadingto the statistical risks of dependency ofdata. We separately compiled studiesthat provided results for 1 data point foreach subject for our sensitivity analysis(Fig 3). The results were similar to theoverall results with no significant het-erogeneity noted.

CONCLUSIONS

The results of this systematic reviewfound that the currently available TcBdevices, particularly JM-103 and Bili-Check, measure TSB values in pretermpopulations with reasonable accuracy,including in infants ,32 weeks’ ges-tational age. The performance of thesedevices in preterm populations issimilar to those in term and near-terminfants. Incorporating the use of TcBdevices in clinical practice, as per oursuggestions outlined here, could helpreduce the need for blood sampling forthe management of neonatal jaundicein preterm infants. The results of thisreview do not apply to preterm infantsundergoing phototherapy or post-phototherapy.

TABLE 4 Pooled Estimates of Bias and Precision for Studies Reporting Data as Bland-AltmanDifference Plots

Variable No. ofStudies

No. ofMeasurements

No. ofSubjects

Biasa

(mmol/L)Precision(SD)

According to site ofmeasurementb

Forehead 11 1361 912 20.06 29.46Sternum 5 424 265 3.8 26.06

According to device usedJM-103 6 827 522 3.14 24.30Bilicheck 6 958 655 1.10 31.98

,32-wk GAdata

4 303 170 1.57 26.86

GA, gestational age.a TcB – TSB difference.b Analysis restricted to data from JM-103 and BiliCheck devices.

REVIEW ARTICLE

PEDIATRICS Volume 132, Number 5, November 2013 879

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

REFERENCES

1. Bhutani VK, Stark AR, Lazzeroni LC, et al;Initial Clinical Testing Evaluation and RiskAssessment for Universal Screening forHyperbilirubinemia Study Group. Pre-discharge screening for severe neonatalhyperbilirubinemia identifies infants whoneed phototherapy. J Pediatr. 2013;162(3):477–482.e1

2. Keren R, Tremont K, Luan X, Cnaan A. Visualassessment of jaundice in term and latepreterm infants. Arch Dis Child Fetal Neo-natal Ed. 2009;94(5):F317–F322

3. Sarici SU, Serdar MA, Korkmaz A, et al. In-cidence, course, and prediction of hyper-bilirubinemia in near-term and termnewborns. Pediatrics. 2004;113(4):775–780

4. Kramer LI. Advancement of dermal icterusin the jaundiced newborn. Am J Dis Child.1969;118(3):454–458

5. Moyer VA, Ahn C, Sneed S. Accuracy ofclinical judgment in neonatal jaundice.Arch Pediatr Adolesc Med. 2000;154(4):391–394

6. Szabo P, Wolf M, Bucher HU, Fauchère JC,Haensse D, Arlettaz R. Detection of hyper-bilirubinaemia in jaundiced full-term neo-nates by eye or by bilirubinometer? Eur JPediatr. 2004;163(12):722–727

7. Anand KJ. Pain, plasticity, and prematurebirth: a prescription for permanent suf-fering? Nat Med. 2000;6(9):971–973

8. el-Beshbishi SN, Shattuck KE, MohammadAA, Petersen JR. Hyperbilirubinemia andtranscutaneous bilirubinometry. Clin Chem.2009;55(7):1280–1287

9. Bhutani VK, Gourley GR, Adler S, Kreamer B,Dalin C, Johnson LH. Noninvasive mea-surement of total serum bilirubin ina multiracial predischarge newborn pop-ulation to assess the risk of severehyperbilirubinemia. Pediatrics. 2000;106(2). Available at: www.pediatrics.org/cgi/content/full/106/2/e17

10. Engle WD, Jackson GL, Stehel EK, SendelbachDM, Manning MD. Evaluation of a trans-cutaneous jaundice meter following hospitaldischarge in term and near-term neonates.J Perinatol. 2005;25(7):486–490

11. Kazmierczak SC, Robertson AF, Briley KP,Kreamer B, Gourley GR. Transcutaneousmeasurement of bilirubin in newborns:comparison with an automated Jendrassik-Grof procedure and HPLC. Clin Chem. 2004;50(2):433–435

12. Maisels MJ, Ostrea EM Jr, Touch S, et al.Evaluation of a new transcutaneous bilirubin-ometer. Pediatrics. 2004;113(6):1628–1635

13. Rubaltelli FF, Gourley GR, Loskamp N, et al.Transcutaneous bilirubin measurement:

a multicenter evaluation of a new device.Pediatrics. 2001;107(6):1264–1271

14. American Academy of Pediatrics Sub-committee on Hyperbilirubinemia. Manage-ment of hyperbilirubinemia in the newborninfant 35 or more weeks of gestation. Pedi-atrics. 2004;114(1):297–316

15. Watchko JF, Maisels MJ. Jaundice in lowbirthweight infants: pathobiology and out-come. Arch Dis Child Fetal Neonatal Ed.2003;88(6):F455–F458

16. Cashore WJ. Bilirubin and jaundice in themicropremie. Clin Perinatol. 2000;27(1):171–179, vii

17. Bland JM, Altman DG. Statistical methodsfor assessing agreement between twomethods of clinical measurement. Lancet.1986;1(8476):307–310

18. Whiting PF, Rutjes AW, Westwood ME, et al;QUADAS-2 Group. QUADAS-2: a revised toolfor the quality assessment of diagnosticaccuracy studies. Ann Intern Med. 2011;155(8):529–536

19. Peyton PJ, Chong SW. Minimally invasivemeasurement of cardiac output duringsurgery and critical care: a meta-analysisof accuracy and precision. Anesthesiology.2010;113(5):1220–1235

20. Ahmed M, Mostafa S, Fisher G, Reynolds TM.Comparison between transcutaneous bilir-ubinometry and total serum bilirubinmeasurements in preterm infants ,35weeks gestation. Ann Clin Biochem. 2010;47(pt 1):72–77

21. Badiee Z, Mohammadizadeh M, Shamee M.Diagnostic usefulness of transcutaneousbilirubinometry in very preterm newborns.Int J Prev Med. 2012;3(4):262–265

22. Bhardwaj HP, Narang A, Bhakoo ON.Evaluation of Minolta jaundicemeter andicterometer for assessment of neonataljaundice. Indian Pediatr. 1989;26(2):161–165

23. De Luca D, Zecca E, de Turris P, Barbato G,Marras M, Romagnoli C. Using BiliCheck forpreterm neonates in a sub-intensive unit:diagnostic usefulness and suitability. EarlyHum Dev. 2007;83(5):313–317

24. Donzelli G, Pratesi S. Transcutaneousbilirubinometry in healthy preterm new-borns. Clin Biochem. 2000;33(6):505–508

25. Jangaard K, Curtis H, Goldbloom R. Esti-mation of bilirubin using BiliChektrade mark,a transcutaneous bilirubin measurement de-vice: effects of gestational age and use ofphototherapy. Paediatr Child Health (Oxford).2006;11(2):79–83

26. Karen T, Bucher HU, Fauchère JC. Comparisonof a new transcutaneous bilirubinometer

(Bilimed) with serum bilirubin measure-ments in preterm and full-term infants.BMC Pediatr. 2009;9:70

27. Karolyi L, Pohlandt F, Muche R, Franz AR,Mihatsch WA. Transcutaneous bilirubinometryin very low birthweight infants. Acta Paediatr.2004;93(7):941–944

28. Knüpfer M, Pulzer F, Braun L, Heilmann A,Robel-Tillig E, Vogtmann C. Transcutaneousbilirubinometry in preterm infants. ActaPaediatr. 2001;90(8):899–903

29. Namba F, Kitajima H. Utility of a newtranscutaneous jaundice device with twooptical paths in premature infants. PediatrInt. 2007;49(4):497–501

30. Palmer DC, Zenner EM, Drew JH. Trans-cutaneous bilirubinometry: use in Australia.Aust Paediatr J. 1982;18(4):273–276

31. Nanjundaswamy S, Petrova A, Mehta R,Hegyi T. Transcutaneous bilirubinometry inpreterm infants receiving phototherapy. AmJ Perinatol. 2005;22(3):127–131

32. Sanpavat S, Nuchprayoon I. Transcutaneousbilirubin in the pre-term infants. J MedAssoc Thai. 2007;90(9):1803–1808

33. Schmidt ET, Wheeler CA, Jackson GL, Engle WD.Evaluation of transcutaneous bilirubinometryin preterm neonates. J Perinatol. 2009;29(8):564–569

34. Siu LY, Siu LW, Au SK, et al. Evaluation ofa transcutaneous bilirubinometer with twooptical paths in Chinese preterm infants.HK J Paediatr. 2010;2(15):132–140

35. Stillova L, Matasova K, Mikitova T, Stilla J,Kolarovszka H, Zibolen M. Evaluation oftranscutaneous bilirubinometry in preterminfants of gestational age 32-34 weeks.Biomed Pap Med Fac Univ Palacky OlomoucCzech Repub. 2007;151(2):267–271

36. Stillova L, Matasova K, Zibolen M, Stilla J,Kolarovszka H. Transcutaneous bilirubinometryin preterm neonates. Indian Pediatr. 2009;46(5):405–408

37. Szabo P, Wolf M, Bucher HU, Haensse D,Fauchère JC, Arlettaz R. Assessment ofjaundice in preterm neonates: comparisonbetween clinical assessment, two trans-cutaneous bilirubinometers and serumbilirubin values. Acta Paediatr. 2004;93(11):1491–1495

38. Tan KL, Mylvaganam A. Transcutaneousbilirubinometry in preterm very low birth-weight infants. Acta Paediatr Scand. 1988;77(6):796–801

39. Willems WA, van den Berg LM, de Wit H,Molendijk A. Transcutaneous bilirubinometrywith the Bilicheck in very premature new-borns. J Matern Fetal Neonatal Med. 2004;16(4):209–214

880 NAGAR et al

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

40. Yasuda S, Itoh S, Isobe K, et al. New trans-cutaneous jaundice device with two opticalpaths. J Perinat Med. 2003;31(1):81–88

41. Lodha R, Deorari AK, Jatana V, Paul VK. Non-invasive estimation of total serum bilirubinby multi-wavelength spectral reflectance inneonates. Indian Pediatr. 2000;37(7):771–775

42. Sanpavat S, Nuchprayoon I. Noninvasivetranscutaneous bilirubin as a screeningtest to identify the need for serum bilirubinassessment. J Med Assoc Thai. 2004;87(10):1193–1198

43. Sanpavat S, Nuchprayoon I. Comparison of twotranscutaneous bilirubinometers—MinoltaAirShields Jaundice Meter JM103 andSpectrx Bilicheck—in Thai neonates. South-east Asian J Trop Med Public Health. 2005;36(6):1533–1537

44. Teran CG, Mohamed T, Casey J. Transcutaneousbilirubinometry: comparison of two multiwave-length devices in healthy term newborns. Eur JPediatr. 2011;170(11):1485–, author reply 1487

45. Grohmann K, Roser M, Rolinski B, et al.Bilirubin measurement for neonates:

comparison of 9 frequently used methods.Pediatrics. 2006;117(4):1174–1183

46. Maisels MJ, Kring E. Transcutaneousbilirubinometry decreases the need forserum bilirubin measurements andsaves money. Pediatrics. 1997;99(4):599–601

47. Recognition and treatment of neonataljaundice. National Institute for Health andCare Excellence (NICE) 2010. Available at:www.nice.org.uk/CG98. Accessed May 29,2013

GENES AND OBESITY: A friend of mine has wrestled with being overweight foryears. She is very careful about her diet, exercises maniacally, and limits heralcohol consumption. Despite these efforts, her BMI has always been . 30. Herweight does not seem to be related to lack of self-control and is not easily un-derstood. My friend’s weight problems may be due to her genes.As reported in The New York Times (July 19, 2013), researchers have known fora long time that some aspects of weight gain are hereditary. Twins raised aparttend to have the same weight, while adopted children tend to have the body massof the biologic parents, not their adopted parents. Evidence has accumulated overtime to suggest that dozens of genes may be involved in increasing appetite, andnew research suggests that at least one gene is associated not only with appetitebut with a change in mammalian metabolic rate.The investigators developed knockoutmice deficient in brain- and body-expressed“melanocortin receptor accessory protein 2” (MRAP2). When allowed to eat asmuch food as desired, the MRAP2 deficient mice were voracious and quicklybecame extremely obese. When MRAP2 deficient mice were fed the same numberof calories as normalmice however, only theMRAP2 deficientmice became obese.The MRAP2 mice had to be fed 10% to 15% fewer calories in order to demonstratethe same weight gain as their normal siblings. How MRAP2 controls weight gainis not entirely understood, but researchers suspect that MRAP2 regulates“melanocortin 4 receptor” (Mc4r), a protein previously implicated in mammalianobesity. Without MRAP2 production, appropriate appetite and energy metabolismregulated by Mc4r is impaired. Interestingly, four children in a registry of 500severely obese children were found to have alterations in the MRAP2 gene whilenone of the healthy controls in the same study did.While still very preliminary, researchers are now looking for alterations in theMRAP2 gene that could lead to partial expression and hence, explain some of thevariance in weight gain among people consuming the same number of calories.While it will not helpmy friend loseweight, I think shewill appreciate learning thatobesity is not always about loss of self-control.

Noted by WVR, MD

REVIEW ARTICLE

PEDIATRICS Volume 132, Number 5, November 2013 881

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

DOI: 10.1542/peds.2013-1713 originally published online October 14, 2013; 2013;132;871Pediatrics 

Gaurav Nagar, Ben Vandermeer, Sandra Campbell and Manoj KumarReview

Reliability of Transcutaneous Bilirubin Devices in Preterm Infants: A Systematic

ServicesUpdated Information &

http://pediatrics.aappublications.org/content/132/5/871including high resolution figures, can be found at:

Supplementary Material

013-1713.DCSupplementalhttp://pediatrics.aappublications.org/content/suppl/2013/10/09/peds.2Supplementary material can be found at:

Referenceshttp://pediatrics.aappublications.org/content/132/5/871.full#ref-list-1This article cites 46 articles, 11 of which you can access for free at:

Permissions & Licensing

https://shop.aap.org/licensing-permissions/in its entirety can be found online at: Information about reproducing this article in parts (figures, tables) or

Reprintshttp://classic.pediatrics.aappublications.org/content/reprintsInformation about ordering reprints can be found online:

ISSN: . 60007. Copyright © 2013 by the American Academy of Pediatrics. All rights reserved. Print American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois,has been published continuously since . Pediatrics is owned, published, and trademarked by the Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from

DOI: 10.1542/peds.2013-1713 originally published online October 14, 2013; 2013;132;871Pediatrics 

Gaurav Nagar, Ben Vandermeer, Sandra Campbell and Manoj KumarReview

Reliability of Transcutaneous Bilirubin Devices in Preterm Infants: A Systematic

http://pediatrics.aappublications.org/content/132/5/871located on the World Wide Web at:

The online version of this article, along with updated information and services, is

ISSN: . 60007. Copyright © 2013 by the American Academy of Pediatrics. All rights reserved. Print American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois,has been published continuously since . Pediatrics is owned, published, and trademarked by the Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it

by guest on February 10, 2018http://pediatrics.aappublications.org/Downloaded from