Transcutaneous bilirubin measurement to estimate serum bilirubin in neonates in a multi-ethnic cohort: a literature review

Abstract

Neonatal jaundice (NNJ) is common in Singapore and is usually monitored with serum bilirubin (SB). This paper reviews literature on the accuracy of transcutaneous bilirubin (TcB) measurement compared to SB to monitor jaundice in neonates of major ethnic groups of Singapore, i.e. Chinese, Malay, and Indian. 15 studies with Chinese babies, 5 with Malay babies, and 15 with Indian babies were identified (total of 28 articles, some with more than one race). Correlation coefficient is good for all three races. Chest is a better site for TcB measurement than forehead, though both are good. Infants with TcB levels more than 160–200 μmol/L should have SB checked. While post-phototherapy infants may benefit from TcB 24 hours after cessation of phototherapy, more needs to be done to ascertain its usefulness. Premature babies should have SB measurements. TcB measurement is suitable to monitor NNJ in term and healthy Chinese, Malay, and Indian babies.

Keywords

Transcutaneous bilirubin neonatal jaundice hyperbilirubinemia serum bilirubin Singapore

Introduction

Neonatal jaundice (NNJ) is common in neonates in Singapore. The majority of monitoring of NNJ is done in the Polyclinics, Hospital Emergency Departments, and postnatal nurseries. This is done by heel prick to measure serum bilirubin (SB). However, a more efficient way is by transcutaneous bilirubin (TcB) measurement. First described by Yamanouchi et al.,¹ there have been numerous studies supporting the use of TcB measurement for NNJ monitoring. Besides cost and time savings, neonates are spared of frequent heel pricks.

NICE has introduced recommendations regarding devices for measuring TcB,² following detailed literature review. However, among the 31 articles reviewed, only 3 studies had Chinese, Malay, or Indian neonates. A few had “Asian” sub-group populations, but were not specific about the actual race.

This topic review examines the evidence available for the use of TcB measurement in Chinese, Malay, and Indian neonates (the three major races in Singapore).

Aim

Primary aim

To review the correlation coefficient (r) of TcB compared with SB at various sites of TcB measurement in Chinese, Malay, and Indian neonates.

Secondary aim

To review the action levels, coefficient of variation, inter-ethnic group variation, effects of phototherapy, and prematurity on TcB and nomograms to predict risk of developing severe hyperbilirubinemia.

Action levels

Action levels are useful in translating the TcB level above which babies with NNJ should be admitted or subject to SB confirmation, and the TcB level below which babies can be safely discharged without immediate risk.

In the management of neonatal jaundice using TcB, sensitivity and negative predictive value (NPV) are more important than specificity and positive predictive value, respectively.

Coefficient of variation

Repeated measurements of TcB on the same subject, at the same site, at the same setting should show similar measurements. Ideally, there should be 0% variation.

Inter-ethnic group variation

Studies that included all 3 ethnic groups provide good inter-ethnic group comparison.

Phototherapy

The American Academy of Pediatrics (AAP) recommended that “Because phototherapy ‘bleaches’ the skin, both visual assessment of jaundice and TcB measurements in infants undergoing phototherapy are not reliable.”³ However, this statement is re-examined in this review.

Premature babies

The NICE guidelines recommended to “always use serum bilirubin measurement to determine the bilirubin level in babies less than 35 weeks gestational age.”² This statement was also re-examined in this review.

Predicting risk of developing severe hyperbilirubinemia

The bilirubin values in the first few hours of life may predict which babies are most at risk of developing severe hyperbilirubinemia, requiring further treatment such as phototherapy or exchange transfusion to prevent complications. This may be in the form of nomograms, which charts hour specific bilirubin levels. This was first described by Bhutani et al.⁴

Methodology

Search for articles were done using PubMed, Google, hand searches from journals, and cross-referencing from articles. Search words included “transcutaneous bilirubin measurement,” “transcutaneous bilirubin,” “transcutaneous bilirubinometer,” “transcutaneous bilirubinometry,” “jaundice meter,” “BiliChek,” “JM-101,” “JM-102,” and “JM-103” as text words and corresponding MeSH terms. There were no limitations on date. It was limited to articles in English language.

Study selection

Articles were included if they involved Chinese, Malay, or Indian neonates with TcB measurements. Individual studies, review articles and guidelines were included.

Statistical power of the studies was calculated with Statistical Package for Social Sciences (SPSS), version 22.0, based on sample size and correlation coefficient (effect size). As the power of all studies was 0.99 and above, no studies were excluded based on sample size.

As the measurements were objective (TcB and SB), risk of assessor bias was deemed to be low and was not considered in the selection of the studies.

Articles (or portions of articles) were excluded if TcB measurement was compared to investigations other than laboratory bilirubin measurement.

Statistical analysis

Meta-analyses were performed using SPSS v22.0. Correlation coefficients and their confidence intervals were derived using the Hunter–Schmidt method and I-square statistic was calculated to quantify heterogeneity across studies. Forest plots were created using MS Excel.

Results

Search results

The search was carried out in March 2013 and updated in February 2014. The articles were reviewed and relevant articles were identified; 273 articles were identified. However, only 28 were eligible for review. 15 articles included Chinese, 5 included Malay, and 15 included Indian neonates. Three articles had more than one race included. The articles reviewed are summarized in Tables 1 to 3.^5–32 Full text was not obtained for 3 articles.^11,14,15 They were analyzed from abstracts,^11,14,15 together with additional information from the NICE review.^11,15

Table 1.

Studies of Chinese neonates.

Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	Results					Other relevant findings / comments by authors
Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	r, Correlation coefficient	Sn	Sp	PPV	NPV	Other relevant findings / comments by authors
Yu et al.,⁵ 2014, China	Healthy term and late-preterm infants	9174	JM-103	Forehead and mid-sternum Mean of both were calculated		Total serum bilirubin (TSB) measurement using UNI-STAT reflectance bilirubinometer		Not mentioned.					Objective was to validate a previously constructed TcB nomogram.Multicenter study in 8 hospitals of China.Combining TcB nomogram and clinical risk factors could improve the predictive accuracy for severe hyperbilirubinemia, which was not assessed in the study
Yu et al.,⁶2010,China	Healthy neonates (gestation age ⩾ 35 weeks and birth weight ⩾ 2000 g).Excluded sick babies and those requiring phototherapy.	6035	Minolta JM-103	Forehead and mid-sternum	36921	All TcB values ⩾ 204 μmol/L were confirmed with a TSB measurement by heel stick, using UNISTAT reflectance bilirubinometer within 1-2 hours.	At 40^th centile,Sn = 100%,NPV = 100%The 75^th percentile curve of the nomogram may be an ideal cutoff point for intensive follow-up of the neonate for hyperbilirubinemia as it carries very high sensitivity (78.7%) and NPV (98.5%).	TcB < 250 μmol/L,r = 0.92TcB ⩾ 250 μmol/L,r = 0.51(these data had beenpublished in authors’ other paper)					Objective was to construct a TcB nomogram.
Lam et al.,⁷2008,Hong Kong	Chinese neonates (>35 weeks gestation)3–7 days old.A&E dept in a regional hospital	113	Minolta JM-103	3 mid-sternum and 3 forehead(per neonate)		TSB using Unistat bilirubinometer (Model 10311C) within 5 min.	<230 μmol/L can return home. (100% sensitivity)>298 μmol/L, admit without TSB.230–298 μmol/L, confirm with TSB.	Mean TcB = 0.80Forehead (max) = 0.79Sternum (max) = 0.80Overall (max) = 0.83(highest with overall max TcB reading)	100% at TcB 230 μmol/L	100% at TcB 298 μmol/L			TSB at (or more than) 250 μmol/L was the admission criteria.The mean difference between transcutaneous and total serum bilirubin was 14 μmol/L (standard deviation, 28 μmol/L; p < 0.001).The JM-103 tended to overestimate TSB. The 95% limits of agreement were between −40 and 69 μmol/L.
Boo and Ishak,⁸2007,Malaysia	Healthy jaundiced term infants	102	BiliChek	Forehead and sternum		Venous blood for TSB within 30 minutes using diazo method using the Cobas Integra system	TcB underestimated TSB more markedly above 250 μmol/L (average of TcB and TSB).	Forehead = 0.84 (p < 0.0001)Sternum = 0.86 (p < 0.0001)Overall (Malay, Chinese and Indian infants)forehead = 0.80 (p < 0.0001),sternum = 0.86 (p < 0.0001)					Study also has Malay and Indian infants (see Malay and Indian neonates tables).Original objective was to determine the sensitivity and specificity of different levels of bilirubin measured by the transcutaneous bilirubinometer BiliChek on forehead and sternum for predicting severe hyperbilirubinemia of TSB ⩾ 300 μmol/L in Malay, Chinese, and Indian infants.Total of 95 infants (27.5%) (Chinese, Malay, and Indians) had TSB ⩾ 300 μmol/L.
Ho et al.,⁹2006,Hong Kong	Chinese neonates with gestation above 32 weeks, 2–9 days old, with neonatal jaundice.	83(77 term babies,6 near-term with gestation 32–37 weeks)	Minolta JM-102	forehead, sternum	Forehead = 100,sternum = 99	Blood samples taken within half an hour, processed within 1 h.Bedside: AO Unistat bilirubinometer or Lab: Diazo reaction.		Forehead = 0.718,sternum = 0.814					Study objective was to verify usefulness of BiliChek compared with Minolta JM 102.Serum bilirubin > 250 μmol/L is the level commonly used for initiation of treatment such as phototherapy (in the investigators’ setting).For reproducibility of tests performed on seven newborns, coefficients of variation for Minolta AirShields JM-102 ranged from 0.61% to 2.85%, slightly better than that of BiliChek (2.63–6.85%).
Ho et al.,¹⁰2006,Hong Kong	Chinese newborns, healthy, term and near term (gestational age > 35 weeks).Not received phototherapy.	997	Minolta JM-103	Mid-sternal area	997 TcB readings were paired with TSB	TSB done with Unistat reflectance bilirubinometer within 30 minutes(only done if initial TcB is more than 40th centile level in Bhutani nomogram; i.e. if NOT low risk of getting severe hyperbilirubinemia subsequently).	< 200 μmol/L	r = 0.83 (p < 0.001).					Bhutani nomogram is an hour-specific bilirubin nomogram to predict, in the first few hours of life, which babies are most at risk of getting severe hyperbilirubinemia.JM-103 bilirubinometer overestimated TSB by a mean of 21.7 μmol/L in our Chinese newborns, which was significant (p = 0.001).Combining the use of TcB and the 75th centile in Bhutani’s nomogram as cut-off level could help identify all cases of severe hyperbilirubinemia.
Chang et al.,¹¹2006,Taiwan	Taiwanese neonates (including pigmented neonates).Healthy term and near term babies born in a tertiary hospital	447	Minolta JM-103	forehead, right and left side of the anterior chest wall (3 each), and their mean	447	TSB within 1 hour of TcB.	To detectTSB > 255 μmol/L	0.835 (p < 0.0001)					Abstract and NICE results only.By setting a cut-off TcB value of 9.4 mg/dL (around 161 μmol/L), it was possible to reduce the frequency of blood sampling by up to 31.3% in this study.
Tan and Dong,¹²2003,Singapore	healthyfull-term Chinese infants	240	Minolta AirShields	Forehead (glabella) and sternum.3 readings for each site.Mean reading recorded.		Capillary blood sampling		Chest = 0.88,forehead = 0.80					Study objective was to evaluate the reliability of transcutaneous bilirubinometry during and after phototherapy.Study had 70 neonates exposed to phototherapy. Correlation coefficients:Forehead (non-exposed) = 0.74Forehead (exposed) = 0.60Chest = 0.60Recovery of skin color occurred within 18–24 h after cessation of exposure.Post phototherapy: r = 0.60 (chest), r = 0.60 (forehead)Day 1 post phototherapy: r = 0.80 (chest), r = 0.70 (forehead)Day 2 post phototherapy: r = 0.84 (chest), r = 0.80 (forehead)Day 1 & 2: r = 0.81 (chest), r = 0.76 (forehead)
Tan et al.,¹³1996,Singapore	Healthy full-term infants	253	Minolta / AirShields	forehead, sternum	Forehead = 540,sternum = 540	Capillary blood sampling (taken together)	Less than 250 μmol/L.Safer if cut-off at 200 μmol/L	Forehead = 0.73<250 μmol/L, r = 0.74>250 μmol/L, r = 0.07p < 0.00001;Chest = 0.78<250 μmol/L, r = 0.80>250 μmol/L. r = −0.20					Study also has Malay and Indian infants (see Malay and Indian neonates tables).TcB underestimates at high values (above 250 μmol/L).
Lin et al.,¹⁴1993,Taiwan	Full term newborns during first 6 days of life.	305	unsure	Forehead and sternum.	3217	Serum bilirubin levels		Forehead = 0.82,sternum = 0.86					Abstract only.TcB is a reliable device for monitoring jaundice in Chinese full-term newborns
Tsai and Lu, ¹⁵1988,Taiwan	Term healthy babies >37 weeks and lessthan 7 days old who had jaundice or TSB measurement	98	Minolta JM-102	8 sites: forehead, cheek, sternum, abdomen, upper back, lower back, palm, and sole.	178(paired observations from each of the 8 sites)	Laboratory TSB levels at the same time as TcB measured.	When JMI was:<15, no SB determination was needed>16, SB measurement was indicated;15 or 16, the infant should be observed, with indication for SB measurement by serial JMI data and by clinical judgement	Foreheadr = 0.87, p < 0.001;cheekr = 0.76, p < 0.001;sternumr = 0.78, p < 0.001;for all other sitesr from 0.47 to 0.76					From NICE guidelines and abstract only
Fok et al.,¹⁶1986, Hong Kong	Full term infants	202	Minolta / AirShields	Forehead and mid-sternum.Mean of 3 readings.	605 readings forehead291 mid-sternum	Serum bilirubin with AO Unistat Bilirubinometer. Blood drawn within 15 min of TcB and analyzed within 1 hour.	A SB level ⩾ 255 μmol/L is unlikely to be missed if TcB Index ⩾ 22.	Forehead = 0.86 (p < 0.001);sternum = 0.912 (p < 0.001);post-phototherapy:forehead (shaded) = 0.79;sternum = 0.26					Study includes 57 receiving phototherapy.Coefficient of variation on 6 infants using 20 consecutive readings ranged from 2.2% to 3.5%.
Tan,¹⁷ 1985,Singapore	Chinese infants, first 2 weeks of life	68	Minolta / AirShields	forehead, sternum and upper back	Forehead = 102;sternum = 69;back = 53		not exceeding 200 μmol/L	Forehead = 0.9327<200 μmol/L = 0.905>200 μmol/L = 0.755 (p < 0.05);sternum = 0.9371<200 μmol/L = 0.851>200 μmol/L = 0.849;upper back = 0.8612					Study also had 57 Malay full-term infants (see Malay neonates table) 18 Chinese premature infants, and 9 full-term Chinese infants exposed to phototherapy.High degree of reproducibility was demonstrated.The TcB is a useful screening device for neonatal jaundice.
Boo and Bakar,¹⁸1984,Malaysia	Jaundiced neonates who have not been treated with phototherapy or exchange transfusion	37	Minolta / AirShields	forehead(mean of 3 readings)	37 mean readings	SB (venous blood) in lab immediately before or after TcB was taken.		Forehead = 0.66					Study also had 38 Malay infants (see Malay neonates table) and 30 Indian infants (see Indian neonate table).
Tan,¹⁹1982,Singapore	Full term, first 2 weeks of life, born in KKH.	68	Minolta / AirShields	Forehead, sternum and back (between scapulae)	Forehead = 102,sternum = 68,back = no mention	Capillary bilirubin (AO bilirubinometer) during or immediately after readings were made.	not exceeding 200 μmol/L	Forehead = 0.9327,sternum = 0.9372,back = 0.861					Study also has Malay infants (see Malay neonates table).The coefficient of variation was less than 5% in all the 10 infants in each of whom 10 measurements were made in rapid succession.In Chinese and Malay infants combined:when SB less than 200, r = 0.905;when SB more than 200, r = 0.755 (p < 0.05).

Table 2.

Studies of Malay neonates.

Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	Results					Other relevant findings / comments by authors
Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	r, correlation coefficient	Sn	Sp	PPV	NPV	Other relevant findings / comments by authors
Boo and Ishak,⁸2007,Malaysia	Healthy jaundiced term infants	214	BiliChek	Forehead and sternum		Venous blood for TSB within 30 minutes using diazo method using the Cobas Integra system	TcB underestimated TSB more markedly above 250 μmol/L (average of TcB and TSB).	Forehead = 0.79 (p < 0.0001)Sternum = 0.86 (P < 0.0001)Overall (Malay, Chinese, and Indian infants):Forehead = 0.80 (p < 0.0001)Sternum = 0.86 (p < 0.0001)	See Chinese neonates table to detect TSB ⩾ 300 μmol/L for overall (Malay, Chinese, and Indian infants)				Study also has Chinese and Indian infants (see Chinese and Indian neonates tables).Original objective was to determine the sensitivity and specificity of different levels of bilirubin measured by the transcutaneous bilirubinometer BiliChek on forehead and sternum for predicting severe hyperbilirubinemia of TSB ⩾ 300 μmol/L in Malay, Chinese, and Indian infants.Total of 95 infants (27.5%) (Chinese, Malay, and Indians) had TSB ⩾ 300 μmol/L.
Tan et al.,¹³1996,Singapore	Healthy full-term infants	169	Minolta / AirShields	forehead, sternum	Forehead = 282Sternum = 282	Capillary blood sampling (taken together)	Less than 250 μmol/L.Safer if cut-off at 200 μmol/L	Forehead = 0.84Chest = 0.87					This study also has Chinese and Indian infants (see Chinese and Indian neonates tables).
Tan,¹⁷ 1985,Singapore	Malay infants, first 2 weeks of life	57	Minolta / AirShields	forehead, sternum and upper back	Forehead = 103Sternum = 103Back = 97		not exceeding 200 μmol/L	Forehead = 0.8724<200 μmol/L = 0.872>200 μmol/L = 0.670 (p < 0.01)Sternum = 0.8984(similar situation as forehead for difference between 200 μmol/L as cut-off)Upper back = 0.8666					Study also had 68 Chinese full-term infants (See Chinese neonates table), 18 Chinese premature infants, and 9 full-term Chinese infants exposed to phototherapy.High degree of reproducibility was demonstrated.The TcB is a useful screening device for neonatal jaundice.
Boo and Bakar,¹⁸1984,Malaysia	Jaundiced neonates who have not been treated with phototherapy or exchange transfusion	38	Minolta / AirShields	forehead(mean of 3 readings)	37 mean readings	Serum bilirubin (venous blood) in lab immediately before or after TcB was taken.		Forehead = 0.83					Study also had 37 Chinese infants (see Chinese neonates table) and 30 Indian infants (see Indian neonate table).
Tan,¹⁹1982,Singapore	Full term, first 2 weeks of life, born in KKH.	28	Minolta / AirShields	forehead, sternum, and back (between scapulae)	Forehead = 48Sternum = 28Back = no mention	Capillary bilirubin (AO bilirubinometer) during or immediately after readings were made.	not exceeding 200 μmol/L	Forehead = 0.8835Sternum = 0.8975Back = 0.932					This study also had Chinese infants (See Chinese neonates table).The coefficient of variation was less than 5% in all the 10 infants in each of whom 10 measurements were made in rapid succession.In Chinese and Malays infants combined:when SB less than 200, r = 0.905;when SB more than 200, r = 0.755 (p < 0.05).

Table 3.

Studies of Indian neonates.

Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	Results					Other relevant findings / comments by authors
Study	Patient (relevant to review)	Sample size	Equipment	Site	Number of readings	Reference standard	Threshold for accuracy	r, correlation coefficient	Sn	Sp	PPV	NPV	Other relevant findings / comments by authors
Kaur et al.,²⁰2012,India	Healthy neonates, gestation ⩾35 weeks or birth weight ⩾2000 g.	997	BiliChek	No mention		Capillary or venous blood for SB, using spectrophotometer.	NA	NA					Main objective was to evaluate efficacy of predischarge TcB measurement and clinical risk assessment in predicting hyperbilirubinemia needing treatment.
Mishra et al.,²¹2010,India	Healthy term and late preterm Indian neonates,gestation 35–41 weeks,0–72 hours old	625	BiliChek	Forehead. Average of 5 readings.	925	NIL	NA	NA					Main objective was to provide normative data for TcB measurements in healthy term and late-preterm Indian neonates during first 72 h of age using a multi-wavelength reflectance transcutaneous bilirubinometer.
Mishra et al.,²²2009,India	Northern Indian newborns ⩾35 weeks gestation	55	BiliChek	No mention		Peripheral venipuncture. Two micro blood samples wereassayed, for the average serum total bilirubin, using twin-beam spectrophotometry		r = 0.76, p < 0.001Infants underwent serum bilirubin measurements only if TcB deemed necessary (55 of 314 infants in the TcB arm of this randomized clinical trial).					This was a randomized clinical trial to compare the usefulness of transcutaneous bilirubinometry (compared to visual assessment of bilirubin) to decrease the need for blood sampling to assay TSB in the management of jaundiced healthy Indian neonates.Results used for this topic review are secondary outcomes.
Bhat and Rao,²³2008,India	Term, birth weight between 2500 g to 4000 g, gestational age 37–42 weeks.	461	Not specified, but likely JM as values are in TcB index.	Forehead		NIL	NA	NA					Main objective of study was to assess the utility of 24 and 48 h TcBI for predicting subsequent significant hyperbilirubinemia (⩾ 17 mg/dL, around 291 μmol/L) in healthy term neonates.135 babies (29.3%) had TcBI < 5 at 24 hours and 200(43.3%) had index < 8 at 48 hours. None of them had later hyperbilirubinemia (100 % NPV).
Boo and Ishak,⁸2007,Malaysia	Healthy jaundiced term infants	18	BiliChek	Forehead and sternum		Venous blood for TSB within 30 min using diazo method using the Cobas Integra system	TcB underestimated TSB more markedly above 250 μmol/L (average of TcB and TSB).	Forehead = 0.83 (p < 0.0001)Sternum = 0.94 (p < 0.0001)Overall (Malay, Chinese and, Indian infants):Forehead = 0.80 (p < 0.0001)Sternum = 0.86 (p < 0.0001)	See Chinese neonates table to detect TSB ⩾ 300 μmol/L for overall (Malay, Chinese. and Indian infants)				Study also has Chinese and Malay infants (see Chinese and Malay neonates tables).Original objective was to determine the sensitivity and specificity of different levels of bilirubin measured by the transcutaneous bilirubinometer BiliChek on forehead and sternum for predicting severe hyperbilirubinemia of TSB > 300 μmol/L in Malay, Chinese, and Indian infants.Total of 95 infants (27.5%) (Chinese, Malay, and Indians) had TSB > 300 μmol/L.
Mahajan et al.,²⁴2005,India	Term neonates within first 10 days of life.Babies with major congenital malformations, blood group incompatibility, conjugated hyperbilirubinemia, post exchange transfusion and those dying during the study were excluded from the study group.	28	Minolta AirShields JM-101	2 over forehead and2 over mid sternum.	52	SB from peripheral vein, within 30 min using Wako Bilirubin tester Model SE 101 D II.		Forehead = 0.934(p = 0.001)Sternum = 0.914(p = 0.001)					Term and pre-term neonates within first 10 days of life in original study.Babies appropriate and small for date (by weight) were included.Babies with and without phototherapy were also included.For this review, only term babies, appropriate for date (by weight) and who did not receive phototherapy were included.
Harish and Sharma,²⁵1998,India	2–20 days old.Babies <24 h old or who have undergone exchange transfusion were excluded.	60	Minolta AirShields jaundice meter	Forehead (mean of 5 readings)		Venous blood within 30 min.		r = 0.83 (p < 0.001)	Predictive value of jaundice meter index 18 for SB ⩾ 13 mg/dL (222 μmol/L)				Original study had 145 neonates in total, which also included preterm (r = 0.76) and small for gestational age (r = 0.76) and neonates on phototherapy (r = 0.65).
Harish and Sharma,²⁵1998,India		60	Minolta AirShields jaundice meter	Forehead (mean of 5 readings)		Venous blood within 30 min.		r = 0.83 (p < 0.001)	96.2	58.8	64.1	95.2
Tan et al.,¹³1996,Singapore	Healthy full-term infants	120	Minolta / AirShields	forehead, sternum	Forehead = 181Sternum = 182	Capillary blood sampling (taken together)	Less than 250 μmol/L.Safer if cut-off at 200 μmol/L	Forehead = 0.81Chest = 0.83					This study also has Malay and Indian infants (see Malay and Indian neonates tables).
Kumar et al.,²⁶1994,India	Full term babies beginning with the detection of jaundice, till serum bilirubin fell to normal levels or till the babies required phototherapy.	100	Minolta AirShields JM-101	Forehead (mean of 3 recordings)		SB by conventional diazo method		r = 0.9090(p < 0.001)
Kumar et al.,²⁷1992,India	Full term	100 (serial bilirubin levels)25 (cord blood at birth for SB estimation)	Minolta AirShields JM-101	Forehead (3 readings)	822	SB by diazo method.	From the graph, TcBI readings over-estimate serum bilirubin when SB > 15mg/dL (257 μmol/L)	r = 0.9090 (p < 0.001)					Original study compared filter paper technique (serum bilirubin index, SBI) and TcBI against diazo method.
Sharma et al.,²⁸1988,India	Newborn babies including 80 preterm (less than 37 weeks gestation) and 90 low birth weight (less than 2.5 kg). Birth to 10 days old.	200 (total)	Minolta AirShields JM-101	Forehead (mean of 3 consecutive readings)	200 (mean readings)	Serum bilirubin by SEAC-5267 photo analyzer.		r = 0.87, p ⩽0.001By gestational age,Term = 0.74,Preterm = 0.69(Statistically significant difference, p < 0.001)By birth weight (BW).Normal BW = 0.85Low BW = 0.81(Statistically significant difference, p < 0.001)
Christo et al.,²⁹1988,India	Neonates, excluding only post-exchange transfusion infants.Preterm infants (gestation 32-37 weeks), high bilirubin levels (⩾ 13 mg/dL, around 222 μmol/L) included.	138	Minolta AirShields JM-101	Forehead and sternum (1 each).Mean reading recorded.	138 (mean readings)	Serum bilirubin within half an hour, using Technicon RA1000 autoanalyzer.		All infants (138) = 0.90Not on phototherapy (91) = 0.89On phototherapy (26) = 0.85Term (91) = 0.89Preterm (17) = 0.90Small for gestational age (27) = 0.94High bilirubin (35) = 0.83
Bhat et al.,³⁰1987India	Mature (Term) South Indian neonates, 1–4 days old	63	Minolta AirShields JM-101	Forehead, 3 consecutive readings		SB (automated analyzer using diazo reagent method)		r = 0.72 (term babies)By weight (including preterm babies),Above 2500 gr = 1.03Up to 2500 gr = 0.94Difference was not significant (p > 0.05)		Study also included 37 preterm babies (gestation age <37 weeks)
Boo and Bakar,¹⁸1984,Malaysia	Jaundiced neonates who have not been treated with phototherapy or exchange transfusion	30	Minolta / AirShields	forehead(mean of 3 readings)	30 mean readings	SB (venous blood) in lab immediately before or after TcB was taken.		Forehead = 0.63					Study also had 37 Chinese infants (see Chinese neonate table) and 38 Malay infants (see Malay neonates table).
Guha et al.,³¹1984,India	Infants of all races and various gestational ages.	70 full term infants and 40 preterm infants	Minolta AirShields JM-101	Forehead – mean of 3 readings		SB using diazo method at the same sitting.		r = 0.854	Using TcBI 27 to predict SB 12.9 mg/dL				This study also included filter paper technique (r = 0.9048).
Guha et al.,³¹1984,India	Infants of all races and various gestational ages.	70 full term infants and 40 preterm infants	Minolta AirShields JM-101	Forehead – mean of 3 readings		SB using diazo method at the same sitting.		r = 0.854	59.09%	88.6%	56.5%	89.6%
Narang and Buche, ³²1983,India	Jaundiced babies.	17 term and 18 preterm babies.	Minolta AirShields jaundice meter	Forehead and sternum	TermForehead = 21Sternum = 21PretermForehead = 33Sternum = 33	Plasma bilirubin measured simultaneously by micro method using American Optical bilirubinometer.		Term (17 babies)Forehead = 0.893Sternum = 0.789Preterm (18 babies)Forehead= 0.94Sternum = 0.79Phototherapy (25 of 35 babies)Forehead = 0.77Sternum = 0.84					This study included 18 premature babies. 25 babies out of the 25 received phototherapy.

The main instruments for TcB measurement in NNJ were the Minolta Jaundicemeter (JM-101, JM-102, and JM-103) and BiliChek.

Analysis results

Primary outcome – correlation coefficient

The results are summarized in Table 4 and Figures 1 to 3.

Table 4.

Range of correlation coefficients of TcB and SB in Chinese, Malay, and Indian neonates (to 2 decimal places).

	Chinese	Malay	Indian
Correlation coefficient	0.73–0.94	0.79–0.93	0.72–0.94

Figure 1.

Correlation coefficient of TcB with SB for Chinese neonates.

Figure 2.

Correlation coefficient of TcB with SB for Malay neonates.

Figure 3.

Correlation coefficient of TcB with SB for Indian neonates.

For Chinese neonates, chest was the better site except for one study.¹⁵ The back had the poorest correlation coefficient comparatively.^17,19

For Malay neonates, chest also was better than the forehead.^8,13,17,19 The upper limit suggested for TcB was 200–250 μmol/L as the four studies overlap with the studies for Chinese neonates.^8,13,17,19

For Indian neonates, forehead was better than chest, i.e. 0.934 vs 0.914 and 0.893 vs 0.789.^24,32)

Secondary outcomes

Action levels

Upper limits of TcB for good correlation were reported to be around 160 μmol/L,¹¹ 200 μmol/L,^10,13,17,19 and 250 μmol/L,^6,8,13,16 above which TcB underestimates SB. In one study,¹⁹ there was a sub-analysis of the correlation coefficients between lower and higher bilirubin levels. For forehead readings, when SB was less than 200 μmol/L (n = 58), r = 0.905; when SB was more than 200 μmol/L (n = 44), r = 0.755; p < 0.001 (statistically significant). In this study, for overall forehead readings, Chinese r = 0.93, Malay r = 0.88.

Coefficient of variation

Coefficient of variation for JM-101 was 2.41–4.52% and 2.2–3.5%,^16,17,19 for JM-102 was 0.61–2.85%,⁹ and for BiliChek was 2.63–6.85%;⁹ all were satisfactory.

Inter-ethnic group variation

The results are summarized in Table 5.

Table 5.

Correlation coefficient of TcB and SB in studies which included the three different ethnic groups.

	Singapore(JM-101),¹³ 1996			Malaysia(JM-101),¹⁸ 1984		Malaysia(BiliChek),⁸ 2007
	Forehead	Chest	N/n	Forehead	n	Forehead	Chest	n
Chinese	0.73	0.78	540/253	0.66	37	0.84	0.86	106
Malay	0.86	0.86	282/169	0.83	38	0.79	0.86	220
Indians	0.84	0.84	182/120	0.63	30	0.83	0.94	19

N = number of readings, n = number of neonates.

Overall, Malays had the highest and most consistent correlation coefficient across studies. Chinese had comparatively lower correlation coefficients. However, overall correlation coefficient was good except for the earliest small study.¹⁸

Chest measurements were equal or better than forehead measurements in the two studies that included chest readings.^8,13

However, with BiliChek, Indian neonates had the highest correlation coefficient over chest but Chinese were highest over forehead.

Phototherapy

The results are summarized in Table 6.

Table 6.

Correlation coefficient reduction with phototherapy (to 2 decimal places).

	No phototherapy	Post-phototherapy	Difference	Reference
Chinese (chest)	0.88	0.60	0.28	Tan and Dong¹²
	0.91	0.26	0.65	Fok et al.¹⁶
	0.94	0.71	0.22	Tan^17,19
Chinese (forehead)	0.88	0.74	0.14	Tan and Dong¹²
	0.86	0.79	0.07	Fok et al.¹⁶
	0.94	0.71	0.22	Tan^17,19
Indian (chest)	0.91	0.54	0.38	Mahajan et al.²⁴
Indian (chest)	0.79	0.84	−0.05	Narang and Buche³²
Indian (forehead)	0.93	0.33	0.60	Mahajan et al.²⁴
	0.83	0.65	0.18	Harish and Sharma²⁵
	0.91	0.69	0.22	Kumar et al.²⁶
	0.89	0.85	0.04	Christo et al.²⁹
	0.89	0.77	0.12	Narang and Buche³²

In Chinese neonates, correlation coefficient was reduced in phototherapy in chest and forehead measurements. However, in one study there was rapid return of skin color after 18–24 hours after cessation of exposure to phototherapy.¹² Correlation coefficient increased from post-phototherapy of r = 0.60 (chest), r = 0.60 (forehead), to Day 1 post-phototherapy of r = 0.80 (chest), r = 0.70 (forehead), to Day 2 post-phototherapy of r = 0.84 (chest), r = 0.80 (forehead). In the study, infants without phototherapy (“control”) had r = 0.88 (chest), r = 0.80 (forehead).

Indian neonates also had similar findings on reduction of correlation coefficient with phototherapy. Christo et al. found the least reduction in correlation coefficients (to 0.85 from 0.89).²⁹ Conversely, Narang and Buche reported a rise from 0.79 to 0.84 in forehead readings.³² However, this was a small and preliminary study.³²

There were no studies on Malay neonates.

(v) Premature babies

The results are summarized in Table 7.

Table 7.

Correlation coefficient reduction in premature infants (to 2 decimal places, except Indian (chest) in Narang and Buche).³²

	Term	Premature	Difference	Reference
Chinese (chest)	0.93	0.84	0.09	Tan¹⁷
Chinese (forehead)	0.93	0.63	0.30	Tan¹⁷
Indian (chest)	0.95	0.88	0.07	Mahajan et al.²⁴
Indian (chest)	0.789	0.790	−0.001	Narang and Buche³²
Indian (forehead)	0.91	0.87	0.04	Mahajan et al.²⁴
	0.83	0.76	0.07	Harish and Sharma²⁵
	0.74	0.69	0.05	Sharma et al.²⁸
	0.89	0.90	−0.01	Christo et al.²⁹
	0.89	0.94	−0.05	Narang and Buche³²

Correlation coefficient was reduced in Chinese premature neonates.¹⁷

Indian neonates had mixed reports of correlation coefficient reduction,^24,25,28 together with an increase in premature babies.^29,32 Bhat et al. found a reduction (to 0.52 from 0.72) but not significant (p > 0.05).³⁰ Narang and Buche found an increase in both forehead (0.94 vs 0.893) and chest (0.79 vs 0.789).³² However, it was a small study.

Predicting risk of developing severe hyperbilirubinemia

Sensitivity and NPV were 100% at the 40th centile on nomograms obtained by Yu et al.^5,6 Using Bhutani et al.’s nomogram,⁴ the 75th centile had 100% sensitivity.

For Indian neonates, sensitivity and NPV were 100% at the 25th centile on the nomogram obtained by Mishra et al.²¹ In addition, TcB index <5 at 24 hours and <8 at 48 hours had 100% NPV.²³

Discussion

Primary outcome – correlation coefficient

This review supports the guidelines of NICE and AAP that TcB is suitable to monitor jaundice in neonates,^2,3 within certain limits. In Chinese, Malay, and Indians, the overall correlation coefficient of TcB is good. Although the chest appears to be the most accurate site, it has to be balanced with convenience, as the forehead is more accessible and the difference is not great.

The forest plot for the all three races’ babies shows heterogeneity. This could be due to differences in lab tests used, technique of measurement, setting of studies, and other variations in methodology. However, the correlation coefficient remains good.

Secondary outcomes

Action levels

TcB measurement underestimates SB above 160,¹¹ 200,^10,13,17,19 or 250 μmol/L.^6,8,13,16 Clinically, this means that, at higher levels (which are clinically most important), TcB may miss neonates with high SB, leading to bad outcomes from bilirubin encephalopathy. Thus, clinicians may order SB for neonates with TcB readings above 160–200 μmol/L, the actual value depending on individual centers’ risk management. NICE guidelines recommend that neonates with TcB greater than 250 μmol/L should have SB checked.² AAP recommends that TcB can replace SB particularly for SB less than 15 mg/dL (257 μmol/L).³

Coefficient of variation

The JM-101, JM-102, and BiliChek all showed very good coefficients of variation.

Inter-ethnic variation

While there is paucity of studies using JM-102 and JM-103 on Malay and Indian infants in Singapore, it is likely to be good. This is because Chinese had lower correlation coefficient than Malays,⁸ but studies on Chinese infants with JM-102 and JM-103 showed good correlations.

Phototherapy

Phototherapy reduces correlation of TcB with SB. However, the reduced correlation is generally still good. More can be done to determine when TcB may be useful post-phototherapy.

Premature babies

While correlation coefficient is reduced in premature babies in most studies, it is still good. However, the studies are small. As prematurity is associated with many potential complications, it is difficult to generalize the applicability of TcB, just by gestational age. For instance, hematocrit affects TcB index.¹⁵ There are many variables in premature babies which need to be studied, before determining when TcB can safely replace SB. As such, SB should be used in premature babies, together with other blood tests which they frequently need.

Predicting risk of developing severe hyperbilirubinemia

When a neonate has a bilirubin reading below the level for 100% sensitivity and NPV (of picking up babies at risk of developing severe hyperbilirubinemia) category, according to hour-specific age, the neonate may be safely discharged without further follow-up with regard to NNJ. However, where TcB and/or SB are accessible, it may be safer to continue monitoring all neonates. But if there are other barriers to SB monitoring, those at low risk according to nomograms, can potentially discontinue follow-up. At present, there are no reported nomograms in Singapore for this purpose.

Limitations

Language of articles in this review was limited to English. There may be reports in other languages which were not found, since countries where Chinese, Malay, and Indian neonates predominate the respective population, may not use English as the main language.

Few full articles were not found and information was based on abstracts and secondary sources. However, their findings largely concur with the rest of the articles.

Various centers vary in their definition of action levels, preterm babies, birth weight to define small for age, and their TcB measurement protocol (site and number of measurements per site). Despite that, the findings are generally similar in their respective definitions.

The laboratory SB measurements were performed with different equipment and varying collection methods. This could potentially cause variation in the SB levels, to which the TcB readings were correlated.

Many articles did not report blinding. However, they were also included in this review because both SB and TcB measurements are objective measurements, and less prone to assessor bias.

This review included previous generation bilirubinometers like JM-101 which did not negate the interference of skin pigmentation to derive some conclusion.

Conclusion

Transcutaneous bilirubin (TcB) measurement is a suitable alternative to serum bilirubin (SB) measurement in the monitoring of NNJ in term and healthy infants in Singapore. This would save cost, time and invasive pricks. JM-101, JM-102, JM-103, and BiliChek are suitable equipment to be used, though further validation of JM-102 and JM-103 in Malays and Indians would be useful.

Chest is a better site for TcB measurement than forehead, though both are good. Infants with TcB levels more than 160–200 μmol/L should proceed to have SB checked.

While post-phototherapy infants may also benefit from TcB 24 hours after cessation of phototherapy, more needs to be done to ascertain its usefulness.

Premature babies should have SB determination until stronger evidence of the usefulness of TcB is demonstrated.

Ethnic-specific nomograms in Singapore can be developed to more accurately risk-stratify neonates.

Footnotes

Acknowledgements

The author expresses appreciation for Ms Eileen Koh Yi Ling, Biostatistician, SingHealth Polyclinics, Department of Research, for her assistance in the biostatistics portions of this review.

Declaration of conflicting interest

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Yamanouchi

Yamauchi

Igarashi

. Transcutaneous bilirubinometry: preliminary studies of noninvasive transcutaneous bilirubin meter in the Okayama National Hospital. Pediatrics 1980; 65: 195–202.

National Institute for Health and Clinical Excellence (NICE). Jaundice in newborn babies under 28 days [CG98]. London: NICE, 2010.

American Academy of Pediatrics (AAP). Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation. Pediatrics 2004; 114: 297–316.

Bhutani

Johnson

Sivieri

. Predictive ability of a predischarge hour-specific serum bilirubin for subsequent significant hyperbilirubinemia in healthy term and near-term newborns. Pediatrics 1999; 103: 6–14.

Han

. Validation of transcutaneous bilirubin nomogram for identifying neonatal hyperbilirubinemia in healthy Chinese term and late-preterm infants: a multicenter study. J Pediatr 2014; 90: 273–278.

Dong

Han

. Transcutaneous bilirubin nomogram for predicting neonatal hyperbilirubinemia in healthy term and late-preterm Chinese infants. Eur J Pediatr 2011; 170: 185–191.

Lam

Tsui

Kam

. Evaluation of a point-of-care transcutaneous bilirubinometer in Chinese neonates at an accident and emergency department. Hong Kong Med J 2008; 14: 356–360.

Boo

Ishak

. Prediction of severe hyperbilirubinaemia using the BiliChek transcutaneous bilirubinometer. J Paediatr Child Health 2007; 43: 297–302.

Lee

Chow

. BiliChek transcutaneous bilirubinometer: a screening tool for neonatal jaundice in the Chinese population. Hong Kong Med J 2006; 12: 99–102.

10.

Tsui

. Evaluation of a new transcutaneous bilirubinometer in Chinese newborns. Arch Dis Child Fetal Neonatal Ed 2006; 91: F434–F438.

11.

Chang

Hsieh

Chou

. The effectiveness of a noninvasive transcutaneous bilirubin meter in reducing the need for blood sampling in Taiwanese neonates. Clin Neonatol 2006; 12: 60–63.

12.

Tan

Dong

. Transcutaneous bilirubinometry during and after phototherapy. Acta Paediatr 2003; 92: 327–331.

13.

Tan

Chia

Koh

. Transcutaneous bilirubinometry in Chinese, Malay and Indian infants. Acta Paediatr 1996; 85: 986–990.

14.

Lin

. The clinical application of transcutaneous bilirubinometry in full-term Chinese infants. Zhonghua Min Guo Xiao Er Ke Yi Xue Hui Za Zhi 1993; 34: 69–76.

15.

Tsai

. Clinical evaluation of transcutaneous jaundice meter in full-term newborns. Zhonghua Min Guo Xiao Er Ke Yi Xue Hui Za Zhi 1988; 29: 376–382.

16.

Fok

Lau

Hui

. Transcutaneous bilirubinometer: its use in Chinese term infants and the effect of haematocrit and phototherapy on the TcB index. Aust Paediatr J 1986; 22: 107–109.

17.

Tan

. Transcutaneous bilirubinometry in Chinese and Malay neonates. Ann Acad Med Singapore 1985; 14: 591–594.

18.

Boo

Bakar

. Transcutaneous bilirubinometry in Malay, Chinese and Indian term neonates. Med J Malaysia 1984; 39: 35–37.

19.

Tan

. Transcutaneous bilirubinometry in full-term Chinese and Malay infants. Acta Paediatr Scand 1982; 71: 593–596.

20.

Kaur

Chawla

Pathak

. Predischarge non-invasive risk assessment for prediction of significant hyperbilirubinemia in term and late preterm neonates. J Perinatol 2012; 32: 716–721.

21.

Mishra

Chawla

Agarwal

. Transcutaneous bilirubin levels in healthy term and late preterm Indian neonates. Indian J Pediatr 2010; 77: 45–50.

22.

Mishra

Chawla

Agarwal

. Transcutaneous bilirubinometry reduces the need for blood sampling in neonates with visible jaundice. Acta Paediatr 2009; 98: 1916–1919.

23.

Bhat

Rao

. Transcutaneous bilirubin in predicting hyperbilirubinemia in term neonates. Indian J Pediatr 2008; 75: 119–123.

24.

Mahajan

Kaushal

Sankhyan

. Transcutaneous bilirubinometer in assessment of neonatal jaundice in northern India. Indian Pediatr 2005; 42: 41–45.

25.

Harish

Sharma

. Transcutaneous bilirubinometry in neonates: evaluation of Minolta AirShields jaundice meter. Indian Pediatr 1998; 35: 264–267.

26.

Kumar

Faridi

Singh

. Transcutaneous bilirubinometry in the management of bilirubinemia in term neonates. Indian J Med Res 1994; 99: 227–230.

27.

Kumar

. Micro-invasive management of neonatal bilirubinemia. Indian Pediatr 1992. 29: 1101–1106.

28.

Sharma

Singh

Lodha

. Transcutaneous bilirubinometry in newborns. Indian Pediatr 1988; 25: 757–760.

29.

Christo

Kamath

Aroor

. Transcutaneous bilirubinometry in newborns. Indian Pediatr 1988; 25: 1073–1077.

30.

Bhat

Srinivasan

Usha

. Correlation of transcutaneous bilirubinometry with serum bilirubin in south Indian neonates. Indian J Med Res 1987; 86: 49–52.

31.

Guha

Verma

Bhatia

. A new bed side technique of bilirubin estimation. Indian Pediatr 1984; 21: 447–451.

32.

Narang

Buche

. Evaluation of the Minolta jaundice meter as a screening device in Indian babies: a preliminary communication. Indian Pediatr 1983; 20: 583–585.