Acta Paediatr 85: 986-90. 1996

Transcutaneous bilirubinometry in Chinese, Malay and Indian infants K L T a n ' , HP Chia2 and BC Koh3

Tan KL, Chia HP, Koh BC. Transcutaneous bilirubinometry in Chinese, Malay and Indian infants. Acta Plediatr 1996;85:986-90. Stockhlm. ISSN 0803-5253 The objective of this study was to investigate the correlation of transcutaneous bilirubinometry (TcB) and plasma bilirubin concentrations in full-term Chinese, Malay and Indian infants. TcB was performed with the Minolta Airshields bilirubinometer on Chinese, Malay and Indian full- term infants. The readings were taken on the chest (sternum) and forehead (glabella) when capillary blood was sampled for bilirubin determination. Five hundred and forty TcB indices in 253 Chinese infants, 282 in 169 Malay infants, and 182 in 120 Indian infants were obtained over the sternum and forehead. A good correlation between the TcB indices and the bilirubin concentra- tions was observed in Chinese, Malay and Indian infants: r = 0.78 (chest), r = 0.73 (forehead); r = 0.86 (chest), r = 0.84 (forehead); and r = 0.84 (chest), r = 0.82 (forehead). The correlation was just as good when the combined values were evaluated together: r = 0.80 (chest) r = 0.75 (fore- head). In Chinese infants, correlation at values below 250 pmol 1 ' was significantly better than that at values over 250 pmol I- ' r = 0.80 versus r = -0.20, p < 0.00001 (chest), and r = 0.74 versus r = 0.07, p < 0.00001 (forehead). However, a safer cut-off point clinically would be 200 pmol 1 I,

since only relatively few higher bilirubin values were encountered when TcB indices were below 200itmol I - ' . The same pattern was noticed with the other two groups, and the combined group. Thus, TcB provides a non-invasive, cost-effective screening method for significant neonatal jaundice, sparing infants and parents physical and emotional stress, and medical and nursing nersonnel extra work and inconvenience. 0 Chinese, Malay, Indian, i i ) ~ ~ ~ ~ ~ r h i l i r i i I ~ i 1 7 ~ i i , ~ i i ( t , /rmi.siw- tuneouJ hiliruhinornetry

K L Tun, Departnfent of Neonatologj Singupore

A reliable method for monitoring bilirubin con- centrations of infants with neonatal jaundice is essen- tial for the effective management of neonatal hyperbilirubinaemia. Visual evaluation of the degree of hyperbilirubinaemia even with adequate clinical practice and experience often requires laboratory con- firmation. Repeated blood sampling causes pain and blood loss to the infant (though only microsamples are required), as well as parental stress and expense, and inconvenience and extra work to all involved; there is also some delay before the results become available.

Transcutaneous bilirubinometry (TcB), being non- invasive, provides a safe, painless, convenient and inex- pensive alternative, with the added advantage of prompt results. This method has been proven to be fairly reliable as a screening tool (1-3), but not during conventional phototherapy (l-3), due to the bleaching of the infant skin. The natural skin pigmentation of the darker ethnic groups might interfere with the reliability of TcB. Though its reliability has been demonstrated in white and black infants (2, 4), the latter in those studies are usually only lightly pigmented in the neonatal period. To the best of our knowledge, a study of the

Nutiand Universitj2 Hospital, Singuporc 119074, Ripihlic of'

accuracy of TcB in darkly pigmented infants has never been documented before.

Material and methods Healthy full-term infants presenting with neonatal jaun- dice were recruited for this study. The infants were from the three main ethnic groups in Singapore: Chinese, Malay and Indian. The Chinese infants were usually mildly pigmented a t birth, with the Malay infants almost similar o r slightly more brown; the Indian infants, usually of southern Indian parents, were already darkly pigmented a t birth, being generally only slightly lighter than their parents. None of the infants had been exposed to phototherapy or sunlight. Capillary blood was sampled for bilirubin determina- tion when deemed necessary by clinical evaluation, or when jaundice was observed at a n unusually early age. especially the first day of life.

A t capillary blood sampling, TcB using the transcu- taneous bilirubinometer (Minolta Airshields) was done over the chest (sternum) and forehead (glabella) as previously described (3). All the readings together

' ( ' Scandinavian University Press 1996. ISSN 0803-5253

Tcihk. I . Clinical data of the infants studied.

Indian infants

n (M:F) 253 (140:113) 169 (90:79) 120 (61:59)

Chinese infants Malay infants

0. 35.

Birthweights Mean SD

Mean SD

Range

Gestational age (weeks)

Postnatal age (days)

Bilirubin concentration (pmol I - ’ )

c < o.ooo1

Indian

3229 478

38.6 1.5

2- 8 100-360

3210 508

38.2 1.4

2-7 35-300

30x9 517

38.2 1.4

2 8 50 ~ 300

with the capillary blood sampling were performed in the morning. This was repeated if deemed necessary on successive day(s), the same procedure being followed; hence, these repeats would result in the number of indices (readings) exceeding the number of infants

recruited. However, since the time interval was appreci- ably long ( > 24 h), and the bilirubin concentrations had changed, it was deemed that the repeat TcB indices and associated bilirubin values could be accepted as separate distinct entities. On 10 of the infants (4 Chinese, 3 Malays,

Y = 10.08 + 0.04 X I = 0.78 n-540 P < 0 . m 1

30. Malay

25.

Y I 8.05 +O.WX I I 0.87 n I 2 8 2 p < o.ooo1

0. 35

oJ I 0 50 100 150 200 250 300 350 400

BillNbln (JIn’IO111)

Fi,q 1. Correlation curves with 95% confidence intervals for TcB-C indices and bilirubin concentrations for the three ethnic groups.

30{ Chinese _. _- ./.-

r I 0.73 n-640 p < om01

OJ 35,

.I.- .

_/-- # - -

I” I/---- Y = 10.65 + 0.05 X , I 0 1 1

5 n .iii P < o.ooo1

0 0 SO 100 150 200 250 300 350

BlllNbln (JImol/l) 0

Fig. 2. Correlation between TcB-F indices and bilirubin concentrations for the three ethnic groups.

35 1 I 35, I

5

25.

Y-lO.?P+O.@dX I -0.75 n.1003 p < 0.mOl

Indian ..

- Chlnese

25

15

10 Malay

0 8 I-

25.

2 0

Y I 9.81 + 0.04 X r -0.80 n. 1004 P ' O . ~ I 5 -

01 I 0 50 100 150 200 250 300 350 400

Bilirubin Bilirubin

Fix. 3. Superimposition of the three regression curves of the three ethnic groups.

Fig. 4 . Combined regression correlation curves with 9S%, conlidcnce intervals between TcB and bilirubin concentrations for the ihree ethnic groups.

and 3 Indians) studied, 10 consecutive readings were taken at the same site, to evaluate the coefficient of variation.

Our practice of starting phototherapy at bilirubin levels above 255 pmol I - ' , or above 222 pmol 1-' in those below 48 h of age (3, limited the study mainly to bilirubin values below 255 pmol 1-', though some higher values were observed, especially before the start of phototherapy.

The capillary blood samples were placed in labelled red-coloured straws, and kept in a light-proof box until the moment of bilirubin determination. The bilirubin concentration was determined under standard condi- tions by the A 0 Bilirubinometer (American Optical, USA) regularly calibrated against known standards. Direct bilirubin was also determined in random samples as previously described (6). Informed consent was obtained. The study was approved by the Director of Medical Affairs of the National University Hospital.

The data were evaluated statistically. The Pearson correlation coefficient was used to determine the strength of association between the variables. The assessment of the equality of the gradient and intercepts of the linear regression curves was carried out by means of multiple regression analysis.

during and after the study. Over 1000 TcB readings from 542 infants were taken at the chest (TcB-C) and forehead (TcB-F) when capillary sampling for bilirubin determination was performed: 540 from 253 Chinese infants, 282 from 169 Malay infants and 182 from 120 Indian infants. The bilirubin concentrations ranged from 100 to 360pmol I - ' in Chinese, 35 to 300pmol1~' in Malay, and 50 to 300pmol 1 ' in Indian infants, with a few values scattered beyond this range, especially in Chinese infants. Due to our photo- therapy guidelines, the bulk of the values were in the 100-250pmol 1-' range, with relatively few values exceeding 250pmol I - ' . The coefficient of variation of 10 consecutive readings taken over the same site in the 10 infants with bilirubin concentrations of 150- 250pmol 1-' (4 Chinese, 3 Malay and 3 Indian) ranged from 1.4 to 2.5%.

There was good linear correlation between the bilir- ubin concentrations and the TcB-C and TcB-F indices respectively in the Chinese, Malay and Indian infants (Figs 1 and 2). The gradients and intercepts of the regression lines of both bilirubin-TcB-C and biliru- bin-TcB-F indices were statistically different between Malay and Indian infants 0, < 0.01) and between Malay and Chinese infants 0, < 0.OOl); there was no statistical difference between Chinese and Indian infants in the intercepts, but there was a significant difference in the gradient for the bilirubin-TcB-F curve 0) < 0.02). How- ever, since the values overlapped to a great extent (Fig. 3), evaluation of all the combined values was

Results All the infants (Table 1 ) studied were healthy full-term infants in the first 10 days of life; they remained well

attempted; the resultant common regression curves (Fig. 4) demonstrated a correlation highly comparable to those of the individual curves.

Correlation between TcB indices and bilirubin con- centrations below 250 pmol I-’ was good compared with concentrations above 250 pmol I-’ ( r = 0.80 versus r = -0.20, p < 0.00001 (TcB-C); r = 0.74 versus r = 0.07, p < 0.00001 (TcB-F)) in Chinese infants. Cor- relation beyond this level was poor: the scatter of values was wide, with relatively more of the high values falling below the regression curve. Thus, about 68 and 70% of those with bilirubin concentrations above 250 pmol I F ’ had TcB-C and TcB-F indices, respectively, indicating less than this level. With TcB indicating bilirubin levels below 200 pmol I - ’ , the proportion affected was about 23 and 17%, respectively; the values reached a max- imum bilirubin concentration of 250 and 280 pmoll-’ respectivcly, with two “stray” values being even higher in the former. This was also the case in the other two ethnic groups. The common regression curves for the combined group revealed very much the same pattern; with the TcB-C and TcB-F indices indicating bilirubin levels below 200 pmol I - ’ , the proportion affected was 33 and 32%, with the bilirubin concentrations reaching 250 and 280pnol I - ’ , respectively, again with the two stray valucs being much higher in the former.

The validity of the assumption that the repeat mea- surements done at least 24 h apart on some infants could bc treated as distinct separate entities was evaluated. The regression curve of the total number of measure- ments was compared with that using only the first measurement on each infant; the results were highly compardbk ( Y = 0.80 versus r = 0.78 (chest) and Y = 0.74 versus r = 0.73 (forehead)). A second test involved the comparison of the regression curve of the total number of measurements, and that obtained from only one measurement per infant, taken at random in those with repeat measurements; again highly compar- able results ( r = 0.80 versus r = 0.79 (chest) and r = 0.74 versus r = 0.75 (forehead)) were observed. The slopes of the resulting regressions were very similar to those of the regressions which included the repeat measurements.

The direct acting bilirubin concentration did not exceed 15 pmol I - ‘ in any of the samples determined.

Discussion Neonatal jaundice is especially common in infants of the three major races in Singapore (6); this observation was independent of any native herbs consumed (7). Repeated capillary blood sampling for bilirubin deter- mination is frequently required, a procedure attended by infant trauma and pain, and parental distress; furthermore, repeated samplings might lead to serious infections (8). TcB, because of its non-invasive nature, would appear to be ideal in this situation.

The chromogens present in the neonatal skin could

possibly interfere with the accuracy of TcB, especially in Asian infants with their increased skin carotene content, and in particular Indian infants with their obviously marked melanin pigmentation making them distinctly darker than the other two ethnic groups. The Indian infants were only slightly less pigmented than their parents, who are mainly Tamils from southern India, and are much darker than their northern counterparts; they would most probably be much darker than thc group of Indian infants described in an earlier report (9). Those infants were apparently northern Indians with much lighter skin pigmentation; the numbers stu- died were also relatively small. Our study demonstratcd that TcB-C and TcB-F indices correlated closely with bilirubin concentrations in all three ethnic groups despite different degrees of skin pigmentation. This would seem to be the first report of TcB in darkly pigmented Indian infants; TcB correlation with the intensity of jaundice in such dark Indian infants was comparable to those in the other two lighter-coloured groups of infants. This might appear surprising since the infants’ dark colour could conceivably have “masked” the “weaker” yellow of the jaundice. In spite of this, the degree of yellow staining could still be measured. Clin- ical observation of jaundice in such dark infants would also support this concept. Unlike preterm infants whosc skin characteristics change with increasing age during the first week of life, thus affecting the reliability of TcB (lo), the skin of full-term infants remains fairly stable during this period in all three ethnic groups. Good correlation has also been reported for American Black infants of African origin (2, 4), but these infants have a much lighter complexion at birth, which allows the jaundice to be more easily detected.

The observation of a combined common curve for all three ethnic groups with such good correlation being still possible was indeed surprising, despite the differing gradients of the individual curves. The high degree of overlap of all the values made this possible. The differ- ent degrees of pigmentation of the three groups would naturally seem to exclude this possibility; apparently the TcB could differentiate between the yellow colour and the pigmentation. This further explains the good corre- lation of TcB and bilirubin concentrations in Indian infants in spite of the obviously dark pigmentation.

The accuracy of the TcB apparently declined at bilir- ubin concentrations above 250 pmol 1 - I as in other studies (2, I I ) ; because infants with bilirubin concentra- tions above 255 pmol I F ’ were routinely treated with phototherapy ( 5 ) , the number of high bilirubin values was limited, especially for the Malay and Indian infants. Even though correlation was good up to 250 pmol 1- I, the substantial numbers of bilirubin concentrations exceeding this level when TcB indices indicated a bilir- ubin level of below 250 pmol I-’ make this cut-off point unacceptable. However, with the TcB index indicating bilirubin concentrations below 200 mol 1- ’ , relatively few concentration measurements exceed this level,

990 K L Tun et d.

with the maximum reaching 250 p,mol I- ' in TcB-C and 280 pmol I- ' in TcB-F, levels still relatively safe. Such a cut-off point would be less useful, but safer, and would still spare many infants unnecessary capillary blood sampling.

The decreased reliability of TcB at high bilirubin concentrations was mainly the effect of increasing depth of colour from increasing jaundice making further increases not as easily quantifiable as in the milder stage, the difference in colour intensity being much less obvious in the more severe situation. Hence the reduced correlation at high bilirubin concentrations, a limitation of the instrument; the few low TcB indices with bilirubin values above 300 pmol I-' was difficult to explain. However, the number of observations was too small for significant conclusions to be drawn. In spite of this limitation, TcB would still be useful in screening for significant neonatal jaundice. A substantial number of infants would be spared unnecessary laboratory biliru- bin determinations; only when TcB indices indicate significant values need laboratory determinations be performed.

The repeat measurements made in some infants on different days could possibly cause some bias in the results; however, statistical evaluation of the regression curves with and without the repeat measurements con- firmed that inclusion of these measurements did not introduce any bias. Indeed, as the situation and bilir- ubin concentrations had changed, any bias in the read- ings would be very unlikely. The evaluation of the relationship of TcB and bilirubin measurements can therefore include such measurements on successive days on the same infant(s).

TcB can therefore serve as a simple, convenient and labour-saving method for screening for significant neo- natal jaundice, especially in situations where this condition is common ( 5 , 7). Only when TcB indices

ACTA PRDIATR X S (1996)

indicate significant bilirubin levels need laboratory determinations be performed; this will result in greater efficiency and cost-effectiveness.

A~knon.k.c~~rnenrs.~ We would like to thank the Director of Medical ARairs of the National University Hospital for permission lor thia study. the ward medical and nursing personnel for their kind cooperation, L E Lee for laboratory support and J Ang for secretcirial assistance.

References I .

2.

3 .

4.

5.

6 .

7.

8.

9.

10.

11.

Yamanouchi I , Yamauchi Y , Igarashi I . T'ntiiacutancous bilirti- binometry: preliminary studies of non-invasivc tr;inacutiineous bilirubin meter in the Okayama National Hospital. Pediatrics 1980;65: 195-212 Heygi T, Hiatt M, lndyk L. Transcutaneous bilil.iihiii~)iiietry. I Correlations in term infants. J Pediatr 1981:98:454 7 Tan KL. Transcutaneous bilirubinometry i n Chinese and Malay Infants. Acta Paediatr Scand 1982;71:593 6 Hannemann RE, Schreiner RL, DeWitt DP et ;iI. Evaluation of the Minolta Bilirubin Meter as a screening device i n white and black infants. Pediatrics 1982;69: 107 -9 Tan KL. Phototherapy for neonatal jaundice. Acta Pxdiatr 1996;85:277 9 Brown W, Wong HB. Ethnic group differences in plasma bilirubin levels of fullterm healthy Singapore newborna. Pediatrics

Tan KL. Glucose 6 phosphate dehydrogenase stcttus and neonatal jaundice. I98 1;56:874-7 Lilien LD. Harris VJ, Rainamurthy RS. Pildes RS. Nconntnl osteomyelitis of the calcaneous: complication o1'heel pitncturc. J Pediatr I976;88:478-80 Guha DK. Verma KK, Bhatia Set al. A new bedside tcchniyue 01' bilirubin estimation. Indian Pediatr 19842 l:447 5 I Tan KL. Mylvaganum A. Transcutaneous bilirubinomctry i n preterm very low birth weight infuits. Act Pxdiatr S c a d 1988;77:796-80 I Maisels MJ, Conrad S. Transcutaneous bilirubin measurements i n fullterm infants. Pediatrics 1982;70:464 7

3965;36:745 ~9

Received Dec. 5, 1995. Accepted i n revised form March 22, I996