Introduction

Hyperbilirubinemia needing therapeutic intervention is one of the most common morbidities in preterm neonates [7, 22]. Both developmental immaturity and management practices play a role in causing increased incidence and severity of hyperbilirubinemia. Preterm neonates are also at higher risk of bilirubin-induced brain damage because of lower serum albumin concentration with weaker bilirubin binding sites and more permeable blood–brain barrier [6]. This necessitates periodic monitoring of severity of hyperbilirubinemia by measurement of serum bilirubin. Serum total bilirubin (STB) measurement, although an objective method, is painful, time-consuming, and needs expertise. In addition, significant inter-laboratory and intra-laboratory variability has been reported with measurement of STB [14, 31]. Measurement of transcutaneous bilirubin (TcB) concentration has been recommended for initial assessment of jaundice in term neonates [15, 16]. Due to excellent correlation and agreement between TcB and STB in neonates born at term gestation, TcB measurement-based screening strategy has been associated with reduction in number of invasive blood tests to measure STB and decreased health costs [20]. Although there is some published data on role of TcB measurement in preterm neonates, studies conducted have been limited to white or East Asian neonates and the results have been inconsistent [2–5, 8, 11, 12, 17, 21, 24, 26, 28, 29, 32]. Accuracy of multi-wavelength transcutaneous bilirubinometer has not been proven in Indian neonates who may have darker skin color and higher rate of rise of bilirubin during first week of life [1, 23]. In addition only a few studies have included preterm, very low-birth-weight infants [12, 21, 27, 32]. Effects of postnatal age at measurement of TcB, different sites of measurement, and birth weight on correlation between STB and TcB has also not been reported. Main objective of this study was to evaluate correlation and agreement between TcB and STB among preterm low-birth-weight Indian neonates. In addition, we planned to examine the effect of postnatal age at measurement of TcB, different sites of TcB measurement, and birth weight on correlation between STB and TcB.

Material and methods

This prospective observational study was conducted in a tertiary care teaching hospital of north India from December 2009 to June 2011. The study protocol was cleared by ethics Committee of the hospital and written informed consent was obtained from parents of enrolled infants.

Inclusion and exclusion criteria

Neonates born at less than 35 weeks of gestation and weighing <2,000 g were eligible for enrolment. Neonates with major congenital malformation, gestation <28 completed weeks or birth weight <750 g were excluded. We planned to enroll 100 neonates each in following two groups: gestation <32 weeks with birth weight <1,500 g (VLBW group) and gestation >32 weeks or birth weight 1,500–1,999 g (LBW group). We did not include preterm neonates born at <28 weeks of gestation in the study because this is a distinct group of extremely preterm neonates and we did not expect to be able to enroll sufficient number of these neonates to have a meaningful set of observations.

Intervention and measurement of outcomes

In each enrolled neonate TcB was measured at 24 ± 6 and 72 ± 12 h of postnatal age and when icterus involved arms or legs (Kramer zone 4 or 5 corresponding to bilirubin levels of >12 mg/dL) [13]. Visual assessment of jaundice was done by clinical care team consisting of pediatricians. Rationale behind choosing these different time-points was to ensure inclusion of wide range of serum bilirubin levels in the study. Phototherapy was started when bilirubin reached about 1 % of birth weight [18]. Bilirubin measurements after starting phototherapy were excluded as bleaching effect of phototherapy may interfere with correlation between TcB and STB [30].

TcB was measured with a multi-wavelength transcutaneous bilirubinometer (BiliChek®, coefficient of variation <5 %). The transcutaneous bilirubinometer was placed against the forehead, sternum, and abdomen of the infant in supine position and average of five readings at each of these three sites was recorded. STB was measured by a bedside spectrophotometric microbilimeter (NEO-BIL plus®, das srl, Italy) and by high-performance liquid chromatography (HPLC). STB estimation by spectrophotometry was done for all three TcB measurements. Peripheral venous blood was collected within 30 min of TcB measurement in two pre-heparinized capillaries and serum was separated by centrifugation. Average of the readings in two capillaries was obtained. Internal calibration of spectrophotometer was done daily. External calibration was done at start of the study and then at three monthly interval as per recommendation of the manufacturer.

High-performance liquid chromatography

STB estimation by HPLC was done for one of the three planned TcB measurements in each neonate. TcB measurement in a neonate for which STB was to be measured by HPLC was chosen by simple random selection method. Random number sequence was generated by computer program and sequence was placed in serially number opaque sealed envelopes to be opened at the time of enrolment. A 100-μL aliquot of serum sample was frozen at less than −20 °C for HPLC. Serum bilirubin level was quantified by modified method of Zhang following reverse-phase HPLC procedure [33]. The HPLC technician was blinded to the results of the laboratory STB and the TcB measurements.

Statistical analysis

Data was entered in a Microsoft Access database and analyzed using Stata 11.1 software. Agreement between STB and TcB is presented as median difference between STB and TcB and its interquartile range in subsequent tables and as Bland–Altmann plots in figures. Bland–Altmann plots depict difference between STB (by microbilimeter or HPLC) and STB on Y-axis against mean of STB and TcB on X-axis. Distribution of points representing difference between STB and TcB against the line of zero difference visually represents the extent of agreement between STB and STB. Pair-wise correlation between STB and TcB is represented as correlation coefficient and its P value.

Results

During the study period (from 1st December 2009 to 31st December 2011) a total of 7,780 neonates (mean birth weight: 2,600 ± 588 g; median gestation: 38 completed weeks; interquartile range, 37–39) were born alive in the hospital. Consecutively, born neonates fulfilling inclusion criteria were enrolled in the study. In VLBW group, among 207 neonates born between first and last study enrolment 116 were enrolled (Fig. 1). In LBW group among 210 neonates born between first and last study enrolment 140 were enrolled.

Fig. 1
figure 1

Flow diagram

Full size image

Neonates enrolled in the study were born at a median of 34 completed weeks of gestation (interquartile range, 32–35) with a birth weight of 1,522 ± 288 g (Table 1). As per protocol, neonates were randomized at enrolment to STB estimation by HPLC at 24, 72, or at Jaundice in Kramer zone 4/5. By randomization, 87 (34 %) neonates were assigned to STB estimation by HPLC at 24 h, 84 (33 %) were assigned to HPLC measurement at 72 h and 85 (34 %) were assigned to HPLC measurement at Jaundice zone 4/5.

Table 1 Baseline characteristics
Full size table

STB measurement by microbilimeter was done in all neonates at all the three time-points. A total of 1,619 observations were made in 256 neonates. Overall there was excellent correlation and agreement between TcB and STB-M (Figs. 2 and 3) with TcB on forehead being most accurate (r = 0.84, mean difference, 0.3 ± 1.9 mg/dL) followed by TcB on abdomen (r = 0.73, mean difference, 1.5 ± 2.6 mg/dL) and sternum (r = 0.72, mean difference, 1.5 ± 2.6 mg/dL). TcB performed well at all three points of measurement with best correlations being observed at icterus level 4/5. Correlation between TcB and STB-H measured by HPLC was less strong but significant (r = 0.59 to 0.69 at different time points of measurement).

Fig. 2
figure 2

Correlation between serum total bilirubin and transcutaneous bilirubin measured at 24 h, 72 h, and at icterus involving arms/legs

Full size image
Fig. 3
figure 3

Bland–Altmann plot depicting difference between serum total bilirubin (STB) and transcutaneous bilirubin (TcB) measured at forehead, sternum, and abdomen on Y-axis against mean of STB and TcB on X-axis

Full size image

TcB performed well for both LBW (birth weight ≥1,500) and VLBW (birth weight <1,500 g) neonates although values of correlation coefficient were lower in the latter group (TcB on forehead, abdomen, and sternum, 0.86 vs. 0.78, 0.81 vs. 0.57, and 0.82 vs. 0.64).

Measurement at 24 h

At 24 h, most of neonates (86.7 %) had no visible yellow staining of skin. Measurement of STB by microbilimeter (STB-M) was done for all neonates at 24 ± 6 h of age. Transcutaneous bilimeter was unable to measure TcB in two neonates on forehead and in four neonates on sternum and abdomen each. STB by HPLC (STB-H) was measured in 81 neonates. Based on STB-M measurement at this time-point, phototherapy was started in 29 (11.3 %) neonates. Mean TcB on forehead, sternum, and abdomen was 7.4 ± 2.1, 8.2 ± 2.6, and 8.8 ± 2.8 mg/dL, respectively (Table 2). Mean STB-M was 7.1 ± 1.9 mg/dL and mean STB-H was 7.5 ± 2.7 mg/dL. Closest agreement and correlation between TcB and STB was observed for measurements obtained from forehead (Table 3).

Table 2 Transcutaneous bilirubin (TcB) and serum total bilirubin (STB) measurements at different time-points in enrolled neonates
Full size table
Table 3 Correlation* and agreement matrix between serum total bilirubin (STB) and transcutaneous bilirubin (TcB)
Full size table

Measurement at 72 h

At 72 h, most of neonates (93.8 %) had some degree of jaundice. Measurement of STB by microbilimeter was done for 225 neonates at 72 ± 12 h of age. Transcutaneous bilimeter was unable to measure TcB in two neonates on forehead and in eight neonates on sternum and abdomen each. STB by HPLC was measured in 90 neonates. Based on STB-M measurement at this time-point, phototherapy was started in 93 (41.3 %) neonates. Mean TcB on forehead, sternum, and abdomen was 11.3 ± 3.3, 12.2 ± 4.0, and 12.2 ± 3.8 mg/dL, respectively (Table 2). Mean STB-M was 10.9 ± 2.7 mg/dL and mean STB-H was 11.5 ± 2.9 mg/dL. Closest agreement between TcB and STB-M was observed for measurements obtained from forehead (Table 3).

Measurement at Jaundice in Kramer zone 4/5

Among 225 neonates who underwent study measurement at 72 h, 93 were started on phototherapy based on results of STB measurement at that time. Sixty-one neonates either died or never developed jaundice extending to zone 4/5 and therefore did not need this measurement. The remaining 71 neonates underwent TcB measurement. Measurement of STB by microbilimeter was done for 71 neonates at mean age of 121.5 ± 29.8 h. Transcutaneous bilimeter was unable to measure TcB in one neonate on sternum and four neonates on abdomen. STB by HPLC was measured in 19 neonates. Based on STB-M measurement at this time-point, phototherapy was started in 29 (40.9 %) neonates. Mean TcB on forehead, sternum, and abdomen was 12.0 ± 3.5, 12.2 ± 4.8, and 12.5 ± 3.9 mg/dL, respectively. Mean STB-M was 12.0 ± 3.0 mg/dL and mean STB-H was 12.1 ± 3.4 mg/dL. Closest agreement and correlation between TcB and STB-M again observed for measurements obtained from forehead (Table 3).

Discussion

In this study, we evaluated accuracy of TcB in preterm low-birth-weight neonates. TcB values were obtained at three time-points (at 24 h, at 72 h, and at jaundice in Kramer zone 4/5) and at three different sites (forehead, sternum, and abdomen). Good correlation and agreement were observed between TcB and STB at all the three time points. TcB values at forehead were most accurate.

As visual assessment of severity of jaundice can be erroneous, TcB measurement has been recommended as a screening tool to assess severity of neonatal hyperbilirubinemia [15, 16]. Accuracy of TcB obtained with multi-wavelength transcutaneous bilimeter has been well proven in neonates born at 35 or more weeks of gestation [10, 19]. However, studies investigating accuracy of TcB in preterm neonates have reported inconsistent results among neonates born at <28 weeks of gestation. Willems et al. reported that BiliCheck was as reliable in very preterm infants of gestational age less than 30 weeks [32]. Schmidt et al. reported excellent correlation between TcB and STB in preterm neonates born at 24–34 weeks of gestation [26]. However, TcB underestimated STB by 2–3 mg/dL. Ho et al. recommended that BiliCheck is a useful screening tool for neonatal jaundice only in the neonates with gestation above 32 weeks [8]. Namba and Kitajima measured TcB over forehead in preterm and VLBW infants and concluded that the TcB measurements were safe and accurate in these infants [21]. However, they reported a low reliability in neonates whose birth weights were lower than 1,000 g or whose gestational ages were shorter than 28 weeks. Rubaltelli et al. recommended BiliCheck as a reliable substitute of STB determination in newborns more than 30 weeks of gestation [24]. Knupfer et al. in their study concluded that reliable TcB values are obtained only in newborns older than 30 weeks of gestation [12]. In present study, good correlation and agreement between TcB and STB has been observed. We could not find any consistent difference in correlation or agreement between two birth weight groups (1,500–2,000 g and <1,500 g) enrolled in the present study. As a result TcB measurement may be used for initial screening of hyperbilirubinemia in preterm neonates. However, bilirubin values near treatment threshold need to be confirmed by STB measurement.

Use of TcB measurement has been shown to decrease need of blood sampling in term and late preterm neonates. Mishra et al. reported 34 % lower need of blood sampling with TcB-based screening strategy as compared to protocol-based visual assessment in neonates born at 35 or more weeks of gestation [20]. Willems et al. measured TcB and STB in preterm neonates born at less than 30 weeks of gestation and developed a model to estimate the possible reduction in need of blood sampling. In absence of skin edema or hypoperfusion, TcB measurement had potential to reduce blood sampling by up to 40 % [32]. In our study, phototherapy was started in 10 % neonates who underwent STB estimation at 24 h of age and in about 40 % neonates who underwent STB estimation at 72 h of age or at icterus level 4/5. Based on these figures, we speculate that TcB measurement can reduce need of blood sampling in moderately preterm neonates also.

Forehead is the most commonly recommended site for TcB measurement. However, performance of transcutaneous bilimeter in other body sites has not been assessed conclusively in preterm neonates. Sajjadian et al. measured TcB in preterm neonates and observed that correlation between STB and TcB was better at forehead than at sternum [25]. On the other hand, Stillova et al. reported better correlation between TcB and STB at sternum [27]. However, bilirubin was measured only at the time of clinical indication in a small group of preterm neonates. Similarly, Holland et al. reported that TcB measured at forehead is more likely to underestimate the STB concentration, whereas the TcB measured at sternum is more likely to overestimate the STB concentration [9]. In the present study, TcB at all the sites correlated well with the STB. However, TcB measured at forehead was closest to STB.

Effect of postnatal age on correlation between TcB and STB is difficult to separate from effect of increase in bilirubin concentration. In the present study, value of correlation coefficient is higher for measurements obtained at icterus level 4/5 or at 72 h of postnatal age than the value at 24 h of age. In a previous study among term and late preterm neonates TcB has been shown to perform best for STB values around 7–10 mg/dL [19]. As treatment is started in preterm neonates at values beyond 12–14 mg/dL, decreased performance of TcB with rising STB may not be reflected in studies enrolling only preterm neonates.

Strengths of this study include large sample size (1,619 TcB measurements in 256 neonates), prospective design, and assessment of accuracy of TcB against gold standard measurement of bilirubin by high-performance liquid chromatography. We also assessed the accuracy of TcB at three different sites of which presently recommended site of forehead was observed to be most accurate. We sampled each neonate at three different time-points to assess accuracy of TcB at different levels of STB and different intensities of clinical jaundice. Due to large number of TcB measurements, STB estimation by the gold standard method of HPLC was not done for all assessments, but by choosing one of the three time-points using a random number table. Smaller number of observations of STB with HPLC may also be the reason behind observing weaker correlation coefficients between TcB and STB by HPLC as compared to between TcB and STB by microbilimeter.

To conclude, this study demonstrates good correlation and agreement between TcB and STB in preterm neonates born at ≥28 weeks of gestation. Further studies are needed to evaluate role of TcB measurement for initial screening of hyperbilirubinemia in preterm neonates.