Pediatrics International (2012) 54, 81–85 doi: 10.1111/j.1442-200X.2011.03457.

Original Article ped_3457 81..85

Is bilirubin/albumin ratio correlated with unbound

bilirubin concentration?

Yumi Sato,1* Ichiro Morioka,1* Akihiro Miwa,1 Tomoyuki Yokota,1 Kiyomi Matsuo,1 Tsubasa Koda,1 Kazumichi Fujioka,1
Satoru Morikawa,1,2 Akio Shibata,1 Naoki Yokoyama,1 Kaoru Takahashi,2 Hisahide Nishio1,2 and Masafumi Matsuo1
Departments of 1Pediatrics and 2Community Medicine and Social Healthcare Science, Kobe University Graduate School of
Medicine, Kobe, Japan

Abstract Background: The American Academy of Pediatrics guidelines recommend that the total bilirubin (TB)/albumin (Alb)
ratio (B/A ratio), instead of serum concentration of unbound bilirubin (UB), can be used with TB for determining
treatment modality for jaundiced newborns 335 weeks of gestation. It is unknown, however, whether the B/A ratio is
actually correlated with serum UB.
Methods: Four hundred and ninety-seven serum samples were obtained from 209 newborns 335 weeks of gestation,
who were admitted to Kobe University Hospital. Serum UB concentration was measured using the glucose oxidase–
peroxidase method. Serum TB and Alb concentrations were measured on spectrophotometry. B/A ratios were calculated
and were linearly compared with serum UB. Furthermore, the accuracy of the B/A ratio was evaluated.
Results: The B/A ratio was significantly correlated with serum UB concentration. A serum UB concentration of
0.6 mg/dL was in agreement with a B/A ratio of 0.5. For comparison of the number of newborns who had serum UB
concentrations 3 or <0.6 mg/dL and B/A ratios 3 or <0.5, we found the following characteristics: the concordance rate
between serum UB concentrations and the B/A ratio was 94%, sensitivity was 51%, and specificity was 99%.
Conclusions: The B/A ratio is significantly correlated with serum UB concentration in newborns 335 weeks of
gestation. The B/A ratio, however, is underestimated when serum UB concentrations are >0.6 mg/dL.

Key words albumin, bilirubin, bilirubin–albumin binding, neonatal jaundice, serum.

Serum concentration of unbound bilirubin (UB), which is
bilirubin not bound to albumin (Alb), is used in conjunction
with serum total bilirubin concentration (TB) for diagnosing
neonatal jaundice. This is because UB has been suggested to
identify, more strongly than TB alone, those infants at risk for
developing bilirubin-induced neurologic dysfunction, such as
acute bilirubin encephalopathy, and its sequelae, kernicterus.1–5
Identification of jaundiced newborns using serum UB is not
widely used, however, because instruments for measuring
serum UB are not marketed for routine clinical use in the
world. Therefore, another index is required for estimating
serum UB concentration.
UB is the free fraction of unconjugated bilirubin as follows.
⎯⎯
[frcc Alb ] + [ UB] ← ⎯
→ [ Alb-bound bilirubin ]
⎯
K
The relationship between serum UB, Alb, and Alb-bound
bilirubin concentrations can be expressed as the binding constant
(K) = [Alb-bound bilirubin]/[free Alb] ¥ [UB]. Because [TB] =
Correspondence: Ichiro Morioka, MD, PhD, Department of Pediatrics,
Kobe University Graduate School of Medicine, 7-5-1 Kusunoki-cho,
Chuo-ku, Kobe 650-0017, Japan. Email: ichim@med.kobe-u.ac.jp
*Both of these authors contributed equally to this work.
Received 10 March 2011; revised 2 July 2011; accepted 15 August
2011.
[Alb-bound bilirubin] + [UB], and [UB] is a very small quan-
tity relative to [TB], [Alb-bound bilirubin] is almost equal to
[TB]. [free Alb] is equal to [Alb] – [bilirubin-bound Alb].
[bilirubin-bound Alb] is also almost equal to [TB]. Therefore, the
equation can be expressed as [UB] = [TB]/K ¥ ([Alb]–[TB]).4,6
[UB] theoretically correlates with [TB]/[Alb] (bilirubin/albumin
[B/A] ratio). Therefore, the 2004 American Academy of
Pediatrics (AAP) guidelines recommend that in newborns 335
weeks of gestation, the B/A ratio, instead of serum UB, can be
used with serum TB for determining treatment modality, such as
exchange transfusion, for severely jaundiced newborns.7,8
K, however, is variable in human newborns.4,9,10 In addi-
tion, bilirubin is combined with not only Alb, but also erythro-
cyte membranes, high-density lipoproteins, a-fetoproteins, and
others in blood, and serum UB is affected by various bilirubin
displacers, such as some hormones and drugs.5,11 Therefore, it is
unknown whether the B/A ratio is correlated with serum UB
concentration in human newborns. The objectives of the present
study were therefore to (i) investigate the correlation of the B/A
ratio and UB concentrations measured in serum human
samples of newborns 335 weeks of gestation; and (ii) determine
whether the B/A ratio can be used for the clinical screen-
ing of severely jaundiced newborns with high serum UB
concentration.

© 2011 The Authors

Pediatrics International © 2011 Japan Pediatric Society
82 Y Sato et al.

Methods Comparison of number of newborns with B/A ratios 3 or

<0.5 and serum UB concentrations 3 or <0.6 mg/dL
Patients and serum samples
Serum UB, TB, Alb, and direct bilirubin (DB) concentrations are The number of newborns with a B/A ratio 3 or <0.5 was com-
routinely measured with informed consent of the parents of the pared with that of newborns who had a serum UB 3 or
newborns, and assessed for the management of neonatal hyper- <0.6 mg/dL to evaluate the accuracy of the B/A ratio, and the
bilirubinemia in all newborns who are cared for at Kobe Univer- concordance rate, sensitivity, specificity, negative predictive
sity Hospital. Serum Alb and DB concentrations need to be value, and positive predictive value were calculated.
measured and assessed because hypoalbuminemia is a risk factor Comparison of K and ([Alb]–[TB])/[Alb] between
for bilirubin-induced neurologic dysfunction,11 and the UB- different serum UB levels
Analyzer occasionally measures higher values when serum DB
Serum samples were divided into two groups based on serum UB
exceeds 2 mg/dL.1 Blood samples were obtained from 209 new-
levels (<0.6 mg/dL and 30.6 mg/dL). K and ([Alb]–[TB])/[Alb]
borns 335 weeks of gestation without phototherapy, who were
were calculated and compared to evaluate the differences
admitted to a level II transitional care nursery or well-baby
between the groups.
nursery of Kobe University Hospital, Kobe, Japan, from 2007 to
2009. All enrolled newborns were healthy or not severely sick, Statistical analysis
that is, they had such conditions as transient tachypnea of the
Data are expressed as median (range) or n (%). Regression analy-
newborn, meconium aspiration syndrome with mild symptoms,
sis was performed to linearly compare between the B/A ratio and
initial vomiting, and newborns delivered from a mother with
serum UB concentration, and regression equations and correla-
well-controlled autoimmune disease or diabetes mellitus, and
tion coefficients (R2) were calculated. Statistical analysis was
premature rupture of the membranes without infectious signs.
performed with the Mann–Whitney non-parametric rank test for
Multiple blood samples on different days from 0 to 14 days
comparison of K and ([Alb]–[TB])/[Alb] between serum UB
after birth were taken from each newborn, because samples of
levels <0.6 mg/dL and 30.6 mg/dL. Correlations and differences
various serum bilirubin concentrations were required. Blood
were deemed statistically significant for P < 0.05.
samples were promptly centrifuged and serum was obtained.
Serum UB, TB, Alb, and DB concentrations were then measured. Results
A total of 497 serum samples were analyzed.
Clinical characteristics of the newborns
Measurement of serum UB, TB, Alb, and DB The clinical characteristics of the 209 enrolled newborns without
concentrations phototherapy are listed in Table 1. Fluids with glucose and elec-
Serum TB and UB were measured using an automated trolytes were used for infusion therapy, and only ampicillin was
UB-Analyzer (Arrows, Osaka, Japan) using spectrophotometry used as the antibiotic therapy to prevent bacterial infections, such
and the glucose oxidase–peroxidase method, respectively, as pre- as Escherichia coli and/or Group B Streptococcus.
viously described.12,13 Serum TB and UB concentrations are given
Correlation of B/A ratio and serum UB concentrations
as mg/dL (1 mg/dL = 17.1 mmol/L) and mg/dL (1 mg/dL =
17.1 nmol/L), respectively. Serum UB was measured at a single The correlation of the B/A ratio and serum UB concentrations in
peroxidase concentration as recommended by the manufacturer. all serum samples (n = 497) is shown in Figure 1. The B/A ratio
Serum Alb and DB concentrations were measured using the was significantly correlated with serum UB concentrations (R2 =
modified bromocresol purple method and the bilirubin oxidase 0.88, P < 0.0001).
method, respectively.14,15 When the enrolled newborns were divided into two groups
based on gestational age 338 weeks (n = 303) or 35–37 weeks
Calculation of the B/A ratio, K, and ([Alb]–[TB])/[Alb]
For calculations, TB (mg/dL), UB (mg/dL), and Alb (g/dL) were
Table 1 Clinical characteristics of the enrolled newborns
converted to mmol/L by dividing by 0.0546, 54.6, and 0.0066741,
respectively. The B/A ratio and K were calculated using TB, UB, Enrolled newborns (n = 209),
and Alb concentrations as follows: n (%) or median (range)
Gestational age (weeks) 39 (35–41)
B / A ratio = [ TB] / [ Alb ]
35–37 83/209 (40)
338 126/209 (60)
K = [ TB] / [ UB] × ([ Alb ] − [ TB])
Birthweight (g) 2854 (1592–4076)
To determine to what degree [Alb]–[TB] can be considered to be <2500 52/209 (25)
32500 157/209 (75)
equal to [Alb], ([Alb]–[TB])/[Alb] was calculated.
Apgar score 1 min 9 (5–10)
Correlation of B/A ratio and serum UB concentration 5 min 9 (8–10)
Male 114/209 (55)
Correlation of the B/A ratio and serum UB concentrations was Oxygen therapy 67/209 (32)
investigated in serum samples of all enrolled newborns, including Infusion therapy 114/209 (55)
newborns 338 weeks and those 35–37 weeks gestational age. Antibiotics therapy 30/209 (14)

© 2011 The Authors

Pediatrics International © 2011 Japan Pediatric Society
 Bilirubin/albumin and unbound bilirubin 83

Comparison of K and ([Alb]–[TB])/[Alb] between serum

1.4
UB concentrations of <0.6 mg/dL and 30.6 mg/dL
Both K and ([Alb]–[TB])/[Alb] in serum with a UB 30.6 mg/dL
1.2 were significantly lower than those in serum with a UB
Serum UB concentration (µg/dL)

<0.6 mg/dL (P < 0.0001; Table 3).

0.8 (a)
1.4
0.6 1.2

0.4 1

0.8
0.2
0.6
0
0.4
Serum UB concentration ( µg/dL)
0 0.2 0.4 0.6 0.8
B/A ratio 0.2
Fig. 1 Correlation of the total bilirubin concentration/albumin con- 0
centration (B/A) ratio and serum unbound bilirubin (UB) concentra-
tions in serum of newborns 335 weeks of gestation (n = 497). The
0 0.2 0.4 0.6 0.8
B/A ratio is significantly correlated with serum UB concentration
(y = 1.35x – 0.089, R2 = 0.88, P < 0.0001). (b)
1.4

1.2
(n = 194), the B/A ratio was significantly correlated with serum
UB concentration in serum samples of both groups of newborns
(R2 = 0.88 and 0.87; P < 0.0001; Fig. 2). 1

Serum DB concentration
0.8
Serum DB concentration was <2 mg/dL in all serum samples.
0.6
Comparison of number of newborns with B/A ratios 3 or
0.4
<0.5 and serum UB concentrations 3 or <0.6 mg/dL
A serum UB 30.6 mg/dL is widely used as a criterion for initiation 0.2
of phototherapy in Japan. When a serum UB (= y) of 0.6 was
substituted for y = 1.35x – 0.089, which is the regression equation
for the B/A ratio versus serum UB in all serum samples, the B/A 0
ratio (= x) was 0.51.
0 0.2 0.4 0.6 0.8
To evaluate the accuracy of the B/A ratio, we evaluated the B/A ratio
number of newborns with a serum UB 3 or <0.6 mg/dL and a B/A
ratio 3 or <0.5. The test performance characteristics were as Fig. 2 Correlation of the total bilirubin concentration/albumin con-
follows: concordance rate was 94%, sensitivity was 51%, speci- centration (B/A) ratio and serum unbound bilirubin (UB) concentra-
ficity was 99%, negative predictive value was 95%, and positive tion in serum of newborns (a) 338 weeks (n = 303; y = 1.40x – 0.090,
R2 = 0.88, P < 0.0001) and (b) 35–37 weeks of gestational age (n =
predictive value was 77%. Twenty-two of 45 samples (49%) in 194; y = 1.31x – 0.088, R2 = 0.87, P < 0.0001). The B/A ratio was
which serum UB concentrations were >0.6 mg/dL, had a B/A found to be significantly correlated with serum UB concentration in
ratio <0.5 (Table 2). serum samples of both groups of newborns.

© 2011 The Authors

Pediatrics International © 2011 Japan Pediatric Society
84 Y Sato et al.

Table 2 No. newborns with B/A ratios 3 or <0.5 and serum UB reasons: (i) K in newborns 35–37 weeks was similar to that in
concentrations 3 or <0.6 mg/dL newborns 338 weeks of gestation; (ii) K was constant because
Serum UB concentration (mg/dL) Total
healthy or not severely sick newborns were enrolled, although K
is often lower in premature and sick newborns;4,9,10 and (iii) the
30.6 <0.6
most potent bilirubin-displacing drugs, such as sulfisoxazole and
B/A ratio 30.5 23 7 30
ibuprofen, were not used in enrolled newborns,5 and ampicillin,
<0.5 22 445 467
Total 45 452 497 which has only a weak bilirubin-displacing effect, was the only
antibiotic used in the present study.5 Another reason for the cor-
Concordance rate, 94%; sensitivity, 51%; specificity, 99%; negative
relation of the B/A ratio and serum UB concentration was that
predictive value, 95%; positive predictive value, 77%.
B/A ratio, total bilirubin concentration/albumin concentration; serum Alb concentrations might not be affected, even though
UB, unbound bilirubin. 114/209 (55%) newborns had fluids with glucose and electro-
lytes. A previous report found that infusion therapy does not
contribute to reductions in serum Alb concentrations.11
Discussion We found that a serum UB of 0.6 mg/dL was in agreement
We have found that the B/A ratio was significantly correlated with a B/A ratio of 0.5 and that the concordance rate between
with serum UB concentration in newborns 335 weeks of gesta- serum UB concentration and the B/A ratio was 94%, verifying
tion. A serum UB concentration of 0.6 mg/dL was in agreement the accuracy of the B/A ratio as a surrogate for serum UB con-
with a B/A ratio of 0.5. The present results suggest that the B/A centration. Furthermore, we found that the B/A ratio had a high
ratio can be used for estimating serum UB concentrations unless specificity and a negative predictive value. These results demon-
serum UB concentration can be measured directly. A low sensi- strate that a B/A ratio <0.5 is equivalent to a serum UB
tivity, however, was found when comparing the number of new- <0.6 mg/dL in almost all serum samples. This indicates that new-
borns who had a serum UB 30.6 mg/dL and B/A ratio 30.5, borns with a B/A ratio <0.5 do not require phototherapy, because
suggesting that caution is advisable when the B/A ratio is applied a serum UB 30.6 mg/dL is widely used as the criterion for the
to clinical screening to identify severely jaundiced newborns with initiation of phototherapy in Japan.
high serum UB concentration. The low sensitivity and positive predictive value, however,
Newborns who were 335 weeks of gestation were enrolled in may indicate that the B/A ratio cannot be relied upon in routine
the present study because the AAP guidelines state that in the clinical use for screening of severely jaundiced newborns with
management of hyperbilirubinemia in newborns 335 weeks of high serum UB concentration. The equation [UB] = [TB]/K ¥
gestation, the B/A ratio, instead of serum UB, can be used with ([Alb]–[TB]) suggests that serum UB concentration does not
serum TB for determining the introduction of exchange transfu- correlate with the B/A ratio if K is not constant and/or
sion.7 Furthermore, because the AAP guidelines suggest that a [Alb]–[TB] is not equal to [Alb]. We found that K in serum
gestation <38 weeks is one of the most important risk factors,7 the with a UB 30.6 mg/dL was significantly lower than in that with
enrolled newborns were divided into two groups (338 weeks and a UB <0.6 mg/dL (Table 3). Alb has multiple binding sites for
35–37 weeks) and analyzed. In all serum samples of enrolled bilirubin, drugs, and other substances: there is a high-affinity
newborns 335 weeks of gestation and in those 35–37 weeks of binding site (K1) and low-affinity binding sites (K2 and K3).4,9
gestation, the B/A ratio was significantly correlated with serum When K1 is saturated with bilirubin at high serum TB concen-
UB concentrations (Figs 1,2). These findings provide useful trations, K2 and K3 are used for bilirubin binding. K decreases
information for the management of neonatal jaundice without as serum TB increases.4 Furthermore, to determine whether
serum UB measurements in clinics and hospitals in which there [Alb]–[TB] can be considered equal to [Alb], we calculated
are term or late pre-term newborns. ([Alb]–[TB])/[Alb]. We found that ([Alb]–[TB])/[Alb] was
The present findings that the B/A ratio was significantly cor- much lower than 1, and ([Alb]–[TB])/[Alb] in serum with a UB
related with serum UB concentration might be because bilirubin- 30.6 mg/dL was significantly lower than that with a UB
Alb binding was constant in enrolled newborns for the following <0.6 mg/dL (Table 3). Because this equation uses molar concen-
trations, [Alb]–[TB] cannot be considered to be equal to [Alb],
especially because serum TB is increasing. When serum UB
Table 3 K and ([Alb]–[TB])/[Alb] vs serum UB <0.6 mg/dL and concentrations are high in healthy term newborns who are not
30.6 mg/dL being treated with bilirubin-displacing drugs, serum TB con-
Serum UB concentration, P centrations are also high. Therefore, we suggest that the B/A
median (range) ratio does not correlate with serum UB concentration when UB
<0.6 mg/dL 30.6 mg/dL concentration is high because of high serum TB concentration.
K (L/mmol) 98.1 78.1 <0.0001 Serum UB concentration should be measured in newborns with
(33.0–324.9) (51.0–143.9) high serum TB concentration.
([Alb]–[TB])/[Alb] 0.78 0.50 <0.0001 Two mg/dL of DB is non-physiological. Because DB is
(0.40–0.97) (0.25–0.62) easily resolved into UB by peroxidase, the UB-analyzer pro-
Alb, albumin; K, binding constant; TB, total bilirubin; UB, unbound duces a higher UB value than actual serum UB concentration in
bilirubin. sera with DB levels 32 mg/dL.1 We confirmed that enrolled

© 2011 The Authors

Pediatrics International © 2011 Japan Pediatric Society

patients did not have any diseases such as metabolic and/or
liver disorders with serum DB levels 32 mg/dL.
The present study had some limitations. First, the B/A ratio was
linearly correlated with serum UB concentration in this study
because of the small number of serum samples with UB concen-
tration 30.6 mg/dL. When including many serum samples with
high UB concentration, the equation for the B/A ratio and serum
UB concentration should not be linear, but curved: [UB] = 0.58 ¥
B/A ratio/(1 – B/A ratio). Second, serum levels of free fatty acid,
which affect serum UB concentration, were not measured and
analyzed. Last, it is unknown whether cytokines affect serum
UB concentration in some newborns with meconium aspiration
syndrome.
Conclusion
The present study has shown that the B/A ratio is correlated with
serum UB concentration in human newborns 335 weeks of ges-
tation. We consider that caution is advisable, however, when the
B/A ratio is used to identify severely jaundiced newborns with
high serum UB concentration, because of the low sensitivity
when comparing a B/A ratio 30.5 and a serum UB 30.6 mg/dL.
Acknowledgments
This work was supported by grants for Scientific Research from
the Morinaga Hoshi-kai Foundation and from the Ministry
of Education, Culture, Sports, Science and Technology in
Japan (IM).
References
1 Lee YK, Daito Y, Katayama Y, Minami H, Negishi H. The sig-
nificance of measurement of serum unbound bilirubin concentra-
tions in high-risk infants. Pediatr. Int. 2009; 51: 795–9.
2 Ahlfors CE, Parker AE. Unbound bilirubin concentration is asso-
ciated with abnormal automated auditory brainstem response for
jaundiced newborns. Pediatrics 2008; 121: 976–8.
Bilirubin/albumin and unbound bilirubin 85
3 Calligaris SD, Bellarosa C, Giraudi P, Wennberg RP, Ostrow JD,
Tiribelli C. Cytotoxicity is predicted by unbound and not total
bilirubin concentration. Pediatr. Res. 2007; 62: 576–80.
4 Wennberg RP, Ahlfors CE, Bhutani VK, Johnson LH, Shapiro SM.
Toward understanding kernicterus: A challenge to improve the
management of jaundiced newborns. Pediatrics 2006; 117: 474–
85.
5 Amin SB. Clinical assessment of bilirubin-induced neurotoxicity
in premature infants. Semin. Perinatol. 2004; 28: 340–47.
6 Ahlfors CE, Wennberg RP, Ostrow JD, Tiribelli C. Unbound (free)
bilirubin: Improving the paradigm for evaluating neonatal jaun-
dice. Clin. Chem. 2009; 55: 1288–99.
7 American Academy of Pediatrics. Management of hyperbilirubine-
mia in the newborn infant 35 or more weeks of gestation. Pediat-
rics 2004; 114: 297–316.
8 Ahlfors CE. Criteria for exchange transfusion in jaundiced new-
borns. Pediatrics 1994; 93: 488–94.
9 Bender GJ, Cashore WJ, Oh W. Ontogeny of bilirubin-binding
capacity and the effect of clinical status in premature infants born
at less than 1300 grams. Pediatrics 2007; 120: 1067–73.
10 Cashore WJ. Free bilirubin concentrations and bilirubin-binding
affinity in term and preterm infants. J. Pediatr. 1980; 96:
521–7.
11 Miwa A, Morioka I, Hisamatsu C et al. Hypoalbuminemia follow-
ing abdominal surgery leads to high serum unbound bilirubin con-
centrations in newborns soon after birth. Neonatology 2011; 99:
202–7.
12 Nakamura H, Lee Y. Microdetermination of unbound bilirubin in
icteric newborn sera: An enzymatic method employing peroxidase
and glucose oxidase. Clin. Chim. Acta 1977; 79: 411–17.
13 Shimabuku R, Nakamura H. Total and unbound bilirubin determi-
nation using an automated peroxidase micromethod. Kobe J. Med.
Sci. 1982; 28: 91–104.
14 Muramoto Y, Matsushita M, Irino T. Reduction of reaction differ-
ences between human mercaptalbumin and human nonmercaptal-
bumin measured by the bromcresol purple method. Clin. Chim.
Acta 1999; 289: 69–78.
15 Kimura S, Iyama S, Yamaguchi Y, Hayashi S, Yanagihara T.
Enzymatic assay for conjugated bilirubin (Bc) in serum using
bilirubin oxidase (BOD). J. Clin. Lab. Anal. 1999; 13: 219–23.
© 2011 The Authors
Pediatrics International © 2011 Japan Pediatric Society