See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259490241

Development of an Equation to Correct for

Hemolysis in Direct Bilirubin Measurements.

Article in Clinica chimica acta; international journal of clinical chemistry · December 2013
DOI: 10.1016/j.cca.2013.12.023 · Source: PubMed

CITATIONS READS

3 124

6 authors, including:

Dina N Greene Daniel T Holmes

University of Washington Seattle St. Paul's Hospital, Vancouver, Canada
58 PUBLICATIONS 343 CITATIONS 81 PUBLICATIONS 622 CITATIONS

SEE PROFILE SEE PROFILE

Joy Liang Robert Schmidt

Kaiser Permanente University of Utah
3 PUBLICATIONS 7 CITATIONS 94 PUBLICATIONS 674 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Dina N Greene on 09 October 2015.

The user has requested enhancement of the downloaded file.

 Clinica Chimica Acta 429 (2014) 194–197

Contents lists available at ScienceDirect

Clinica Chimica Acta

journal homepage: www.elsevier.com/locate/clinchim

Development of an equation to correct for hemolysis in direct

bilirubin measurements
Dina N. Greene a, Daniel T. Holmes b, Maun-Jan Lin a, Joy Y. Liang a, Thomas S. Lorey a, Robert L. Schmidt c,⁎
a
Kaiser Permanente, TPMG Northern California Regional Laboratory, Berkeley, CA, United States
b
University of British Columbia, Department of Pathology and Laboratory Medicine, Vancouver, BC, Canada
c
University of Utah School of Medicine, Department of Pathology, Salt Lake City, UT, United States

a r t i c l e i n f o a b s t r a c t

Article history: Background: Direct bilirubin is measured for the investigation of pediatric and adult jaundice. Package inserts
Received 19 August 2013 suggest that hemolysis decreases direct bilirubin measurements, but no published studies have adequately
Received in revised form 29 November 2013 described the extent of interference.
Accepted 16 December 2013 Methods: The influence of hemolysis on direct bilirubin quantification (Beckman AU680) was evaluated by
Available online 27 December 2013
titrating increasing amounts of hemoglobin into specimens with variable starting concentrations of direct bili-
rubin. An equation was derived to predict the nominal interference-free concentration of direct bilirubin as a
Keywords:
Neonatal hyperbilirubinemia
function of measured concentration and hemolysis-index.
Hemolysis Results: Hemolysis decreased the direct bilirubin concentration reported by the AU680. The extent of interference
Interference is a function of both the interference-free concentration of direct bilirubin and the degree of hemolysis.
Bilirubin Conclusions: The concentration of direct bilirubin in hemolyzed specimens can be predicted.
Screening © 2013 Elsevier B.V. All rights reserved.

1. Introduction direct bilirubin concentration plays an important role in the diagnosis of

Direct bilirubin is quantified to assist in the evaluation of patients
with suspected liver dysfunction and/or jaundice. Isolated elevation of
direct bilirubin generally points to cholestasis. In adults, cholestasis
can result from medication reaction, autoimmune conditions or me-
chanical obstruction. In neonates and pediatric patients, cholestasis
can result from biliary atresia, infantile hepatitis, or total parenteral nu-
trition. Direct bilirubin is rarely elevated (N5% of total bilirubin) in neo-
nates, but elevated direct bilirubin is always pathological [1]. Hospitals
commonly screen all neonates for total bilirubin, but the measurement
of direct bilirubin or conjugated bilirubin (herein denoted Bc) is re-
served for cases where there is clinical suspicion of specific pathologies
[2–4].
Direct bilirubin is not usually evaluated in isolation but is measured
in conjunction with other liver markers, importantly alanine amino-
transferase, aspartate aminotransferase, alkaline phophatase, and
gamma-glutamyl-transpeptidase (GGT). Depending on the suspected
pathology, the results will display a specific pattern. For example, recent
studies have outlined that neonates with biliary atresia will have elevat-
ed direct bilirubin with concurrent elevations in GGT that are signifi-
cantly higher than that of the GGT concentrations seen in other
cholestatic conditions [5]. Accurate measurements are required because
⁎ Corresponding author at: Department of Pathology, 15 N Medical Drive East,
University of Utah, Salt Lake City, UT 84112, United States. Tel.: +1 801 585 2010;
fax: +1 801 585 2805.
E-mail address: Robert.Schmidt@hsc.utah.edu (R.L. Schmidt).
liver pathologies. This is especially true in neonates.
Hemolysis is a common source of interference in the measurement
of bilirubin. The effect of hemolysis on total bilirubin determination
has been extensively investigated [6–10]. In contrast, there is relatively
little information on the falsely decreased direct bilirubin results
observed on hemolyzed specimens [11]. Given the importance, many
institutions have strict criteria for rejecting direct bilirubin neonatal
samples with elevated concentrations of hemoglobin. Strict criteria
cause high rejection rates of neonatal samples, which increase laboratory
expenses, delay diagnosis and necessitate repeated difficult collections.
These consequences could be reduced if sample rejection criteria were
relaxed. Unfortunately, limited data hampers the evidence-based relaxa-
tion of these criteria.
2. Materials and methods
2.1. Study approval
This study was considered a quality assessment project and
was therefore deemed exempt by the Northern California Kaiser
Permanente Institutional Review Board.
2.2. Direct bilirubin quantification
Bilirubin was measured in all samples using the Beckman AU680
(Beckman Coulter Inc.). The AU680 uses a diazo-based method (uses

0009-8981/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cca.2013.12.023
 D.N. Greene et al. / Clinica Chimica Acta 429 (2014) 194–197
diazonium salt of 3,5-dichloroaniline) to quantify direct bilirubin (no
accelerator) and total bilirubin (caffeine salt added as accelerator).
Both the total and direct assays are calibrated against materials
traceable to the National Institute of Standards and Technology (NIST)
Standard Reference Material.
2.3. Hemolysis interference
Hemolysates were prepared from whole blood with a known
hemoglobin concentration. Samples were centrifuging it at 3000 rpm
(1610 g) for 10 min. The resulting plasma was aspirated and the
volume removed was measured. The red cells were then washed in qua-
druplicate with saline. After washing, the appropriate volume of deion-
ized water was added to make the final hemoglobin concentration
20 mg/dl. The cells were freeze-thawed three times before centrifuging
to separate the cellular debris from the hemolysate. The resulting super-
natant was removed and used to titrate hemoglobin into the serum
samples.
Residual, non-hemolyzed adult serum samples (n = 32; starting
H-index = 0) with varying concentrations of direct bilirubin concen-
tration (range of starting concentrations 0.49–9.76 mg/dl; mean =
3.05 mg/dl, SD = 2.30) were selected from samples to be discarded
and appropriately deidentified. Hemolysis was induced by titrating
hemoglobin into these samples, correcting for dilution, to create a se-
ries of six–seven samples having identical direct bilirubin concentra-
tions, but varying amounts of hemoglobin. Each sample therefore
Fig. 1. The relationship between the initial direct bilirubin concentration (mg/dl) as a function of the measured direct bilirubin concentration (mg/dl) and the hemoglobin concentration.
Each graph corresponds to a different concentration of hemoglobin.
195
had 6–7 levels of hemolysis associated with a single, non-hemolyzed
direct bilirubin quantitation. Total bilirubin, direct bilirubin, and the
H-index were determined in these samples using the AU680.
The Vitros assay was used to confirm that the direct bilirubin was
primarily composed of conjugated bilirubin (i.e. delta bilirubin was
less than 10% of the direct bilirubin). The Vitros assay uses reflectance
spectrophotometry to quantify glucuronide and albumin conjugates of
bilirubin independently. This provided confirmation that the AU680
results were attributable to glucuronide conjugates of bilirubin.
2.4. Assessment of sample hemolysis
The extent of hemolysis was classified using a hemolysis-index
(H-index) of 1+ through 5+ based upon transmission spectropho-
tometry on the AU680 [12,13]. Parameters were user defined as
follows: normal (N), optical density (OD) b 0.0850; 1+, OD = 0.0850–
0.1699 (~31 mg/dl hemoglobin); 2+, OD = 0.1700–0.349 (~62 mg/dl
hemoglobin); 3+, OD = 0.350–0.6999 (~125 mg/dl hemoglobin); 4+,
OD = 0.7000–0.9999 (~250 mg/dl hemoglobin); 5 +, OD ≥ 1.000
(~500 mg/dl hemoglobin); abnormal (ABN), OD ≥ 3.000 (~1000 mg/dl
hemoglobin).
2.5. Statistical analysis
Statistics associated with the performance characteristics (precision,
linearity, method comparison) were calculated using Microsoft Excel
196 D.N. Greene et al. / Clinica Chimica Acta 429 (2014) 194–197

(Microsoft Corp) and EP Evaluator (Data Innovations). Regression expressed the relationships between the constants and hemoglobin
analysis was used to find a relation between initial direct bilirubin, concentration:
measured direct bilirubin and hemoglobin. Regression for patient com-
parisons was performed using cp-R, a graphical user interface to the R lnðaÞ ¼ lnð0:334Þ þ 0:334 lnðH Þ ð3Þ
statistical programming language (http://sourceforge.net/projects/
cprchempath/). lnðbÞ ¼ lnð1:56Þ–0:14lnðHÞ: ð4Þ

Substituting Eqs. (3) and (4) into Eq. (1) gives the following equa-
3. Results
tion which predicts the initial direct bilirubin concentration based on
the measured direct bilirubin and the hemoglobin concentration:
3.1. Hemolysis and direct bilirubin
  −0:14
0:334 1:56H
Addition of hemoglobin caused a negative interference in direct D0 ¼ 0:334H D ð5Þ
bilirubin. The degree of interference depended on the initial direct
bilirubin concentration and the hemoglobin concentration (Fig. 1). Where:
We found that the relationship between initial direct bilirubin concen-
tration (i.e. direct bilirubin concentration without hemoglobin inter- D0 the corrected direct bilirubin concentration (mg/dl)
ference; D0) and measured direct bilirubin concentration (D) could H the hemoglobin concentration (g/dl) b 1000 mg/dl
be fit to an exponential relationship: D the measured direct bilirubin concentration (mg/dl).

b
D0 ¼ aD : ð1Þ The correction equation showed good correspondence between the
predicted, D0, and actual direct bilirubin concentrations, D (Fig. 3;
Eq. (1) was linearized: F(1,190) = 11892, p b 0.001, R2 = 0.98) for hemoglobin concentra-
tions between 31 and 1500 mg/dl.
lnðD0 Þ ¼ lnðaÞ þ blnðDÞ: ð2Þ
4. Discussion
The linear form provided a good fit with the data as shown in Fig. 2.
The constants, a and b, varied with the hemoglobin concentration Preanalytical interferences are a common motivation for sample
(Table 1). We found the following implicitly defined functions recollection. Reduction of redraws in all populations is important,

Fig. 2. The relationship between the logarithm of the initial direct bilirubin concentration (mg/dl) as a function of the logarithm of the measured direct bilirubin (mg/dl) and the hemo-
globin concentrations is linear. Each graph corresponds to a different concentration of hemoglobin. Regression coefficients and coefficients of determination (R2) are presented in Table 2.
 D.N. Greene et al. / Clinica Chimica Acta 429 (2014) 194–197
Table 1
Maximum corrected direct bilirubin concentration (mean & 95% confidence interval) as
a function of the measured direct bilirubin and approximate hemoglobin concentration
(as measured by H-index). Abbreviations — direct bilirubin (d-bil); hemoglobin (Hb).
Measured d-bil Maximum predicted d-bil based on H-index
H index = 1–2 H-index = 3–4 H-index = 5-Abn
0.00–0.18 – 0.49 0.73 (0.61–0.85)
0.18–0.60 0.60 (0.52–0.65) 0.85 (0.69–1.02) 1.70 (1.51–1.90)
0.61–1.17 1.05 (0.93–1.18) 1.52 (1.40–1.64) 2.80 (2.45–3.10)
1.18–1.67 1.59 (1.52–1.65) 2.25 (2.00–2.50) –
1.68–2.85 2.51 (2.33–2.69) 3.07 (2.73–3.41) 5.01 (4.63–5.38)
2.85–5.50 4.68 (4.31–5.05) 5.24 (4.88–5.60) 6.11 (5.30–6.91)
N 5.50 7.91 (6.97–8.88) 8.47 (7.30–9.64) 9.21 (8.12–10.30)
particularly in neonates. Hemoglobin causes spurious reductions in
direct bilirubin concentration and this effect is of particular concern
for neonatal collections. Accordingly, direct bilirubin requests on hemo-
lyzed specimens are often (appropriately) rejected by the laboratory.
Approaching the problem systematically, we have developed an empir-
ical equation that accurately estimates the nominal interference-free
concentration of direct bilirubin as a function of the measured concen-
tration and extent of hemolysis. It is important to note that these
studies were all completed on the AU680, which uses diazotized 3,5-
dichloroaniline. Other platforms may use different diazo compounds
and the equations shown are unlikely to apply, though the generic
mathematical approach may be replicable.
As a matter of practicality, we were forced to use adult specimens
because of the large volume of specimen required to complete the in-
vestigations. However, the primary motivation for this study was to
establish evidence-based criteria for requesting a redraw for neonates
with suspected cholestatic disease. This is a limitation of the study and
assumes that neonatal specimens behave similarly to adult specimens.
Despite this limitation, we believe our approach can be used to
reduce specimen rejection. In our own institution, we developed a mod-
ified approach based on the prediction equation. We felt that the equa-
tion was too complex for our laboratory information system or to be
used for bench-side calculations. However, we did use the equation to
guide us in composing footnotes that alert the physician of (1) the extent
of hemolysis and (2) the threshold concentration of direct bilirubin at the
specific H-index that could be grounds for redrawn if clinically indicated.
Rules were derived after calculating the mean and the 95% confi-
dence interval for the predicted direct bilirubin concentration using
the correction formula (Table 1). The rules were designed to be more
conservative than the calculation suggests, allowing for the laboratory
to be increasingly confident that the physician is alerted of the potential
for false negative results in any hemolyzed specimen. For example, if a
Fig. 3. Regression of mathematically corrected direct bilirubin (y axis) with initial direct
bilirubin from hemolysis-free sample (x axis). The dashed line indicates perfect agreement
(y = x).
View publication stats
197
Table 2
Relationship between coefficients (a and b; used in Eqs. (3) and (4)) and approximate
hemoglobin concentration, based on the H-index. Each concentration of hemoglobin is
accompanied by the linear coefficients derived from the logarithm of the initial and
measured direct bilirubin concentrations. R2 is the coefficient of determination for each
least squares fit. Abbreviations — hemoglobin (Hb). *omitted outlier.
Hb concentration ln(a) b R2
31 0.131 0.945 0.999
62 0.252 0.890 0.996
125 0.452 0.807 0.991
250 0.714 0.696 0.981
500 0.981 0.634 0.970
1000 1.267 0.603 0.944
1500* 1.217 0.674 0.965
neonatal direct bilirubin specimen has + 2H-index, we will append a
comment that states “Specimen mildly hemolyzed; hemolysis will
falsely decrease direct bilirubin results. For results N 1.6 mg/dl consider
redraw if clinically indicated”. This is a conservative approach, as the
data shown in Table 2, suggests that the concentration could be closer
to 1.8 mg/dl before redraw would be deemed necessary.
It is also noteworthy that we append these concentration specific
comments only until 14 days of life, at which point our reference range
for direct bilirubin remains ≤2.0 mg/dl (the 99th percentile for direct
bilirubin in the first 2 weeks of life [14]). For patients N 14 days of
age, the reference interval is decreased to 0.3 mg/dl, at which point it
is unlikely that hemolysis would impact pathological results (usually
N1.0 mg/dl) to an extent where they would be unflagged (b 0.3 mg/dl).
However, a generic comment is appended to samples from patients
N14 days old to inform the physician that hemolysis was present and
can lead to falsely decreased direct bilirubin results.
In conclusion, this manuscript shows an approach that can be used
to quantify the extent of hemolysis-related interference on direct biliru-
bin quantitation. The results are applicable to both basic analytics and to
rejection criteria for neonatal specimens.
Acknowledgments
The authors thank Dr Thomas B Newman for his excellent editorial
assistance and clinical opinion.
References
[1] Brumbaugh D, Mack C. Conjugated hyperbilirubinemia in children. Pediatr Rev
2012;33:291–302.
[2] Ip S, Chung M, Kulig J, et al. An evidence-based review of important issues concerning
neonatal hyperbilirubinemia. Pediatrics 2004;114:e130–53.
[3] Bhutani VK, Vilms RJ, Hamerman-Johnson L. Universal bilirubin screening for severe
neonatal hyperbilirubinemia. J Perinatol 2010;30:S6–S15 [Suppl.].
[4] American Academy of Pediatrics Subcommittee on Hyperbilirubinemia. Manage-
ment of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation.
Pediatrics 2004;114:297–316.
[5] Sun S, Chen G, Zheng S, et al. Analysis of clinical parameters that contribute to the
misdiagnosis of biliary atresia. J Pediatr Surg 2013;48:1490–4.
[6] Christensen RD, Lambert DK, Henry E, et al. Unexplained extreme hyperbilirubinemia
among neonates in a multihospital healthcare system. Blood Cells Mol Dis 2013;50:
105–9.
[7] McDonagh AF, Lo SF, Maisels MJ, Watchko JF. Serum bilirubin measurements in
hemolyzed samples. Clin Chim Acta 2011;412:1275 [author reply 1276].
[8] Algeciras-Schimnich A, Cook WJ, Milz TC, Saenger AK, Karon BS. Evaluation of hemo-
globin interference in capillary heel-stick samples collected for determination of
neonatal bilirubin. Clin Biochem 2007;40:1311–6.
[9] Lo SF, Jendrzejczak B, Doumas BT. Total or neonatal bilirubin assays in the Vitros 5,1
FS: hemoglobin interference, hemolysis, icterus index. Clin Chem 2007;53:799–800.
[10] Brunori P, Masi P, Faggiani L, et al. Evaluation of bilirubin concentration in hemolysed
samples, is it really impossible? The altitude-curve cartography approach to inter-
fered assays. Clin Chim Acta 2011;412:774–7.
[11] Mori L. Modified Jendrassik–Grof method for bilirubins adapted to the Abbott
Bichromatic Analyzer. Clin Chem 1978;24:1841–5.
[12] CLSI. Hemolysis, icterus, and lipemia/turbidity indices as indicators of interference in
clinical laboratory analysis; approved guideline. Wayne, PA: Clinical and Laboratory
Standards Institute; 2012.
[13] Beckman Coulter. OSR 62166; LIH; 2009.
[14] Davis AR, Rosenthal P, Escobar GJ, Newman TB. Interpreting conjugated bilirubin
levels in newborns. J Pediatr 2011;158 [562–5 e561].