INGR-16 / INGREDIENTS

SPSFAM EXPERT REVIEW PANEL

DO NOT DISTRIBUTE

SM
AOAC® Official Method xxx

Spectrophotometric Method (iCheck® CHROMA)

for determination of Vitamin A in Vegetable Oil

For approval by: STAKEHOLDER PANEL ON STRATEGIC FOOD ANALYTICAL METHODS

(SPSFAM)

Submission date: 07.06.2012

Applicability:

Quantitative measurement of vitamin A is defined as 13-cis and all-trans retinol, retinyl esters
(retinyl palmitate, and retinyl acetate) in food ingredients such as vegetable oil at a concentration
range of 2.5 – 30 ppm. The upper limit can be increased by adjusting the dilution level.

Caution statement:

No special precautions needed as the method is developed for safe use inside as well as outside a
laboratory setting.

Interlaboratory study result table:

Unavailable at this point in time.

Principle:

Vegetable oil is directly injected into the sealed glass vial containing antimony trichloride (SbCl3)
and chloroform (CHCl3). The determination of vitamin A in the oil is based on the reaction of SbCl3
and the double bonds of the retinol molecule in CHCl3 that forms anhydroretinylic and retinylic
cation yielding transient brilliant blue color. The absorption of the solution is immediately measured
in the iCheck® CHROMA (BioAnalyt, Germany) photometer. The device automatically calculates
and displays the concentration of vitamin A as retinyl equivalents/RE (retinol and retinyl esters) in
mg/kg oil.

Apparatus and Materials:

iCheck CHROMA – portable photometer

designed specifically for determination of
vitamin A in vegetable oil. The optical design
has been developed for the iEx™ ELAN
reagent vial and cuvette. The unit is
equipped with the latest light emitting diode
(LED) technology. Modern LEDs emit high
and extremely stable light energy together
with very low energy consumption at no heat
development. Therefore, user calibration is
not required for the anticipated shelf life of
the iCheck™. Based on the half life of
30,000 hours of the LED, approximately 10
million measurements can be conducted. iCheck™ software has programmed algorithm that
converts the absorption units of the blue-color reaction of several automatic measurements into
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

RE/kg. For quality control purposes the device conducts and auto-control to verify the emitter and
receptor performance. A vitamin A standard to control the device is not needed.

iEx™ ELAN - disposable reaction vial and cuvette used for the measurement with iCheck™. iEx™
ELAN contains 2 mL of mixture of solvents containing essentially SbCl3 and CHCl3 developed to
determine Vitamin A in the sample. BioAnalyt guarantees that each vial contains the mixture of
reagents safely and permanently crimped (shelf life 12 months).

Reagents:

iEx™ ELAN - disposable reaction vial. No additional chemicals are required.

Preparation of Test Sample:

For sample preparation the oil sample may be warmed up to 30° C in order to avoid any
precipitate. No additional preparation required.

Procedure and Determination:

Oil is drawn up with the applicator (syringe) at a usual volume of 500 µl. The complete volume is
injected slowly into the sealed glass vial iEx™ ELAN through the red rubber septum. Sealed vial
prevents toxic exposure to reagents. Briefly invert the vial and immediately measure the absorption
of the concentration-dependent blue colour of the solution at 610 nm with iCheck® CHROMA
photometer. The measurement is accompanied by the device displaying the instructions for the
investigator of each single step. The concentration of vitamin A is displayed as retinyl equivalents
(= retinol and retinyl esters) in mg/kg oil.

Reference:

Int. J. Vitam. Nutr. Res., 81 (5), 2011, p.335-342.

Method Performance Requirements

AOAC requirements iCheck® CHROMA

Analytical range 100 ppm – 100% 2.5 -30 ppm [100%*]

0.01% ≤ 4% 1.62%
Repeatability
1% ≤ 2% N/A
(RSDr)
100% ≤ 1% N/A

Recovery 90-110% 99-101%

0.01% ≤ 8% 3.92%
Reproducibility N/A
1% ≤ 4%
(RSDR)
100% ≤ 2% N/A

*100% can be reached by adjusting the dilution.

 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

INTERNAL VALIDATION OF iCheck® CHROMA

Working and linear range

Based on the injection of 500µl sample the linear range was determined. Palm oil fortified with
retinyl palmitate was gradually diluted with five equivalent concentrations (0, 20, 40, 60, 80 and
100%) of non fortified palm oil and measured in triplicates. The linear range is 2.5 - 30 mg/kg.

Figure 1. Linearity of fortified palm oil spiked with increasing amounts of non fortified palm oil. Each sample was measured
in triplicates. Each dot represents one measurement.

LOD & LOQ

Limits of detection and quantitation were assumed to be interchangeable for this method and
determined by functional sensitivity (Clin. Chem. 44, p. 2217). Unfortified palm oil was increasingly
fortified with retinyl palmitate, and measured, until a coefficient of variation (CV) consistently below
20% was reached. Because of the algorithms programmed in the portable device, a separate
definition of LOD and LOQ is not possible. The threshold where the portable device can reliably
distinguish the signal from the background noise (<20% CV) is 2.5 mg RE/kg and is representative
of LOD.

Repeatability

Intra-assay (precision within series) and day-to-day precision was determined by measuring 3
fortified palm oil samples at three different days by one person. Palm oil fortified with retinyl
palmitate was mixed with non-fortified palm oil in three concentrations (20, 40 and 80%) and
measured ten times:
Fortified Expected Expected
Non-fortified
Concentration concentration in concentration
palm oil palm oil
% (mg RE/kg)
Low 20% 80% 0.00065 6.50

Medium 40% 60% 0.0013 13.0

High 80% 20% 0.0025 25.0

The intra-assay and day-to-day precision show that all concentration levels (Low, Medium, High)
comply with the expected concentrations. The repeatability relative standard deviation (%RSDr)
among the results of the three days ranges between 0.21 and 3.34%. The intra-assay %RSDr was
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

7.14%, 5.86% and 3.24% for the Low, Medium and High concentration on Day 1 (10
measurements by one person / the same day).
Vitamin A [mg RE / kg]
Oil
Sample RSDr
Day 1 Day 2 Day 3 MW SDr
(%)
Mean 6.53 6.58 6.18 6.43 ±0.21 3.34
Low SDr ±0.47 ±0.70 ±0.38
RSDr (%) 7.14 10.60 6.20
Mean 13.09 13.05 13.10 13.08 ±0.03 0.21
Medium SDr ±0.77 ±0.38 ±0.50
RSDr (%) 5.86 2.92 3.78
Mean 25.24 24.71 24.45 24.80 ±0.04 1.62
High SDr ±0.82 ±1.29 ±0.42
RSDr (%) 3.24 5.24 1.73

Reproducibility

Inter-person precision was determined by measuring 3 fortified palm oil samples at three different
days by three persons. Palm oil fortified with retinyl palmitate was mixed with non-fortified palm oil
in three concentrations (20, 40 and 80%) and measured in triplicates.

The inter-person precision shows that all three measured concentrations (Low, Medium, High)
comply with the expected concentrations. The % reproducibility relative standard deviation
(%RSDR) coefficient of variation among the results of the three days ranges between 1.07 and
3.92%.
Vitamin A [mg RE / kg]
Oil RSDR
Person 1 Person 2 Person 3 MW
Sample (%)
Mean 6.58 6.72 6.67 6.66 1.07
Low
RSDR (%) 10.60 3.08 2.81
Mean 13.05 13.67 14.11 13.61 3.92
Medium
RSDR (%) 2.92 2.38 2.35
Mean 24.71 25.43 25.82 25.32 2.24
High
RSDR (%) 5.24 6.87 1.22

Recovery

Recovery was calculated as the percentage of expected/spiked concentration of vitamin A in oil

recovered using iCheck® CHROMA and ranged between 98.9% and 100.6%.

Expected Measured
Concentration concentration concentration Recovery (%)
(mg RE/kg) (mg RE/kg)
Low 6.50 6.43 98.9

Medium 13.0 13.08 100.6

High 25.0 24.80 99.2

 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

EXTERNAL VALIDATION OF iCheck® CHROMA

iCheck® CHROMA method performance internally was further validated against external well-
established HPLC reference method performed by Potsdam University (Germany), and against
external use by technicians with limited training in a low-resource setting in West Africa (Helen
Keller International, Abidjan, Côte d’Ivoire).

Validation of internally performed iCheck® CHROMA method against externally performed

HPLC method

To validate the iCheck® CHROMA device against the HPLC reference method, 200 samples
between 0 and 16 mg RE/kg as previously determined by HPLC in Côte d’Ivoire were randomly
selected from 2500 samples collected in a oil fortification household coverage survey in Côte
d’Ivoire. The vitamin A concentrations of the selected oil samples ranged from 0-16 mg RE/kg. All
oil samples were collected in November 2009 and stored in cool, dark, conditions (+4°C) before
and after analysis in-country, and during shipment of the aliquots to Germany. Upon arrival in
Germany, the oil aliquots were split again, and: a) after warming to 30°C for 5 minutes and mixing,
analyzed using the iCheck® CHROMA by experienced technicians involved in the development of
the device, and b) after warming to 30°C for 5 minutes and mixing, analyzed at the University of
Potsdam by HPLC as described below. All technicians were blinded to the results of the other
analyses until results were merged in a database.

All reagents purchased were of HPLC-grade: Methanol, ethyl-acetate, iso-propanol (Sigma,

Deisenhofen, Germany). Oil samples were warmed to 30°C and mixed thoroughly. Samples were
diluted in isopropanol at a ratio of 1:10 (oil: isopropanol; v:v) and the diluted oil sample (20 µL) was
injected without prior saponification and analysed on a reverse-phase column (Repro Sil 70 C18
(5µm; 200x3) Dr. Maisch GmbH, Germany). The solvent system consisted of methanol (A) and
ethyl acetate (B) which were used as the eluent at a flow rate of 0.5 mL/min (HPLC pump
Shimadzu LC 20 AD, Duisburg, Germany). Each analytical run consisted of the following gradient
profile: 3.0 min 0 % B; 3.1 min 50% B; 10.0 min 50% B, 10.1 min 0% B, 15.0 min rinsing time. The
column temperature was 40°C. Absorption of vitamin A esters (palmitate) were identified by
comparison of retention times to the external standard (retinyl palmitate) by use of a photodiode
array detector at a wavelength of 325 nm (PDA Shimadzu SPD-M 20A, Duisburg, Germany). The
exact concentration of the external standard - retinyl palmitate - was measured on a weekly basis
using a UV-VIS spectrophotometer (Specord 205, Analytik Jena AG, Germany). The detection limit
for retinyl pamitate is 50 µg RE/kg.

Data processing and statistics were conducted using Excel version 2007. Besides plotting the two
data sets and calculating the Spearman coefficient and the regression equation, the Bland and
Altman plot was used. The absolute differences of the values in the samples analyzed by the two
methods were calculated. Limits of agreement (LOA) were calculated using:
∆ - 2s = LOAlow
∆ + 2s = LOAhigh
where ∆ is the mean of the difference between the two methods, and s is the standard deviation
(SD) of this difference.

Of the 200 samples collected, sufficient volume for both analyses was available in 189 samples.
The results of this comparison are presented graphically in Figure 2. The equation of the
correlation is y = 0.83x + 0.08, with y being the reference HPLC method. The corresponding
2
Spearman coefficient for the relationship is R = 0.92. The limits of agreement are as follows:
LOAlow is -1.24 mg RE/kg
LOAhigh 2.53 mg RE/kg
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

Figure 1 shows the regression (A) and Bland and Altman plot (B) of the comparison between the iCheck™-iEx and the
reference method (HPLC). The dashed line in A is the line of identity, whereas in B they represent the ± 2 SD.

The Bland-Altman plot (Figure 2) indicates

indicates that there are data points that are outside the 2
standard deviation lines and although tests for identifying outliers exist, no such outliers were
excluded.

Among the samples analyzed, using a target range to define an “adequate fortification level” of 6.4
– 9.6 mg RE/kg (which represents the legally stated fortification concentration in the Ivorian
fortification standards, i.e. 8 mg RE retinol
retinol palmitate/kg oil, plus a fortification precision of ± 20%),
the difference in the percentage of oil samples
samples that were categorized as adequately fortified by the
HPLC reference method and the portable device was 0.6%.

Analysis of oil samples from Côte D’Ivoire using the portable device and by the reference HPLC
method yielded good correlation and agreement, with the portable device measuring consistently
0.6mg RE/kg higher than the reference method. As AOL show, 97.5% of the results obtained by
iCheck® CHROMA do not differ more than 2.8 mg RE/kg (-1.24 to 1.53 mf RE/kg) from the
reference method.

Validation of internally against externally performed iCheck® CHROMA method

Two local technicians in Côte d’Ivoire underwent a brief training module in Côte d’Ivoire, and were
then asked to analyze the samples on-site and then ship those samples to Germany to be
analyzed internally by BioAnalyt. The training included the information on the test, the technology,
and the operation of the instrument and the testing of 10 samples by each technician. Technicians
were blinded to the HPLC results previously obtained with the same samples.

Comparison between the iCheck® CHROMA conducted externally in the field and internally in the
Bioanalyt laboratory: For this comparison, all 200 collected samples contained sufficient amounts
(over 3 mL) and could be measured both by the technicians in Côte d’Ivoire and in Germany. The
results of this comparison are presented graphically in Figure 2. The equation of the correlation is y
= 0.93x + 0.08, with y being the portable device method conducted in the field; the corresponding
2
Spearman coefficient is R =0.94. The LOAlow is -0.60 mg RE/kg and the LOAhigh 1.99 mg RE/kg.
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

Figure 2 shows the regression (A) and Bland and Altman plot (B) of the comparison between the iCheck™-iEx conducted in
the field and in the laboratory. The dashed line in A is the line of identity, whereas in B they represent the ± 2 SD.

The comparison between the analysis of the same oil samples by the iCheck® CHROMA used in
the field by technicians with only brief training, and expert technicians in the BioAnalyt laboratory
gave good agreement, proving the device to be easy to use and to deliver reliable results. This is
an important aspect, as HPLC-based or other spectrophotometric methods require a high level of
expertise and a well-established laboratory infrastructure (including high equipment and
maintenance cost).
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

Int. J. Vitam. Nutr. Res., 81 (5), 2011, 335 –342 335

Original Communication

Quantification of Vitamin A in
Palm Oil Using a Fast and
Simple Portable Device: Method
Validation and Comparison to
High-Performance Liquid
Chromatography
Fabian Rohner1, Simone K. Frey2, Ralf Mothes2, Andrea Hurtienne 3, Simone
Hartong1, Patrice Emery Bosso4, Mai Bui5, Florian J. Schweigert2,3, and
Christine Northrop-Clewes1
1
GAIN – Global Alliance for Improved Nutrition, Geneva, Switzerland
2
BioAnalyt GmbH, Teltow, Germany
3
University of Potsdam, Potsdam-Rehbrücke, Germany
4
Helen Keller International, Abidjan, Côte d’Ivoire
5
Swiss Vitamin Institute, Epalinges, Switzerland

Received: May 8, 2011; Accepted: September 25, 2011

Abstract: Vitamin A deficiency continues to be a global public health problem. Fortification of oil with
vitamin A is considered a cost-effective, feasible strategy to prevent this problem but quality control
poses a challenge to program implementation. To overcome this, we have validated a newly developed
device that quantitatively measures the content of retinyl palmitate in refined palm oil, is simple to use,
and yields immediate results.
Linearity of analysis ranged from 2.5 – 30 mg retinol equivalents (RE)/ kg of palm oil, with 2.5 mg RE/
kg being the determination limit; inter- and intra-assay precision ranged from 1.4 – 7.1 %. Comparison
with a high-performance liquid chromatography method showed high agreement between the methods
(R2 = 0.92; Limits of Agreement: –1.24 mg to 2.53 mg RE/kg), and further comparisons illustrate that
the new device is useful in low-resource settings. This device offers a field- and user-friendly solution
to quantifying the vitamin A content in refined palm oil.

Key words: Vitamin A, retinyl palmitate, fortification, monitoring, rapid test kit, palm oil

DOI 10.1024/0300 – 9831/a000081 Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

336 F. Rohner et al.: Quantification of Vitamin A in Palm Oil

Introduction regulatory monitoring and for program evaluation at

the household level. BioAnalyt GmbH has developed
Data collected by the World Health Organization a simple portable device to rapidly quantify vitamin A
(WHO) between 1995 – 2005 estimates that 190 million concentration in fortified refined palm oil (iCheck™-
pre-school children in 122 countries and 19.1 million iEx , hereafter referred to as “portable device”).
pregnant women in 88 countries have vitamin A de- The objectives of the present study were 1) to vali-
ficiency (VAD). The estimates of VAD are based on date the portable device used by well-trained techni-
the specified population groups having plasma retinol cians against an HPLC reference method, and 2) to
concentrations < 0.7μmol/L. Low vitamin A intake evaluate the field-friendliness and robustness of the
during nutritionally demanding periods in life greatly portable device through its use by technicians with lim-
increases the risks of adverse health consequences, ited training in a low-resource setting in West Africa.
and VAD is therefore considered an important de-
terminant of child survival and is of public health
concern [1].
Of the most commonly used strategies to help re- Material and methods
duce VAD (dietary diversity; food fortification; vita-
min A supplementation through capsules; other public Portable device
health measures such as breastfeeding promotion),
the fortification approach is thought to be the most Method
sustainable and cost-efficient way to reduce VAD [2].
Among the most commonly fortified food vehicles The portable device determines the concentration of
are vegetable oils, sugar, and cereal flours. Vegetable vitamin A (retinol, retinyl acetate, retinyl palmitate)
oils, in particular palm, peanut, and cottonseed oils, in palm oil based on a color reaction described by
are widely consumed in many African and Asian Carr and Price [8]. Antimony trichloride (SbCl3) in
countries, where the prevalence of VAD is of public chloroform and the double bonds of the retinol mol-
health concern [1]. Oil presents an ideal matrix for ecule react in the formation of an anhydroretinylic
the addition of fat-soluble vitamin A, and methods and a retinylic cation, yielding a brilliant blue color.
of fortifying vegetable oil are fairly simple and easy The absorption of the blue color is dependent on
to implement at low cost. Furthermore, oil stabilizes the concentration of the solution and is measured
the fortificant, retinyl acetate or retinyl palmitate, and at 610 nm in the spectrophotometric measuring unit.
delays oxidation of the vitamin [3]. Results are displayed and saved in the measuring
For a successful and sustainable fortification pro- unit, and can be downloaded via a universal serial
gram, ongoing quality assurance monitoring and bus (USB) as a text file (.txt) to a computer. The
regulatory quality control enforcement is required to device can be battery-operated or be connected to
ensure that appropriately fortified oil is produced and the mains (100 – 240 V). The device’s software has a
marketed, and that vitamin A retains its stability over programmed algorithm that converts the absorption
time. In addition, a population-based assessment of units of the blue-color reaction of several measure-
the coverage of adequately fortified oil and its impact ments into mg RE/ kg.
on the vitamin A status of the population is needed to
inform and guide program implementation. Current
semi-quantitative and quantitative analysis methods Material
for assessment of retinyl palmitate in oil are techni-
cally quite demanding and can be expensive; e. g. the The portable device consists of two units, the measur-
qualitative method based on the formation of anhy- ing unit (iCheck™ CHROMA) and the disposable
droretinol [4], the quantitative spectrophotometric [5], reagent vial (iEx™ ELAN), in which the reaction
or high-performance liquid chromatography (HPLC) is performed (Figure 1). The disposable reagent vi-
methods [6]. Qualitative methods do not measure the als contain 2 mL of a mixture of reagents, which are
adequacy of fortification; rather they provide informa- needed for completion of the Carr-Price reaction [8];
tion on presence or absence of a fortificant, which is no additional reagents are required. Both the measur-
often insufficient information [7]. ing unit and reagent vials are commercially available
To ensure adequate fortification levels, simple-to- (www.bioanalyt.com). Currently, shelf life at 25 °C
use quantitative methods that yield immediate and re- has been verified for 12 months. For quality control
liable results are needed at the points of production for purposes the device conducts an auto-control to verify

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

F. Rohner et al.: Quantification of Vitamin A in Palm Oil 337

(30,000 hours), approximately 10 million measure-

ments can be conducted.

Reference method
Method

To conduct the reference method, a gradient reverse-

phase, HPLC system was used; the method is based
on that described by Gimeno and colleagues [9] for
the determination of vitamin E and was adopted for
vitamin A.

Material

All reagents purchased were of HPLC-grade: Metha-

nol, ethyl acetate, and isopropanol (Sigma, Deisen-
hofen, Germany).

Working conditions
Figure 1: Photograph of the prototype of the portable de-
vice that has been validated as described in this paper. To
conduct the analysis, 500 μL of the oil sample is injected Oil samples were warmed to 30 °C and mixed thor-
into the reagent vial using a syringe; the vial is inverted and oughly. Samples were diluted in isopropanol at a ratio
inserted into the device for measurement. As an indication of 1:10 (oil:isopropanol; v:v) and the diluted oil sample
of size, a match has been placed in the picture.
(20 μL) was injected without prior saponification and
analyzed on a reverse-phase column [Repro Sil 70
the emitter and receptor are working correctly. A vi- C18 (5μm; 200 × 3) Dr. Maisch GmbH, Germany].
tamin A standard to control the device is not needed. The solvent system consisted of methanol (A) and
ethyl acetate (B), which were used as the eluent at a
flow rate of 0.5 mL/minute (HPLC pump Shimadzu
Working conditions LC 20 AD, Duisburg, Germany). Each analytical
run consisted of the following gradient profile: 0 – 3
The reagent vials and measuring unit were stored at min – 0 %B; 3 – 3.1 min –50 %B; 3.1 – 10 min – 50 %B;
room temperature (20 – 30 °C) prior to analysis. With- 10 – 10.1 min – 0 %B; and 15.0 min rinsing time. The
out any further preparation, palm oil samples (500 column temperature was 40 °C. Absorption of vitamin
μL) were drawn using a syringe, included as part of A esters (palmitate) were identified by comparison
the portable device test kit, and the 500 μL of oil were of retention times to the external standard (retinyl
injected slowly through the red rubber septum into the palmitate) by use of a photodiode array detector at a
reagent vial. After inverting the vial once (max. 1 sec- wavelength of 325 nm (PDA Shimadzu SPD-M 20 A,
ond), it was placed immediately into the photometric Duisburg, Germany). The exact concentration of the
measuring chamber of the measuring unit. The device external standard – retinyl palmitate – was measured
displays instructions to the analyst at each step of the on a weekly basis using a UV-VIS spectrophotometer
analysis. After five minutes, the photometer displays (Specord 205, Analytik Jena AG, Germany). The de-
the calculated vitamin A concentration in mg RE/kg. tection limit for retinyl palmitate is 50 μg RE/kg. To
During the production process, the measuring unit ensure validity of the results obtained by this reference
was calibrated and good quality control procedures method, the results of a subset (n = 30) of fortified palm
were used at all stages. Because the measuring unit oil samples (range: 0 – 30 mg RE/kg) were compared
uses a stable light source (LED), user calibration is with those obtained by a third laboratory, the Swiss
not required for the anticipated shelf life of the por- Vitamin Institute (Epalinges, Switzerland; external
table device. Based on the half-life of the light source validation), using saponification, followed by retinol

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

338 F. Rohner et al.: Quantification of Vitamin A in Palm Oil

analysis using an HPLC method [10]. Agreement be- using the portable device by experienced technicians
tween the HPLC reference method and the external involved in the development of the device, and b) after
validation was very good [HPLC = 0.9485*(external warming to 30 °C for 5 minutes and mixing, analyzed
validation) + 0.3092 mg RE/kg; R2 = 0.98]. at the University of Potsdam by HPLC as described
above. All technicians were blinded to the results of the
other analyses until results were merged in a database.
Method validation of the portable device
g) Comparison of the use of the portable device in
a) Linearity of the portable device was determined the field and in the laboratory: prior to shipment of
by the injection of a 500-μL sample of fortified palm aliquots to Germany, two local technicians in Côte
oil (fortificant: retinyl palmitate) with the following d’Ivoire underwent a one-hour training module in Côte
calculated concentrations: 2.8, 5.6, 11.2, 16.8, 22.4, and d’Ivoire. The training included information on the test,
28 mg RE/kg; each sample was measured in triplicate. the technology and operation of the instrument, and
the testing of 10 samples by each technician. After-
b) Limits of detection and quantitation were assumed wards they were asked to analyze the same samples
to be interchangeable for this method and determined on-site as described in section f) above. Technicians
by functional sensitivity [11]: unfortified palm oil was were blinded to the HPLC results previously obtained.
increasingly fortified with retinyl palmitate, and mea-
sured, until a coefficient of variation (CV) consistently
below 20 % was reached. Statistical analysis
c) Intra-assay precision: Palm oil fortified with retinyl For the laboratory validation of the method, standard
palmitate was prepared to achieve three concentra- protocols were followed, unless otherwise described.
tions of 6.5, 13.0, and 25.0 mg RE/kg, respectively. The For the field validation, there were two levels of com-
same technician measured each of the three samples parison: a) the comparison of the results obtained by
ten times on the same day, and the CV of the results the portable device (performed in Germany) com-
was calculated. pared to the HPLC reference method, and b) the
comparison of the results obtained using the portable
d) The inter-assay precision was determined by one device by local technicians in-country in a hot climate
person measuring the same three concentrations of with those obtained by well-trained technicians in Ger-
fortified palm oil in triplicate on three different days, many. Data processing and statistics were conducted
and the CV of the results was calculated. Samples were using Excel version 2007. Besides plotting the two data
stored in the dark and at room temperature. sets and calculating the Spearman coefficient and the
regression equation, the Bland and Altman plot was
e) The inter-person precision was determined by three used [12]. The absolute differences of the values in the
people measuring the same three concentrations of samples analyzed by the two methods were calculated.
fortified palm oil in triplicate, over three days, and Limits of agreement (LOA) were calculated using
the CV was calculated. Δ – 2 s = LOAlow
Δ + 2 s = LOAhigh
f) Comparison of the portable device to the reference where Δ is the mean of the difference between the
method: To validate the portable device against the two methods, and s is the standard deviation (SD) of
HPLC reference method, 200 samples between 0 and this difference.
16 mg RE/kg, as previously determined by HPLC in
Côte d’Ivoire, were randomly selected from 2500 re-
fined palm oil samples collected in an oil fortification
household coverage survey in Côte d’Ivoire. The vita- Results
min A concentrations of the selected oil samples ranged
from 0 – 16 mg RE/kg. All oil samples were collected Method validation of the portable device
in November 2009 and stored in cool, dark, conditions
(+4 °C) before and after analysis in-country, and during a) Linearity: The linear range of the portable device
shipment of the aliquots to Germany. Upon arrival in was determined to be 2.5 – 30 mg RE/kg; over this
Germany, the oil aliquots were split again, and: a) after range, R2 was 0.996 and the regression equation was
warming to 30 °C for 5 minutes and mixing, analyzed y = 0.9916x – 0.1182.

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

F. Rohner et al.: Quantification of Vitamin A in Palm Oil 339

Figure 2: Regression (A) and Bland and Altman plot (B) of the comparison between HPLC and the iCheck™-iEx con-
ducted in the laboratory (above), and between the iCheck™-iEx conducted in the field and in the laboratory. The dashed
line in A is the line of identity, whereas in B the dashed lines represent the ± 2 SD.

b) Limits of detection and quantitation: Because of the covery rates of 100.4, 100.7, and 100.9 % for each of
algorithms programmed in the portable device, a sepa- the three levels, respectively (recovery rate: measured
rate definition of limit of detection and quantitation is value/expected value * 100).
not possible. The threshold where the portable device
can reliably distinguish the signal from the background d) Inter-assay precision: The CV for the results of
noise (< 20 % CV) is 2.5 mg RE/kg. one person measuring 3 fortified palm oil samples
over three days were 8.82, 1.35, and 4.00 % for the
c) Intra-assay precision: the CV of the 10 measure- three concentrations (6.5, 13.0, and 25.0 mg RE/kg),
ments conducted within one day by one technician for respectively. The mean concentration of each fortified
the three concentrations of retinyl palmitate (6.5, 13.0, sample was 6.4, 13.1 and 25.3 mg RE/kg, again close
and 25.0 mg RE/kg) were 7.14 %, 5.86 %, and 3.24 % to the expected concentrations (see above). This cor-
respectively, and the mean of the measurements were responds to recovery rates of 98.6, 100.9 and 101.2 %
close to the expected concentrations: 6.5, 13.1, and for each of the three levels, respectively.
25.2 mg RE/kg, respectively. This corresponds to re-

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

340 F. Rohner et al.: Quantification of Vitamin A in Palm Oil

e) The inter-person precision, obtained by calculating governmental food control authorities, non-govern-
the CV from three technicians measuring the three ment organizations (NGOs), and development or-
concentrations of fortified palm oil in triplicate, over ganizations. To fulfill the requirements of users in
three days were 4.53, 3.82, and 4.79 % respectively; low-resource settings, micronutrient assays should be
the respective mean concentrations obtained were 6.6, inexpensive, accurate, reliable, rugged, and well suited
13.6 and 25.5 mg RE/kg. This corresponds to recovery to the context of the developing world.
rates of 100.7, 104.7 and 102.0 % for each of the three To our knowledge, this is the first time a simple,
levels, respectively. quantitative, portable field-friendly device has been
validated for the quantitative analysis of retinyl palmi-
f) Comparison between the portable device and the tate in palm oil. In this paper, the recently developed
reference method: Of the 200 samples collected, suf- portable device has been described. In addition to
ficient volume for both analyses was available in 189 standard method validation tests, our objective was
samples. The results of this comparison are presented to validate the portable device against an HPLC ref-
graphically in Figure 2. The equation of the correlation erence method and to evaluate the field-friendliness
is y = 0.83x + 0.08, with y being the reference HPLC and robustness of the portable device by testing the
method. The corresponding Spearman coefficient for potential difference in analytical performance by com-
the relationship is R2 = 0.92. The limits of agreement paring its use by an experienced laboratory technician
are as follows: LOAlow is -1.24 mg RE/kg and the LOA- to that of technicians only briefly trained.
high 2.53 mg RE/kg. Analysis of oil samples from Côte D’Ivoire using the
The Bland-Altman plot (Figure 2) indicates that portable device and by the reference HPLC method
there are data points that are outside the 2 SD line yielded good correlation and agreement, with the
and although tests for identifying outliers exist, no portable device measuring consistently 0.6 mg RE/
such outliers were excluded. kg higher than the reference method. As the Limits
Among the samples analyzed, using a target of Agreement show, 97.5 % of the results obtained
range to define an “adequate fortification level” of by the device do not differ more than 2.8 mg RE/kg
6.4 – 9.6 mg RE/kg (which represents the legally stated (–1.24 to 1.53 mg RE/kg) from the reference method.
fortification concentration in the Ivorian fortification Using a target range for defining an adequate fortifica-
standards; i. e. 8 mg RE retinyl palmitate/kg oil, plus tion level of 6.4 – 9.6 mg RE/kg, the portable device
a fortification precision of ± 20 %), the difference in showed 27.2 % of oil samples were adequately forti-
the percentage of oil samples that were categorized as fied, compared to 27.8 % by HPLC.
adequately fortified by the HPLC reference method The comparison between the analysis of the same
and the portable device was 0.6 %. oil samples by the portable device used in the field
by technicians with only brief training, and expert
g) Comparison between the portable device method technicians in the BioAnalyt laboratory, showed good
conducted in the field and in the laboratory: For this agreement, proving the device to be simple to use, but
comparison, all 200 collected samples contained suffi- still able to deliver valid results. This is an important
cient amounts (over 3 mL) and could be measured both aspect, as HPLC-based or other spectrophotometric
by the technicians in Côte d’Ivoire and in Germany. methods require a high level of expertise and a well-
The results of this comparison are presented graphi- established laboratory infrastructure (including high
cally in Figure 3. The equation of the correlation is y = equipment and maintenance cost).
0.93x + 0.08, with y being the portable device method The linear range of the portable device was deter-
conducted in the field; the corresponding Spearman mined to be 2.5 – 30 mg RE/kg palm oil and according
coefficient is R2 = 0.94. The LOAlow is 0.60 mg RE/kg to recently published guidelines by WHO/FAO, the
and the LOAhigh 1.99 mg RE/kg. levels of vitamin A currently added to oil are in the
range of 5 – 15 mg/kg [13], a range for which the device
shows high linearity. Hence, the portable device is able
to deliver useful results as to whether oil samples are
Discussion adequately fortified or not.
The validation of the instrument showed that intra-
Not having field-friendly and cost-effective tools to assay precision was between 3 and 7 % and the inter-
monitor micronutrient levels in fortified foods at the assay precision between 1 and 9 %. The inter-person
production site, at national borders, and in popula- precision was also acceptable at between 4 and 5 %.
tions is a common problem among food laboratories, With the variations being below 11.3 %, the precision

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

F. Rohner et al.: Quantification of Vitamin A in Palm Oil 341

can be judged satisfactorily for field applications [14]. Author disclosures

A limitation of the portable device to date is that the
device has only been validated for palm oil. In many FR, AH, PEB, MB, and CC declare no conflict of
countries, the use of other types of oil or mixtures interest; SKF, RM, and FJS are employees of Bio-
of palm oil with other types of vegetable oils is com- Analyt GmbH, the holder of patents related to the
mon. Although a laboratory validation of other oils device presented in this manuscript. The latter con-
such as sunflower, rapeseed, and peanut oil (data not ducted the basic laboratory validation of the method
presented) has been performed and holds promise, a and were involved in planning the field study in Côte
rigorous and systematic validation will need to take d’Ivoire. However, SKF, RM, and FJS were not in-
place to state the device can be used for all types of volved in the analysis using the portable device in
vegetable oil. Côte d’Ivoire or using the reference method. SKF,
When compared to other methods based on the RM, and FJS were involved in conducting the parallel
Carr-Price principle, the use of hazardous reagents analysis using the portable device in Germany, but
(e. g., antimony trichloride) is considerably reduced they were blinded to the results from the reference
in volume in this system, but still, these reagents are method until the results in the databases were locked.
necessary and hence, an appropriate waste manage- SKF, RM, and FJS were not involved in any of the
ment needs to be implemented. statistical analysis.
In conclusion, the portable device presented in this
paper offers a viable solution to the current problem
of lack of quantitative and field- and user-friendly
analytical methods to analyze vitamin A content in References
refined palm oil. Further research and use in the field
is warranted to expand its use for other vegetable oils. 1. WHO (2009) Global prevalence of vitamin A defici-
ency in populations at risk 1995 – 2005: WHO global
database on vitamin A deficiency. Geneva.

2. Edejer, T.T., Aikins, M., Black, R., Wolfson, L., Hu-

Abbreviations and nomenclature tubessy, R. and Evans, D.B. (2005) Cost effectiveness
analysis of strategies for child health in developing
CV Coefficient of variation countries. BMJ 331, 1177.
FAO Food and Agriculture Organization
HPLC High-performance liquid chromatography 3. Dary, O. and Mora, J.O. (2002) Food fortification to
reduce vitamin A deficiency: International Vitamin A
LOA Limit of agreement
Consultative Group recommendations. J. Nutr. 132,
RE Retinol equivalents (or: μg retinol/g, equal 2927S-2933S.
to 3.3 IU of vitamin A)
VAD Vitamin A deficiency 4. Bayfield, R.F. (1971) Colorimetric determination
WHO World Health Organization of vitamin A with trichloroacetic acid. Anal. Bio-
chem. 39, 282 – 287.

5. Kamangar, T. and Fawzi, A.B. (1978) Spectrophoto-

metric determination of vitamin A in oils and fats. J.
Acknowledgments Assoc. Off. Anal. Chem. 61, 753 – 755.

The authors thank Helen Keller International, Côte 6. Bui, M.H. (1994) Simple determination of retinol,
d’Ivoire for allowing additional testing on samples [alpha]-tocopherol and carotenoids (lutein, all-trans-
lycopene, [alpha]- and [beta]-carotenes) in human
collected for a household coverage survey. We also
plasma by isocratic liquid chromatography. Journal
thank the National Health Institute of Côte d’Ivoire of Chromatography B: Biomedical Sciences and Ap-
(Institut National de Santé Publique) who provid- plications 654, 129 – 133.
ed technicians to conduct the field analysis on the
portable device, as well as colleagues at the Global 7. Pandav, C.S., Arora, N.K., Krishnan, A., Sankar, R.,
Alliance for Improved Nutrition (GAIN) for their Pandav, S. and Karmarkar, M.G. (2000) Validation of
spot-testing kits to determine iodine content in salt.
support in the logistics.
Bull. World Health Organ. 78, 975 – 980.

8. Carr, F.H. and Price, E.A. (1926) Colour Reactions

Attributed to Vitamin A. Biochem. J. 20, 497 – 501.

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE

342 F. Rohner et al.: Quantification of Vitamin A in Palm Oil

9. Gimeno, E., Castellote, A.I., Lamuela-Raventos, 13. Bland, J.M. and Altman, D.G. (1986) Statistical me-
R.M., de la Torre, M.C. and Lopez-Sabater, M.C. thods for assessing agreement between two methods
(2000) Rapid determination of vitamin E in vegetable of clinical measurement. Lancet 1, 307 – 310.
oils by reversed-phase high-performance liquid chro-
matography. J. Chromatogr. A 881, 251 – 254. 14. Fraser, C.G., Hyltoft Petersen, P., Libeer, J.C. and
Ricos, C. (1997) Proposals for setting generally appli-
10. European Committee for Standardization (2000)
Determination of vitamin A in foodstuffs by high cable quality goals solely based on biology. Ann. Clin.
performance liquid chromatography – Part 1: Measu- Biochem. 34 ( Pt 1), 8 – 12.
rement of all-trans-retinol and 13-cis-retinol: EN
12823 – 1 – 2000. Fabian Rohner

11. Schambeck, C.M., Bedel, A. and Keller, F. (1998) GAIN – Global Alliance for Improved Nutrition
Limit of quantification (functional sensitivity) of the PO Box 55
new IMx Tacrolimus II microparticle enzyme immu- CH-1211 Geneva
noassay. Clin. Chem. 44, 2217.
Switzerland
12. Allen, L.H., de Benoist, B., Dary, O. and Hurrell, Tel.: +41 (0)22 749 18 05
R.F. (2006) Guidelines on food fortification with mi- Fax: +41 (0)22 749 18 51
cronutrients. World Health Organization, Geneva. [email protected]

Int. J. Vitam. Nutr. Res. 81 (5) © 2011 Hans Huber Publishers, Hogrefe AG, Bern
 INGR-16 / INGREDIENTS
SPSFAM EXPERT REVIEW PANEL
DO NOT DISTRIBUTE
TM
Instruction Guide
Read all of the procedures before starting. Be sure to study the
separate iEx™ TECH Syringe Use & Mixing guide first.

1.1 1.2 1.1 – BLANK MEASUREMENT

Place your iCheck™ CHROMA on a stable flat surface
and switch on. Take a new vial and place it in the
measurement chamber. Press the measurement key
for the blank measurement, wait some seconds.
1.2 – TAKE UP 0.5 mL SAMPLE
Use the green syringe without a needle attached and
take up at least 0.5 mL of an oil sample. Clean the end
of the syringe with a tissue and attach a needle. Turn
the syringe in order to release air bubbles. Adjust the
volume of the sample to exact 0.5 mL by ejecting
excessive volume out, into the tissue.

1.3 1.4 1.3 – INJECT THE SAMPLE.

Now slowly inject the sample into the iEx™ ELAN vial
which has been used for blank measurement. Take
special care to inject the sample along the side of the
glass, and above the clear liquid in order to produce
two distinct liquids.
1.4 - If you have injected the sample correctly, the
contents of your vial should look very similar to the
image at left. Note: A very small amount of blue liquid
may appear where the sample has met the clear liquid.

1.5 1.6 1.5 – PRESS MEASUREMENT

Now that your sample is ready, press the
measurement button on the iCheck™ CHROMA and
follow the procedures as indicated by the count-down
on the display.
1.6 – INVERT THE VIAL.
At ZERO as indicated on the display of the iCheck™
CHROMA, invert the vial overhead twice and insert the
vial into the iCheck™ within 2 seconds. Note: The
content of the vial will instantly turn into a bluish
colour, if the sample contains vitamin A.

1.7 1.8 1.7 – INSERT THE VIAL INTO THE iCHECKTM.

Insert the vial into your iCheck™ CHROMA device as
long as the display indicates “Insert vial”.

1.8 - Your iCheck™ CHROMA will now analyse your

sample. This will take approx. 30 seconds. The result is
displayed in mg RE/kg.

Note: For detailed instructions on using the iCheck™

please refer to the iCheck™ CHROMA User Manual.