lable at ScienceDirect

Clinical Nutrition 39 (2020) 3512e3519

Contents lists avai

Clinical Nutrition

journal homepage: http://www.elsevier.com/locate/clnu

Original article

Vitamin A and iron status of children before and after treatment of
uncomplicated severe acute malnutrition

Suvi T. Kangas a, b, *, C?ecile Salp?eteur b, Victor Niki?ema c, Leisel Talley d, Andr?e Briend a, e,
Christian Ritz a, Henrik Friis a, Pernille Kaestel a

a Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark
b Expertise and Advocacy Department, Action Against Hunger (ACF), Paris, France
c Nutrition and Health Department, Action Against Hunger (ACF) Mission, Burkina Faso
d Centers for Disease Control and Prevention, Atlanta, GA, USA
e Center for Child Health Research, University of Tampere School of Medicine, Tampere University, FIN-33014, Tampere Finland

a r t i c l e i n f o

Article history:
Received 15 January 2020
Accepted 10 March 2020

Keywords:
Vitamin A
Iron
Micronutrient
Severe acute malnutrition
Children
Ready-to-use therapeutic food

* Corresponding author. Expertise and Advocacy
Hunger (ACF), 14/16 Boulevard Douaumont – CS 80
France.

E-mail address: [email protected] (S.T. Ka

https://doi.org/10.1016/j.clnu.2020.03.016
0261-5614/© 2020 The Author(s). Published by Elsevie

s u m m a r y

Background & aims: Treatment of children with uncomplicated severe acute malnutrition (SAM) is based
on ready-to-use therapeutic foods (RUTF) and aims for quick regain of lost body tissues while providing
sufficient micronutrients to restore diminished body stores. Little evidence exists on the success of the
treatment to establish normal micronutrient status. We aimed to assess the changes in vitamin A and
iron status of children treated for SAM with RUTF, and explore the effect of a reduced RUTF dose.
Methods: We collected blood samples from children 6e59 months old with SAM included in a rando-
mised trial at admission to and discharge from treatment and analysed haemoglobin (Hb) and serum
concentrations of retinol binding protein (RBP), ferritin (SF), soluble transferrin receptor (sTfR), C-reac-
tive protein (CRP) and a1-acid glycoprotein (AGP). SF, sTfR and RBP were adjusted for inflammation (CRP
and AGP) prior to analysis using internal regression coefficients. Vitamin A deficiency (VAD) was defined
as RBP < 0.7 mmol/l, anaemia as Hb < 110 g/l, storage iron deficiency (sID) as SF < 12 mg/l, tissue iron
deficiency (tID) as sTfR > 8.3 mg/l and iron deficiency anaemia (IDA) as both anaemia and sID. Linear and
logistic mixed models were fitted including research team and study site as random effects and adjusting
for sex, age and outcome at admission.
Results: Children included in the study (n ¼ 801) were on average 13 months of age at admission to
treatment and the median treatment duration was 56 days [IQR: 35; 91] in both arms. Vitamin A and iron
status markers did not differ between trial arms at admission or at discharge. Only Hb was 1.7 g/l lower
(95% CI ?0.3, 3.7; p ¼ 0.088) in the reduced dose arm compared to the standard dose, at recovery. Mean
concentrations of all biomarkers improved from admission to discharge: Hb increased by 12% or 11.6 g/l
(95% CI 10.2, 13.0), RBP increased by 13% or 0.12 mmol/l (95% CI 0.09, 0.15), SF increased by 36% or 4.4 mg/l
(95% CI 3.1, 5.7) and sTfR decreased by 16% or 1.5 mg/l (95% CI 1.0, 1.9). However, at discharge, micro-
nutrient deficiencies were still common, as 9% had VAD, 55% had anaemia, 35% had sID, 41% had tID and
21% had IDA.
Conclusion: Reduced dose of RUTF did not result in poorer vitamin A and iron status of children. Only
haemoglobin seemed slightly lower at recovery among children treated with the reduced dose. While
improvement was observed, the vitamin A and iron status remained sub-optimal among children treated
successfully for SAM with RUTF. There is a need to reconsider RUTF fortification levels or test other
potential strategies in order to fully restore the micronutrient status of children treated for SAM.
© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Department, Action Against
060, 75854 Paris Cedex 17,

ngas).

r Ltd. This is an open access article

1. Introduction

Severe acute malnutrition (SAM), defined as severe wasting, low
mid-upper arm circumference (MUAC) and/or oedema, is wide-
spread among children in low-income countries. While incidence

under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Abbreviations

AGP a1-acid glycoprotein
CMAM community-based management of acute

malnutrition
CRP C-reactive protein
Hb haemoglobin
IDA iron deficiency anaemia
MAM moderate acute malnutrition
MUAC mid-upper arm circumference
RBP retinol binding protein
RUTF ready-to-use therapeutic food
SAM severe acute malnutrition
SF serum ferritin
sID storage iron deficiency
tID tissue iron deficiency
sTfR soluble transferrin receptor
VAD vitamin A deficiency
WHO World Health Organisation
WHZ weight-for-height z-score

S.T. Kangas et al. / Clinical Nutrition 39 (2020) 3512e3519 3513

data are lacking, the number of children with severe wasting alone,
at any time, was estimated at 16.6 million children in 2018 [1]. The
treatment of SAM without medical complications consists of ready-
to-use-therapeutic food (RUTF) and one week of antibiotic treat-
ment [2]. RUTFs are fortified energy-dense pastes designed to fulfil
all the nutritional needs of children during recovery from SAM [3].
The aim of the treatment is to enable a rapid regain of lost body
tissues while providing sufficient micronutrients to restore
diminished body stores. However, little evidence exists on the
success of the treatment to restore sufficient micronutrient status
by recovery.

Vitamin A deficiency (VAD) affects 190 million children under 5
years of age worldwide and is associated with increased risk of
morbidity and mortality, and in the most severe form can lead to
blindness [4]. In Brazil 41.2% and in Congo up to 98% of hospitalised
malnourished children had VAD [5,6]. RUTFs contain 0.7e1.0 mg of
vitamin A per 92 g sachet [3] and each child typically receives 2e3
sachets per day, which translates to 3e8 times the daily recom-
mended intake of 0.4 mg for healthy children in this age group [7].
However, no data exist on the vitamin A status of children with
uncomplicated SAM at admission to and discharge from
community-based treatment. Ensuring a normal vitamin A status
by discharge would seem crucial, considering its role in healthy
growth [8e10].

Similarly, anaemia and iron deficiency are common world-
wide; WHO estimates that nearly half of all children under 5 years
of age are anaemic and/or iron deficient [11]. Iron deficiency in
childhood is associated with impaired growth and development
[12,13]. Previous studies on children with SAM have reported high
prevalence of anaemia at admission: up to 95% in inpatient set-
tings in India [14,15], 80% in Burkina Faso [16], and 49% among
uncomplicated cases in Malawi [17]. RUTFs contain 9e11 mg of
iron per sachet [3] with 2 sachets equalling the daily recom-
mended intake of 18.6 mg (assuming a low bioavailability) for
healthy children [7]. A study in Malawi showed that 25% of chil-
dren were still anaemic when discharged from SAM treatment
[17]. Additionally, they observed that the proportion of children
with iron deficiency (defined as soluble transferrin receptor
concentrations > 8.3 mg/l when adjusted for inflammation) did
not decrease from admission (56%) to discharge (58%). Doc-
umenting the response to SAM treatment in different contexts

would be important in order to guide possible protocol revisions
aimed at improving the iron status of treated children.

The objective of this paper is to assess the change in vitamin A
and iron status of children treated for uncomplicated SAM with
RUTF, and explore the effect of a reduced RUTF dose on the vitamin
A and iron status of recovered children.

2. Methods

2.1. Study area and participants

This study was a cohort study nested in to the MANGO study, a
randomised non-inferiority trial testing the efficacy of a reduced
RUTF dose compared to standard RUTF dose in the treatment of
uncomplicated SAM among children 6e59 months of age [18]. We
previously reported a non-inferior weight gain velocity [18] and a
similar tissue gain pattern (in press) with a reduced RUTF dose
compared to the standard dose.

The study was conducted from October 2016 to December 2018
in the east of Burkina Faso, where the prevalence of wasting was
10% in 2016 [19]. Intestinal parasites are common with 86% of
school age children in neighbouring regions infected [20] and up to
16% are estimated to present with haemoglobinopathies [21]. Six-
monthly vaccination campaigns with vitamin A supplementation
(100 000 IU to 6e11 month olds; 200 000 IU to 12e59 month olds)
were organised in the area throughout the study period. Commu-
nity and health centre level sensitisation on appropriate infant and
young child feeding (IYCF) practices including promotion of the use
of fortified infant flours were supported by various NGOs working
in the area.

As described previously [18], participants were recruited at 10
health centres. Eligibility criteria consisted of having a weight-for-
height z-score (WHZ) < ?3 and/or a MUAC < 115 mm, no oedema
and the absence of medical complications as per the national
protocol [22]. Children failing the appetite test, having received
SAM treatment in the past 6 months, with known peanut or milk
allergy, disability affecting food intake or whose caregiver were
unable to comply with the weekly visit schedule were excluded.

2.2. Randomisation and study procedures

For a full description of the methodology, please refer to Ref.
[18]. In brief, after obtaining caregiver consent, children were
randomised individually to receiving either a standard dose of RUTF
throughout treatment or a reduced dose of RUTF from the 3rd
treatment week onwards (Table 1). The vitamin A and iron content
per daily dose and per arm is described in Table 1. The content has
been calculated based on RUTF nutrient specifications [3]. Medical
treatment was provided for all children per the national protocol
[22] including a 7-day course of amoxicillin at admission
(50e100 mg/kg/d) and albendazole at the second treatment visit
for children ? 12 months of age (200 mg to 12e23-month-olds;
400 mg to ?24-month-olds). Any missed routine vaccinations or 6-
monthly vitamin A supplementations were caught-up at admis-
sion. Children were treated until reaching anthropometric recovery
criteria or until defaulting from treatment, being transferred to
inpatient care, diseased or declared non-responding after a
maximum of 16 weeks of treatment. Anthropometric recovery was
defined as a WHZ ? ?2 for those admitted with a WHZ < ?3 only,
MUAC ? 125 mm for those admitted with a MUAC < 115 mm only,
or WHZ ? ?2 and MUAC ? 125 mm for those admitted with
WHZ < ?3 and MUAC < 115 mm, on two consecutive visits and
absence of any illness.

Venous blood was collected in Vacutainer® (BD, New Jersey,
USA) clot activator tubes at admission and at discharge from

Table 1
Vitamin A and iron content per daily RUTF dose with reduced and standard dose.

Weight (kg) Standard RUTF dosea Reduced RUTF dosea

RUTF quantity/day
(sachets)

Vitamin A/daily RUTF
dose (mg)

Iron/daily RUTF
dose (mg)

RUTF quantity/day
(sachets)

Vitamin A/daily
RUTF dose (mg)

Iron/daily RUTF
dose (mg)

3.0e3.4 1.1 1.1 11.4 1.0 1.0 10.0
3.5e4.9 1.4 1.4 14.3 1.0 1.0 10.0
5.0e6.9 2.1 2.1 21.4 1.0 1.0 10.0
7.0e9.9 2.9 2.9 28.6 2.0 2.0 20.0
10.0e14.9 4.3 4.3 42.9 2.0 2.0 20.0

RUTF, ready-to-use therapeutic foods.
a Reduced dose arm received a standard dose of RUTF for the first 2 weeks and then the reduced dose from 3rd treatment week onwards.

S.T. Kangas et al. / Clinical Nutrition 39 (2020) 3512e35193514

treatment. Two attempts were made for blood collection from a
total of 3 possible sites. If infeasible, a finger prick was used for a
rapid diagnostic test for malaria (SD Bioline, Abbott, Illinois, USA)
and to measure Hb with a HemoCue® 301 device (Hemocue AB,
Sweden). The accuracy of the HemoCue® was monitored monthly
using commercial controls (Eurotrol, Kentucky, USA) and the values
were within certified values. Blood samples were transported in a
cold box at 2e8 ?C to the field laboratory where samples were
stored in a fridge at 2e10 ?C for maximum 24 h. Serum was isolated
following centrifugation at 3000 rotations per minute for 5 min
(EBA 20 S Hettich, Germany) and stored at ?20 ?C until shipment to
VitMin Lab in Willstaedt, Germany for analysis. Serum ferritin (SF),
soluble transferrin receptor (sTfR), retinol binding protein (RBP), C-
reactive protein (CRP), a1-acid glycoprotein (AGP), were deter-
mined using a combined sandwich enzyme linked immunosorbent
assay [23].

A thorough medical history and evaluation of morbidities was
performed by a study nurse at admission and at each weekly visit.
Fever was defined as an arm pit temperature of ?37.5 ?C and led
systematically to rapid testing for malaria with a positive result
defining malaria. Acute respiratory illness (ARI) was defined as
cough reported by caregiver in the past week or diagnosed by study
nurse during visit. Diarrhoea included acute, persistent or dysen-
teric forms and was defined as 3 or more loose stools per day as
reported by caregiver in the past week or diagnosed by study nurse.
Medical treatment offered included a 3-day course of arthemeter
(2 ? 20 mg/d) and lumefantrine (2 ? 120 mg/d) in case of malaria
and a 7-day course of amoxicillin (50e100 mg/kg/d) in case of ARI
or diarrhoea.

2.3. Outcomes and adjustment for inflammation

As SF, sTfR and RBP are affected by inflammation they were
adjusted prior to analysis using internal regression coefficients as
previously described [24e26]. Log-transformation was applied for
RBP, SF, sTfR, CRP and AGP due to non-normal distribution. The
coefficients were 0.178 for log transformed CRP (logCRP) and 0.167
for log transformed AGP (logAGP) when adjusting log transformed
SF, 0.004 for logCRP and 0.228 for logAGP when adjusting log
transformed sTfR and ?0.071 for logCRP and ?0.028 for logAGP
when adjusting log transformed RBP. Based upon recommenda-
tions from the BRINDA study group [24e26], adjustments were not
applied below first deciles of CRP and AGP corresponding to
0.20 mg/l for CRP and 0.43 g/l for AGP in the current data.

Anaemia was defined as Hb < 110 g/L [27], storage iron defi-
ciency (sID) as inflammation adjusted SF (SFadj) < 12 mg/L [28],
tissue iron deficiency (tID) as inflammation adjusted sTfR >8.3 mg/l
and iron-deficiency anaemia (IDA) as Hb < 110 g/L and SFadj < 12 mg/
L. Vitamin A deficiency (VAD) was defined as inflammation
adjusted RBP <0.7 mmol/l. For descriptive purposes, inflammation
categories were defined as proposed by Thurnham et al. [29].

2.4. Statistical analysis

Data were collected via tablets using Open Data Kit software. All
statistical analyses were carried out using Stata 15 (Stata Corp,
Texas, USA). Characteristics of the study population were summa-
rized as percentages and means ±SDs or, if not normally distributed
as median (IQR). Linear and logistic mixed models were used to
assess differences in means and proportions at admission, with
study team and health centre included as random effects.

Linear and logistic mixed models were used to evaluate change
from admission to discharge in mean biomarker concentrations or
proportions of deficiencies, as appropriate. Study team, health
centre, and child id were included as random effects. Both unad-
justed and adjusted models (including sex and age) were fitted.
Similarly, the effect of RUTF dose on mean biomarker concentra-
tions and proportions of deficiencies were analysed with linear and
logistic mixed models. Study team and health centre were included
in the models as random effects. Unadjusted and adjusted models
(including sex, age, and outcome measure at admission) were
fitted.

Results were presented as estimated mean differences with 95%
CI in means and proportions. Right-skewed outcomes were
logarithm-transformed prior to analysis. Subsequently, back
transformation was applied to log transformed variables to esti-
mate mean differences in original units [30]. Model checking was
based on residual and normal probability plots.

2.5. Ethical considerations

Children not included in the study but diagnosed with SAM
were referred to standard care at the health centre. Children who
did not recover from SAM within 16 weeks of treatment were
subsequently referred to standard care. The study was carried out
in accordance with the Declaration of Helsinki. Field registries were
kept in a locked facility. The study was approved by the national
Ethics Committee of Burkina Faso (deliberation number 2015-12-
00) and the national clinical trials board of Burkina Faso (Direction
G?en?erale de la Pharmacie, du M?edicament et des Laboratoires
(DGPML)). The trial was registered in the International Standard
Randomized Controlled Trial Number (ISRCTN) registry as
ISRCTN50039021.

3. Results

Out of the total 801 children included in the trial, 402 were
randomised to the reduced RUTF dose and 399 to the standard
dose. Hb was analysed for all admitted children and additional
biomarkers including SF, sTfR, RBP, CRP and AGP were analysed for
714 (89%) of admitted children. At discharge, we analysed Hb for
425 (98%) of recovered children and additional biomarkers were
analysed for 383 (90%) of recovered children (see Fig. 1).

S.T. Kangas et al. / Clinical Nutrition 39 (2020) 3512e3519 3515

As previously detailed [18], non-recovered children represented
a heterogeneous group of children referred to inpatient care (20%),
defaulters (12%), lost to follow-up (0.1%), deaths (0.1%), non-
responders (13%) and false discharges (3%). Because of ethical and
operational constraints, vitamin A and iron biomarker data were
only obtained from 30% of these children, mostly from the non-
responders.

Baseline characteristics of children did not differ between the
study groups in terms of morbidity, inflammatory markers, and
vitamin A and iron status markers (Table 2). For the full cohort, the
mean age was 13.4 months at admission and 49% were male.
Approximately 78% of children reported or were diagnosed with an
illness at admission with 33% presenting with positive malaria
rapid test. Most children had elevated serum CRP (42%) or AGP
(64%) at admission. The median length of stay in treatment was 56
days [IQR: 35; 91] in both arms.

Mean concentrations of all vitamin A and iron status biomarkers
improved from admission to discharge: Hb increased by 12%, RPB
by 12% and SF by 36% while sTfR decreased by 16%. These changes
resulted in fewer children being under the deficiency cut-offs at
discharge (Table 3). Vitamin A deficiency (RBP < 0.7 mmol/l)
decreased from 25% at admission to 9% at discharge (Fig. 2).
Anaemia (Hb < 110 g/l) decreased from 77% at admission to 55% at
discharge. Storage iron deficiency (SF < 12 mg/l) decreased from 50%
at admission to 35% at discharge. Tissue iron deficiency
(sTfR > 8.3 mg/l) decreased from 55% at admission to 41% at
discharge. Iron deficiency anaemia decreased from 42% at admis-
sion to 21% at discharge.

At recovery, no differences were found in the mean concentra-
tions of RBP, SF or sTfR or on the percentages of VAD, sID, tID or IDA
between the study arms (Table 4). However, the reduced dose was
associated with a 1.7 g/l lower haemoglobin concentration and 9%
higher anaemia prevalence, although these differences were only

Table 2
Characteristics of children at admission to SAM treatment receiving a reduced or a stand

Characteristics n Red

Age, months 801 13.
Male, % 801 49.
Morbidity

Any illness, % 801 79
Malaria, % 801 33
Acute respiratory illness, % 801 31
Diarrhoea, % 801 25
Other, % 801 11
Fever, % 801 27

Inflammation
C-reactive protein (CRP), mg/l 714 2.5
>5 714 41

a1-acid glycoprotein (AGP), g/l 714 1.3
>1 714 62

Inflammation categories
CRP ? 5 mg/l and AGP ? 1 g/l 714 33
CRP > 5 mg/l and AGP ? 1 g/l 714 5 (1
CRP > 5 mg/l and AGP > 1 g/l 714 36
CRP ? 5 mg/l and AGP > 1 g/l 714 26

Vitamin A and iron
Retinol binding protein, mmol/l 714 0.8
<0.7 mmol/l, % 714 26

Haemoglobin, g/l 801 96.
<110 g/l, % 801 78

Ferritin, mg/l 714 11.
<12 mg/l, % 714 52

Soluble transferrin receptor, mg/l 714 9.0
>8.3 mg/l, % 714 55

Iron deficiency anaemia, % 714 43

Data are mean ± SD, median [IQR] or proportion (n) with p-value for difference using log
RUTF, ready-to-use therapeutic food.

marginally significant. Similar results were found when including
all discharge categories as opposed to only recovered (S1 Table) or
without adjustments for age, sex and outcome measure at
admission.

4. Discussion

In this study, a high proportion of children with SAM had sub-
optimal vitamin A and iron status at admission and these de-
ficiencies were only partly corrected by discharge with 56%
anaemic, 21% IDA and 9% VAD. There was no difference in mean
RBP, SF or sTfR at discharge between children who had received the
reduced and the standard RUTF dose. However, the mean Hb con-
centration was slightly lower in children receiving the reduced
compared to standard dose, although this difference was only
marginally significant.

High rates of anaemia have been reported in previous studies
among malnourished children [17,31,32], however few have re-
ported change during treatment. In our study, anaemia decreased
significantly from 78% to 56%, similarly to Cichon’s observation
among children with MAM treated for 12 weeks where anaemia
decreased from 70% at admission to 53% at discharge [31].

Iron deficiency explains only half of the anaemia; 40% were IDA
while 78% were anaemic at admission, and 21% were IDA and 56%
anaemic at discharge. RUTFs contain many nutrients such as vita-
mins A, C, D, E, B2, B6, B12, folate, copper and zinc, whose defi-
ciency can cause anaemia [33e39]. Thus, treatment with RUTF
could correct nutritional anaemia not caused by iron deficiency. Yet,
haemoglobinopathies and inflammation may also cause anaemia
[36], meaning nutritional supplementation may not fully correct it.

We observed a trend towards 1.7 g/l lower mean Hb and 9-
percentage points more anaemia among children who received
the reduced compared to standard RUTF dose. A similar trend was

ard RUTF dose.

uced RUTF Standard RUTF p-value

3 ± 8.6 13.4 ± 8.9 0.79
5 (199) 49.4 (197) 0.97

(316) 78 (311) 0.82
(134) 32 (129) 0.75
(126) 31 (125) 0.94
(101) 25 (99) 0.93
(45) 13 (53) 0.34
(108) 25 (100) 0.57

[0.6e12.6] 3.3 [0.7e13.2] 0.59
(149) 42 (150) 0.74
[0.8e1.9] 1.3 [0.8e2.0] 0.19
(225) 65 (229) 0.48

(118) 31 (111) 0.72
8) 4 (13) 0.39
(131) 39 (137) 0.49
(94) 26 (92) 0.99

8 [0.70e1.12] 0.90 [0.71e1.11] 0.94
(93) 25 (87) 0.73
6 ± 16.9 94.7 ± 17.8 0.11
(314) 79 (314) 0.85
4 [5.1e31.6] 12.9 [5.1e31.2] 0.58
(187) 48 (168) 0.26
[6.4e13.8] 9.3 [6.5e14.2] 0.50
(200) 57 (200) 0.76
(156) 40 (141) 0.37

istic or linear mixed models with study site and research team as random variables.

Table 3
Change in vitamin A and iron status biomarkers from admission to discharge from SAM treatment.

Outcome Admission Discharge Changea

n values n values mean (95% CI) p-value

RBP, mmol/l 714 0.89 [0.70e1.11] 473 1.00 [0.83e1.21] 0.12 (0.09; 0.15) <0.001
Hb, g/l 801 95.7 ± 17.4 537 106.8 ± 13.5 11.6 (10.1; 13.0) <0.001
SF, mg/l 714 12.1 [5.1e31.6] 474 16.1 [9.6e28.6] 4.4 (3.1; 5.7) <0.001
sTfR, mg/l 714 9.2 [6.5e14.1] 474 7.8 [6.3e10.7] ?1.5 (?1.9; ?1.0) <0.001

Values are median [IQR] for RBP, SF and sTfR and mean ± SD for Hb.
RBP, retinol binding protein adjusted for inflammation; Hb, haemoglobin; SF, serum ferritin adjusted for inflammation; sTfR, soluble transferrin receptor adjusted for
inflammation.

a Change in concentrations when adjusting for sex and age and using linear mixed models with id, study site and research team as random effects.

Fig. 1. Patient flow chart. * including serum ferritin (SF), soluble transferrin receptor (sTfR), retinol binding protein (RBP), C-reactive protein (CRP), a1-acid glycoprotein (AGP). Hb,
haemoglobin; RUTF, ready-to-use therapeutic food.

Fig. 2. Deficiencies in vitamin A and iron status biomarkers at admission to and
discharge from SAM treatment. Data are means with 95% CI when using logistic mixed
models including study team, health centre and id as random factors and adjusting for
sex and age. VAD: vitamin A deficiency defined as retinol binding protein adjusted for
inflammation < 0.7 mmol/l; sID: storage iron deficiency defined as serum ferritin
adjusted for inflammation < 12 mg/l; tID: tissue iron deficiency defined as soluble
transferrin receptor adjusted for inflammation > 8.3 mg/l; IDA: iron deficiency
anaemia defined as haemoglobin < 110 g/l and serum ferritin adjusted for
inflammation < 12 mg/l.

S.T. Kangas et al. / Clinical Nutrition 39 (2020) 3512e35193516

not observed in SF nor sTfR. The observed effect on Hb might be due
to other micronutrients in the RUTF that, when given at smaller
quantities, become insufficient to correct low Hb.

While we observed a favourable change in all iron status
markers, it was insufficient to achieve normal values upon
discharge. In Malawi non-standard formulations of RUTF were

tested including three times the iron of the standard RUTF [17]. The
alternative formulations led to significantly lower rate of anaemia
by discharge; 12e18% compared to 25% with standard RUTF [17].
These results combined with our observation of high rates of
anaemia and iron deficiency at discharge from standard treatment
suggest that children with SAM might benefit from RUTF formu-
lations with higher iron content. Currently, a sachet of RUTF con-
tains 9e11 mg of iron with 2e3 sachets providing a maximum of
twice the recommended daily intake (12e19 mg/d) of a healthy
7e59 month old child [7]. Most of the iron requirements for chil-
dren under 3 years of age are related to growth [7]. Children
recovering from malnutrition, with an estimated weight gain ve-
locity 3e5 times higher [40] than normal children [41], would be
expected to have proportionally high iron requirements. This said,
it is usually accepted that correcting iron deficiency takes 3e6
months [42,43] and therefore post-discharge interventions
designed to correct the remaining micronutrient deficiencies
should be considered.

However, iron interventions raise several issues. First, potential
effects of additional iron on morbidity should be investigated as
iron can increase the risk and severity of diarrhoea, fever, vomiting
and hospitalisations [44e47]. Second, additional iron might have a
negative effect on the microbiota [47e50] that is already altered in
malnourished children [51,52]. Third, additional iron might nega-
tively affect growth especially among iron replete children
[44,53e55]. Slower weight gain velocity was also observed in the
Malawian study among children with SAM receiving RUTF with
higher than standard iron content [56]. Fourth, high iron content in

Table 4
Vitamin A and iron status at discharge among children recovered from SAM and treated with a reduced or a standard RUTF dose.

Outcome n Reduced RUTF dose Standard RUTF dose Differencea

mean (95% CI) p-value

RBP, mmol/l 382 0.99 [0.83e1.23] 0.99 [0.83e1.17] 0.02 (?0.03; 0.08) 0.38
<0.7 mmol/l, % 382 10 (18) 9 (17) 0 (?5; 4) 0.87

Hb, g/l 425 107.1 ± 11.8 108.6 ± 11.2 ?1.7 (?3.7; 0.3) 0.088
<110 g/l, % 425 59 (124) 51 (110) 9 (?1; 19) 0.074

SF, mg/l 383 16.1 [9.4e27.2] 16.5 [10.0e28.4] 0.8 (?1.8; 3.3) 0.56
<12 mg/l, % 383 36 (69) 31 (60) 2 (?8; 13) 0.65

sTfR, mg/l 383 7.8 [6.4e11.2] 8.1 [6.4e10.8] ?0.1 (?0.8; 0.7) 0.82
>8.3 mg/l, % 383 43 (82) 43 (83) ?3 (?13; 8) 0.63

IDA, % 376 23 (43) 19 (35) 1 (?7; 9) 0.75
Values are median [IQR] for RBP, SF and sTfR, mean ± SD for Hb and proportions (n).
RBP, retinol binding protein adjusted for inflammation; Hb, haemoglobin; SF, serum ferritin adjusted for inflammation; IDA, iron deficiency anaemia (defined as Hb < 110 g/l
and SF < 12 mg/l); sTfR, soluble transferrin receptor adjusted for inflammation.

a Difference when adjusting for age, sex and outcome measure at admission using linear and logistic mixed models including research team and study site as random factors.

S.T. Kangas et al. / Clinical Nutrition 39 (2020) 3512e3519 3517

infant formula has been associated with impairment of cognitive
development [57,58]. Finally, additional iron could interfere with
the absorption of other trace elements such as copper and zinc
[59e62]. The choice of the form of iron would also be crucial in
minimising the harmful and maximising the positive effects [63].

To our knowledge this is the first study reporting the response of
vitamin A status to treatment with RUTF of children with uncom-
plicated SAM; VAD decreased from 25% at admission to 9% at
discharge. Previous studies conducted in inpatient settings have
reported high rates of VAD; 41% in Brazil [5] and 81% in Bangladesh
[64] had VAD. However, neither study adjusted serum retinol for
inflammation and therefore may have overestimated VAD preva-
lence [65].

That 9% of children remained VAD while discharged as recov-
ered from SAM deserves attention. This was observed despite daily
RUTF intake for a mean duration of 2 months and in the context of
twice a year high-dose vitamin A supplementation campaigns. The
16-percentage point decrease in VAD during treatment reflects an
effective response to RUTF, but the question remains whether a
higher vitamin A fortification level of RUTF could eliminate VAD. A
sub-optimal vitamin A status is also related to anaemia and iron
deficiency and thus eradicating VAD might also help in further
decreasing anaemia [36,66].

However, there remains a lot of uncertainty as to the upper safe
limit of vitamin A intake: the WHO advises against intakes over
0.9 mg/d [7] among infants but without evidence from studies
showing adverse effects, and the IOM gives an upper-limit of
0.6 mg/d for children under 3 years of age [67]. Toxicity has been
observed with daily intakes of 0.45 mg/kg for 6 months [68] which
would translate to around 4 mg/d for a normal 1 year old. These
limits are for healthy children with normal stores but considering
that malnourished children seem to start with deficiencies, higher
safety cut-offs would probably apply. The vitamin A content of RUTF
is about 0