Content uploaded by Sergio Paredes-Solís
Author content
All content in this area was uploaded by Sergio Paredes-Solís on Oct 17, 2014
Content may be subject to copyright.
PLEASE SCROLL DOWN FOR ARTICLE
This article was downloaded by:
[Paredes-Solis, Sergio]
On:
27 September 2010
Access details:
Access Details: [subscription number 927315525]
Publisher
Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
Vulnerable Children and Youth Studies
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t724921266
Childhood malnutrition during the Bosnian conflict: Four linked cross-
sectional studies
Neil Andersson
a
; José Legorreta-Soberanis
a
; Sergio Paredes-Solís
a
; Lorraine Sherr
b
; Anne Cockcroft
c
a
Centro de Investigación de Enfermedades Tropicales, Universidad Autónoma de Guerrero, Acapulco,
Mexico
b
Research Department of Infection & Population Health, Royal Free and University College
Medical School, London, UK
c
CIET Trust, Johannesburg, South Africa
Online publication date: 27 September 2010
To cite this Article Andersson, Neil , Legorreta-Soberanis, José , Paredes-Solís, Sergio , Sherr, Lorraine and Cockcroft,
Anne(2010) 'Childhood malnutrition during the Bosnian conflict: Four linked cross-sectional studies', Vulnerable
Children and Youth Studies, 5: 3, 197 — 207
To link to this Article: DOI: 10.1080/17450128.2010.507801
URL: http://dx.doi.org/10.1080/17450128.2010.507801
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Vulnerable Children and Youth Studies
Vol. 5, No. 3, September 2010, 197–207
ISSN 1745-0128 print/ISSN 1745-0136 online
© 2010 Taylor & Francis
DOI: 10.1080/17450128.2010.507801
http://www.informaworld.com
RVCH1745-01281745-0136Vulnerable Children and Youth Studies, Vol. 1, No. 1, Jul 2010: pp. 0–0Vulner able Childr en and Youth S tudies
Childhood malnutrition during the Bosnian conflict: Four linked
cross-sectional studies
Vulner able Childr en and Youth S tudiesN. Andersson
Neil Andersson
a
, José Legorreta-Soberanis
a
, Sergio Paredes-Solís
a
, Lorraine Sherr
b
and Anne Cockcroft
c
*
a
Centro de Investigación de Enfermedades Tropicales, Universidad Autónoma de Guerrero,
Calle Pino, El Roble, Acapulco, Mexico;
b
Research Department of Infection & Population Health,
Royal Free and University College Medical School, London, UK;
c
CIET Trust, Johannesburg,
South Africa
(Received 8 August 2009; final version received 7 July 2010)
In the first years of the Bosnia conflict (1992–1995), a number of small local studies
failed to confirm the expected widespread malnutrition that was the basis of humanitarian
appeals. At the request of relief agencies, four population surveys from 1994 to 1997
measured childhood malnutrition during and immediately after the conflict as well as
potential risk factors. The four surveys visited a random sample of clusters from popula-
tion registers in Bosnia and Herzegovina (BiH) and the Republica Srpska (RS). All sur-
veys measured mid-upper arm circumference (MUAC) in children 6–59 months old in
31 BiH clusters and in 10 RS clusters (last three surveys). An administered questionnaire
documented potential risk factors, including breastfeeding, receipt of food aid and socio-
economic variables. Analysis relied on a cluster-adjusted multivariate Mantel–Haenszel
procedure. In BiH, the proportion of children with MUAC less than 125 mm increased
between 1994 and 1997: 5.5%, 6.8%, 14.2% and 8.6% (c
2
= 23.2; 2 d.f.); using z-scores
(>-2SD), the increase was 2.8%, 5.6%, 7.5% and 5.7% (c
2
= 11.9; 2 d.f.). In the third
year of life, the risk of malnutrition was significantly higher in children from households
receiving food aid (OR
a
=2.38, 95% CI
ca
= 1.36–4.15), whereas in the fourth year of
life the risk of malnutrition was higher among children in male-absent households
(OR
a
= 4.42, 95% CI
ca
= 1.99–9.83). The risk of malnutrition was not related to ethnic-
ity, sex of the child or urban/rural residence. The increased childhood malnutrition
between 1994 and 1997 confirms increased vulnerability of some segments of the Bosnian
population over the last years of the war, despite the humanitarian aid programme.
Keywords: malnutrition; conflict; vulnerability
Background
Armed conflict in Bosnia began in 1992 after a referendum in Bosnia and Herzegovina in
favour of independence, with Serbian military laying siege to Sarajevo. The conflict spread,
with widespread involvement of civilians and ‘ethnic cleansing’, mainly of Muslim popu-
lations. The United Nations declared Sarajevo and several other enclaves ‘safe areas’ in
1993. In 1994, NATO aircraft attacked Serb positions around Sarajevo, following a Serb
mortar shell attack on a Sarajevo market place, and Croatian and Bosniak forces took or
retook positions held by the Serbs in Bosnia. Figure 1 shows areas of control in the region
*Corresponding author. Email: acockcroft@ciet.org
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
198 N. Andersson et al.
in September 1994. At the end of 1995, the Dayton Peace Accord formally divided the coun-
try into the Republica Srpska (RS) and the Federation of Bosnia and Herzegovina (BiH). Out-
breaks of violence persisted for several years, with a further peace plan agreed in June 1999.
Before 1992, the main nutrition concerns in the former Yugoslavia were about obesity
(Zec, Telebak, Slijepcevic & Filipovi-Hadziomeragic, 1995). But the war led to serious con-
cerns about malnutrition, especially among the displaced and besieged populations. A small
study in Sarajevo found little malnutrition among children but claimed ‘fairly high levels’
among adults (WHO, 1992); another report suggested malnutrition and related infections could
be responsible for 20–30 additional deaths per day in Srebrenica (WHO, 1993b). A 1993 study
found a low body mass index in some mothers, especially among those who were refugees or
displaced, or living in the Muslim parts of Sarajevo and Zenica (WHO Nutrition Unit, 1994b).
In 1993, UNHCR claimed that adults in Central Bosnia had lost ‘an average of 10 kg over the
course of the war’ (UNHCR, 1993), leading to UN concerns that the population of central Bos-
nia was ‘on the verge of chronic energy deficiency’ (WHO, 1993a). A successful fund-raising
effort generated funds to provide full dietary replacement for an estimated three million people.
Some studies concluded that donations were well used and prevented malnutrition
(Zec, Dzumhur, Buljina & Terzic, 1994; WHO, 1993c). A number of small studies between
1992 and 1994 produced no convincing evidence of serious malnutrition. The WHO Winter
Monitoring scheme in Sarajevo found no severe adult malnutrition; those who lost weight
in winter quickly regained it (WHO, 1994b). Winter monitoring in three besieged cities
Figure 1. Areas of control in the conflict region in September 1994. VRS areas were controlled by
Serbs; HVO areas were controlled by Croatians and BiH areas by Bosnians (Muslim).
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
Vulnerable Children and Youth Studies 199
(Sarajevo, Tuzla and Zenica) found no malnutrition in children (Vespa & Watson,
1995), but more adult malnutrition in Zenica than in Sarajevo and Tuzla (WHO Nutri-
tion Unit, 1994a). A 1993 cluster study found no malnutrition in children or mothers
(Robertson et al., 1995).
There were some reports of micronutrient deficiencies. WHO reported anaemia in up
to one in four mothers (haemoglobin levels less than 11 g/dL) and one in ten children
(haemoglobin levels under 8 g/dL) (WHO, 1993d). A study in Serbian Sarajevo reported a
third or more of mothers and children were anaemic based on a clinical definition (WHO,
1994b). Micronutrient analysis found lower than normal levels of folate, Vitamin D and
beta-carotene among Sarajevo residents (Mardel et al., 1995). Another study in Sarajevo
in 1993 reported 7% of children below the fifth percentile of the normal haemoglobin
range (Cemerlic-Zecevic, Catibusic & Milicic, 1995). A cluster study in East Mostar con-
cluded that women and children had normal nutritional status, although one in three of the
mothers was anaemic (WHO, 1994a).
Thus by 1994 there was no convincing evidence of the widespread malnutrition ini-
tially feared. The relief agencies concerned with the conflict commissioned four annual
population surveys from 1994 to 1997 to evaluate the targeting and impact of the huge
food aid programme, including measuring rates of childhood malnutrition. Reports pro-
duced immediately after each survey described the main findings, including the rates of
childhood malnutrition, and formed the basis for discussions with the aid agencies and
other stakeholders. These reports were not published in a retrievable form. This article
presents for the first time in a retrievable form an analysis of the survey data about
childhood malnutrition and describes the changing levels of childhood malnutrition dur-
ing and immediately after the conflict. It also examines the actionable factors related to
malnutrition in this humanitarian emergency.
Methods
The sample
The four surveys used the same stratified, random, cluster sample. In 1994, UNHCR regional
relief teams provided their population registers as a sample frame. This comprehensive
official listing of potential food aid beneficiaries could be divided into blocks of 1000
people in nine operational areas. Stratifying by operational area, numbering the blocks and
using a random number table, we randomly drew 41 blocks (clusters) – one cluster for
every 100,000 people on the register. Because of inflation of beneficiary population figures,
in retrospect this may have been closer to one cluster per 50,000. If the local military situation
allowed it, subsequent surveys revisited exactly the same clusters.
Figure 2 shows the clusters included in each of the four surveys. The 1994 survey
included households in the selected 41 clusters in BiH. The 1995 survey revisited 37 of the
same clusters (military conditions did not permit revisiting of four clusters) and an addi-
tional 15 clusters in RS, a total of 52 clusters. In 1996, population relocation required by the
Dayton Accord emptied the five Serb-held Sarajevo clusters surveyed in 1994 and 1995;
five new clusters were selected in Eastern Bosnia. The 1996 survey covered 66 clusters,
including 30 in RS. The 1997 survey covered 55 communities, 45 in BiH and 10 in RS.
We approached the missing data resulting from these changes in the sample by limit-
ing BiH trend analysis to sites covered in all four surveys and RS trend analysis to sites
covered in all of the last three surveys.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
200 N. Andersson et al.
The surveys
The first author of this article was commissioned to undertake the four surveys, and two of
the other authors helped to conduct the surveys. For each survey, we trained four to six
field teams, each of 8–10 members, with a majority female membership. The same trainer
conducted classroom and practical training sessions over one week in each of the four sur-
veys and more than half the team members undertook all four surveys.
The surveys used mid-upper arm circumference (MUAC) in children under the age of
60 months to estimate the extent of childhood malnutrition. Training of field workers to
undertake and record MUAC measurements included careful and specific instructions to
reduce errors. After classroom practice, interviewers checked their own repeatability of
measurements in field situations. Training continued, changing subjects in a community-
based pilot, until those measuring could achieve the standard measure of the trainer every
time. The final teams included only those who showed themselves capable of measuring
and recording MUAC properly. The training required field workers to record in millime-
tres, to limit the effect of any rounding down.
During data collection in each cluster the trained field teams covered approximately 100
contiguous households, radiating from a random starting point, with no sub-sampling within the
cluster. This included all households, not only those with children aged under 60 months old.
The directly administered questionnaire in ‘Serbo-Croat’ (Bosnian) documented the house-
hold structure (age and sex of each member), sex of the head of the household, and how she or
he earned a living in the last month. The interview documented refugee or displaced persons in
the household, receipt of remittances from abroad, receipt of food aid, whether they had
enough food in the last week, stores of wheat and presence of coffee (a luxury item).
Figure 2. Chart of the clusters covered in the four surveys.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
Vulnerable Children and Youth Studies 201
Because of the security implications of questions about ethnicity in the context of eth-
nic cleansing and conscription, interviewers coded ethnicity of households by observation,
without direct questioning. For each child under 60 months of age, the interviewers asked
the mother about breast feeding and measured MUAC. In households with more than one
child under 60 months old, the interviewers attempted to collect information and measure
MUAC for all the children in this age group. In practice, few households had more than
one child aged under 60 months.
Statistical analysis
Data entry relied on Epi-Info; double data entry and verification of discordant records
minimised key-stroke errors. Further cleaning looked for logical inconsistencies and out
of range responses, with checking back against the original paper records as necessary.
Analysis relied on CIETmap, open source freeware that combines epidemiological ana-
lysis with raster and vector mapping (Andersson & Mitchell, 2006).
Analysis focused on children aged 6–59 months. We categorised children as malnour-
ished if they had a MUAC of less than 125 mm, the standard in emergency situations to
identify moderate acute malnutrition. At the time of design, there was no prior belief about
the likely level of malnutrition, the degree to which this could change in the course of the
conflict, or how long the conflict might last. A baseline frequency of 10% malnourished
would have to increase by 40% to be detectable among 1000 children in 50 clusters at a
95% confidence level with 80% power.
Although this was not available at the time to potential users of the survey results, we
verified the MUAC estimates using z-scores and a cut-off of –2 SD as the indicator of
moderate acute malnutrition, allowing for a more precise classification across age and sex.
We used the WHO freeware (WHO Anthro Version 3) in combination with the April 2009
WHO growth curves.
We tested the significance of changes in malnutrition rates over time using tests for
trend. We adjusted for clustering using the method devised by Gilles Lamothe; this applies
a published variance estimator to weight the Mantel–Haenszel OR for cluster-correlated
data (Bieler & Williams, 1995; Williams, 2000). Bivariate analysis examined the strength
and statistical significance of associations between individual variables and childhood
malnutrition. Multivariate analysis of the simultaneous effects of different variables on
the risk of malnutrition relied on simultaneous stratification using the Mantel–Haenszel
procedure (Mantel & Haenszel, 1959). The multivariate analyses began with saturated
models and stepped down to the point where all the remaining variables in the model were
significantly related to the outcome. In the initial models we included variables about the
child (age, sex, ever breastfed, measles vaccination), variables about the household
(female-headed household, displaced or refugee, ethnicity, employment of head, crowd-
ing, receipt of remittances from abroad, receipt of food aid, sufficient food in last week,
stores of wheat, coffee in the house). We express the magnitude and statistical signifi-
cance of associations as an unadjusted Mantel–Haenszel OR from bivariate analysis, and
adjusted OR (OR
a
) from multivariate analysis. The 95% CI around the OR and OR
a
are
those of Cornfield, adjusted for clustering (CI
ca
).
In 1997, UNHCR introduced priority categories for targeting food aid, based on
income and several vulnerability criteria. Households in priority categories 1 and 2 were
supposed to receive food aid ahead of those in category 3. We compared the level of
childhood malnutrition among children from households in the different priority
categories.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
202 N. Andersson et al.
Results
The population of children and MUAC measurements
Table 1 summarizes the numbers of children aged 6–59 months of age in BiH and RS with
MUAC measurements over the four surveys, a total of 4450 children. Using the z-score
and cut-off of –2 SD as the reference test, MUAC with a cut-off of 125 mm had 100%
sensitivity, correctly detecting 203/203 cases of moderate acute malnutrition identified by
z-scores, and 97% specificity, correctly identifying 3658/3772 non-cases.
Changes in malnutrition during the conflict
In children in clusters visited in all four surveys, there was a steady and significant
increase in malnutrition in BiH between 1994 and 1996, with a levelling off or fall in the
proportion malnourished between 1996 and 1997 (Table 2). The main increase in mal-
nutrition was in children aged 6–35 months. In RS there was no corresponding increase in
malnutrition between 1995 and 1996 (Table 2).
Risk factors for malnutrition
Table 3 summarizes bivariate associations between potential risk factors and malnutrition
among all children aged 6–59 months over the four surveys. None of the potential risk factors
we examined was significantly associated with malnutrition among all the children aged
6–59 months. In particular, boys were as likely as girls to be malnourished, Muslim children
were no more likely to be malnourished than non-Muslim children, and there was no overall
difference in the risk of malnutrition in children in BiH compared with children in RS.
Given that risk factors for malnutrition are likely to differ by age of the child, separate
multivariate analyses examined factors related to malnutrition in children of different ages.
In children aged 6–11 months, those aged 12–23 months and those aged 48–59 months,
none of the risk factors we examined was related to malnutrition. In children aged 24–35
months, the risk of malnutrition was significantly higher from households receiving food
aid, taking into account other variables (final model OR
a
= 2.38, 95% CI
ca
= 1.36–4.15).
Children aged 36–47 months were significantly more likely to be malnourished in female-
headed households (final model OR
a
= 4.42, 95% CI
ca
= 1.99–9.83).
Malnutrition and official priority categories in 1997
In 1997, childhood malnutrition was not higher among the higher priority households for
the food aid programme than among the lower priority households: 7.4% (29/392) of chil-
dren in Priority 1 had MUAC <125 mm, 7.6% (21/277) in Priority 2 and 7.7% (21/272) in
Table 1. Numbers of children in the surveys aged 6–59 months with MUAC measurements.
1994
a
1995 1996 1997 Total
Bosnia and Herzegovina (BiH)
All children 1065 663 809 891 3428
Children in sites visited in all four surveys 609 436 506 499 2050
Republica Srpska (RS)
All children – 223 598 201 1022
Children in sites visited in all three surveys – 70 136 105 311
a
In 1994, the survey did not include any sites in RS.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
Vulnerable Children and Youth Studies 203
Table 2. Proportions of children aged 6–59 months malnourished (MUAC <125 mm) over time in BiH and RS, restricted to children in sites visited in all surveys.
Area and age group
1994
a
1995 1996 1997
c
2
for trend
1994–1996
b
Fraction (%) with MUAC <125 mm
Bosnia and Herzegovina (BiH)
All children 6–59 months 30/545 (5.5) 28/412 (6.8) 66/464 (14.2) 38/442 (8.6) 23.17 p < 0.005
Children 6–11 months 4/40 (10.0) 9/29 (31.0) 13/45 (28.9) 7/21 (33.3) 4.10 p > 0.10
Children 12–23 months 12/84 (14.3) 7/73 (9.6) 30/78 (38.5) 16/66 (24.2) 13.82 p < 0.005
Children 24–35 months 9/117 (7.7) 7/94 (7.4) 17/101 (16.8) 13/104 (12.5) 4.56 p > 0.10
Children 36–47 months 5/150 (3.3) 3/113 (2.7) 2/126 (1.6) 1/121 (0.8) 0.82 NS
Children 48–59 months 0/154 (0) 2/103 (1.9) 4/114 (3.5) 1/130 (0.8) 5.13 p < 0.10
Republica Srpska (RS)
All children 6–59 months (numbers too
small for age-specific breakout)
3/55 (5.7) 9/117 (7.7) 7/81 (8.6) 0.45 1 d.f. NS
Fraction (%) with z-score >–2 SD c
2
for trend 1994–1996
BiH children aged 6–59 months 15/544 (2.8) 23/409 (5.6) 35/464 (7.5) 25/442 (5.7) 11.86 2 d.f. p < 0.05
a
In 1994, the survey did not include any sites in RS.
b
Tests for trend 1994–1996 with 2 d.f.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
204 N. Andersson et al.
Priority 3. Households who did not receive food aid in 1997 (about two-thirds of the
households) similarly showed no trend for more malnutrition in lower priority categories;
among these households 6.5% (12/184) of children were malnourished in Priority 1, 6.6%
(13/197) in Priority 2 and 6.8% (15/221) in Priority 3.
Discussion
Between 1992 and 1994, a plethora of small studies conducted or sponsored by aid agencies
produced sometimes conflicting results but no clear evidence of malnutrition in adults or
children. The larger representative surveys reported here show nutritional status in the
1994 survey was well within expected ranges. The steady and significant increase in the
proportion of children with moderate acute malnutrition in the same clusters and using the
same anthropometric parameter is convincing evidence of a real increase in childhood
malnutrition in BiH between 1994 and 1996. This pattern of increased malnutrition was
not apparent between 1995 and 1996 in the smaller sample of children in RS.
Food aid was associated with malnutrition in the third year of life – children from
households that received food aid were more likely to malnourished. This could reflect
targeting of food aid to the most vulnerable households where children were more likely
to be malnourished, but the food aid nevertheless failing to prevent the malnutrition. The
finding also raises the question of how children remained malnourished in households that
received more food every month. It is conceivable that food aid contributed in some way
to childhood malnutrition in children aged 24–35 months. Food aid contributed by most
countries, including the United States, until 1999 did not consider micro-nutrients
(Marchione, 2000). Yamano and colleagues found that food aid was effective in
improving malnutrition up to the age of 24 months, but not thereafter (Yamano,
Alderman, & Christiaensen, 2005). Following this logic, it is not so much that the food
aid caused the measured malnutrition in Bosnian 24–35 month olds so much as that it
failed to prevent it in a hungry population. Yamano et al. (2005) also found that children
in the second year of life grew faster when receiving food aid than those who did not; this
fits with several other studies that found children aged 12–24 months were especially
vulnerable to income shocks (Hoddinott & Kinsey, 2001; Martorell, 1999).
A FAO food nutritional assessment in The Gambia identified another but possibly related
mechanism: ‘The fact that a household has enough food does not ensure good nutritional
Table 3. Bivariate associations with malnutrition (MUAC < 125 mm) in children aged 6–59
months in all surveys.
Variable
Fraction (%) with MUAC <125 mm
OR (95% CI
ca
)
cluster adjusted
Children without
factor
Children with
factor
Sex (female) 120/1767 144/1739 1.24 (0.95–1.62)
Ethnicity (Muslim) 120/1658 142/1845 1.07 (0.75–1.51)
Never breastfed 239/3190 17/257 0.87 (0.52–1.47)
Rural residence 213/2734 51/773 0.84 (0.60–1.17)
Household (displaced or refugee) 174/2280 90/1225 0.96 (0.70–1.30)
Female-headed household 243/3242 21/265 1.06 (0.59–1.91)
Household insufficient food
in last week
219/2762 44/734 0.74 (0.48–1.15)
Household receipt of food aid 77/1217 187/2285 1.32 (0.94–1.84)
Living in RS rather than BiH 205/2688 59/819 0.94 (0.61–1.44)
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
Vulnerable Children and Youth Studies 205
status of the children. Illnesses and inappropriate diets are also associated with acute
malnutrition’ (FAO, 2002). Awokuse (2006) commented similarly on several studies of
the impact of food aid on nutritional status: ‘Other factors may contribute to sub-optimal
caloric intake and increased prevalence of malnutrition. These factors include poor treatments
for infectious diseases, nutritional imbalances in local diets and various social and cultural
conditions that give priority to adult males rather than mothers and children.’
In the fourth year of life, children living in male-absent (female-headed) households
were significantly more likely to be malnourished. Theoretically, female-headed house-
holds – more correctly, male-absent households – might be expected to be more vulnerable
because they lack economically active male purchasing or representational power. The
higher risk of malnutrition among older children in female-headed households may reflect
the reduced ability of mothers to care for toddlers while trying to eke out a survival for the
household in emergency and immediately post-emergency conditions. In a review of 61
studies examining the relationship of female headship to poverty, 38 found female-headed
households over-represented in the poor (Buvinic & Gupta, 1997). About one-half of the
studies considering childhood malnutrition reported negative consequences of female
headship. The authors hypothesised that women may spend a higher proportion of their
available income on food. Kennedy and colleagues found that, although female-headed
households are typically poorer, they have less childhood malnutrition (Kennedy & Peters,
1992). Again, the hypothesis was that women spend a higher proportion of income on
food. Rogers noted that, although children in male-absent households consumed fewer
calories, their growth was significantly better than that of children in male-headed house-
holds (Rogers, 1996). In Bosnia it is possible that, even if female-headed households had
discretionary income and the evidence points to them having considerably less than male-
headed households, for most of the conflict there simply was not enough food to purchase.
In 1997 the official vulnerability categories (the UNHCR priority categories for targeting
of food aid) were not associated with malnutrition. One possible explanation is that the
targeting worked, and that the assistance strategy had reduced malnutrition selectively in
children from Priority 1 households. Against this explanation is the finding of no gradient
of rates of childhood malnutrition across priority categories among those households that
did not actually receive food aid (about two-thirds did not receive food aid in 1997).
Limitations
Although the sample of clusters was diligently drawn from official lists, it later turned out
that these lists inflated the beneficiary population. Even if the sample had at one point
been representative of the population, dramatic population movements meant this was
unlikely to be the case for all four surveys. In the face of this uncertainty, following the
same clusters over the four years still offers an interpretable picture of changes over time.
However, restricting the analysis to those clusters that could be visited in all four surveys
meant we excluded clusters that could not be visited in some surveys because of active
conflict in the area; the restricted analysis probably underestimated the increase in childhood
malnutrition and other health consequences of the conflict.
MUAC is a robust indicator of malnutrition in the analysis of risk factors if measurement
errors are more or less constant between subgroups. There is no reason to believe that
MUAC measurement errors might be different between, for example, households that
reported sufficient food in the last week and those that reported insufficient food. In deter-
mining the actual level of moderate acute malnutrition, there is an argument that z-scores give
a more accurate picture than the MUAC cut-off (WHO, 2000). During the humanitarian
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
206 N. Andersson et al.
emergency in Bosnia, relief managers had access to malnutrition rates as assessed using
the MUAC cut-off, not z-scores. As it turned out, the MUAC cut-off had a high sensitivity
and specificity in relation to z-scores (>–2 SD) calculated with the latest WHO growth
curves.
Conclusions
MUAC remains a robust measure of malnutrition in young children in humanitarian emer-
gencies. The increase of childhood malnutrition in BiH between 1994 and 1996 confirms
increased vulnerability in some segments of the Bosnian population in the final years of
the war.
Acknowledgements
The 1994, 1995 and 1996 surveys were commissioned in three separate contracts by the World Food
Programme (WFP) and United Nations High Commissioner for Refugees (UNHCR), who also pro-
vided logistical support in Sarajevo, Tuzla, Banja Luka, Bihac and Medugorje. This support made
the fieldwork possible in the difficult conditions of Bosnia during the conflict. The 1997 survey was
commissioned by the World Bank.
Kate Galt, Indira Kulenovic, Nicole Massoud, Sharmila Mhatre, Lorenzo Monasta, Manuel Pascual
Salcedo, Aparna Swaminathan and Charles Whitaker supported the data collection in one or more of
the four surveys. Mary Cameron and Jeff Jarabek helped with retrieval of the literature. Gilles
Lamothe developed a variance estimator that weights the odds ratio for cluster-correlated data using
the Mantel–Haenszel procedure, and Mario Beauchamp developed an open source application that
made this available for the analysis.
References
Andersson, N., & Mitchell, S. (2006). Epidemiological geomatics in evaluation of mine risk education
in Afghanistan: Introducing population weighted raster maps. International Journal of Health
Geographics, 5, 1.
Awokuse, T. (2006). Assessing the impact of food aid on recipient countries: A survey. The Food
and Agriculture Organization of the UN. Retrieved from ftp://ftp.fao.org/docrep/fao/009/
ag431t/ag431t00.pdf.
Bieler, G.S., & Williams, R.L. (1995). Cluster sampling techniques in quantal response teratology
and development toxicity studies. Biometrics, 51(2), 764–776.
Buvinic, M., & Gupta, G.R. (1997). Female-headed households and female-maintained families:
Are they worth targeting to reduce poverty in developing countries? Economic Development
and Cultural Change, 45(2), 259–280.
Cemerlic-Zecevic, E., Catibusic, F., & Milicic, D. (1995). Nutritional influences on growth and
prevalence of anaemia in children during the war in Sarajevo. European Journal of Clinical
Nutrition, 49(Suppl. 2), S43–S45.
FAO. (2002). Special Report FAO/WFP Crop and Food Supply Assessment Mission to the Gambia.
Retrieved from http://www.fao.org/docrep/005/y8162e/y8162e00.HTM.
Hoddinott, J., & Kinsey, B. (2001). Child growth in times of drought. Oxford Bulletin of Economics
and Statistics, 63, 409–436.
Kennedy, E., & Peters, P. (1992). Household food security and child nutrition: The interaction of
income and gender of the household head. World Development, 20(8), 1077–1087.
Mantel, N., & Haenszel, W. (1959). Statistical aspects of the analysis of data from retrospective
studies of disease. Journal of the National Cancer Institute, 22, 719–748.
Marchione, T. (2000). Title II food aid and the nutrition of children in Latin America and the
Caribbean. Food and Nutrition Bulletin, 21(1), 65–70.
Mardel, S., Hailey, P., & Mahmutovic, A. (1994). Anaemia survey in Sarajevo. In: Summaries of a
nutrition seminar in Sarajevo: The impact of war on nutrition in Sarajevo (p. 12). Sarajevo:
Institute of Public Health and WHO.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010
Vulnerable Children and Youth Studies 207
Mardel, S., Hailey, P.C., Mahmutovic, A., Broom, J., Golden, M.H., Franklin, M., & Mills, C.F.
(1995). Micronutrient status of the besieged residents of Sarajevo: May 1993. European Journal
of Clinical Nutrition, 49(Suppl. 2), S46–S61.
Robertson, A., Fronczak, N., Jaganjac, N., Hailey, P., Copeland, P., & Duprat, M. (1995). Nutrition
and immunization survey of Bosnian women and children during 1993. International Journal of
Epidemiology, 24, 1163–1170.
Rogers, B.L. (1996). The implications of female household headship for food consumption and
nutritional status in the Dominican Republic. World Development, 24(1), 113–128.
UNHCR. (1993). Information notes on former Yugoslavia (Report No. 10/93). Washington, DC:
UNHCR.
Vespa, J., & Watson, F. (1995). Who is nutritionally vulnerable in Bosnia-Hercegovina? BMJ 311,
652–654.
WHO Anthro. (2010). Version 3.1 Retrieved from http://www.who.int/childgrowth/software/en/.
WHO Nutrition Unit. (1994a). In search of the most vulnerable. WHO Winter Nutrition Monitoring
Summary Report (December 1993–February 1994). Zagreb: WHO.
WHO Nutrition Unit. (1994b). Nutrition and health survey in Serbian areas of Sarajevo (February
1994). Zagreb: WHO.
WHO. (1992). Nutrition monitoring in former Yugoslavia (December 1992 report). Zagreb: WHO.
WHO. (1993a). 12th report from the office of the WHO special representative (December 6–January 2,
1993). Zagreb: WHO.
WHO. (1993b). 14th report from the office of the WHO special representative. (March 1993).
Zagreb: WHO.
WHO. (1993c). 18th report from the office of the WHO special representative (July/August 1993–1
September 1993). Zagreb: WHO.
WHO. (1993d). Bihac, Sarajevo, Zenica and Tuzla surveys (Report of WHO Nutrition Unit, Zagreb,
24 September 1993). Zagreb: WHO.
WHO. (1994a). Nutrition and health survey in Mostar East (April 3–11, 1994). Zagreb: WHO.
WHO. (1994b). WHO winter nutrition monitoring: Results from the fourth round of data collection.
Zagreb: WHO.
WHO. (2000). The management of nutrition in major emergencies. Geneva: WHO.
Williams, R.L. (2000). A note on robust variance estimation for cluster-correlated data. Biometrics,
56, 645–646.
Zec, S., Dzumhur, Z., Buljina, A., & Terzic, R. (1995). Therapeutic feeding in Sarejevo during the
war. European Journal of Clinical Nutrition, 49(Suppl. 2), S40–S42.
Zec, S., Telebak, B., Slijepcevic, O., & Filipoviv-Hadziomeragic, A. (1995). Nutrition in pre-war
Sarajevo. European Journal of Clinical Nutrition, 49
(Suppl. 2), S6–S10.
Downloaded By: [Paredes-Solis, Sergio] At: 17:34 27 September 2010