civil war and child health: regional and ethnic dimensions of child immunization and malnutrition in...

Social Science & Medicine 56 (2003) 2515–2527

Civil war and child health: regional and ethnic dimensions ofchild immunization and malnutrition in Angola

Victor Agadjaniana,*, Ndola Pratab

aDepartment of Sociology, Arizona State University, Tempe, AZ 85387-2101, USAbBay Area International Group (BIG), Institute of Human Development and School of Public Health, University of California,

1213 Tolman Hall, Berkeley, CA 94720, USA

Abstract

This study arises from a general proposition that different levels and types of exposure to war are crucial in shaping

health outcomes in a population under war-induced duress. We analyze civil war-related regional and ethnolinguistic

differentials in age-adequate immunization (complete vaccination for age) and levels of malnutrition in Angola. Our

analysis is based on data from a nationally representative survey conducted in 1996, some 2 years after the end of one of

the most destructive periods of hostilities in the history of Angolan civil war. The data show that despite Angola’s

unique mineral wealth, the nation’s levels of child age-adequate immunization is lower and malnutrition rates are higher

than in most of sub-Saharan Africa. To examine age-adequate immunization and chronic malnutrition we fit logistic

regression models that include the regional degree of war impact and ethnolinguistic group, in addition to rural–urban

differences and other conventional sociodemographic characteristics. The tests reveal a significant disadvantage of rural

children relative to urban children in both immunization and chronic malnutrition. Net of the rural–urban differences,

we also detect a significant disadvantage of children residing in parts of the country that had been most affected by the

fighting. The tests also point to a lower level of immunization and higher level of chronic malnutrition among children

from the ethnolinguistic group commonly identified with the opposition. These associations tend to be stronger among

children who were born and/or grew up during war than among children who were born after peace was re-established.

r 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Angola; War; Child health; Malnutrition; Immunization; Sub-Saharan Africa

Introduction and conceptualization

Children have been among the main victims of civil

wars in sub-Saharan Africa. Despite a common percep-

tion that military conflicts negatively affect child health,

and survival, the complexity of these effects and of their

longer-term consequences is poorly understood and

rarely studied, largely because adequate data are often

lacking. Whenever systematic information does become

available, it generally paints pictures of disaster (Cliff &

Noormahomed, 1988; Garenne, Coninx, & Dupuy,

1996; Gessner, 1994; Gubhaju, 1997; Levy & Sidel,

1996). The available literature also indicates that health

and nutritional insecurity among children is a common

consequence of military conflicts (for example, African-

European Institute, 1990; AlDoori, Armijo-Hussein,

Fawzi, & Herrera, 1994; Brentlinger, Hern!an, Hernan-

dez-D!ıaz, Azaraff, & McCall, 1999; Cutts et al., 1996;

Garenne, 1997; Kinfu, 1999). However, it is important

to remember that in sub-Saharan Africa, as in other

parts of the developing world, child malnutrition and

health problems caused by poverty and economic crises

are pervasive even under peaceful conditions (Boerma,

Sommerfelt, Rutstein, & Guillermo, 1990; Martin-Pr!evel

et al., 2000). In addition to a direct impact of war on

child health through malnourishment, war affects it

indirectly by undermining a country’s public health

system and specifically its immunization programs. This

*Corresponding author. Tel.: +1-480-965-3804; fax: +1-

480-965-0064.

E-mail address: [email protected] (V. Agadjanian).

0277-9536/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.

PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 2 8 6 - 1

consequence of war is particularly deleterious in poorer

countries where comprehensive immunization coverage

can lead to dramatic improvements in child health and

survival (Amin, 1996; Desgr!ees du Lo #u, 1996; Garenne,

Coninx, & Dupuy, 1996; Samb, Aaby, Whittle, Seck, &

Simomdon, 1997).

In this study we contribute to an understanding of

these matters by using the case of Angola, a sub-

Saharan country with a population of 13 million that

has been in and out of civil war for almost three decades.

We examine Angolan children’s immunization and

malnutrition levels by focusing on their regional and

ethnic dimensions and attempting to relate these

dimensions to the effects of civil war.

Angola, mired in military conflicts since before its

independence from Portugal in 1975, offers a glaring

example of the devastation caused by war. The warfare

that dragged on for the first decade and a half of

Angola’s independent history was brought to a halt by

the signing of a ceasefire between the ruling MPLA (the

Portuguese acronym for the Popular Movement for the

Liberation of Angola) and its long-term rival UNITA

(the National Union for the Total Independence of

Angola) in 1991 only to resume a year later, after

UNITA refused to accept the results of the general

elections held in September 1992. The stretch of

hostilities that followed was particularly brutal and

generalized, engulfing a large portion of the Angola’s

countryside as well as some urban areas. An accord

signed in Lusaka, Zambia, in November 1994 put an end

to the fighting and started the process of disarmament

and political integration. However, this process failed,

and new fighting erupted in the end of 1998 (Spears,

1999). It took almost three and a half years of fierce

combats and the sudden death of UNITA’s charismatic

leader, for a cease-fire agreement between the warring

parties to be signed in April of 2002 (The New York

Times, 2002). At the time of this writing, the chances for

a durable pacification in Angola are looking increasingly

brighter.

The decades of civil war have caused massive loss of

lives and destruction, reducing the Gross National

Income per capita of this oil-rich country to a mere

USD 290 by the turn of the century (World Bank, 2002).

Hundreds of thousands of people have been killed in

fighting or died from war-related diseases and starva-

tion, and according to some estimates, over 3 millions

persons have been displaced just since the 1998

resumption of hostilities (Chelala, 1999; MSF, 2000;

UN Security Council, 2001). Not surprisingly, children

have been the primary victims of the war; suffice it to say

that a few years ago UNICEF gave Angola a grim

distinction of the worst country in the world to be a

child (UNICEF, 1999).

Although war has been part of Angolan reality for

decades, it has not affected the country evenly. Some

parts of the country—both rural and urban—have

suffered direct and generalized devastation, especially

during the 1992–1994 fighting, whereas other parts have

been largely spared direct fighting and have been

affected indirectly (Anstee, 1996; Brittain 1998). The

effects of war on child health may differ greatly between

directly and indirectly affected areas, and we chose this

regional dimension of the war impact as one of the two

main foci of our analysis. Understanding war-related

regional variations in child immunization and nutrition

may also help design better targeted and tailored post-

war policy interventions.

Another important peculiarity of Angolan war is its

ethnic dimension. As in many other similar settings,

ethnic divisions, loyalties, and rivalries are often

mentioned in accounts and analyses of civil conflict in

Angola and can be traced back to colonial times

(Heywood, 2000; Malaguias, 2000). In the post-colonial

era, the Kimbundu-speaking population, concentrated

in and around the area of the capital city Luanda as well

as inland areas to the east, has been traditionally seen as

the ethnic mainstay of the ruling MPLA, whereas

Umbundu-speakers (also known as the Ovimbundu),

Angola’s largest ethnolinguistic group spread over the

middle part of Angola, have been assumed to support

the UNITA opposition. Although the Angolan conflict

cannot be reduced to ethnic strife, the ethnic factor in it

nevertheless has been significant; ideologies and external

political alliances of the MPLA and UNITA have

changed over time but their ethnic identifiers, however

vehemently denied by the warring parties themselves,

have endured. By focusing on ethnic differentials in

immunization and malnutrition, our study shifts the

focus of analysis from the ethnic causes of war to a no

less important—but a much less studied—issue of war’s

ethnically differential consequences. Again, we believe

that this focus will not only expand our understanding

of how war affects society, but also may usefully inform

national strategies and priorities in dealing with the

public-health legacy of war. Importantly, the ethnic map

of Angola and that of the strength and pattern of the

war impact (greater and direct vs. lesser and indirect) are

not strongly correlated, allowing for an analysis of the

independent effects of each of the two factors.

Our conceptual model rests on a general proposition

that different levels and types of exposure to war are

crucial in shaping health outcomes in a population

under war-induced duress. The two outcomes of

interest—immunization and malnutrition levels—are

known to inversely correlate (Sommerfelt & Stewart,

1994). At the same time, they represent two rather

different dimensions of child health. Malnutrition

results from a large number of factors, the most

common being inadequate food intake and the incidence

and severity of infectious diseases. Hence, it is only

indirectly related to government policies and actions.

V. Agadjanian, N. Prata / Social Science & Medicine 56 (2003) 2515–25272516

Immunization, in contrast, is a direct result of a

government-run public health system.

Our hypotheses are as follows. We expect that child

immunization levels are lower and malnutrition levels

are higher in the segments of the population—either

regionally or ethnically defined—that have suffered most

from war. We also hypothesize that regional and ethnic

differences associated with war are more pronounced

among children who were born and/or grew up during

hostilities, compared to those who were born after

hostilities stopped. Although the disadvantage of rural

areas relative to urban areas is generally to be expected

in both immunization and nutrition, rural–urban

differences should also be stronger among children born

during war than among children born during peace,

because rural areas, regardless of their regional location,

suffer from war disproportionately.

Data and methods

We use data from the Multiple Indicators Cluster

Survey (MICS) conducted in Angola in the last trimester

of 1996, at the height of the peaceful period—2 years

after the Lusaka peace protocol was signed and 2 years

before the next round of hostilities started. MICS was a

nationally representative survey of 4440 households. Its

questionnaire focused on household composition, edu-

cation, mortality, fertility, health, anthropometry, im-

munization, and sanitation (see INE, 1998 for details of

the survey design and sampling methodology). The

MICS sampling frame excluded camps set up for

displaced people during the 1992–1994 fighting. How-

ever, because the survey took place 2 years after the

cessation of hostilities, when many displaced families

had already left the camps, this omission should not

create any major bias. In sum, the MICS provides rich

and unique information on child health and nutritional

status in the aftermath of a major outbreak of civil war.

Information on immunization recorded in MICS

comes from both the vaccination registration card and

mother’s recall. The survey recorded compliance sepa-

rately for each immunization included in the Angolan

vaccination schedule. For our test of immunization

status we construct a variable that we define as age-

adequate immunization. This variable takes into ac-

count the vaccination schedule used in Angola (BCG

and POLIO 0 at birth; POLIO 1 and DTP 1 at 2 months;

POLIO 2 and DTP 2 at 4 months; POLIO 3 and DTP 3

at 6 months; and measles at 9 months) and is formulated

as a dichotomy—children who had all the necessary

vaccines for their age vs. those who missed one or more

of the required vaccines. The analysis of immunization

includes all children aged zero to 59 months.

Following the standard procedure, we define chil-

dren’s nutritional status on the basis of anthropome-

trical measures. MICS recorded age, sex, weight, and

height of all children aged 6 to 59 months in the sampled

households. From these data we created two indices

expressed as z-scores. Weight-for-height and height-for-

age z-scores are the numbers of standard deviations

(SD) from the median International Reference Popula-

tion (WHO/NCHS) (for details on this procedure see,

US Food and Nutrition Board, 1974; WHO, 1983). The

z-score cut-off point of –2SD was used to classify

children as wasted (inadequate weight-for-height, or

acute malnutrition) or stunted (inadequate height-for-

age, or chronic malnutrition) in the respective distribu-

tions. Wasting and stunting reflect the immediate and

cumulative effects, respectively, of poor health and

nutritional deficiency. Our multivariate analyses focus

on these two forms of malnutrition. Our dependent

variables are dichotomous—whether or not a child is

wasted and whether or not a child is stunted.

Our three main independent variables are area of

residence (rural vs. urban), degree of war impact in the

region of residence (region of heavier impact vs. region

of lighter impact), and ethnolinguistic group. Based on a

perusal of contemporary media reports and more

systematic accounts of Angolan war, especially its

1992–1994 outbreak (for example, Anstee, 1996; Brit-

tain, 1998), we subdivide Angola into two parts—one

with a more direct and heavy impact of war, and the

other with a more indirect and relatively mild impact.

We define these two parts using the regional classifica-

tion employed in MICS. In the part with a heavier war

impact, we include the North, East, and Centre-South

regions (the provinces of Uige, Zaire, Malange, Lunda

Norte, Lunda Sul, Moxico, Huambo, Bie, and Kuando-

Kubango). That part of Angola was characterized, in

general, by fierce fighting and bombardments, especially

in the aftermath of the 1992 elections, a vast destruction

of the existing infrastructure, a paralysis of the economy

and basic services such as electricity, running water, and

sewer, and a near complete collapse of the health and

educational systems. The part of the country with a

relatively light impact of war comprises Capital, West

and South regions (the provinces of Luanda, Bengo,

Cabinda, Kwanza Norte, Kwanza Sul, Benguela, Huila,

Namibe, and Cunene). These areas were largely con-

trolled by government forces and saw only sporadic and

isolated clashes. The effect of war there was largely

indirect, manifested in economic stagnation and fre-

quent and prolonged interruptions of basic services.

As can be seen in Fig. 1, each group of regions

constitutes a contiguous area; lighter-impact regions are

in the western part of the country, reflecting the

government’s greater control in the coastal and adjacent

areas, whereas heavier-impact regions encompass the

parts of the Angolan hinterland, where the opposition

has been traditionally strong. Although this simple

classification may not reflect all the regional and local

V. Agadjanian, N. Prata / Social Science & Medicine 56 (2003) 2515–2527 2517

nuances of the war impact, we nevertheless believe that

it is sufficient to highlight the most important differ-

ences.

Fig. 1 also presents a rough ethnolinguistic map of

Angola. In our study, ethnolinguistic group is operatio-

nalized as the language spoken in the household as

reported in the survey by the household’s head. We

define five ethnolinguistic groups: Umbundu, Kimbun-

du, Kikongo, Portuguese, and a residual group of other

languages. Although Portuguese-speakers do not repre-

sent a separate ethnicity and may be of different ethnic

and racial roots, they are a culturally distinct group

reflecting the centuries-long political, cultural, and

demographic presence of the Portuguese in Angola. As

to other indigenous ethnolinguistic groups, although the

ethnographic literature points to considerable differ-

ences among them not only in language but also in

various aspects of traditional economies, material and

spiritual cultures, and social organization, it does not

identify any systematic long-standing ethnic inequalities

in nutrition and health status (for example, Redinha,

1975). In the following text we refer to ethnolinguistic

group as ethnicity for brevity.

Although with our cross-sectional data we cannot

follow trends in immunization and malnutrition over

different stages of the Angolan politico-military situa-

tion, we can approximate the effects of war and peace by

stratifying the MICS sample of children into two groups

according to the time of birth—those who were born

during war (or those who were born before war restarted

in 1992 but grew up after that), and those who were born

during the ensuing period of peace. For the analysis of

immunization status the peace cohort includes all

children born after the end of hostilities in the autumn

of 1994 (age 24 months or younger). However, because

of potential environmental influences on early intrau-

terine growth (for example, through maternal nutri-

tional status), for malnutrition analyses the peace birth

cohort includes children who were conceived after the

war ended (age 18 months or younger). The size of the

resulting peace cohort is relatively small yet, still

sufficient for meaningful statistical tests.

For both immunization and malnutrition the results

of the inter-cohort comparisons should be interpreted

with caution because it is practically impossible to

separate the war-cohort effect from that of age alone.

The goal of our analyses, however, is not so much to

argue that the peace cohort had better health than did

the war cohort, but to try to show, even if indirectly,

that the magnitudes of regional and ethnic differences in

Fig. 1. Main ethnolinguistic groups and regions of war impact in Angola.


immunization and malnutrition changed (decreased)

between the period of hostilities and that of relative

peace.

We start with the analysis of immunization and then

move on to acute and chronic malnutrition. For each of

the three outcomes we first present bivariate associations

and then employ multivariate logistic regression to

analyze the effects of area of residence, region of war

impact, and ethnolinguistic group. For each outcome we

first fit a baseline model that includes conventional

socioeconomic and demographic predictors: urban vs.

rural residence; child’s age (as a set of dummy variables);

sex of household’s head; the average educational level of

household’s adult members (the MICS data do not

allow us to link children’s characteristics and their

mothers’ educational levels), which is broken down into

a trichotomy—none, 1 to 4 year, 5 or more years;

household economic status, approximated by radio

ownership (whether or not the household owns a radio

set); and household sanitary status, approximated by

whether or not water is regularly treated or boiled before

consumption. For the nutritional models we also control

for age-adequate immunization status, as a proxy for the

risk of infections that may affect nutritional conditions.

We then add region of war impact and, finally,

ethnicity, to the baseline model for each of the three

outcomes. In these comprehensive models, Umbundu-

speakers (the Ovimbundu), who are commonly asso-

ciated with the opposition, are the reference category.

For each model, we conduct separate tests for the entire

sample and for each of the two birth cohorts. To control

for possible age variations within each cohort, the

cohort-specific models include age in months as a

continuous variable (preliminary tests showed that

because of each birth cohort’s short age range, within

each cohort the association of age with the outcomes of

interest is more or less linear). To save space and time,

we only present and discuss the odds ratios for the

comprehensive models, acknowledging whether the

addition of region of war impact and ethnicity alters

the effects of predictors in the baseline models.

We use the STATA statistical package for logistic

regression analyses. Due to clustering of observations in

surveys like MICS, the standard errors produced by

commonly employed multivariate tests tend to be

artificially small. To account for the effect of clustering

we obtain robust estimates of standard errors (STATA,

2001).

Results

Age-adequate immunization

As can be seen in Table 1, Angola as a whole has a

dismally low level of age-adequate immunization—only

one in ten of the surveyed children had all the vaccines

required for age. However, the MICS data also show

considerable variations in immunization. Thus immuni-

zation levels reveal a U-shape association with age. The

most striking imbalance, however, is between urban and

rural areas—the odds of age-adequate immunization are

more than three times as high in the former than in the

latter. Although in almost all developing countries

urban children have higher immunization coverage than

their rural counterparts, in Angola the urban–rural

disparity is extreme even by poorest countries’ standards

(Boerma et al., 1990; Sommerfelt & Piani, 1997). The

urban–rural gap is noticeably narrower among children

born after the war than among the war-born cohort,

primarily because of the difference in immunization

rates in rural areas.

Differences in immunization levels according to the

severity of the war impact in the region of residence are

also very pronounced and in the predicted direction:

children living in heavier-impact regions are less likely to

be fully immunized for their age. Again, these differ-

ences diminish in the peace cohort thanks to the

improvement in more affected regions. Among the

ethnolinguistic groups, the Ovimbundu are the worst-

off: only six percent of children in Umbundu-speaking

households had full immunization for their age, less than

half the proportions of adequately immunized children

in Portuguese-, Kimbundu- or Kikongo-speaking fa-

milies. Surprisingly, the ‘‘other’’ group is much more

similar to Umbundu-speakers, but as with all residual

categories consisting of a diverse array of smaller

ethnolinguistic groups, it is difficult to interpret this

similarity. In all groups, with the exception of Kimbun-

du-speakers, immunization rates tend to be higher

among children of the peace cohort. Umbundu-speakers

display the largest improvement, but even so remain at

the bottom of the ethnic list.

The results of the multivariate test of immunization

are presented in Table 2. This test confirms the

considerable advantage of urban areas; urban children

have twice the odds of rural children of being fully

immunized for their age. The urban–rural differences are

particularly large for children conceived and born

during wartime: in that cohort the corresponding odds

ratio exceeds 2.5. In the peace cohort these diminish

considerably, but even so among urban children the

odds of having age-adequate immunization are nearly

twice as high as among rural children.

The analysis underscores the disadvantage of more

heavily affected parts of Angola, both urban and rural.

Overall, the odds of age-adequate immunization for

children living in more affected regions are only two-

thirds of those for children in less affected regions.

However, when we break the sample down by birth

cohort, the regional differences become even stronger

among children born during war, but are no longer


significant among those born during peace. Notably,

even after the inclusion of the war-impact region in the

model, the disadvantage of rural children remains

noticeable. Finally, the inclusion of war-impact region

significantly improves the overall fit of the model but

barely alters the effects of other factors, suggesting that

the influence of region is largely independent.

The urban–rural and regional differences are barely

affected by the inclusion of ethnicity. In line with the

bivariate pattern, children in Umbundu-speaking house-

holds, UNITAs alleged ethnic ‘‘backbone,’’ have the

smallest odds of having age-adequate immunization.

However, while the multivariate test increases the

distance between Umbundu-speakers and ‘‘others,’’

which seemed rather slight at the bivariate level, it also

nearly erases the once salient difference between

Umbundu and Portuguese-speakers. Although there is

no straightforward explanation for the latter result, a

possible clue may lie in the sociocultural (rather than

purely ethnolinguistic) nature of the Portuguese-speak-

ing group: once education and material status are

controlled for, the advantage of Portuguese-speakers

disappears. A comparison of the two birth cohorts

suggests that the Ovimbundu’s disadvantage decreased

after hostilities ended.

Our tests indicate that the effects of the three

predictors of interest on having age-adequate immuni-

zation are largely independent from those of other

sociodemographic variables. The effects of these other

variables, however, are also instructive. As is often the

case, the youngest children have the highest level of age-

adequate immunization, but the second-youngest group

is much more different from them than are older

children. Households with a radio are significantly more

likely to have their children adequately immunized than

are households without one. This also accords with the

universal pattern, although in this case a radio may not

just be an indicator of household socioeconomic status

Table 1

Descriptive statistics: age-adequate immunization and child malnutrition, Angola MICS, 1996 (percent)

Immunization Malnutrition

Variables Age-adequate immunization Acute Chronic

All

children

War

cohort

Peace

cohort

All

children

War

cohort

Peace

cohort

All

children

War

cohort

Peace

cohort

Angola total 10.0 9.4 10.7 5.8 5.2 7.6 53.9 57.3 44.9

Area of residence

Urban 17.3 18.0 16.6 4.8 4.0 6.9 47.2 48.9 43.0

Rural 5.1 3.9 6.5 6.6 6.0 8.1 58.4 62.8 46.3

Region of war impact

Heavier impact 5.6 4.0 7.2 6.5 5.9 8.0 59.7 63.9 49.6

Lighter impact 13.5 13.5 13.6 5.3 4.6 7.2 49.3 52.3 40.6

Language spoken at home

Umbundu 6.1 4.7 7.7 7.7 6.8 10.1 64.1 68.9 51.1

Kimbundu 15.6 17.1 14.2 4.0 2.5 7.7 50.5 49.9 52.1

Kikongo 13.6 12.0 15.4 5.1 5.5 3.9 48.5 52.9 35.0

Portuguese 13.7 12.4 15.0 4.7 3.8 6.8 41.5 45.7 30.6

Other languages 7.7 7.7 7.8 5.5 5.6 5.2 47.7 51.6 36.5

Age

o12 monthsa 14.9 — — 3.3 — — 36.8 — —

12–23 months 6.1 — — 10.4 — — 53.5 — —

24–35 months 9.8 — — 4.5 — — 50.0 — —

36–59 months 9.3 — — 4.4 — — 60.7 — —

Fully immunized for age

Yes — — — 3.3 2.2 6.9 36.7 40.3 23.6

No — — — 6.1 5.5 7.7 55.5 59.0 46.5

a0–11 months for immunization, 6–11 months for malnutrition.


but also a medium through which household members

are exposed to vaccination campaign messages (Som-

merfelt & Piani, 1997). Households that regularly treat

drinking water display higher odds of having their

children adequately immunized. Child’s sex shows no

significant effect on the odds of immunization—a result

that conforms to the cross-national evidence (Sommer-

felt & Piani, 1997). Also in line with the experience of

other countries, education of household adults, espe-

cially beyond the primary level, tends to increase the

likelihood of children having all the necessary vaccines.

Finally, living in female-headed households proves more

conducive to age-adequate immunization than living in

male-headed ones, which suggests that in female-headed

families, the absence of men, while increasing women’s

vulnerability, may also make women more aware of

Table 2

Odds ratios of being fully immunized for age, children aged 0–59 months, Angola MICS 1996

Predictors All children Birth cohorts

Born in war Born in peace

Odds

ratios

95% CI Odds

ratios

95% CI Odds

ratios

95% CI

Child’s age (grouped)

[6–11 months] 1 — —

12–23 months 0.32*** (0.21, 0.47) — —

24–35 months 0.58*** (0.40, 0.83) — —

36–59 months 0.54*** (0.40, 0.74) — —

Child’s age (continuous) — 0.99 (0.97, 1.01) 0.91*** (0.88, 0.94)

Child’s sex

[Female] 1 1 1

Male 0.87 (0.68, 1.12) 0.81 (0.57, 1.16) 0.91 (0.64, 1.31)

Sex of household’s head

[Female] 1 1 1

Male 0.58*** (0.44, 0.78) 0.64** (0.43, 0.97) 0.51*** (0.33, 0.78)

Mean education of household adults

[Less than 1 year] 1 1 1

1–4 years 1.21 (0.92, 1.61) 1.13 (0.76, 1.67) 1.36 (0.89, 2.09)

5+years 1.45*** (1.20, 1.74) 1.44*** (1.11, 1.86) 1.51*** (1.14, 2.00)

Household radio ownership

[Household owns no radio] 1 1 1

Household owns a radio 1.53*** (1.16, 2.02) 1.68*** (1.15, 2.46) 1.44* (0.96, 2.15)

Treatment of drinking water

[Water is not treated or boiled] 1 1 1

Water is treated or boiled 1.50*** (1.13, 1.99) 1.46* (0.98, 2.17) 1.52** (1.01, 2.29)

Area of residence

[Rural] 1 1 1

Urban 2.17*** (1.63, 2.87) 2.58*** (1.73, 3.84) 1.91*** (1.28, 2.85)


[Lighter impact] 1 1 1

Heavier impact 0.65*** (0.49, 0.88) 0.52*** (0.34, 0.80) 0.81 (0.53, 1.24)


[Umbundu] 1 1 1

Kimbundu 1.71*** (1.22, 2.42) 2.09*** (1.27, 3.44) 1.5 (0.91, 2.46)

Kikongo 2.05*** (1.29, 3.26) 2.28** (1.12, 4.64) 1.93** (1.03, 3.61)

Portuguese 1.08 (0.69, 1.69) 1.02 (0.54, 1.92) 1.16 (0.62, 2.17)

Other languages 1.43* (0.98, 2.09) 1.97** (1.13, 3.43) 1.09 (0.64, 1.85)

Likelihood ratio chi-square 268*** 148*** 120***

Number of cases 3702 1953 1749

Notes: Reference categories in brackets; — not applicable; CI confidence intervals; significance level at �po0:1; � � po0:05;� � �po0:01:


their children’s health needs and give them the necessary

decision-making autonomy in trying to meet those

needs.

Acute malnutrition

Table 1 presents levels of malnutrition by two types—

wasting and stunting. Wasting is generally the most

dramatic consequence of malnutrition for a child. It is

also the shortest-lasting, and once the supply of basic

food is re-established, it tends to recede relatively

quickly (Scrimshaw, Taylor, & Gordon, 1971). Because

MICS was conducted 2 years after the politico-military

situation had stabilized, the prevalence of wasting

among surveyed children may have passed its peak.

Also, the levels of acute malnutrition in our tests may be

underestimated because MICS did not collect nutri-

tional status information for children younger than 6

months, who are more likely to suffer from wasting than

are older children. Furthermore, the WHO/NCHS

definition of malnutrition used in this study has some

limitations when applied to sub-Saharan Africa nations,

as the wasting indicator tends to underestimate the

prevalence of malnutrition (Sommerfelt & Stewart,

1994).

Acute malnutrition in Angola, as registered in MICS,

is high by international standards, although not

exceptionally so for sub-Saharan Africa (Madise, Zo .o,

& Margetts, 1999; Sommerfelt & Stewart, 1994). As is

commonly the case, the levels of wasting are higher in

rural areas than in urban areas. Differences are also

noticeable in the two war-impact regions, with more

affected regions having slightly higher levels. Each of the

two birth cohorts replicates the overall pattern, but the

cross-cohort differentials display an advantage of the

war-born cohort, ostensibly defying our expectations.

Yet this latter pattern reflects more age differences than

it does different contexts of birth; other things being

equal, wasting usually has higher prevalence among

younger children. The ethnolinguistic breakdown pro-

duces one clear outlier—Umbundu speakers—whose

level of wasting greatly exceeds those of other groups.

The results of the multivariate tests of wasting, that

include the same set of predictors as the immunization

models and add to it immunization status (whether or

not a child is fully immunized for age), are presented in

Table 3. The models fit the data poorly, and few

statistically significant effects on the likelihood of being

wasted can be observed. Among the few that do exert

non-negligible impacts are the highest level of household

adults’ education (relative to the lowest schooling level),

and ethnicity. Children of Umbundu-speakers tend to

suffer from wasting more than do other children, but

only in two cases is this tendency statistically significant,

and no meaningful pattern of cohort differences can be

discerned.

Chronic malnutrition

Chronic malnutrition patterns, by definition, take a

longer time to establish and are more enduring than

those of wasting. The overall level of stunting in Angola,

54 percent, is high even by the unenviable sub-Saharan

standards (Sommerfelt & Stewart, 1994). As Table 1

shows, the levels of stunting generally rise with age.

Almost half of urban children are stunted, whereas in

rural areas the share of chronically malnourished

children approaches 60 percent. The gap between

regions of heavier and lighter war-impact is almost

equally wide and in favor of the latter. Notably, the

levels of stunting decline considerably in both areas and

in both regions between the war and peace birth cohorts.

Yet, while the gap between urban and rural children

narrows dramatically in the cohort born after the war,

the gaps between children in more and less affected

regions remains exactly the same as in the war birth

cohort.

As in the case of wasting, the stunting statistics reveal

a glaring disadvantage for the Ovimbundu—almost two-

thirds of children in Umbundu-speaking households

suffered from chronic malnutrition. All the groups

improve their nutritional situation between the peace

and war cohorts, with a surprising exception of

Kimbundu-speakers, whose level of stunting in the

peace cohort is even slightly higher than that of

Umbundu-speakers.

Table 4 presents the results of the same type of logistic

regression as in the analysis of wasting. Compared to the

wasting models, the stunting models have a much

stronger predictive power, and the effects of key

predictors lend themselves to meaningful inferences.

Controlling for age and other sociodemographic

factors, the advantage of urban residence, which we

detected at the bivariate level, is not statistically

significant for the whole sample. However, area of

residence performs differently in each of the birth cohort

models. Whereas rural–urban differences have no

significant influence on the likelihood of being stunted

among children conceived and born in peacetime, the

disadvantage of rural children born or conceived during

wartime is statistically significant.

The region effect underscores the vulnerability of

children living in more affected parts of Angola: their

odds of being stunted are a quarter higher than those for

children in less affected areas. The regional differences

loom larger when we look only at children who were

born during war. In contrast, the region’s effect in the

peace cohort model is not statistically significant (due

chiefly to a large standard error).

With respect to ethnicity the multivariate test confirms

the pattern elicited at the bivariate level. Children

in Umbundu-speaking households reveal a profound

disadvantage even after controlling for region of war


impact, area of residence, and other characteristics. The

differences between them and all other ethnolinguistic

groups are highly significant, and the variation among

the latter seems rather minor. In fact, the group that is

noticeably closer to the Ovimbundu is Kimbundu-

speakers; that is, the group that is commonly seen as

the ethnic mainstay of the MPLA government. The

disadvantage of the Ovimbundu is more pronounced

among the war cohort, but, even among children born

after the war, Umbundu-speakers are more likely to be

stunted than any other groups but Kimbundu-speakers

(as was already suggested by the bivariate comparisons).

In sum, although, as in the case of age-adequate

immunization, the strength of ethnolinguistic differentials

Table 3

Odds ratios of being wasted, children aged 6–59 months, Angola MICS 1996



Odds

ratios

95% CI Odds

ratios

95% CI Odds

ratios

95% CI


[6–11 months] 1 — —

12–23 months 3.40*** (1.73, 6.69) — —

24–35 months 1.32 (0.62, 2.79) — —

36–59 months 1.3 (0.65, 2.60) — —

Child’s age (continuous) — 0.98** (0.96, 0.996) 1.13*** (1.06, 1.22)

Child’s sex

[Female] 1 1 1

Male 0.98 (0.69, 1.40) 0.85 (0.55, 1.33) 1.25 (0.68, 2.32)


[Female] 1 1 1

Male 1.02 (0.68, 1.55) 1.09 (0.65, 1.81) 0.99 (0.5, 1.98)

Mean education of household’s adults

[Less then one year] 1 1 1

1–4 years 0.91 (0.72, 1.15) 0.83 (0.62, 1.11) 1.1 (0.71, 1.69)

5+years 0.85* (0.71, 1.02) 0.87 (0.71, 1.08) 0.85 (0.62, 1.18)



Household owns a radio 0.85 (0.56, 1.29) 0.75 (0.45, 1.26) 1.04 (0.52, 2.05)



Water is treated or boiled 0.89 (0.52, 1.53) 0.68 (0.33, 1.38) 1.38 (0.61, 3.13)


[No] 1 1 1

Yes 0.75 (0.35, 1.59) 0.52 (0.18, 1.51) 1.15 (0.37, 3.63)

Area of residence

[Rural] 1 1 1

Urban 0.97 (0.62, 1.50) 1.05 (0.59, 1.85) 0.94 (0.48, 1.85)



Heavier impact 1.09 (0.75, 1.58) 1.12 (0.7, 1.79) 1.17 (0.65, 2.12)


[Umbundu] 1 1 1

Kimbundu 0.57** (0.33, 0.99) 0.43** (0.2, 0.95) 0.86 (0.38, 1.93)

Kikongo 0.59 (0.31, 1.14) 0.72 (0.33, 1.56) 0.42 (0.13, 1.43)

Portuguese 0.68 (0.36, 1.30) 0.63 (0.27, 1.45) 0.79 (0.29, 2.17)

Other languages 0.68* (0.44, 1.06) 0.77 (0.45, 1.32) 0.51* (0.25, 1.04)

Likelihood ratio chi-square 51*** 23* 25**




in the probability of stunting tends to decline between

the two cohorts, this decline is somewhat less manifest

than in the case of immunization.

The effects of other predictors included in this model

also merit attention. As the bivariate distribution

already suggested, the likelihood of stunting tends to

rise with age. Like immunization, stunting is signifi-

cantly associated with household material and sanitary

status: children in radio-owning (i.e., economically

better-off) households, and in households where drink-

ing water is regularly treated, have lower odds of

stunting. Although these effects hold for the war cohort

and the peace cohort alike, the impact of both predictors

is stronger in the peace cohort model, suggesting that the

condition (or the imprint) of war may attenuate the

conventional socioeconomic differentials in nutrition.

Table 4

Odds ratios of being stunted, children aged 6–59 months, Angola MICS 1996



Odds

ratios

95% CI Odds

ratios

95% CI Odds

ratios

95% CI


[6–11 months] 1 — —

12–23 months 2.13*** (1.51, 30.2) — —

24–35 months 1.92*** (1.35, 2.73) — —

36–59 months 2.95*** (2.12, 4.09) — —

Child’s age (continuous) — 1.01*** (1.004, 1.02) 1.12*** (1.07, 1.18)

Child’s sex

[Female] 1 1 1

Male 1.1 (0.91, 1.32) 1.04 (0.84, 1.29) 1.27 (0.88, 1.82)


[Female] 1 1 1

Male 1.14 (0.93, 1.39) 1.18 (0.93, 1.49) 1.11 (0.73, 1.68)

Mean education of household’s adults

[Less then one year] 1 1 1

1–4 years 1 (0.88, 1.15) 1.03 (0.88, 1.20) 0.93 (0.71, 1.21)

5+years 0.97 (0.88, 1.07) 0.99 (0.89, 1.11) 0.89 (0.73, 1.07)



Household owns a radio 0.80** (0.66, 0.98) 0.84 (0.66, 1.07) 0.74 (0.5, 1.09)



Water is treated or boiled 0.66*** (0.52, 0.83) 0.74** (0.56, 0.97) 0.45*** (0.28, 0.73)


[No] 1 1 1

Yes 0.62*** (0.46, 0.84) 0.67** (0.48, 0.94) 0.45** (0.22, 0.92)

Area of residence

[Rural] 1 1 1

Urban 0.85 (0.69, 1.05) 0.76** (0.6, 0.98) 1.13 (0.75, 1.70)



Heavier impact 1.27** (1.03, 1.53) 1.29** (1.003, 1.65) 1.27 (0.85, 1.89)


[Umbundu] 1 1 1

Kimbundu 0.71** (0.53, 0.97) 0.59*** (0.41, 0.84) 1.19 (0.67, 2.11)

Kikongo 0.52*** (0.39, 0.71) 0.51*** (0.36, 0.73) 0.56* (0.3, 1.04)

Portuguese 0.54*** (0.4, 0.75) 0.54*** (0.37, 0.78) 0.55* (0.29, 1.04)

Other languages 0.49*** (0.39, 0.63) 0.48*** (0.37, 0.64) 0.49*** (0.30, 0.78)

Likelihood ratio chi-square 152*** 97*** 63***




As with immunization, child’s sex has no influence on

the probability of being stunted. The effect of household

head’s sex is not significant either, but at least its

direction (the opposite of that in the immunization test)

suggests—not surprisingly—that male-headed house-

holds may have better access to food. Interestingly, the

likelihood of chronic malnutrition does not decrease

with education. Maternal education is known to have a

strong negative association with child malnutrition

(Sommerfelt & Stewart, 1994), but because we cannot

link children’s nutritional status to their mothers’

schooling levels directly, we can only guess that the lack

of association in our case is due to our using the average

schooling level of household adults. Finally, as one

could expect, being fully immunized for age significantly

lowers the odds of being stunted.

Discussion and conclusion

In this study we set out both to document the state of

child immunization and malnutrition in Angola, and to

explore possible associations between civil war and these

two aspects of child health. The 1996 MICS survey,

conducted 2 years after one of the worst bouts of the war

ended and the nation had begun to savour the

precarious peacefulness, captured a very low level of

immunization and widespread malnutrition among

Angolan children. The state of Angolan children’s

nutrition compares unfavorably with most sub-Saharan

nations; Angola’s immunization coverage is even worse

by comparison. Whereas malnutrition is largely a

testimony to how war can undermine normal food

production and distribution, the failure of immunization

exemplifies how war paralyses the normal functioning of

the state. The situation of child immunization and

nutrition in Angola is particularly appalling if we take

into account that Angola, generously endowed with

mineral and agricultural resources, is potentially one of

the richest countries in sub-Saharan Africa.

We should emphasize that due to the nature of the

data our analyses were confined to children born in

1991–1996 who were still alive at the time of the survey.

Since children with no or incomplete immunization and

malnourished children have a higher probability of

dying, our results may underestimate the severity of

malnutrition and of immunization undercoverage in

Angola. However, at least with respect to malnutrition,

it has been shown that the survivor bias is small and

does not considerably distort the picture of malnutrition

prevalence (Boerma et al., 1990).

Our analytic approach allowed us to highlight both

the universal and unique factors that have shaped

Angola’s child immunization and malnutrition patterns.

First, we found that immunization coverage and

malnutrition are determined by standard sociodemo-

graphic factors that have proven their importance in

other, more ‘‘normal’’ settings. Yet, we also found

associations that are specific to Angola. Thus, support-

ing our hypothesis, children living in the parts of the

country where fighting had been particularly ferocious

and generalized, and the devastation most profound,

exhibited significantly lower levels of age-adequate

immunization and higher levels of malnutrition, espe-

cially of stunting, than did children living in the

country’s less affected parts. Controlling for other

factors, the disadvantage of those areas remained

sweeping both in immunization and in chronic malnu-

trition.

The ethnolinguistic factor added another important

nuance to our results. The Ovimbundu, the group that is

often described as the UNITA’s ethnic base, displayed

by far the worst nutritional situation and immunization

coverage. We do not have any evidence to argue that the

Ovimbundu-populated areas have been discriminated

against by the government-run immunization programs,

or have been intentionally denied food and other

assistance by the Luanda regime on the basis of their

ascribed loyalty to the opposition. After all, our analysis

also showed that the government’s supposed backers—

even if by ascription only, Kimbundu-speakers and

perhaps Portuguese-speakers—were not particularly

privileged. The historical and ethnographic studies offer

no evidence of any ecologically or culturally rooted

differences in the amount and allocation of available

household resources or war-unrelated differences in

child caloric intake that might have shaped the observed

disadvantage of Umbundu-speakers.

We believe that an explanation of the Ovimbundu’s

disadvantage lies in the fact that most of them just

happened to live where UNITA control was stronger—

or where UNITA made stronger claims to such

control—and therefore where people had a dispropor-

tionately limited access to food supply and elementary

public health services, regardless of the intensity of the

fighting. As typically happens, children belonging to a

group that is dragged—however unwillingly—into the

centre of the conflict pay a particularly high toll. The

revealed regional and ethnolinguistic differences in

immunization and chronic malnutrition are long-stand-

ing; remarkably, in the levels of wasting—a more

extreme but less enduring form of malnutrition,

especially in the age range considered—no such differ-

ences could be observed. It is also important to note that

the detected effects of region and ethnicity were largely

independent from one another and from the effects of

other, more conventional factors.

With data from just one cross-sectional survey we, of

course, could not make any direct causal inferences

regarding the regional and ethnic patterns of relation-

ship between war and child health. We attempted,

however, to explore such causality indirectly, by


comparing children who were born and/or grew up

during hostilities with those who were born after

hostilities stopped. Lending support to our hypothesis,

the differences in levels of age-adequate immunization

that could be associated with war—rural–urban, regio-

nal, and ethnolinguistic—were most pronounced among

the war-born cohort than among the peace-born cohort.

The analysis of chronic malnutrition revealed a similar

trend. There, however, the inter-cohort differences in the

effects of area, region, and ethnicity were not as

pronounced as in the case of immunization, suggesting

that nutritional equity may take longer to establish after

hostilities cease. Notably, no comparable patterns of

association emerged from the tests of wasting, suggest-

ing again that it may not be a good measure of lasting

war-related differences in children’s nutritional status.

In closing, we want to situate our findings within the

context of Angola’s politico-military situation. The

persistence of war-related regional and ethnic differences

2 years after hostilities ceased is troubling in itself, but

subsequent political developments in Angola did not

give much hope that these differences might disappear

soon. The peace process initiated by the 1994 Lusaka

accord was slow to progress, as the government and the

UNITA opposition were trying to outmanoeuvre each

other in their struggle for power. The general and

nutritional health of the nation’s children, especially of

those most affected by war, was not high on the political

leaders’ agendas. In addition, the worst-off among

Angolan children were concentrated mainly in areas

where the supply of food and health care was hampered

by inadequate infrastructure (all but paralysed during

fighting), and by the opposition’s constant suspicions of

the government’s actions. This reminds us that a post-

war reconstruction, especially in a country with such a

long history of hatred and mistrust between the warring

parties, is a slow process, in which health is likely to

have a lower priority.

The April 2002 cease-fire agreement brought the

country closer than ever before to ending the seemingly

endless civil conflict. When this longest-running African

war is finally over, Angola’s leaders will have to face the

enormous challenge of rebuilding the national economy

and health care system. This challenge, daunting in

itself, will be further magnified by the enduring

unevenness of the war legacy. To meet this challenge

effectively, efficiently, and equitably the government

should build this unevenness into its strategy of national

recovery.

References

African-European Institute (1990). Child survival on the front-

line. Amsterdam, The Netherlands: African-European

Institute.

AlDoori, W., Armijo-Hussein, N., Fawzi, W., & Herrera, . M.

G. (1994). Child nutrition and armed conflicts in Iraq.

Journal of Tropical Pediatrics, 40(1), 32–36.

Amin, R. (1996). Immunization coverage and child mortality in

two rural districts of Sierra Leone. Social Science &

Medicine, 42(11), 1599–1604.

Anstee, M. J. (1996). Orphan of the cold war: The inside story of

the collapse of the Angolan peace process, 1992–93. London:

Macmillan Press.

Boerma, J. T., Sommerfelt, E., Rutstein, S., & Guillermo, R.

(1990). Immunization: levels, trends and differentials. Co-

lumbia, MD: Demographic and Health Surveys, Compara-

tive studies Studies, No.1 Institute for Resource

Development/Macro Systems, Inc.

Brentlinger, P., Hern!an, M., Hernandez-D!ıaz, S., Azaraff, L. S.,

& McCall, M. (1999). Childhood malnutrition and postwar

reconstruction in rural El Salvador: A community-based

survey. Journal of the American Medical Association, 28(2),

184–190.

Brittain, V. (1998). Death of dignity: Angola’s civil war. London,

Chicago: Pluto Press.

Chelala, C. (1999). Angola: A forgotten emergency. The Lancet,

354, 1365.

Cliff, J., & Noormahomed, A. R. (1988). Health as a target:

South Africa’s destabilization of Mozambique. Social

Science & Medicine, 27(7), 717–722.

Cutts, F. T., dos Santos, C., Novoa, A., David, P., Macassa,

G., & Soares, A. C. (1996). Child and maternal mortality

during a period of conflict in Beira City, Mozambique.

International Journal of Epidemiology, 25(2), 349–356.

Desgr!ees du Lo #u, A. (1996). Sauver les enfants: le r #ole des

vaccinations. une enqu#ete longitudinale en milieu rural "a

Bandafassi au S!en!egal. Paris, France: Les Etudes du

CEPED, No. 12, Centre Fran-cais sur la Population et le

D!eveloppement [CEPED].

Garenne, M. (1997). Political crises and child survival: Five case

studies in sub-Saharan Africa. Paper presented at the 23rd

IUSSP General Conference. Beijing, China, 11–17 October

1997.

Garenne, M., Coninx, R., & Dupuy, C. (1996). Effets de la

guerre civile au Centre-Mozambique et !evaluation d’une

intervention de la Croix Rouge. Les Dossiers du CEPED,

No. 38.

Gessner, B. D. (1994). Mortality rates, causes of death, and

health status among displaced and resident populations of

Kabul, Afghanistan. Journal of the American Medical

Association, 272(5), 382–385.

Gubhaju, B. (1997). L’Ethiopie au temps des troubles. In J.-P.

Jean-Claude Chasteland, & J.-C. Chesnais (Eds.), La

population du monde: enjeux et probl"emes (pp. 149–161).

Paris, France: Presses Universitaires de France.

Heywood, L. (2000). Contested power in Angola: 1840 to the

present. Rochester, NY: University of Rochester Press.

INE [Instituto Nacional de Estat!ıstica] (1998). Inqu!erito de

indicadores multiplos: Demografia, !agua e saneamento, sa !ude

materno infantil, nutri-c *ao, vacina-c *ao, fecundidade, mortali-

dade. Luanda, Angola: Instituto Nacional de Estat!ıstica-

UNICEF.

Kinfu, A. Y. (1999). Child undernutrition in war-torn society:

The Ethiopian experience. Journal of Biosocial Science,

31(3), 403–418.


Levy, B. S., & Sidel, V. W. (1996). War and public health.

Oxford, England: Oxford University Press.

Madise, N. J., Zo .o, M., & Margetts, B. (1999). Heterogeneity of

child nutritional status between households: A comparison

of six Sub-Saharan Africa countries. Population Studies,

53(3), 331–343.

Malaguias, A. (2000). Ethnicity and conflict in Angola:

Prospects for reconciliation. In J. Cilliers, & C. Dietrich

(Eds.), Angola’s war economy: The role of oil and

diamonds. Pretoria, South Africa: Institute for Security

Studies.

Martin-Pr!evel, Y., Delpeuch, F., Massamba, J.-P., Adoua-

Oyila, G., Coudert, K., & Tr"eche, S. (2000). Deterioration

in the nutritional status of young children and their mothers

in Brazzaville, Congo, following the 1994 devaluation of the

CFA franc. Bulletin of the World Health Organization, 78(1),

108–117.

MSF [M!ed!ecins Sans Fronti"eres] (2000). Fa-cade of normality

in Angola hides manipulation, violence, and neglected

population. MSF Report and Press Release (November

2000). MSF web site, www.msf.org. Accessed on 3 October

2001.

Redinha, J. (1975). Etnias e Culturas de Angola. Luanda,

Angola: Instituto de Investiga-c*ao Cient!ıfica de Angola.

Samb, B., Aaby, P., Whittle, H., Seck, A. M. C., & Simondon,

F. (1997). Decline in measles case fatality ratio after the

introduction of measles immunization in rural Senegal.

American Journal of Epidemiology, 145(1), 51–57.

Scrimshaw, N. S., Taylor, C. E., & Gordon, J. E. (1971).

Interaction between nutrition and infections. World Health

Organization Monograph No. 57. Geneva, Switzerland:

World Health Organization.

Sommerfelt, A. E., & Piani, A. L. (1997). Childhood Immuniza-

tion: 1990–1994. Calverton, MD, USA: Demographic and

Health Surveys, Comparative studies No.22. Macro Inter-

national.

Sommerfelt, A. E., & Stewart, K. (1994). Children’s nutritional

status. Columbia, MD, USA: Demographic and Health

Surveys, Comparative studies No. 12 Macro International,

Institute for Resource Development.

Spears, I. A. (1999). Angola’s elusive peace: The collapse of the

Lusaka accord. International Journal, 54(4), 562–581.

STATA (2001). STATA reference manual, release 7, Vol. 2.

College Station, TX Stata Press.

The New York Times (2002). Angolans cheer the peace and

hope it will stay awhile. The New Times, April 5.

The UN Security Council (2001). Report of the secretary-

general on the United Nations office in Angola (UNOA), 10

October 2001, United Nations.

UNICEF (1999). The progress of nations, 1999. The UNICEF

annual report. New York: UNICEF.

US Food and Nutrition Board (1974). Comparison of body

weights and body heights of groups of children. Atlanta, GA:

US Department of Health, Education and Welfare.

World Bank (2002). World development indicators database.

Washington, DC: The World Bank.

World Health Organization (1983). Measuring change in

nutritional status. Guidelines for assessing the nutritional

impact of supplementary feeding programmes for vulnerable

groups. Geneva: WHO.


http://www.msf.org

civil war and child health: regional and ethnic dimensions of child immunization and malnutrition in...

Documents