European Journal of Clinical Nutrition (2000) 54, 463472

2000 Macmillan Publishers Ltd All rights reserved 09543007/00 $15.00

www.nature.com/ejcn

Undernutrition in relation to childhood infections: a prospective

study in the Sudan
J Kossmann1, P Nestel1, MG Herrera1,2, A El Amin3 and WW Fawzi2*
1

Harvard Institute for International Development, Cambridge, MA, USA; 2Department of Epidemiology and Nutrition Harvard School of
Public Health, Boston, MA, USA; and 3Ministry of Health, Khartoum, Sudan

Objective: The aim of the study was to examine the relationships between nutritional status and diarrhoea and
respiratory infections.
Design: Prospective cohort study within the framework of a randomized double-blind placebo-controlled
intervention trial.
Setting: In rural communities in the Khartoum and Gezira regions, in Northern Sudan.
Subjects: 28,753 Sudanese pre-school children between 6 months and 6 y old.
Methods: Relative risks of subsequent diarrhoea and respiratory infections in relation to nutritional status
measured by anthropometry (Z-scores of height-for-age (H=A), weight-for-height (W=H), and weight-for-age
(W=A), which reect stunting, wasting and underweight, respectively) were estimated using odds ratios from
logistic regression adjusting for various covariates.
Results: H=A, W=H and W=A were signicantly and inversely associated with subsequent diarrhoea and febrile
diarrhoea (P for trend < 0.001) with risks being 2.00 times higher (95% condence interval, CI (1.64, 2.43))
among children with W=A Z-scores below 74 Z, and 1.75 times higher (95% CI (1.56, 1.96)) among those with
a W=A Z-score between 74 and 73 Z compared with children having a W=A Z-score  1. Age, gender, region
of residence and seasonality modied these associations. Also, febrile cough was inversely associated with W=A
and W=H (P < 0.03), with risks ranging from 1.41 times higher (95% CI (1.02, 1.97)) to 1.21 times higher (95%
CI (1.04, 1.41)) in the group of underweight children with W=A Z-scores below 74 and between 72 and 71 Z,
all compared with normally nourished children (  71 Z).
Conclusions: The reduction of severe but also mild and moderate undernutrition is necessary through nutrition,
health and socio-economic improvement in order to prevent morbidity.
Sponsorship: This study was carried out under cooperative agreement no. DAN-00450G-SS-6067 of the Ofce
of Nutrition, US Agency for International Development, Washington DC, and the Harvard Institute for
International Development.
Descriptors: nutritional status; anthropometry; diarrhoea; respiratory infections; socio-economic factors
European Journal of Clinical Nutrition (2000) 54, 463472

Introduction
Diarrhoea and respiratory infections are leading causes of
child deaths, and the majority of these occur in developing
countries (Tulloch & Richards, 1993). Undernutrition is
known to adversely affect specic and non-specic defence
mechanisms resulting in increased susceptibility to infections. This, in turn, can cause further deterioration in
nutritional status through reduced food intake, malabsorption, mobilization of body stores, increased losses and other
systemic reactions that can affect linear and ponderal
growth (Scrimshaw, 1992). Scrimshaw (1992), Santos
(1994), and Rivera and Martorell (1988) have comprehensively reviewed evidence of the synergistic relationship
between undernutrition and infection. Community-based
studies investigating associations between nutritional
*Correspondence: WW Fawzi, Department of Nutrition, Harvard School
of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA.
Guarantors: P Nestel, MG Herrera and WW Fawzi.
Contributors: JK contributed to the design of this analysis, prepared the
drafts of the paper and did the statistical analysis. P Nestel designed and
implemented the eld study, MH and with WF initiated the study, and
provided advice in data analysis and commented on the manuscript. A ElA contributed to the implementation of the study in the eld.
Received 27 September 1999; revised 19 December 1999; accepted
13 January 2000

status and morbidity from diarrhoea and respiratory infections have reported inconsistent results. A study from
Nigeria showed that moderate and severe undernutrition
were associated with both the incidence and duration of
diarrhoeal disease (Tomkins, 1981), while three other
studies observed positive associations between undernutrition and the incidence of diarrhoea only (Sepulveda et al,
1988; El-Samani et al, 1988; Lindtjrn et al, 1993). Yet
another three studies found positive associations between
severe and moderate undernutrition and the duration or
percentage of annual time with diarrhoeal illness, but not
the incidence of disease (Trowbridge et al, 1981; Black
et al, 1984; Bairagi et al, 1987). In two studies, no
associations were observed between nutritional status and
the incidence or duration of diarrhocal disease (Chen et al,
1981; Mathur et al, 1985).
Few community-based prospective studies have examined the relationship between nutritional status and acute
respiratory infections. Signicant positive associations
have been observed between stunting and wasting and the
incidence (Tupasi et al, 1988, 1990; Ballard et al, 1995;
Smith et al, 1991) or duration of respiratory infections
(James, 1972). Such associations were not found in a study
in Southern Ethiopia (Lindtjrn et al, 1993). Some of the
inconsistencies in the results may have been due to the

Undernutrition and childhood infections

J Kossmann et al

464

differences in the design, demographic structure of the

study populations, choice and classication of the nutritional indices and morbidity, or the analytical procedures
used in the studies.
To investigate the relationships between nutritional
status and morbidity from diarrhoea and respiratory infections, we analysed data from a cohort of Sudanese children
collected within the framework of the Sudan vitamin A
study (Herrera et al, 1992).
Methods
The Sudan vitamin A study was conducted in ve rural
councils in Northern Sudan and included 17,031 eligible
households with 29,615 children between 6 and 72 months
old. Approximately 3% of the children were treated for
signs of vitamin A deciency at baseline and excluded from
the study (Herrera et al, 1992). The 28,753 children
enrolled in the study were visited after 6 months (round
2) 12 months (round 3) and 18 months (round 4). At each
round the children enrolled received either 200,000 IU
vitamin A with 40 IU of vitamin E or 40 IU of vitamin E
alone. Follow-up rates were 92% for the rst round, 87%
for the second, and 84% for the third round. Information on
socio-economic factors collected at baseline included
maternal and paternal literacy, the availability of water
and a latrine in the house, region of residence, and a
subjective assessment of household wealth (on a fourpoint scale). Seasonal variation was reected by calendar
month of each visit. Further details of the study design have
been published elsewhere (Herrera et al, 1992; Fawzi et al,
1994).
Anthropometric measurements of weight (to the nearest
100 g) and height (to the nearest 0.1 cm) were taken at each
round at a central location using a Salter-hanging-scale and
a locally-made anthropometer, respectively. Z-scores for
height-for-age (H=A) reecting stunting, weight-for-height
(W=H) reecting wasting, and weight-for-age (W=A)
reecting underweight were calculated using the Centre
for Disease Control Anthropometric Software Package
(CASP), based on the growth curves of the National
Centre for Health Statistics (Hamill et al, 1977). Z-scores
were used as a continuous variable and as discrete categories using ve one-Z-score intervals ranging from below
74 Z to 71 Z or higher. Children with a Z-score below
72Z were considered undernourished, those with Z-scores
between 73 and 72 Z moderately undernourished, those
with Z-scores between 74 and 73 severely undernourished, and those below 74 Z as very severely undernourished (Beaton et al, 1990). Children having Z-scores of
72 Z or above were considered not undernourished or as
normally nourished.
At each round information on dietary vitamin A intake
was obtained by asking the mothers whether the child had
eaten any of 30 listed foods or a composite dish containing
any of those foods during the day preceding each visit. The
approximate dietary vitamin A intake of preformed vitamin
A, carotenoid precursors of vitamin A, and total vitamin A
grouped in quintiles, was calculated for each child by
multiplying the nutrient content of each food item by an
assumed averaged portion as described in detail by Fawzi
et al (1994, 1995). Mothers were also asked whether the
child was exclusively breast or bottle-fed, fed by a combination of both, breast-fed and given complementary food,
or fully weaned. Few infants and young children were

European Journal of Clinical Nutrition

exclusively breast-fed; thus, fully or partially breast-fed

children were grouped into one category.
At each round morbidity was recorded based on the
reported presence of specied symptoms during the week
preceding each visit. Morbidity was grouped into seven
mutually exclusive categories: no morbidity, diarrhoea
alone (dened as three or more loose or watery stools per
24 h period), diarrhoea with fever, cough alone (lasting
more than 24 h), complicated cough (cough with fever
and=or diarrhoea), fever alone, and measles. In some
analyses the category `complicated cough' was subdivided
into cough with fever, cough with diarrhoea, and cough
with both diarrhoea and fever.
Each child was considered as a new observation at the
beginning of each round, conditional on his=her presence at
the beginning of the subsequent round, resulting in 73,363
observations, each contributing six child-months. Nutritional status based on anthropometric measurements was
used to prospectively predict morbidity 6 months later. The
analyses were also done separately for each round and also
limited to children free of symptoms at the beginning of
each round. Data on morbidity and anthropometry at the
beginning of the rounds were missing for approximately
5% of children, who tended to be from poorer households,
at rounds two and three.
Risk estimates with 95% condence intervals (CI) for
morbidity from the specied symptoms in relation to
nutritional status were computed using odds ratios (OR)
from multivariate logistic regression models, with the
reference category being that which reected the best
nutritional status in the respective analysis (  1 Z or
 72 Z). Adjustments were made for variables that were
for most cases signicantly associated with the specied
symptoms, that is age (four indicator variables, reference
category:  48 months), gender, socio-economic variables
like the presence of a latrine or piped water in the house
or maternal literacy (all yes=no), region of residence (four
indicator variables, reference category: region Abu Dileig),
household wealth (ordinal with four levels), periodic vitamin A supplementation during the study (yes=no), dietary
vitamin A intake (ordinal in quintiles), feeding practices
(breast-feeding under the age of 24 months: yes=no) and
season (ve indicator variables, reference category: Jan
Feb). Household wealth, which was assessed by the interviewers on a four-point scale, was positively correlated
with the presence of a latrine and piped water in the house
as well as maternal literacy, and was included in the model
as an independent indicator of socio-economic status. We
expected recent morbidity would affect acute nutritional
status and thus included information on the presence of the
specied symptoms (using three indicator variables: diarrhoea and febrile diarrhoea combined, cough and complicated cough combined, and measles and fever combined) in
the regression models. Tests for trends in the relationship
between nutritional status and morbidity were performed
with nutritional status as a continuous variable. We controlled for H=A (continuous variable) when examining the
relationship between wasting and subsequent diarrhoea or
respiratory symptoms, and vice versa, to reect the effects
of past and current insults on nutritional status (Beaton
et al, 1990). The relationship between undernutrition and
measles was not examined in detail due to the small
number of cases.
The relationships between of H=A, W=H and W=A and
diarrhoea, diarrhoea with fever, cough, and cough with

Undernutrition and childhood infections

J Kossmann et al

fever were examined across categories of age (ve levels:

6 11, 12 23, 24 35, 36 47, and 48 months or older),
gender, vitamin A supplementation (placebo=vitamin A),
dietary vitamin A intake (quintiles), breast-feeding
(yes=no), region (ve geographical regions: Abu Dilieg,
Ri Genoub, El Jaeli, Ri Shamal, and Gezira) and season
(six 2-month intervals). The risk estimates derived from the
prior analyses did not markedly differ in the two categories
 71 Z or  72 Z, thus, we combined these two categories as one in the stratied analysis. To test formally for
interactions, product terms for the nutritional indicator and
the variable of interest were included in the models and
log-likelihood-ratio tests were done.
Statistical signicance was dened as a probability level
below 0.05. The analyses were done using Statistical
Analysis System (SAS), version 6.04. The study was
approved by the Committee on the Use of Human Subjects
in Research of the Harvard School of Public Health, the
Director General of Primary Health Care of the Ministry of
Health in the Sudan, and both the Directors of Health for
Khartoum and Central Regions.
Results
Morbidity rates among all children (per 1000 child-months)
were 9.6 for diarrhoea alone, 1.6 for diarrhoea with fever,
10.7 for diarrhoea and diarrhoea with fever combined, 26.7
for cough alone, 6.5 for complicated cough with rates of 3.6
for children with cough and fever, and 11.5 for fever alone.
The prevalence of diarrhoea, diarrhoea with fever, and
cough with fever was highest in the age-group 6 23
months and cough alone in the age groups 24 35 and
36 47 months.
All three nutritional indices H=A, W=H and W=A
reecting nutritional status were signicantly associated
with age, region of residence as well as seasonality and

signicantly and inversely with breast-feeding, the availability of water in the house and household wealth. Gender
was only signicantly associated with W=A and W=H, and
maternal literacy signicantly and inversely with H=A and
W=A only. The prevalence of stunting and of wasting was
signicantly higher among children from poorer households, among those living in households that did not have
piped water, as well as among breast-fed children (Table 1).
Breast-feeding was protective against wasting and stunting
among the under 2-y-old children with 42% higher risk of
stunting (95% CI (33%, 51%) comparing non-breast-fed
children with those being breast-fed).
An inverse relationship between diarrhoea and febrile
diarrhoea and nutritional status was observed, and because
the associations went in a similar direction and were only
slightly more pronounced for febrile diarrhoea than for
diarrhoea alone, the two categories were collapsed into one,
namely diarrhocal disease. Results from the multivariate
logistic regression analyses for diarrhoeal disease in relation to nutritional status, after adjusting for covariates, are
presented in Table 2.
Very severely stunted children were 1.50 times (95% CI
(1.33, 1.69)) more likely to have had subsequent diarrhoeal
disease than children whose H=A Z-score was 71 Z or
above, after adjusting for age and gender (Table 2, I, model
a). The multivariate association after adjustments for socioeconomic variables, morbidity at the beginning of the
respective round, dietary variables and season was stronger
with 69% higher risk of subsequent disease (95% CI (49%,
92%), Table 2, I, model b). Including W=H as a continuous
variable in the models did not improve the predictive power
(Table 2, I, model c), although the association was statistically signicant (P < 0.001).
Mild and moderate and severe wasting increased the risk
of subsequent diarrhoeal disease by 9% (95% CI (2%,
17%)) and 34% (95% CI (21%, 48%), respectively), even

465

Table 1 Stunting and wasting by age, gender, socioeconomic variables and breast-feeding at baseline

Age (months)
6 11
12 23
24 35
36 47
 48
Gender
Boys
Girls
Literacy
Yes
No
Water
Yes
No
Wealth
Well
Average
Poor
Very poor
Breast-feeding
Yes
No

No. of childrena

Stuntingb

1396
4367
4499
4409
8607

13.4
32.4
28.6
27.1
25.3

11624
11653

P-valueb

Relative risk (95% CI)c

Wastingb

P-valueb

Relative risk (95% CI)c

P < 0.01

1.0
6.02
7.08
7.26
6.02

7.17)
8.41)
8.58)
7.12)

6.8
10.4
5.8
4.4
5.8

P < 0.01

1.0
1.56
0.83
0.63
0.68

26.6
26.2

P 0.10

1.0
0.97 (0.92, 1.03)

7.0
6.1

P < 0.01

1.0
0.82 (0.76, 0.89)

11357
11207

23.0
28.7

P < 0.01

1.0
1.10 (1.03, 1.19)

5.1
0.7

P < 0.01

1.0
1.09 (0.99, 1.22)

10116
13161

23.3
29.2

P < 0.01

1.0
1.19 (1.11, 1.27)

5.3
7.6

P < 0.01

1.0
1.22 (1.11, 1.35)

2322
11556
7241
2142

19.9
24.2
29.9
33.0

P < 0.01

1.0
1.28 (1.16, 1.41)
1.64 (1.47, 1.83)
1.81 (1.57, 2.08)

3.2
6.2
7.6
10.0

P < 0.01

1.0
1.79 (1.50, 2.14)
1.91 (1.59, 2.31)
2.38 (1.93, 2.94)

4129
19149

31.2
25.2

P < 0.01

1.0
1.42 (1.33, 1.51)

9.0
6.1

P < 0.01

1.0
1.31 (1.16, 1.48)

(5.09,
(5.96,
(6.14,
(5.09,

(1.31,
(0.68,
(0.52,
(0.56,

1.85)
1.01)
0.78)
0.83)

Each child contributing six child-months.

Stunting and wasting (per 1000 child-months) reected by height for age (H=A) and weight for height (W=H), respectively, based on the
cut-off of 72 Z; 4% of the children had both H=A and W=H Z-scores below 72 Z. P-values are from Maentel Haenszel w2 tests.
c
Risk estimates and 95% condence intervals from logistic regression model including the characteristics age (four indicator variables),
gender, piped water in the house, latrine in the house, maternal literacy, breast-feeding (all yes=no), household wealth (three indicator
variables) and dietary vitamin A intake (four indicator variables) adjusting for the respective other characteristics.
b

European Journal of Clinical Nutrition

Undernutrition and childhood infections

J Kossmann et al

466

Table 2 Diarrhoeal disease in relation to nutritional status

Undernutrition
I. Height-for-age (H=A)
No. of children with diarrhoeal disease
Morbidity rates according to nutritional
status (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season
Model c: b weight-for height Z-score

Very severe
<74 Z
433
7.5

Severe
74 to 73 Z
681
6.4

Moderate
73 to 72 Z

Mild
72 to 71 Z

 1 Z

1239
6.0

1329
5.2

1047
4.5

1.50 (1.33, 1.69) 1.46 (1.32, 1.61)

1.69 (1.49, 1.92) 1.49 (1.34, 1.65)

1.37 (1.25, 1.49)

1.36 (1.24, 1.49)

1.18 (1.08, 1.28)

1.15 (1.05, 1.25)

1.0
1.0

1.68 (1.48, 1.90) 1.47 (1.33, 1.63)

1.35 (1.23, 1.47)

1.14 (1.05, 1.25)

1.0

566
7.2

1878
5.5

2211
8.9

1.32 (0.81, 2.13) 1.28 (0.96, 1.72)

1.53 (0.93, 2.50) 1.25 (0.92, 1.69)

1.31 (1.19, 1.45)

1.34 (1.21, 1.48)

1.09 (1.02, 1.16)

1.09 (1.02, 1.17)

1.0
1.0

1.68 (1.02, 2.84) 1.24 (0.92, 1.68)

1.30 (1.19, 1.44)

1.08 (1.01, 1.47)

1.0

1614
5.8

1536
4.9

733
4.4

1.30 (1.19, 1.42)

1.13 (1.21, 1.45)

1.13 (1.04, 1.24)

1.13 (1.03, 1.24)

1.0
1.0

Trend b

P < 0.001

II. Weight-for-height (W=H)

No. of children with diarrhoeal disease
Morbidity rates according to nutritional
status (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season
Model c: d height-for-age Z-score

20
10.4

54
8.6

P < 0.001

III. Weight-for-age (W=A)

No. of children with diarrhoeal disease
Morbidity rates according to nutritional
status (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season

153
10.1

698
8.1

1.74 (1.44, 2.10) 1.62 (1.46, 1.81)

2.00 (1.64, 2.43) 1.75 (1.56, 1.96)

P < 0.001

Multivariate models included nutritional status: height-for-age (H=A), weight-for-height (W=H) and weight-for-age (W=H) respectively (four indicator
variables, reference: Z-score of 71 Z or above), age (0 11, 12 23, 24 35, 36 47 and  48 months), gender, and variables assessing socioeconomic
status (SES), that is piped water in the house, latrine in the house, maternal literacy (all yes=no), household wealth (ordinal with four levels) as well as
region of residence (four indicator variables) and variables accounting for vitamin A intake: capsule (vitamin A=placebo), dietary vitamin A intake (ordinal
in quintiles) as well as morbidity in the previous round (three indicator variables), season (ve indicator variables) and H=A and W=H Z-score (continuous)
as further specied in the text.
b
P-values of tests for trend from logistic regression models adjusting for the specied covariates and using nutritional status as continuous variable.

after controlling for the covariates. Among the very severe

wasted children (below 74 Z) this association was stronger
but lacked signicance (Table 2, II, model b). Adjusting for
H=A resulted in a slightly stronger and signicant association, although only among the very severely undernourished children (model c).
The age and gender adjusted risk of diarrhoeal disease
among the very severely and severely underweight children
was 74% higher (95% CI (44%, 110%)) and 62% higher
(95% CI (46%, 81%)), respectively, than among children
with W=A Z-scores  1 (Table 2, III, model a). This
association was stronger after adjustments for further covariates and the risks of diarrhoeal disease among the very
severely and severely underweight children increased by
100% (95% CI (64%, 143%)) and 75% (95% CI (56%,
96%)), respectively (Table 2, III, model b), compared with
children having a Z-scores of  1. Among children with
febrile diarrhoea only (data not shown), those who were
very severely or severely underweight were at a 182%
higher risk (95% CI (82%, 335%)) and a 114% higher risk
(95% CI (61%, 185%)) of subsequent disease, respectively.
The risk of diarrhoeal disease among moderately undernourished children was increased by 30 36% even after
adjusting for covariates (Table 2, I III, models b). Mild
stunting increased the risk of diarrhoeal disease by 15%
(95% CI (5%, 25%)) similar to the risk for mild underweight, and both risks were higher than the 9% increase in
risk among the mildly wasted children, all compared with
children with Z-scores  1 (Table 2, II, models b and c).
European Journal of Clinical Nutrition

Tests for trend of diarrhoeal disease in relation to H=A,

W=H and W=A were signicant for all predictors
(P < 0.001, model b). Limiting the analyses to children
free of symptoms at baseline or repeating the analyses
using the cut-off 72 Z or examining these associations by
round did not change the results substantially for any of the
symptoms.
Including W=H in the models to estimate the associations between morbidity and HA, and vice versa, did not
substantially change the results, which remained signicant; thus, the models without these variables were used in
the stratied analyses, which were done to examine the
potential modication of the association between diarrhoeal
disease and undernutrition by other variables. The results
are presented for underweight (Table 3), because the
associations were strongest with weight for age in this
study, although the relationships between H=A or W=H and
subsequent diarrhoea were similar.
Age, region and season were strong independent predictors of diarrhoeal disease (P < 0.001 in all cases).
Among normally nourished infants (Table 3, column 3)
those over 12 months old were at a 6.38 times greater risk
of subsequent diarrhoeal disease (95% CI (5.12, 7.95))
while those of 12 23, 24 35, and 36 47 months of age
were at a 4.46 (95% CI (3.89, 5.12)), 2.77 (95% CI (2.43,
3.16)), and 1.78 (95% CI (1.54, 2.05)) times greater risk,
respectively, compared with children older than 48 months.
Normally nourished boys were at an 8% lower risk (95% CI
(716%, 1%)) of subsequent disease than girls, but this

Undernutrition and childhood infections

J Kossmann et al

difference was not statistically signicant. Among the

normally nourished children those breast-fed were 23%
less likely (95% CI (733%, 711%)) to have had subsequent diarrhoeal disease compared with non breast-fed
children.
Compared to children living in Abu-Dileig, normally
nourished children living in Gezira were 30% more likely
(95% CI (9%, 57%)) to have had subsequent diarrhoeal
disease, while those in Ri Shamal and Ri Genoub were
at a 29% (95% CI (745%, 79%)) and 33% (95% CI
(746%, 716%)) lower risk of subsequent diarrhoea.
Among normally nourished children, the risk of diarrhoeal
disease was 40 50% lower in the hot dry months (May
December) than in the cool months of January February.
Among undernourished children (Table 3, columns 4
7), severe underweight increased the risk of diarrhoeal
disease in all age-groups, especially among infants under
12 months by 93% (95% CI, 17%, 219%) and children
between 36 and 47 months old by 85% (95% CI, 50%,
128%) compared with better nourished children having a
Z-score of  72. Moderate underweight increased the risk
of diarrhoeal disease in the two youngest age groups by
17 28% compared with their normally nourished cohorts.
The elevated risk of diarrhoea was similar for moderately
underweight girls with 15% (95% CI, 3%, 28%) and 17%
(95% CI, 10%, 29%), respectively, and boys with 15%
(95% CI, 3%, 28%) and 17% (95% CI, 10%, 29%),
respectively, compared with normally nourished boys and
girls.
Severely underweight non breast-fed children under 2 y
old were at 57% greater risk (95% CI (24%, 99%)), and
those breast-fed at 46% greater risk (95% CI (17%, 82%))

of subsequent diarrhoeal disease compared with normally

nourished children.
For children living in any region except for Ri Shamal,
being severely underweight increased the risk of diarrhoeal
disease by 51 75% compared with normally nourished
children living in the same region. In Abu Dileig and
Gezira moderate underweight also increased the risk of
diarrhoeal disease signicantly by 26% (95% CI (2%,
56%)) and 24% (95% CI (12%, 38%)), respectively.
Throughout the study severely underweight children
were at a greater risk of diarrhoeal disease than the
normally nourished underweight children, but those who
were weighed and measured between March and August
were at 76 78% greater risk of diarrhoeal disease than
their normally nourished cohorts measured at the same time
of the year. Similarly, moderately underweight children
weighed and measured between March and August were at
signicantly greater risk of diarrhoeal disease than their
normally nourished cohorts.
Cough alone was signicantly and inversely associated
with wasting and underweight (P for trend < 0.01 for both:
data not shown). Among the very severely wasted and very
severely underweight children, the risk estimates for cough
were 0.59 (95% CI (0.36, 0.97) and 0.67 (95% CI (0.57,
0.80)), respectively, compared with those for children
having a Z-score  1 (data not shown).
Cough with fever was also signicantly and positively
associated with wasting and underweight (tests for trend:
P 0.029 and P 0.025). In multivariate analysis as
shown in Table 4, severe stunting did not predict subsequent febrile cough, while very severe wasting increased
the age and gender adjusted relative risk by 125% (95% CI

467

Table 3 Modication of the relationship between diarrhoeal disease and weight for age (W=A)

Characteristic
Age (months)
 48
36 47
24 35
12 23
0 11
Gender
Boys
Girls
Breast-feedingd
Yes
No
Region
1, Abu Dileig
2, Ri Genoub
3, El Jaeli
4, Ri Shamal
5, Gezira
Season
January February
March April
May June
July August
September October
November December

Stratum-specic
Risk of diarrhoeal disease within subgroups of characteristicsc
Morbidity rates risk of morbidity among
Interaction, w2(d.f.),
among all childrena children > 2 SDb
< 73 Z
73 to 72 Z
> 72 Z
P-valuee
41.0
9.7
15.8
24.6
40.6

1.00
1.78
2.77
4.46
6.38

1.30)
1.31)
1.30)
1.46)
1.29)

1.00
1.00
1.00
1.00
1.00

39.25 (4) < 0.0001

10.6
10.8

0.92 (0.84, 1.01)

1.00

1.55 (1.37, 1.76)

1.64 (1.45, 1.86)

1.17 (1.10, 1.29)

1.15 (1.03, 1.28)

1.00
1.00

0.62 (1) < 0.43

4.4
3.9

0.77 (0.67, 0.89)

1.00

1.46 (1.17, 1.82)

1.57 (1.24, 1.99)

1.10 (0.93, 1.31)

1.45 (1.22, 1.73)

1.00
1.00

1.20 (1) 0.27

11.5
7.6
7.2
7.9
16.5

1.00
0.67
1.02
0.71
1.30

(0.54,
(0.83,
(0.55,
(1.09,

0.84)
1.26)
0.91)
1.57)

1.49
1.70
1.64
1.16
1.75

(1.19,
(1.39,
(1.26,
(0.89,
(1.53,

1.87)
2.08)
2.12)
1.53)
2.01)

1.26
0.96
1.12
1.19
1.24

(1.02,
(0.78,
(0.89,
(0.95,
(1.12,

1.56)
1.19)
1.41)
1.49)
1.38)

1.00
1.00
1.00
1.00
1.00

43.21 (4) < 0.0001

16.7
16.9
12.5
8.1
6.3
9.7

1.00
1.16
0.60
0.44
0.46
0.57

(1.00,
(0.51,
(0.38,
(0.38,
(0.49,

1.35)
0.71)
0.52)
0.56)
0.67)

1.37
1.78
1.76
1.77
1.49
1.69

(1.10,
(1.38,
(1.47,
(1.40,
(1.18,
(1.01,

1.70)
2.29)
2.12)
2.24)
1.89)
2.00)

1.33
1.23
1.35
1.17
0.92
1.02

(1.13,
(1.01,
(1.14,
(0.93,
(0.71,
(0.85,

1.56)
1.48)
1.58)
1.46)
1.86)
1.23)

1.00
1.00
1.00
1.00
1.00
1.00

28.89 (5) < 0.0001

(1.54,
(2.43,
(3.89,
(5.12,

2.05)
3.16)
5.12)
7.95)

1.59
1.85
1.58
1.49
1.93

(1.26,
(1.50,
(1.35,
(1.25,
(1.17,

1.98)
2.28)
1.85)
1.76)
3.19)

1.10
1.08
1.20
1.28
1.17

(0.93,
(0.89,
(0.96,
(1.12,
(1.10,

Per 1000 child months.

From multivariate models including weight-for-age (W=A) Z-scores, age, gender, piped water in the house, latrine in the house, maternal literacy, wealth,
region of residence, vitamin A supplementation during the study, dietary vitamin A intake, morbidity in the previous round, breast-feeding, and season as
further specied in the text.
d
The analysis involving breast-feeding was limited to children under the age of 2 y.
e 2
w as the difference between the 2 log-likelihood of models with and without respective interaction product term; d.f. degrees of freedom.
b, c

European Journal of Clinical Nutrition

Undernutrition and childhood infections

J Kossmann et al

468

Table 4 Cough with fever in relation to nutritional status

Undernutrition
I. Height-for-age (H=A)
No. of children with febrile cough
Morbidity rates (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season
Model c: d weight-for height Z score

Very severe
< 74 Z

Severe
74 to 73 Z

Moderate
73 to 72 Z

Mild
72 to 71 Z

 71 Z

118
3.9

202
3.7

418
3.9

468
3.6

360
3.1

1.22 (0.99, 1.50)

1.05 (0.84, 1.30)

1.19 (1.00, 1.42)

1.10 (0.93, 1.32)

1.30 (1.12, 1.49)

1.22 (1.06, 1.41)

1.23 (1.07, 1.41)

1.19 (1.03, 1.36)

1.0
1.0

1.04 (0.84, 1.30)

1.10 (0.92, 1.31)

1.21 (1.05, 1.40)

1.19 (1.03, 1.36)

1.0

9
8.4

19
5.7

183
4.5

648
3.7

725
3.3

2.25 (1.14, 4.42)

1.91 (0.96, 3.77)

1.56 (0.98, 2.48)

1.26 (0.80, 2.02)

1.35 (1.14, 1.59)

1.16 (0.98, 1.37)

1.14 (1.02, 1.27)

1.05 (0.95, 1.17)

1.0
1.0

1.95 (0.99, 3.86)

1.27 (0.78, 2.03)

1.15 (0.97, 1.36)

1.05 (0.94, 1.17)

1.0

48
6.1

206
4.6

537
3.8

558
3.5

235
2.8

1.96 (1.42, 2.68)

1.41 (1.02, 1.97)

1.60 (1.33, 1.94)

1.34 (1.10, 1.63)

1.39 (1.19, 162)

1.25 (1.07, 1.46)

1.28 (1.10, 1.49)

1.21 (1.04, 1.41)

1.0
1.0

Trend b

P 0.105

II. Weight-for-height (W=H)

No. of children with febrile cough
Morbidity rates (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season
Model c: b height-for-age Z
III. Weight-for-age (W=A)
No. of children with febrile cough
Morbidity rates (per 1000 child months)
Risk of morbidity, 95% CI, adjusting for:a
Model a: age gender
Model b: a SES vitamin A intake
previous morbidity breast-feeding season

P 0.029

P 0.025

Multivariate models included nutritional status: height-for-age (H=A), weight-for-height (W=H) and weight-for-age (W=A) respectively (four indicator
variables, reference: Z-score of 71 or above), age (0 11, 12 23, 24 35, 36 47 and  48 months), sex and successively variables assessing
socioeconomic status (SES) that is, water piped water in the house, latrine in the house, maternal literacy (all yes=no), household wealth (ordinal with four
levels), region of residence (four indicator variables) and variables accounting for vitamin A intake: periodic large dose supplementation during the study
(vitamin A=placebo) and dietary vitamin A intake (ordinal in quintiles), as well as morbidity in the previous round (three indicator variables), season (ve
indicator variables) and H=A and W=H Z-score (continuous), as further specied in the text.
b
P-values of test for trend from logistic regression models adjusting for the specied covariates and using nutritional status as continuous variable.

(14%, 342%)) and very severe underweight by 96% (95%

CI (42%, 168%), models a).
After adjusting for these and the other covariates, the
association was borderline signicant among the severly
wasted children and remained signicant among the very
severely underweight children, whereupon the risk of
febrile cough increased by 41% (95% CI (2%, 97%))
compared with normally nourished children above 71 Z
(Table 4, II and III, model b). The associations between
severe stunting and severe wasting and febrile cough were
not signicant. Mild and moderate stunting and underweight were similar in their predictive ability, while mild
and moderate wasting did not signicantly predict febrile
cough after adjusting for covariates other than age and
gender. Adjusting for H=A and W=H, as in model c, did
not change the results markedly, and neither did repeating
the analyses using 72 Z as cut-off to dene normally
nourished.
Complicated cough (cough and fever and=or diarrhoea)
was signicantly associated with stunting and underweight
(P for trend < 0.001 for H=A and W=A) and the associations were slightly stronger than those between nutritional
status and febrile cough (data not shown).
Results for the stratied analyses of the relationships between febrile cough and underweight are given in
Table 5.
The stratum-specic risk of cough with fever among
better=normally-nourished children were highest in the
12 23 months age-group, non-breast-fed children under
2 y old, children living in Ri Genoub, and among those
weighed and measured in September and October (Table 5,
column 3). For specic variables, severe underweight
increased the risk of febrile cough among children 48
European Journal of Clinical Nutrition

months or older by 55% (95% CI (18%, 103%)), among

those living in Ri Genoub by 33% (95% CI (1%, 75%)),
and among those weighed and measured in July and August
by 47% (95% CI (8%, 100%)), compared with their cohorts
having a Z-score of  72. Similar results were obtained
for moderate underweight. Modications to the relationship
between febrile cough and H=A and W=H followed similar
patterns to those for W=A. Unlike for diarrhoeal disease,
formal tests for interaction were not signicant for any of
the characteristics studied (Table 5, columns 4 7).

Discussion
Signicant and positive associations were observed
between H=A, W=H and W=A and diarrhoeal disease and
febrile cough. In view of the large study population,
reected in narrow condence intervals, chance is an
unlikely explanation of the results. This study, however,
has some limitations. Despite standardized procedures and
the well-trained staff, measurement and recording errors
could have resulted in misclassication of nutritional status
that would have affected the strengths of associations, but
this was most likely of a non-differential nature with
respect to the outcome.
Misclassication of diarrhoea, cough and fever was
possible because these symptoms are susceptible to misinterpretation. Random misclassication of a binary outcome in effect does not affect the risk estimate in a
prospective study but precision decreases with non-differential under-reporting of cases (Rothman, 1986). Underreporting of illness-related events increases with time after

Undernutrition and childhood infections

J Kossmann et al

469

Table 5 Modication of the relationship between cough with fever and weight for age (W=A)

Characteristic
Age (months)
 48
36 47
24 35
12 23
0 12
Gender
Boys
Girls
Breast-feedingd
Yes
No
Region
1, Abu Dileig
2, Ri Genoub
3, El Jaeli
4, Ri Shamal
5, Gezira
Season
January February
March April
May June
July August
September October
November December

Morbidity rates
among all childrena

Stratum-specic risk Risk of diarrhoeal disease within subgroups of characteristicsc

of morbidity among
Interaction, w2e
children > 72 Z b
<73 Z
73 to 72 Z
>72 Z
(d.f.), P-value

2.9
3.6
4.0
5.2
5.3

1.28
1.38
1.53
1.25

1.0
(1.05,
(1.13,
(1.20,
(0.79,

1.56)
1.42)
1.72)
1.17)
3.32)

1.0
1.0
1.0
1.0
1.0

3.84 (4) 0.43

3.7
3.5

0.91 (0.79, 1.05)

1.0

1.24 (1.01, 1.52)

1.17 (0.95, 1.46)

1.11 (0.94, 1.32)

1.28 (1.08, 1.52)

1.0
1.0

0.39 (1) 0.53

6.2
4.4

0.72 (0.54, 0.96)

1.0

1.09 (0.74, 1.62)

1.43 (0.93, 2.21)

1.03 (0.74, 1.44)

0.91 (0.62, 1.34)

1.0
1.0

1.07 (1) 0.70

3.3
4.3
2.3
3.8
3.7

1.47
0.92
1.15
1.34

1.0
(1.04,
(0.65,
(0.78,
(1.00,

2.08)
1.30)
1.70)
1.81)

1.17
1.33
1.27
1.31
1.05

(0.79,
(1.01,
(0.77,
(0.92,
(0.79,

1.73)
1.75)
2.11)
1.86)
1.40)

1.23
1.36
0.99
1.13
1.15

(0.87,
(1.05,
(0.65,
(0.83,
(0.95,

1.74)
1.76)
1.53)
1.53)
1.40)

1.0
1.0
1.0
1.0
1.0

2.47 (4) 0.35

2.7
2.9
3.1
4.7
4.8
3.2

1.45
1.73
1.71
2.27
1.52

1.0
(1.07,
(1.30,
(1.32,
(1.72,
(1.15,

1.97)
2.31)
2.20)
2.98)
2.00)

0.95
0.90
0.98
1.47
1.25
1.42

(0.57,
(0.49,
(0.68,
(1.08,
(0.94,
(0.98,

1.61)
1.65)
1.42)
2.00)
1.66)
2.05)

0.95
1.04
0.94
1.59
1.19
1.39

(0.65,
(0.70,
(0.69,
(1.21,
(0.92,
(1.04,

1.37)
1.53)
1.26)
2.10)
1.54)
1.86)

1.0
1.0
1.0
1.0
1.0
1.0

10.26 (5) 0.07

1.56)
1.69)
1.94)
1.98)

1.55
1.17
0.95
1.20
0.65

(1.18,
(0.83,
(0.69,
(0.88,
(0.14,

2.03)
1.66)
1.29)
1.62)
2.90)

1.27
1.07
1.33
0.89
1.78

(1.04,
(0.80,
(1.03,
(0.68,
(0.95,

Per 1000 child months.

Multivariate models included weight-for-age (W=A) Z-scores, age, gender, piped water in the house, latrine in the house, maternal literacy, wealth,
region of residence, vitamin A supplementation during the study and dietary vitamin A intake, morbidity in the previous round, breast-feeding, and season
as further specied in the text.
d
The analysis involving breast-feeding was limited to children under the age of 2 y.
e 2
w as the difference between the 72 log likelihood between models with and without respective interaction product term; d.f. degrees of freedom.
b, c

the event and decreases with severity if reporting is not

differentially related to covariates such as education and
poverty (Martorell et al, 1976). Substantial bias is unlikely
in this study although severe morbidity may have been
better recalled. Differential over-reporting of morbidity
among undernourished children and=or selective underreporting among better nourished children could have
over-estimated existing or even created spurious associations. Assuming that lower household wealth and less
maternal education, both signicantly and positively
associated with undernutrition, implied `less awareness'
of morbidity, true associations would have been underrather than over-estimated. Children with selectively missing anthropometric measurements or information on morbidity tended to be from poorer households, also potentially
resulting in an underestimation of the strength of the
association. Follow-up rates were high at all three rounds,
therefore, such bias is unlikely to be substantial.
Anthropometric indices of nutritional status have been
validated as indicators for increased risk of mortality in this
study population (Fawzi et al, 1997) and others (Pelletier,
1994). Wasting, as low W=H, reects recent insults to
nutritional status, while stunting (H=A) reects the cumulative effect of previous undernourishment, except for the
very young children where linear growth is also closely
related to recent insults to nutritional status. Underweight
dened as low weight-for-age reects both recent and
previous insults to nutritional status (Waterlow et al,
1977; Beaton et al, 1990). In prospective studies the
predictive power of nutritional indices for morbidity
decreases with increasing length of the follow-up interval

(Pelletier, et al, 1994); thus, this study may have underestimated true associations.
Child age and gender, socio-economic status, season and
breast-feeding are important confounders of the nutritional
status-mortality relationship (Pelletier, et al, 1994).
El-Samani et al (1988) noted higher incidence rates of
diarrhoea among children who had diarrhoea in the previous child-period, which needs to be taken into account
when looking at the underweight-diarrhoea relationship.
Chowdhury et al (1990) found increased risks of diarrhoea
among moderately and severely stunted and wasted children without diarrhoea in the previous follow-up interval.
In this study diarrhoeal disease was signicantly associated
with previous diarrhoea but not with respiratory symptoms
and neither limiting the analysis to children free of symptoms at baseline nor adjusting for the previous presence of
the specied morbidity changed the risk estimates of the
undernutrition-morbidity relationship markedly.
In this study we adjusted for important confounders like
age, maternal literacy, household wealth, the availability of
piped water and a latrine in the house, region of residence,
periodic vitamin A supplementation during the study,
feeding practice, dietary vitamin A intake, season and
W=H or H=A in the models estimating associations
between morbidity and stunting or wasting, respectively.
Maternal nutrition status and birth weight were not considered potential confounders in this study because their
effects are greatest in the rst 6 months and these data were
not available. Moreover all children enrolled in this study
were older than 6 months. Beyond 6 months of age, linear
and ponderal growth are largely determined by feeding
European Journal of Clinical Nutrition

Undernutrition and childhood infections

J Kossmann et al

470

practices, type of diet, and exposure to pathogens, all of

which are related to socio-economic status. Nevertheless,
residual confounding related to socio-economic status,
previous or concurrent morbidity from other diseases,
through incompleteness and measurement error, genetic
factors, the synergy between nutrition status and morbidity
from infectious or other diseases, and other unknown
factors cannot be excluded.
In this study severe, moderate and even mild stunting,
wasting and underweight increased the risk of subsequent
diarrhoeal disease. These results are in accordance with
those from a study observing elevated risks of diarrhoea
among underweight Mexican children (Sepulveda
et al, 1988) with relative risks ranging from 10% to
110% across W=A categories using deciles and adjusting
for some important confounders. Stunting and wasting
were also associated with the incidence and duration of
diarrhoea in a Guatemalan study and an Ethiopian study
population (Delgado et al, 1983; Lindtjrn et al, 1993).
Studies among Bangladeshi pre-school children found
positive associations between wasting and underweight
and the duration but not the incidence of diarrhoea (Black
et al, 1984; Bairagi et al, 1987) while another found
duration of diarrhoea only to be signicantly associated
with underweight (Chowdhury et al., 1990). The studies
examining associations between undernutrition and infections differed substantially in design and analyses, demographic factors of the study population and adjustments
for covariates. Moreover, very high infectious loads could
reduce the effect of poor nutritional status on the incidence or duration of diarrhoea, which may partly explain
why some studies did not observe any associations
between undernutrition and the incidence or duration of
diarrhoea in India and Bangladesh (Chen et al, 1981;
Mathur et al, 1985).
Our observation of a stronger association between
undernutrition and febrile diarrhoea compared with diarrhoea alone is in accordance with a study showing that
underweight was signicantly associated with both the
duration and the incidence of diarrhoea when the diarrhoea
was accompanied by a very high fever (James, 1972).
Because data on the duration of diarrhoea were not collected, we could not examine the relationships between
nutritional status and the duration of diarrhoeal disease.
In summary, more studies have observed associations
between undernutrition and the duration of diarrhoea than
between undernutrition and the incidence of diarrhoeal
disease. Such associations are biologically plausible
because undernutrition can result in a depressed specic
and unspecic immune system, which increases the susceptibility to disease, requiring a longer time to eliminate
the pathogens, a slower turnover of intestinal cells with
longer recovery of damaged enterocytes, and reduced
intestinal enzyme activity that could prolong diarrhoea
(Sepulveda et al, 1988; Scrimshaw, 1992; Santos, 1994).
Reduced secretory activity associated with prolonged
diarrhoea may be present without severely damaging
the mucosa itself. This and the fact that milder disease
is common in the community may partly explain why
the association between undernutrition and the duration
of diarrhoea has been observed more frequently than for
the incidence of diarrhoeal disease. Also, children with
Z-scores above the conventional cut-off of 72 Z were at
greater risk of diarrhoea, reecting that a statistically
meaningful cut-off does not necessarily reect biological

European Journal of Clinical Nutrition

signicance. Hence, even mild undernutrition should not be

ignored as a risk factor for morbidity.
In the Sudan study non-febrile cough was inversely
associated with the severity of wasting and underweight
while febrile cough was positively associated with underweight and to a lesser extent also with wasting or stunting.
The lower incidence rates of non-febrile cough among very
severely and severely wasted children compared with better
and normally nourished children, might reect the
relationship between undernutrition and impaired immunity. In other words, producing the symptom non-febrile
cough might require at least a partially functioning localized non-specic and specic immune response to infectious agents that may prevent the promotion of more
generalized severe disease, mechanisms which may not
be present among very severely and severely wasted
children. Substantial impairment of mechanical borders
and humoral and cellular immunity is known to be associated with severe undernutrition (Scrimshaw, 1992;
Santos, 1994). The categories `cough' and `cough with
fever' were mutually exclusive in the Sudan study. The
higher incidence of cough alone among better and normally
nourished than among severely undernourished children,
and the higher incidence of febrile cough among the
undernourished than among the better nourished children,
might reect more generalized reactions resulting from the
incapacity to produce the localized cough reaction. In this
study dietary vitamin A intake in contrast to periodic
vitamin A supplementation was signicantly and positively
associated with cough alone and inversely associated with
febrile cough (Fawzi et al, 1995). Vitamin A deciency is
known to adversely effect mechanical barriers and both
humoral and cellular immune functions, similar to the
effects of moderate and severe undernutrition (Scrimshaw,
1992; Fawzi et al, 1995).
The observed signicant positive associations between
underweight, mild and moderate stunting and the nonsignicant positive association between severe wasting
and febrile cough, all after adjusting for covariates, are in
accordance with ndings from other community-based
studies (Tupasi et al, 1988, 1990; Smith et al, 1991; Ballard
et al, 1995; Zaman et al, 1996). Among Filipino preschool-children, being moderately underweight was associated with acute respiratory infections (ARI) and substantially more so with acute lower respiratory infection
(ALRI) than being normally nourished children (Tupasi
et al, 1988, 1990). Undernutrition, particularly stunting and
underweight, was positively associated with the subsequent
risk of ALRI among pre-school-children in Kenya and
Papua New Guinea (Smith et al, 1991; Ballard et al,
1995), as was stunting, wasting and underweight among
pre-school children in Bangladesh (Zaman et al, 1996).
Less protection against pathogens and exposure to contaminated food in environments with poor hygiene during
the weaning period are major contributors to the high
incidences of morbidity, particularly diarrhoea during the
rst 2 y of life (Black et al, 1982). Maternal education and
other socio-economic factors are related to child malnutrition, notably stunting (Bertrand et al, 1988). Our results
from the stratied analyses are in accordance with these
ndings. Different risks of diarrhoeal disease and febrile
cough across the region of residence reect environmental
sanitation and other factors related to access to food.
In this study breast-feeding was observed to be protective against diarrhoeal disease and febrile cough among

Undernutrition and childhood infections

J Kossmann et al

better and normally nourished children and to be a modier

of the relationship between nutritional status and diarrhoeal
disease. Among these Sudanese children breast-feeding
also was a strong modier of the relationship between
undernutrition and all-cause mortality (Fawzi et al, 1997).
These ndings are in accordance with those from a study
examining breast-feeding in relation to mortality from
diarrhoea and respiratory infections among Filipino
children (Yoon et al, 1996). The smaller difference
between the risk of subsequent diarrhoeal disease across
breast-feeding status among the severely underweight compared with the moderately underweight children suggests a
larger protective effect of breast-feeding among moderately
undernourished children. Formal tests for interaction, however, were not signicant. Due to the small number of
children being exclusively breast-fed the category `breastfeeding' in this analysis comprised also children being only
partially breast-fed, which may have attenuated a true
modifying association. Dietary vitamin A intake was not
an important modier in this study but was signicantly
inversely associated with diarrhoea and respiratory infections and mortality, as well as linear and ponderal growth,
in studies examining these relationships in this study
population in detail (Fawzi et al, 1994, 1995, 1997).
Seasonal variations with higher incidences of respiratory
infections during the dry months (Lindtjrn et al, 1993) and
of diarrhoea during the rainy season have been reported
(Sepulveda et al, 1988).
Most of the potential biases identied in this study
would have under rather than over-estimated true associations. The ndings from this study are in accordance with
results from studies similar in design and analyses, and
support the notion of a causal relationship between
moderate and severe and even mild undernutrition and
diarrhoeal disease and respiratory infections. In view of
the synergistic relationship between undernutrition and
infections, preventive interventions to ensure proper child
feeding practices, substantially improve socio-economic
conditions, and provide basic preventive and curative
health services are as necessary among the mildly and
normally nourished children as among the moderately and
severely undernourished children in developing countries.
Acknowledgements We are grateful to the Task Force Sight and Life
(Hoffmann-La Roche) for providing the vitamin A capsules. We would
also like to thank the eld teams, and the mothers and children who made
the study possible. JK was sponsored by German Academic Exchange
Service (DAAD).

References
Bairagi R, Chowdhury MK, Kim YJ, Curlin GT & Gray RH (1987): The
association between malnutrition and diarrhoea in rural Bangladesh.
Int. J. Epidemiol. 16, 477 481.
Ballard TJ, Neumann PH & Neumann C (1995): The effects of malnutrition, parental literacy and household crowding on acute respiratory
infections in young Kenyan children. Trop Pediate 41, 8 13.
Beaton G, Kelly A, Kevaney J, Martorell R & Mason J (1990): Appropriate uses of anthropometric indices in children. ACC=SCN state-of-the
art-series nutrition policy discussion paper no. 7, pp 7 11.
Bertrand WE, Mock NB & Franklin RR (1988): Differential Correlates of
nutritional status in Kinshasa, Zaire. Int. J. Epidemiol. 17, 556 567.
Black RE, Brown KH, Becker S, Abdul Alim ARM & Merson MH (1982):
Contamination of weaning foods and transmission of enterotoxigenic
Eseherischia coli diarrhoea in children in rural Bangladesh. Trans. R.
Soc. Trop. Med. Hyg. 76, 259 264.

Black RE, Brown KH & Becker S (1984): Malnutrition is a determining

factor in diarrhoeal duration, but not incidence, among young children
in a longitudinal study in rural Bangladesh. Am. J. Clin. Nutr. 39,
87 94
Chen LC, Huq E & Huffman SL (1981): A prospective study of the risk of
diarrheal diseases according to the nutritional status of children. Am.
J. Epidemiol. 114, 284 292.
Chowdhury MK, Gupta VM, Baigari R & Bhattacharya BN (1990): Does
malnutrition predispose to diarrhoea during childhood? Evidence from
a longitudinal study in Matlab, Bangladesh. Eur. J. Clin. Nutr. 44,
515 525.
Delgado HI, Valverde V, Belizan JM & Klein RE (1983): Diarrhoeal
disease and nutritional status and health care: analysis for their relationship. Ecol. Food Nutr. 12, 229 234.
El-Samani EZ, Willett WC & Ware JH (1988): Association of malnutrition
and diarrhoea in children aged under ve years a prospective followup study in a rural Sudanese community. Am. J. Epidemiol. 128,
93 105.
Fawzi WW, Herrera MG, Willett WC, Nestel P, El Amin A, Lipsitz S &
Mohamed KA (1994): Dietary vitamin A intake and the risk of
mortality among children. Am. J. Clin. Nutr. 59, 401 408.
Fawzi WW, Herrera MG, Willett WC, Nestel P, El Amin A & Mohamed
KA (1995): Dietary vitamin A intake and the incidence of diarrhoea
and respiratory Infection among Sudanese children. J. Nutr. 125,
1211 1221.
Fawzi WW, Herrera G, Spiegelman DL, El-Amin A, Nestel P & Mohamed
KA (1997): A prospective study of malnutrition in relation to child
mortality in the Sudan. Am. J. Clin. Nutr. 65(4), 1062 1069.
Hamill PVV, Drizd TA, Johnson CL, Reed RB & Roche AF (1977): NCHS
growth curves for children birth 18 years. Hyattsville, USA: National
Center for Health Statistics (DHEW publication, PHS).
Herrera MG, Nestel P, El-Amin A, Fawzi WW, Mohamed KA & Weld L
(1992): Vitamin A supplementation and child survival. Lancet 340,
267 271.
James JW (1972): Longitudinal study of the morbidity of diarrheal and
respiratory infections in malnourished children. Am. J. Clin. Nutr. 25,
690 694.
Lindtjrn B, Alemu T & Bjorvatn B (1993): Nutritional status and risk of
infection among Ethiopian children. J. Trop. Pediat. 39, 76 81.
Martorell R, Habicht JP, Yarbrough C, Lechtig A, Klein RE & Western
KA (1976): Underreporting in Fortnightly Recall Morbidity Surveys.
J. Trop. Pediat. Environ. Child Health June, 129 134.
Mathur R, Reddy V, Naidu AN, Ravikumar & Krishnamachari KAVR
(1985): Nutritional status and diarrhoeal morbidity: a longitudinal
study in rural Indian preschool children. Human Nutr. Clin. Nutr.
39C, 447 454.
Pelletier DL, Low JW, Johnson FC & Msukwa LAH (1994): Child
anthropometry and mortality in Malawi: testing for effect modication
by age and length of follow-up and confounding by socio-economic
factors. J. Nutr. 124, 2082S 2105S.
Pelletier DL (1994b): The relationship between child anthropometry and
mortality in developing countries: implications for policy, programs and
future research. J. Nutr. 124, 2047S 2081S.
Rivera J & Martorell R (1988): Nutrition, infection, and growth part II:
effects of malnutrition on infection and general conclusions. Clin. Nutr.
7, 163 167.
Rothman KJ (1986): Modern Epidemiology, pp 87 88. Boston, MA: Little
Brown.
Santos JI (1994): Nutrition, infection, and immunocompetence. Infect. Dis.
Clin. N. Am. 8, 243 267.
Scrimshaw NS (1992): Effect of infection on nutritional status. Proc. Nat.
Sci. Council 16, 46 64.
Sepulveda J, Willett W & Munoz A (1988): Malnutrition and diarrhoea, a
longitudinal study among urban Mexican children. Am. J. Epidemiol.
127, 365 367.
Smith TA, Lehmann D, Coackley C, Spooner V & Alpers MP (1991):
Relationships between growth and acute lower respiratory infections in
children aged less than 5 years in a highland population of Papua-New
Guinea. Am. J. Clin. Nutr. 53, 963 970.
Tomkins A (1981): Nutritional status and severity of diarrhoea among preschool children in rural Nigeria (1981). Lancet 1, 860 862.
Trowbridge FL, Newton LH & Campbell CC (1981): Letter to the editor.
Lancet 1, 1375.
Tulloch J & Richards L (1993): Childhood diarrhoea and acute respiratory
infections in developing countries. Med. J. Austral. 159, 46 51.
Tupasi TE, Velmonte MA Sanvictores MEG, Abraham L, de Leon LE,
Miguel CA & Saniel MC (1988): Determinants of morbidity and
mortality due to respiratory infections: implications for interventions.
J. Infect. Dis. 157, 615 623.

471

European Journal of Clinical Nutrition

Undernutrition and childhood infections

J Kossmann et al

472

Tupasi TE, Mangubat NV, Sunico MES, Magdangal DM, Navarro EE,
Leonor ZA, Lupisan S, Medalla F & Lucero MG (1990): Malnutrition
and acute respiratory tract infections in Filipino children. Rey. Infect.
Dis. 12, 1047 1054.
Waterlow JC, Buzina R, Keller W, Lane JM, Nichmann MZ & Tanner JM
(1977): The presentation and use of height and weight data for
comparing the nutritional status of groups of children under the age
of 10 years. Bull. WHO 55, 489 498.

European Journal of Clinical Nutrition

Yoon PW, Black RE, Moulton LH & Becker S (1996): Effect of not
breast-feeding on the risk of diarrheal and respiratory mortality in
children under 2 years of age in Metro Cebu, The Philippines. Am. J.
Epidemiol. 143, 1142 1148.
Zaman K, Baqui AH, Yunus MD, Sack RB, Bateman OM Chowdury HR
& Black RE (1996): Association between nutritional status, cellmediated immune status and acute lower respiratory infections in
Bangladeshi children. Eur. J. Clin. Nutr. 50, 309 331.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.