Mortality Rates, Prevalence of Malnutrition, and Prevalence of Lost Pregnancies among the Drought-Ravaged Population of Tete Province, Mozambique

ORIGINAL RESEARCH
Mortality Rates, Prevalence of Malnutrition,
and Prevalence of Lost Pregnancies among
the Drought-Ravaged Population of Tete
Province, Mozambique
Dr. Andre M.N. Renzaho, PhD, MPH
Abstract
Program Quality Advisor, World Vision
Background: Tete Province, Mozambique has experienced chronic food inse-
Australia and Honorary Fellow, School of
curity and a dramatic fall in livestock numbers due to the cyclic problems
Health and Social Development, Deakin
characterized by the floods in 2000 and severe droughts in 2002 and 2003.
University, Melbourne, Australia
The Province has been a beneficiary of emergency relief programs, which
have assisted >22% of the population. However, these programs were not
Correspondence:
based on sound epidemiological data, and they have not established baseline
Dr Andre M.N. Renzaho, PhD, MPH
data against which to assess the impact of the programs.
Senior Research Fellow
Objective: The objective of this study was to document mortality rates, caus-
School of Health and Social Development
es of death, the prevalence of malnutrition, and the prevalence of lost preg-
Faculty of Health, Medicine, Nursing and
nancies after 2.5 years of humanitarian response to the crisis.
Behavioural Sciences
Methods: A two-stage, 30-cluster household survey was conducted in the
Deakin University
Cahora Bassa and Changara districts from 22 October to 08 November 2004.
221 Burwood Highway
A total of 838 households were surveyed, with a population size of 4,688 people.
Burwood 3125
Results: Anthropometric data were collected among children 6–59 months of
Victoria, Australia
age. In addition, crude mortality rates (CMRs), under five mortality rates
E-mail: andre.renzaho@deakin.edu.au
(U5MRs), causes of deaths, and prevalence of lost pregnancies were deter-
mined among the sample population. The prevalence of malnutrition was
Funding
8.0% (95% confidence interval (CI) = 6.2–9.8%) for acute malnutrition,
This study was fully funded by World Vision
26.9% (95% CI = 24.0–29.9%) for being underweight, and 37.0% (95% CI =
Australia, and there was no external funding
33.8–40.2%) for chronic malnutrition. Boys were more likely to be under-
source.
weight than were girls (odds ratio (OR) = 1.34; 95% CI = 1.00, 1.82; p <0.05)
after controlling for age, household size, and food aid beneficiary status.
Keywords: human immunodeficiency
Similarly, children 30–59 months of age were significantly less likely to suffer
virus/acquired immunodeficiency syndrome
from acute malnutrition (OR = 0.45; 95% CI = 0.26, 0.79; p <0.01) and less
(HIV/AIDS); lost pregnancies; malnutrition;
likely to be underweight (OR = 0.37; 95% CI = 0.27, 0.51; p <0.01) than chil-
mortality; Mozambique; Tete Province
dren 6–29 months of age, after adjusting for the other, aforementioned fac-
tors. The proportion of lost pregnancies was estimated at 7.7% (95% CI =
Abbreviations:
4.5–11.0%). A total of 215 deaths were reported during the year preceding the
AIDS = acquired immunodeficiency
survey. Thirty-nine (18.1%) children <5 years of age died. The CMR was
syndrome
1.23/10,000/day (95% CI = 1.08–1.38), and an U5MR was 1.03/10,000/day
CMR = crude mortality rate
(95% CI = 0.71–1.35). Diarrheal diseases, malaria, tuberculosis, and human
CSB = Corn and Soya Blend
immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS)
H/A = height-for-age
accounted for more than two-thirds of all deaths.
HIV = human immunodeficiency virus
Conclusions: The observed CMR in Tete Province, Mozambique is three times
MCH = Maternal and Child Health
higher than the baseline rate for sub-Saharan Africa and 1.4 times higher than
NGO = non-governmental organization
the CMR cut-off point used to define excess mortality in emergencies.The cur-
TB = tuberculosis
rent humanitarian response in Tete Province would benefit from an improved
U5MR = under 5 mortality rate
alignment of food aid programming in conjunction with diarrheal disease con-
W/A= weight-for-age
trol, HIV/AIDS, and malaria prevention and treatment programs. The impact
W/H = weight-for-height
of the food programs would be improved if mutually acceptable food aid pro-
WFP = World Food Programme
gram objectives, verifiable indicators relevant to each objective, and beneficiary
targets and selection criteria are developed. Periodic re-assessments and evalu-
Received: 01 June 2006
ations of the impact of the program and evidenced-based decision-making
Accepted: 10 July 2006
urgently are needed to avert a chronic dependency on food aid.
Revised: 14 July 2006
Renzaho AMN: Mortality rates, prevalence of malnutrition, and prevalence
Web publication: 16 February 2006
of lost pregnancies among the drought-ravaged population of Tete Province,
Mozambique. Prehosp Disast Med 2007;22(1):26–34.
Prehospital and Disaster Medicine
http://pdm.medicine.wisc.edu
Vol. 22, No. 1

---

Renzaho
27
Introduction
During the last three decades, Mozambique has been the
scene of multiple disasters. In 1992, following 20 years of
civil war, Mozambique experienced a number of disasters
caused by naturally occurring hazards. In May 1996, the
Food and Agriculture Organization/World Food
Programme (WFP) Crop and Food Supply Assessment
Mission to Mozambique found that cyclone Bonita result-
ed in exceptionally heavy rains, that caused widespread
flooding, destroyed 44,200 hectares of crops, and affected
83,528 people. Similarly, in 1994, cyclone Nadia left thou-
sands of people on Mozambique’s northern coast homeless
and hungry for an extended period of time.1 Cyclone Eline
in February 2000 and cyclone Hudah in April 2000 left
450,000 people homeless, 160,000 displaced, and 500
dead.2 The year 2002 was characterized by sporadic and
insufficient rains in southern Mozambique. Several years of
flooding and droughts combined with a loss of productivi-
Renzaho © 2007 Prehospital and Disaster Medicine
ty due to the impact of human immunodeficiency
Figure 1—Tete Province showing the study area
virus/acquired immunodeficiency syndrome (HIV/AIDS),
have left the community vulnerable to food insecurity and
dren >5 years of age, and pregnant and lactating
malnutrition.3 By 2003, it was estimated that people in 43
mothers. From February 2003 to June 2004, the
of the 128 (33%) districts in the country suffered from food
average quantity of CSB distributed per eligible ben-
shortages, and the number of people needing food assis-
eficiary was 5.5 kg.
tance escalated between 2002 and 2003.2
3. A home-based feeding program for the chronically
Large, semi-arid areas and cyclic problems associated
ill (mainly for people living with HIV/AIDS and
with food insecurity have severely affected the population of
tuberculosis)—By 01 October 2004, the average
Tete Province.3 Anthropometric data indicate that in 2003,
quantity of food distributed per eligible beneficiary
the prevalence of acute malnutrition was 9.9% for Tete
was 25.8 kg of maize, 3.1 kg of sorghum, and 8.2 kg
Province (Figure 1), 3.4% for Manica, 3.6% for Sofala, 2.9%
of CSB.
for Inhambane, and 2.4% for Gaza, averaging 6.4% nation-
4. The Vulnerable Group Feeding Program—By 01
ally (Figure 2).4 Since 2000, the south and central regions of
October 2004, the average quantity of food distrib-
Tete Province have experienced irregular and insufficient
uted per eligible beneficiary was 8.6 kg of maize, 3.8
rains. Maize yields were reduced greatly, and the following
kg of sorghum, 0.4 kg of beans, 0.7 kg of rice, and 0.1
season for grains and vegetables was affected adversely.
kg of oil. This program is intended for orphans and
Although sorghum and cassava are more resistant to
vulnerable children, child- or female-headed house-
drought conditions, an overall reduction in productivity
holds, the elderly, and the disabled.
from plantations due to personnel with HIV/AIDS5 has
Despite such intensive and comprehensive interventions,
resulted in at least four years of a meager harvest in Tete. In
these programs have not been based on sound epidemiologi-
September 2002, World Vision and the WFP implemented
cal data, nor have they established baseline data against which
nutritional support programs aimed to reduce mortality and
to assess the impact of the programs.The purpose of this study
malnutrition among the most severely affected districts of
was to document the mortality rates, causes of death, preva-
Tete Province: the Cahora Bassa, Changara, Moatise, and
lence of malnutrition, and prevalence of lost pregnancies after
Mutarara districts. These programs included: (1) the Food-
2.5 years of ongoing humanitarian response to the crisis.
for-Work Program; (2) the Corn and Soya Blend (CSB)
Supplementary Food Program; (3) a home-based feeding
Methods
program; and (4) the Vulnerable Group Feeding Program.
Study Design
1. The Food-for-Work Program—From October 2002
The study design was cross-sectional, and based on a two-
to July 2004, the average quantity of food distributed
stage, 30-cluster household survey in the Cahora Bassa and
per eligible beneficiary was 10.5 kg of maize, 4 kg of
Changara districts. When the relief programs were imple-
sorghum, 0.9 kg of beans, 2.7 kg of rice, and 0.3 kg
mented, population-based data on mortality or nutritional
of oil. The work associated with this scheme resulted
status were available, and existing data had been collected
in the construction of >47 schools, 12 small dams,
on an ad hoc basis. Available data on malnutrition indicat-
three houses for nurses, two maternity houses, and
ed that the prevalence of acute malnutrition in 2003 was
five houses for teachers. Roads were rehabilitated,
9.9% in Tete Province.4 However, given that the Cahora
and >17 farmers’ associations were established to
Bassa and Changara districts were the most severely affect-
promote the multiplication of sweet potato and cassava
ed, in this study, it was assumed that the prevalence of acute
stakes, vegetable production, and fruit tree plantations;
malnutrition in those districts would be higher than the
2. The Corn and Soya Blend (CSB) Supplementary
average of the Province. Due to a lack of data on malnutri-
Food Program—This program is intended for chil-
tion in these districts, a prevalence of acute malnutrition
January–February 2007
http://pdm.medicine.wisc.edu
Prehospital and Disaster Medicine

---

28
Mortality Rates, Prevalence of Malnutrition
Prior to data collection, the heads of households or
guardians were told that participation in the study was vol-
untary, that they were free to withdraw their participation
at any stage during the study, and that the data would be
presented as aggregate. They were assured that all data pro-
vided would be treated with strict confidentiality. All
selected households consented to participate in the study.
Trained enumerators administered the questionnaire. In
the event that data collectors reached the border of the vil-
lage before the required number of children for a cluster
was reached, they returned to the center of the village and
repeated Steps 1 and 2. In Step 2, if household members
>
_15 years of age were absent at the time of the interview,
neighbors were asked to assist in locating the members. If
the occupants could not be found, the data from the house-
hold was eliminated and replaced with data from the next
closest household; this only occurred twice.
Renzaho © 2007 Prehospital and Disaster Medicine
Data collectors who spoke both the local language (main-
Figure 2—Death rates (10,000/day) and their 95% CI
ly Barwa, but also referred to as Balke or Cibalke) and
reported in Cahora Bassa and Changara districts over the
English, administered the questionnaire after receiving 3.5
recall period
days of training on sampling methodology, data collection,
anthropometric measurements, and data recording. The
was estimated as 20% for the sample size calculation so that
questionnaire was translated into the local language, trans-
the sample size was not under-estimated. In calculating the
lated back into English by a different person to validate the
sample size, a design effect of two and a precision of 4% were
accuracy of translation, and field-tested prior to official data
assumed.6–8 A sample size of 768 children was obtained.This
collection. Upon completion of each cluster, the group leader
sample size was adjusted to account for contingencies such as
checked all questionnaires for completeness and accuracy.
non-response or recording errors, and the corresponding fig-
ure was rounded to 820 children,6,8 or 4,713 people, assum-
Measurements
ing a proportion of 17.4% children <5 years of age.9,10 With
Weight was measured by Salter-type spring hanging scales,
an average household size estimated at 5.6 persons,9,10 the
with a capacity of 25 kg and 100 gram increments (DK-
total number of households that needed to be visited was esti-
2100 Copenhagen, Denmark). Height was measured to the
mated at 842. This sample was more than adequate to deter-
nearest millimeter using a measuring board (Shorr
mine the crude mortality rate (CMR), when a CMR of 1.5
Productions, Maryland). Three types of malnutrition were
deaths/10,000 people/day and a recall period of 373 days,
considered in this study: (1) underweight, measured by
with a 95% CI of ±0.5/10,000/day, and a design effect of two
weight-for-age (W/A); (2) chronic malnutrition, measured
were assumed (calculated n = 3,753 persons).
by height-for-age (H/A); and (3) acute malnutrition, mea-
sured by weight-for-height (W/H). The Z-scores were
Procedure
used as indicators of the nutritional status of children: acute
During the first stage, the smallest population unit in each
malnutrition was defined as a W/H <-2 Z-score; chronic mal-
district (village) was determined. The total number of villages
nutrition was defined as a H/A <-2 Z-score; and underweight
that compose the Cahora Basa and Changara districts, as well
was defined as a W/A <-2 Z-score.
as the population size of each village, were determined using
The computation of the mid-point population assumed
the 2003 Mozambique Demographic and Health Survey.9,10
that deaths and births occurred at a constant rate—half of
Thirty clusters were assigned randomly in proportion to the
deaths and half of births did not occur by the midpoint of
population of the village. During the second stage, house-
the recall period.12,13 This is an accurate estimation, given
holds were selected using the standard Expanded Program on
that the population movement had remained stable during
Immunization methods.11 Data collectors went to the center
the five years preceding the evaluation. After the war ended
of the village and chose a direction randomly (e.g., by spin-
in 1992, the majority of Mozambicans returned from
ning a bottle) (Step 1). They counted the total number of
neighboring countries, especially Zimbabwe, after years in
households (t) in the chosen direction from the center to the
refugee camps,5 and were attracted by the coal mines at
edge of the village. The first number surveyed was selected
Moatize. By the late 1990s, the population movement was
randomly by choosing a number between 1 and t using a table
stable. Thus, three questions were asked: (1) the number of
of random numbers. After the first household was identified,
people that slept in the house the night preceding the sur-
the next closest household was selected and the process con-
vey; (2) the number of pregnancies and their outcomes dur-
tinued until the number of children required per cluster were
ing the recall period; and (3) the number of people living in
questioned (820/30 = 27; Step 2). A household was defined as
the household who died during the recall period. For each
an aggregate of persons who either live together under the
identified death, respondents were asked to identify the
same roof or in different units in the same compound, but eat
cause of death or to list three symptoms the victim experi-
together or share the household food.
enced before their death.
Prehospital and Disaster Medicine
http://pdm.medicine.wisc.edu
Vol. 22, No. 1

---

Renzaho
29
The proportion of lost pregnancies was calculated using the
CI = 0.26, 0.79; p <0.01) and to be underweight (OR = 0.37;
following formula:14
95% CI = 0.27, 0.51; p <0.01) than children aged 6–29
% lost pregnancies =
months after adjusting for the other factors.
(abortion + stillbirths)/(abortion + stillbirths + live births)
Prevalence of Lost Pregnancies
Mortality rates were computed as follows:12,13
A total of 339 pregnancies were reported during the recall
period. Of these, 80 (23.6%) women still were pregnant,
Number of deaths
CMR =
x 10,000
239 (70.1%) resulted in live births, and 20 (5.9%) resulted
Recall period in days x mid-point population
stillbirths, miscarriages, or abortions, corresponding to a the
Where: mid-point population = number of living in the
proportion of 7.7% (95% CI = 4.5–11.0%) lost pregnancies.
sample + 1/2 deaths in the sample – 1/2 live births in the
sample.
Mortality
Number of deaths among
A total of 215 deaths were reported during the recall peri-
those <5 years of age
od. Of these deaths, 39 (18.1%) victims were children <5
U5MR =
Number of living <5 years of age + 1/2
years of age. These deaths resulted in a CMR of
deaths among those <5 years of age x recall period
1.23/10,000/day (95% CI = 1.08–1.38) and an U5MR of
1.03/10,000/day (95% CI = 0.71–1.35).The CMR and U5MR
Data Processing
did not vary significantly by sex (Table 2). The number of
Data were entered into SPSS for Windows, version 12.0
deaths was higher in the last quarter of 2003, then
(SPSS Inc., Chicago, IL) and processed using Stata version
decreased up to October 2004.
7.0 (Stata Corporation, College Station, TX). The propor-
tion of malnutrition and its 95% confidence interval (CI)
Causes of Death
was computed while mortality data were expressed per
Diarrheal diseases, malaria, and TB accounted for over half
10,000/day and stratified by age and gender. In Stata, the
(61.1%) of all of the deaths (Figures 3a and 3b). Causes of
“svyset command” was used to specify clustering within the
death did not vary by gender, but varied considerably by age
household, stratification, and weight prior to analysis.
group (p <0.01). For children <5 years of age, the top five
When the outcome was binary, logistic regression was
main causes of deaths were: (1) diarrheal diseases; (2)
used. The relationship between the two categorical vari-
malaria; (3) TB; (4) HIV/AIDS; and (5) traffic crashes. In
ables was examined using the chi-square test. The level of
contrast, the top five main causes of death for persons >
_5 years
statistical significance was set at a probability of p <0.05 for
of age were: (1) diarrheal diseases; (2) malaria; (3) TB; (4)
all tests.
anemia; and (5) hypertension.
Results
Discussion
Demographics and Children Anthropometric Measurements
Mortality
A total of 838 households were surveyed, resulting in a
This study is the first to explore health and nutrition out-
total population size of 4,688 people. The average house-
comes in the Cahora Bassa and Changara districts of the
hold size was 5.53 persons (95% CI = 5.48, 5.58; minimum
Tete Province. The reported CMR is 1.5 times higher than
= 1; maximum = 14). One in three (37.1%) households sur-
the 2003 national average, and 2.5 times higher than the
veyed had >5 family members living in the house. More
2003 rate for Tete Province.4 It also is >3 times higher than
than half (53.7%) of the sample was male. The median age
the baseline for sub-Saharan Africa15 and 1.4 times higher
was 14 years (range = 0–86 years). One in five people
than the CMR cut-off used by the Centers for Disease
(21.2%) were children <5 years of age. Anthropometric
Control and Prevention (CDC) to define excess mortality in
data were obtained for 874 children (425 girls and 449
emergencies.16 Using the 2000 CMR of 0.60/10,000/day17
boys) aged 6–59 months, representing 18.6% of the total
as the pre-drought baseline and total population of 177,226
population. The calculated Z-score for boys and girls aver-
in Cahora Bassa and Changara districts,18 it was estimated
aged –1.5 (95% CI = –1.7, –1.2) and –1.3 (95% CI = –1.5,
that about 16,301 excess deaths occurred in both of the dis-
–1.0) for H/A, 0.1 (95% CI = –0.1, 0.2) and 0.3 (95% CI =
tricts between 2000 and 2004. Nevertheless, the U5MR was
0.1, 0.4) for W/H, and –1.1 (95% CI = –1.3, –0.9) and for
41% lower than the national average.4 The U5MR also was
W/A –0.9 (95% CI = –1.1, –0.7) respectively.
lower than the 1.14/10,000/day baseline for sub-Saharan
Africa and the 2.23/10,000/day used to define excessive
Malnutrition
mortality in persons <5 years of age during emergencies.15
The prevalence of malnutrition (acute malnutrition,
Although data have indicated that children <5 years of
underweight, and chronic malnutrition) is summarized in
age account for the majority of deaths during emergen-
Table 1. The prevalence of acute and chronic malnutrition
cies,19,20 the opposite was found in the current study with
did not vary by sex, but boys were statistically significantly
increased mortality in children >5 years of age. Such a pat-
more likely to be underweight than were girls (OR = 1.34; 95%
tern is consistent with the situation that has characterized
CI = 1.00, 1.82; p <0.05) after controlling for age, household
Tete Province over the last four years. In partnership with
size, and food aid beneficiary status. Similarly, children
World Vision and the Red Cross, the WFP has been pro-
30–59 months of age were statistically significantly less
viding feeding programs for children <5 years of age as well
likely to suffer from acute malnutrition (OR = 0.45; 95%
as lactating and pregnant mothers. These programs have
January–February 2007
http://pdm.medicine.wisc.edu
Prehospital and Disaster Medicine

---

30
Mortality Rates, Prevalence of Malnutrition
W/H Z-Score
W/A Z-Score
H/A Z-Score
%
Adjusted OR
%
Adjusted OR
%
Adjusted OR
n
(95% CI)a
(95% CI)#
(95% CI)b
(95% CI)#
(95% CI)c
(95% CI)#
All
874
8.0%
NA
26.9%
NA
37.0
NA
(6.2%, 9.8%)
(24.0%, 29.9%)
(33.8, 40.2)
Sex
Girls
425
7.6
Ref
23.8
Ref
35.1
Ref
(5.0, 10.1)
(19.7, 27.8)
(30.5, 39.6)
Boys
449
8.5
1.14
30.0
1.34
38.8
1.19
(5.9, 11.0)
(0.68, 1.92)
(25.7, 34.2)
(1.00, 1.82)* (34.4, 43.5)
(0.90, 1.58)
Age Group
6–29 months
438
9.8
Ref
36.2
Ref
40.4
Ref
(7.0, 12.6))
(31.7, 40.8)
(35.8, 45.0)
30–59 months
436
4.7
0.45
17.5
0.37
34.3
0.76
(2.7, 6.7)** (0.26, 0.79)** (13.9, 21.1)*** (0.27, 0.51)*** (29.8, 38.8)
(0.58, 1.00)
Beneficiary Status
Exbeneficiary of
food aid
298
9.4
Ref
27.9
Ref
34.7
Ref
(6.1, 12.7)
(22.8, 33.1)
(29.3, 40.1)
Current
beneficiary of
120
6.7
0.80
30
1.21
44.2
1.57
food aida
(2.2, 11.2)
(0.35, 1.85)
(21.8, 38.2)
(0.75, 1.96)
(35.2, 53.1) (1.01, 2.43)*
Never been
beneficiary of
456
7.5
0.73
25.5
0.89
36.7
1.11
food aidb
(5.1, 9.9)
(0.42, 1.27)
(21.5, 29.5)
(0.63, 1.25)
(32.3, 41.1)
(0.82, 1.51)
Household Size
?5 people
400
8.5
Ref
27.2
Ref
38.9
Ref
(5.8, 11.2)
(22.8, 31.5)
(34.1, 43.7)
?6 people
474
7.6
0.85
26.8
0.94
35.5
0.84
(5.2, 10.0)
(0.50, 1.42)
(22.7, 30.8)
(0.69, 1.28)
(31.1, 39.8)
(0.63, 1.11)
Renzaho © 2007 Prehospital and Disaster Medicine
Table 1—Prevalence of chronic malnutrition, underweight, and acute malnutrition and adjusted odd ratios (OR) as a
function of demographic variables (H/A = height-for-age; Ref = reference; W/A = weight-for-age; W/H = weight-for-
height)
# Model adjusted for factors in the table; a Prevalence of Acute malnutrition (WHZ <-2); b Prevalence of underweight
(WAZ <-2) c Prevalence of chronic malnutrition (HAZ<-2); *p <0.05; **p <0.01; ***p <0.001
aBeneficiary at the time of the survey; bDischarged from the program
Prehospital and Disaster Medicine
http://pdm.medicine.wisc.edu
Vol. 22, No. 1

---

Renzaho
31
Male
Female
Total
CMR
Number of people living in
the sample
2,519
2,169
4,688
Number of deaths during
recall period
105
110
215
Number of live-births
during recall period
192
47
239
Mid-point population
2,476
2,201
4,676
Deaths/10,000/day
1.14
1.34
1.23
(95%CI) (0.93, 1.35)
(1.10, 1.58)
(1.08, 1.38)
U5MR
Number of deaths during
recall period
18
21
39
Total <5 years
509
486
995
Mid-point population
518
497
1,015
Deaths/10,000/day
0.93
1.13
1.03
(95% CI) (0.51, 1.35)
(0.66, 1.61)
(0.71, 1.35)
Renzaho © 2007 Prehospital and Disaster Medicine
Table 2—Crude (CMR) and under five mortality rate (U5MR) and (95%) confidence interval by gender
included supplementary feeding programs providing free
they would have identified one of their members as having
CSB, school lunches, and vulnerable group feeding, target-
died from HIV/AIDS; it also may be due to the lack of
ing orphans, the elderly, and child- and female-headed
HIV testing. Thus, many people may be dying from
households. Therefore, findings of a CMR above the emer-
HIV/AIDS before being aware of their HIV status. Further
gency threshold and the U5MR below the threshold for
studies to explore the impact of HIV/AIDS are required.
this sub-group, suggest that there is an urgent need to
In this study, the proportion of lost pregnancies is simi-
improve and target life-saving interventions, specifically
lar to that reported in countries experiencing chronic emer-
among the adult population.
gencies.14 Since 3.2% of deaths are due to pregnancy termi-
While various studies have identified malnutrition to be
nation/abortion/miscarriage (Figure 3a), it is possible that
a major cause of death during emergencies,16,19,21–24 this
the poverty and hunger-induced sex trade is used as a sur-
was not the case in the current study. However, there is no
vival strategy. This social and public health issue leads to
doubt that the consequences of the drought on the people
unwanted or unplanned pregnancies. However, other fac-
and the economy have been far-reaching. The drought has
tors have been shown to increase the risk of miscarriage or
limited access to water and sanitation facilities, has impact-
stillbirth. These include the age of the mother, pregnancy
ed the economy negatively, has reduced household pur-
order, pregnancy history, and the mother’s social character-
chasing power, and has restricted access to health care. The
istics and environment.28 In addition, the current literature
findings that diarrheal diseases and malaria as major caus-
suggests that an environment characterized by famine and
es of death are consistent with the current litera-
malnutrition significantly increases the risk of miscarriage
ture.16,19,25–27 Despite the high prevalence of HIV/AIDS,3
and stillbirths,28,29 so the effect of famine and starvation
the proportion of deaths caused by HIV/AIDS was esti-
cannot be ruled out in explaining the observed prevalence
mated at 6.3% in adults and 9.2% in children <5 years of
of lost pregnancies. Further studies are required to elucidate
age, placing this cause fifth and fourth, respectively, after
this theory.
diarrheal diseases, malaria, and TB. It is possible that TB,
acute respiratory diseases, and the non-specificity of “diar-
Malnutrition
rheal diseases” are likely to be linked to HIV/AIDS infection,
By applying the cut-off points for assessing the severity of
and that the mortality due to HIV/AIDS was underestimat-
malnutrition of the World Health Organization,30 the
ed. This may have been exacerbated by the stigma, isola-
observed malnutrition prevalence could be said to be medi-
tion, and discrimination the family would experience if
um for acute malnutrition, but very high for underweight
January–February 2007
http://pdm.medicine.wisc.edu
Prehospital and Disaster Medicine

---

32
Mortality Rates, Prevalence of Malnutrition
Renzaho © 2007 Prehospital and Disaster Medicine
Renzaho © 2007 Prehospital and Disaster Medicine
Figure 3a—Proportionate casuse-specific mortality for all
Figure 3b—Proportionate cause-specific mortality for >5
ages
years of age (TB = tuberculosis)
and chronic malnutrition. However, a long time period is
exports as well as the expansion of cassava cultivation. This
needed to observe changes in chronic malnutrition and
could explain why the number of deaths was very high dur-
underweight while humanitarian responses are geared
ing the last quarter of 2003, but fell in a stepwise fashion in
toward addressing acute malnutrition. Thus, this discussion
2004. However, given the long recall period, it also is pos-
will focus on acute malnutrition.
sible that a recall bias might have been more evident for
The prevalence of acute malnutrition in Cahora Bassa
deaths occurring at the beginning of the October 2003 to
and Changara remains similar to the average for all of Tete
January 2004 quarter.
Province, but higher than the prevalence reported in the
Third, food distributed to eligible households was cal-
neighboring provinces. The prevalence of acute malnutri-
culated on the basis of an average of five persons per house-
tion for this Province has remained static since 2002, esti-
hold, but 37.1% of the surveyed households had >5 family
mated at 7.8% in 20024 versus 8.0% reported in the current
members living in the house. Two implications are associ-
study. Many factors could explain why the prevalence of
ated with this criterion: (1) more than one-third of benefi-
acute malnutrition in the Cahora Bassa and Changara dis-
ciary households received an inadequate ration; and (2) the
tricts of Tete Province has not declined, despite the inten-
high likelihood that beneficiary households shared their
sive humanitarian responses to the crisis. First, there were
often inadequate and irregular food rations with non-ben-
frequent delays in food delivery that were attributable
eficiary families and neighbors. Despite the initial reduc-
mainly to the ineffectiveness or irregularity of the food
tion in food for the beneficiary families, the act of sharing
pipeline; this resulted in the reduction of food entitlements.
food builds reciprocity of support among a community in
The situation was exacerbated by poor targeting strategies
times of stress,5 and this practice can result in significant,
that led to chronic mismatching between the actual and
short-term nutritional security gain for beneficiary house-
proposed beneficiaries, which resulted in a less than ade-
holds. That is, when these families are confronted with a
quate coverage rate (Table 3).
food pipeline problem, neighbors return the favor, and
Second, use of the World Vision Commodity Tracking
share their food. Thus, the impact of food sharing and rec-
System indicated that a large number of beneficiaries did
iprocity should be included in the indicators of any evalu-
not receive their entitlements and actually received less
ation of the impact of food aid programs in Tete Province.
maize than they were entitled to receive. This was more
The pattern observed could help explain why no difference
pronounced between October 2002 and September 2003.
was found in the prevalence of malnutrition between chil-
In contrast, from October 2003 to September 2004 (a peri-
dren from small versus large families.
od within the recall period of the current study), the num-
Finally, the prevalence of HIV/AIDS and the number
ber of actual beneficiaries increased in linear fashion and
of orphans in the various communities in Tete Province
they received a ration that provided more energy than they
remains high.3 Those with symptoms of AIDS may fail to
were entitled to receive due to back-payment of rations that
gain weight because of opportunistic infections.31 Food aid
were overdue. In addition, the agricultural outlook for 2004
programs in Tete Province have not been proactive in
was encouraging, with cereal production increasing by 11%
designing strategies geared toward mitigating the effect of
from 2003, and an increase in formal and informal maize
complications due to HIV/AIDS.
Prehospital and Disaster Medicine
http://pdm.medicine.wisc.edu
Vol. 22, No. 1

---

Renzaho
33
Phase Ia
Phase IIb
Phase IIIc
Phase IVd
Oct 02-Mar 03
Apr 03-Sept 03
Oct 03-Mar 04
Apr 04-Sept 04
Total number of proposed
beneficiaries
316,800
484,800
484,800
258,000
Energy profile of pledged ration
Energy
(Kcal/beneficiary/day)
2,139
12,139
2,139
1,754
% of energy from protein
11.4
11.4
11.2
11.2
% of energy from fat
19.3
19.3
19.3
10.4
Total number of actual beneficiaries
196,162
470,356
376,420
465,644
Energy profile of distributed ration
Energy
1,871
1,925
2,408
1,871
% of energy from protein
12.8
9.9
12.6
13.3
% of energy from fat
10.0
9.5
11.0
9.5
Coverage (%)
61.9
97.0
77.6
180.5
Fulfillment of energy pledges (%)
87.5
90.0
112.6
106.7
Renzaho © 2007 Prehospital and Disaster Medicine
Table 3—Number of proposed food program beneficiaries vs. actual beneficiaries and quantity of food beneficiaries
received (g/beneficiary/day) vs. pledged ration by period offeraPledged ration (g/beneficiary/day): maize: 500 g, beans: 50
g, oil: 25 g. Received (g/beneficiary/day): maize: 473 g, beans: 40 g, corn-soya blend (CSB): 22 g. bPledged ration
(g/beneficiary day): maize: 500 g, beans: 50 g, oil: 25 g. Received (g/beneficiary/day): maize: 135 g, beans: 21 g, CSB: 41
g, oil: 11 g, rice 314 g. cPledged ration (g/beneficiary day): maize: 500 g, beans: 20 g, oil: 25 g. Received
(g/beneficiary/day): maize: 376 g, beans: 20 g, CSB: 108 g, oil: 5 g, rice: 90 g, sorghum: 267 g. dPledged ration (g/benefi-
ciary day): maize: 1,200 g, beans: 200 g, CSB: 600 g, oil: 100 g. Received (g/beneficiary/day): maize: 214 g, beans: 10 g;
CSB: 51 g, oil: 5 g, sorghum: 267 g
Source: Summarized from World Vision Commodity Tracking System
Limitations
ty data, the use of a calendar of events and the use of an
The reported prevalence of acute malnutrition did not include
immunization card (given the adequate Maternal and
bilateral leg edema (indicating severe malnutrition irrespective
Child Health (MCH) observed in the region) were adopt-
of W/H). Data on bilateral edema were not collected due to
ed to check the last date the child attended the MCH cen-
the lack of skilled data collectors to detect edema. This is due
ter prior to death to validate the reported date of birth.
to the educational level of the data collectors, and the high
likelihood of bias and misclassification. Thus, the prevalence
Conclusions
of acute malnutrition may have been under-estimated.
In spite of the different food aid programs in Tete Province,
In the absence of an epidemiological surveillance sys-
the observed mortality rates and prevalence of malnutrition
tem, a “verbal autopsy” was used to determine morbidity
remain high when compared to accepted international cut-
and causes of death through in-depth interviews with the
off points. Relief programs in Tete Province are not based
head of the household or next of kin;12 this may have led
on sound epidemiological data. The decision to intervene
to misclassification. However, the risk of misclassification
and the type of intervention often are determined political-
was reduced by developing focus-tested traditional terms
ly. Thus, the government potentially is relieved from fulfill-
used to describe causes of death and by asking respondents
ing its responsibility to provide much-needed services to its
to validate their answer by providing 1–3 symptoms that
constituency. In order to avoid these problems, it is advised
the person experienced in the week before the death.
that non-governmental organizations (NGOs) be granted
Finally, to account for the effect of the fluctuation in food
permission and the opportunity by the relevant stakehold-
supply, mortality was estimated over a period of 373 days.
ers in Mozambique, including the WFP to:
Such a long recall period may have affected the accuracy of
1. develop, from inception, food aid programming
the estimates. In order to improve the accuracy of mortali-
objectives with verifiable indicators relevant to each
context, beneficiary targets, and selection criteria;
January–February 2007
http://pdm.medicine.wisc.edu
Prehospital and Disaster Medicine

---

34
Mortality Rates, Prevalence of Malnutrition
2. collect baseline data; and
WFP or NGOs may finance and implement improved
3. give due consideration and support to improving
complementary programs may depend on their capacities
monitoring and reporting, periodic re-assessments
to marshal support from their respective donors and/or
and evaluations to confirm progress, and the impact
other partners.
of the project.
Unless current food aid program selection criteria are
Acknowledgements
reviewed and made more objective, the project’s impact on
The author thanks Gabrielle Mahony, World Vision
malnutrition will remain nullified and a surge in the
Australia, Duncan Campbell, FPMG, and World Vision
reliance, if not dependency, on food aid will be evident. The
Mozambique for helping with the coordination of data col-
findings of this study suggest that relief programs in Tete
lection and the logistics. A special thank you to the com-
Province would benefit from an improved alignment of
munity in Chora Basa and Chingara districts for endorsing
food aid programming with diarrheal disease control and
the study and participating. Thank you to Dr. Ben
HIV/AIDS and malaria programs. One example of a
Coghlan, Centre for International Health, Macfarlane
potentially high impact, preventive intervention is the dis-
Burnet Institute for Medical Research and Public Health,
tribution of treated mosquito nets. The extent to which the
for the statistical advice and feedback on the early draft.
References
1. Food and Agriculture Organization/World Food Program (FAO/WFP):
14. Coghlan B, Brennan R, Ngoy P, et al: Mortality in the Republic of Congo:
Special Report: FAO/WFP Crop and Food Supply Assessment Mission to
Results from a Nationwide Survey. Burnet Institute and International
Mozambique, May 1996. Available at http://www.fao.org/documents/show_
Rescue Committee: 2004.
cdr.asp?url_file=/docrep/004/w1549e/w1549e00.htm. Accessed 20 April
15. The Sphere Project: The Humanitarian Charter and Minimum Standards in
2004.
Disaster Response. Geneva: Sphere Project, 2004.
2. Schlosser S, Saulnier V: Mozambique: A country ravaged by civil war and nature.
16. Centers for Disease Control and Prevention: Famine-affected, refugee, and
Journal of Mine Action. Version 4.9. Available at http://www.maic.jmu.edu/jour-
displaced populations: Recommendations for public health issues. MMWR
nal/4.2/Features/Mozambique/mozambique.htm. Accessed 12 February 2004.
1992;41:1–76.
3. McEwan M: HIV and AIDS and Food and Nutrition Security Linkages in
17. UNAIDS/WHO Working Group on Global HIV/AIDS: Mozambique:
Mozambique: Background Document for Mainstreaming HIV and AIDS
Epidemiological Fact-Sheets on HIV/AIDS and Sexually Transmitted
Considerations into the National Food and Nutrition Security Strategy.
Infection. Geneva: World Health Organization, 2002.
Technical Secretariat for Food and Nutrition Security: Maputo, 2004.
18. Instituito Nacional de Estatística (INE) Mozambique: Demographic and
4. Martel P: Summary of the Additional Analysis of the Demographic,
Health Survey 2003. Measure DHS+/ORC Macro.; INE: Maputo, 2004.
Nutrition and Health Findings: Multi-Sectoral Assessment of the Impact of
19. Salama P, Assefa F, Talley L, et al: Malnutrition, measles, mortality and the
the Humanitarian Situation on the Lives of Mozambican Children and
humanitarian response during a famine in Ethiopia. JAMA 2001;286:563–571.
Women. Maputo: SETSAN and UNICEF, 2003.
20. Toole MJ, Waldman RJ: Prevention of excess mortality in refugee and dis-
5. Sylvester K, Fidalo L: Changara Livelihoods Study: 20 years of Change in
placed populations in developing countries. JAMA 1990;263:3296–3302.
Chingara, Tete Province. Maputo: World Vision Mozambique, 2004.
21. Médecins sans Frontières: Refugee Health: Approach to Emergency
6. FAO: Conducting Small-Scale Nutrition Surveys. A field manual. Nutrition
Situations. London: MacMillan Eductaion Ltd, 1995.
in Agriculture No. 5. Rome: Policy and Nutrition Division, 1990.
22. Pelletier DL, Frongillo EA, Schroeder DG, Habicht JP: The effect of mal-
7. Médecins sans Frontières: Nutrition Guidelines to Facilitate the Application
nutrition on child mortality in developing countries. Bull World Health
of Fundamental Concepts and Principles Necessary for the Assessment of
Organ 1995;73:443–448.
Nutritional Problems and the Implementation of Nutritional Programmes in
23. Fauveau V, Briend A, Chakraborty J, Sarder A: The contribution of severe
Emergency Situations 1st ed; Paris: Médecins sans Frontières, 1995.
malnutrition to child mortality in rural Bangladesh: Implications for target-
8. Magnani R: Sampling guide. IMPACT Food Security and Nutrition
ing nutritional interventions. Food Nutr Bull 1990;12:215–228.
Monitoring Project. Arlington: International Science and Technology
24. Schofield C, Ashworth A: Why have mortality rates for severe malnutrition
Institute, 1997.
remained so high? Bull World Health Organ 1996;74:223–229.
9. United Nations Population Division: 2000 World Population Prospects:
25. Bouma MJ, Parvez SD, Sondorp E: Treatment of tents with permethrin for
Population Database: The 2002 Revision; http://esa.un.org/unpp. Accessed
the control of vector-borne diseases. Medical News 1992;1:5–8.
on 01 September 2002.
26. Decludt B, Pecoul B, Biberson P, et al: Malaria surveillance among displaced
10. World Health Organization/African Regional Office Mozambique:
Karen population in Thailand, April 1984 to February 1989, Mae Sot,
Population profile. Available at http://www.afro.who.int/malaria/country-
Thailand. Southeast Asian J Trop Med Public Health 1991;22:504–508.
profile/mozambique.pdf. Accessed 03 June 2004.
27. Shears P, Berry A, Murphy R, Nabil MA: Epidemiological assessment of the
11. Henderson R, Sundaresan T: Cluster sampling to assess immunisation cov-
health and nutrition of Ethiopian refugees in emergency camps in Sudan,
erage: A review of experience with a simplified sampling method. Bull World
1985. BMJ 1987;295:314–317.
Health Organ 1982;60:253–260.
28. Cai Y, Feng W: Famine, social disruption, and involuntary fetal loss:
12. Checchi F, Roberts L: Interpreting and Using Mortality Data in
Evidence from Chinese survey data. Demography 2005;42:301–322.
Humanitarian Emergencies: A Primer for Non-Epidemiologists. London:
29. World Health Organization: Assessment of Iodine Deficiency Disorders and
Humanitarian Practice Network, 2005.
Monitoring their Elimination. Geneva: World Health Organization, 2001.
13. Coghlan B, Brennan RJ, Ngoy P, et al: Mortality in the Democratic Republic
30. de Onis M, Blössner M: WHO Global Database on Child Growth and
of Congo: A nationwide survey. Lancet 2006;367:44–51.
Malnutrition. Geneva: World Health Organization, 1997.
31. WHO, UNHCR, International Federation of Red Cross and Red Crescent
Societies, World Food Program: The Management of Nutrition in Major
Emergencies. Geneva: WHO, 2000.
Prehospital and Disaster Medicine
http://pdm.medicine.wisc.edu
Vol. 22, No. 1

---