Medical Journal of Zambia, Vol. 37, No.

3 (2010)

ORIGINAL PAPER

Interactions between HIV, dietary diversity and socio-

economic position in an urban African setting
*J Harris1, P Kelly2, S Filteau3
1
Poverty Health and Nutrition Division, International Food Policy Research Institute, Washington, DC
2
University of Zambia School of Medicine, University Teaching Hospital, Lusaka, Zambia
3
Department of Nutrition and Public Health Interventions Research, London School of Hygiene and Tropical
Medicine, Keppel Street, London WC1E 7HT

ABSTRACT INTRODUCTION

Design and objectives: A cross-sectional field study Food security and HIV
was undertaken to assess the impact of HIV on Food security is a highly complex and fluid concept,
dietary diversity in an urban context. involving the availability of foods, economic and
social access to foods, and the stability of
Setting and subjects: The study interviewed the availability and access over time and through
food preparers of 47 households in Misisi 'shocks' such as illness or drought. Stable access to
Compound, a poor area of Lusaka, Zambia. nutritious food is particularly important in HIV as
Participants were stratified by the sex and HIV status nutrients are needed by the body to fight the disease1,
of the nominal household head. and the effectiveness of and adherence to
antiretroviral drugs used to combat the virus is
Outcome measures: Outcome measures were 2
affected by sub-optimal nutrition . It has been
difference in mean Household Dietary Diversity recognised in previous research that the HIV
Score (HDDS, 12 food groups), and achievement of epidemic tends to impact negatively on household
HDDS target for assessing food security. and individual food security in Sub-Saharan Africa,
and in turn food insecurity increases both biological
Results: The HIV status of the household head was and social susceptibility of individuals to HIV,
associated with socio-economic position in female- lowering immunity and in some cases increasing
headed households, with HIV-positive individuals high-risk behaviour in order to obtain food. While
significantly more likely to head a household of most research into these interactions has taken place
higher socio-economic position (P=0.037). HIV in rural communities, little is known about the
status was not associated with dietary diversity or impacts in an urban setting.
any other measure in the study.
Dietary diversity
Conclusions: This study found no association Dietary diversity, the variety of foods or food groups
between HIV infection in the household head and consumed by a household (or individual), has been
access to a diverse diet. This may be because while suggested as a useful proxy indicator of food
wealthier households are disproportionately security (access), and increased household dietary
affected by HIV in this population, they are also diversity is shown to be associated with more
better placed to cope with the pressures of the disease complex measures of food security such as per
and maintain food security. Coping strategies capita consumption and per capita caloric
employed to maintain food security in urban acquisition4. Dietary diversity has been shown to
populations should be further investigated. increase with socio-economic status, in both rural
*Corresponding author and urban areas, with consistently much higher
Jody Harris, diversity seen in urban areas and in wealthier
[Link]@[Link] households within urban areas5.
Poverty Health and Nutrition Division, International Food
Policy Research Institute, Washington, DC Key words: HIV, Food security Dietary diversity, Urban

180
 Medical Journal of Zambia, Vol. 37, No. 3 (2010)

Food security and HIV in Zambia household invited to participate attended their
This study was undertaken to assess the impact of interview. Six households were found not to have the
HIV on dietary diversity in a poor area of Lusaka, relevant HIV records and excluded, leaving 47
Zambia's capital city. Dietary calorie availability per households in the study. Sample sizes were
capita per day in Zambia is low compared to calculated to assess the difference between mean
recommendations (1900kcal/day), although food dietary diversity scores at 90% power and 5%
availability is generally good in Lusaka. The 2002 significance.
Demographic and Health Survey (DHS) for Zambia
found 19% of Zambians to be food insecure Definitions
(measured by asking how often households have Household: 'Those whose main meals are prepared
enough food to eat), but only 6.5% of Lusaka by the same person, living under one roof at the time
residents7. of the study'.

HIV prevalence in Zambia is thought to be around Head of household: It was assumed for the purposes
14% (around 1.9 million people). In the last of this study that if a husband was present in the
Zambian survey, HIV was found to be twice as family, then he was the head of the household. If not,
prevalent in urban areas, at 20% compared to 10% in the prime-age or elder male was considered the head.
rural areas8. Zambia introduced 'opt-out' HIV testing If neither of these criteria was met, the prime-age or
elder female was considered the head. Child-headed
in 2005, with the aim of testing all visitors to health
households were not included in this study.
centres in the country. Anti-retroviral drugs have
been provided free since 2005, and currently an Measures
estimated 46% of those in need of the drugs receive This study assessed household access to foods, and
9
them . the stability of that access through the shock of HIV,
using Household Dietary Diversity Scores (HDDS)
METHODS as a proxy measure of food security (access). The
HDDS questionnaire elicits which of a range of
The study comprised a cross-sectional field study to foods or food groups have been eaten by anyone in
quantify and compare dietary diversity in the household over a reference period (24 hours); a
households stratified by the sex and HIV status of the simple sum of the number of food groups consumed
household head. gives the score, which is then compared to the
average score of the top-scoring tercile, which
Participants and sampling indicates achievable dietary diversity in the
11
Participants were drawn from a population living in population and is taken as a food security target .
Misisi settlement, Lusaka, in July 2008. Misisi is a HDDS thus looks at food consumption at household
poor squatter area close to the city centre with a level, and can assess household access to foods and
population of around 23,000, high housing density major availability issues in the study area. The
and no space for food cultivation. Participants were HDDS questionnaire was adapted to local conditions
identified through an existing cohort enrolled in an from the validated FANTA HDDS questionnaire11
ongoing study of micronutrient nutrition and using the standard 12 food groups (Table 1). Foods to
gastrointestinal health with the Tropical be listed under each of the food groups were taken
G a s t ro e n t e ro l o g y a n d N u t r i t i o n G ro u p from the FAO's food balance sheets for Zambia, and
this list was checked and added to by programme
(TROPGAN), at the University Teaching Hospital,
staff with local knowledge of foods available in
Lusaka. Participants were food-preparers of the
Zambia and Lusaka.
households (mainly women) where the household
head was enrolled in the TROPGAN study. This study was undertaken within an existing cohort
Sampling was purposive, with households identified all of whom had been tested for HIV within the last
from the TROPGAN group chosen by the sex and year, and HIV status of the household head was
HIV status of the household head. Representatives therefore based on confirmed positive blood tests.
of 53 households were interviewed, and every Individuals found to be HIV-positive had previously

181
 Medical Journal of Zambia, Vol. 37, No. 3 (2010)

been referred to a local clinic, where those with Data collection and analysis
CD4+ cell-counts of <200 cells/mm3 could access Dietary diversity questionnaires were administered
anti-retroviral drugs, which are available at no cost. by TROPGAN study nurses, simultaneously
translating from English to Nyanja as necessary.
Socio-economic position (SEP) has been found to be Nurses were trained on the study aims, the informed
independently associated with Household Dietary consent procedure, and the HDDS questionnaire.
Diversity Scores12, so SEP data were collected for Information on the proposed study was read to each
control at the analysis stage. Assessments of prospective participant in Nyanja, and opportunity
household SEP and demographics in this study used for questions given, before informed consent was
various measures, including the sex, occupation and sought and witnessed.
educational attainment of the household head (to
Ethical approval was granted by both the London
assess social standing); two scores, derived from
School of Hygiene and Tropical Medicine ethical
lists of household assets (out of six assets) and
board in London, and the University Teaching
household characteristics (describing household Hospital ethical board in Lusaka.
access to water and sanitation, the type of flooring
and the method of cooking used in the household, Data recording, cleaning and transformation, and
out of a maximum of 29) (to assess socio-economic qualitative analysis, were performed on Excel.
standing); and household crowding and dependency Statistical analysis was performed using the
ratio. Categories for each measure or score were STATA10 statistical package, with Fisher's exact
taken from the Zambia Demographic and Health test used to assess association between categorical
Surveys. variables.

Other data collected relating to food security RESULTS

included how a household obtained food and any
food aid received; whether the previous 24 hours Household characteristics
represented a 'typical' day for the household in terms 34% of the sample listed the occupation of the
of food; whether anybody in the household ate a household head as 'casual labour', and 45% as 'own
meal away from home in the previous 24 hours; and unregistered business'. This was significantly split
how many meals were prepared for the household in by sex, with males more likely to be labourers
the reference period. (54%), and women more likely to run their own
small business (70%) (P = 0.002). 10% of the sample
Table 1: Food groups used in the HDDS questionnaire received remittances, and only 2% (one household)
received food aid of any kind; 98% of households
Food groups (n=12) purchased food to eat, and no household grew or
foraged food. 20% of sampled households prepared
Cereals and grains only one meal the previous day, while 40% prepared
Roots and tubers 2 meals and 40% prepared 3.
Legumes, nuts and pulses
Milk and dairy products The HIV status of the household head was
associated with socio-economic position defined by
Eggs
both socio-economic score and asset score (table 2),
Meat, poultry and insects
with HIV-positive individuals more likely to head a
Fish and seafood
household of higher socio-economic position. On
Fruits further investigation of the data, the association
Vegetables between socio-economic score and HIV status was
Oils and fats explained by the female-headed households (HIV-
Sugar, honey, sweets and snacks negative: 100% scoring <20; HIV-positive: 46%
Other / miscellaneous scoring <20; P = 0.037). HIV status was not
associated with any other measure in the study.

182
 Medical Journal of Zambia, Vol. 37, No. 3 (2010)

Diet scores and food security Dietary patterns

98% of respondents claimed that responses given Patterns of consumption were similar between
represented a typical day's food in the household. households; most protein-rich food groups (dairy,
Households consumed between 2 and 11 food eggs and meat) were consumed by a larger
groups; the average HDDS was 8 ± 2. HDDS was proportion of HIV-positive households, while other
not significantly associated with any measure of micronutrient-rich food groups (fish, fruits,
socio-economics or demographics, or with the HIV vegetables) did not differ significantly between
status or sex of the household head even after households (figure 1). Every food group was
stratification by socio-economic position. There consumed by several households, indicating
was a non-significant trend towards higher HDDS in availability in the area.
the HIV-positive households, with all 7 of the
highest scores (HDDS=11) found in HIV-positive Figure 1: Proportion of households consuming each food group
households.

The HDDS target was 10 (average HDDS for the

highest-scoring tercile). 34 households (72%) failed
to reach the target. This did not differ significantly
by the HIV status of the household head, but was
associated with a lower asset score, with those not
meeting the target having a lower asset score (Not
meeting target: 88% scoring <3 assets; Meeting
target: 41% scoring <3 assets; P = 0.05). The cut-off
for absolute food insecurity at emergency levels,
suggested in the literature, is <4 food groups
regularly consumed, giving an overall prevalence of
severe food insecurity in the study population of 6%,
in agreement with previous findings of 6.5% in DISCUSSION
Lusaka.
Overall, this study paints a fairly positive picture of
Table 2: Household characteristics and diet scores the impacts of HIV on household food consumption
in a poor, urban area of Zambia, with little difference
found between households and
a high average dietary
All HIV- HIV+ P-value*
diversity. However, within the
n (%) 47 22 (47) 25 (53) - data were some households
with very low dietary diversity,
Sex of household head n (%) regardless of HIV status. The
m 24 (51) 12 (55) 12 (48) -
f 23 (49) 10 (45) 13 (52) - vulnerability and resilience of
a household to food insecurity
Socio-economic score / 29 mean (SD) 17.86 20.04 0.044 depends largely on the
(2.1) (2.79) livelihood system relied on for
Asset score / 6 mean (SD) 1.36 2.64 0.021
income or food production,
(1.5) (1.58) and the coping strategies
employed to maintain food
HDD score / 12 mean (SD) 7.94 7.55 8.28 0.087 security in the face of a shock
(2.33) (2.04) (2.54)
such as HIV14. Coping
strategies were not assessed in
* Fisher's exact test for significant difference between HIV this study, although both men and women seem to
positive- and HIV negative- headed households have the capacity to produce income for the
household in this population either through work or
receipt of remittances (and the term 'head of

183
 Medical Journal of Zambia, Vol. 37, No. 3 (2010)

household' may therefore be more flexible in terms between coping and increased vulnerability due to
of earning potential than was assumed in this study). HIV.
This is important in urban areas which tend to be far
more cash-intensive, with food purchased rather While much evidence is available on the livelihood
than grown. links between HIV and lower food security in rural
19
areas , the dearth of information on the impact of
Cut-offs classifying food security from HDDS are HIV in urban areas makes evidence-based policy
not well defined, nor are they standardised between and programming difficult. There are a minority of
studies. The measure of relative food security households in and around Lusaka and other African
11
recommended and used here showed almost three cities which are experiencing extreme food
quarters of households attaining less than optimal insecurity, and this figure is likely to rise with ever
dietary diversity, whatever the HIV status of the increasing urban migration, higher rates of HIV,
household head (but this may be an artefact of the weaker links to food-producing rural counterparts,
measure itself, as in taking the average score of the and rising food prices. Evidence from other
top third of respondents as a target, roughly two countries indicates that social grants for poorer
thirds are likely to miss the target). Overall, diets in households20 and the practice of urban agriculture21,
the study were monotonous and based on filling among other mechanisms, effectively buffer the
carbohydrates in all households,with only one or socio-economic impacts of HIV, and may therefore
two foods in each food group consumed, even if improve both food security and general livelihoods.
overall dietary diversity as measured by a count of Policy makers should be aware that those
food groups consumed was high. Increased diversity households most vulnerable to both HIV and food
within as well as between food groups should be insecurity may now be not in rural areas but in urban
pursued in order to improve diet quality. and peri-urban neighbourhoods, and further
attention and resources should be directed to these.
Perhaps counter intuitively, it was households of
higher socio-economic position (by both measures) Study limitations
that were more likely to be affected by HIV in this The impacts of HIV on household food security are
study (although it should be noted that this was not a many and varied, and this study chose to look at just
random sample of households). While it has been one element (the current infection of the household
found in numerous studies that HIV negatively head with HIV). This does not allow for full analysis
affects the productivity and socio-economic status of the impact of HIV on a household through
of households15,16, this finding makes sense in the morbidity, mortality and the altered demographic
context of the uniqueness of the HIV epidemic in load that are the major burdens of the HIV epidemic.
southern Africa, which has been seen to affect Particularly, the household may have experienced
wealthier men over the poorer, who may have less an AIDS death or have adopted AIDS orphans, but
access to different partners17. While all participants not have a currently affected head, so would be
in this study were poor by most definitions, this classified as non-HIV-affected in this study. This
relative difference in socio-economic position is study did not look specifically at coping strategies,
important in understanding the lack of association of but these are important in a full understanding food
HIV with dietary diversity; while wealthier security and are under-researched in an urban
households are disproportionately affected by HIV context.
they are also better placed to cope with the pressures
of the disease and maintain food security. HIV REFERENCES
prevalence is often negatively associated with
measures such as child underweight and mortality, 1. World Health Organisation. Nutrient
potentially explained by the generally better requirements for people living with
nutrition situation but higher HIV prevalence in HIV/AIDS: report of a technical consultation:
towns compared to rural contexts. However, World Health Organization, Geneva, 2003; 13-
underweight has been shown to increase more 15.
dramatically in urban areas in reaction to other 2. Cantrell, R.A., Sinkala, M., Megazinni K.,
18
shocks such as drought , suggesting a fine line Lawson-Marriott, S., Washington, S., Chi

184
 Medical Journal of Zambia, Vol. 37, No. 3 (2010)

B.H., et al. A pilot study of food 12. Ruel, M. Is dietary diversity an indicator of
supplementation to improve adherence to food security or dietary quality? A review of
antiretroviral therapy among food-insecure measurement issues and research needs, FCND
adults in Lusaka, Zambia. J Acquir Immune discussion paper #140. IFPRI, Washington DC.
Defic Syndr. 2008; 49(2):190-195. 2002.
3. Loevinsohn, M., Gillespie, S. HIV/AIDS, food 13. IPC Global Partners. Integrated Food Security
security and rural livelihoods: Understanding Phase Classification Technical Manual Version
and responding: Food consumption and 1.1: Food and Agriculture Organisation, Rome
nutrition division discussion paper 157. IFPRI, 2008.
Washington DC. 2003. 14. Gillespie S, Kadiyala, S. HIV/AIDS and food
4. Hoddinott, J., Yohannes, Y. Dietary diversity as and nutrition security, Food Policy Review 7.
a food security indicator. Food and Nutrition IFPRI, Washington DC. 2005.
Technical Assistance (FANTA), Washington, 15. Fox, M.P., Rosen, S., MacLeod, W.B.,
DC. 2002. Wasunna, M., Bii, M., Foglia, G., et al. The
5. Hatloy, A., Hallund, J., Diarra, M.M., Oshaug, impact of HIV/AIDS on labour productivity in
A. Food variety, socioeconomic status and Kenya. Trop Med Int Health. 2004 ;9(3):318-
nutritional status in urban and rural areas in 324.
Koutiala (Mali). Public Health Nutr. 2000; 16. Bachmann, M.O., Booysen, F.L. Relationships
3(1): 57-65. between HIV/AIDS, income and expenditure
6. Food and Agriculture Organisation. FAO over time in deprived South African
Country Profiles: Zambia. 2005. households. AIDS Care. 2004;16(7):817-826.
7. Central Statistical Office [Zambia], Central 17. Chapoto, A., Jayne, T. Socio-economic
Board of Health [Zambia], Macro ORC. characteristics of individuals affected by prime-
Zambia Demographic and Health Survey 2001- age AIDS-related mortality in Zambia:
2002. Calverton, Maryland, USA: Central Ministry of Agriculture and Cooperatives,
Statistical Office, Central Board of Health, and Agricultural Consultative Forum, Michigan
ORC Macro. 2003. State University 2005.
8. Central Statistical Office, Ministry of Health, 18. Mason J.B., Bailes, A., Mason, K.E., Yambi, O.,
Tropical Diseases Research Centre, University Jonsson, U., Hudspeth, C., et al. AIDS, drought,
of Zambia, Macro International Incorporated. and child malnutrition in southern Africa.
Zambia Demographic and Health Survey Public Health Nutr. 2005;8(6):551-563.
2007-2009. 19. Various. AIDS, poverty and hunger: Challenges
9. World Health Organisation. Towards universal and responses. Gillespie, S., editor:
access: scaling up priority HIV/AIDS International Food Policy Research Institute,
interventions in the health sector. WHO, Washington, D.C.; 2006.
Geneva. 2008. 20. Booysen F. Social grants as safety net for
10. Baboo, K.S., Zulu, I., Sianongo, S., Kaonga, HIV/AIDS-affected households in South
K., Mwenya, D., Mwansa, J.C.L. et al. Africa. SAHARA J. 2004;1(1):45-56.
Intestinal infection in adults in an urban 21. Bukusuba, J., Kikafunda, J.K., Whitehead,
Zambian population in relation to immune R.G. Food security status in households of
status. Med J Zambia 2004;6:104-110. people living with HIV/AIDS (PLWHA) in a
11. Food and Nutrition Technical Assistance. Ugandan urban setting. Br J Nutr.
Household Dietary Diversity Score (HDDS) 2007;98(1):211-217.
for Measurement of Household Food Access:
Indicator Guide, Version 2. FANTA,
Washington, DC. 2006.

185