You are on page 1of 7

The HIV epidemic in Zambia: socio-demographic

prevalence patterns and indications of trends


among childbearing women
Knut Fylkesnes*†, Rosemary Mubanga Musonda‡, Kelvin Kasumba*†§,
Zacchaeus Ndhlovu*†§, Fred Mluanda‡, Lovemore Kaetano‡
and Chiluba C. Chipaila‡

Objective: To examine socio-demographic HIV prevalence patterns and trends


among childbearing women in Zambia.
Design: Repeated cross-sectional surveys.
Methods: Personal interviews and unlinked anonymous testing of blood samples of
women attending antenatal care in selected areas.
Results: The 1994 data includes information from 27 areas and a total of 11 517
women. The HIV prevalence among urban residents appeared with moderate
variation at a very high level (range 25–32%, comparing provinces). The
geographical variation was more prominent in rural populations (range 8–16%) and
was approximately half the prevalence level of the urban populations. With the
exception of the 15–19 years age-group, HIV infection was found to rise sharply
with increasing educational attainment (odds ratio, 3.1; confidence interval,
2.6–3.8) when contrasting extreme educational levels. Although the assessment of
trends is somewhat restricted, the available information indicates stable prevalence
levels in most populations over the last 2–4 years.
Conclusions: The data showed extremely high HIV prevalence levels among
childbearing women. Longer time-intervals between surveys are needed, however,
in order to verify the stability in prevalence identified by this study. The tendency to
changing differentials by social status is suggested as a possible sign of an ongoing
process of significant behavioural change.

AIDS 1997, 11:339–345

Keywords: Zambia, HIV, childbearing women, prevalence, trends,


educational attainment, socio-economic status

Introduction ed (15–44 years age-group) ranges from 15–30%, with


striking urban/rural differentials [1]. Outside this belt
Globally, a number of countries found in a contiguous the HIV prevalence levels are still relatively low, except
belt stretching from Uganda southward to Zambia and from a few countries in West Africa. The available
Botswana, have experienced HIV prevalence rates far information on trends in HIV infection from sub-
above any others. In this main HIV belt, the propor- Saharan Africa indicate that prevalence rates may be
tion of the population in urban areas that is HIV infect- stabilising in several countries both within and outside

From the *Epidemiology and Research Unit, National AIDS, STD, TB & Leprosy Programme, Lusaka, Zambia, the †Institute of
Community Medicine, University of Tromsø, Norway, the ‡Immunology Unit, Tropical Disease Research Centre, Ndola and
the §University Teaching Hospital, Lusaka, Zambia.
Sponsorship: Supported financially by the Norwegian Agency for Development Cooperation (NORAD), the Ministry of
Health, Zambia, the Swedish International Development Authority (SIDA), and Irish AID.
Requests for reprints to: K. Fylkesnes, Institute of Community Medicine, University of Tromsø, 9037 Tromsø, Norway.
Date of receipt: 20 June 1996; revised: 7 November 1996; accepted 11 November 1996.

© Rapid Science Publishers ISSN 0269-9370 339


340 AIDS 1997, Vol 11 No 3

the main HIV belt [2–5]. Signs of declining prevalence collected for syphilis screening as part of the routine
rates appear in the northernmost part of the main HIV standards of care. Personal interviews were carried out
belt [6–7], contrasting with the recent and sharp oppo- by specially trained nurses using a pre-tested, structured
site trend in South Africa, Swaziland and Namibia questionnaire. The main information collected was
[8–9]. socio-demographic characteristics, and in this study,
age, marital status (married versus unmarried), residence
Childbearing women (CBW) have been identified as and educational attainment were employed in the
the key sentinel population for monitoring the HIV analyses. Residence (urban versus rural) was measured
epidemic in the sexually active population [4,10], and at individual level, and urban was defined according to
most data on the epidemic in sub-Saharan Africa stems the definition employed by the Zambia Central
from surveillance systems among CBW. These systems Statistical Office, namely, a locality with at least 5000
are primarily intended to monitor trends of infection inhabitants, having half the population in non-agricul-
from sequential cross-sectional surveys of the same tural activities, and with attributes such as piped water
population, however, and are rarely designed to pro- and electricity. Educational attainment was measured
vide national representative estimates. In Zambia, an by the number of years in school. Complete informa-
HIV sentinel surveillance system among CBW was tion on HIV status and residence was obtained from a
established in 1990 [5,11,12], and in 1994 it was con- total of 11 517 women. A total of 208 did not provide
siderably expanded in order to improve the area of information on age, and in addition 1069 did not
geographical representation and to obtain more detailed respond to the question about the number of years in
information on the socio-demographic characteristics. school.
In this study we examine the socio-demographic pat-
terns of HIV infection and describe prevalence trends HIV serology 1994
based on sequential cross-sectional surveys during the A venous blood sample was taken from each person
period 1990–1994. and serum separated in the laboratory at the local hos-
pital. All sera were tested for antibodies to HIV
(Capillus Rapid Test, HIV-1/HIV-2, Cambridge
Biotechnology, Galway, Ireland) by the local laboratory
Material and methods technician, who had been specially trained and super-
vised by laboratory personnel from the two national
Study population and methods of data reference laboratories. The sera were frozen and trans-
collection ported to one of the two national reference laborato-
The HIV sentinel surveillance system in Zambia has ries. Those from the sites in the four northern
gradually been developed since 1990, and the criteria provinces were sent to the laboratory in Ndola, and
for selecting the geographical areas (coverage of an samples from the other sites were sent to the laboratory
antenatal clinic) included in the system before 1994 in Lusaka.
have been described elsewhere [5,11,12]. In order to
further improve the geographical coverage of the sys- The accuracy of tests performed in the local laborato-
tem, 15 new areas were included in 1994 (total 27), ries was assessed as follows: 50% of all HIV antibody
and each of the nine provinces was represented by at reactive sera and 5% of non-reactive sera (randomly
least one area with primarily rural residents and one selected) were re-tested using RTD (recombigen HIV-
representing the provincial headquarters, with mostly 1/HIV-2 Rapid Test Device, Cambridge Biotech-
urban/peri-urban residents. The exception was the nology). Discordant results were tested again using
capital, Lusaka, where four sites were selected, each Western blot or an enzyme-linked immunosorbent
representing the coverage of an urban health clinic. assay (HIV-1, ELISA), (both supplied by Cambridge
The choice of the new rural areas which were added to Biotechnology). No further testing of sera was per-
the system in 1994 was a random selection from a list formed in geographical sites showing no discrepancies
of health clinics at province level. between the first and second test (among 50% of the
sites). False positives were the most common form of
All first attendees for antenatal care during a maximum discrepancy detected. In all sites where false positive
period of 4 months were enrolled. The required and tests appeared the remaining 50% reactive samples were
maximum sample size was 500, but the limit of 4 also tested again, and in sites where false negatives were
months on the data collection period resulted in sample detected, the remaining 95 non-reactive samples were
sizes below the required level from clinics with low retested.
population coverage. For most sites the data collection
took place from August to November 1994. Data analysis
Data were entered in EpiInfo (Centers for Disease
Procedures were instituted to fulfil the requirement Control and Prevention, Atlanta, Georgia, USA) and
of unlinked anonymous HIV testing of blood samples later converted to SPSS–Windows (Statistical Package
The HIV epidemic in Zambia Fylkesnes et al. 341

for the Social Sciences, Chicago, Illinois, USA) for Table 1. Prevalence of HIV infection among childbearing women
analysis. Confidence intervals (CI) were calculated by province and area, Zambia 1994.
(using one–way ANOVA) assuming a random sample Province/Area No. HIV-positive 95% CI for
tested (%) % HIV-positive
of individuals. Missing data due to non-response to the
question on education were not found to represent any Western
Mongu 479 28.4 24.3–32.4
selection bias when it comes to HIV status. A logistic Kalabo 293 10.2 6.7–13.7
regression model was used to estimate the probability Southern
of HIV infection with the inclusion of age-groups as Livingstone 595 31.9 28.2–35.7
dummy variables (15–19 years as reference), residential Macha 497 9.1 6.5–11.6
Lusaka urban
status, marital status and educational attainment as inde- Chelstone 462 24.7 20.7–28.6
pendent variables (full model). Supplementary analyses Chilenje 456 35.3 30.9–39.7
were performed in order to test the assumption that the Kalingalinga 512 21.7 18.1–25.3
Matero 394 28.4 24.0–32.9
impact of educational attainment on infection is less Central
prominent in young versus older age-groups. When Kabwe 492 29.5 25.4–33.5
performing separate analyses of the 15–19 years age- Kapiri Mopshi 500 13.0 10.0–16.0
Eastern
group, the years in school were re-grouped (from 0–4, Chipata 498 30.3 26.3–34.4
5–6, 7, 8–9, 10+ as used in the full model) to 0–4, 5–6, Minga 489 9.6 7.0–12.2
7 and 8+, and the age-adjustment was done in terms of Northern
Kasama 470 23.8 20.0–27.7
age in years in the logistic regression. Chilonga 311 16.7 12.6–20.9
Isoka 498 10.6 7.9–13.4
Mporokosso 379 12.9 9.5–16.3
Luapula
Mansa 462 23.6 19.7–25.7
Results Kashikishi 508 14.6 11.5–17.6
Kasaba 499 12.0 9.2–14.9
The 1994 findings Samfya 404 20.0 16.1–24.0
Copper belt
Table 1 shows the HIV prevalence estimates for all Ndola 501 27.5 23.6–31.5
areas by province, with the provincial headquarters list- Ibenga 361 11.4 8.1–14.6
ed first. The proportion of HIV-seropositive individu- Kamuchanga 402 23.1 19.0–27.3
als in the Lusaka areas or provincial headquarters North-western
Solwezi 231 23.8 18.3–29.3
ranged from 22–35%. The areas which were mainly Mukinge 380 9.5 6.5–12.4
rural were shown in general to have about half the Kabompo 318 5.0 2.6–7.4
level of provincial headquarters (approximately 13%), Chitokoloki 126 7.1 2.6–11.7
Total 11517 19.8 19.1–20.6
although with some striking variation from area to area
CI, Confidence interval.
(range 5–23%). Provincial estimates by urban versus
rural residence, i.e., residence measured at the individ-
ual level, showed the urban residents to be at generally Table 2 shows the overall distribution of HIV infection
uniform prevalence levels in all provinces (range by educational attainment and Table 3 the same distrib-
25–32%) with some more distinct differences between ution for the different age-groups. Among both urban
the provinces among rural residents (range 8–16%). and rural residents the seroprevalence was found to be
The results of a logistic regression model after adjust- rising significantly with increasing educational attain-
ment for age only, revealed an estimated odds ratio ment. However, among adolescents (15–19 years) HIV
(OR) of 2.60 (95% CI, 2.36–2.87) when comparing infection was not found to differ by educational level
urban with rural residence. (Table 3). As shown from the results of the logistic
regression model (Table 4), the estimated OR was 3.13
The pattern of HIV infection by age was found to be (95% CI, 2.59–3.79) when comparing the two most
closely similar in the selected areas, and is given in extreme levels of educational attainment. Separate
Table 2 for the total material according to residential
analysis for urban and rural residents yielded estimated
status. In both urban and rural areas the highest sero-
OR of 2.46 and 4.23 for urban and rural respectively
prevalence was in the 25–29 years age-group (34% ver-
sus 17%), although not significantly higher than the (Table 4). Separate logistic regression analyses per-
prevalence of the 20–24 years age-group. In all age- formed for particular age-groups supported the assump-
groups the urban : rural ratio was approximately 2 : 1. tion of less prominent impact of education attainment
The age distribution of infection from the 15–19 years on infection in young individuals. No statistically sig-
age-group appeared closely linear in shape and with a nificant impact of education was revealed within the
clearly steeper slope in urban compared to rural popu- 15–19 years age-group; i.e., separate analysis for urban
lations; i.e., from 11% to 28% versus 4% to 11% in and rural strata revealed an OR of 1.11 (CI, 0.63–1.98)
urban and rural areas respectively. in urban and 1.58 (CI, 0.88–2.84) in rural strata.
342 AIDS 1997, Vol 11 No 3

Table 2. Distribution of HIV seroprevalence by residence, age and years in school of childbearing women in Zambia, 1994.
Urban Rural
N HIV-positive N HIV-positive
(% in category) (%) 95% CI (% in category) (%) 95% CI
Total* 5251 28.2 27.0–29.4 6058 12.9 12.1–13.8
Age-group (years)
15–19 1147 (22) 20.4 18.1–22.7 1446 (24) 9.0 7.5–10.5
20–24 1867 (36) 32.2 30.1–34.4 2012 (33) 15.2 13.6–16.7
25–29 1097 (21) 33.9 31.1–36.7 1147 (19) 16.5 14.3–18.6
30–34 718 (14) 27.6 24.2–30.9 861 (14) 12.5 10.3–14.8
35–39 355 (7) 18.9 14.8–23.0 442 (7) 9.3 6.5–12.0
40–44 67 (1) 10.4 2.9–18.0 150 (3) 6.7 2.6–10.7
Years in school
0–4 516 (11) 20.2 16.7–23.6 1682 (32) 8.0 6.7–9.3
5–6 695 (14) 23.5 20.3–26.6 1138 (21) 11.4 9.6–13.3
7 1511 (31) 26.0 23.8–28.2 1446 (27) 14.2 12.4–16.0
8–9 1255 (25) 30.3 27.7–32.8 753 (14) 20.8 17.9–23.8
10–11 442 (9) 40.6 35.5–44.6 178 (3) 20.2 14.3–26.2
12+ 508 (10) 39.4 35.1–43.6 121 (2) 39.7 30.8–48.5
*The total number is less when comparing different educational groups due to non-response to the question on years in school. CI,
Confidence interval.

Table 3. HIV infection and educational attainment by residence according to age-group.


Age group (years)
15–19 20–24 25–29 30–44
Years in Urban Rural Urban Rural Urban Rural Urban Rural
school N (%) N (%) N (%) N (%) N (%) N (%) N (%) N (%)
0–4
18.4 6.8 21.7 9.5 20.7 9.2 20.0 7.2
(98) (385) (161) (483) (82) (282) (120) (362)
5–6
20.9 10.1 27.1 11.3 27.3 15.6 19.3 11.4
(182) (298) (247) (379) (121) (192) (114) (193)
7
19.8 11.0 30.6 16.2 29.9 17.4 19.8 11.7
(339) (318) (517) (463) (311) (293) (273) (281)
8+
22.1 10.4 38.4 28.2 40.9 31.7 29.1 17.7
(457) (250) (812) (390) (509) (221) (309) (136)

Table 4 shows an estimated OR of 2.03 of HIV infec- north-western province, shows a rather unlikely single
tion when comparing urban and rural residents and result from 1990, and the two latest results indicate an
after adjustment for age, marital status and educational increasing trend. The remaining five areas shown in
attainment. Further, it also shows the likelihood of Table 5 represent rural population. Four of them reveal
being infected is slightly higher among unmarried a stable prevalence during the last 2–3 years. Only one
women than married. In the 15–24 years age-group no area appears to have a tendency to increasing HIV
difference was revealed between currently married and prevalence (Kalabo).
not married, while the age-specific analysis in the
25–44 years age-group revealed that those unmarried Unfortunately, age-specific trend analyses are seriously
were twice as likely to be infected as those married hampered since only the 1993 and 1994 sentinel sur-
(OR, 2.18; CI, 1.62–2.94). veillance system provided information on the age of
individuals. Change in prevalence within the 15–19
years age-group might be indicative of trends in inci-
Indications of trends of HIV infection dence, however, and the finding within this age-group
1990–1994 of a decline in the HIV prevalence in all four Lusaka
Table 5 shows the HIV prevalence among antenatal areas is worth mentioning (down from 27.6% to 22.5%
women from areas where at least two results from when proportions of each area are weighted equally).
repeated measurements are available since 1990. The Parallel age-specific comparisons from the four rural
results from Lusaka, as the only urban area with avail- areas with age-specific data showed the prevalence to
able repeated measurements, indicate a stable preva- be stable in two of them (Macha and Minga), a decline
lence during the period 1990–1994 when considering in one (Minga) and an increase in Kalabo (1993: no
all areas together. The only peri-urban area, Solwezi in information on age from Mukinge and Solwezi).
The HIV epidemic in Zambia Fylkesnes et al. 343

Table 4. Results of logistic regression analysis of HIV infection among childbearing women in Zambia in 1994*.
Total Urban Rural
odds odds odds
Variable ratio 95% CI ratio 95% CI ratio 95% CI
Age-group:
15–19 (ref.) 1 1 1
20–24 1.85 1.60–2.14 1.89 1.57–2.27 1.79 1.41–2.26
25–29 1.95 1.66–2.29 1.91 1.55–2.35 2.03 1.57–2.62
30–34 1.42 1.18–1.71 1.49 1.18–1.88 1.32 0.97–1.78
35–39 0.98 0.76–1.26 0.93 0.67–1.29 1.05 0.70–1.56
40–44 0.73 0.42–1.27 0.55 0.23–1.31 0.92 0.45–1.89
Residence
Rural (ref.) 1 – –
Urban 2.03 1.83–2.26 – –
Marital status
Not married (ref.) 1 1 1
Married 0.85 0.72–1.00 0.81 0.67–0.99 0.95 0.71–1.27
Years in school
0–4 (ref.) 1 1 1
5–6 1.42 1.17–1.72 1.26 0.95–1.69 1.48 1.14–1.91
7 1.66 1.40–1.96 1.39 1.08–1.79 1.80 1.42–2.28
8–9 2.18 1.82–2.26 1.67 1.29–2.16 2.87 2.22–3.71
10+ 3.13 1.59–3.79 2.46 1.89–3.20 4.23 3.08–5.80
CI, Confidence interval. *Urban: n = 4846; rural: n = 5191.

Discussion An assumption relevant to the first 5–10 years of the


epidemic in sub-Saharan Africa has been that of a
Due to the comprehensive geographical coverage, the social-status bias in transmission towards the higher
findings from 1994 are expected to show the main social-status groups. When employing educational
geographical patterns of HIV infection among child- attainment as a proxy variable for social status, studies
bearing women in Zambia reasonably well. The selec- conducted in the early 1990s, within the main HIV
tion of areas at province level was arbitrary, however, belt did not conclusively suggest an increased risk
and prevalence estimates presented at provincial and among the higher educational groups [14–16]. The
national levels should be interpreted with caution. present results indicate strong social-status differentials
CBW seem to provide overall prevalence estimates in HIV seroprevalence in Zambia. Prominently higher
very close to those of the population within the 15–44 rates of infection were found in groups with higher
years age-group [13], but somewhat underestimate the educational achievement compared to those less well-
population prevalence of women within the age group educated. A previous report based on Zambian blood
15–34 [2]. The results from a recent population-based donors revealed the same distribution pattern of HIV
survey in an urban Zambian population revealed the infection at that very early stage of the epidemic [17].
same relationship: an overall HIV prevalence for both Given the relevance of a theoretical assumption that
sexes within the 15–39 years age-group of 26% the process of behavioural change takes place through
(n = 2170) compared to 25% among pregnant women diffusion of innovation from higher social-status groups
(n = 462) recruited from the same population (unpub- and downwards, the well educated are expected to be
lished data). the first to receive and to employ information on HIV
protection. Accordingly, it is likely that the socio-eco-
Our new data-set from Zambia provides some impor- nomic pattern in transmission rates will gradually
tant information leading towards a better understanding change. In most age-groups it might be too early for
of the dynamics of the epidemic. One of the major disclosing this type of change by prevalence rates, since
uncertainties has been the degree to which HIV infec- the majority of adults contracted the virus during the
tion will spread from urban to rural populations [12]. It first 5–10 years of the epidemic. A second factor is that
is noteworthy in our present findings that the HIV the progression time from HIV infection to
prevalence levels appeared surprisingly homogeneous AIDS/death is presumably longer among the higher
among urban residents in all provinces. Furthermore, educational groups. In this regard the 15–19 years age-
the surveillance data available provides some indications group represents an exception regarding our data, in
that the urban : rural ratio in HIV prevalence might particular when considering that this group became
have stabilized at a level of approximately 2–2.5 : 1. sexually active at a time when the information available
However, several more years of surveillance is needed about protection had significantly improved in compar-
for verification, and the data presented here are the first ison with the 1980s. Accordingly, if a significant
which is seen as an acceptable baseline for future behavioural change has started, we should expect to
repeated measurements, putting the urban/rural differ- find a less prominent difference in HIV infection by
ences under continuous scrutiny. education in the youngest age-group. Findings from
344 AIDS 1997, Vol 11 No 3

Table 5. Prevalence of HIV among childbearing women by area, contrast between the same countries and communities
Zambia 1990–1994 (areas with two or more measurements). with regard to when the prevalence levels stabilize
Province/site Year N HIV-positive 95% Cl [2–4,18]. Within the main HIV belt, however, the
%
level at which stabilization occurs seems the same. In
Lusaka: Zambia, the available data for most areas are still ham-
Chelstone 1994 462 24.7 20.7–28.6
1993 299 26.8 21.7–31.8 pered by short time-intervals between surveys.
1992 208 22.6 16.9–39.7 Accordingly, the revealed overall tendency of stable
Chilenje 1994 456 35.3 30.9–28.3 prevalence levels based on data from ten areas with
1993 287 22.0 17.1–26.8
1992 445 27.0 22.9–31.1 repeated measurements will have to be verified. With
Matero 1994 394 28.4 24.0–32.9 the extremely high prevalence levels revealed in our
1993 288 27.1 21.9–32.2
1992 323 29.7 24.7–34.7
study in mind, however, it should be emphasized that
Kalingalinga 1994 512 21.7 18.1–25.3 even a situation with stable HIV prevalence is likely to
1993 442 23.5 19.6–27.5 mask high incidence.
Lusaka+ 1990 750 24.5

Southern:
Macha 1994 497 9.1 6.5–11.6
1993 500 10.0 7.4–12.6
1992 418 7.9 5.3–10.5 Acknowledgements
Eastern:
Minga 1994 489 9.6 7.0–12.2
1993 497 17.7 14.3–21.1 We acknowledge the support received from the staff at
1992 497 12.9 10.0–15.8 the two reference centre laboratories, the University
Luapula: Teaching Hospital, Lusaka and the Immunology Unit,
Kashikishi 1994 508 14.6 11.5–17.6
1993 490 15.1 11.9–18.3 Tropical Disease Research Centre, Ndola, and all the
1992 202 11.4 9.2–13.6 staff at the various sentinel areas.
Western:
Kalabo 1994 293 10.2 6.7–13.7
1993 306 4.9 2.5–7.3
1992 220 5.9 2.8–9.0
North-western:
Solwezi 1994 231 23.8 18.3–29.3 References
1993 380 15.0 11.4–18.6
1990 100 30.0 21.0–40.0
1. Center for International Research, US Bureau of the Census:
Mukinge 1994 380 9.5 6.5–12.4
Recent HIV seroprevalence levels by country: June 1994.
1993 568 9.7 7.2–12.1
Washington, DC: Health Studies Branch; Research note no. 13,
1991 149 7.5 3.2–11.8
1994.
1990 100 13.0 7.0–21.0
2. Kigadye R-M, Klokke A, Nicoll A, et al.: Sentinel surveillance for
*All results from 1992 are corrected, based on results from retested, HIV among pregnant women in a developing country: 3 years’
stored 1992 blood sera. Measured accuracy of test performed in experience and comparison with a population serosurvey. AIDS
1992: sensitivity 96.0%, specificity 85.8%, positive predictive 1993, 7:849–855.
value 76.6%, negative predictive value 97.8%. †In 1990, samples 3. Sokal DC, Buzingo T, Nitunga N, Kadende P, Standaert B:
were collected from three sites in Lusaka (250 samples from each Geographic and temporal stability of HIV seroprevalence
site). It has not been possible to trace their identity. The HIV preva- among pregnant women in Bujumbura, Burundi. AIDS 1993,
lence revealed ranged between 24.0% and 25.0%. CI, Confidence 7:1481–1484.
interval. 4. Batter V, Matela B, Nsuami M, et al.: High HIV-1 incidence in
young women masked by stable overall seroprevalence among
childbearing women in Kinshasa, Zaïre: estimating incidence
this study of no social-status differentials among 15–19- from serial seroprevalence data. AIDS 1994, 8:811–817.
year-olds might thus be interpreted as a positive indica- 5. Fylkesnes K, Musonda RM, Luo NP, Msiska R: HIV infection
tion. This type of assessment might be used as a among antenatal women in Zambia 1990–1993 [letter]. AIDS
1996, 10:555–556.
supplementary method in trend assessments, but the 6. Asiimwe-Okiror G, Musinguzi J, Tembo G, et al.: Declining
suitability of the present type of data in this regard will trends in HIV infection in urban areas in Uganda. IX
have to be explored in a population-based data setting. International Conference on AIDS and STD in Africa. Kampala,
December 1995 (abstract WeC206).
7. Mulder D, Nunn A, Kamali A, Kengeya-Kayondo J: Decreasing
The monitoring of trends of HIV infection in sub- HIV-1 seroprevalence in young adults in a rural Ugandan
Saharan Africa is most often based on sequential cross- cohort. BMJ 1995, 311:833–836.
sectional surveys from the same population using CBW 8. Irlam J, Stuart J, Steinberg M, et al.: Surveillance of HIV/AIDS in
Kwazulu-Natal 1990–94. IX International Conference on AIDS
as the indicator group [2–4,18]. Reported trends are and STD in Africa . Kampala, December 1995 (abstract
based on varying and often very short time-intervals MoC495).
between surveys. Combined with critically small sam- 9. Ministry of Health & Social Services: 1994 Second National HIV
Serosurvey. Survey report. Winhoeck: Ministry of Health and
ple sizes, the prevalence changes will have to be very Social Services; 1994.
marked in order to be shown to be statistically signifi- 10. Chin J: Public health surveillance of AIDS and HIV infections.
cant. An interesting observation based on available Bull WHO 1990, 68:529–539.
11. World Health Organization, Global Programme on AIDS: HIV
reports is, apart from the striking magnitudinal preva- sentinel surveillance in Zambia (1990). Wkly Epi Rec 1992,
lence differences between neighbouring countries, the 67:221–228.
The HIV epidemic in Zambia Fylkesnes et al. 345

12. Fylkesnes K, Brunborg H, Msiska R: Zambia: the Current of a population-based cohort study. AIDS 1994, 8:1707–1713.
HIV/AIDS Situation and Future Demographic Impact. Lusaka: 16. Dalabetta GA, Motti PG, Chipanwi R, et al.: High socio-eco-
Ministry of Health (NASTLP, Epidemiology and Research Unit); nomic status is a risk factor for human immunodeficiency virus
1994. type 1 (HIV-1) but not for sexually transmitted diseases in
13. Borgdorff M, Borongo L, van Jaarsveld E, et al.: Sentinel surveil- women in Malawi: implications for HIV control. J Infect Dis
lance for HIV-1 infection: how representative are blood donors, 1993, 167:36–42.
outpatients with fever, anaemia, or sexually transmitted dis-
17. Melbye M, Njelesany EK, Bayle A, et al.: Evidence of heterosex-
eases, and antenatal clinic attenders in Mwanza Region,
Tanzania? AIDS 1993, 7:567–572. ual transmission and clinical manifestations on HIV infection
14. Nunn AJ, Kengeya-Kayondo JF, Malambo SS, Seeley JA, Mulder and related conditions in Lusaka, Zambia. Lancet 1986,
DW: Risk factors for HIV-1 infection in adults in a rural ii:1113–1115.
Ugandan community: a population study. AIDS 1994, 8:81–86. 18. Nzilambi N, De Cock KM, Forthal DN, et al.: The prevalence of
15. Sewankambo NK, Wawer MJ, Grey RH, et al.: Demographic infection with human immunodeficiency virus over a 10-year
impact of HIV infection in rural Rakai District, Uganda: results period in rural Zaire. N Engl J Med 1988, 318:276–279.

You might also like