Secondary Logo

Journal Logo


HIV Incidence Among Women of Reproductive Age in Malawi and Zimbabwe

Kumwenda, Newton PhD*; Hoffman, Irving; Chirenje, Mike; Kelly, Clifton§; Coletti, Anne; Ristow, Aleen*; Martinson, Francis; Brown, Joelle; Chilongozi, David**; Richardson, Barbra§††; Rosenberg, Zeda‡‡; Padian, Nancy; Taha, Taha MD, PhD*

Author Information
doi: 10.1097/01.olq.0000223283.27142.9f
  • Free

UNAIDS ESTIMATES THAT IN SUB-SAHARAN Africa, 57% of adult HIV-1 infections are in women and 75% of young people infected are women and girls (UNAIDS, 2004). Two countries that have been hard hit by the HIV epidemic are Malawi and Zimbabwe with estimates of HIV prevalence by end of 2003 among adults 15 to 49 years being 14.2% and 24.6%, respectively.1 Several studies during the past decade have also reported high HIV-1 incidence among women in these 2 countries. In Malawi, the overall incidence rate among antenatal and postnatal urban women was 4.2 per 100 person years (p-y) (95% confidence interval [CI] = 3.2–5.3) during 1990 to 1995,2 and among men working in a semirural sugar estate, the average annual incidence was 3.5% during 1996 to 1999.3 In Zimbabwe, among antenatal and postnatal women, the HIV incidence rate was 4.8 per 100 p-y (95% CI = 3.1–6.5).4 Among a cohort of male factory workers in Harare, Zimbabwe, HIV incidence rate was 2.9 per 100 p-y during the period 1993 to 95.5

In southern Africa, subtype C is the dominant HIV-1 clade. The populations of Malawi and Zimbabwe share several demographic characteristics, both are of Bantu origin, and although not confirmed, it is unlikely that genetic susceptibility to HIV infection could be different in these 2 countries. Therefore, socioeconomic and behavioral factors and genital tract infections could be the major determinants of HIV acquisition and may account for any potential differences between the 2 countries. Data from the 1997 Zimbabwe Sexual Behavior and Condom Use Survey on a randomly selected nationally representative sample of men and women found high levels of condom use among steady and casual partners (>60%); condom use, however, was low among legal and informal spousal relationships (approximately 10%).6 In Malawi, reported condom use was low and inconsistent in the early 1990s,7 and based on recent UNAIDS estimates, condom use appears to have increased to more than 30% among young men and women.1

The health and socioeconomic impact of HIV/AIDS is evident in these countries and preventive interventions are urgently needed. Clinical trials being actively considered for implementation in sub-Saharan Africa are use of female microbicides, antiretroviral treatment, and HIV vaccines. The design and monitoring of these trials, however, will require knowledge of baseline HIV incidence. We conducted a multisite prospective study in Blantyre and Lilongwe, Malawi, and Harare, Zimbabwe, to provide HIV counseling, promote condom use, and determine incidence of HIV and other sexually transmitted infections (STIs). Unlike prior studies, the current investigation focused on intensive counseling and active condom promotion as the main activity.

Materials and Methods

Study Design and Clinical Procedures

This study was conducted between August 1999 and April 2001. In 1999, a phase III clinical trial of the vaginal microbicide Nonoxynol-9 (N-9) was being considered. In preparation for this trial, 2 pilot studies were implemented in Malawi and Zimbabwe: the first study to assess safety of N-9 and the second to start an intensive condom counseling and promotion activity. The results of the safety study have been previously reported.8 The objective of the intensive condom counseling and promotion activity was to educate women on consistent condom use as a standard HIV prevention method that should accompany the phase III vaginal microbicide trial. This condom promotion phase continued for 8 weeks. During this period, 3 sessions of intensive condom counseling and promotion were provided using standard scripts every 2 weeks. Before each counseling session, data on frequency of intercourse and condom use were collected through a structured questionnaire. While this condom promotion phase was in progress, results of a phase III vaginal microbicide trial from South Africa showed that N-9 gel had no effect in protecting against heterosexual HIV acquisition and might have adverse effects on the vaginal mucosa.9 The original design of proceeding to a phase III N-9 clinical trial was therefore abandoned. Women participating in the pilot study of condom promotion were reconsented to continue in the current observational study.

Women were recruited from postnatal or family planning clinics in Blantyre and Lilongwe in Malawi and Harare, Zimbabwe. Eligibility criteria included age 18 years or older, willing to give written informed consent, and willing to adhere to study follow-up schedule. At enrollment, after obtaining informed consent, women received HIV pre- and posttest counseling. Demographic, sexual, behavioral, and contraceptive use data were collected through interviews using a standard questionnaire in a local language. Follow-up visits were conducted quarterly for 9 months. At each of these visits, a speculum-aided pelvic examination was performed to detect preexisting genital lesions and to collect specimens for laboratory testing for STIs. At each quarterly visit, venous blood was collected and tested for HIV-1 and syphilis. STIs treatment was available whenever indicated based on local guidelines. Referral for other health problems was also available in all clinics.

Laboratory Tests

Serum samples collected were tested for HIV-1 and syphilis. For HIV-1 testing, samples were initially tested using site-specific testing protocols (2 different conventional enzyme-linked immunosorbent assay tests followed by Western blot of all reactive samples). For syphilis testing, each sample was tested using the rapid plasma reagin (RPR) test (Macro-Vue; Becton Dickinson, Cockeysville, MD) and RPR-positive samples were confirmed using the Treponema pallidum hemagglutination assay. Vaginal and cervical specimens were collected to screen for chlamydia (Syva EIA, Trenton, NJ), gonorrhea (standard culture), trichomoniasis, candidiasis, and bacterial vaginosis (vaginal wet mount, standard morphology). The definition of bacterial vaginosis was based on Amsel's clinical criteria.10 Herpes simplex virus type 2 tests were not performed in this study.

Data Analysis

Enrollment demographic and behavioral characteristics, pelvic examination, and laboratory results were summarized by site. Women were followed to time of HIV seroconversion or were censored at their final HIV-negative results within the 9-month follow-up period. Incident cases of STIs and genital lesions were calculated based on the number of women who were not infected with these conditions at the start of the quarterly follow-up period. For baseline comparisons, Fisher exact 2-sided tests were used to compare proportions by country, and a Wilcoxon rank sum test was used to compare the continuous variable, age. Overall HIV seroincidence rates were calculated as number of seroconversions divided by total woman-years (w-y) under observation (per 100). Time of seroconversion was calculated as the time from the baseline HIV-negative test result to the HIV-positive test result. Univariate and multivariate Cox proportional hazard models were used to examine correlates of HIV-1 seroconversion. Unadjusted and adjusted hazard ratios (HRs) and 95% confidence intervals were estimated. In the Cox proportional hazard models, we examined both fixed (baseline) and time-varying covariates. Selection of variables for inclusion in the multivariate model to adjust for potential confounding was based on univariate statistical significance and biologic plausibility. We also controlled for site in these analyses. SAS version 8.2 (Cary, NC) was used for all statistical analyses.

This study was approved by ethical committees in Malawi and Zimbabwe and at the University of North Carolina, Johns Hopkins University, and the University of California at San Francisco in the United States.


A total of 2016 HIV-negative women (657 from Blantyre, 685 from Lilongwe, and 674 from Harare) were enrolled in the condom promotion and counseling pilot phase of the study. Table 1 displays selected demographic and behavioral characteristics of the women enrolled by site. Women from the 2 sites in Malawi (Blantyre and Lilongwe) showed comparable characteristics but were significantly different from the Harare site in Zimbabwe for several baseline characteristics. Women enrolled in Zimbabwe were slightly older, more likely to be married and living with their husbands, more educated, of higher socioeconomic status (based on available utilities and income), more sexually active and used male condoms with last sex act, and more likely to be using contraceptives compared with women enrolled in Malawi (all P values <0.001). Women from the 3 sites also showed some similarities and differences in STI frequency at enrollment. As shown in Table 2, the proportions of women who reported genital symptoms and those with detectable trichomonas infection were comparable between the 2 Malawi sites but significantly higher than rates at the Zimbabwe site (both P values <0.001). The prevalence of chlamydia was low at all 3 sites and significantly lower in Malawi (approximately 1%) compared with Zimbabwe (approximately 2%) (P = 0.05). The prevalence of gonorrhea was comparable and low at the 3 sites, averaging approximately 1%. Likewise, there were no significant differences at the 3 sites in prevalence of genital lesions in the month before enrollment (approximately 2% in Malawi and 1% in Zimbabwe; P = 0.08) or syphilis (approximately 3% in Malawi and 2% in Zimbabwe; P = 0.16).

Demographic and Behavioral Characteristics of HIV-Negative Women at Enrollment by Site
Prevalence of Sexually Transmitted Infections and Genital Findings Among HIV-Negative Women at Enrollment by Site

Of the 2016 HIV-negative women enrolled in the study, 1679 (559 in Blantyre, 549 in Harare, and 571 in Lilongwe) were tested for HIV during the quarterly follow-up visits for a total of 2429.2 w-y. The overall median duration of follow up was 1.4 years (interquartile range [IQR] 1.2–1.8); the median follow-up times at the 3 sites were similar (1.4 in Blantyre [IQR 1.1–1.7], 1.4 [IQR 1.2–1.7] in Harare, and 1.6 [IQR 1.4–1.9] in Lilongwe). A total of 113 women seroconverted, resulting in an overall HIV incidence rate of 4.7 per 100 w-y (95% CI = 3.8–5.6). Incidence rates (per 100 w-y) were similar across sites: 4.9 (95% CI = 3.4–6.8), Blantyre; 4.9 (3.5–6.7), Harare; and 4.3 (3.0–5.8), Lilongwe.

Of 1575 women who were negative for genital lesions (no genital sores or ulcers on pelvic examination at the beginning of the follow-up period), 1459 women had at least one follow-up pelvic examination. A total of 28 women had new genital lesions for an incidence rate of 2.7 (28 of 1034.7 w-y; 95% CI = 1.8–3.9) per 100 w-y. The incidence rates per site were 2.3 (0.9–4.8) in Blantyre, 1.6 (0.6–3.6) in Harare, and 4.0 (2.3–6.7) in Lilongwe.

Estimates of unadjusted HIV incidence and the association of various factors with incident HIV infection for the 3 sites combined are shown in Table 3. The overall incidence rate was highest among young women (<25 years) and there was a clear trend of decreasing risk with older age. There was also a clear trend of increase in HIV incidence with higher educational level, with the risk being highest (6.34 per 100 w-y) among women who attended more than secondary education. Other factors related to increased risk of HIV acquisition (albeit some not statistically significant) were being unmarried (11.36 per 100 w-y, P <0.01), having ≥2 sexual partners (16.38 per 100 w-y, P = 0.14), having any STIs or genital lesions (genital sores or ulcers) during follow up (11.00 per 100 w-y, P <0.01), or having bacterial vaginosis (7.34 per 100 w-y, P = 0.03) (Table 3). Despite self-reports of full (100%) condom use during the follow up, the HIV incidence rate was still relatively high at 4.37 per 100 w-y (Table 3).

Factors Associated With Incident HIV-1 Infection

In the univariate Cox proportional hazard models, age, educational level, being unmarried, having STIs or genital lesions during follow up, or having bacterial vaginosis during the quarterly visits were all significantly associated with incident HIV infection (data not shown). These variables, with the exception of bacterial vaginosis, remained significantly associated with incident HIV infection in the multivariate Cox proportional hazard model (Table 4). Study site was not associated with incident HIV infection in either unadjusted or adjusted models.

Final Multivariable Model: Association of Selected Risk Factors With Incident HIV-1 Infection Among Women in Malawi and Zimbabwe


This multisite study shows that the HIV incidence rate is high in Malawi and Zimbabwe (overall 4.7 per 100 w-y) and was not different among women of reproductive age from the 3 sites. An interesting observation in both Malawi and Zimbabwe is that the prevalence of some conventional STIs (e.g., gonorrhea, chlamydia, and syphilis) was low at time of enrollment. Additionally, the incidence of genital sores or ulcers was low—average incidence rate of 2.7 per 100 w-y (95% CI = 1.8–3.9). This corroborates results of previous studies from both countries that reported declining trends of STIs during the past few years.2,11 However, the incidence of HIV, as shown in this study and compared with prior studies, has not changed. For example, among women of reproductive age in Malawi, the incidence rate was 4.2 (95% CI = 3.2–5.3) per 100 p-y during 1990 to 1995 and was highest among young women <20 years (incidence rate of 5.9 per 100 p-y.2 Likewise, earlier studies among antenatal and postnatal women in Zimbabwe reported HIV incidence rate of 4.8 per 100 p-y (95% CI = 3.1–6.5) and the rate was highest among those 17 years or less (6.3 per 100 p-y).4 The current study showed similar findings: overall incidence rate 4.7 (95% CI = 3.8–5.6) per 100 w-y and among women <25 years the incidence rate was 5.8 per 100 p-y. This lack of change in HIV acquisition over time, even with intensive condom counseling and availability, and frequent STI diagnosis and treatment should therefore be explored to maximize opportunities for interventions. It emphasizes the need for alternative interventions such as microbicides, antiretroviral prophylaxis, or vaccines. Another explanation for the apparent decline in STIs without a decrease in HIV incidence is that changes in STI incidence may not necessarily be good surrogates for HIV incidence in these mature epidemics. A potential limitation in this study is that the conventional laboratory tests we used to detect chlamydia and gonorrhea infections might have lower sensitivity compared with other molecular diagnostic tests. However, other studies in Malawi that used ligase chain reaction for chlamydia testing also reported low rates (0.7–2%) among high-risk men.12,13

The major determinants of HIV seroconversion in this study were mostly demographic or socioeconomic (young age, higher education level, or being unmarried). The increasing trend of HIV incidence with increasing level of education is rather unexpected; however, this might reflect a higher socioeconomic status, a factor reported to be associated with HIV infection in this region of Africa.14 Higher socioeconomic status may be an underlying factor for having multiple sexual partners. Consistent with other studies, the only biologic factor that increased the hazard of HIV infection by 2.5-fold was any STI, including genital lesions. Of the STIs, Trichomonas vaginalis infection was the most common at the 3 sites (Table 2). Of the behavioral factors, condom use was not significantly associated with HIV acquisition either in the univariate or multivariate analyses, but there is evidence that self-reported condom use in this population was overestimated.15 The association of demographic and socioeconomic factors with HIV incidence in this study appears to reflect the lower status of women, their inability to make decisions, and the economic difficulties they encounter.1 However, we did not specifically measure these more distal factors.

Although previous studies reported differences in risk factors that contribute to uneven prevalence of HIV-1 in different countries and regions of Africa,16–18 no comparisons of HIV incidence were made. These studies, as a result of their cross-sectional nature, could only suggest regional differences and likewise variations in risk factors. In our multisite study, comparison of HIV incidence showed that the rate of HIV acquisition is not different among women of reproductive age. Our study also shows that differences in underlying characteristics (Tables 1 and 2) appear to have no major influence on rates of HIV acquisition. However, time to detect an impact in this study might be short because some of the behavioral changes may be more recent to influence HIV incidence. In Malawi and Zimbabwe, it appears that this unchanging incidence is the main reason for the stable prevalence. In both countries, mortality associated with HIV/AIDS is high. Other explanations for the differences in HIV prevalence between the 2 countries, despite similar incidence, may include factors that we did not measure (e.g., herpes simplex virus type 2), differences in population structure, and sampling variations.

Clinical trials to evaluate the effectiveness of measures to reduce heterosexual HIV transmission (e.g., microbicides, treatment of STIs, promotion of behavior change, and HIV vaccines) are being planned. The design and monitoring of these trials require knowledge of baseline incidence of HIV. Current estimates of incidence, however, are rarely available in sub-Saharan Africa. This study provided incidence estimates from 3 sites in southern Africa. Innovative interventions that take into account the vulnerability of women of reproductive age are needed. Safe measures that can be controlled by women will be most appropriate.


1. 2004 Report on the Global AIDS Epidemic. 4th Global Report. Geneva, Switzerland: UNAIDS, June 2004.
2. Taha TE, Dallabetta GA, Hoover DR, et al. Trends of HIV-1 and sexually transmitted diseases among pregnant and postpartum women in urban Malawi. AIDS 1998; 12:197–203.
3. Kumwenda NI, Taha TE, Hoover DR, et al. HIV-1 incidence among male workers at a sugar estate in rural Malawi. J Acquir Immun Defic Syndr 2001; 27:202–208.
4. Mbizvo MT, Kasule J, Mahomed K, Nathoo K. HIV-1 seroconversion incidence following pregnancy and delivery among women seronegative at recruitment in Harare, Zimbabwe. Cent Afr J Med 2001; 47:115–118.
5. Mbizvo MT, Machekano R, McFarland W, et al. HIV incidence and correlates of seroconversion in a cohort of male factory workers in Harare, Zimbabwe. AIDS 1996; 10:895–901.
6. Adetunji J, Meekers D. Consistency in condom use in the context of HIV/AIDS in Zimbabwe. J Biosoc Sci 2001; 33:121–138.
7. Taha TE, Canner JK, Chiphangwi JD, et al. Reported condom use is not associated with incidence of sexually transmitted diseases in Malawi. AIDS 1996; 10:207–212.
8. Hoffman IF, Taha TE, Padina NS, et al. Nonoxynol-9 100 mg gel: Multisite safety study from sub-Saharan Africa. AIDS 2004; 18:2191–2195.
9. Van Damme L, Ramjee G, Alary M, et al. Effectiveness of COL-1492, a nonoxynol-9 vaginal gel, on HIV-1 transmission in female sex workers: A randomized controlled trial. Lancet 2002; 360:971–977.
10. Amsel R. Totten PA, Speigel CA, Chen KCS, Eschenbach D, Holmes KK. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiological associations. Am J Med 1983; 74:14–22.
11. Decosas J, Padian N. The profile and contest of the epidemics of sexually transmitted infections including HIV in Zimbabwe. Sex Transm Infect 2002; 78(suppl 1):140–146.
12. Price MA, Zimba D, Hoffman IF, et al. Addition of treatment for trichomoniasis to syndromic management of urethritis in Malawi: A randomized clinical trial. Sex Transm Dis 2003; 30:516–522.
13. Cohen MS, Hoffman IF, Royce RA, et al. Reduction of concentration of HIV-1 in semen after treatment of urethritis: Implications for prevention of sexual transmission of HIV-1. Lancet 1997; 349:1868–1873.
14. Dallabetta GA, Miotti PG, Chiphangwi JD, et al. High socioeconomic status is a risk factor for human immunodeficiency virus type 1 (HIV-1) infection but not for sexually transmitted diseases in women in Malawi: Implications for HIV control. J Infect Dis 1993; 167:36–42.
15. Martinson FEA, Chilongozi D, Kelly CW, et al. Self-reported condom use and HIV/STD incident cases: HIV Prevention Trials Networks Study 016A. Microbicides 2004 Conference, London, March, 28–31, 2004; Abstract 02527.
16. Boerma JT, Gregson S, Nyamukapa C, Urassa M. Understanding the uneven spread of HIV within Africa. Comparative study of biologic, behavioral, and contextual factors in rural populations in Tanzania and Zimbabwe. Sex Transm Dis 2003; 30:779–787.
17. Asamoah-Odei E, Calleja JMG, Boerma JT. HIV prevalence and trends in sub-Saharan Africa: No decline and large subregional differences. Lancet 2004; 364:35–40.
18. Auvert B, Buve A, Ferry B, et al. Ecological and individual level analysis of risk factors for HIV infection in urban populations in sub-Saharan Africa with different levels of HIV infection. AIDS 2001; Suppl.4:S15–30.
© Copyright 2006 American Sexually Transmitted Diseases Association