SEARCH

SEARCH BY CITATION

Keywords:

  • human immunodeficiency virus;
  • human immunodeficiency virus preventive behaviour;
  • Zimbabwe;
  • population-based survey
  • VIH;
  • comportements préventifs du VIH;
  • Zimbabwe;
  • enquête de population
  • VIH;
  • comportamiento preventivo frente al VIH;
  • Zimbabwe;
  • estudio basado en la población

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Objective  To assess reported HIV knowledge and attitudes, sexual behaviours and HIV testing in Zimbabwe.

Methods  Representative household surveys of all 18–24 year olds and a proportion of 25–44 year olds were conducted in six purposefully selected rural districts in two provinces in 2007 and 2009. Both surveys used the same methods and questionnaires. We compared differences in reported HIV knowledge, sexual behaviours and HIV testing, controlling for differences in socio-demographics at baseline, using cross-tabulations and multivariate regression analyses.

Results  Analysis was restricted to districts included in both baseline (n = 1891) and mid-term (n = 2746) surveys. Comparisons indicate increased reports of HIV knowledge (35%vs. 22% had high knowledge) and more favourable individual attitudes towards HIV. There was an increase in reported HIV testing (men: 41%vs. 31%, women: 55%vs. 36%) and condom use (men: adjusted odds ratio (AOR) = 1.35, women: AOR = 1.22) and a decrease in number of sexual partners (men: 67%vs. 49% reported 1 partner/previous 6 months, women: 77%vs. 68%).

Conclusions  As Zimbabwe continues to document declines in HIV prevalence, this analysis offers insight into recent and continuing positive changes in knowledge, attitudes and behaviours among the rural population.

Objectif:  Analyser la connaissance rapportée et les attitudes sur le VIH, les comportements sexuels et le dépistage du VIH au Zimbabwe.

Méthodes:  Des enquêtes représentatives auprès des ménages sur toutes les personnes âgées de 18 à 24 ans et une proportion de celles âgées de 25 à 44 ans ont été menées dans six districts ruraux délibérément choisis dans deux provinces en 2007 et 2009. Les deux enquêtes ont utilisé les mêmes méthodes et questionnaires. Nous avons comparé les différences dans la connaissance déclarée sur le VIH, les comportements sexuels et le dépistage du VIH, en ajustant pour les différences sociodémographiques au départ, à l’aide des tableaux croisés et des analyses de régression multivariée.

Résultats:  L’analyse a été limitée aux districts inclus à la fois dans les sondages du départ (n = 1891) et à mi-terme (n = 2746). Les comparaisons indiquent un report accru de la connaissance sur le VIH (35% vs 22% avaient une connaissance élevée) et des attitudes individuelles plus favorables à l’égard du VIH. Il y avait une augmentation des cas de dépistage du VIH (hommes: 41% vs 31%, femmes: 55% vs 36% :) et l’utilisation du préservatif (hommes: AOR = 1,35; femmes: AOR = 1,22) et une diminution du nombre de partenaires sexuels (hommes: 67% vs 49%; femmes: 77% vs 68%, ont déclaré 1 partenaire au cours 6 mois précédents).

Conclusions:  Alors que le Zimbabwe continue d’enregistrer une diminution de la prévalence du VIH, cette analyse offre un aperçu des changements positifs récents et continus dans la connaissance, les attitudes et les comportements au sein de la population rurale.

Objetivo:  Evaluar los conocimientos y actitudes frente al VIH reportados, el comportamiento sexual y el testaje del VIH en Zimbabwe.

Métodos:  Encuestas representativas en hogares, a todos los habitantes con edades entre los 18–24 años y una proporción de los de 25–44 años, en seis distritos rurales seleccionados a propósito en dos provincias, durante el 2007 y 2009. Ambas encuestas utilizaban los mismos métodos y cuestionarios. Hemos comparado las diferencias en el conocimiento reportado del VIH, el comportamiento sexual y el testaje del VIH, controlando para las diferencias socioeconómicas al inicio del estudio, utilizando tabulaciones cruzadas y análisis de regresión logística multivariada.

Resultados:  El análisis estaba restringido a los distritos incluidos tanto en el estudio de base (n = 1891) como en el estudio intermedio (n = 2746). Las comparaciones indicaban un aumento de reportes sobre conocimiento del VIH (35%vs.22% tenían un nivel alto de conocimientos), y una actitud individual más favorable hacia el VIH. Había un aumento en el testaje para VIH reportado (hombres: 41% vs. 31%, mujeres: 55% vs. 36%:) y en uso de preservativos (hombres: AOR = 1.35, mujeres: AOR = 1.22), y una disminución en el número de parejas sexuales (hombres: 67% vs. 49% reportando 1 pareja/últimos 6 meses, mujeres: 77% vs. 68%).

Conclusiones:  Zimbabwe continúa documentando una disminución en la prevalencia del VIH, y este análisis ofrece una perspectiva de los cambios recientes y positivos en el conocimiento, las actitudes y el comportamiento de la población rural.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Zimbabwe has one of the largest and most sustained HIV epidemics in the world (Stirling et al. 2008; Gregson et al. 2010). While the overall HIV prevalence among adults has declined from 25.3% in 1997 to 13.7% in 2009 (Zimbabwe Ministry of Health and Child Welfare 2010), one in seven adults still lives with HIV (Gregson et al. 2010). These unacceptably high HIV rates continue to severely impact morbidity and mortality straining the social fabric of the country (Foster 2002, 2006).

Recent analyses have been conducted to explain the decrease in HIV prevalence in Zimbabwe (Gregson et al. 2010, 2011; Halperin et al. 2011; Muchini et al. 2011). Mathematical modelling suggests that while mortality rates were high, the natural dynamic of the epidemic is unlikely to explain this decline in prevalence (Hallett et al. 2006). Reported behaviour change has been documented, specifically reduction in numbers of multiple and concurrent sexual partnerships and increase in reported condom use with non-regular partners (Zimbabwe Central Statistical Office & Macro International Inc. 2006, 2010; Gregson et al. 2010, 2011). However, data informing these analyses were collected in the early to mid-2000s. As HIV prevalence has continued to decline to the end of the 2000s, it is important to examine whether this decrease is the result of further changes in behaviours (e.g. sexual behaviour, HIV testing). This is explored in this study.

In early 2007, Zimbabwe launched its National Behaviour Change Programme (NBCP) promoting behaviour change to reduce HIV transmission. The programme is evaluated using three population-based cross-sectional surveys conducted in 2007 (baseline prior to implementation), 2009 (mid-term), and 2011 (final). The main NBCP evaluation will be conducted using baseline and end-line data, and its results will be reported elsewhere. This study presents data from the baseline and mid-term surveys that explored changes in knowledge, behaviours and attitudes among rural Zimbabweans between 2007 and 2009.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Study design

The baseline survey was carried out in September-December 2007. Four provinces were purposively selected to ensure national representation/coverage (Masvingo, Matabeleland North, Mashonaland East, Midlands). Four districts were randomly selected in each province; two districts were scheduled to receive the programme in 2007 (Phase I) and two districts in 2010 (Phase II). Ten enumeration areas (EAs) were randomly selected from all EAs in each selected district (according to Census listing), of which three were urban/peri-urban and seven rural (according to Zimbabwe Central Statistics Office).

The mid-term survey was conducted in October-November 2009 to provide an interim assessment of programme exposure in selected areas. Of the sixteen districts selected for the baseline, two Phase II districts and four Phase I districts were purposively selected as they were in provinces where the NBCP had been most comprehensively implemented. Five EAs were selected from each district: in Phase I districts the EAs where the programme had been most intensively implemented (according to NBCP monitoring data) and in Phase II districts EAs that included a school or health centre were randomly selected. This study examines change over time using data from the six districts that contribute to both surveys.

Study population

The baseline and mid-term surveys were population-based surveys conducted among 18–44 year olds. In both surveys, all 18–24 year olds living in the selected communities were eligible to participate. As the number of adults aged 25–44 was anticipated to be greater than that of younger participants, approximately one-third of adults were randomly sampled using a pre-specified sampling frame.

In both surveys, after written informed consent, participants completed a questionnaire using audio computer-assisted survey instrument (ACASI). This study received ethics approval from the Medical Research Council of Zimbabwe and University College London.

Data analysis

Data from the two surveys were merged and only included data from the six districts surveyed in both rounds. Data were analysed using STATA 10 (STATA Corp., College Station, TX, USA). Given the small number of districts, in multivariate analyses, we corrected for district-level clustering by controlling for district in all significance tests. Comparisons in socio-demographic characteristics and multivariate analyses were conducted using unweighted data. All other univariate and bivariate analyses employed weighted data, to adjust for having oversampled youth, and urban/rural (only for baseline) and gender representation (using the 2002 Census population distribution within each selected district).

We examined the change between the two survey rounds for a number of dependent variables (DV) (Tables 1–6) and conducted bivariate and multivariate analyses. We employed logistic regression for dichotomous DVs, multinomial logit regression for DVs with more than two categories and Cox proportional hazard regression for age at first sex (Table 4), controlling for district and the socio-demographic characteristics shown to be significantly different between the two rounds (see table footnotes). For example, for comparisons using the entire sample, we controlled for district, age (continuous), ethnic group (Shona/non-Shona), religion (Christian/non-Christian), marital status (married/previously married/never married), occupation (subsistence farming/commercial farming/formally employed/other employment/unemployed) and number of sellable assets (0–1/2–4/5–9). To ensure enough variability in the multivariate models (a minimum of 10 cases per dummy variable for each category of the outcome variable), control variables were dropped from the models (starting with the variables least associated with the outcome variable) (see table footnotes). Analyses presented in Table 2 were also stratified by sex and age; these results are only mentioned if there was a difference in the pattern observed when examining the entire sample. We also examined the change over time in sexual behaviour indicators while stratifying for sex, age and marital status (Tables 5 and 6).

Table 1.   Knowledge and attitudes to HIV by survey round (%)
 2007 (n = 1997)2009 (n = 2746)AOR95% CIP-value
  1. AOR, adjusted odds ratio; CI, confidence interval, bivariate analyses employed weighted data, and multivariate analyses employed unweighted data but controlled for district, age, ethnic group, religion, marital status, occupation, sellable assets.

  2. *An index measuring HIV knowledge ranging from 0 to 6 was created by adding answers (1 = yes, 0 = no/don’t know) to six questions: 1) If you look carefully, you can know if someone has HIV (recoded), 2) Using condoms can prevent you from being infected by HIV, 3) A person who looks strong and healthy can have HIV, 4) A mother can transmit HIV through breastfeeding, 5) You can get HIV if you share utensils with someone who is infected (recoded), 6) If a mosquito bites you, it can infect you with HIV (recoded). The index was categorised into low (0–2), medium (3–4) and high knowledge (5–6).

  3. †An index measuring attitudes for HIV ranging from 1 to 12 was computed by adding answers (1 = positive attitude, 0 = negative attitude) to 12 questions: 1) would buy food from an HIV-positive shopkeeper, 2) HIV-positive teachers who are not ill should not be allowed to teach in school, 3) HIV-positive health workers should not be allowed to treat patients, 4) if a family member would become HIV-positive, I would want it a secret, 5) HIV/AIDS is the result of sinning, 6) it is a waste of money to train/educate someone with HIV, 7) one would be foolish to marry someone with HIV, 8) people with HIV/AIDS should not be ashamed, 9) health workers should treat people with AIDS as people with other illnesses, 10) people with HIV should be allowed to participated in social events, 11) it is reasonable for an employer to fire someone with AIDS, 12) people with HIV/AIDS do not deserve any support. The index was categorised into negative (1–4), medium (5–8) and positive attitudes towards HIV (9–12).

HIV knowledge index*
 Low (0–2 questions)15.69.71.00 <0.001
 Medium (3–4 questions)62.555.01.491.23–1.80
 High (5–6 questions)21.935.32.702.18–3.34
Attitudes to HIV index†
 Negative (1–4)6.65.31.00 <0.001
 Medium (5–8)50.542.01.190.92–1.54
 Positive (9–12)42.952.71.571.21–2.04
Table 2.   HIV testing by survey round (%)
 2007 (n = 1997)2009 (n = 2746)AOR95% CIP-value
  1. AOR, adjusted odds ratio; CI, confidence interval, bivariate analyses employed weighted data, and multivariate analyses employed unweighted data and controlled for district, age, ethnic group, religion, marital status, occupation, sellable assets.

Knows someone who is HIV+53.068.11.931.70–2.20<0.001
Knows someone who is HIV+ and on antiretroviral treatment38.562.62.782.44–3.16<0.001
What are your chances of getting HIV?
 No risk at all31.926.41.00 <0.001
 Small or moderate29.937.31.311.12–1.54 
 Great23.419.80.960.81–1.14 
 Already know my status14.916.51.341.11–1.63 
The participant thinks s/he can access ARVs if needed25.456.43.543.10–4.04<0.001
The participant thinks community members can access ARVs if needed39.959.12.161.90–2.44<0.001
Number of times tested for HIV
 Not tested66.351.01.00 <0.001
 117.721.31.461.25–1.72 
 29.613.41.811.49–2.21 
 3+6.414.32.822.26–3.52 
Ever tested for HIV with their partner11.724.72.351.97–2.79<0.001
Among those tested for HIV(n = 673)(n = 1345)   
Received result of last HIV test84.395.34.192.98–5.90<0.001
Table 3.   Sexual behaviour by survey round (%)
 2007 (n = 1997)2009 (n = 2746)AOR95% CIP-value
  1. AOR, adjusted odds ratio; CI, confidence interval, bivariate analyses employed weighted data, and multivariate analyses employed unweighted data and controlled for district, age, ethnic group, religion, marital status, occupation, sellable assets.

Ever had sex85.682.30.610.50–0.76<0.001
Among those who ever had sex
Lifetime sex partners(n = 1708)(n = 2258)  0.500
 151.953.61.00 
 216.717.61.050.87–1.27
 3–414.312.60.880.71–1.09
 5 or more17.216.21.040.84–1.28
Sex partners in the last 12 months(n = 1708)(n = 2258)   
 019.815.91.00 0.216
 164.271.01.060.87–1.28
 2 or more16.113.10.880.69–1.12
Sex partners in the last 6 months(n = 1708)(n = 2258)   
 031.922.61.00 <0.001
 160.273.01.321.21–1.56
 2 or more8.04.50.730.53–1.00
Current sex partners(n = 1708)(n = 2258)   
 017.520.71.00 <0.001
 175.873.90.640.53–0.78
 2 or more6.75.40.700.51–0.96
Condom use (last sex with last partner)(n = 1699)(n = 2255)   
 No74.575.11.00 0.004
 Yes25.524.91.281.08–1.52
Table 4.   Age at first sex by birth year and survey round among 18–25 year olds (median, interquartile range)
 TotalMenWomen
  1. P-value – results of test of significance from Cox regressions using unweighted data; controlled for district, ethnic group, religion, marital status, occupation, sellable assets (results from adjusted and unadjusted survival analyses were very similar).

Birth year2007 (n = 1120)2009 (n = 1292)P-value2007 (n = 472)2009 (n = 512)P-value2007 (n = 648)2009 (n = 780)P-value
1982–198319 (17–22) 0.00320 (18–22) 0.00219 (17–21) 0.186
1984–198519 (17–20)19 (17–21) 20 (18–21)20 (18–22) 18 (17–20)19 (17–21) 
1986–198718 (16–19)19 (17–20) 18 (16–19)19 (17–21) 18 (16–19)19 (17–20) 
1988–198917 (15–18)18 (17–19) 17 (15–18)18 (16–19) 17 (16–18)18 (17–19) 
1990–1991 17 (16–18)  17 (15–18)  17 (16–18) 
Table 5.   Sexual behaviour by survey round stratified by gender and age group, respectively (%)
Gender2007 (n = 873)2009 (n = 1211)AOR95% CIP-value2007 (n = 1123)2009 (n = 1535)AOR95% CIP-value
Male*    Female†    
  1. AOR, adjusted odds ratio; CI, confidence interval.

  2. Bivariate analyses employed weighted data.

  3. *Employed unweighted data and controlled for district, age, ethnic group, religion, marital status, occupation.

  4. †Employed unweighted data and controlled for district, age, ethnic group, religion, marital status, occupation, sellable assets.

  5. ‡Not controlled for religion, occupation.

  6. §No control variables (including district).

  7. ¶No control variables.

  8. **Employed unweighted data and controlled for district, ethnic group, religion, marital status, occupation, sellable assets.

  9. ††Employed unweighted data and controlled for district, ethnic group, religion, marital status, occupation.

  10. ‡‡Not controlled for ethnic group, religion, occupation.

Ever had sex80.776.40.580.44–0.76<0.00189.386.90.610.43–0.860.006
Lifetime sex partners(n = 705)(n = 925)   (n = 1003)(n = 1334)   
 127.626.91.00 0.38669.072.21.00 0.514
 215.218.01.100.78–1.57 17.717.40.970.77–1.21 
 3 or more57.355.10.900.68–1.18 13.310.50.850.65–1.12 
Sex partners in the last 12 months(n = 705)(n = 925)   (n = 1003)(n = 1334)   
 024.116.61.00  16.715.41.00 0.008‡
 147.756.41.120.82–1.530.56075.781.10.980.77–1.25 
 2 or more28.226.90.970.70–1.35 7.63.60.540.36–0.82 
Sex partners in the last 6 months(n = 705)(n = 925)   (n = 1003)(n = 1334)   
 035.823.31.00 0.00229.022.11.00 <0.001§
 149.367.41.391.05–1.83 68.076.91.501.25–1.80 
 2 or more14.99.40.760.52–1.12 3.01.00.450.24–0.84 
Current sex partners(n = 705)(n = 925)   (n = 1003)(n = 1334)   
 020.022.51.00 0.00215.719.41.00 0.350¶
 168.667.00.550.39–0.76 80.978.70.970.79–1.19 
 2 or more11.410.50.650.43–0.99 3.41.90.680.41–1.15 
Condom use (last sex, last partner)(n = 703)(n = 925)   (n = 996)(n = 1330)   
 No63.464.91.00 0.02282.382.21.00 0.092
 Yes36.635.11.351.05–1.76 17.717.81.220.97–1.53 
Age group18–24 years**25–44 years††
(n = 851)(n = 1161)   (n = 1145)(n = 1585)   
Ever had sex70.963.30.560.44–0.71<0.00196.596.20.750.47–1.220.253‡‡
Lifetime sex partners(n = 603)(n = 735)   (n = 1105)(n = 1524)   
 154.462.11.00 0.29650.549.61.00 0.793
 218.917.11.010.76–1.36 15.517.91.060.83–1.34 
 3 or more26.820.80.810.62–1.07 34.032.60.960.79–1.18 
Sex partners in the last 12 months(n = 603)(n = 735)   (n = 1105)(n = 1524)   
 022.116.81.00 0.19918.515.41.00 0.642
 158.368.61.040.77–1.39 67.372.20.980.76–1.26 
 2 or more19.614.60.780.54–1.12 14.212.40.860.62–1.21 
Sex partners in the last 6 months(n = 603)(n = 735)   (n = 1105)(n = 1524)   
 040.126.81.00 0.002‡‡27.420.51.00 0.005‡‡
 151.367.91.311.03–1.67 65.075.51.140.92–1.42 
 2 or more8.65.30.630.40–0.99 7.64.00.580.38–0.90 
Current sex partners(n = 603)(n = 735)   (n = 1105)(n = 1524)   
 021.428.71.00 <0.001‡‡15.416.81.00 0.039‡‡
 169.365.30.550.41–0.73 79.478.00.720.56–0.93 
 2 or more9.36.00.590.38–0.93 5.35.20.750.48–1.17 
Condom use (last sex, last partner)(n = 594)(n = 734)   (n = 1105)(n = 1520)   
 No68.168.91.00 0.04177.978.11.00 0.028
 Yes31.931.11.311.01–1.70 22.121.91.291.03–1.62 
Table 6.   Sexual behaviour by survey round and marital status (%)
 MarriedFormerly marriedNever married
2007 (n = 930)2009 (n = 1562)AOR*95% CIP-value2007 (n = 439)2009 (n = 392)AOR†95% CIP-value2007 (n = 600)2009 (n = 740)AOR‡95% CIP-value
  1. AOR, adjusted odds ratio; CI, confidence interval.

  2. *Employed unweighted data and controlled for district, age, ethnic group, religion, occupation, sellable assets.

  3. †Employed unweighted data and controlled for district, age, education, occupation.

  4. ‡Employed unweighted data and controlled for district, age, religion, occupation.

  5. §Not controlled for occupation.

  6. ¶No control variables.

  7. **No control variables (including district).

  8. ††Not controlled for religion, occupation.

Ever had sex    56.541.20.570.45–0.72<0.001
Lifetime sex partners          (n = 339)(n = 305)   
 157.758.61.00 0.80848.546.11.00 0.54840.438.11.00 0.040
 214.615.81.070.83–1.38 20.319.40.850.59–1.24 17.724.71.450.92–2.27 
 3 or more27.825.70.980.78–1.22 31.234.61.060.75–1.51 41.937.10.820.56–1.19 
Sex partners in the last 12 months          (n = 339)(n = 305)   
 010.38.51.00 0.58328.130.81.00 0.163§35.034.51.00 0.426
 179.281.31.160.85–1.57 53.755.80.960.69–1.35 36.337.70.990.67–1.46 
 2 or more10.510.21.070.71–1.61 18.213.40.650.40–1.05 28.727.70.780.52–1.18 
Sex partners in the last 6 months          (n = 339)(n = 305)   
 019.112.51.00 <0.001¶41.642.61.00 0.175**52.848.41.00 <0.135††
 174.584.11.561.23–1.97 50.153.61.070.80–1.42 35.441.30.940.66–1.32 
 2 or more6.43.50.780.48–1.26 8.43.90.570.30–1.11 11.910.30.590.35–0.99 
Current sex partners          (n = 339)(n = 305)   
 05.37.91.00 0.089¶30.642.31.00 0.007**34.158.21.00 <0.001††
 191.787.90.710.51–1.00 59.551.60.650.48–0.87 53.430.80.450.31–0.65 
 2 or more3.04.20.750.43–1.31 9.96.10.550.31–0.97 12.611.00.650.39–1.08 
Condom use (last sex, last partner)(n = 928)(n = 1559)   (n = 438)(n = 391)   (n = 354)(n = 285)   
 No85.684.91.00 0.52671.567.71.00 0.11847.634.51.00 <0.001
 Yes14.415.11.080.84–1.39 28.532.31.290.94–1.78 52.465.52.021.43–2.84 

Results are presented as percentages. Patterns noticed at the bivariate level were confirmed by multivariate analyses; odds ratios are mentioned only when there is a discrepancy between the bivariate and multivariate analyses.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

A total of 4,979 individuals participated in the baseline (79% of eligibles) and 2746 in the mid-term survey (87%). Analysis was restricted to districts included in both the baseline and mid-term surveys (baseline n = 1891 and all mid-term participants).

Baseline and mid-term survey participants had similar gender distributions (P = 0.153), educational status (P = 0.498) and onset of sexual activity (P = 0.933) (data not shown). Overall, women comprised 63% of participants; most participants had completed 4 years of secondary schooling (45%) or less (46%) and were sexually active (83%). Participants were younger in the baseline than in the mid-term survey (55%vs. 43% below 25 years, P < 0.001), less likely to be married (44%vs. 59%, P < 0.001) and unemployed (27%vs. 37%, P < 0.001). Participants at baseline were more likely to report being Christian (80%vs. 73%, P < 0.001) and less likely to be Shona (93%vs. 96%, P < 0.001). Mid-term survey participants were poorer, as indicated by the fewer number of sellable items in their household (43%vs. 38% had fewer than two of the nine items listed, P < 0.001).

Knowledge and attitudes to HIV

Human immunodeficiency virus knowledge had improved between 2009 and 2007 (35%vs. 22% answered correctly 5–6 of 6 questions, adjusted odds ratio (AOR) = 2.70, confidence interval (CI) = 2.18–3.34). Results suggested more favourable attitudes towards HIV (Table 1); for example, 53% in 2009 vs. 43% in 2007 have relatively positive attitudes towards HIV (9–12 of 12 positive statements, AOR = 1.57, CI = 1.21–2.04).

HIV testing norms and behaviours

Norms related to HIV testing were also more favourable in 2009 compared to 2007. More participants agreed that most people in their communities want to get tested for HIV (56%vs. 45%, P < 0.001) and those who want to get tested discuss this with their spouses/partners (57%vs. 54%, P = 0.060) (data not shown).

A greater proportion reported knowing someone who is HIV positive in 2009 than in 2007 (68%vs. 53%, P < 0.001) and on anti retoviral treatment (ART) (63%vs. 39%, P < 0.001) (Table 2). Confidence about accessing ART had increased; 56% of participants said they would be able to access ART in 2009 vs. 25% in 2007 (P < 0.001). More people had tested for HIV (49%vs. 34%, P < 0.001) and received the result of their last HIV test (95%vs. 84%, P < 0.001). Repeat HIV testing was reported more commonly (28% in 2009 vs. 16% in 2007, P < 0.001).

Reported HIV testing increased between 2007 and 2009 for both women (36%vs. 55%, P < 0.001) and men (41%vs. 31%, P = 0.021). In 2009, women were more likely to report testing with a partner (27%vs. 22%, P = 0.002), repeat testing (33%vs. 21% tested more than once, P < 0.001) and confidence about accessing ART (65%vs. 46%, P < 0.001) than men.

Sexual behaviour

Over this two-year period, there was a change in reported sexual behaviour (Table 3). There was a reduction in the reported number of sexual partners in the last 6 months and at the time of the survey. Specifically, 73% of mid-term sexually active survey participants compared to 60% of baseline participants reported being monogamous in the previous 6 months (P < 0.001). In addition, there was a significant increase in reported condom use between 2007 and 2009 (AOR = 1.28, CI = 1.08–1.52).

In Table 4, we compared the reported age at first sex of 18–25 year olds by survey round. Given the 2-year gap between the surveys, we compared age at first sex of men and women from the same 2-year birth cohorts. Overall, younger participants had sex at an earlier age compared to those from later cohorts. For example, the median age at first sex among women born in 1990–1991 was 17 years, compared to those born in 1982–1983 (19 years). Theoretically, participants from the same birth cohorts should report similar ages at first sex across the two surveys (as they basically represent approximately the same populations at different times). However, both men and women from the same birth cohort tend to report older ages at first sex in the mid-term survey (e.g. 1988–1989 birth cohort reports a median age of 17 years in 2007 and 18 years in 2009).

In 2009, sexually active women reported fewer lifetime partners than men (28%vs. 73% had ≥2 partners) and partners in the last 6 (1%vs. 9%) and 12 months (4%vs. 27%) (Table 5). Women reported less frequent condom use at last sex than their male peers (18%vs. 35%). Changes in sexual behaviours were most marked among young people; for example, 29% of sexually active youth had no sexual partner at the time of the survey in 2009 compared to 21% in 2007 (P < 0.001), while there was little change among adults (17% in 2009 vs. 15% in 2007, P = 0.039). Table 6 presents reported sexual behavioural data by marital status and shows some statistically significant positive changes for never married participants: 31% of them had one sexual partner at the time of survey in 2009 compared to 53% in 2007, and 58% of them had no partners in 2009 vs. 34% in 2007 (P < 0.001) (among previously married 52%vs. 60% and 42%vs. 31%, respectively, P = 0.007). Moreover, while there was little change in reported condom use among married (15%vs. 14%, P = 0.526) and previously married participants (32%vs. 29%, P = 0.118), 66% of never married participants reported using a condom at last sex in 2009 compared to 52% in 2007 (P < 0.001).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Data from these two surveys offer us an opportunity to examine recent changes taking place within Zimbabwe where HIV prevalence rates have been falling (Gregson et al. 2010). The data show an increase in voluntary HIV testing, safer sexual behaviours (most marked among young people), increase in HIV knowledge and more favourable attitudes towards HIV. These data support modelling studies that suggest that the decline in HIV prevalence observed in Zimbabwe from 17.7% in 2006 to 16.1% in 2009 among antenatal attendees (Zimbabwe Central Statistical Office & Macro International Inc. 2006; Ministry of Health & Child Welfare 2009) is likely partly attributable to behaviour change (Gregson et al. 2006; Zimbabwe Central Statistical Office & Macro International Inc. 2006).

We note an important shift in HIV testing practices with individuals being almost twice as likely to have tested for HIV in 2009 than in 2007, as well as being over twice as likely to have tested as a couple or to have tested more than once. This suggests that HIV testing is becoming normalised, likely as a result of the increased availability of antiretroviral treatment, the introduction of provider initiated testing and counselling at all health facilities and the increased coverage of prevention of mother-to-child transmission programmes (Makwiza et al. 2006).

Our data indicate an overall reduction in reporting of number of sexual partners in the previous 6 months and concurrent sexual partnerships and an increase in reported condom use. For example, condom use at last sex was twice as high in 2009 vs. 2007 among those who had never been married. Previous studies in the region place the bulk of sexual behaviour change among young people (Gregson et al. 2006, 2010; Mahomva et al. 2006). Our data largely support these findings.

Strengths of this study include its sample size coupled with its population-based survey design. We also found similar populations living in the districts over this two-year period. Together, these provide strength to this analysis. The data were collected primarily in rural areas, which limit their generalisability to urban populations. Nevertheless, Zimbabwe remains a predominantly rural country. In 2009, the NBCP had been implemented in four of the six districts included in this analysis, which might explain some of the reported behaviour change. However, in 2007, the NBCP started being implemented in 26 districts throughout Zimbabwe and in 2010, expanded to additional 34 districts. Hence, data presented here are representative of the coverage of NBCP activities as a whole.

While these data are encouraging, they need to be interpreted with minor caution as they are self-reported and hence subject to recall bias and social desirability (Cowan et al. 2002; Mensch et al. 2008). Recall bias may be particularly salient when comparisons are made between youth and adults regarding age of sexual debut. While we attempted to minimise these biases by using appropriate questionnaire delivery methods, they are unlikely to have been excluded completely (Langhaug et al. 2009; Minnis et al. 2009). Despite similar populations, there were differences in the characteristics of participants from the two survey rounds; but these differences were controlled for during multivariate analyses. The interval between data collection times was relatively short, which makes it difficult to provide a longer-term picture of the potential changes taking place within the country. However, as Zimbabwe went through dramatic economic challenges during this short time, it is reassuring to know that behaviours seem to have continued on a positive trajectory. For the mid-term survey, we purposively selected the EAs from Phase I districts where the NBCP had been most intensively implemented. Hence, some of the reported behaviour change may be due to effects of the NBCP. However, the data presented in this study cannot be used to evaluate the programme. The NBCP evaluation using the baseline and end line surveys is underway.

In summary, comparative results from two surveys conducted in the same districts in Zimbabwe in 2007 and 2009 indicate an increase in HIV testing and some reduction in risky sexual practices. The level of change observed here is impressive given the short time frame in which these changes occurred.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References