Evaluation of bias in HIV seroprevalence estimates from national household surveys

HIV estimation using population-based surveys with non-response: A partial identification approach

HIV estimation using data from the demographic and health surveys (DHS) is limited by the presence of non-response and test refusals. Conventional adjustments such as imputation require the data to be missing at random. Methods that use instrumental variables allow the possibility that prevalence is different between the respondents and non-respondents, but their performance depends critically on the validity of the instrument. Using Manski's partial identification approach, we form instrumental variable bounds for HIV prevalence from a pool of candidate instruments. Our method does not require all candidate instruments to be valid. We use a simulation study to evaluate and compare our method against its competitors. We illustrate the proposed method using DHS data from Zambia, Malawi and Kenya. Our simulations show that imputation leads to seriously biased results even under mild violations of non-random missingness. Using worst case identification bounds that do not make assumptions about the non-response mechanism is robust but not informative. By taking the union of instrumental variable bounds balances informativeness of the bounds and robustness to inclusion of some invalid instruments. Non-response and refusals are ubiquitous in population based HIV data such as those collected under the DHS. Partial identification bounds provide a robust solution to HIV prevalence estimation without strong assumptions. Union bounds are significantly more informative than the worst case bounds without sacrificing credibility.

Trend and principal components of HIV/AIDS among adults in SSA

This study aimed to identify the most important principal components (PCs) that contribute to the prevalence and change of HIV/AIDS in 44 SSA and data from different national and international datasets. The study estimated HIV prevalence, trend, and principal component analysis (PCA). Using the elbow method, the number of important PCs and contributions was identified. The quality of representation was checked, and more contributing variables for most important PCs were identified. Finally, the status by prevalence, the progress by trend, the more influenced component by PCA, and the more influenced variable with quality of representation by PCs were reported. The study found that HIV prevalence varied significantly, with 30 of the countries showed good progress/decline. Four PCs accounted for 51% of the total variance. Literacy, cohabitation, media exposure, and HIV status awareness are highly contributing factors. Based on these findings, a gap-based response will help reduce the burden of HIV.

Estimation of HIV prevalence and burden in Nigeria: a Bayesian predictive modelling study

Background

The cost of population-based surveys is high and obtaining funding for a national population-based survey may take several years, with follow-up surveys taking up to five years. Survey-based prevalence estimates are prone to bias owing to survey non-participation, as not all individuals eligible to participate in a survey may be reached, and some of those who are contacted do not consent to HIV testing. This study describes how Bayesian statistical modeling may be used to estimate HIV prevalence at the state level in a reliable and timely manner.

Methods

We analysed national HIV testing services (HTS) data for Nigeria from October 1, 2020, to September 30, 2021, to derive state-level HIV seropositivity rates. We used a Bayesian linear model with normal prior distribution and Markov Chain Monte Carlo approach to estimate HIV state-level prevalence for the 36 states +1 FCT in Nigeria. Our outcome variable was the HIV seropositivity rates and we adjusted for demographic, economic, biological, and societal covariates collected from the 2018 Nigeria HIV/AIDS Indicator and Impact Survey (NAIIS), 2018 Nigeria Demographic and Health Survey (NDHS) and 2016-17 Multiple Indicator Cluster Surveys (MICS). The estimated population of 15-49 years olds in each state was multiplied by estimates from the estimated prevalence to generate state-level HIV burden.

Findings

Our estimated national HIV prevalence was 2.1% (95% CI: 1.5-2.7%) among adults aged 15-49 years in Nigeria, which corresponds to approximately 2 million people living with HIV, compared to previous national HIV prevalence estimates of 1.4% from the 2018 NAIIS and UNAIDS estimation and projection package PLHIV estimation of 1.8 million in 2022. Our modelled HIV prevalence in Nigeria varies by state, with Benue (5.7%, 95% CI: 5.0-6.3) having the highest prevalence, followed by Rivers (5.2%, 95% CI: 4.6-5.8%), Akwa Ibom (3.5%, 95% CI: 2.9-4.1%), Edo (3.4%, 95% CI: 2.9-4.0%) and Taraba (3.0%, 95% CI: 2.6-3.7%) placing fourth and fifth, respectively. Jigawa had the lowest HIV prevalence (0.3%), which was consistent with prior estimates.

Interpretation

This model provides a comprehensive and flexible use of evidence to estimate state-level HIV seroprevalence for Nigeria using program data and adjusting for explanatory variables. Thus, investment in program data for HIV surveillance will provide reliable estimates for HIV sub-national monitoring and improve planning and interventions for epidemiologic control.

Funding

This article was made possible by the support of the American people through the United States Agency for International Development (USAID) under the U.S. President's Emergency Plan for AIDS Relief (PEPFAR).

Analytical methods used in estimating the prevalence of HIV/AIDS from demographic and cross-sectional surveys with missing data: a systematic review

BACKGROUND

Sero- prevalence studies often have a problem of missing data. Few studies report the proportion of missing data and even fewer describe the methods used to adjust the results for missing data. The objective of this review was to determine the analytical methods used for analysis in HIV surveys with missing data.

METHODS

We searched for population, demographic and cross-sectional surveys of HIV published from January 2000 to April 2018 in Pub Med/Medline, Web of Science core collection, Latin American and Caribbean Sciences Literature, Africa-Wide Information and Scopus, and by reviewing references of included articles. All potential abstracts were imported into Covidence and abstracts screened by two independent reviewers using pre-specified criteria. Disagreements were resolved through discussion. A piloted data extraction tool was used to extract data and assess the risk of bias of the eligible studies. Data were analysed through a quantitative approach; variables were presented and summarised using figures and tables.

RESULTS

A total of 3426 citations where identified, 194 duplicates removed, 3232 screened and 69 full articles were obtained. Twenty-four studies were included. The response rate for an HIV test of the included studies ranged from 32 to 96% with the major reason for the missing data being refusal to consent for an HIV test. Complete case analysis was the primary method of analysis used, multiple imputations 11(46%) was the most advanced method used, followed by the Heckman's selection model 9(38%). Single Imputation and Instrumental variables method were used in only two studies each, with 13(54%) other different methods used in several studies. Forty-two percent of the studies applied more than two methods in the analysis, with a maximum of 4 methods per study. Only 6(25%) studies conducted a sensitivity analysis, while 11(46%) studies had a significant change of estimates after adjusting for missing data.

CONCLUSION

Missing data in survey studies is still a problem in disease estimation. Our review outlined a number of methods that can be used to adjust for missing data on HIV studies; however, more information and awareness are needed to allow informed choices on which method to be applied for the estimates to be more reliable and representative.

HIV surveillance based on routine testing data from antenatal clinics in Malawi (2011-2018): measuring and adjusting for bias from imperfect testing coverage

Supplemental Digital Content is available in the text

Objective:

The use of routinely collected data from prevention of mother-to-child transmission programs (ANC-RT) has been proposed to monitor HIV epidemic trends. This poses several challenges for surveillance, one of them being that women may opt-out of testing and/or test stock-outs may result in inconsistent service availability. In this study, we sought to empirically quantify the relationship between imperfect HIV testing coverage and HIV prevalence among pregnant women from ANC-RT data.

Design:

We used reports from the ANC Register of all antenatal care (ANC) sites in Malawi (2011–2018), including 49 244 monthly observations, from 764 facilities, totaling 4 375 777 women.

Methods:

Binomial logistic regression models with facility-level fixed effects and marginal standardization were used to assess the effect of testing coverage on HIV prevalence.

Results:

Testing coverage increased from 78 to 98% over 2011–2018. We estimated that, had testing coverage been perfect, prevalence would have been 0.4% point lower (95% CI 0.3–0.5%) than the 7.9% observed prevalence, a relative overestimation of 6%. Bias in HIV prevalence was the highest in 2012, when testing coverage was lowest (72%), resulting in a relative overestimation of HIV prevalence of 15% (95% CI 12–17%). Overall, adjustments for imperfect testing coverage led to a subtler decline in HIV prevalence over 2011--2018.

Conclusion:

Malawi achieved high coverage of routine HIV testing in recent years. Nevertheless, imperfect testing coverage can lead to overestimation of HIV prevalence among pregnant women when coverage is suboptimal. ANC-RT data should be carefully evaluated for changes in testing coverage and completeness when used to monitor epidemic trends.

Geospatial assessment of the voluntary medical male circumcision programme in Tanzania, 2011-2016

Introduction

Tanzania is one of the 14 priority countries in sub-Saharan Africa scaling up voluntary medical male circumcision (VMMC) for HIV prevention. In this study, we assessed the progress of VMMC by evaluating changes in the spatial structure of male circumcision (MC) prevalence and identifying age groups with low MC uptake.

Methods

We use data from two waves of the Demographic and Health Survey (DHS) conducted in Tanzania in 2011–2012 and 2015–2016. MC incidence rate was estimated using a method developed to calculate incidence rates from two successive cross-sectional surveys. Continuous surface maps of MC prevalence were generated for both DHS waves and compared with identified areas with high MC prevalence changes and high density of uncircumcised males.

Results

National MC prevalence in Tanzania increased from 73.5% in 2011–2012 to 80.0% in 2015–2016. The estimated national MC incidence rate was 4.6 circumcisions per 100 person-years (py). The lowest circumcision rate was observed in males aged 20–24 years, with 0.61 circumcisions per 100 py. An estimated 1 567 253 males aged 15–49 years residing in low-MC prevalence areas were uncircumcised in 2015–2016.

Conclusion

Tanzania has shown substantial progress in the implementation of VMMC. However, extensive spatial variation of MC prevalence still exists in the country, with some areas having an MC prevalence <60%. Here, we identified locations where VMMC needs to be intensified to reach the ~1.5 million uncircumcised males age 15–49 living in these low-MC areas, particularly for men aged 20–34.

Refusal bias in HIV data from the Demographic and Health Surveys: Evaluation, critique and recommendations

Non-response is a commonly encountered problem in many population-based surveys. Broadly speaking, non-response can be due to refusal or failure to contact the sample units. Although both types of non-response may lead to bias, there is much evidence to indicate that it is much easier to reduce the proportion of non-contacts than to do the same with refusals. In this article, we use data collected from a nationally representative survey under the Demographic and Health Surveys program to study non-response due to refusals to HIV testing in Malawi. We review existing estimation methods and propose novel approaches to the estimation of HIV prevalence that adjust for refusal behaviour. We then explain the data requirement and practical implications of the conventional and proposed approaches. Finally, we provide some general recommendations for handling non-response due to refusals and we highlight the challenges in working with Demographic and Health Surveys and explore different approaches to statistical estimation in the presence of refusals. Our results show that variation in the estimated HIV prevalence across different estimators is due largely to those who already know their HIV test results. In the case of Malawi, variations in the prevalence estimates due to refusals for women are larger than those for men.

Towards UNAIDS Fast-Track goals: targeting priority geographic areas for HIV prevention and care in Zimbabwe

INTRODUCTION

Zimbabwe has made substantial progress towards the Joint United Nations Programme on HIV/AIDS (UNAIDS) targets of 90-90-90 by 2020, with 73% of people living with HIV diagnosed, 87% of those diagnosed on antiretroviral therapy (ART) and 86% of those on ART virally suppressed. Despite this exceptional response, more effort is needed to completely achieve the UNAIDS targets. Here, we conducted a detailed spatial analysis of the geographical structure of the HIV epidemic in Zimbabwe to include geographical prioritization as a key component of their overall HIV intervention strategy.

METHODS

Data were obtained from Zimbabwe Demographic and Health Survey (ZDHS) conducted in 2015 as well as estimations from the Zimbabwe Population-Based HIV Impact Assessment (ZIMPHIA) 2016 report, and other published literature. Data were used to produce high-resolution maps of HIV prevalence. Using these maps combined with the population density maps, we mapped the HIV-infected population lacking ART coverage and viral suppression.

RESULTS

HIV maps for both sexes illustrated similar geographical variation of HIV prevalence within the country. HIV-infected populations lacking ART coverage and viral suppression were concentrated in the main cities and urban settlements such as Bulawayo, Harare, Ruwa and Chitungwiza.

CONCLUSION

Our study showed extensive local variation in HIV disease burden across Zimbabwe for both women and men. The high-resolution maps generated here identified areas wherein high density of HIV-infected individuals are lacking ART coverage and viral suppression. These results suggest that there is need to tailor HIV programmes to address specific local needs to efficiently achieve epidemic control in Zimbabwe.

The association between being currently in school and HIV prevalence among young women in nine eastern and southern African countries

INTRODUCTION

Interventions to keep adolescent girls and young women in school, or support their return to school, are hypothesised to also reduce HIV risk. Such interventions are included in the DREAMS combination package of evidence-based interventions. Although there is evidence of reduced risky sexual behaviours, the impact on HIV incidence is unclear. We used nationally representative surveys to investigate the association between being in school and HIV prevalence.

METHODS

We analysed Demographic and Health Survey data from nine DREAMS countries in sub-Saharan Africa restricted to young women aged 15-19 (n = 20,429 in total). We used logistic regression to assess cross-sectional associations between being in school and HIV status and present odds ratios adjusted for age, socio-economic status, residence, marital status, educational attainment and birth history (aOR). We investigated whether associations seen differed across countries and by age.

RESULTS

HIV prevalence (1.0%-9.8%), being currently in school (50.0%-72.6%) and the strength of association between the two, varied between countries. We found strong evidence that being currently in school was associated with a reduced odds of being HIV positive in Lesotho (aOR: 0.37; 95%CI: 0.17-0.79), Swaziland (aOR: 0.32; 95%CI: 0.17-0.59), and Uganda (aOR: 0.48: 95%CI: 0.29-0.80) and no statistically significant evidence for this in Kenya, Malawi, Mozambique, Tanzania, Zambia or Zimbabwe.

CONCLUSIONS

Although the relationship is not uniform across countries or over time, these data are supportive of the hypothesis that young women in school are at lower risk of being HIV positive than those who leave school in some sub-Saharan African settings. There is a possibility of reverse causality, with pre-existing HIV infection leading to school drop-out. Further investigation of the contextual factors behind this variation will be important in interpreting the results of HIV prevention interventions promoting retention in school.

Role of survey response rates on valid inference: an application to HIV prevalence estimates

Background

Nationally-representative surveys suggest that females have a higher prevalence of HIV than males in most African countries. Unfortunately, these results are made on the basis of surveys with non-ignorable missing data. This study evaluates the impact that differential survey nonresponse rates between males and females can have on the point estimate of the HIV prevalence ratio of these two classifiers.

Methods

We study 29 Demographic and Health Surveys (DHS) from 2001 to 2010. Instead of employing often used multiple imputation models with a Missing at Random assumption that may not hold in this setting, we assess the effect of ignoring the information contained in the missing HIV information for males and females through three proposed statistical measures. These measures can be used in settings where the interest is comparing the prevalence of a disease between two groups. The proposed measures do not utilize parametric models and can be implemented by researchers of any level. They are: (1) an upper bound on the potential bias of the usual practise of using reported HIV prevalence estimates that ignore subjects who have missing HIV outcomes. (2) Plausible range intervals to account for nonresponses, without any additional parametric modeling assumptions. (3) Prevalence ratio inflation factors to correct the point estimate of the HIV prevalence ratio, if estimates of nonresponders’ HIV prevalences were known.

Results

In 86% of countries, males have higher upper bounds of HIV prevalence than females, this is consonant with males possibly having higher infection rates than females. Additionally, 74% of surveys have a plausible range that crosses 1.0, suggesting a plausible equivalence between male and female HIV prevalences.

Conclusions

It is quite reasonable to conclude that there is so much DHS nonresponse in evaluating the HIV status question, that existing data is plausibly generated by the situation where the virus is equally distributed between the sexes.

Reaching the 'first 90': Gaps in coverage of HIV testing among people living with HIV in 16 African countries

BACKGROUND

UNAIDS has recently proposed a set of three ambitious targets that, if achieved, are predicted to end the AIDS epidemic by 2030. The targets, known as 90-90-90, call for 90% of people living with HIV (PLHIV) to know their status, 90% of PLHIV to receive antiretroviral therapy, and 90% of those on antiretroviral therapy to achieve viral suppression by the year 2020. We examine the first of these targets, focusing on sub-Saharan Africa, the region of the world most affected by HIV, to measure the proportion of PLHIV estimated to know their HIV status, and to identify background and behavioral characteristics significantly associated with gaps in ever testing among PLHIV.

METHODS AND FINDINGS

We analyze cross-sectional population-based data from the Demographic and Health Surveys (DHS) and AIDS Indicator Surveys (AIS) fielded since 2010 in 16 sub-Saharan African countries where voluntary serological testing was recently conducted: Burkina Faso, Cameroon, Chad, Cote d'Ivoire, Ethiopia, Gabon, Lesotho, Malawi, Namibia, Rwanda, Sierra Leone, Tanzania, Togo, Uganda, Zambia, and Zimbabwe. Survey response rates averaged 95.0% (range 89.3-99.5%), while consent to serotesting averaged 94.9% (range 88.7-99.6%). This study, which includes more than 14,000 respondents living with HIV, finds that 69% of PLHIV in the average study country have ever been tested for HIV (range 34-95%). Based on timing of the last test and on ART coverage, we estimate that 54% of PLHIV in the average country are aware of their status (range 26-84%). Adjusted logistic regression finds that men (median adjusted odds ratio [AOR] = 0.38), adults with less than primary education (median AOR = 0.31), and adolescents (median AOR = 0.32) are consistently less likely to have ever been tested for HIV than women, adults with secondary and above education, and adults age 30-39, respectively. In most countries unadjusted logistic regression also finds significant gaps in testing among the poorest groups and those reporting never having had sex.

CONCLUSION

The fact that an average of 54% of PLHIV in these 16 countries are estimated to know their status reflects encouraging progress. However, not only is this average far short of the 90% target set by UNAIDS for 2020, but it also implies that in the average study country nearly one-half of PLHIV are unable to access lifesaving care and treatment because they are unaware that they are HIV-positive. Several gaps in HIV testing coverage exist, particularly among adolescents, the least educated, and men. While the need to target demographic groups at greatest risk of HIV continues, additional interventions focused on reaching men and on reaching socially vulnerable populations such as adolescents, the poorest, and the least educated are essential.

Mapping the spatial variability of HIV infection in Sub-Saharan Africa: Effective information for localized HIV prevention and control

Under the premise that in a resource-constrained environment such as Sub-Saharan Africa it is not possible to do everything, to everyone, everywhere, detailed geographical knowledge about the HIV epidemic becomes essential to tailor programmatic responses to specific local needs. However, the design and evaluation of national HIV programs often rely on aggregated national level data. Against this background, here we proposed a model to produce high-resolution maps of intranational estimates of HIV prevalence in Kenya, Malawi, Mozambique and Tanzania based on spatial variables. The HIV prevalence maps generated highlight the stark spatial disparities in the epidemic within a country, and localize areas where both the burden and drivers of the HIV epidemic are concentrated. Under an era focused on optimal allocation of evidence-based interventions for populations at greatest risk in areas of greatest HIV burden, as proposed by the Joint United Nations Programme on HIV/AIDS (UNAIDS) and the United States President's Emergency Plan for AIDS Relief (PEPFAR), such maps provide essential information that strategically targets geographic areas and populations where resources can achieve the greatest impact.

Geographical Patterns of HIV Sero-Discordancy in High HIV Prevalence Countries in Sub-Saharan Africa

INTRODUCTION

Variation in the proportion of individuals living in a stable HIV sero-discordant partnership (SDP), and the potential drivers of such variability across sub Saharan Africa (SSA), are still not well-understood. This study aimed to examine the spatial clustering of HIV sero-discordancy, and the impact of local variation in HIV prevalence on patterns of sero-discordancy in high HIV prevalence countries in SSA.

METHODS

We described the spatial patterns of sero-discordancy among stable couples by analyzing Demographic and Health Survey data from Cameroon, Kenya, Lesotho, Tanzania, Malawi, Zambia, and Zimbabwe. We identified spatial clusters of SDPs in each country through a Kulldorff spatial scan statistics analysis. After a geographical cluster was identified, epidemiologic measures of sero-discordancy were calculated and analyzed.

RESULTS

Spatial clusters with significantly high numbers of SDPs were identified and characterized in Kenya, Malawi, and Tanzania, and they largely overlapped with the clusters with high HIV prevalence. There was a positive correlation between HIV prevalence and the proportion of SDPs among all stable couples across within and outside clusters. Conversely, there was a negative, but weak and not significant, correlation between HIV prevalence and the proportion of SDPs among all stable couples with at least one HIV-infected individual in the partnership.

DISCUSSION

There does not appear to be distinct spatial patterns for HIV sero-discordancy that are independent of HIV prevalence patterns. The variation of the sero-discordancy measures with HIV prevalence across clusters and outside clusters demonstrated similar patterns to those observed at the national level. The spatial variable does not appear to be a fundamental nor independent determinant of the observed patterns of sero-discordancy in high HIV prevalence countries in SSA.

Closer to 90-90-90. The cascade of care after 10 years of ART scale-up in rural Malawi: a population study

INTRODUCTION

The antiretroviral therapy (ART) programme supported by Médecins Sans Frontières in the rural Malawian district of Chiradzulu was one of the first in sub-Saharan Africa to scale up ART delivery in 2002. After more than a decade of continuous involvement, we conducted a population survey to evaluate the cascade of care, including population viral load, in the district.

METHODS

A cross-sectional household-based survey was conducted between February and May 2013. Using a multistage cluster sampling method, we recruited all individuals aged 15 to 59 years living in 4125 randomly selected households. Each consenting individual was interviewed and tested for HIV at home. All participants who tested positive had their CD4 count and viral load measured. The LAg-Avidity assay was used to distinguish recent from long-term infections. Viral suppression was defined as a viral load below 1000 copies/mL.

RESULTS

Of 8271 individuals eligible for the study, 7269 agreed to participate and were tested for HIV (94.1% inclusion for women and 80.3% for men). Overall HIV prevalence and incidence were 17.0% (95% CI 16.1 to 17.9) and 0.39 new cases per 100 person-years (95% CI 0.0 to 0.77), respectively. Coverage at the other steps along the HIV care cascade was as follows: 76.7% (95% CI 74.4 to 79.1) had been previously diagnosed, 71.2% (95% CI 68.6 to 73.6) were under care and 65.8% (95% CI 62.8 to 68.2) were receiving ART. Finally, the proportion of participants who were HIV positive with a viral load ≤ 1000 copies/mL reached 61.8% (95% CI 59.0 to 64.5).

CONCLUSIONS

This study demonstrates that a high level of population viral suppression and low incidence can be achieved in high HIV prevalence and resource-limited settings.

Are Geographical "Cold Spots" of Male Circumcision Driving Differential HIV Dynamics in Tanzania?

BACKGROUND

Growing evidence suggests significant geographic clustering of male circumcision (MC) in Tanzania. The impact of spatial heterogeneity of MC prevalence on HIV transmission dynamics in this country is not well documented. The aim of this study was to assess the spatial association between MC and HIV infection in Tanzania.

METHODS

Data from three Demographic and Health Survey rounds conducted in Tanzania were analyzed to identify spatial associations between MC and HIV using bivariate local indicators of spatial association (LISA). Spatial clusters with low MC prevalence (MC cold spots) were identified using scan statistics. HIV incidence rates for males and females within and outside the MC cold spots were calculated.

RESULTS

Local indicators of spatial association analysis indicated a significant association between MC and HIV in the northern and southwestern regions of Tanzania. Scan statistics identified two MC cold spots in the same locations. Males located outside the MC cold spots had the lowest HIV incidence rate at 0.28 per 100 person-years at risk (pyar). HIV incidence in females located outside the MC cold spots increased from 0.40/100 pyar during 2004-2008 to 0.68/100 pyar in 2008-2012.

CONCLUSION

Our study provides evidence for a geographic association between MC and HIV in Tanzania. MC could be one of the key factors driving the geographical distribution of the HIV epidemic in the country. Furthermore, in areas where most males are circumcised, the HIV infection burden could be concentrating in the female population. Therefore, along with the voluntary medical MC program, efforts targeting the female population should also be considered.

Factors associated with HIV infection among sexually experienced adolescents in Africa: a pooled data analysis

The article examines the factors associated with HIV status among adolescents aged 15-19 years in 13 African countries: Côte d'Ivoire, Democratic Republic of Congo, Guinea, Kenya, Liberia, Mali, Malawi, Rwanda, Sierra Leone, Swaziland, Tanzania, Zambia and Zimbabwe. The data were derived from demographic and health surveys or AIDS indicator surveys conducted between 2004 and 2009. The levels of HIV prevalence among adolescents varied considerably across the countries. There was significantly higher HIV prevalence among female adolescents as compared with their male counterparts. For male adolescents, circumcision was the only variable significantly associated with HIV status. Nonetheless, the data suggest that the association between male circumcision and HIV status may be exaggerated. Indeed, regional-level random effects became insignificant once male circumcision was introduced into the estimated models, indicating a strong correlation between unmeasured regional-level factors and male circumcision. For female adolescents, multiple sexual partnerships, time elapsed since sexual debut, marital status, household wealth, and the regional prevalence of male circumcision were strongly and positively associated with HIV status. Moreover, for female adolescents there appear to be significant unmeasured variables operating at the regional level which influence the levels of HIV infection. The implications of the findings for HIV-prevention programming, policy and research are discussed.

Association between HCV infection and diabetes type 2 in Egypt: is it time to split up?

PURPOSE

There is a conflicting evidence about the association between hepatitis C virus (HCV) infection and diabetes mellitus. The objective of this study was to assess this association in Egypt, the country with the highest HCV prevalence in the world.

METHODS

The source of data was from the Egypt Demographic and Health Survey conducted in 2008. Using multivariable logistic regression analyses to account for known confounders, the association was investigated at two levels']: (1) HCV exposure (HCV antibody status) and diabetes mellitus and (2) diabetes mellitus and chronic HCV infection (HCV RNA status) among HCV-exposed individuals.

RESULTS

We found no evidence for an association between HCV antibody status and diabetes (adjusted odds ratio [OR] = 0.87; 95% confidence interval [CI], 0.63-1.19). However, among HCV-exposed individuals, we found an evidence for an association between diabetes and active HCV infection (adjusted OR = 2.44, 95% CI, 1.30-4.57).

CONCLUSIONS

Although it does not appear that HCV exposure and diabetes are linked, there might be an association between diabetes and chronic HCV infection. The HCV-diabetes relationship may be more complex than previously anticipated. Therefore, a call for an "amicable divorce" to the HCV-diabetes relationship could be premature.

Have the educated changed HIV risk behaviours more in Africa?

Theory predicts that when new health information becomes available, more educated individuals may adopt healthy behaviours sooner, resulting in lower morbidity and mortality among the highly educated. This may be the case for HIV in sub-Saharan Africa: Recent empirical work shows that incidence is falling and the reduction is concentrated in more educated populations. However, it is unclear whether the educated have indeed adopted HIV risk-reducing behaviours to a greater extent than the less educated. I used two rounds of demographic and health surveys (DHS) in eight African countries to examine whether HIV-related behavioural change over time is greater among the more highly educated. I examined changes in condom use, age of marriage, number of partners, extramarital partnerships, and HIV testing. Results showed that education has a robust positive association with condom use and HIV testing, but also with having more sexual partners. I found that the probability of HIV testing increased more between rounds among the more educated, relative to the less educated. More educated men also appeared to have larger reductions in the number of sexual partners and there was evidence that younger, more educated women may be marrying earlier than their predecessors did. The education gradient did not change significantly over time for condom use. These changes in behaviour may signal a shift in the future burden of the epidemic towards more marginalised and less educated populations.

Trends in socioeconomic inequalities in HIV prevalence among young people in seven countries in eastern and southern Africa

Background

In Eastern and Southern Africa, HIV prevalence was highest among higher socioeconomic groups during the 1990s. It has been suggested that this is changing, with HIV prevalence falling among higher-educated groups while stable among lower-educated groups. A multi-country analysis has not been undertaken.

Methods

We analysed data on socio-demographic factors and HIV infection from 14 nationally representative surveys of adults aged 15-24 (seven countries, two surveys each, 4-8 years apart). Sample sizes ranged from 2,408-12,082 (72,135 total). We used logistic regression to assess gender-stratified associations between highest educational level attended and HIV status in each survey, adjusting for age and urban/rural setting. We tested for interactions with urban/rural setting and age. Our primary hypothesis was that higher education became less of a risk factor for HIV over time. We tested for interaction between survey-year and the education-HIV association in each country and all countries pooled.

Findings

In Ethiopia and Malawi, HIV prevalence was higher in more educated women in both surveys. In Lesotho, Kenya and Zimbabwe, HIV prevalence was lower in higher educated women in both surveys. In Ethiopia, HIV prevalence fell among no and secondary educated women only (interaction p<0·01). Only among young men in Tanzania there was some evidence that the association between education and HIV changed over time (p=0·07). Pooled analysis found little evidence for an interaction between survey year and the education-HIV association among men (p=0·60) or women (p=0·37).

Interpretation

The pattern of prevalent HIV infection among young adults by level of education in different sub-Saharan African countries was heterogeneous. There was little statistical evidence that this pattern changed between 2003-5 and 2008-12. Explanations for the social epidemiology of HIV in Africa will need to account for time-trends and inter-country differences.

On the assumption of bivariate normality in selection models: a Copula approach applied to estimating HIV prevalence

BACKGROUND

Heckman-type selection models have been used to control HIV prevalence estimates for selection bias when participation in HIV testing and HIV status are associated after controlling for observed variables. These models typically rely on the strong assumption that the error terms in the participation and the outcome equations that comprise the model are distributed as bivariate normal.

METHODS

We introduce a novel approach for relaxing the bivariate normality assumption in selection models using copula functions. We apply this method to estimating HIV prevalence and new confidence intervals (CI) in the 2007 Zambia Demographic and Health Survey (DHS) by using interviewer identity as the selection variable that predicts participation (consent to test) but not the outcome (HIV status).

RESULTS

We show in a simulation study that selection models can generate biased results when the bivariate normality assumption is violated. In the 2007 Zambia DHS, HIV prevalence estimates are similar irrespective of the structure of the association assumed between participation and outcome. For men, we estimate a population HIV prevalence of 21% (95% CI = 16%-25%) compared with 12% (11%-13%) among those who consented to be tested; for women, the corresponding figures are 19% (13%-24%) and 16% (15%-17%).

CONCLUSIONS

Copula approaches to Heckman-type selection models are a useful addition to the methodological toolkit of HIV epidemiology and of epidemiology in general. We develop the use of this approach to systematically evaluate the robustness of HIV prevalence estimates based on selection models, both empirically and in a simulation study.

Using interviewer random effects to remove selection bias from HIV prevalence estimates

BACKGROUND

Selection bias in HIV prevalence estimates occurs if non-participation in testing is correlated with HIV status. Longitudinal data suggests that individuals who know or suspect they are HIV positive are less likely to participate in testing in HIV surveys, in which case methods to correct for missing data which are based on imputation and observed characteristics will produce biased results.

METHODS

The identity of the HIV survey interviewer is typically associated with HIV testing participation, but is unlikely to be correlated with HIV status. Interviewer identity can thus be used as a selection variable allowing estimation of Heckman-type selection models. These models produce asymptotically unbiased HIV prevalence estimates, even when non-participation is correlated with unobserved characteristics, such as knowledge of HIV status. We introduce a new random effects method to these selection models which overcomes non-convergence caused by collinearity, small sample bias, and incorrect inference in existing approaches. Our method is easy to implement in standard statistical software, and allows the construction of bootstrapped standard errors which adjust for the fact that the relationship between testing and HIV status is uncertain and needs to be estimated.

RESULTS

Using nationally representative data from the Demographic and Health Surveys, we illustrate our approach with new point estimates and confidence intervals (CI) for HIV prevalence among men in Ghana (2003) and Zambia (2007). In Ghana, we find little evidence of selection bias as our selection model gives an HIV prevalence estimate of 1.4% (95% CI 1.2% - 1.6%), compared to 1.6% among those with a valid HIV test. In Zambia, our selection model gives an HIV prevalence estimate of 16.3% (95% CI 11.0% - 18.4%), compared to 12.1% among those with a valid HIV test. Therefore, those who decline to test in Zambia are found to be more likely to be HIV positive.

CONCLUSIONS

Our approach corrects for selection bias in HIV prevalence estimates, is possible to implement even when HIV prevalence or non-participation is very high or very low, and provides a practical solution to account for both sampling and parameter uncertainty in the estimation of confidence intervals. The wide confidence intervals estimated in an example with high HIV prevalence indicate that it is difficult to correct statistically for the bias that may occur when a large proportion of people refuse to test.

Socioeconomic gradients in internalized stigma among 4,314 persons with HIV in sub-Saharan Africa

The stigma attached to HIV is a major public health problem given its adverse impacts on HIV prevention and on the psychosocial wellbeing of persons with HIV. In this study, I apply a novel method to data from the Demographic and Health Surveys to identify persons with HIV who were aware of their seropositivity at the time of the survey. The pooled dataset includes 4,314 persons with HIV in Cameroon, Ethiopia, Gabon, Kenya, Lesotho, Malawi, Rwanda, Swaziland, Tanzania, Uganda, Zambia, and Zimbabwe. My findings indicate that nearly one-fifth of study participants provided survey responses consistent with internalization of stigmatizing beliefs. Furthermore, in multivariable regression models, striking socioeconomic gradients in internalized stigma were observed. A clear implication of my findings is that the adverse health and psychosocial impacts of HIV stigma are likely concentrated among those with the fewest socioeconomic resources for managing and resisting it.

Assessing and adjusting for differences between HIV prevalence estimates derived from national population-based surveys and antenatal care surveillance, with applications for Spectrum 2013

Objective(s):

To assess differences between HIV prevalence estimates derived from national population surveys and antenatal care (ANC) surveillance sites and to improve the calibration of ANC-derived estimates in Spectrum 2013 to more appropriately account for differences between these data.

Design:

Retrospective analysis of national population survey and ANC surveillance data from 25 countries with generalized epidemics in sub-Saharan Africa and 8 countries with concentrated epidemics.

Methods:

Adult national population survey and ANC surveillance HIV prevalence estimates were compared for all available national population survey data points for the years 1999–2012. For sub-Saharan Africa, a mixed-effects linear regression model determined whether the relationship between national population and ANC estimates was constant across surveys. A new calibration method was developed to incorporate national population survey data directly into the likelihood for HIV prevalence in countries with generalized epidemics. Results were used to develop default rules for adjusting ANC data for countries with no national population surveys.

Results:

ANC surveillance data typically overestimate population prevalence, although a wide variation, particularly in rural areas, is observed across countries and survey years. The new calibration method yields similar point estimates to previous approaches, but leads to an average 44% increase in the width of 95% uncertainty intervals.

Conclusion:

Important biases remain in ANC surveillance data for HIV prevalence. The new approach to model-fitting in Spectrum 2013 more appropriately accounts for this bias when producing national estimates in countries with generalized epidemics. In countries with concentrated epidemics, local sex ratios should be used to calibrate ANC surveillance estimates.

Spatial epidemiology of hepatitis C virus infection in Egypt: analyses and implications

Egypt has the highest hepatitis C virus (HCV) prevalence in the world (14.7%). The drivers of the HCV epidemic in Egypt are not well understood, but the mass parenteral antischistosomal therapy (PAT) campaigns in the second half of the 20th century are believed to be the determinant of the high prevalence. We studied HCV exposure in Egypt at a microscale through spatial mapping and epidemiological description of HCV clustering. The source of data was the 2008 Egypt Demographic and Health Survey. We identified clusters with high and low HCV prevalence and high and low PAT exposure using Kulldorff spatial scan statistics. Correlations across clusters were estimated, and each cluster age-specific HCV prevalence was described. We identified six clusters of high HCV prevalence, three clusters of low HCV prevalence, five clusters of high PAT exposure, and four clusters of low PAT exposure. HCV prevalence and PAT exposure were not significantly associated across clusters (Pearson correlation coefficient [PCC] = 0.36; 95% confidence interval [CI] -0.12 to 0.71). Meanwhile, there was a strong association between HCV prevalence in individuals older than 30 years of age (who could have been exposed to PAT) and HCV prevalence in individuals 30 years of age or younger (who could not have been exposed to PAT) (PCC = 0.81; 95% CI 0.55-0.93).

CONCLUSION

The findings illustrate a spatial variation in HCV exposure in Egypt. The observed clustering was suggestive of an array of iatrogenic risk factors, besides past PAT exposure, and ongoing transmission. The role of PAT exposure in the HCV epidemic could have been overstated. Our findings support the rationale for spatially prioritized interventions.

Refusal bias in the estimation of HIV prevalence

In 2007, UNAIDS corrected estimates of global HIV prevalence downward from 40 million to 33 million based on a methodological shift from sentinel surveillance to population-based surveys. Since then, population-based surveys are considered the gold standard for estimating HIV prevalence. However, prevalence rates based on representative surveys may be biased because of nonresponse. This article investigates one potential source of nonresponse bias: refusal to participate in the HIV test. We use the identity of randomly assigned interviewers to identify the participation effect and estimate HIV prevalence rates corrected for unobservable characteristics with a Heckman selection model. The analysis is based on a survey of 1,992 individuals in urban Namibia, which included an HIV test. We find that the bias resulting from refusal is not significant for the overall sample. However, a detailed analysis using kernel density estimates shows that the bias is substantial for the younger and the poorer population. Nonparticipants in these subsamples are estimated to be three times more likely to be HIV-positive than participants. The difference is particularly pronounced for women. Prevalence rates that ignore this selection effect may be seriously biased for specific target groups, leading to misallocation of resources for prevention and treatment.

Spatial variability in HIV prevalence declines in several countries in sub-Saharan Africa

Evidence suggests substantial declines in HIV prevalence in parts of sub-Saharan Africa. However, the observed aggregate declines at the national level may obscure local variations in the temporal dynamics of the infection. Using spatial scan statistics, we identified marked spatial variability in the within-country declines in HIV prevalence in Tanzania, Malawi, Kenya, and Zimbabwe. Our study suggests that the declines in the national HIV prevalence in some of the SSA countries may not be representative of downward trends in prevalence in areas of high HIV prevalence, as much as the result of sharp declines in prevalence in areas of already low HIV prevalence. Our findings provide insights for resource allocation and HIV prevention interventions in these countries.

Adjusting the HIV prevalence for non-respondents using mortality rates in an open cohort in northwest Tanzania

Objective

To estimate HIV prevalence in adults who have not tested for HIV using age‐specific mortality rates and to adjust the overall population HIV prevalence to include both tested and untested adults.

Methods

An open cohort study was established since 1994 with demographic surveillance system (DSS) and five serological surveys conducted. Deaths from Kisesa DSS were used to estimate mortality rates and 95% confidence intervals by HIV status for 3‐ 5‐year periods (1995–1999, 2000–2004, and 2005–2009). Assuming that mortality rates in individuals who did not test for HIV are similar to those in tested individuals, and dependent on age, sex and HIV status and HIV, prevalence was estimated.

Results

In 1995–1999, mortality rates (per 1000 person years) were 43.7 (95% CI 35.7–53.4) for HIV positive, 2.6 (95% CI 2.1–3.2) in HIV negative and 16.4 (95% CI 14.4–18.7) in untested. In 2000–2004, mortality rates were 43.3 (95% CI 36.2–51.9) in HIV positive, 3.3 (95% CI 2.8–4.0) in HIV negative and 11.9 (95% CI 10.5–13.6) in untested. In 2005–2009, mortality rates were 30.7 (95% CI 24.8–38.0) in HIV positive, 4.1 (95% CI 3.5–4.9) in HIV negative and 5.7 (95% CI 5.0–6.6) in untested residents. In the three survey periods (1995–1999, 2000–2004, 2005–2009), the adjusted period prevalences of HIV, including the untested, were 13.5%, 11.6% and 7.1%, compared with the observed prevalence in the tested of 6.0%, 6.8 and 8.0%. The estimated prevalence in the untested was 33.4%, 21.6% and 6.1% in the three survey periods.

Conclusion

The simple model was able to estimate HIV prevalence where a DSS provided mortality data for untested residents.

Sources of HIV incidence among stable couples in sub-Saharan Africa

Introduction

The recent availability of efficacious prevention interventions among stable couples offers new opportunities for reducing HIV incidence in sub-Saharan Africa. Understanding the dynamics of HIV incidence among stable couples is critical to inform HIV prevention strategy across sub-Saharan Africa.

Methods

We quantified the sources of HIV incidence arising among stable couples in sub-Saharan Africa using a cohort-type mathematical model parameterized by nationally representative data. Uncertainty and sensitivity analyses were incorporated.

Results

HIV incidence arising among stable concordant HIV-negative couples contribute each year, on average, 29.4% of total HIV incidence; of those, 22.5% (range: 11.1%–39.8%) are infections acquired by one of the partners from sources external to the couple, less than 1% are infections acquired by both partners from external sources within a year and 6.8% (range: 3.6%–11.6%) are transmissions to the uninfected partner in the couple in less than a year after the other partner acquired the infection from an external source. The mean contribution of stable HIV sero-discordant couples to total HIV incidence is 30.4%, with most of those, 29.7% (range: 9.1%–47.9%), being due to HIV transmissions from the infected to the uninfected partner within the couple. The remaining incidence, 40.2% (range: 23.7%–64.6%), occurs among persons not in stable couples.

Conclusions

Close to two-thirds of total HIV incidence in sub-Saharan Africa occur among stable couples; however, only half of this incidence is attributed to HIV transmissions from the infected to the uninfected partner in the couple. The remaining incidence is acquired through extra-partner sex. Substantial reductions in HIV incidence can be achieved only through a prevention approach that targets all modes of HIV exposure among stable couples and among individuals not in stable couples.

Mapping HIV clustering: a strategy for identifying populations at high risk of HIV infection in sub-Saharan Africa

Background

The geographical structure of an epidemic is ultimately a consequence of the drivers of the epidemic and the population susceptible to the infection. The ‘know your epidemic’ concept recognizes this geographical feature as a key element for identifying populations at higher risk of HIV infection where prevention interventions should be targeted. In an effort to clarify specific drivers of HIV transmission and identify priority populations for HIV prevention interventions, we conducted a comprehensive mapping of the spatial distribution of HIV infection across sub-Saharan Africa (SSA).

Methods

The main source of data for our study was the Demographic and Health Survey conducted in 20 countries from SSA. We identified and compared spatial clusters with high and low numbers of HIV infections in each country using Kulldorff spatial scan test. The test locates areas with higher and lower numbers of HIV infections than expected under spatial randomness. For each identified cluster, a likelihood ratio test was computed. A P-value was determined through Monte Carlo simulations to evaluate the statistical significance of each cluster.

Results

Our results suggest stark geographic variations in HIV transmission patterns within and across countries of SSA. About 14% of the population in SSA is located in areas of intense HIV epidemics. Meanwhile, another 16% of the population is located in areas of low HIV prevalence, where some behavioral or biological protective factors appear to have slowed HIV transmission.

Conclusions

Our study provides direct evidence for strong geographic clustering of HIV infection across SSA. This striking pattern of heterogeneity at the micro-geographical scale might reflect the fact that most HIV epidemics in the general population in SSA are not far from their epidemic threshold. Our findings identify priority geographic areas for HIV programming, and support the need for spatially targeted interventions in order to maximize the impact on the epidemic in SSA.

Estimating HIV prevalence from surveys with low individual consent rates: annealing individual and pooled samples

Many HIV prevalence surveys are plagued by the problem that a sizeable number of surveyed individuals do not consent to contribute blood samples for testing. One can ignore this problem, as is often done, but the resultant bias can be of sufficient magnitude to invalidate the results of the survey, especially if the number of non-responders is high and the reason for refusing to participate is related to the individual’s HIV status. One reason for refusing to participate may be for reasons of privacy. For those individuals, we suggest offering the option of being tested in a pool. This form of testing is less certain than individual testing, but, if it convinces more people to submit to testing, it should reduce the potential for bias and give a cleaner answer to the question of prevalence. This paper explores the logistics of implementing a combined individual and pooled testing approach and evaluates the analytical advantages to such a combined testing strategy. We quantify improvements in a prevalence estimator based on this combined testing strategy, relative to an individual testing only approach and a pooled testing only approach. Minimizing non-response is key for reducing bias, and, if pooled testing assuages privacy concerns, offering a pooled testing strategy has the potential to substantially improve HIV prevalence estimates.

National HIV prevalence estimates for sub-Saharan Africa: controlling selection bias with Heckman-type selection models

Objectives

Population-based HIV testing surveys have become central to deriving estimates of national HIV prevalence in sub-Saharan Africa. However, limited participation in these surveys can lead to selection bias. We control for selection bias in national HIV prevalence estimates using a novel approach, which unlike conventional imputation can account for selection on unobserved factors.

Methods

For 12 Demographic and Health Surveys conducted from 2001 to 2009 (N=138 300), we predict HIV status among those missing a valid HIV test with Heckman-type selection models, which allow for correlation between infection status and participation in survey HIV testing. We compare these estimates with conventional ones and introduce a simulation procedure that incorporates regression model parameter uncertainty into confidence intervals.

Results

Selection model point estimates of national HIV prevalence were greater than unadjusted estimates for 10 of 12 surveys for men and 11 of 12 surveys for women, and were also greater than the majority of estimates obtained from conventional imputation, with significantly higher HIV prevalence estimates for men in Cote d'Ivoire 2005, Mali 2006 and Zambia 2007. Accounting for selective non-participation yielded 95% confidence intervals around HIV prevalence estimates that are wider than those obtained with conventional imputation by an average factor of 4.5.

Conclusions

Our analysis indicates that national HIV prevalence estimates for many countries in sub-Saharan African are more uncertain than previously thought, and may be underestimated in several cases, underscoring the need for increasing participation in HIV surveys. Heckman-type selection models should be included in the set of tools used for routine estimation of HIV prevalence.

Prevalences of sexually transmitted infections in young adults and female sex workers in Peru: a national population-based survey

Background

We assessed prevalences of seven sexually transmitted infections (STIs) in Peru, stratified by risk behaviours, to help to define care and prevention priorities.

Methods

In a 2002 household-based survey of the general population, we enrolled randomly selected 18–29-year-old residents of 24 cities with populations greater than 50 000 people. We then surveyed female sex workers (FSWs) in these cities. We gathered data for sexual behaviour; vaginal specimens or urine for nucleic acid amplification tests for Neisseria gonorrhoeae, Chlamydia trachomatis, and Trichomonas vaginalis; and blood for serological tests for syphilis, HIV, and (in subsamples) herpes simplex virus 2 (HSV2) and human T-lymphotropic virus. This study is a registered component of the PREVEN trial, number ISRCTN43722548.

Findings

15 261 individuals from the general population and 4485 FSWs agreed to participate in our survey. Overall prevalence of infection with HSV2, weighted for city size, was 13·5% in men, 13·6% in women, and 60·6% in FSWs (all values in FSWs standardised to age composition of women in the general population). The prevalence of C trachomatis infection was 4·2% in men, 6·5% in women, and 16·4% in FSWs; of T vaginalis infection was 0·3% in men, 4·9% in women, and 7·9% in FSWs; and of syphilis was 0·5% in men, 0·4% in women, and 0·8% in FSWs. N gonorrhoeae infection had a prevalence of 0·1% in men and women, and of 1·6% in FSWs. Prevalence of HIV infection was 0·5% in men and FSWs, and 0·1% in women. Four (0·3%) of 1535 specimens were positive for human T-lymphotropic virus 1. In men, 65·0% of infections with HIV, 71·5% of N gonorrhoeae, and 41·4% of HSV2 and 60·9% of cases of syphilis were in the 13·3% who had sex with men or unprotected sex with FSWs in the past year. In women from the general population, 66·7% of infections with HIV and 16·7% of cases of syphilis were accounted for by the 4·4% who had been paid for sex by any of their past three partners.

Interpretation

Defining of high-risk groups could guide targeting of interventions for communicable diseases—including STIs—in the general Peruvian population.

Funding

Wellcome Trust-Burroughs Wellcome Fund Infectious Disease Initiative and US National Institutes of Health.

Distinct HIV discordancy patterns by epidemic size in stable sexual partnerships in sub-Saharan Africa

Objective

To describe patterns of HIV infection among stable sexual partnerships across sub-Saharan Africa (SSA).

Methods

The authors defined measures of HIV discordancy and conducted a comprehensive quantitative assessment of discordancy among stable partnerships in 20 countries in SSA through an analysis of the Demographic and Health Survey data.

Results

HIV prevalence explained at least 50% of the variation in HIV discordancy, with two distinct patterns of discordancy emerging based on HIV prevalence being roughly smaller or larger than 10%. In low-prevalence countries, approximately 75% of partnerships affected by HIV are discordant, while only about half of these are discordant in high-prevalence countries. Out of each 10 HIV infected persons, two to five are engaged in discordant partnerships in low-prevalence countries compared with one to three in high-prevalence countries. Among every 100 partnerships in the population, one to nine are affected by HIV and zero to six are discordant in low-prevalence countries compared with 16–45 and 9–17, respectively, in high-prevalence countries. Finally, zero to four of every 100 sexually active adults are engaged in a discordant partnership in low-prevalence countries compared with six to eight in high-prevalence countries.

Conclusions

In high-prevalence countries, a large fraction of stable partnerships were affected by HIV and half were discordant, whereas in low-prevalence countries, fewer stable partnerships were affected by HIV but a higher proportion of them were discordant. The findings provide a global view of HIV infection among stable partnerships in SSA but imply complex considerations for rolling out prevention interventions targeting discordant partnerships.

The effect of changes in condom usage and antiretroviral treatment coverage on human immunodeficiency virus incidence in South Africa: a model-based analysis

This study aims to assess trends in human immunodeficiency virus (HIV) incidence in South Africa, and to assess the extent to which prevention and treatment programmes have reduced HIV incidence. Two models of the South African HIV epidemic, the STI (sexually transmitted infection)-HIV Interaction model and the ASSA2003 AIDS and Demographic model, were adapted. Both models were fitted to age-specific HIV prevalence data from antenatal clinic surveys and household surveys, using a Bayesian approach. Both models suggest that HIV incidence in 15-49 year olds declined significantly between the start of 2000 and the start of 2008: by 27 per cent (95% CI: 21-32%) in the STI-HIV model and by 31 per cent (95% CI: 23-39%) in the ASSA2003 model, when expressed as a percentage of incidence rates in 2000. By 2008, the percentage reduction in incidence owing to increased condom use was 37 per cent (95% CI: 34-41%) in the STI-HIV model and 23 per cent (95% CI: 14-34%) in the ASSA2003 model. Both models also estimated a small reduction in incidence owing to antiretroviral treatment by 2008. Increased condom use therefore appears to be the most significant factor explaining the recent South African HIV incidence decline.

Factors associated with repeated refusal to participate in longitudinal population-based HIV surveillance in rural South Africa: an observational study, regression analyses

BACKGROUND

For many estimation purposes, individuals who repeatedly refuse to participate in longitudinal HIV surveillance pose a bigger threat to valid inferences than individuals who participate at least occasionally. We investigate the determinants of repeated refusal to consent to HIV testing in a population-based longitudinal surveillance in rural South Africa.

METHODS

We used data from two years (2005 & 2006) of the annual HIV surveillance conducted by the Africa Centre for Health and Population Studies, linking the HIV surveillance data to demographic and socioeconomic data. The outcome for the analysis was "repeated refusal". Demographic variables included sex, age, highest educational attainment, and place of residence. We also included a measure of wealth and the variable "ever had sex". To compare the association of each variable with the outcome, unadjusted odds ratios and standard errors were estimated. Multivariable logistic regression was used to estimate adjusted odds ratios and their standard errors. Data were analyzed using STATA 10.0.

RESULTS

Of 15,557 eligible individuals, 46% refused to test for HIV in both rounds. Males were significantly more likely than females to repeatedly refuse testing. Holding all other variables constant, individuals in the middle age groups were more likely to repeatedly refuse testing compared with younger and older age groups. The odds of repeated refusal increased with increasing level of education and relative wealth. People living in urban areas were significantly more likely to repeatedly refuse an HIV test than people living in peri-urban or rural areas. Compared to those who had ever had sex, both males and females who had not yet had sex were significantly more likely to refuse to participate.

CONCLUSIONS

The likelihood of repeated refusal to test for HIV in this longitudinal surveillance increases with education, wealth, urbanization, and primary sexual abstinence. Since the factors determining repeated HIV testing refusal are likely associated with HIV status, it is critical that selection effects are controlled for in the analysis of HIV surveillance data. Interventions to increase consent to HIV testing should consider targeting the relatively well educated and wealthy, people in urban areas, and individuals who have not yet sexually debuted.

Incidence of HIV in Windhoek, Namibia: demographic and socio-economic associations

Objective

To estimate HIV incidence and prevalence in Windhoek, Namibia and to analyze socio-economic factors related to HIV infection.

Method

In 2006/7, baseline surveys were performed with 1,753 private households living in the greater Windhoek area; follow-up visits took place in 2008 and 2009. Face-to-face socio-economic questionnaires were administrated by trained interviewers; biomedical markers were collected by nurses; GPS codes of household residences were recorded.

Results

The HIV prevalence in the population (aged>12 years) was 11.8% in 2006/7 and 14.6% in 2009. HIV incidence between 2007 and 2009 was 2.4 per 100 person year (95%CI = 1.9–2.9). HIV incidence and prevalence were higher in female populations. HIV incidence appeared non-associated with any socioeconomic factor, indicating universal risk for the population. For women a positive trend was found between low per-capita consumption and HIV acquisition. A HIV knowledge score was strongly associated with HIV incidence for both men and women. High HIV prevalence and incidence was concentrated in the north-western part of the city, an area with lower HIV knowledge, higher HIV risk perception and lower per-capita consumption.

Discussion

The HIV incidence and prevalence figures do not suggest a declining epidemic in Windhoek. Higher vulnerability of women is recorded, most likely related to economic dependency and increasing transactional sex in Namibia. The lack of relation between HIV incidence and socio-economic factors confirms HIV risks for the overall urban community. Appropriate knowledge is strongly associated to lower HIV incidence and prevalence, underscoring the importance of continuous information and education activities for prevention of infection. Geographical areas were identified that would require prioritized HIV campaigning.

Women's and men's fertility preferences and contraceptive behaviors by HIV status in 10 sub-Saharan African countries

This article draws on biomarker data from Demographic and Health Surveys (2003-2007) in 10 sub-Saharan African countries to examine differences in fertility preferences and contraceptive behaviors by HIV status for women and men, taking into account whether or not they probably know their HIV status. The objective is to determine if there are common patterns in the associations between these variables across several countries. Women's and men's fertility preferences and contraceptive behaviors are relatively similar across HIV status and probable knowledge of that status. However, two consistent differences emerge in some of the countries: HIV-positive women who probably know their status are less likely to want more children and are more likely to be using male condoms than women who are HIV-negative and probably know it. A similar association is observed for men for condom use but not for limiting childbearing. Other factors unrelated to HIV status seem to be shaping women's and men's unmet demand for contraception and use of methods other than the condom.

Monitoring trends in HIV prevalence among young people, aged 15 to 24 years, in Manicaland, Zimbabwe

Background

In June 2001, the United Nations General Assembly Special Session (UNGASS) set a target of reducing HIV prevalence among young women and men, aged 15 to 24 years, by 25% in the worst-affected countries by 2005, and by 25% globally by 2010. We assessed progress toward this target in Manicaland, Zimbabwe, using repeated household-based population serosurvey data. We also validated the representativeness of surveillance data from young pregnant women, aged 15 to 24 years, attending antenatal care (ANC) clinics, which UNAIDS recommends for monitoring population HIV prevalence trends in this age group. Changes in socio-demographic characteristics and reported sexual behaviour are investigated.

Methods

Progress towards the UNGASS target was measured by calculating the proportional change in HIV prevalence among youth and young ANC attendees over three survey periods (round 1: 1998-2000; round 2: 2001-2003; and round 3: 2003-2005). The Z-score test was used to compare differences in trends between the two data sources. Characteristics of participants and trends in sexual risk behaviour were analyzed using Student's and two-tailed Z-score tests.

Results

HIV prevalence among youth in the general population declined by 50.7% (from 12.2% to 6.0%) from round 1 to 3. Intermediary trends showed a large decline from round 1 to 2 of 60.9% (from 12.2% to 4.8%), offset by an increase from round 2 to 3 of 26.0% (from 4.8% to 6.0%). Among young ANC attendees, the proportional decline in prevalence of 43.5% (from 17.9% to 10.1%) was similar to that in the population (test for differences in trend: p value = 0.488) although ANC data significantly underestimated the population prevalence decline from round 1 to 2 (test for difference in trend: p value = 0.003) and underestimated the increase from round 2 to 3 (test for difference in trend: p value = 0.012). Reductions in risk behaviour between rounds 1 and 2 may have been responsible for general population prevalence declines.

Conclusions

In Manicaland, Zimbabwe, the 2005 UNGASS target to reduce HIV prevalence by 25% was achieved. However, most prevention gains occurred before 2003. ANC surveillance trends overall were an adequate indicator of trends in the population, although lags were observed. Behaviour data and socio-demographic characteristics of participants are needed to interpret ANC trends.

The Spectrum projection package: improvements in estimating incidence by age and sex, mother-to-child transmission, HIV progression in children and double orphans

Background

The Spectrum program is used to estimate key HIV indicators from the trends in incidence and prevalence estimated by the Estimation and Projection Package or the Workbook. These indicators include the number of people living with HIV, new infections, AIDS deaths, AIDS orphans, the number of adults and children needing treatment, the need for prevention of mother-to-child transmission and the impact of antiretroviral treatment on survival. The UNAIDS Reference Group on Estimates, Models and Projections regularly reviews new data and information needs, and recommends updates to the methodology and assumptions used in Spectrum.

Methods

The latest update to Spectrum was used in the 2009 round of global estimates. This update contains new procedures for estimating: the age and sex distribution of adult incidence, new child infections occurring around delivery or through breastfeeding, the survival of children by timing of infection and the number of double orphans.

Fertility of HIV-infected women: insights from Demographic and Health Surveys

Objective

To analyse the data from Demographic and Health Survey (DHS) linking HIV serological data and fertility, and to recommend new age-specific fertility rate (ASFR) ratios of HIV-positive women to HIV-negative women to be used in the WHO/UNAIDS HIV estimates.

Methods

The authors obtained ASFRs and ASFR ratios from HIV-infected women and HIV-uninfected women from 20 DHS surveys collected between 2003 and 2007. The authors then computed mean and median ASFR ratios, and performed general linear modelling to estimate the overall ASFR ratio for each age group, and finally performed an analysis to identify determinants to national ASFR ratio for women aged 15–19 years old.

Results

The overall ASFR ratio estimated by regression was highest among women aged 15–19 years old (ASFR ratio of 1.20), and decreased with age (ASFR ratio of 0.76, 0.71, 0.65, 0.59 and 0.53, respectively for every 5-year bracket of 20–44 years old). The mean and median of ASFR ratios among 20 countries were similar to the results above. The analysis suggested that country-specific proportion of sexually active women aged 15–19 years old could be a determinant of the ASFR ratio for this group. The higher proportion of sexually active women in this group suggested a lower ASFR ratio in young women.

Conclusion

This study confirmed previous findings of a high ASFR ratio among women younger than 20 years old, and of decreasing ASFR ratios with age among those aged 20 years or older, based on nationally representative data from a large number of countries. The study updated the assumptions used in the annual WHO/UNAIDS HIV epidemic estimates. Countries should compute their own ASFR ratio of 15–19-year-old women accounting for the sexually active rate among young females.

Comparison of HIV prevalence estimates for Zimbabwe from antenatal clinic surveillance (2006) and the 2005-06 Zimbabwe Demographic and Health Survey

Objective

To assess whether HIV surveillance data from pregnant women attending antenatal care (ANC) clinics in Zimbabwe represent infection levels in the general population.

Methods

HIV prevalence estimates from ANC surveillance sites in 2006 were compared with estimates from the corresponding Zimbabwe Demographic and Health Survey 2005–06 (ZDHS) clusters using geographic information systems.

Results

The ANC HIV prevalence estimate (17.9%, 95% CI 17.0%–18.8%) was similar to the ZDHS estimates for all men and women aged 15–49 years (18.1%, 16.9%–18.8%), for pregnant women (17.5%, 13.9%–21.9%), and for ANC attendees living within 30 km of ANC surveillance sites (19.9%, 17.1%–22.8%). However, the ANC surveillance estimate (17.9%) was lower than the ZDHS estimates for all women (21.1%, 19.7%–22.6%) and for women living within 30 km catchment areas of ANC surveillance sites (20.9%, 19.4%–22.3%). HIV prevalence in ANC sites classified as urban and rural was significantly lower than in sites classified as “other”.

Conclusions

Periodic population surveys can be used to validate ANC surveillance estimates. In Zimbabwe, ANC surveillance provides reliable estimates of HIV prevalence among men and women aged 15–49 years in the general population. Three classifications of ANC sites (rural/urban/other) should be used when generating national HIV estimates.

Adjusting HIV prevalence for survey non-response using mortality rates: an application of the method using surveillance data from Rural South Africa

Background

The main source of HIV prevalence estimates are household and population-based surveys; however, high refusal rates may hinder the interpretation of such estimates. The study objective was to evaluate whether population HIV prevalence estimates can be adjusted for survey non-response using mortality rates.

Methodology/Principal Findings

Data come from the longitudinal Africa Centre Demographic Information System (ACDIS), in rural South Africa. Mortality rates for persons tested and not tested in the 2005 HIV surveillance were available from routine household surveillance. Assuming HIV status among individuals contacted but who refused to test (non-response) is missing at random and mortality among non-testers can be related to mortality of those tested a mathematical model was developed. Non-parametric bootstrapping was used to estimate the 95% confidence intervals around the estimates. Mortality rates were higher among untested (16.9 per thousand person-years) than tested population (11.6 per thousand person-years), suggesting higher HIV prevalence in the former. Adjusted HIV prevalence for females (15–49 years) was 31.6% (95% CI 26.1–37.1) compared to observed 25.2% (95% CI 24.0–26.4). For males (15–49 years) adjusted HIV prevalence was 19.8% (95% CI 14.8–24.8), compared to observed 13.2% (95% CI 12.1–14.3). For both sexes (15–49 years) combined, adjusted prevalence was 27.5% (95% CI 23.6–31.3), and observed prevalence was 19.7% (95% CI 19.6–21.3). Overall, observed prevalence underestimates the adjusted prevalence by around 7 percentage points (37% relative difference).

Conclusions/Significance

We developed a simple approach to adjust HIV prevalence estimates for survey non-response. The approach has three features that make it easy to implement and effective in adjusting for selection bias than other approaches. Further research is needed to assess this approach in populations with widely available HIV treatment (ART).

A decline in new HIV infections in South Africa: estimating HIV incidence from three national HIV surveys in 2002, 2005 and 2008

BACKGROUND

Three national HIV household surveys were conducted in South Africa, in 2002, 2005 and 2008. A novelty of the 2008 survey was the addition of serological testing to ascertain antiretroviral treatment (ART) use.

METHODS AND PRINCIPAL FINDINGS

We used a validated mathematical method to estimate the rate of new HIV infections (HIV incidence) in South Africa using nationally representative HIV prevalence data collected in 2002, 2005 and 2008. The observed HIV prevalence levels in 2008 were adjusted for the effect of antiretroviral treatment on survival. The estimated "excess" HIV prevalence due to ART in 2008 was highest among women 25 years and older and among men 30 years and older. In the period 2002-2005, the HIV incidence rate among men and women aged 15-49 years was estimated to be 2.0 new infections each year per 100 susceptible individuals (/100pyar) (uncertainty range: 1.2-3.0/100pyar). The highest incidence rate was among 15-24 year-old women, at 5.5/100pyar (4.5-6.5). In the period 2005-2008, incidence among men and women aged 15-49 was estimated to be 1.3/100 (0.6-2.5/100pyar), although the change from 2002-2005 was not statistically significant. However, the incidence rate among young women aged 15-24 declined by 60% in the same period, to 2.2/100pyar, and this change was statistically significant. There is evidence from the surveys of significant increases in condom use and awareness of HIV status, especially among youth.

CONCLUSIONS

Our analysis demonstrates how serial measures of HIV prevalence obtained in population-based surveys can be used to estimate national HIV incidence rates. We also show the need to determine the impact of ART on observed HIV prevalence levels. The estimation of HIV incidence and ART exposure is crucial to disentangle the concurrent impact of prevention and treatment programs on HIV prevalence.

Heterogeneous and decreasing HIV prevalence among women seeking antenatal care in Kinshasa, Democratic Republic of Congo

BACKGROUND

We examined HIV prevalence trends over 4.5 years among women receiving antenatal care in Kinshasa, Democratic Republic of Congo, by geographic location, clinic management and urbanicity.

METHODS

Quarterly proportions and 95% confidence intervals (CIs) of pregnant women with HIV positive results were determined using aggregate service provision and uptake data from 22 maternity units that provided vertical HIV prevention services from October 2004 to March 2009. Assuming linearity, proportions were assessed for trend via the Cochran-Armitage test. Multivariable binomial regression was used to describe detailed prevalence trends.

RESULTS

HIV testing was offered to 220,006 pregnant women; 210,348 (95.6%) agreed to be tested and 191,216 (90.9%) received their results. A total of 3999 women were found to be HIV positive, a prevalence of 1.90% (95% CI: 1.84-1.96%). The median quarterly proportion of women testing positive for HIV was 1.94% (range: 1.44-2.44%). Prevalence was heterogeneous in terms of maternity management, urbanicity and geographic location. Modeling suggested that the overall prevalence dropped from 2.04% (95% CI: 1.92-2.16%) to 1.77% (95% CI: 1.66-1.88%) over 4.5 years, a relative decrease of 13.2% (95% CI: 3.53-22.9%). Trend testing corroborated this decline (P < 0.01).

CONCLUSIONS

The decreasing HIV prevalence among Kinshasa antenatal care seekers is robust and encouraging. The relatively low prevalence and the weak existing healthcare system require prevention of mother-to-child transmission interventions that strengthen maternal and child healthcare service delivery. Complacency would be unwarranted: assuming a uniform national crude birth rate of 50/1000 and 1.8% antenatal HIV prevalence, approximately 7000 pregnant HIV infected women in Kinshasa, and 60,000 nationwide, are in need of care and prevention services yearly.

Measuring the HIV/AIDS epidemic: approaches and challenges

In this article, the author reviews current approaches and methods for measuring the scope of the human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) epidemic and their strengths and weaknesses. In recent years, various public health agencies have revised statistical estimates of the scope of the HIV/AIDS pandemic. The author considers the reasons underlying these revisions. New sources of data for estimating HIV prevalence have become available, such as nationally representative probability-based surveys. New technologies such as biomarkers that indicate when persons became infected are now used to determine HIV incidence rates. The author summarizes the main sources of errors and problems with these and other approaches and discusses opportunities for improving their reliability. Changing methods and data sources present new challenges, because incidence and prevalence estimates produced at different points in time are not directly comparable with each other, which complicates assessment of time trends. The methodological changes help explain the changes in global statistics. As methods and data sources continue to improve, the development of statistical tools for better assessing the extent to which changes in HIV/AIDS statistics can be attributed to changes in methodology versus real changes in the underlying epidemic is an important challenge.

Estimates of HIV incidence from household-based prevalence surveys

OBJECTIVE

To estimate HIV incidence in the general population in countries where there have been two recent household-based HIV prevalence surveys (the Dominican Republic, Mali, Niger, Tanzania, and Zambia).

METHODS

We applied a validated method to estimate HIV incidence using HIV prevalence measurement in two surveys.

RESULTS

We estimate incidence among men and women aged 15-44 years to be: 0.5/1000 person-years at risk in the Dominican Republic 2002-2007, 1.1/1000 in Mali 2001-2006, 0.6/1000 in Niger 2002-2006, 3.4/1000 in Tanzania 2004-2008, and 11.2/1000 in Zambia 2002-2007. The groups most at risk in these epidemics are typically 15-24-year-old women and 25-39-year-old men. Incidence appears to have declined in recent years in all countries, but only significantly among men in the Dominican Republic and Tanzania and women in Zambia.

CONCLUSION

Using prevalence measurements to estimate incidence reveals the current level and age distribution of new infections and the trajectory of the HIV epidemic. This information is more useful than prevalence data alone and should be used to help determine priorities for interventions.