Progress towards the UNAIDS 90–90-90 goals by age and gender in a rural area of KwaZulu-Natal, South Africa: a household-based community cross-sectional survey

We conducted a cross-sectional household-based community survey between July and October 2013.

A two-stage stratified cluster probability sampling strategy was used for the selection of households according to the 2011 Census [9]. In total, 125 clusters of 25 households each were selected from 14 administrative units called Wards. Google Earth maps from 2011 with exhaustive identification of the households were used to sample the households to be visited by choosing randomly the first household and then sequentially the closest to the first/previous one. Field staff used Global Positioning System (GPS) receivers to find the geographic coordinates of each household.

People aged 15–59 years old living in Mbongolwane and Eshowe Health Service Areas were eligible for enrolment in the study. Those who signed a written informed consent were included.

Mbongolwane, a rural area, and Eshowe, the main town of the municipality, account for approximately 120,000 inhabitants [9]. Decentralization of ART care to the peripheral level was implemented gradually in this area. In 2011, the KZN province embraced the notion of nurse-initiated and managed ART (NIM-ART). MSF support to the KZN Department of Health (DOH) included prevention activities such as condom distribution, voluntary medical male circumcision, community mobilisation, large-scale community-based HIV counselling and testing, implementation of point of care CD4 testing, linkage to care, and training and mentoring of health staff in facilities in support of NIM-ART. In 2013, two district hospitals and their linked 10 primary healthcare facilities were ART-initiating centres. The survey was conducted 8 years after the initiation of the HIV program in the area. At the time of the survey the CD4 threshold for ART initiation was 350 cells/μL.

Prior to starting the survey, we conducted community information and mobilization activities through several channels: information on radio spots, meetings with community leaders and health facilities workers, information in schools, leaflets and posters. In order to reach a maximum of eligible individuals in their houses the survey teams visited the houses from Tuesday to Sunday. Time slots from early morning to late evening were covered in different days of the week in order to maximize the possibilities of finding the eligible participants at home. Due to the importance that blood has in the Zulu culture, the survey teams made a particular effort in explaining the purpose of collecting and storing blood and the use of it. The survey teams used face-to-face interviewer-administered questionnaires to collect information at the participant’s home on socio-demographics and history of HIV testing and care (see Additional files 1, 2 and 3). Questionnaires were developed for the Demographic and Health Surveys [10] and adapted for the study. Certified lay counsellors performed rapid HIV testing on site and provided pre and post-test counselling to the participants willing to test at home. Counsellors used Determine Rapid HIV-1/2 Antibody test kit for screening, and if positive, Unigold Rapid HIV test kit for confirmation according to the South African National guidelines for HIV Counselling and Testing. The tests were standardised and validated for this use. In addition, HIV-positivity was confirmed by ELISA at the laboratory. Survey nurses collected venous blood specimens from HIV-positive participants for antiretroviral (ARV) drug presence test, CD4 count and viral load. Venous blood samples were transported every evening to Global Clinical and Viral Laboratory in Durban. CD4 count was performed using a FACSCalibur™ device from Becton, Dickinson and Company (BD) according to standard manufacturer’s instructions on samples reported as HIV positive. Two dry blood spots (DBS) samples were prepared using the venous blood samples from each participant and transported in batches to the Department of Pharmacology laboratory at Groote Schuur Hospital, University of Cape Town, for ARV drug levels. Qualitative testing for ARV drug levels was performed for the presence of nevirapine, efavirenz and lopinavir which covered all ARV regimens in use in the public sector in the area. A liquid chromatography tandem mass spectrometry assay with a limit of quantification of 0.04 μg/mL was used for all drugs. The assay was developed and validated at the Division of Clinical Pharmacology, University of Cape Town. Viral load was performed for participants on ART for more than 6 months (determined by questionnaire) at Global Clinical and Viral Laboratory in Durban using a NucliSens EasyQ HIV-1v2.0 assay from Biomerieux according to manufacturer’s instructions. The test could quantify HIV-1 RNA over the range of 20 copies to 20 million copies for 0.5 mL sample.

We calculated progression towards the 90–90-90 UNAIDS targets by estimating three proportions: HIV positive individuals who knew their status (first 90), those diagnosed who were on ART (second 90), and those on ART who were virally suppressed (third 90). Viral suppression was defined as having less than 1000 copies/mL. In addition, we calculated five steps of the HIV cascade of care using the total number of HIV positive individuals as a common denominator. ‘Diagnosed’ were the individuals who knew their HIV positive status prior to the survey; ‘Linked to care’ were those who declared having sought care for their HIV infection; ‘In care’ were those who were still receiving HIV care at the time of the survey; ‘On ART’ were those who had ARV detected in blood; ‘Virally suppressed’ were those with viral load less below 1000 copies/mL. All statistical analyses were adjusted for clustering at the level of Ward and household. Descriptive analyses are presented here with 95% confidence intervals (CI). Categorical variables were compared using proportional test. Analyses were primarily performed using Stata 13 (™StataCorp, College Station, Texas, USA).

The protocol was approved by the University of Cape Town Human Research Ethics Committee (HREC), the Health Research Committee of the Health Research and Knowledge Management Unit of KZN Department of Health, and the Comité de Protection de Personnes de Paris in France. All participants provided written informed consent. Participants under 18 years provided assent and their parents, guardians or caregivers provided written informed consent for them.

We visited 2377 households and we included 5649 (84.5%) participants among 6688 eligible individuals. Inclusion rate was: 3518/4008 (87.8%) among women and 2131/2680 (79.5%) among men. The remaining individuals were not included due to: refusal (8.7%), not being at home (4.9%), being incapacitated (1.0%) and other reasons (0.9%). The median age of the participants was 26 years (IQR: 19–40), 62.3% were women, 83.4% lived in rural areas, 78.8% were not living with a partner, 49.7% had completed at least secondary school, 36.3% declared no occupation and 16.6% had moved their residence in the 10 years prior to the survey or were visitors (Table 1). Thirty-two per cent of the men were under 19 years, compared to 22% of the women, possibly reflecting out-migration of adult men to seek work.

In total, 2548 (72.4%) women had ever given birth. The median number of children per women was 2 (IQR: 1–4). At the time of the survey, 134 (3.8%) of the women were pregnant and 308 (8.8%) were breastfeeding. Of the 1259 women who had delivered in the 5 years prior to the survey (2008–2013), 1214 (96.4%) had had at least one medical antenatal care (ANC) consultation, and 920 (73.1%) had had 3 or more ANC consultations. The median number of ANC consultations was 6 (IQR: 5–7). The median month of pregnancy at the first ANC consultation was 4 months (IQR: 3–5). Out of the 799 women who had delivered in the 2 years prior to the survey, 745 (93.2%) had had an HIV test as part of their ANC.

In total, 1423 participants were HIV positive. The overall prevalence was 25.2% (95% CI: 23.6–26.9). Prevalence in women was higher than in men: 30.9% (95% CI: 29.0–32.9) vs 15.9% (95% CI: 14.0–18.0). Peak prevalence was 56.5% (95% CI: 50.9–62.0) in women at age 30–34 years and 45.5% (95% CI 37.3–54.0) in men at age 35–39 years (Fig. 1). Prevalence for age 15–29 years crudely averaged 22.3% (95% CI: 20.5–24.3) in women and 6.2% (95% CI: 5.1–7.6) in men, increasing dramatically from age 15 to 29 (3.9% to 55.0% in women and 1.5% to 26.7% in men). HIV positive mothers who had delivered in the 5 years prior to the study had a higher proportion of children who had died than HIV negative mothers: 5.3% vs 2.5% (p = 0.010). Of the 1400 HIV-positive participants with a CD4 count, 130 (9.3%) had a CD4 count below 200 cells/μL, 255 (18.2%) between 200 and 349 cells/μL, 363 (25.9%) between 350 and 499 cells/μL and 652 (46.6%) over or equal to 500 cells/μL. Median CD4 count was 483 cells/μL (IQR: 332–665). Among the 655 individuals not on ART, 78 (11.9%) had a CD4 count below 200 cells/μL and 138 (21.1%) between 200 and 349 cells/μL. Among the 741 individuals on ART, 52 (7.0%) had a CD4 count below 200 cells/μL and 115 (15.5%) between 200 and 349 cells/μL. Of the participants with viral load below 1000 copies/mL, 5.7% (95% CI: 4.2–7.5) had a CD4 below 200 cells/μL and 19.7% (95% CI: 17.2–22.6) a CD4 below 350 cells/μL (Table 2).

UNAIDS targets were partially achieved with 76.4% (95% CI: 74.1–78.6) of all HIV-positive who knew their status, 69.9% (95% CI: 67.0–72.7) of them being on ART, and 93.1% (95% CI: 91.0–94.8) of the people treated being virally suppressed (Fig. 2). Progress towards the first target differed by sex and age. HIV diagnosis was: 79.0% (95% CI: 76.4–81.4) in women versus 68.3% (95% CI: 63.1–73.1) in men (p < 0.001); and 83.3% (95% CI: 80.6–85.7) in individuals aged 30–59 years versus 64.0% (95% CI: 59.9–67.9) in those aged 15–29 years (p < 0.001). Progress towards the second and third targets differed by age but not by sex. ART among individuals diagnosed was: 70.5% (95% CI: 67.7–73.2) in women versus 67.9% (95% CI: 60.8–74.2) in men (p = 0.443); and 75.1% (95% CI: 71.5–78.4) in individuals aged 30–59 years versus 57.6% (95% CI: 51.8–63.2) in those aged 15–29 years (p < 0.001). Viral suppression in individuals on ART was: 93.4% (95% CI: 91.1–95.1) in women versus 92.1% (95% CI: 86.0–95.7) in men (p = 0.584); and 94.5% (95% CI: 92.3–96.0) in individuals aged 30–59 years versus 89.0% (95% CI: 82.7–93.1) in those aged 15–29 years (p = 0.011).

We also looked at other components of the cascade of care and the proportions of ART and viral suppression among all HIV-positive individuals. Of the 1384 HIV-positive participants with complete information on HIV diagnosis, ARV presence in blood and viral load, 71.0% (95% CI: 68.6–73.4) were linked to care, 62.7% (95% CI: 59.8–65.6) were in care, 53.5% (95% CI: 50.6–56.3) were on ART, 57.4% (95% CI: 54.6–60.1) were virally suppressed. The largest gaps in the cascade of care occurred in men and people aged 15–29 years (Table 3). Regarding viral suppression, 60.3% (95% CI: 57.4–63.1) of all HIV-positive women were virally suppressed compared to 47.9% (95% CI: 41.7–54.1) of the HIV-positive men and 66.2% (95% CI: 63.0–69.2) of the HIV-positive individuals aged 30–59 years were virally suppressed compared to 41.3% (95% CI: 36.9–45.9) of those aged 15–29 years. Of the 590 HIV-positive participants virally unsuppressed, 279 (47.3%) were undiagnosed and 260 (44.1%) were diagnosed but not on ART. A breakdown by age and sex is given in Fig. 3.