Linkage to care, mobility and retention of HIV-positive postpartum women in antiretroviral therapy services in South Africa

Introduction: Linkage to care and mobility postpartum present challenges to long-term retention after initiating antiretroviral therapy (ART) in pregnancy, but there are few insights from sub-Saharan Africa. We aimed to describe postpartum linkage to care, mobility, retention and viral suppression after ART initiation in pregnancy. Methods: Using routine electronic data we assessed HIV-specific health contacts and clinic movements among women initiating ART in an integrated antenatal care (ANC) and ART clinic in Cape Town, South Africa. The local care model includes mandatory transfer to general ART clinics postpartum. We investigated linkage to care after leaving the integrated clinic and mobility to new clinics until 30 months on ART. We used Poisson regression to explore predictors of linkage, retention (accessing care at least once at both 12 [6 to <18] and 24 [18 to <30] months on ART), and viral suppression (HIV viral load [VL] ≤50 and ≤1000 copies/mL after 12 months on ART). Results: Among 617 women, 23% never linked to care; 71% and 65% were retained at 12 and 24 months on ART respectively, with 59% retained in care at both times. Those who linked (n = 485) accessed HIV care at 98 different clinics and 21% attended ≥2 clinics. Women >25 years, married/cohabiting or presenting early for ANC were more likely to link. Younger and unemployed women were more likely to attend ≥2 clinics (adjusted risk ratio [aRR] 1.10 95% confidence interval [CI] 1.02 to 1.18 and aRR 1.06 95% CI 0.99 to 1.12 respectively). Age >25 years (aRR 1.17 95% CI 1.02 to 1.33) and planned pregnancy (aRR 1.20 95% CI 1.09 to 1.33) were associated with being retained. Among 338 retained women with VL available, attending ≥2 clinics reduced the likelihood of viral suppression when defined as ≤50 copies/mL (aRR 0.81 95% CI 0.69 to 0.95). Distance moved was not associated with VL. Conclusions: These data show that a substantial proportion of women do not link to postpartum ART care in this setting and, among those that do, long-term retention remains a challenge. Women move to a variety of clinics and young women appear particularly vulnerable to attrition. Interventions promoting linkage and continued retention for women initiating ART during pregnancy warrant urgent consideration. This is a secondary analysis of women enrolled into the Maternal & Child Health – Antiretroviral Therapy (MCH‐ART) study, which investigated optimal ART services for pregnant and postpartum women (ClinicalTrials.gov {“type”:”clinical-trial”,”attrs”:{“text”:”NCT01933477″,”term_id”:”NCT01933477″}}NCT01933477). This study was conducted at a large primary healthcare clinic in Cape Town, South Africa in an area with high rates of unemployment and poverty 17. ANC coverage is high (approximately 95%) and the antenatal HIV seroprevalence is approximately 30% 18. The clinic serves over 4000 women annually from a wide catchment area. Women from neighbouring areas of Cape Town as well as from other provinces are known to access services here 19. ART initiation and follow‐up are provided with ANC by nurse‐midwives throughout pregnancy. During the study period, ART eligibility was based on local public‐sector guidelines (WHO stage III/IV disease or CD4 count ≤350 cells/μL until June 2013, and thereafter universal ART for pregnant women regardless of disease stage). All women initiated a fixed‐dose combination of efavirenz, emtricitabine and tenofovir, and initiation usually occurred within a week of presentation for ANC. Per local standard of care, all women were transferred out to general ART services after delivery. They were provided with up to 3 months’ supply of ART and a transfer letter to their new clinic, chosen based on preference or proximity to where a woman lived. Women were instructed to attend the new clinic before the end of her ART supply but no additional support for linkage occurs in this setting. Data for this analysis came from multiple sources. Retrospective data from available routine electronic data sources were assembled for all enrolled women through a minimum of 30 months on ART. Additional baseline data for all women, and for a subset of women additional prospectively collected data, were obtained from the parent study. The data sources are described in detail below. The parent study methods have been described previously 20. Briefly, between April 2013 and June 2014, 628 ART‐eligible pregnant women were consecutively enrolled when they presented for ANC at the integrated clinic. Study measurement visits occurred prospectively through one month postpartum in all women and through 18 months postpartum in a subset of breastfeeding women (n = 471). Mandatory transfer out of the integrated clinic occurred at six weeks postpartum for most women per local standard of care. By study design, 233 women remained in the integrated clinic for up to 12 months postpartum (median 7 months, IQR 2 to 12). Data from the parent study provided details on baseline demographics, timing of ART initiation, delivery outcomes and last visit in the integrated clinic. As part of the parent study, routine electronic health data were abstracted retrospectively through at least 30 months after ART initiation for all women (final data point December 2016). Data were abstracted from the National Health Laboratory Services (NHLS) database, which provides laboratory data for public health facilities in all provinces of South Africa. In addition, electronic data on pharmacy dispensing and clinic contacts, including facility recorded deaths were obtained from the Provincial Health Data Centre (PHDC) of the Western Cape Department of Health. These data are linked with a unique patient identifier and include all public health facilities in the Western Cape Province. Contacts at hospitals and other non‐HIV services were excluded. We brought together the above data sources to measure the following constructs. First, we defined linkage to care after leaving the integrated clinic based on evidence of at least one HIV‐specific contact (routine ART clinic visit, antiretroviral (ARV) pharmacy refill or a CD4 cell count or HIV VL laboratory test) between the last visit at the integrated clinic and 30 months after ART initiation. Second, we assessed mobility, by determining the location and counting each clinic attended after leaving the integrated clinic. This was analysed as a binary variable of one versus ≥2 different clinics. Third, we created a global measure of retention in HIV care based on evidence of at least one HIV‐specific contact at both 12 (6 to <18) and 24 (18 to <30) months after ART initiation at any clinic (including the integrated clinic for any women who linked to care prior to 30 months on ART but had not been transferred out of the integrated clinic by 12 months on ART). In sensitivity analyses we also examined evidence of HIV‐specific contact at only 24 (18 to <30) months after ART initiation and at 18 (12 to <24) months postpartum. A 12‐month window was used in all definitions as, although routine ART visits and ARV dispensing are expected more regularly, routine HIV laboratory results (our only nationally available data source) are only expected annually in this setting. Fourth, among women considered to be retained in HIV care, we investigated HIV viral suppression based on any HIV RNA VL taken nearest to 24 months on ART and at least 12 months after ART initiation. These were primarily routine care VL results from the NHLS database. However, if no routine VL was found, available VL results from the MCH‐ART study were used. VLs were found for 338 women; 61% from NHLS. VL thresholds of ≤50 and ≤1000 copies/mL were used to define suppression based on definitions of suppression and flags for treatment failure in the South African National ART guidelines 21. Analyses were conducted in STATA 14 (STATA Corporation, College Station, TX). Descriptive analysis used frequencies and proportions, means with standard deviations (SD) or medians with interquartile ranges (IQR) with chi squared tests, Fisher's exact test, t‐tests or rank sum tests as appropriate. ArcMap 10.3.1 (Esri, Inc., Redlands, CA, USA) was used to describe the spatial distribution of continued care after the integrated clinic. Multivariable Poisson regression models with robust standard errors were used to estimate the relative risk of each outcome 22. Covariates that reached p < 0.10 in bivariate analyses were included in model building using a step‐wise approach. Although the parent MCH‐ART trial intervention was not the focus of this analysis, the MCH‐ART intervention did impact retention in HIV care at 12 months postpartum in the primary trial analysis 23 and some differences were seen for the retention outcomes in this analysis (Table S6). To account for differences in subgroups of women who received continued prospective follow‐up and/or delayed transfer out of the integrated clinic as part of the MCH‐ART study, all multivariable models were adjusted for design status in the MCH‐ART study in. Results are presented as crude or adjusted relative risks (RR or aRR) with 95% confidence intervals (CI). In this exploratory secondary analysis which may not have had sufficient power to detect small associations, all associations reaching p < 0.10 were discussed. All women included in this analysis completed written informed consent that included consent to review their routine medical records. Ethical approval was obtained from both the University of Cape Town Human Research Ethics Committee and the Columbia University Medical Centre Institutional Review Board.