Factors associated with non-attendance at scheduled infant follow-up visits in an observational cohort of HIV-exposed infants in South Africa, 2012–2014

Optimal management of HIV in infants and children has been a challenge in low-middle income countries, resulting in missed opportunities for initiating timely treatment and averting adverse outcomes [1]. The reasons for sub-optimal care, despite well-structured guidelines for monitoring and follow-up, are complex and intertwined between the user and provider-related factors. Despite having made good progress in adult HIV treatment in South Africa, the country is still faced with a high loss to follow-up of HIV-exposed infants mainly during the high-risk breastfeeding period, with concomitant low coverage of infant HIV treatment [2, 3]. Data from a 2010 national survey in South Africa illustrates that 38% of early infant HIV testing opportunities are missed as a result of poor documentation of infant HIV exposure status or no maternal request for a test [4]. Other factors that contribute to unsatisfactory HIV care in children include poor parental uptake of HIV services, fear of stigmatization, limited knowledge about the modes of vertical HIV transmission (MTCT), young maternal age and low socioeconomic status [4‐6].

Notwithstanding improved prevention of MTCT (PMTCT) policies from 2002 until 2015, the infant rapid HIV test at 18 months continues to mark the end point of the PMTCT programme, serving as a measure of PMTCT effectiveness [7]. Additionally, since 2005 the Integrated Management of Childhood Illness (IMCI) strategy, recommends that all infants, regardless of HIV exposure status, receive monthly routine growth monitoring and care (well-baby visits) during the first year of life, and every 3 months thereafter. There is little data on factors associated with non-attendance at scheduled child health care visits in the context of HIV.

We aimed to measure non-attendance at intermittently scheduled study follow-up visits, and at the final 18-month exit visit, amongst HIV-exposed uninfected (HEU) infants enrolled in a nationally-representative survey to measure postnatal PMTCT effectiveness. We also sought to identify factors associated with non-attendance at these visits to shed light on possible interventions to improve postnatal follow-up.

Using data from a nationally representative observational cohort study which aimed to synchronise study follow-up visits for HEU infants with their recommended or planned routine postnatal follow-up visits where possible, we demonstrate that, ~ 59% of HEU infants did not attend one or more scheduled visits between 3 and 18 months postpartum. The lowest MV occurred at 3 and 18 months where just under a third of participants did not attend. Uptake of the 3 months visit could have been influenced by either the 14-weeks immunization uptake and/or that it was the first follow-up after enrolment [13]. Better uptake of the 18 months visit was probably linked to the fact that this is a major PMTCT end point visit and an immunization time point. Furthermore, study inconvenience allowances increased from USD$2.50 to USD$6.00 to cover participant travel and time costs, at eight calendar months before the study closure. The change in allowance amount could have motivated for better attendance thereafter at which most upcoming visits were for the 15 and 18-month points. It is also likely that maternal uptake of infant postnatal care is better when postnatal visits are fixed to specific activities such as vaccination rather than being labelled as a ‘general baby check-up’. It is also important to note that the rate of non-attendance could be lower under routine settings because participants were given inconvenience allowances during these study visits, which could have motivated some to return for follow-up.

The low uptake of study visits after 3 months is consistent with what has been observed in routine settings of other African countries. A study by Godfrey et al. (2015) for example, found extremely low retention rates for both mothers and infants between 4 and 12 months postpartum in Kenya, Malawi and Swaziland [14]. Sub-optimal retention in care has also been a challenge across different PMTCT regimen eras indicating that improving the regimen alone is not a solution but new and innovative strategies to promote better uptake of HIV care are also needed [15‐18].

Not receiving maternal ART by 6 weeks postpartum was significantly associated with non-attendance regardless of how it was defined. Most importantly it was associated with missing the 18-month visit, which is a critical PMTCT visit as, in most countries, the 18-month HIV test is used to measure the final outcome of PMTCT interventions. Data collection for this study began when Option A regimen was in use during pregnancy and ended when the local policy had moved to Option B. For both Option A and Option B; ART was initiated in pregnant women with a CD4 count of 350 and less. The results seen here could reflect the differences in women’s attitudes then, between women whose lives were deemed to be at risk and hence on ART and the rest who otherwise probably did not find it necessary to attend as many visits. Unknown infant nevirapine-intake status at 6 weeks was associated with poor uptake of postnatal visits. This could reflect general neglect of child health issues and requires continued education and encouragement of caregivers during routine child health care visits. When looking at the breakdown of proportions of MV per time-point (see Additional file 1), the group with unknown infant nevirapine-intake had much higher MV until the 9 months time-point, and this difference was significant. From 12 months onwards, there were no longer significant differences in MV proportions between those who were on infant prophylaxis at 6 weeks, those who were not on infant prophylaxis at 6 weeks, and those who did not know the prophylaxis intake history. This finding, could mirror a change from the use of Option A regimen (cessation of infant prophylaxis at 6 weeks post cessation of breastfeeding) to Option B (infant prophylaxis for 6 weeks postpartum). That is, while Option A was active, infant prophylaxis continued during breastfeeding and therefore promoted infant visits to healthcare, to collect infant nevirapine, until breastfeeding cessation. Yet after Option B was implemented, infants were not required to be on nevirapine prophylaxis beyond 6 weeks postpartum, and thus not required to attend health facilities for HIV-care and prophylaxis. The influence of nevirapine intake on postnatal visits was therefore lost with Option B regimen. The significant results from the regression analyses therefore show that the strength of association between infant nevirapine prophylaxis intake and MV rates during the era of Option A was very strong.

Mothers older than 24 years, compared to adolescent and younger mothers were significantly associated with fewer MV in both definitions of MV. This finding is not surprising, given that provision of appropriate care for adolescents and young mothers have been identified as a critical gap in many countries [19‐23]. Some earlier studies have consistently shown better awareness and uptake of HIV care among older women across different countries [23‐25], and thus appropriate interventions for the younger age group are needed urgently. Interventions focussing on creating friendly attitudes and safe environments for adolescent-mothers in public healthcare facilities, aiming at making them feel as comfortable as older women in healthcare facilities, are needed urgently.

It was interesting to find that participants who took less time traveling to healthcare facilities, within 30 min, by walking or using public transport, had better outcomes compared to those who owned cars. The latter were not different from public transport users or walkers who took longer than 30 min to reach the facilities. Although it is not clear why those who owned cars performed poorer than the second group, the findings show that shorter time spent travelling to healthcare facilities by walking or public transport is an important factor for achieving good retention in care among mothers and their young children. This finding of better uptake among those with quicker access to facilities is not unique to South Africa. Accessibility to health facilities, in particular, availability of transport, transport costs and longer distances is a known obstacle to uptake and retention in care in other African settings [26, 27]. Investing in sustainable mobile clinics for the remote settings could facilitate uptake of health care among communities living far from healthcare facilities.

Significant differences between provinces were also observed for both types of MV, raising the need to investigate province-specific barriers.

We did not find any association between missing postnatal study visits and knowledge of MTCT modes or discrimination and disclosure of maternal HIV status in contrast to what has been reported by other earlier studies [28]. The findings related to disclosure and discrimination, in particular, are probably because disclosure was largely towards trusted close family or friends only, thus did not affect uptake of health-related care in any way. Also, a majority of the participants (three-quarters of the sample) had disclosed their HIV status and only less than 10% of participants reported facing discrimination indicating general improvements in societal attitudes towards HIV and the stigmas which were associated with it in the past. There was also no clear association of MV rate with the type of caregiver at baseline or socio-economic status. Lack of association with the type of caregiver could simply be because 97% of the participants were infant mothers.

We measured non-attendance at scheduled study visits, with a particular focus on the 18-month time point, and used these study visits as a proxy for attendance at routine postnatal visits. However, we do not have data to confirm that scheduled study visits coincided with routine postnatal visits. Given that we offered infant HIV testing at each time point, and thus earlier referral into care, we believe that this limitation was diluted by the offer of earlier infant HIV diagnoses, as each visit could be a proxy for attendance or uptake of earlier infant HIV testing, a critical component of preventative child health programmes in high HIV prevalence settings. The lack of routine tracking systems did not allow us to establish whether a child who missed their study visit attended a different health facility for their routine child health visit. Our findings raise the importance of patient tracking systems, across rural and urban settings to ensure that no mother-baby pair is lost from care.

The PMTCT regimen changed from Option A to Option B guidelines mid-way of the cohort study, in April 2013, which was the last month of baseline (6-weeks post-partum) enrolment of participants. Although we investigate association of MV with only baseline (including antenatal experiences) characteristics, which were all under the same regimen (Option A), follow-up to study end-points over-lapped with Option B guidelines. The data collection process did not include individual-level information on the specific time that they were moved from Option A to Option B. For this reason, it was not possible to accurately disentangle, nor adjust in the regression models, the effects of treatment policy change on MV, if any.

Recent South African studies, following the introduction (in 2015) of birth PCR HIV testing in infants, have demonstrated poorer infant follow-up postnatally among infants tested at birth [29, 30]. Therefore the retention seen in this study may be an over-estimate of true current retention.

In this nationally-representative sample, we studied non-attendance at scheduled study visits where we offered additional HIV testing (at 3, 6, 9, 12 and 15 months postpartum) and routine HIV testing at the final PMTCT endpoint of 18-months. Despite us offering these services, which facilitate earlier access to care, we demonstrate that > 50% HEU did not attend one or more scheduled study visits. Poor accessibility of clinics, young age, poor maternal uptake of ART at 6 weeks postpartum and child healthcare unawareness were risk factors for non-attendance at the scheduled study visits during the era of PMTCT Option A and B PMTCT regimens. On-going exploration is needed to understand and remove the barriers for postnatal care uptake in the current era of lifelong ART. Strengthening community-based health care linkages with the facilities could be considered to facilitate tracing and ease of access if similar challenges persist.