Page 393 - SAHCS HIVMed Journal Vol 20 No 1 2019

Page 6 of 8 Original Research

factors may not be a major determinant of the success of ART Although the level of knowledge shown by our participants
in Africa is suggested by the low rates of viraemia in is reassuring, it is of concern that some participants were not
ART-treated populations reported from Uganda of 11% at a aware about post-exposure prophylaxis and emergency
threshold of 1000 copies/mL in a general population and contraception in the event of condoms breaking, nor about
46
7% at a threshold of 400 copies/mL in pregnant patients, as how to safely negotiate condom use when planning
50
51
well as a rate of 5% in an Ethiopian general population. conception. We found our cohort to lack sufficient knowledge
Both these countries are of lower income than South Africa. about the danger of non-adherence. The knowledge shown
McMahon et al. also found viral suppression rates in low- by our participants is consistent with that shown in other
income countries to be comparable to high-income countries KwaZulu-Natal studies. 55,56 Haffejee et al., in their study of
after 12 months of non-nucleoside reverse-transcriptase women in a low-income community, found that although
inhibitors (NNRTI)-based ART regimens. Thus, although women had general knowledge about HIV they lacked
52
56
our inability to show an association between adverse socio- knowledge of specific prevention behaviour. Mindry et al.
economic circumstances and the probability of transient or found that ART clinic patients had insufficient knowledge on
persistent viraemia may have been methodological, there are safe conception methods. 55
indeed reasons to suspect that the lack of association is real.
Ninety-three per cent of pregnancies in our study whose
Adherence was the only factor we found to be associated outcomes were known resulted in live births (Table 2). We
with viraemia in our cohort, suggesting that poor adherence were unable to determine the outcomes of 18% of pregnancies
is likely to be the major factor inducing transient viraemia. in our cohort. These were patients who delivered at
Non-adherence is known to be a strong predictor of viraemia, institutions other than the study ANC or its referral hospital,
with significant odds ratios for non-adherence in four African whose infant PCR could not be identified through laboratory
studies ranging from 2.4 to 3.4. 5,41,46,51 records on a search by mother’s identifying data and who
were not contactable by telephone.
We found a significant association between initial viraemia None of the infant birth PCR results showed inutero HIV
and self-reported adherence as measured with a VAS, with a transmission (Table 2). This study did not measure viraemia
high specificity (98.5%) and low sensitivity (37.7%). at the time of delivery. Confidence intervals are wide given
Furthermore, of six participants who reported full adherence our small sample size, and thus caution is needed in
but were viraemic on initial presentation, three (50%) interpreting our finding of zero MTCT. However, we believe
resuppressed following adherence counselling (Figure 2). it probable that very few if any of our participants would
It is possible that they too were non-adherent, particularly have been viraemic at the time of labour, given that the
because self-reported measures are known to overestimate patients with transient viraemia had resuppressed, while
adherence. Of eight patients with initial viraemia who those with persistent viraemia were referred for further
53
admitted partial adherence on VAS, only three (38%) intervention, including a change in regimen. Even the
resuppressed following counselling. This suggests either that historical perinatal transmission rate of 2.3% reported by our
they were persistently non-adherent, or that some other participants for their previous pregnancies is much lower
factor, including drug resistance, was operative. These than the rate of MTCT of approximately 15% – 45% recorded
samples will be tested for resistance. An important in the pre-ART era, in keeping with much improved
20
consideration is the reported evidence suggesting that viral outcomes with modern approaches.
suppression is more tolerant of partial adherence following a
prolonged period of viral suppression. This implies that We believe our study is valuable in that it prospectively
41
rigorous attention to adherence is particularly valuable early followed a group of patients on PCART, as opposed to those
in the course of treatment. It is possible that some of the commencing ART in pregnancy. It also sought to correlate
reportedly adherent participants with a detectable VL had viraemia with possible causal factors, including adherence
failed to disclose their non-adherence, but in the absence of and socio-economic disadvantage. However, it is subject to a
resistance testing this cannot be taken further. number of limitations, particularly in that it was a cross-
sectional study and that only the viraemic group had follow-
We found that the VAS was a more reliable and practical up VL taken to determine persistence or transience of viraemia.
measure of adherence than pill counts and 4- and 30-day recall. Serial VL measurement in all patients at regular intervals until
Chaiyachati et al. undertook a study comparing the predictive delivery would be necessary to understand the natural history
accuracy of five measures of non-adherence in predicting of viraemia in pregnancy fully. The factors driving loss of viral
treatment failure in a community in rural KwaZulu-Natal. suppression were not rigorously identified. Infant PCR results
54
They found the VAS to be highly specific (98% – 100%, were incomplete, a consequence of a fragmented healthcare
dependent on the threshold used), although with low system with poor continuity of care.
sensitivity (0% – 7%) in predicting adherence. Indeed, they
found that no method had both acceptable sensitivity and With an increasing number of perinatally infected girls born
acceptable specificity, and the accuracy of the VAS was not in the era before comprehensive ART coverage now reaching
inferior to any other method in terms of either sensitivity or childbearing age, and increasing national ART coverage
specificity. expected with the Universal Test and Treat initiative, the
57
http://www.sajhivmed.org.za 386 Open Access