Page 390 - SAHCS HIVMed Journal Vol 20 No 1 2019

Page 3 of 8 Original Research

were then counselled on adherence and instructed to bring Ethical consideration
their pills at the next visit for a pill count.
Permission to conduct the study was obtained from the
Patients returned one week after the initial interview. A pill University of KwaZulu-Natal Biomedical Research Ethics
count was performed and they were informed of their initial Committee (BE286/16).
VL results. Participants with any detectable VL received
additional adherence counselling. Results

Three hundred and eighty-six pregnant women presented
All participants were requested to return for a final visit after for booking at the study site within the study period. One
four weeks to repeat the pill count and perform a follow-up hundred and seventy three of these patients (44.8%) were
VAS. A second VL sample was drawn on those who had a HIV-positive and 94 (54.3%; 95% confidence interval [CI]
baseline viraemia above 50 copies/mL. Persistently viraemic 47% – 62%) were on PCART. Eighty-two participants (86% of
participants with a VL exceeding 1000 copies/mL were those who potentially met inclusion criteria), selected on the
switched to second-line ART according to standard treatment basis of attendance on specific study days, were invited to
protocols. For the purposes of this study, viral suppression was participate and all enrolled. The mean age of the participants
defined as a VL less than 50 copies/mL, transient viraemia was was 30.4 years (SD 6.1, range 18–43 years). The median
defined as a VL at any level above 50 copies/mL on the initial gestational age at booking was 17 weeks (IQR 12–23, range
test but where viral suppression was achieved on the repeat 4–30). Eleven were first pregnancies (13.4%), and 72 were
test, and persistent viraemia as a VL above 50 copies/mL, subsequent pregnancies (86.6%). Fifty-nine pregnancies
present on both the initial and the repeat test. (72%) were unplanned. The median duration of ART was
46 months (IQR 24–72, range 9–216). Seventy-two participants
Blood was drawn into EDTA-containing tubes and (87.8%) had disclosed their HIV status to their partner. Sixty-
transported to the National Health Laboratory Service two participants (75.6%) were aware of their partner’s status.
(NHLS), where HIV VL testing was conducted by the Of these, 42 partners were seropositive (67.7%) and 29 of the
Cobas® 8800/6800 HIV-1 test methodology. In the case of seropositive partners (69.0%) were themselves on ART.
neonates, whole blood was collected at birth onto filter
cards and then air dried as dried blood spots. Samples were VL at enrolment and at the 4-week reassessment is shown in
stored on-site in a refrigerator and transported to the Table 1. Of 15 patients initially viraemic, seven showed
regional reference laboratory within the time frame evidence of viral suppression at the follow-up visit (Figure 1).
recommend by the NHLS. HIV polymerase chain reaction Information on the outcome of pregnancy was available in 65
(PCR) testing was performed using the Roche COBAS® patients (79.3%) and is summarised in Table 2. Our
TaqMan® HIV-1 Qualitative Test Version 2 (Roche Molecular participants reported three perinatal transmissions out of 141
Systems, Inc., Branchburg, NJ, USA).
live births in previous pregnancies, indicating a historical
perinatal transmission rate of 2.1%.
Pregnancy outcomes were determined by accessing the
clinic records to review the postnatal consultation. Where Of the 82 participants, 80 (97.6%) provided information on
necessary, patient records were also accessed from the adherence using the VAS self-reported adherence test. The
regional referral hospital, to which patients are referred for VAS score correlated significantly with the likelihood of
antenatal problems such as miscarriage and complicated viraemia. The median scores were 92.5 (IQR 30–100) and 100
pregnancies and for management of difficult deliveries. The (IQR 90–100) for the initial viraemia and suppressed group,
results of infant HIV PCR tests at birth were obtained from respectively ( p = 0.02) (Table 3). The VAS showed high
both sources as well as from a search of the computerised specificity for initial viraemia on receiver operating
records of the NHLS, which is responsible for the testing of characteristics (ROC) curve analysis (sensitivity 37.7% and
all public sector patients. specificity 98.5% at a cut-off value of 50, area under curve
[AUC] 0.67). Four of the eight patients who demonstrated
Data analysis
TABLE 1: Viral load at baseline and at 4-week follow-up following intervention.
Data were captured on a data sheet and transferred to a Viral load CI n %
Microsoft Access database. Statistical analysis was performed At initial visit (N = 82)
using SPSS version 17.9.7 and MedCalc Statistical Software Suppressed (< 50 copies/mL) CI 71.3% – 89.1% 67 81.7
version 17.9.7. Missing data and outliers were verified for Viraemic (> 50 copies/mL) CI 10.9% – 28.7% 15 18.3
correct entry by checking the original source documents. < 400 copies/mL - 6 7.3
• 400 copies/mL - 9 11.0
Categorical variables were compared with chi-square and At 4-week follow-up (N = 22)
Fisher’s exact tests as appropriate. Ordinal data were Suppressed (< 50 copies/mL) 8.5% of total cohort 7 46.7
(Transient viraemia)
described by mean and standard deviation (s.d.) if normally Viraemic (Persistent viraemia) 9.8% of total cohort 8 53.3
distributed and by median, and by interquartile range (IQR) 50–400 copies/mL 6.1% of total cohort 5 33.3
and range if not. Non-parametric data were compared with • 400 copies/mL 3.7% of total cohort 3 20.0
the Mann-Whitney U test. CI, 95% confidence interval (Wilson’s method corrected for continuity).

http://www.sajhivmed.org.za 383 Open Access