Page 61 - SAHCS HIVMed Journal Vol 20 No 1 2019

Page 6 of 10 Review Article

could include in our meta-analysis and therefore reduced the documented in current data systems. 12,92,93,94 Already a three-
likelihood we could find significant statistical differences in tiered monitoring system exists in the Western Cape Province
LTFU. For instance, we did not find that variation in LTFU that aggregates paper and electronic systems into a single
95
definition impacted overall LTFU estimates at 1 year in our database for reporting purposes ; the ideal or goal is to scale
meta-analysis, and this is likely because of several reasons. this up to a national level and transition to an electronic
Firstly, the small sample size of the analysis; once estimates medical record as resources allow. We hope that our data may
were matched for definition and overlapping cohorts were be useful to South African ART programmes in advancing
removed, the sample size was relatively small. Similarly, these broader goals of improving ART retention for patients
larger estimates of LTFU are notable in smaller cohorts across South Africa.
likely because of outlier effects. Secondly, there was a lack of
standardisation of estimation methodologies for LTFU and Acknowledgements
mortality including length of follow-up time. Thirdly,
inclusion of paediatric cohorts likely also played a role in the S.K. would like to thank the staff at the University of Cape
observed variation. For instance, paediatric patients may be Town Health Sciences Library for assistance with literature
more likely to be retained in care given that they have searches and the staff at the University of Cape Town
caregivers. Additionally, pregnant patients may be more Clinical Research Centre for assistance with organisation of
likely to be lost to follow-up following childbirth, which has data extraction.
been demonstrated in several studies. This may be for a
variety of reasons, including lifestyle changes postpartum as Competing interests
well as changing motivations after preventing HIV
transmission to their infants. 89,90,91 Indeed, the differences in The authors declare that they have no financial or personal
aggregate LTFU estimates at 1 year were different between relationships that may have inappropriately influenced them
adult versus paediatric versus pregnancy cohorts and largely in writing this article.
follow this trend: pregnancy cohorts had higher LTFU
(33.0%), and paediatric lower LTFU (7.5%) than adult cohorts Authors’ contributions
(11.6%). A final limitation was that six randomised controlled G.M., A.B. and S.K. conceived and designed the project. G.M.
studies were included, of which some of the interventions and A.B. provided the overall conceptual and design
were designed to impact adherence and LTFU, which guidance. S.K. was the lead author and conducted literature
therefore could have biased the meta-analysis estimates. searches and led data extraction efforts with the assistance of
K.S.N. who aided in data extraction. N.F. provided conceptual
We likely underestimated and/or misrepresented true input, conducted the meta-analysis and created Figures 2
estimates of LTFU at 5 years in our meta-analysis because and 3. S.K wrote the manuscript with assistance from G.M.
of including only two non-representative cohorts in our All authors reviewed the manuscript, provided edits and
estimate after standardisation. However, both estimated agreed with its final form.
LTFU at 5 years to be > 1 in five patients. Fatti et al. defined
22
LTFU as 187 days without a clinic visit and did not include
81
those who had left care and returned later. Grimsrud et al. Funding
similarly defined LTFU as 6 months without a clinic visit S.K. was supported by the National Institutes of Health
and also did not include patients who had left care and Office of the Director; Fogarty International Center; Office of
returned later. Despite being high crude rates of LTFU, these AIDS Research; National Cancer Center; National Heart,
are lower than estimated by large systematic reviews as Blood, and Lung Institute; and the National Institutes of
described above. Health (NIH) Office of Research for Women’s Health through
the Fogarty Global Health Fellows Program Consortium
In conclusion, going forward in South Africa, our data suggest comprising the University of North Carolina, John Hopkins,
that it would be helpful for policy-makers to recommend and Morehouse and Tulane (R25TW009340). A.B. was supported
programme managers to put into practice a system in which by the NIH (U01AI069924), Médecins Sans Frontières and
the definition of LTFU or ‘default’ from care is standardised National Research Foundation (NRF) incentive funding.
across South African ART programmes. Such standardisation G.M. was supported by the Wellcome Trust (098316), the
would not only aid in comparing outcomes across clinics and South African Research Chairs Initiative of the Department
across the country, especially at defined timeframes, but also of Science and Technology, and NRF of South Africa (Grant
in planning broadly applicable interventions for patient No 64787), NRF incentive funding (UID: 85858) and the
retention. Ideally, data from clinics could be monitored in real South African Medical Research Council through its TB and
time using a standardised definition, with an actionable HIV Collaborating Centres Programme with funds received
reporting system in place to identify patients who require re- from the National Department of Health (RFA# SAMRC-
engagement, or clinics that need interventions to improve RFA-CC: TB/HIV/AIDS-01-2014). The funders had no role
patient retention. Additionally, tracing patients after they are in the study design, data collection, data analysis, data
LTFU may improve outcomes and lower LTFU rates, as many interpretation or writing of this report. The opinions, findings
ART patients are mobile and receiving care at more than one and conclusions expressed in this manuscript reflect those of
clinic, and/or transfers to other clinics may not be sufficiently the authors alone.

http://www.sajhivmed.org.za 54 Open Access