Page 193 - HIVMED_v21_i1.indb

Page 6 of 7 Original Research

are able to interpret results as accurately as trained HCWs. In clinic, with a HCW present, may also present bias, as the app
low bandwidth environments, the requirement to upload is intended to be used independently of a clinic setting.
images may also incur additional data charges and may not Another limitation of the pilot process was that the HCWs
be cost effective. did not record the participants’ interpretation on paper and,
thus, results discordance could not be verified, as was done
With the number of countries adopting HIVST policies being for the other variables.
on the rise, the M&E of these programmes poses a unique set
12
of challenges and measurement of uptake and effectiveness Although recent studies have introduced validated data
45
TM
becomes difficult. The Aspect HIVST app facilitated the collection tools for mHealth usability, at the time of this
capture of HIVST data directly to an operational dashboard, study, there were also no validated data collection tools to
TM
namely Aspect . This dashboard was developed by measure the acceptability and feasibility of mHealth apps for
SystemOne and is currently being used to report tuberculosis HIVST, hence the study-specific questions may not be used to
and HIV viral load results from over 3000 diagnostic reproduce these results in similar settings. Similarly, the use
instruments across 43 countries. For this study, the of only one HIVST kit and its accompanying IFU means that
29
dashboard displayed very basic summary HIV statistics, a these results cannot be generalised across all HIVSTs,
list of individual test results and also supported the especially since many of the errors were related to the
downloading of automated reports. This could allow a interpretation of the IFU.
programme manager to remotely monitor indicators such as
uptake, demographics of the testing population, HIV Conclusions
positivity rates, invalid rates and improve reporting against
key performance indicators. The functionality of the With millions of HIVST kits distributed worldwide without
dashboard also allows for the pushing of automated SMS adequate tracking, the need for M&E of these kits is ever
notifications directly to the tester based on their HIV result, increasing. On an individual level, this may lead to better
which could be used to promote confirmatory testing and linkage to care and follow-up with patients and, on a national
help link them to care. This is especially important for level, tracking can identify areas of need to optimise kit
43
HIVST, as one of the problems with home testing is that distribution, marketing and supplementary information.
people receiving a positive diagnosis are suddenly faced with Despite some challenges with IFU interpretation and data
a serious diagnosis and no immediate access to information, capture via the app, this pilot study has shown that the Aspect
TM
11
counselling or treatment resources. The feasibility of these HIVST app is an acceptable way to upload mobile HIVST
dashboard features should be considered for future research. results and demographic information to a central database.
Data concerns are also an important issue in South Africa, Acknowledgements
with previous mHealth studies highlighting data costs and
phone memory as a barrier to entry. 26,44 Future app The investigators would like to thank the study
development should focus on keeping storage requirements implementation team and the study participants.
minimal to ensure that the app is available for as many
individuals as possible. Furthermore, the necessity to upload Competing interests
images may also incur additional data charges and may not N.G. works for SystemOne and was involved in the design of
be affordable for all users.
the Aspect HIVST app.
Limitations Authors’ contributions

The study had several limitations. Convenience sampling N.G., N.R., M.P. and M.M. designed the study. N.G., N.R. and
from one sub-district from inner-city Johannesburg was used M.P. collected data, N.G., N.R., M.P. and A.E.F. were involved
to recruit participants limiting the generalisability of the in the data cleaning and analysis; A.E.F., S.T.L.-E. and N.G.
findings, and the compensation of participants may have wrote the initial draft of the manuscript. All authors critically
accounted for the very high participation rate. Furthermore, reviewed and approved the final draft.
the majority of participants were under 35 years old, which
may have made it easier for them to navigate a mobile app as
they may be more tech-savvy than older age groups. The Funding information
Aspect HIVST app was only available in English. It was Funding to carry out the research project has been obtained
TM
also only tested on a Samsung phone, and it may not reflect through a Gates Grand Challenges Award to SystemOne
the usability of the app on other phones owned by the general (grant number: OPP1182240) and Grand to Wits RHI received
population, especially across different operating systems and from the Bill and Melinda Gates Foundation (grant number
memory capacities. The discordance between HCW-recorded OPP1132929).
and app-captured demographics may reflect an interviewer
bias, whilst the process of testing in front of a HCW may have Data availability statement
increased the number of forced errors due to the pressures of
being observed. Performing the HIVST with the app in a Data are available upon reasonable request.

http://www.sajhivmed.org.za 185 Open Access