Assessment of a Viral Load Result-driven Automated Differentiated Service Delivery Model for Participants Taking Antiretroviral Therapy in Lesotho

To sustainably provide good quality care to increasing numbers of people living with HIV (PLHIV) receiving antiretroviral therapy (ART), care delivery has to shift from a "one-size-fits-all" approach to differentiated care models. Such models should reallocate resources from patients who are doing well to patient groups who may need more attention, such as those with treatment failure or medical and psycho-social problems. Ideally, such a reallocation allows health systems and patients to save resources while improving quality of care.

One proposed approach to differentiate care and intensity of monitoring is viral load-driven differentiated service delivery. Reducing the intensity of monitoring in patients with suppressed viral load (VL) and no other clinical problems would substantially reduce the workload at health care facilities and save time and transport cost for patients, thus potentially improve long-term engagement in care. Time and resources saved in patients with suppressed VL and no other clinical problems would allow focusing on those participants with elevated viral load and/or other clinical problems (like tuberculosis, which is the most common cause of mortality among PLHIV in sub-Saharan Africa). This may potentially improve PLHIVs' clinical outcome through intensified adherence support, clinical follow-up and timely switches to second-line ART. In many settings in sub-Saharan Africa, however, the potential of VL monitoring to differentiate care is not exploited and thus constitutes a missed opportunity. In Lesotho it was shown that the majority of unsuppressed VLs are not acted upon in a timely manner, be it due to providers and patients not being aware of the results or health care providers not being proficient in the management of treatment failure.

The concept of the proposed automated differentiated service delivery model (aDSDM) is to use VL results, other clinical characteristics (TB screening results and CD4 cell counts) and participants' preference to automatically triage participants into groups requiring different levels of attention and care. Innovatively, triaging of participants will be done automatically capitalising on an existing VL database platform. The implemented aDSDM will differentiate care according to three elements:

• clinical characteristics (with focus on VL measurement)
• sub-population (women, men)
• participants' and health care providers' preferences

To ensure effective flow of information, VL results and other relevant information is sent directly to participants' phones, whereas health care providers receive results directly on their study tablet together with the recommended action. Further features of the platform are preference-based tailored adherence reminders and automated calls to participants for symptomatic tuberculosis screening. The proposed aDSDM is designed for being scaled up at national and regional level as it mainly builds on automated triage and communication with participants and health care workers, thus not requiring additional human resources.