In our formative research, analysis of antiretroviral treatment (ART) data manually entered in the Three Interlinked Electronic Registers (TIER.Net) showed poor viral load monitoring (VLM) and inadequate management of virological failure in HIV-positive patients on ART in rural KwaZulu-Natal, South Africa. ART interruption was high, with nearly half of patients falling out of care within 5 years of starting ART. Non-Nucleoside reverse transcriptase pre-treatment drug resistance exceeds 10% in the setting; the threshold required to trigger in a change in first-line ART using the public health approach. These factors are contributory to increasing HIV drug resistance (HIVDR) in this setting. HIVDR is associated with increased morbidity and mortality with the risk of transmitting drug-resistant HIV to sexual partners. The investigators presented these findings to healthcare providers, policy makers and community representatives with brainstorming of health system challenges and potential interventions. This study aims to complement these findings by investigating the clinical and process impediments in VLM within the health system and to develop a quality improvement package (QIP) to address the gaps. The stakeholders recommended such QIP would utilise the viral load (VL) champion model, a named healthcare provider who would be the focal point for ensuring proper administrative management of viral load tests and results through identification of those who need tests and triaging of results for action. This QIP will be supported by technological enhancement of the routine clinic-based TIER.Net software which will allow daily automatic import of results from the National Health Service Laboratory (NHLS) to TIER.Net and development of a dashboard system to support VLM. In addition, results of contact tracing will be recorded and followed up pro-actively if not initially successful.

The investigators will evaluate the effectiveness of these interventions compared to standard care for improving VLM and virological suppression using an innovative effectiveness-implementation hybrid cluster-randomised design in 10 clinics. A within-trial health economics analysis will be undertaken using recommended methods to examine the cost-effectiveness of the intervention compared to standard care.

South Africa has the biggest HIV treatment programme in the world with 7.7 million individuals living with HIV and 62% currently receiving ART within a stretched health system. VLM has been part of the public ART programme since roll-out in 2004 and requires people with HIV initiating ART to have a VL measured at 6 months and 12 months after ART initiation and 12-monthly thereafter if virologically suppressed. Those with a VL ≥ 1000 HIV-1 RNA copies/mL should be retested after 3 months of adherence counselling support, and then either retained on first-line therapy if re-suppressed or switched to second-line therapy if VL ≥ 1000 copies/mL.

However, little is known about how these VL guidelines are being used in clinical decision-making in public ART programmes in sub-Saharan Africa. In formative research utilising an electronic database, the Three Interlinked Electronic Register (TIER.Net), of a programmatic ART cohort in rural KwaZulu-Natal, the investigators observed infrequent VLM and sub-optimal management of virological failure. The study showed that only 34% of patients had a viral load documented after 12 months on ART. Only 20% of individuals in the cohort were confirmed to have virologic re-suppression or change to second line therapy after virologic failure, and those that did change therapy did so a median of one year after virologic failure. With the expansion in the indications for ART use, such delays are likely to have significant deleterious individual and public health impacts through effects on patient morbidity, accumulation of drug resistance, and persistent risk of HIV transmission in the setting.

The investigators hypothesise that a staff-centred quality improvement package (QIP) and technological augmentation of an existing electronic ART database (TIER.Net) would result in optimal VLM of patients on ART, prompt clinical management of virological failure and an overall improvement in virological suppression.

Main trial objective

The main objective is to evaluate the impact of a combination of interventions that includes a staff-centred quality improvement package, designated viral load monitoring champion, and augmentation of TIER.Net with a dashboard system will lead to improvement in viral load monitoring and virological suppression over a period of 12 months in comparison to the current standard of care.

Secondary objectives

• To identify health system specific gaps in VLM.
• To evaluate the cost and cost effectiveness of the intervention compared to standard care

Inclusion Criteria:

• All nursing care staff employed at the health facilities are eligible for inclusion to receive the quality improvement intervention
• Willing to provide informed consent
• Work in one of the ten facilities randomised for the study
• Involved in HIV patient care
• Patients receiving care from participating clinics must be aged 16 years and above to be included in the study

Exclusion Criteria:

• Expecting to relocate or change jobs during the study duration
• Unwilling or unable to provide informed consent/refusal to participate

The 10 clinics will be randomly allocated to the intervention (5 clinics) and control (5 clinics) arms. Randomisation will be stratified on clinic size, defined by the number of HIV-related visits per month (2 strata). Restricted randomisation will be used to ensure balance between study arms on the following important covariates: proportion of HIV patients who are male; proportion of HIV patients who are aged <25 years. A computer programme will be used to prepare a list of all permissible randomised combinations; community leaders will be invited to make a random selection from the list at a public ceremony. The randomisation list will be prepared by an independent statistician and will be concealed until after randomisation. It will not be possible to blind the research staff and nurses to the intervention.