Evaluation of a health system intervention to improve virological management in an antiretroviral programme at a municipal clinic in central Durban

Background With the largest antiretroviral therapy (ART) programme globally, demand for effective HIV management is increasing in South Africa. While viral load (VL) testing is conducted, VL follow-up and management are sub-optimal. Objectives The objective of this study was to address gaps in the VL cascade to improve VL testing and management. Methods Antiretroviral therapy records were sampled for an in-depth review. The study team then reviewed individual records, focusing on ART management, virological suppression and retention. Multifaceted interventions focused on virological control, including a clinical summary chart for ART care; streamlining laboratory results receipt and management; monitoring VL suppression, flagging virological failure and missed visits for follow-up; down-referral of stable patients eligible for the chronic club system; and training of personnel and patients. Results Pre-intervention, 78% (94/120) of eligible patients had VL tests, versus 92% (145/158) post-intervention (p = 0.0009). Pre-intervention, 59% (71/120) of patients accessed their VL results, versus 86% (136/158) post-intervention (p < 0.0001). Post-intervention, 73% (19/26) of patients eligible for ART change were appropriately managed, versus 11% (4/36) pre-intervention (p < 0.0001). Only 27% had no regimen changes (7/26) post-intervention, versus 81% (29/36) pre-intervention (p < 0.0001). Conclusion Service delivery was streamlined to facilitate HIV services by focusing on VL test monitoring, protocol training and accessibility of results, thereby improving clinical management.


Introduction
South Africa has the largest antiretroviral therapy (ART) programme in the world with over 4.1 million adults on treatment by the end of 2017. 1 Despite expanded ART access, incident HIV infections are still substantial with approximately 270 000 individuals infected with HIV in 2017. 1 In 2014, UNAIDS launched the 90-90-90 targets to progress towards elimination of HIV by 2020. 2 The last '90' of the UNAIDS strategy is to ensure that 90% of individuals receiving ART have achieved virologic suppression. 2 Plasma viral load (VL) is the main driver of new HIV infections. 3 HIV replication is interrupted by ART, thereby reducing VL, decreasing transmission risk and improving morbidity and survival. 4 Post-initiation, VL is indicative of ART response. 5 In practical terms, to achieve the UNAIDS 90-90-90 goals towards HIV elimination, 73% of people living with HIV would need to be virologically suppressed to reduce the reproductive rate of infection to below 1. 2 In a study of South Africa's ART 'treatment cascade' using data from the national health laboratory service (NHLS), almost half of all South Africans infected with HIV were not linked to care. 6 The authors were not able to report the proportion of diagnosed people who were not linked to care. However, among those linked to care and on treatment, only 73.7% had suppressed VLs. The overall http://www.sajhivmed.org.za Open Access proportion of viral suppression among South Africans living with HIV was 23.8%. 6 Part of the challenge of achieving virologic suppression may be the complexity of VL monitoring -from attaining patient blood samples, transport of laboratory specimens, documenting results in patients' files and appropriate management of virological outcomes. 7 Moreover, maintaining viral suppression may be particularly challenging with psychosocial, cultural and economic obstacles to adherence, including efficient systems for laboratory monitoring and adequate, timely drug supply. 8 Given the multifactorial barriers, maintaining drug adherence may necessitate a hierarchical approach with patients at risk triaged for additional clinical and psychosocial support. 8 The primary objective of this health systems intervention was to address gaps in the VL cascade in a real-world setting to improve VL testing and management. In this article, we present the evaluation of results for this health system intervention.

Study design
The study team conducted a health system intervention to improve the quality of HIV services, focusing on virological management.

Setting
The eThekwini district of KwaZulu-Natal (KZN) had an HIV prevalence of 11.4% in 2016 9 and one of the highest antenatal HIV prevalence rates in the country at 41.1% in 2013. 10 The HIV programme implemented at the Lancers Road Primary Healthcare (PHC) clinic is managed by the eThekwini Health Unit, situated near a busy taxi rank in Durban. This PHC offered a unique opportunity to determine if a quality improvement intervention could improve virological management in a setting servicing a highly mobile population. It offers standard PHC services: antenatal care, well-baby clinic and/or immunisations, HIV counselling and testing, ART provision, tuberculosis (TB) screening and treatment and other chronic care services to patients from the greater functional region within 100 km radius.

Routine HIV management
Approximately 60-70 HIV-positive patients were managed daily at the PHC by a single nurse clinician with weekly medical officer support. During the study period from 2011 to 2015, the South African Department of Health (DOH) CD4 + threshold for ART initiation was revised from 200 cells/ mm 3 in 2013 11 to < 500 cells/mm 3 in 2015. 12 Once initiated on ART, patients were followed up monthly by nurses who were responsible for clinical and laboratory monitoring, dispensing of medications, re-enforcing adherence messages and providing psychosocial support. Consultations focused on ART initiation and dispensing, with little time for ensuring the accessing and management of VL results, or for comprehensive clinical care such as management of communicable and non-communicable diseases. In addition, there were few opportunities for doctors to mentor nurses. The guideline also recommended down-referral of stable patients who were suppressed virologically to community-based 'chronic clubs' for ongoing care.
The current South African guideline recommends VL testing following ART initiation at 6 months, 12 months and thereafter annually. 12

Study population and sampling strategy
All patients (n = 1538) enrolled in the ART programme at the PHC facility from 01 September 2011 to 31 March 2014 were included in pre-intervention reviews. Furthermore, the authors systematically sampled 13% of all patients initiated on ART between 01 September 2011 and 31 March 2014 for in-depth file reviews.
Every 10th file was selected for review and recording data captured until end of March 2014. In addition, files of patients commenced on ART in the 3 months following the initial file review period, that is, 01 April 2014-30 June 2014, were reviewed post-intervention.

Health systems assessment
Our research involved file reviews at several time  Table 1).

Pre-intervention (Figure 1)
The study team assessed the baseline operational issues of the clinic to guide our interventions and focus. The review was conducted as follows: (1) In-depth file review (IDR); (2) VL and retention in care review (VLRIC); (3) VL management review (VLMR).

Intervention period
Based on the in-depth review, the following combination of interventions was implemented: (1) training sessions; (2) clinical summary chart insertion ( Figure 2); (3) improved laboratory systems; (4) VL management system; (5) follow-up system for missed appointments and lost to follow-up and (6) chronic club care.

Post-intervention assessment
The effectiveness of our combination of interventions at the clinic was measured as follows: (1) improvement in access to VL results, whether performed and included in the client file; (2) VL management follow-up (reassessed clients found with high VL during the VLRIC); (3) clinical summary chart usage review (uptake of clinical summary sheet on new clients enrolled in the ART programme) and (4) down-referral system (chronic clubs).

Data collection
All pre-and post-intervention data extracted from the patient files were recorded on appropriate questionnaires. The data were entered into a specific database developed by research clinicians. The database was secured using passwordprotected access systems. All data were de-identified to maintain patient confidentiality; all specific laboratory

Pre-intervention
In-depth review Possible reasons for failure to receive a laboratory result were as follows: blood tests not requested nor performed; transport logistics; rejection of blood specimens and misplacing of the printed laboratory result form at the clinic.

Viral load management review
Pre-intervention, in the 68 patients with raised VLs, it was noted that only 30% had a repeat VL. Repeat VLs were seldom done timeously in accordance with DOH protocol with 22% conducted > 6 months ( Table 2).
Of the 68 patients, 53% (36/68) had virologic failure and were eligible for a regimen change (in care with at least two prior raised VL results). Pre-intervention, of the 36 who were eligible for regimen change, 81% (29/36) had no regimen change when needed, 8% (3/36) had an inappropriate regimen change and 11% (4/36) had appropriate regimen change according to standard guidelines. However, the appropriate regimens were only prescribed 15 months after the initial raised VL.
Overall, there was also a significant increase in those who had access to their results (86%; 136/158) (p < 0.0001).

Discussion
This project shows that significant improvements in VL testing and monitoring can be achieved through a combination of multifaceted interventions, including the implementation of training, review of patient files and the strategic addition of monitoring systems. We report improvements in the VL tests conducted and the proportion of patients who had access to VL results, which improved by just over 25%. In addition, improvements in management of HIV-positive patients with raised VLs were noted, with a significant reduction in the number of patients who did not require a regimen change, and a significant increase in appropriate regimen changes postintervention.
In a South African study modelling national and provincial HIV data, 78% of HIV-positive adults on ART had suppressed VLs in 2015; however, only 38% of HIV-positive adults overall had virological suppression. 13 Better virological suppression rates were reported in KwaZulu-Natal with 85% of HIV-positive patients on ART with undetectable VLs, but only 47% HIV-positive adults were virologically suppressed. 13 A notable finding of this study was that 73% of the patients who had a VL result available pre-intervention were virologically suppressed, lower than the national (78.4%) and provincial estimates (excluding Limpopo and Mpumulanga). 13 Virologic suppression was higher in this study than described in other South African ART cohorts. 14 However, under twothirds of patients had an available VL test and patients with high VLs were not appropriately managed.
The outcomes measured showed the effectiveness of the interventions, particularly the availability of VL results for clinical management improving from 59% pre-intervention to 87% post-intervention. There was also an improvement in management of raised VLs, with a significant increase in repeat VL tests from 78% pre-intervention to 92% postintervention. Furthermore, there was a significant increase in appropriate regimen changes from 11% pre-intervention to 73% post-intervention, and timelier follow-up of these patients.
The findings of this study with a quality improvement intervention focusing on the VL monitoring and recognition of treatment failure are similar to other studies in South Africa and Malawi. In a South African study in three clinics in eThekwini in KwaZulu-Natal, a VL champion was designated to focus on virologic management; following the intervention, the VL tests performed significantly increased from under 70% at the three clinics to over 80%. 15 In a study at 13 facilities in Malawi over a 6-month period, changing staff roles with a VL 'focal person' to oversee all VL activities improved VL testing 164% pre-intervention versus post-intervention. 16 Furthermore, patients are vulnerable to non-adherence and disengagement the longer they remain on treatment. 17 In a retrospective cohort study in Cape Town, South Africa, almost 23% of patients disengaged from care within 2 years of ART initiation with 16% requiring hospital admissions and 3% dying following disengagement. 17 Moreover, the WHOdefined VL threshold (VL ≥ 1000 copies/mL) for intervention of virological failure may not adequately identify patients at risk for adverse clinical outcomes. In a recent South African study with over 69 000 patients, HIV-positive patients on ART with low-level viraemia (≥ 50 copies/mL) were threefold more likely to develop virologic failure than patients with undetectable VL; the risk increased to almost fivefold when using VL 400 copies/mL -999 copies/mL. 18 According to the South African ART guidelines, an elevated VL (> 400 RNA copies/mL) indicates either poor treatment adherence or resistance, necessitating adherence support and drug therapy changes for treatment optimisation. 11,19 Low adherence may also lead to virological failure, transmission and drug resistance risk. 20,21,22 South African data highlight the increasing resistance patterns exacerbated by patients with an unsuppressed VL. 23 There were several contributory factors to the success of our intervention. The project was conducted in a well-functioning public health system with an inculcated culture of reflective data analysis and quality improvement. The NHLS system facilitated the use of data to drive improvement as it is linked to patient data that are meaningful to frontline staff. The health system leadership used VL process data feedback to encourage participation in learning opportunities and supported the testing of new ideas and the spread of successful interventions.
Another factor for success was the perceived added value and acceptance of the intervention in the healthcare setting by the staff who were using it. This acceptance is reflected by the large proportion of patient files with clinic summary sheets inserted by clinic staff. These results are also indicative of the feasibility of using summary sheets in busy healthcare settings. The clinical summary sheet also proved useful for the completion of the required governmental forms for audit purposes. While there were fields to assist with integrated care provision such as pregnancy or HIV-positive patients co-infected with TB, the clinical summary chart was mainly used for ART management, and assessment of patient eligibility for down-referral to a chronic club. While the initial investment in summarising patient data may be unwieldy, long term, the summary sheet may mitigate the necessity to peruse through long and often complex clinical notes, saving time for both the patient and provider during the consultation. Moreover, if there is substantial buy-in from staff, a clinical summary sheet could be inserted for all patients newly initiating on ART or pre-inserted into files of patients presenting for care the next day.
There is limited data on the association between summary sheets and improvement in clinical care; however, the integrated summary of prospectively collected clinical information can assist clinical staff in organising complex patient data to facilitate management. 24 In South Africa, HIV care is decentralised to PHC level; ART initiation and follow-up care are primarily nurse-managed with task shifting to less specialised health workers.

Limitations
Assessment of long-term effectiveness of interventions implemented during this operational research is difficult to measure accurately for the following reasons: difficulty of measurable end points; the presence of the research clinicians pre-and post-intervention, with availability for assistance and advice, and a split focus between intervention and review; and the progressive nature of the research, with the need to design tools for measurement while interventions are occurring.
The multifaceted interventions in this study included support from two part-time research clinicians with HIV experience and research assistants who were actively involved in patient care. Hence, while the interventions were fairly simple, it may not be feasible to institute this level of health systems strengthening at all clinics. Moreover, it is difficult to assess which intervention had the most effect given the package of care. Finally, while there was some emphasis on empowering patients in their HIV care, the effect of this on knowledge and behaviour was not measured.
Follow-up of patients not returning to the clinic was difficult to assess because there was no appointment system in place, and missed visits were not easily identified until a patient returned to the clinic and the file was taken out, or, as in this study, all the files reviewed. Despite this, retention in care was 88% and 79% in the in-depth review and the VL retention in care review, respectively, possibly indicative of the commitment of patients to the care they were receiving at the Lancers' Road PHC. In a prior retrospective study of South African ART cohort, overall retention rates fell from 71% at 24 months to 56% at 60 months; notably these study sites were well resourced. 25 Currently, programmatic data are limited regarding retention in care in adults in South Africa; studies evaluating postpartum retention in care in PMTCT also noted challenges in attrition because of patientlevel factors (disclosure to spouse and competing priorities), provider-level factors (clinical knowledge and attitudes) and wider health systems barriers, such as access to care and drug stock-outs. 26 The attrition rate following access to universal ART is unclear and there may be a larger loss to follow-up following the 2016 guideline revision to initiate all patients on ART, irrespective of CD4 + count. 27 The gaps in patient care at the PHC clinic may be accounted for by the high patient volume, and prioritisation of ART initiation and dispensing without the assistance of a pharmacist. Moreover, there were few opportunities for clinicians to mentor nurses. Clinical mentoring has been highlighted by WHO to be a vital strategy for the scale of ART and improving healthcare in developing countries. 28 As this study was conducted at a municipality PHC, it may not be generalisable to other PHC facilities in South Africa. However, it may be feasible to implement most of the lessons learned in improving the quality of virologic management in this study.

Conclusion
In this study, multifaceted interventions focusing on laboratory systems, staff training and mentoring, streamlined documentation and a strong referral and follow-up system contributed to improvements in virologic management in the time period evaluated.
The third '90' UNAIDS 2020 target is therefore achievable by the implementation of simple, appropriate and targeted systems in a busy PHC setting.

Recommendations
While it may not be feasible to place extra clinicians at clinics for the sole purpose of mentoring and documentation review, several of the interventions are easy to implement in a resource-poor setting. This includes the clinical summary chart as a simple way to monitor VL status. Moreover, routine education of clinical staff on basic South African governmental protocols is achievable. Finally, with the introduction of Tier.net, an electronic database for HIV-positive patients and access to the NHLS laboratory results system, improvements to the result availability are feasible.