Task shifting and integration of HIV care into primary care in South Africa: The development and content of the streamlining tasks and roles to expand treatment and care for HIV (STRETCH) intervention

Background Task shifting and the integration of human immunodeficiency virus (HIV) care into primary care services have been identified as possible strategies for improving access to antiretroviral treatment (ART). This paper describes the development and content of an intervention involving these two strategies, as part of the Streamlining Tasks and Roles to Expand Treatment and Care for HIV (STRETCH) pragmatic randomised controlled trial. Methods: Developing the intervention The intervention was developed following discussions with senior management, clinicians, and clinic staff. These discussions revealed that the establishment of separate antiretroviral treatment services for HIV had resulted in problems in accessing care due to the large number of patients at ART clinics. The intervention developed therefore combined the shifting from doctors to nurses of prescriptions of antiretrovirals (ARVs) for uncomplicated patients and the stepwise integration of HIV care into primary care services. Results: Components of the intervention The intervention consisted of regulatory changes, training, and guidelines to support nurse ART prescription, local management teams, an implementation toolkit, and a flexible, phased introduction. Nurse supervisors were equipped to train intervention clinic nurses in ART prescription using outreach education and an integrated primary care guideline. Management teams were set up and a STRETCH coordinator was appointed to oversee the implementation process. Discussion Three important processes were used in developing and implementing this intervention: active participation of clinic staff and local and provincial management, educational outreach to train nurses in intervention sites, and an external facilitator to support all stages of the intervention rollout. The STRETCH trial is registered with Current Control Trials ISRCTN46836853.


Background
South Africa has the largest human immunodeficiency virus (HIV) burden in the world, with an estimated 5.7 million infected people [1]. By the end of 2008, five years after the public sector antiretroviral treatment (ART) programme was launched, an estimated 700,500 people were accessing ART [2]. Although this represents an increase of 53% on the previous year, it constitutes only 40% of those estimated to be in need of ART [3]. Despite policy guidelines recommending that comprehensive HIV care be incorporated into existing primary care services [4], the initial public sector ART rollout in South Africa was implemented as a vertical (stand alone) programme with separate funding, facilities, staff, medical records, and reporting requirements [5]. There are several reasons to justify such an initial vertical approach to comprehensive HIV care, including the need for a rapid response in a weak health system and the need for highly skilled staff to implement a new, complex intervention [6,7]. There are, however, two equally powerful reasons for moving away from vertical HIV care programmes in high HIV-burden countries: that such vertical programmes will be unable to achieve universal ART access because of the sheer numbers of people needing treatment; and that they could draw away financial and human resources from already struggling public health systems in these countries [8,9].
In order to address these concerns, calls have been made to utilise the impetus of new financing, training, and infrastructural support, directed towards the acquired immunodeficiency syndrome (AIDS) epidemic, to strengthen broader health systems [10], and to incorporate current vertical ART programmes into these health systems-a strategy now termed the 'diagonal approach' [11]. Approaches to incorporating HIV care into general health systems include: the referral of patients stabilised on ART from ART clinics to primary care clinics where they could receive monthly supplies of treatment (sometimes referred to as 'down referral') [12,13]; task shifting of aspects of HIV care to lower cadres of healthcare workers [14,15]; setting up nursedriven HIV care programmes [16]; and integration of HIV care into primary care services [17][18][19].
These types of interventions are complex, and there are two important research questions that need to be answered, particularly in low-and middle-income countries [20]: What should be the components of these interventions [21][22][23]? And are these interventions effective in improving access to ART? This article addresses the first question-it describes the content of the STRETCH (Streamlining Tasks and Roles to Expand Treatment and Care for HIV) intervention, including its components, the processes of change used, the conditions in the control clinics, and links to manuals used in the intervention, as suggested in the WIDER recommendations (Workgroup for Intervention Development and Evaluation Research) [24]. The development of the intervention was based on the educational outreach model and our practical experience of engaging with the Free State Department of Health in implementing an earlier nurse training programme called PALSA PLUS (Practical Approach to Lung Health and HIV/AIDS) in the Free State [25][26][27]. The second question is being addressed through a pragmatic cluster randomised controlled trial of the effects of the STRETCH intervention on access to ART conducted in 31 ART clinics randomised in nine strata in the Free State province [28]. This description will supplement the forthcoming trial results.

Context and setting: the Free State public sector ART rollout
The Free State, with a population of 2.8 million [29], has an estimated HIV prevalence of 18.5% among 15 to 49 year olds [30]. The province comprises five districts, divided into 20 local areas, with primary care services offered at 222 nurse-led clinics. The public sector ART rollout commenced in mid-2004 in designated nurse-led ART assessment sites situated in selected primary care clinics. Table 1 summarises the organisation of HIV care in health facilities in the initial rollout. Patients diagnosed as HIV positive in primary care clinics and hospitals are referred to ART assessment sites for further clinical care and assessment of eligibility for ART. Those eligible for ART receive drug readiness training and are then referred to ART treatment sites in local hospitals for initiation of treatment and for threeto six-month reviews of ART prescriptions by a doctor. National regulations require that antiretrovirals (ARVs) be dispensed by or under the direct supervision of a pharmacist. Where assessment sites do not have pharmacists, ARVs have to be dispensed at treatment sites into patient-named packets and transferred to assessment sites where nurses issue them monthly to patients. In some remote areas, assessment and treatment site functions were conducted by combined sites with the support of visiting doctors.
In the first three years of the rollout, achievements included: good patient outcomes amongst patients receiving ART [31,32], a reliable supply of drugs and other medical supplies, and increases in nurse posts [33]. These successes were tempered by high mortality rates among patients waiting for ART [31], increased vacancies in primary care services [34], and high levels of burnout among ART and primary care nurses [35]. Despite opening 57 ART sites, coverage by the end of 2007 remained disappointingly low. Only 25% of new patients estimated to be in need of ART that year were started on treatment [36].
In late 2008, while the STRETCH trial was ongoing, the Free State ART programme was forced to implement a three-month moratorium on selected adult ART initiations to ensure uninterrupted drug supplies for those already on treatment. This moratorium was due in part to chronic underfunding of the ART programme in all provinces, and resulted in a major review and increase in funds for the national ART programme. In early 2010, before the STRETCH trial was completed, the South African government commenced implementation of its accelerated AIDS plan in all provinces. This plan includes nurse prescription of ART and integration of ART into all primary care clinics in an attempt to rapidly scale-up ART access [37] Developing the intervention In 2005, Free State Department of Health managers expressed their concern about high mortality rates among patients waiting for ART, and about the dependence of the programme on doctors, who are in short supply, for ART prescription. Working in the Free State, the Knowledge Translation Unit of the University of Cape Town Lung Institute had piloted and evaluated a training programme for nurses in the use of integrated primary care guidelines covering the management of respiratory diseases and HIV-the PALSA PLUS initiative [25][26][27]38,39]. The provincial department thus requested that nurse prescription of ART be included in the PALSA PLUS guidelines, and that training be rolled out in the province. Because of widespread ambivalence about the ability of nurses to take on the clinical responsibility for ART prescription and the absence of clear national policy, it was decided to pilot the intervention and monitor its outcomes as a pragmatic randomised controlled trial in the province's ART clinics. Meetings were then held over eighteen months between researchers, managers, senior clinicians, and clinic staff to develop the intervention.

Meetings with senior managers and clinicians
In initial meetings with senior managers and clinicians from the ART programme, it was established that delays in people accessing ART were caused not only by the shortage of doctors but also the high caseload of ART nurses at ART assessment sites that were managing growing numbers of patients on ART as well as those not yet eligible for ART. The intervention was therefore designed to be a more complex task-shifting intervention with two main components: shifting ART prescription from doctors to ART nurses and shifting routine HIV care for patients not yet eligible for ART (pre-ART care), from ART nurses to primary care nurses at ART assessment sites.

Meetings with middle managers
Workshops were then held with district and local area managers to further develop the intervention. Managers expressed concern about the ability of nurses to assume these new clinical responsibilities and about how to implement the reorganisation of care required for this type of complex health intervention. It was agreed that in addition to providing nurse training, the intervention would be implemented in phases, and detailed descriptions of the task and role changes needed at intervention clinics in each phase would be included in an implementation 'toolkit' to be developed by the researchers.

Meetings with clinic staff
To obtain feedback from clinic staff on the proposed intervention, the STRETCH coordinator (KU) visited all 31 nurse-led ART assessment clinics selected for the trial and held meetings with staff members. The staff raised a number of problems with functioning of the ART sites that were resulting in difficulties for patients accessing ART. These difficulties included increasing workload, drug transport and storage problems resulting from hospital-based ART dispensing, transport problems for patients, and lack of basic communication infrastructure such as telephones and fax machines (see Table 2). ART nurses were also struggling to cope with providing care for the growing numbers of patients accessing ART as well as those not yet eligible for ART. In one local area where primary care clinics did not offer HIV testing, ART staff had to provide this service too. However, in other districts, increasing workload had already prompted ART sites to integrate pre-ART care into the work of the surrounding primary care clinics. In one district, ART sites were already discussing the integration of drug readiness training, for patients eligible for ART, into primary care services. Thus, in their comments on the proposed intervention and in order to address some of the problems outlined in Table 2, such as nurse workload and transport difficulties for patients, many of the staff felt that more elements of HIV care, including drug readiness training and monthly collection of ARVs, needed to be integrated into primary care services. Furthermore, these elements of care needed to be available not only within the ART clinic but also in surrounding primary care clinics referring patients to these ART sites. Task shifting of pre-ART care from ART nurses to primary care nurses at ART sites, as initially envisaged in discussions with management, was thus reformulated as a step-wise integration of the following six elements of comprehensive HIV care into all primary care services both within the ART clinics and those at clinics referring patients to the ART nurses at the ART sites: voluntary counselling and testing; initial CD4 count; routine HIV care for patients not yet eligible for ART; drug readiness training for patients initiating ART; baseline blood tests for patients initiating ART; and monthly ART care for stable patients. This 'decentralisation checklist' was included in the implementation toolkit.
A meeting was also held to gather the views of primary care nurses in the 16 ART sites. These nurses were concerned about the burden of HIV disease in their patients, were keen to be involved in the programme, and felt capable of providing comprehensive HIV care. However, they were also concerned about the increased workload this would create for healthcare providers in already overloaded and understaffed primary care services.

Components of the intervention
The main components of the intervention are discussed below and are summarised in Table 3, where they are compared with standard of care support at control clinics.

The STRETCH coordinator
A provincial STRETCH coordinator (KU), a family medicine practitioner with experience in the management of HIV/AIDS and tuberculosis, was appointed and had the following responsibilities during the intervention: further developing the intervention in consultation with staff at management and clinic level; involvement in initial training and continuing support of nurse training at intervention sites; teaching in the Free State ART training programme alongside ART programme doctors; helping to provide clinical advice to all ART sites; providing extra support to nurses prescribing ART at the intervention sites; and facilitating the establishment of management teams to oversee the implementation of the intervention. The involvement of the STRETCH coordinator in teaching in the ART programme and helping to provide clinical advice to all ART sites was not initially envisaged as part of the intervention, but was included at the request of the province because of the shortage of doctors available to provide this support.

Regulatory changes
Although there was no official national policy prior to the trial on nurse prescription of ART, two pieces of national legislation supported such prescription [40,41]. The Free State Pharmaceutical and Therapeutics Committee gave permission for professional nurses in the province to initiate and repeat ART prescriptions for adults during the trial. This permission was conditional on these nurses completing appropriate training and Nurse training Table 4 summarises the characteristics of the ART training available to nurses in all clinics across the province and the training offered as part of the intervention. The details of these training programmes are described below.

Standard of care training in all clinics
Since 2005, the Free State Department of Health has been running a regular two-week ART training course for staff in ART and other primary care clinics. This course combines one week of lectures broadcast to classrooms throughout the province and a one-week placement at an existing ART site. Regular maintenance training is also conducted in the districts and in weekly lectures broadcast to staff in these classrooms. Clinical support was available to staff at all ART sites from doctors at treatment sites, specialists at a tertiary level AIDS clinic and the STRETCH coordinator. At the time of the trial, PALSA PLUS training was being rolled out to all provincial primary care clinics, including all ART assessment sites [27]. This model of training involves equipping nurse managers to conduct outreach training for nurses at clinics in their area. Nurse managers are trained in a one week course known as Training the Trainer to Train (TtTtT) [25]. Adult education models are used to fully integrate experiential learning on how to facilitate small group training using case scenarios, while enabling the trainers to become familiar with the contents of the guideline. These nurse managers in turn conduct outreach training onsite, in short sessions over several weeks, using these case scenarios to facilitate nurses engaging with the PALSA PLUS guideline. This training has been shown to be effective in improving quality of care and minimises disruption to clinic services [26,27]. Thirty of the 31 ART sites in the STRETCH trial had completed PALSA PLUS training before the trial began and plans were made to train staff at the outstanding clinic.

Training at intervention clinics
The PALSA PLUS model of training was expanded to include extra training in nurse prescription of ART. One established PALSA PLUS trainer was identified for each of the 16 intervention clinics. All had been trained in ART, and three had experience working in ART sites. These trainers were either clinic supervisors or local programme coordinators regularly visiting these clinics in a supervisory capacity. They participated in a two and one-half-day training on: how to train nurses in the ART protocols contained in the STRETCH edition of the guidelines by using four case scenarios; and the staff role changes needed as part of the intervention, as described in the toolkit. We anticipated that nurse confidence might be severely compromised if patients who were started on ART by nurses developed severe side effects. The case scenarios were therefore also used to impart basic skills for trainers to debrief nurses. The training was led by three facilitators from the research team: two nurses experienced in adult and nurse education who had been involved in developing the PALSA PLUS training (GF and PM), and the STRETCH coordinator.
The trainers then trained all nurses at the 16 intervention clinics, including designated ART nurses and those working in primary care, commencing in August 2007. A minimum of four educational outreach trainings, one of which was supported by the STRETCH coordinator, were conducted at each clinic, and most of these sessions were completed by October 2007. The trainers continued to support the nurses and train those who were newly appointed or had not attended all the initial sessions, but the regularity of these visits varied and depended on their other supervisory responsibilities.
All doctors supporting the intervention sites were oriented by the STRETCH coordinator using the guidelines and case scenarios. Doctors working in the five combined sites were able to provide clinical support to the nurses. However, at the other eleven assessment sites, where doctors only worked at distant treatment sites, they were less able to provide support. Additional clinical support was also provided by the STRETCH coordinator via telephone or during clinic visits. These visits took place typically once every four months in the first twelve months of the trial and less frequently after that.

Patient management guidelines for nurses
Nurses working in all primary care clinics including all ART sites had access to and were receiving training in the use of the PALSA PLUS guidelines (see above). A STRETCH edition of the PALSA PLUS guideline, containing algorithms for nurse initiation and management of adults on ART, was distributed to all nurses in the 16 intervention clinics and used in outreach training by the STRETCH trainers. The algorithms were developed in consultation with clinicians in the province and with reference to the Integrated Management of Adolescent and Adult Illnesses guideline [42]. Thus, adults with a CD4 <50, Stage 4 HIV, previous ARV treatment, who were on tuberculosis (TB) or other chronic medication, were bedbound, or who were pregnant were identified as potentially complicated cases that needed to be initiated onto ART by a doctor. All other adults eligible for ART could be initiated by nurses. Similarly, a decreasing CD4 count, detectable viral load, or clinical problems in a patient already receiving ART were criteria for doctor management, while all other patients could be managed by a nurse. (The ART algorithms are included in Additional file 1)

Phased introduction
The intervention was implemented in phases to support logistical changes such as the dispensing of nurse ART prescriptions and to allow nurses to build confidence and skills in ART prescriptions. The three phases of implementing the intervention were: the training of nurses in ART prescription and setting up of management support teams; nurse re-prescription of ART for stable patients; and nurse initiation of ART for uncomplicated new patients. The timing of progress through the stages was determined by staff in the STRETCH teams at each individual clinic.

Implementation guideline
Because of the complexity of the intervention, the research team developed an implementation guideline called the STRETCH Toolkit and distributed copies to all intervention sites. The Toolkit contained the decentralisation checklist (as outlined above), descriptions of the different phases of the study, as well as details about the changing roles of all staff members in each phase and useful advice on communicating these changes to the community. It also contained important documents and information, such as contact numbers for doctors and nurse managers of all the clinics in the trial and relevant managers in the provincial department, along with copies of documents authorising nurse prescription of ART. (The STRETCH Toolkit is included in Additional file 2)

Management support
Standard support was provided to all ART sites by two to three monthly visits from district ART coordinators (who had district wide responsibility for the ART programme) and monthly visits from clinic supervisors (who were responsible for overall primary care services in a local group of clinics). Meetings between clinic managers (in charge of each clinic) and local area managers (who had overall responsibility for health services in that local area) are typically held at one-or twomonth intervals.
During phase one of the intervention, STRETCH teams were convened by the STRETCH coordinator at each of the intervention clinics. These teams usually comprised the clinic manager, one clinic nurse representing ART services and one representing primary care, and the pharmacist or pharmacy assistant, as well as staff from the treatment site and the district ART coordinator. These teams were given copies of the STRETCH Toolkit and were tasked with implementing changes at the clinic during the intervention. One of these tasks, as outlined in the decentralisation checklist, was to assess the state of integration of comprehensive HIV care into primary care services, and which further elements of HIV care needed to be integrated into these services ( Table 1).
Thirteen of the intervention clinics had patients referred for ART from other primary care clinics in their area. In four of these intervention clinics, local management had already started implementing the integration of all six elements of HIV care into the primary care clinics. In the other nine intervention clinics, the STRETCH team identified the need to integrate further elements of HIV care into these referring clinics. Local area management teams were then convened for seven of the nine clinics. In the remaining two clinics management support was difficult to mobilise. These teams usually comprised the local area manager, the manager of the intervention site, facility managers of all referring primary care clinics, and the local ART pharmacist. They were able to evaluate capacity to integrate further elements of HIV care into the referring clinics by assessing staffing and training needs, space for drug readiness training classes, and ability to store and transport ARVs-all of which were the type of practical issues identified by staff ( Table 2). The STRETCH coordinator's responsibility was to convene these management teams and assist at the first one or two meetings. It was then the team's responsibility to decide which elements of HIV care could be integrated at which primary care clinics and to implement these decisions.

Discussion
One of the distinctive features of this intervention was the participation of clinic staff and all levels of management in many stages of its development and implementation. First, the trial was set up at the request of senior management to address the problem of high mortality rates among patients eligible for ART and awaiting access to treatment. In the national environment of ambivalence to nurse ART-prescription that existed at the start of the trial, senior management support was crucial to developing and implementing the intervention. Second, senior management, middle management, and clinic staff were involved in an iterative process of assessing the barriers facing patients and staff with regard to accessing ART, and then tailoring the intervention to be relevant and implementable. Management concerns about the complexity of the intervention led to the development of an 'Implementation Toolkit.' The types of problems outlined by staff (Table 2) and their insight into possible solutions led to the reformulation of integration in the context of ART rollout as the flexible, progressive integration of pre-ART and ART care into all primary care services referring to intervention sites. Third, staff at local area and clinic level were involved in the teams tasked with implementing the intervention, with support from the STRETCH coordinator. STRETCH teams were tasked with assessing readiness for different phases of the intervention and with implementing the changes at clinic level. Local management teams assessed capacity and arranged for primary care services to take on aspects of pre-ART and ART care.
The strong participation of clinic staff and managers in intervention development and implementation could be seen as an example of how features of participatory action research can be integrated into trial intervention design and implementation. It has been suggested that this approach to intervention design may make complex health interventions both more effective and more easily reproducible in other settings [43]. This is congruent with evidence from a systematic review that suggests that interventions tailored to prospectively identified barriers have a greater likelihood of improving professional practice than interventions with no such tailoring [44]. However the review also notes that further work is needed on methods to identify barriers and tailor interventions to address them. The participatory approach used here is also in line with calls to involve the district health systems in efforts to deliver comprehensive HIV care [8,17,45] One of the weaknesses of the development of this intervention is that, while staff at the ART sites were involved in initial discussions, staff at the primary care clinics referring patients to these sites were not. However, as part of the implementation, managers of these primary care clinics were included as members of local management teams and were then able to give their input, assess capacity issues, and make workable plans for the integration of HIV care into their clinic services.
A second change technique used to facilitate uptake of the intervention was educational outreach. This approach was the basis for the training of professional nurses in the intervention clinics. The PALSA PLUS training model, on which the STRETCH intervention was based, draws on adult education principles and the outreach education approach, and has been shown to be effective in changing nurse clinical practice in study setting and more widely [26,27,46]. The trainers chosen to implement this training were local staff membersanother facet of active participation in the implementation. Many of the 16 STRETCH trainers were themselves clinic supervisors and had also been PALSA PLUS trainers. As part of this trial, they trained the professional nurses at the clinics for which they provided supervision.
The STRETCH coordinator also functioned as an 'agent of change' in this intervention, playing a role in facilitating the active participation of staff in, firstly, the process of developing and reformulating the intervention so that it was implementable and responsive to local conditions in the clinics and, secondly, in establishing local teams to implement the intervention actively. The coordinator was appointed by the research team but based in the provincial health department. This allowed her to facilitate communication between the research team and provincial staff and act as a 'problem solver.' The coordinator was also able to provide ongoing support to nurses, doctors, and trainers because of her previous clinical experience. All of these roles have been acknowledged as important functions of external facilitation in the implementation of complex health interventions [47]. Models of implementation also acknowledge the overlap between outreach educators, which formed one component of this intervention, and facilitation, which formed another component. These models suggest that facilitators take on a wider range of roles than outreach educators, including the use of a greater range of enabling approaches to help support practice change and mediate between stakeholders [48].

Conclusion
This paper describes the development and content of the STRETCH intervention intended to improve access to ART. This complex intervention incorporates three processes: participatory action research, educational outreach, and external facilitation to change the practice of nurses in primary care settings in South Africa. The effects of the intervention are now being evaluated in a pragmatic randomised controlled trial. To evaluate the degree to which the intervention was implemented as intended [43,49], a qualitative process evaluation of the trial was conducted. In addition, the integration of HIV care into primary care services was monitored using a semi-quantitative questionnaire. The findings of these parallel studies will contribute to understanding the effects of the intervention described in this paper.

Additional material
Additional file 1: ART algorithms. Algorithms for initiation and management of patients on antiretroviral therapy included in the STRETCH edition of the PALSA PLUS guideline that was used in intervention clinics during the STRETCH trial.
Additional file 2: STRETCH Toolkit. STRETCH Implementation toolkit developed by the research team to assist clinic staff in implementing the STRETCH intervention.