The STRETCH trial showed that the expansion of primary care nurses’ roles to include ART initiation and re-prescription can be done safely, and can improve health outcomes and quality of care for the duration of care covered by the trial. Nurse-initiation and re-prescription did not, however, reduce time to ART or mortality [16]. The results reported below present the key findings emerging from the qualitative evaluation of this trial. The first three sections report on the general acceptability and fidelity of the implementation process. The last four sections highlight four key factors—pharmacy, human resources, clinical support, and local management input—that affected the implementation of STRETCH.
As noted above, some of the findings reported below were common to both intervention and control sites and others were specific to the STRETCH sites. There were no major differences observed between the STRETCH and control sites along several important dimensions, including management effectiveness, health systems and infrastructure constraints, forms of service organisation, and presence of decentralisation/integration processes. Control sites faced all of the same health systems pressures that STRETCH sites reported, they showed the same kinds of variation across clinics, and they also reported some decentralisation and integration of HIV services independent of the STRETCH intervention.
General acceptability of the STRETCH intervention
There was generally good commitment to STRETCH among management, trainers, clinic staff, and patients. Nurses were comfortable with and enthusiastic about the opportunity to be involved more directly in providing life-saving treatment:
"‘We can ‘STRETCH’ ourselves very far. This is our sisters, our brothers, our mothers we are nursing. Otherwise we would have gone to Australia or UK to work.’ [STRETCH nurse and trainer]"
STRETCH was also seen as acceptable, feasible, and, indeed, urgently needed by staff at the control sites.
Management and political support at the provincial level were strong, though the intensity of involvement of management at clinic level was variable, reflecting a broader weakness in health management in primary care [11, 23]. The attitude among physicians was reported to be more mixed: participants felt that the majority supported decentralisation and nurse initiation of ART, but a significant minority were perceived to be uncertain about the ability of nurses to manage and appropriately refer more complex cases.
Patients were very supportive of STRETCH and appreciated both the improved access to care and the reduction in travel costs and time once on treatment now that they were able to receive HIV care and ART nearer to their homes. STRETCH nurses argued that the decreases in patient travel facilitated by decentralised care were a major factor in patients’ overall acceptance of the intervention. Most patients were satisfied to have their care managed by nurses, but there was still a tension for some between wanting ART to remain a separate service and wanting the benefits of a mainstreamed ART programme:
"‘We don’t have a problem waiting with everyone but we want our files separated and our nurse should just call our names and we go to our specific room.’ [patient]"
"‘Again we want to have our own nurse. Sometimes we experience personal problems that we would like to discuss with our nurse but it is not easy if today you are seen by this one and next time is that one.’ [patient]"
Some patients requested their own section of the clinic where they could avoid the long waiting lines in the general clinic and could be seen by a nurse familiar to them. Others wanted to continue to receive their care from a physician because of both the physician’s higher clinical status as well as the fact that only physicians can medically certify social grant applications, a key source of income for people living with HIV/AIDS in South Africa.
Training and support: strong foundations but inconsistent follow-up
Nurses responded positively to the initial training on STRETCH’s guideline-based approach. Their prior training on and experience with PALSA PLUS was important in facilitating their understanding of the STRETCH approach and its implementation. There was significant variation, however, in the quality and quantity of the ongoing support provided to intervention sites. Some staff felt that STRETCH trainers lacked direct clinical experience and perceived that they did not have sufficient time in their work schedule to travel regularly to sites to provide support. These factors prevented some trainers from fulfilling the role that staff expected of them. Support and mentorship from physicians also varied greatly. Where strong support was available, nurses were more likely to report that they had developed clinical confidence.
Active local management support for implementation also varied, and many participants reported that the trial coordinator often had to step in to address gaps in logistical and management support to sites. This ongoing support should ideally have been provided by the local STRETCH ‘support teams’ composed of local area and site managers, ART and pharmacy coordinators, and physicians and nurses. Some of these teams, however, struggled with leadership and logistical challenges (see below) and did not function effectively.
Variations in the pace of implementing NIMART
The pace with which sites progressed through the three phases of STRETCH—training, re-prescription and decentralisation, and nurse-initiation—varied considerably (Figure1). Though the STRETCH Toolkit recommended a four- to six-month timeline for progressing through these phases, clinics were also encouraged to adapt the pace of roll-out to suit local resources and conditions. However, a number of sites struggled to meet the basic requirements for progressing through the phases. As a result, the start dates for nurse-initiation were spread over 10 months with only six of the 16 sites starting in January 2008 as planned. Two sites never progressed from the second to the third phase (though they remained in the trial).
A number of factors were consistently reported as having influenced the pace with which intervention sites progressed. Some sites had difficulties in implementing the decentralisation aspect of STRETCH despite the fact that ad hoc decentralisation of routine HIV care services to PHC clinics had already occurred in a number of intervention and control sites prior to the STRETCH trial. Barriers to decentralising care included high staff turnover, resource and logistical constraints (e.g. poor drug distribution systems), concern among nurses about the quality of HIV care at some PHC sites, and lack of local area management support for and coordination of decentralisation.
The implementation of nurse-initiation was also influenced by a number of constraints. At some sites where nurses were re-prescribing and some HIV care had been successfully decentralised, a lack of clinical confidence and poor physician support appeared to be barriers to progressing to, and sustaining, full NIMART. In two sites, patient deaths and complications shortly after starting ART undermined nurse confidence and led to the temporary suspension of NIMART:
".’..when you see things happening [clinical complications] then you begin to feel somehow guilty to say, maybe, if you were not given this thing and they have been doing it the old way, maybe this couldn’t have happened. But on the other hand, when we see our patients doing well, then you feel proud.. at times, the clients come back to say, ‘You have brought my life back,’ and thank you for that. So those are some of the things that I have seen…There is the good and there is the other side.’ [STRETCH site manager and nurse]"
In other sites, a number of drug delivery issues and infrastructure deficits affected both initiation and maintenance of patients on ART. These are discussed below.
Logistic and infrastructural constraints
In this section, we consider in more detail some of the above-mentioned logistical and infrastructural constraints. The STRETCH intervention took place against the background of a health system that was contending with a variety of difficulties, including resource constraints, pharmacy re-organisation, information system and transport problems, inadequate size of clinic buildings for the increasing numbers of patients, and, in some facilities, no functioning toilets or telephones. The new workload arising from providing NIMART further strained many of these key areas of logistics and infrastructure.
Nurses in all facilities said repeatedly that paperwork demands in the health system as a whole were onerous, and had been increased by NIMART. It is likely that the increase in administration work for NIMART was related to the increased numbers of ART patients rather than unusually intense recordkeeping for ART patients. This burden was compounded by weak and fragmented information systems that were insufficiently staffed and resourced.
STRETCH did reduce the need for referrals and patient transport between sites, an issue of urgent concern among most nurses and patients. One nurse at a PHC referral site described their difficulties referring patients to other distant sites for ART:
"‘…if the patient is very ill, then the transport is a problem. And the transport leaves early. Four o’clock in the morning…I had a case last week when one of the patients died there at [Clinic X] because she was very [too] ill to go. And then the complaint in the morning was that it was too cold to wait for the transport. So she died there. It was a new patient.’ [PHC referral site nurse]"
However, nurses reported that the decentralisation of monthly medication to local clinics, a key way to ensure fewer referrals and patient journeys, had been particularly stressful. This was a consequence of the national requirement that ART be dispensed under direct supervision of a pharmacist. This requirement was implemented by prepacking specific prescriptions centrally for named patients. These named prescriptions and packets of drugs then had to be transported back and forth between facilities. Poor communication and transport between pharmacy and clinic services made drug supplies unreliable. STRETCH also introduced significant drug storage and management challenges at smaller sites that already had insufficient space allocated to their pharmacy.
In other sites, unreliable delivery of ART drugs from hospitals and the central dispensing unit, as well as infrastructure deficits such as non-functioning telephone lines, were reported to have had significant effects on patients already on ART and made it very difficult for nurses to initiate new patients onto treatment. These were the obstacles cited by the two STRETCH intervention sites that had not progressed to phase three two years into the trial.
The Free State placed a temporary moratorium on ART initiation between November 2008 and February 2009 because of inadequate funding for ART procurement. This also impacted adversely on nurses’ morale. However, depletion of clinics’ drug stocks and other service disruptions are not unusual in this setting and STRETCH sites reported having dealt with the backlog of new patients needing treatment soon after the moratorium ended. Although the trial coordinator reported that some facilities struggled to return to nurse-initiation of ART, the moratorium was not reported as a factor leading to the slow progression of some sites through the three trial phases.
In general, though increased paperwork demands and weak IT systems were the subject of frequent criticism, these were not reported to have significantly affected implementation and were counter-balanced by improvements in other aspects of patient care such as fewer referrals and less transport requirements. Rather, a more important source of frustration and delay was pharmacy-related logistics and infrastructure challenges. These did not seem to significantly affect ongoing implementation of services in the medium-term but did contribute to initial delays and frustrations and the slow progression to full nurse-initiation in some sites.
Human resources: increased workloads but short of spare capacity
The overall shortage of all categories of PHC workers was seen as a critical issue in every site. This included shortages of nurses and physicians as well as of pharmacists, managers, social workers, data clerks, lay counsellors, and administrative support. STRETCH increased nurse workloads through shifts to nurses of physician tasks, and also increased workloads on other team members through improvised shifting of duties as a result of this broader lack of capacity. For example, data capturers performed basic nursing duties (like weighing patients) when nurses were very busy or nurses dispensed when pharmacists were not available.
The initial enthusiasm for NIMART was tempered in some STRETCH sites by the increased workload, in particular from ongoing monthly patient follow-up visits. This workload curbed the drive to place new patients on treatment:
"‘We don’t want to promote it [HIV testing and ART] worldwide and then we can’t handle the load. So the promotion is basically through other people that were successful that’s come in…if we get somebody in the clinic identified as HIV-positive, we give them the information about ARVs. Personally, I think we must have a much bigger promotion for ARVs to make people aware that there is help…but as I say, we are not doing it because we can’t handle the burden if it’s much bigger.’ [local area manager]"
Nurses also reported that ART patients required more complex and comprehensive clinical input thereby increasing the time and effort required per consultation. This extra input was felt to be needed in particular at clinics that already had high patient volumes.
In contrast with increased nurse workloads, there were clear decreases in physicians’ routine workloads, suggesting that the underlying objective of NIMART was attained. For example, one physician was able to start seeing patients at other clinics because his work had been reduced significantly at the intervention site. The focus of physicians’ work also shifted, from managing all new cases and follow-up visits to managing only those cases with clinical complications:
"‘In the beginning … we started in 2006 as an ARV site here, and back then I obviously did virtually everything…But then, after the STRETCH started, it made my work much easier, especially the follow-ups because the sisters are now doing everything, and is just referring the problem cases to me. The workload is much less now … the paperwork, I would say.’ [physician at STRETCH site]"
Despite the growing and increasingly complex workload and the human resource constraints, nurses in the STRETCH sites described a substantial emotional reward from their investment in the NIMART programme and a long-term commitment to their patients (as noted in an earlier study in this setting [24]). This commitment to patients and optimism about treatment, however, was not limited to STRETCH sites. Even in control sites, where decentralisation of HIV services had started to improve treatment access, nurses reported the same kinds of emotional satisfaction:
"‘There’s hope now. In the past we could only give the Bactrim [an antibiotic] and say, ‘there’s no hope and that is it.’ Now at least we can say, ‘okay, if your CD4 is this and this, you can give this and that for it’…So it gives us a little option, and you can see the difference in patients. Some are being pushed in wheelchairs and are malnourished … you know, they are very sick, and then after three months the same patients walk in here and they start working again. You know, looking for work and planning their lives again…That’s making it ….worth it.’ [control site nurse]"
Nurses felt, however, that commitment often had negative effects for their own wellbeing. In particular, they felt that they needed much more support to sustain their clinic work, and that middle and upper management layers were uninterested, or unable to provide this support, or both.
Challenges in supporting the development of clinical confidence among nurses
The STRETCH intervention included several forms of clinical support to ensure quality of care and to develop nurses’ clinical self-confidence. Physicians were supposed to provide support to nurses at clinic level by accepting telephone queries, providing feedback on referred cases, and visiting nurses on-site to discuss cases. Nurses were also expected to support and share skills with each other as HIV care shifted from a specialist to a generalist service.
It was also intended that STRETCH trainers would build clinical knowledge and confidence through ongoing training and support for nurses, both telephonic and in person. The local STRETCH support teams were supposed to complement this direct clinical support with assistance in addressing management and logistics challenges. The STRETCH trial coordinator was available to facilitate and encourage these relationships in this decentralised support system and to provide additional logistical and clinical advice where necessary. Further clinical support was available via a telephone hotline from the provincial ART programme’s ‘Centre of Excellence,’ though some nurses reported being too intimidated to call an expert whom they did not know and found it easier to contact the trial coordinator for advice.
Nurses were generally familiar and satisfied with the STRETCH approach to guidelines, given their prior experience with PALSA PLUS. However, they expressed concerns regarding the complexity of ART. While they agreed that HIV/AIDS should be treated like any other chronic disease, they also maintained it was more complex in terms of time per consultation, medication side-effects, and emotional involvement. There were some nurses who would therefore have preferred NIMART to have remained a vertical programme so that they could develop expertise within this specialty.
For the most part, though, initial resistance to STRETCH soon gave way to an acceptance of the viability and preferability of generalised nurse-initiation of ART. For some, provision of ART also brought with it a feeling of accomplishment and a degree of prestige among the other nurses and the patients for being able to offer this valuable service:
"‘You know, it’s an eye-opener, and it’s giving us an opportunity to be able even to use our brains even further. As I’ve said, you’ll be consulting and doing this. And there are times that you have to think deeper. And it’s even broadening our intelligence…to me, it’s an achievement, and it’s something that really boosted my confidence.’ [STRETCH nurse]"
Of greater concern for most nurses, however, were the volume of patients they had to see and their sense of being at the limit of their capacity to safely manage a rapid roll out. The STRETCH guidelines played an important role in this context, offering a secure platform for developing clinical confidence.
Two important forms of horizontal support within clinics emerged. Firstly, peer support among nurses appeared to be widespread and was effective in sites where no physicians were available. Secondly, nurses generally reported trusting and effective relationships with physicians in the sites where physicians were permanently employed or visited regularly. However, nurses at STRETCH sites that did not have a regular on-site physician presence generally ended up with little senior clinical mentorship.
Key reasons for inconsistent physician support included an insufficient number of physicians, poor coordination between physicians and STRETCH trainers, and a lack of relevant ART experience or interest in ART among physicians (particularly newly qualified physicians). Also some physicians worked only at distant treatment sites and did not visit the local assessment sites. In the early stages of implementation, friction between some physicians and nurses about treatment protocols resulted in insecurity among nurses, but the trial coordinator was able to mediate these conflicts.
Nurses generally felt that this gap in clinical support was not sufficiently addressed by the STRETCH trainers. They argued that STRETCH trainers were often not adequately experienced in ART ‘on the ground’ and were not able to provide the ongoing training and support that nurses needed. These trainers were generally drawn from a pool of middle managers with the expectation that they would supervise and support nurses after the initial training, rather than provide ongoing clinical training. Once nurses had developed some clinical confidence and experience, this supervisory aspect of their relationship with the STRETCH trainers was more effective. Again, the trial coordinator often addressed this gap, serving as an important source of clinical support both telephonically and in person.
Overall, however, it was the trial coordinator who had the most impact on ensuring clinical support and on developing the confidence of nurses at the intervention sites. Even though her efforts to address logistical and management challenges and provide direct clinical support went further than originally planned, her input was most intensive towards the beginning of the trial and tapered off as these issues were addressed. We discuss later the implications of her contribution for the broader sustainability of the STRETCH programme.
The critical role of effective local and district management
Though STRETCH was well regarded by most managers throughout the health system, their input was at times insufficient to address the many day-to-day challenges of logistics, human resources, and clinical management, referral, and support. For example, during the course of the pre-intervention visits, STRETCH support teams were formed within each facility. These teams were composed of nurses, site managers, ART and pharmacy coordinators, and physicians. Management of these teams, however, was inconsistent and their success generally depended on the availability and commitment of the more senior staff members. Similarly, local area managers, whose input was often key to solving small-scale but crucial logistical problems, found it difficult to visit clinics under their jurisdiction, often because they had no budget for vehicles and fuel.
Managers at the facility level were expected to oversee general changes in service organisation and to assist with logistical arrangements, local intervention tailoring, and scheduling, guided by the STRETCH Implementation Toolkit. However, the trial coordinator reported that, on its own, the toolkit was often not enough to promote progress through the trial phases at sites that were not otherwise highly motivated. The coordinator found that she needed to work through many of the logistical and administrative issues directly with sites early in the implementation process.
The direct support by adequately informed and engaged managers at higher levels of the health system, for example at sub-district and district levels, and of district-based ART coordinators also facilitated the roll-out. In some sites, these managers were central to putting in place the elements that would allow for effective decentralisation. For example, decentralising routine HIV care within a district meant moving a number of tasks and responsibilities to new sites and staff. These included initial laboratory workup, drug readiness training, and the monthly supply of ARVs. The involvement of these managers, however, was highly variable, with some sites requiring much more contact with and support from the trial coordinator. Staff turnover within management was also a problem at all levels within the province.
Allowing sites to manage the pace of implementation eased management conflicts and improved problem solving. This bottom-up approach also appeared to increase ownership of the intervention by nurses and managers.