Access to HIV services can be increased by task shifting — but only with support and encouragement

Several studies presented at the 5^th South African AIDS Conference provided reassuring evidence that many health services, including the initiation and management of antiretoviral therapy (ART), can be safely provided by well-trained nurses, while other tasks such as HIV counselling and testing (HCT), patient tracking and adherence support can be shifted to adequately prepared lay staff — all without a loss in quality of care.

However, these and other presentations also reported that task shifting is not the quick and easy solution to sub-Saharan Africa’s human resources for health crisis that many had hoped. Poorly executed and supported task shifting could put stress on both non-physician healthcare workers and the health system, while lack of confidence, poor integration into the local health services and other obstacles can keep newly tasked health staff from realising their full potential.

This doesn’t mean that task shifting won’t work — but effective task shifting could require careful planning, training with appropriate guidelines for referral of difficult cases and, above all, support. However, with close monitoring and evaluation — possibly linked with a quality improvement process to identify and address barriers that keep staff from practising their new skills — as well as ongoing supervision, coaching or mentoring, task shifting could result in a truly significant expansion of the delivery and coverage of high-quality HIV services.

STRETCH-ing to support the nurse-initiation and management of antiretroviral therapy (NIMART)

People living with HIV at sites where nurses initiated and managed ART do just as well as those who start treatment at clinics where only doctors can prescribe ART, according to presentations of the STRETCH (Streamlining Tasks and Roles to Expand Treatment and Care for HIV) study, which evaluated the effect of task shifting ART prescribing to nurse-led clinics in the Free State.

In fact, “viral load responses were equivalent, and there were some indications of more benefit at nurse-led clinics, including better weight gain, and higher rates of TB case detection,” according to Lara Fairall of the Knowledge Translation Unit, University of Cape Town Lung Institute, who gave an oral presentation of the STRETCH study results.¹ However, she also reported that uptake of ART was not significantly better at the nurse-led clinics, contrary to expectations, though it should be pointed out that access to ART improved across the board in the Free State by the end of the study period.

Background on shifting ART management to non-physicians

There is clear evidence of what happens whenever the number of HIV-positive people who need ART outstrips the health system’s will or capacity to deliver it.  Fairall began her talk by describing some of the previous experiences in the Free State’s ART programme in 2004 to 2005.² Over a period of 18 months 14,627 people enrolled in the programme, 48% of whom were eligible for ART. Unfortunately, only 25% received it. Follow-up data were available for 4570 of the programme participants for more than one year. More than half of these individuals (53%) died, 87% before ART could be initiated. Among those with CD4 counts below 200, who did not receive ART, the hazard ratio (HR) for survival was only 0.14 (95% confidence interval [CI] 0.11-0.18).  

One of the factors limiting access to ART was that only doctors could initiate patients on ART. However, in 2008, task shifting was recommended to expand access to ART by both WHO (see WHO document Task Shifting: Global Recommendations and Guidelines), and the South African National AIDS Council (SANAC). (See Technical Task Team (TTT) on Treatment, Care and Support document Building the Capacity of the Primary Health Care System for HIV/AIDS care and treatment in South Africa: Task Shifting Recommendations Document. September 2008).

Fairall reviewed the published literature describing the increasing evidence that task shifting may help expand access to ART. For instance, task shifting ART care to nurses has been increasingly used and described in Botswana³, Lesotho⁴, Rwanda, South Africa (Lusikisiki)⁵, Zambia⁶, and Uganda⁷.

But there have only been a few comparative studies. One from Mozambique, was presented byDr Kenneth Gimbel-Sherr of the University of Washington and Health Alliance International at AIDS 2008 in Mexico City and previously described in HATIP. This non-randomised study reported similar outcomes whether ART was initiated by non-physician clinicians at international NGO-supported clinics or at clinics where doctors initiated patients on ART.8,9 Dr Gimbel-Sherr concluded that the care provided by the non-physician clinicians seemed “equivalent to or slightly better than that provided by MDs,” but he added that “results from large centralised sites with more supervision may not apply to smaller remote sites with less supervision.”

A similar study reported that clinic attendance and patient experience were better with nurse-led antiretroviral treatment based in primary healthcare facilities than with hospital care — with similar health-related outcomes at 15 nurse-led clinics compared to 14 hospitals in Swaziland.¹⁰

Only a few of the reports on task shifting of ART management have come from randomised controlled trials. One was the study from Jinja, Uganda that reported similar rates of virological failure whether follow-up was provided by community-based health workers (home-based carers) or clinics.¹¹ The second reported similar outcomes in South African patients whether follow-up was provided at two nurse-led clinics or clinics where doctors managed ART (see description in What will it take to put millions more on ART?).¹²

Neither of these studies looked at non-physician initiation of ART (NIMART). The STRETCH study is really the first rigorous evaluation of NIMART — and one that evaluated its introduction within the constraints of routine public sector health care.

Learning to STRETCH

Fairall and her colleagues have long been working to increase the ability of nurses in the Free State to manage serious conditions, such as tuberculosis, through the Practical Approach to Lung Health and HIV/AIDS in South Africa (PALSA PLUS) model. Indeed, the nurses at the primary healthcare level were already managing people living with HIV including many who had already been initiated on ART by doctors — however, whenever someone qualified for ART, or needed to have prescriptions renewed, they had to be referred by the nurses to doctors at ART treatment sites. The case for introducing NIMART seemed obvious: if nurses could re-prescribe and prescribe ART, it ought to reduce the number of clinic visits the patient has to make, reduce the likelihood of loss to follow-up during the referral process and thus increase access to ART.

So the PALSA PLUS model was adapted to introduce nurse initiation of ART — with guidelines, materials (the STRETCH toolkit) and training tailored specifically to the nurses, with clear clinical criteria for referral of complex cases to doctors. A ‘change facilitator’ — the STRETCH provincial co-ordinator — would oversee the training and mentoring programme, which involved trainers, experienced fellow nurses, doctors and district ART co-ordinators — as well as a participatory action approach with local facility management teams for the reorganisation of care to enable nurses to prescribe ART at the primary care facility. STRETCH was implemented in three phases, first with training on the guidelines, then scaling up nurse re-prescription, and finally, nurse initiation of ART.

The STRETCH trial and evaluations

STRETCH was evaluated through a pragmatic randomised controlled trial of 31 clinics, sixteen where STRETCH would be introduced, fifteen controls where it would not be. Several evaluations were performed using the data from these sites. One would look at the effects of NIMART on ART access and on quality of ART care, and on “waiting list mortality”. Two poster presentations also described an evaluation to see whether the integration of elements of ART care into primary care had an effect on mortality, and a qualitative process evaluation of those participating in STRETCH over the course of the study.^13,14

The effects of NIMART were evaluated in two cohorts.

• Cohort 1 was a superiority study assessing whether task shifting would reduce the time to death, in other words, could it achieve superior reductions in ‘waiting list’ mortality by improving access to ARTamong patients with CD4 cell counts below 350 (9252 patients at 31 clinics).

• Cohort 2 was an equivalence study to see whether a nurse-led service could provide equivalent ART care (as measured by viral load suppression) in the medium to long term for those patients already on ART ≥ 6 months (6321 patients at the 31 clinics).

Fairall reiterated that this was a pragmatic trial, run in the real world, and within all the usual constraints of the public health system. First, nurse re-prescription of ART took about three months to be introduced into the sixteen clinics, but one year into the study, nurse initiation of ART was only taking place at fourteen out of sixteen of the clinics randomised to the intervention. The training effort was frequently confounded by high staff turnover.

There were major problems with drug distribution to the sites, including the much-publicised difficulties with maintaining ART supplies in the Free State during late 2008 and 2009 — including when a moratorium on new prescriptions was temporarily put in place.

In addition, Fairall pointed out that the fidelity of the control arm of the study was compromised as doctor-supported sites also dramatically increased ART prescribing over the course of the study.

As a result, “contrary to our expectations, STRETCH was not superior in improving access to ART in the study context,” said Fairall. Nor did it appear to cause a profound reduction in ‘waiting list’ mortality.

In cohort 1 (the number of those with CD4 cell counts below or equal to 350), there was no significantly improved survival among those patients at the NIMART clinics followed out to 18 months (with a HR of death of 0.92 [95% CI 0.76-1.15; p 0.532]).

The effect was primarily powered among the patients with CD4 cell counts below 200 (CD4 count ≤ 200 HR 1.00 [95% CI 0.52-1.00; p 0.020]).

However, survival did look somewhat better for the NIMART cohorts among those with baseline CD4 cell counts between 201 and 350, with HR 0.73 (95% CI 0.54-1.00; p = 0.052) interaction term p = 0.050.

Even though the study failed to demonstrate the superiority of nurse-initiated management of ART, it should be pointed out that the intervention hadn’t really fully come online — only about 26% of patients who started ART in the STRETCH group were initiated by a nurse.

It proved relatively easy to demonstrate that nurses provided equivalent management of stable patients on ART for more than six months, however.

Of note, part of STRETCH involved the integration of new ART-related activities into clinic settings, so another evaluation looked at correlations between the integration of HIV care and mortality rates (in Cohort 1) using a semi-quantitative questionnaire including 4 assessments of integration at each clinic during the trial, with questions on:

1. pre-ART care (VCT, CD4 counts, routine HIV care) and
2. ART care (baseline bloods, drug readiness training and monthly ART supply) at two levels of primary care services:
3. integration of above elements of HIV care into primary care services within the ART site (internal integration) and
4. provision of above elements of HIV care by primary care clinics, not designated as ART sites, and referring patients to the trial ART site (mainstreaming HIV care).¹⁵

Cohort 1: ART Provision to people with CD4 ≤ 350 not yet on ART

Cohort 2: On ART ≥ 6 months

Integration scores improved significantly by the third assessment, which the preliminary analysis suggested was associated with a significant overall improvement in survival at all the clinics. This was probably driven by the integration of pre-ART care, which correlated with improved survival again at all of the clinics (intervention and non-intervention), while the integration of ART care and mainstreaming of HIV care was significantly correlated with improved survival at the intervention clinics.

This suggests that as integration (and implementation) of these ART-related activities improves over time, survival also improves — and also that the integration clinics are not the only clinics which are improving their performance.

“It should be remembered that this was a pragmatic trial and that provision of doctor support increased disproportionately during the trial with the result that we compared nurse-led doctor-poor clinics to clinics with substantially more doctor support,” said Fairall.

She noted one factor that proved to be a particular barrier to better performance: “We did not address the complex logistics of drug distribution which our qualitative evaluation highlighted as a key barrier to scaling up services.”

Qualitative evaluations

Fairall said that the “nurses took on ART prescribing without complaint, and didn't even ask for more money.”

The qualitative evaluation, which involved in-depth interviews (n=26) and focus group discussions (n=16) with patients, nurses, doctors and health service managers across the Free State, quoted several of the nurses participating in the programme.

“I think we are very much busier now,” said one.

“And I’m working very much harder, yes,” said another.

“Now, with one patient, you have to exclude a lot of things…But we are enjoying it, because we understand whatever we’ve been taught from the STRETCH. So there is more information, so we are able to explain to our patients better than before,” said another nurse.

But while the evaluation concluded NIMART was generally acceptable to the nurses, patients (who were after all happy for the reduced transport costs and time costs due to decentralised care), doctors and managers, it was not all smooth sailing:

• Nurse confidence grew slowly — and some nurses expressed “hesitation regarding their management of the sickest patients and the lack of clinical support at times.”
• Although doctors were in favour of NIMART, and senior management and political support was strong, the clinical support offered by doctors and local management support varied widely by clinic.
• STRETCH took place against “the background of an already struggling health system” with pressures such as clinic buildings that were too small to accommodate ever-increasing patient numbers. STRETCH increased some of these pressures.
• While decreasing the burden on doctors, STRETCH increased the workload, not only of nurses, but of pharmacists, and, notably, other staff such as lay counsellors or data capturers, who were sometimes called upon to perform basic nursing duties in an improvised manner when the nurses were not available.
• “The pace of rollout varied greatly across clinics, with some clinics unable to easily implement the full task shifting envisaged within STRETCH,” wrote the poster’s authors.

In other words, while NIMART appears feasible, “it results in significant increases in training and clinical support needs, workload and capacity constraints, as well as shifts in the working and referral relationships between health staff,” the authors concluded.

Quite plainly, NIMART can only be scaled up incrementally, requires a great deal of training and support, and “a significant reorganisation of health services to accommodate these shifts in practice.”

Perhaps this is why, in her conclusion, Fairall emphasised “We found no evidence of harm whatsoever, and the nurse-led service delivered equivalent viral suppression rates and a substantial improvement in survival among the patients nurses felt qualified and confident to assist.” She added, “it is important to note that this improvement is not attributable to the provision of ART alone, but rather to better overall care, for which there was robust evidence of multiple improvements — both in processes of care [TB case detection and programme retention] and in actual patient outcomes,” such as CD4 cell count and weight.

Perhaps the take home message is that NIMART in the Free State is a work in progress that has yet to reach its full capacity.

Optimising NIMART services

On the basis of the recommendations and data on task shifting, including the preliminary data from STRETCH, the South African government has moved forward with NIMART in order to meet the demands to scale up ART across the country. But several other presentations also suggested more work is needed to develop confidence and capacity among nursing staff trained to initiate ART.

“After announcing ART expansion, the National Department of Health set a target to train 6000 nurses and initiate 500,000 new patients onto ART by March 2011,” said Elizabeth Mokoka, a nurse with a PhD working for the International Training and Education Centre for Health (I-TECH). (I-TECH is a PEPFAR-supported international NGO working in health capacity development — primarily the development of a skilled health workforce and well-organised national health delivery systems.)¹⁶

The initial results of this training programme were not as successful as hoped.

According to Mokoka, this slow uptake of NIMART was due to delays “in setting up effective mentoring systems for trained nurses; there was a backlog of facilities to be assessed and delays resolving facility infrastructure issues (space, etc.),” she said. Also, there was initially poor co-ordination between the nurses trained in NIMART and facilities being made ready as ART service points.

South Africa doesn’t just need NIMART, it needs nurses who are efficient, competent and confident to prescribe ART safely in practice, who are capable of implementing national policies and guidelines.

Scaling up clinical mentoring

To develop this capacity, the National Department of Health needs to accelerate clinical mentoring of NIMART-trained nurses, according to Mokoka, and develop a standard national curriculum on clinical mentoring. Some of this is already underway, including the development of a clinical mentoring manual, and the beginnings of a standardised curriculum with pilot training.

The training process begins with didactic training, followed by clinical practice with ongoing continuous assessment and support from a clinical mentor. This should lead to competence, and then proficiency as the nurse learns to independently make clinical decisions, developing clinical expertise.

“Clinical mentoring bridges the gap between theory and practice, and supports nurses after NIMART training, supporting the decentralisation of healthcare delivery with high quality of care,” said Mokoka. “It should strengthen problem-solving and decision-making skills and build the capacity of providers to manage or refer unfamiliar or complicated cases… The use of algorithms and guidelines should not replace critical thinking and application of problem-solving skills.”

Clinical mentors should clearly have clinical experience and current practice in the field but they should also be trained as mentors, according to Mokoka, and need key skills, such as the ability to teach, facilitate case discussions and assess clinical skills. In fact, the training process for mentors is not too dissimilar to the training process for nurse prescribers, although it focuses on developing mentoring skills (e.g., how to build relationships, effectively communicate, assess and identify systems issues and provide feedback to the nurse and task teams).

In the past eight months, I-TECH has trained 204 mentors: 107 of the public sector clinical mentors trained in Mpumalanga and Limpopo and 97 mentors from five partner organisations. These will support 235 public facilities where trained nurse providers will practise. Mentoring of the mentors is ongoing, with an average 3 to 4 on-site follow-up visits conducted with each clinical mentor, and support provided by phone, SMS or email.

Although a partial solution to helping NIMART reach its potential, Mokoka pointed out that developing mentors also takes time and funding — and requires a dedicated budget. It can be a challenge to select appropriate trainees (who are willing and committed), and they will need to receive some form of recognition (including a formal position within the DHS). Mentoring also directly consumes resources, such as transportation expenses, and time from the mentor’s usual clinical duties must be allocated for mentoring. Collaboration is essential, so that district trainers and clinical mentors co-ordinate activities — so all the stakeholders need to be oriented on their role in the process.

Expansion of NIMART at the Nkwenkwezi Primary Health Clinic

The scale up of NIMART at one primary health clinic was fraught with challenges and opportunities, according to Dr Sundesh Maharaj, who presented the experience at the Nkwenkwezi Primary Health Clinic in the Eastern Cape, which was assisted by Africare, another US-based NGO supporting health system capacity and other development needs in Africa.¹⁷

Africare’s Injongo Yethu project uses roving clinical support teams to support comprehensive HIV/AIDS programmes in the Eastern Cape, including 24 sites in Makana District, 24 sites in Nkonkobe District and 31 sites in Lukhanji District. It employs a clinical systems mentorship model approach targeted at the provider, team and site level.

The Nkwenkwezi Clinic is a primary health clinic with a catchment area of 4600 in Nkwenkwezi township near Port Alfred. The ANC prevalence in the Eastern Cape is around 23%. Onsite HIV counselling and testing conducted at the Nkwenkwezi Clinic, from October 2010 to March 2011 found a HIV prevalence of 9.3%amongst all tested clients, and a 20% prevalence amongst ANC attendees.

Clinic staff includes a clinic manager, five full-time professional nurses (three of whom have been trained on NIMART, one sessional doctor who visits the clinic once a month (3 hour sessions per visit), one pharmacy assistant, one auxiliary pharmacist, one data capturer, and one cleaner. Clearly, there was limited doctor support for NIMART at the clinic, limited infrastructure and space constraints.

Nevertheless, the clinic was approved as a NIMART site as of 1 April 2010, with two feeder sites: PAL2 and Station Hill Clinics; while the Port Alfred Hospital serves as the referral centre and facilitates the supply of ARVs. Lab services are provided by the NHLS Service via a courier who comes twice daily to collect specimens and deliver results. Support for people living with HIV includes a support group that meets onsite weekly, a dietitian available at the referral hospital who also visits the clinic weekly, and a social worker, available via hospital referral.

Consultations and planning meetings were held with the site management staff, including the district HIV Programme Manager, Site Clinic Manager/Supervisor, the Site Multidisciplinary Team and key staff from the referral hospital to discuss how to support NIMART.

The plan was to hold fortnightly onsite multidisciplinary team meetings on the initiation of new clients on ART. A client initiation checklist was provided to ensure comprehensive preparation for ART initiation — including clinical and social parameters (such as adherence support). There would be a doctor review of patients conducted at months 6 and 12, with complicated cases reviewed as they arose, including participation of staff from the two feeder clinics.

There would also be monthly HIV stakeholder meetings including the PHC’s and key referral hospital staff and support services (the social worker, dietitian, and pharmacy staff) to discuss referrals, drug supply, support services, revisions to guidelines, and case reviews.

As a result, there was an increase in provider-initiated counselling and testing from 50% of patients in October to 90% in March, an increase in cases discussed, roughly 77% of whom have since been initiated on ART. So far, a total of 96 patients have been started on ART including twelve who were pregnant, six children and nine TB-coinfected patients. Four individuals started on ART have since passed away.

Dr Maharaj highlighted the CD4 cell counts at which patients are now being initiated on ART. At the start of the programme, there was a sudden drop in the CD4 cell count at initiation, as the most ill patients who were waiting were initiated on treatment, but since September last year the CD4 cell count at initiation has been a median of 134.6, which is substantially higher than the norm in South Africa (~100). 

“The Antiretroviral Treatment in Lower-Income Countries (ART-LINC) collaborative has shown that the most important predictor of a patient’s CD4 response on ART is the baseline CD4 count at the time treatment is initiated,” he said.

Even so, the infrastructure and shortage of human resources continue to challenge the clinic, which now has added stress due to the high influx of patients from the hospitals to the ‘new’ more convenient NIMART sites, according to Dr Maharaj.

“All the nurses need to be trained in NIMART with certification,” he said. In addition, there continues to be limited doctor support, and the number of patients who must be reviewed every six months is simply getting unwieldy for just one doctor to manage. Another issue is there are poor community links to support and follow-up on patients — who are very mobile between clinics.

More mentorship and supervision will be needed to make NIMART sustainable at this and similar clinics, Dr Maharaj believes, He said that clinic supervisors will need capacity development to act as mentors, with more nurse mentor support from the Department of Health. Finally, they will need to arrange more outreach support visits with hospital doctors.

Quality improvement to empower nurses at PHC facilities to deliver ART

A final presentation, touched on briefly in the last HATIP, described how the quality improvement process could be used to improve the delivery of nurse-initiated ART to pregnant women at clinics in Ugu District in KwaZulu Natal.¹⁸

The district has an antenatal HIV prevalence of 40.2%, one of the highest rates in the country. In 2009, 51% of the maternal deaths in the district were HIV-related. Sixty-one per cent of pregnant women who were eligible for ARV therapy in the Ugu district were referred for treatment initiation, and only 44% of those referred for treatment were successfully initiated.

ART initiation was dependent on doctors. Nurses have been trained but lacked confidence to prescribe, according to Mpume Shibe, PMTCT Co-ordinator for the district who presented the report. 

So the district multidisciplinary team set a goal to increase the proportion of primary health clinics in districts where nurses initiate ART from 32% to 75% over a six-month period (July 2010 to January 2011), building on an existing quality initiative The Masibavikele Campaign.

A data-driven ‘dashboard’ was created to track monthly progress and test changes — small non-threatening tests of change that wereintroduced to build confidence in the nurses who had been trained on NIMART, according to Shibe. Again, teams of doctors helped capacitate, motivate and mentor the PHC nurses to start initiating ART.

There were three other key interventions worth noting — the pre-packaging of drugs, the development and scale-up of a ‘change package’ and the engagement of community health workers.

“The hospital pharmacy, from the Mother Hospital, began pre-packing of the antiretrovirals that the women needed to be initiated with. They pre-packed the drug for the clinic so that it became easier for them to initiate it,” said Shibe.

“We were also able to work with the team to develop and scale up a ‘change package’ — a collection of ideas that are developed by the frontline staff in order to facilitate the implementation of the intervention of the programme,” she said.

Finally, community health workers were assigned to each pregnant woman, and made responsible for arranging collection/follow-up of CD4 cell counts, and tracking of women lost to follow-up.

Results

“By November 2010, we already reached our target of 75% of the clinics where nurses now initiated pregnant women on ART,” said Shibe. The numbers of pregnant women initiated on ART also jumped sharply (though Shibe did not show what percentage of pregnant women with CD4 cell counts below 350 were starting ART). By December 2010, all of the clinics in Ugu district were initiating ART for pregnant women. Over this same period, there were also sharp drops in the perinatal transmission rate at six weeks (although this was due to multiple factors and interventions). There also appears to have been a recent drop in maternal mortality, although these rates were not being monitored prior to 2009.

Shibe believes that the quality improvement process was critical to build nurse confidence in NIMART.

“A quality improvement approach empowers nurses because it recognises nurses as local leaders. It promotes the use of data-driven decision making; it encourages nurses to creatively design local solutions to their challenges and nurtures a sense of teamwork among facility staff,” she said.

There’s also a better chance that the approach will lead to sustained improvement, because the “team work creates a culture of improvement… and ensures that improvement discussions are ‘hard-wired’ into the system,” she concluded.