To our knowledge, this is the first assessment of nurse-centered ART within a context of a nationally coordinated program seeking to achieve universal access to AIDS care and treatment. Our evaluation suggests that with adequate preparation and support nurses can effectively prescribe first-line ART and monitor noncomplex patients. Nurses participating in the initiative achieved high compliance with national guidelines and excellent patient outcomes. The experience demonstrates the feasibility and suggests effectiveness of nurse-centered task shifting for decentralized ART services without compromising the quality of care.
Our findings bear out positive reports from other African countries. While Morris et al. 
describe a task-shifting strategy used in clinics in Lusaka, Zambia, Stringer and colleagues 
report robust treatment adherence and patient outcomes from a large cohort at urban primary care sites in the city. The treatment services relied primarily on nonphysician clinicians to provide initial patient management with rotating physicians consulting complex patients. Specialist HIV clinicians outside of Zambia could be consulted as well via internet 
. While confirming the feasibility of involving nonphysician clinicians, the Zambian program relied heavily on clinical officers, who have received substantial medical training but are of limited availability in many African countries. In addition, no formal assessment of the performance of the nonphysician clinicians was reported and no details were provided on the practical organization (training, task-distribution), implementation, and supervision within the program.
In South Africa, an expanded role for nurses in AIDS care, including nurse-initiated ART, was tried in order to enhance service delivery in rural communities 
. An assessment of program data from Lusikisiki showed significantly increased enrollment of patients on treatment through decentralized rural services, whereas the hospital ART enrollment capacity was rapidly exceeded. Furthermore, a comparison of 1-y outcome data on retention, CD4 monitoring and response, and viral load consistently showed outcomes for rural clinic patients to be significantly better than patients treated at hospitals. No assessment of nurses' performance was reported, however, and details of the step-by-step organization of nurse-initiated ART was lacking.
That the task-shifting initiative in Rwanda was carried out within a national framework with the explicit objective to evaluate its safety and potential for policy development represents an important addition to the existing literature. Our report provides a detailed description of the structure and content of the intervention as well as additional evidence that nurse-initiated ART may be a viable strategy for expanding access to AIDS treatment to rural populations. In combination with an independent evaluation commissioned by the Ministry of Health in early 2008 
, the present evaluation forms the basis of national plans to adopt this task-shifting strategy for national-level ART scale-up.
Compared to a recent evaluation of the national ART program in Rwanda during 2004–2005, our study found that patient retention at the facility level was 89% at 12 mo of ART, close to the national estimate of 87% for adult patients attending similar size health centers in Rwanda, where care is generally provided by physicians 
. Whereas documented mortality at 12 mo was slightly higher in our program (8.5% versus 5.7%), LTFU was substantially lower (0.3% versus 3.2%). Comparable changes in weight and CD4 cell count were also observed. Overall, key patient outcomes from this task-shifting model were also comparable to findings from other cohorts in sub-Saharan Africa. For example, in a recent systematic review of ART programs in sub-Saharan Africa, retention at 1 y after ART initiation was estimated at 70% 
. Similar mortality rates, stratified by baseline CD4 count or WHO stage, were reported in large programs in Zambia 
, Malawi 
, and numerous other countries in the region 
The limitations of this study include the fact that data presented are descriptive and no direct comparison of outcomes within this nurse-centered model of care was made with those from traditional physician-centered models, which makes it difficult to ascertain if patients' outcomes were as a result of the nurses' role or due to MDs' intensive supervision. Furthermore, we should exercise caution when interpreting the retention results given that the median follow-up time for patients in this sample was 8.3 mo. Although the sites selected were generally comparable to other PHCs in Rwanda, the fact that neither the sites nor the patients were randomly selected implies that our findings cannot necessarily be generalized to other sites or even to other patients at these particular sites. The criteria used to identify pilot sites favored PHCs that offered relatively favorable conditions, particularly with strong management and adequate staffing. Also, we selected nurses with substantial clinical experience prior to the pilot initiative. Finally, participation in a closely monitored pilot project can itself increase commitment from both nurses and physicians. Future sites will need to meet a minimum set of criteria for safe and effective nurse-initiated ART delivery.
To avoid over-reliance on nursing skills for HIV care and treatment, the tasks and responsibilities of nurses and physicians will also need to be clear. As we task shift some clinical responsibilities to nurses, we must also consider the modified role of physicians, which will require provision of care for complex patients while at the same time providing overall oversight of patient management by the nurses. Not only provider roles but skill levels (of both nurses and physicians) will need to be reviewed, periodically revised, and upgraded to meet evolving clinical challenges. In particular, treatment failure will be an important challenge in the future, as will provision of treatment and follow-up to children. Both will require careful monitoring and an appropriate mix of physician and nurse skills. To emphasize the complementary nurse-physician skill mix necessary to provide high quality patient care, it might be more appropriate to conceptualize this approach as “task sharing” rather than task shifting. Specific guidelines to scale-up the initiative should thus focus on defining roles and responsibilities of the nurse-physician patient care team. Additionally, this nurse-physician task shifting is best framed within a whole context of task transfers for patient care, including to community health workers, family members, and patients themselves.
In addition, regulatory frameworks integrating task shifting within national policy on human resources for health should be established 
. Given the severe shortage of human resources for health, task shifting in isolation is not recommended but rather as combined with wider investment in human resources and health systems strengthening more generally 
. High quality preservice training with regular in-service updates, focusing not only on theory but actual clinical practice, will be pivotal for the long-term success of task shifting 
; training curricula will need to be tailored to different nursing and nonphysician practitioners providing care in different African settings 
. Adapted medical records, tools, and protocols to assist nurses and other health professionals to perform in this expanded AIDS care role will also be necessary 
. Further research on successful and cost-effective implementation models, as well as studies on patient and health care worker perspectives will also be of interest 
The potential benefit of this task-shifting model likely extends beyond HIV patient care, as indicated by Chung and colleagues 
. Including time estimates for physician supervision and consultation in complicated cases, Chung and colleagues estimate that nurse-initiated ART for noncomplex cases saved approximately 45 min of physician time for every 1 h worked by a prescribing nurse. If task shifting were applied nationally to the future roll-out of ART in Rwanda, their simulation model suggested a 76% reduction in the HIV care burden for physicians. Physician time spared through ART task shifting could be used for other patient care, mentoring, and quality improvement activities. On this point, specific guidance and systems may be in order; whereas the task of the nurses was clearly defined and evaluated in this pilot initiative, the new roles that physicians could play was not explored. On the other hand, the impact of this model on the overall health system, and specifically on other activities where the nurses would have been used, was not explored either.
Given the shortage of physicians in many African countries 
, nurses have long played an important role in the management and treatment of key diseases, such as tuberculosis and malaria, typically with well-established task descriptions, clear referral indications, and documented good results 
. This report adds to the evidence base that nurses can play a similarly effective role in HIV care. It further advocates for decentralized HIV care and treatment at the primary health clinic level 
. If carefully conceived and monitored, task shifting for HIV care offers a promising strategy to address physician shortages in many resource-limited countries and will likely have benefits across the health sector.