In fact, low- and middle-income countries not only have the highest rates of HIV worldwide, but also face a severe shortage of health-care workers. Sub-Saharan Africa is home to 11% of the global population and nearly two-thirds of people living with HIV/AIDS, but only 3% of the world's health-care workers 14. Staff shortages in resource-limited settings are a major obstacle to the scale-up of HIV care and treatment, including PMTCT. In 2006, the World Health Organization (WHO) was one of the first international bodies to devote an entire report to the global health-care workforce crisis, thus renewing interest in this topic 14. The shortage of trained health care providers in resource-limited settings has existed for decades, but in recent years the number of workers has been inadequate at almost all levels of the health system. One reason is that the number of people completing health-related training has not kept up with the increasing needs 15. It is therefore critical to use the available health-care personnel in a cost-effective manner.

An estimated 45 countries worldwide, of which an overwhelming majority are in sub-Saharan Africa, will not meet the WHO target of skilled health-care attendants delivering 80% of the live births by 2015 16. The added workload brought on by the HIV epidemic has increased the strain on fragile health systems and already over-stretched health-workers 1718. In Tanzania, the average staff workload at MNCH clinics ranges from 37.8% at facilities that do not provide PMTCT to 50.5% at clinics in which trained health-care workers provide PMTCT services 19. This suggests that health professionals may be less motivated to undertake PMTCT-related activities. A performance-based system for providing PMTCT services may therefore improve MNCH and long-term HIV care and treatment.

According to economic theories, individuals only choose to work if the overall benefits of their labor outweigh the associated opportunity costs. Benefits may include financial and non-financial rewards 2021. Professional motivation is defined as an individual's willingness to work towards the goals of an organization. In order for health-care workers to be effective, they must be motivated and willing to implement the reforms proposed by health policy makers 22. The problem of under-performance and de-motivation of health-workers has been often explained in terms of an asymmetric principal-agent relationship where agents (health-workers) lack incentives to perform and principals (employers and patients) lack sanction and accountability mechanisms 23. HRM and support systems can increase motivation by improving working conditions and involving staff in management decisions. Reviews of health-care worker motivation in resource-limited settings have shown that a wide range of factors on both the individual and group level, as well as in various social and cultural contexts, are likely to influence their expectations and effectiveness when providing care. However, these studies have rarely assessed HIV care or PMTCT specifically 24252627.

The WHO has put forth four criteria for evaluating performance of health-care workers: availability, competence, responsiveness and productivity 14. Several international development organizations have also adapted a "Performance Improvement Model" for use in low- and middle-income settings. This model outlines five key factors that influence performance outcomes: job expectations; performance feedback; environment and tools; motivation and incentives; and knowledge and skills. One group used this model in Armenia to assess the factors affecting performance among MNCH health-care providers 28. They found that among people providing antenatal and postpartum care, performance was associated with practical knowledge and the ability to use everyday tools, recognition for good work, and performance feedback. These findings suggest that a set of incentives could help in introducing and scaling-up effective PMTCT programs (Figure 1).

First, without adequate remuneration, health-care workers will not perform their tasks effectively, nor will they commit to their jobs on the long term, regardless of their responsibilities and initial level of commitment 2930. It may therefore be possible to create explicit financial incentives, for instance by increasing salaries or providing bonuses to those working in underserved communities or rural regions 31.

Although wage concerns are important, they are not the only problem for already over-stretched health-care workers. Other important factors include professional fulfillment and opportunities for career advancement. Non-financial incentives may include compensation payments that improve living conditions, such as free housing, transportation or electricity, professional support and supervision, streamlined management, opportunities for continuing education and training, relocations, or promotions 3233 (Figure 1). However, it is important to note that non-financial incentives weigh as much as the financial ones and both need to be linked together.

One potential approach to improve PMTCT coverage is opt-out, provider-initiated HIV testing and counseling (PITC) 34. HIV screening practices are currently shifting from voluntary, patient-initiated testing to routine, provider-initiated testing that includes group pre-test counseling, a rapid HIV test, individual post-test counseling, and encouragement of community and spousal involvement 3536. PITC has thus become the standard practice in many resource-limited countries: in 2008, of the 53 countries with generalized HIV epidemics that reported information on PITC, half reported that its introduction had encouraged health-care workers to propose testing and counseling to everyone, regardless of presenting symptoms and type of facility 37.

Another method for improving the provision of PMTCT services may be the formal reorganization of tasks among health-care workers. International bodies such as the WHO and the United States President's Emergency Plan For AIDS Relief (PEPFAR) advocate alternative models of care delivery that rely more on nurses and lay health-workers than physicians 38. Reports from rural and primary health-care centers in Rwanda, Lesotho and elsewhere demonstrate that nurses can prescribe antiretroviral treatment (ART) safely and effectively to HIV-infected adults and children as long as they receive adequate training and support 3940414243. A cost-minimization model has also shown that task-shifting during the follow-up of patients on ART can substantially reduce costs as well as the need for physicians. Replacing the standard physician-intensive follow-up with nurse-intensive follow-up, for instance, could save 90 full-time-equivalent physicians, or 4.1% of the national physician workforce in Uganda 44. Innovative staffing strategies that use off-duty workers, trained traditional birth attendants and HIV-infected female community health-care workers to provide MNCH and PMTCT services could help alleviate the shortage of human resources in many MNCH centers 45464748.

Despite the growing use of PITC and alternative models of care delivery, translating PMTCT research findings into practice remains difficult in resource-limited settings 49. A carefully thought-out package of incentives must be proposed to health professionals, to ensure that they not only continue working in the field, but also agree to relocate to rural areas if necessary. Indonesia has successfully attracted physicians to peripheral islands and retained them there, using a financial incentive package. Once they complete their public service requirement, health-workers who agree to work in very remote areas receive higher salaries -isolation pay-as well as a guaranteed career in the civil service, which is highly valued because it allows for complementary private practice and free access to specialist training 50.

Patients can also receive incentives to present to care. In rural Nepal, midwives received financial incentives to outreach pregnant women in neighboring villages 51. Pregnant women were also reimbursed for the cost of transportation to the clinic if they gave birth with the help of skilled health-care attendants. The authors of this study acknowledged that close supervision and strong relations between the target community and the health-care workers probably contributed to the success of this program. Furthermore, the sustainability of such incentives over time is unknown and results likely depend on context.

Public health-care worker salaries are set according to local budgets, which rely on the national budget and international funding agencies, rather than worker productivity. As a result, public sector health-care workers are often paid too little compared to the local cost of living and generally develop individual strategies to increase their incomes 29.

Many studies have shown that financial incentives are rarely the only factor that motivates staff and improves performance. In fact, health-care workers may agree to work for less pay if their job is perceived to be attractive enough 3132. Few studies, however, have evaluated the quality of supervision and continuing education, which could both help improve the quality of health services over time.

Financial incentives are particularly important when salaries are very low. Health-care workers are at risk, however, of considering them as bonuses rather than incentives. Even when salaries are reasonable, financial incentives alone are often not enough. A study conducted in public and private facilities in rural Benin and Kenya evaluated the impact of non-monetary incentives. It showed that health-workers were strongly guided by their professional conscience and ethos 52. Many health-workers were unmotivated and frustrated, because of the lack of means and supplies, as well as inadequate or inappropriately used HRM tools such as supervision, recognition, training, career planning, and leadership, preventing them from pursuing their vocation and reaching their full potential. A study on the association between motivation and performance among health-care workers in Mali showed that after salary, responsibility, training and recognition were the main motivators 53. Health-care workers declared that they gained professional satisfaction from receiving recognition for their work, having responsibilities, and performing the work itself. Factors that led to discontent were related to poor working conditions such as inadequate equipment, no job description, and subjective performance evaluation. Interviewees added that their motivation would likely increase if they had more responsibilities and recognition for their work. Access to and funding for education and promotion is also important for motivating and retaining workers 26335455. These published reports confirm our hypothesis that non-monetary incentives and HRM tools play an important role in motivating health professionals.

Financial incentives have not been shown to improve patient health when health systems and their personnel are not equipped to handle the necessary quantitative and qualitative changes. For instance, the role of traditional birth attendants (TBAs) in MNCH centers and uptake of PMTCT services is unclear, with some studies reporting an increase and others reporting no change in uptake after implementation of TBA-directed incentives 4656. Furthermore, a system of incentives and allowances that differs from those at non-HIV health centers could lead the most qualified personnel to "defect" to HIV care programs, thus reducing the quality of health services in other sectors 57.

In Haiti, a United States Agency for International Development (USAID)-funded bilateral health project used pay-for-performance to increase the quantity and quality of health services. The project was monitored and evaluated, and showed that paying for results could also work in a resource-limited country. This pilot program started in 1999 in three facilities. It gradually shifted from reimbursing the health centers for documented expenditures- pay-for-input - to paying them in part based on their attainment of specific performance targets- pay-for-performance. Over the six-year phasing-in of this payment-for-performance scheme, the health centers showed remarkable improvements in a range of performance indicators, including the percentage of pregnant women who attend prenatal visits at least three times and the percentage of deliveries conducted by trained attendants 58. Performance for health facilities in performance-based payment improved substantially, moving from provision of at least three prenatal care visits to 31% of pregnant women in the first contract period to an average of 60% in 2005. A slight increase in the proportion of assisted trained deliveries from 78.5% to 80% was concomitantly observed.

A pilot program in Rwanda demonstrated that performance-based financing for HIV services such as voluntary counseling and testing (VCT) is also feasible and sustainable in sub-Saharan Africa 59. The program emerged from an agreement between the Rwandan Ministry of Health, USAID, and PEPFAR and aimed to improve access to and quality of HIV care throughout the country 59. In the first nine months, HIV VCT at targeted health centers increased by 155%. It is important to note, however, that independent funding agencies oversaw the district health system at a decentralized level and ensured the separation of purchases, service delivery, and regulation by local health authorities from contract negotiation to fund disbursement. Local community groups used patient surveys to monitor the performance of health facilities as well as consumer satisfaction. These new structures provided an opportunity to redistribute roles within the Rwandan health system and make it more transparent and efficient. The performance-based financing initiative showed an increase in health services use, financial accessibility, and health-care worker motivation. By incorporating the private sector into the health system, it also strengthened public-private partnerships in Rwanda. More to the point, in the province of Cyangugu, a performance-based financing initiative aiming to create checks and balances at the district level, showed good results in terms of motivation of health staff, financial accessibility as well as use of services. In fact, not only out-of-pocket health expenditure decreased by 62% from US$ 9.05 to US$ 3.45 but more outstandingly the proportion of women delivering in a health facility concomitantly increased from 25% to 60%-offering a good opportunity of linkage with PMTCT services 60.

The Rwandan experience may provide lessons for other countries, as it shows that the combination of output-based payment and greater autonomy is a feasible and effective strategy for improving the performance of public health services at a local level 61. Nonetheless, the Rwandan experiment has been described by the targeted health-care workers as having put additional stress on an already overstretched health-care system. Some staff members also complained of the ethical issues associated with choosing between activities seen as necessary (e.g. intensive care) and those required to attain performance targets (e.g. filling out forms) 62. Moreover, this might disrupt the whole system by making health professionals turn away from perceived under-financed infectious diseases to HIV-related activities, for instance.

Few studies have explored performance-based financing for PMTCT. In 2008, the Elizabeth Glaser Pediatric AIDS Foundation (EGPAF) initiated a pilot performance-based financing scheme in collaboration with private and faith-based health facilities in Côte d'Ivoire, the country with the highest HIV prevalence in West Africa. Côte d'Ivoire did not have a national framework for performance-based financing at that time. EGPAF recently reported the first results of this initiative, but its sustainability over time remains to be assessed 63. In brief, the program team designed 22 indicators to evaluate performance-based financing of PMTCT and HIV care and treatment. When they compared data from the year before initiating the program to data from the year after its inception, they found that performance-based financing was associated with improved quantity and quality of HIV prevention, care and treatment, as well as a 100% increase in health-care worker effort. To work, performance-based schemes should be built upon reliable monitoring and evaluation systems that routinely collect process data allowing evaluation of the strategy implemented. The effectiveness of performance-based financing of PMTCT, its sustainability, and its impact on the efficacy of PMTCT have never been practically evaluated on a global scale.