There is an emerging consensus that a lack of financial resources explains only part of the slow progress towards improved health indicators made by most developing countries 21. In India, a little more than 73% of all health spending is out-of-pocket, 6% from third-party insurers and employers, and the remainder from government 22. States typically account for about two thirds of these public expenditures, and the central government the remaining one third 23.

The largely privatized nature of the spending has contributed to huge inequities among the states. In 2005, for instance, overall health spending in Himachal Pradesh, at USD 98 per capita, was almost five times Tamil Nadu's annual health expenditure, at USD 20 per person.

Interestingly, spending levels appear to have only the most general correlation with health indicators. In 2005, Tamil Nadu's infant mortality rate (IMR) was 9% lower than that of Himachal Pradesh; under-four mortality was 31% lower, and life expectancy was 3.4 years longer (Table 1).

How can we explain these differences in health indicators, given the enormous disparity in resources? There is growing evidence that health system components (e.g. financing, human resources and governance) determine in large part the success or failure of health systems 24. Among these, management of human resources has been cited as the most crucial factor for success of developing country health systems 25.

WHO, in its World health report 2000, identified three principal health system inputs: human resources (HR), physical capital and consumables 26. While each of these is important to the delivery of health services, HR is critical to the success of any health system. Put simply, the ultimate impact of any health programme hinges on whether health care workers actually deliver those services. Not surprisingly, human capital is one of the largest assets available within a health system and is frequently the single greatest expense in any national health care budget. In many countries it represents as much as two thirds of the total recurring costs 26.

In spite of its central position in health care systems, HR typically receives less attention than investment in buildings and technology. Since 1951 the government of India has focused heavily on capital infrastructure without any comparable investment in human capital. While the country's rural health system is impressive, with almost 146,000 subcentres, 23,000 primary health centres (PHCs) and just over 3,000 community health centres (CHCs), shortages of human resources are apparent at every level 27. More than 7% of subcentres operate without an auxiliary nurse midwife (ANM) and 50% without a male health worker 28. More than 800 PHC have no physician 17, and CHCs face deep shortages of obstetricians and gynaecologists (56%), paediatricians (67%) and surgeons (56%) 27.

Unfortunately, in today's increasingly globalized world, many HR challenges have moved beyond the control of individual health care systems. India is not untypical in facing a crisis of emigration of doctors and nurses to Australia, Canada, the United Kingdom and the United States of America. Among developing countries, it is one of the largest exporters of health care professionals, with India-trained physicians accounting for approximately 4.9% of practising physicians in the United States, and 10.9% in the United Kingdom 29. One study estimated that almost 11% of graduates for all medical schools in India emigrated to other countries to practise 29. The situation is similar for nurses. A recent survey carried out at two large nursing schools in India showed that approximately 50% of graduating students migrate out of the country 30. This has huge implications for staffing and training within the public health system. Studies have shown that India has lost up to USD 5 billion in training costs since 1951 because of emigration 31.

Researchers exploring the linkages between human resources and maternal mortality have reached contradictory findings. Robinson and Wharrad 3233 showed that density of doctors was significantly related to maternal outcomes. In contrast, Cochrane et al. reported that physicians per capita had no effect on maternal mortality 34. Similarly, neither Kim and colleagues nor Hertz et al. found a significant association between doctor density and maternal death 3536. Most recently, Anand and Bärnighausen, using new data from WHO, found a strong negative correlation between the concentration of physicians and maternal mortality 1. Interestingly, all six studies showed no association between nurse density and improvement in maternal outcomes.

Given the conflicting data, what is the takeaway lesson about physician density and its relationship to maternal mortality? While all the studies have strengths and weaknesses; Anand and Bärnighausen's analysed newer WHO data from 198 countries and is the largest and most comprehensive to date. Their findings suggest that doctors appear best able to address the largest proportion of conditions putting mothers at risk. In addition, such a conclusion would also be consistent with findings showing that developing countries with a shortage of doctors but a large cadre of nurses have had more success with lowering under-five mortality, a health care challenge requiring less specialized interventions, than they have with lowering maternal mortality 1.

Since it seems likely that emigration of physicians and nurses will be a continuing problem, given the low salaries and poor working conditions in developing countries, how can policy-makers address shortages and skill-mix discontinuities? Sibbald and colleagues, in a recent literature review, suggest seven strategies that have been used to realign human resources in health systems 37:

• Enhancement: upgrading a particular job by increasing the skill level of workers or enhancing the role with additional responsibilities;

• Substitution: exchanging one type of worker for another. This might mean for instance, training nurses to take on the role of doctors in primary health care delivery;

• Delegation: moving particular tasks up or down a traditional role ladder;

• Innovation: creating new jobs by introducing a new type of worker with a different role;

• Transfer: moving particular jobs from one health care sector to another;

• Relocation: shifting particular services from one healthcare sector to another;

• Liaison: using specialists in one health system sector for support workers in another.

Developing countries have tried all these strategies, with mixed results. During the 1970s and 1980s, traditional birth attendants (TBA) were trained in midwifery (enhancement) but this appeared to have little impact on maternal outcomes 38. While there is evidence from developing countries that appropriately trained nurses can replace doctors in many care settings (substitution) 39, previously mentioned econometric studies throw serious doubt on whether this strategy is effective in other settings – particularly in developing countries, where nurse and midwife training is often inadequate 1.

The use of TBAs in managing postpartum haemorrhage using the drug Misoprostol has been documented in several resource-poor countries 4041. Since this traditionally would be carried out by a doctor or trained nurse, this task has been shifted down the role ladder (delegation).

There have also been efforts to create new categories of workers (innovation). One particularly successful example is the use of lay health workers to promote immunization and improve outcomes for acute respiratory infections and malaria 42.

There have been a variety of efforts to transfer primary health care functions and sometimes even government staff (transfer/relocation), from the public sector to nongovernmental organizations and private providers when there was a critical need for additional capacity 43. Finally, government health care workers have been used extensively in Africa and Asia to train and support private practitioners 44, an example Sibbald et al. would label a "liaison" strategy.

Considering the scope of the problem, surprisingly little attention has been given to HR management in India. Most efforts have been focused on pilot projects using community health workers in HIV education and testing 45, child nutrition and survival 46, pneumonia management 47 and malaria screening and treatment 48. While some efforts have shown promise, sustainability has been poor because of limited funding from external sponsors. More recently, the government has been experimenting with community health workers called "accredited social health activists" (ASHA) to carry out a variety of health initiatives as part of the National Rural Health Mission 27, but the impact of this strategy is not yet clear. In contrast, on the state level there are a number of innovative and successful programmes realigning human resources, some even decades old. This paper will focus on two very different approaches successfully employed by the states of Gujarat and Tamil Nadu to realign human capital and reduce maternal mortality.

Gujarat, one of India's leading industrial states, is located on the western tip of the country. Despite its ranking among the top five states in the country in per capita income, social and health indicators have lagged far behind those of many of its less well-off neighbours. In 2005, the state had an MMR of 172 per 100 000 live births. While that number was lower than the all-India MMR of 301, it still came in well above Kerala and Tamil Nadu, at 110 and 134, respectively 49. In that year, the state also had an infant mortality rate (IMR) of 54 per 1000 births, almost on par with the all-India average of 58. In contrast, Kerala had an IMR of 14, Maharashtra 36, Tamil Nadu 37, West Bengal 38, and Uttaranchal 4250.

With those grim statistics in mind, Gujarat set out in 2005 to lower maternal and infant mortality. The primary obstacle to the state's efforts was a shortage of human resources. Shockingly, there were only seven public sector obstetrician/gynaecologists (OB/GYN) providing services to a rural population of almost 32 million. In contrast, Gujarat had more than 700 private OB/GYN practising in rural areas. The disparity is not surprising, since private sector specialists receive salaries typically five times higher than those earned in comparable positions in government service 51. Following a series of consultations with both public and private stakeholders, the government developed a Public Private Partnership (PPP) called "Chiranjeevi Yojana" which realigned health system human resources by relocating obstetric gynaecology services from the public sector to the private sector in Gujarat 52.

The scheme was first pilot-tested in five predominantly rural districts, and then scaled up across the state. Under the scheme, the Gujarat Health & Family Welfare Department recruited providers who had postgraduate qualifications in obstetrics and gynaecology; owned their own hospital with a labour room, operating theatre and blood bank; and had access to anaesthesiology services. In return, the state reimbursed physicians approximately USD 40 per delivery. Rather than pay providers directly, the Chiranjeevi Yojana scheme distributed vouchers to all pregnant women living below the poverty line (approximately USD 9 to USD 14 per person per month). Eligible women could choose a local OB/GYN and exchange the voucher for delivery services, free medicines and transport reimbursement 5253.

Through November of 2007, "Chiranjeevi Yojana" enrolled 843 providers and provided for almost 143,000 deliveries. While 642 maternal deaths might have been anticipated in the programme through then, only 31 were reported. Strikingly, only 454 infants died, against an expectation of 6561 in the absence of the programme. Even more impressive, Gujarat was able to deliver these results through the direct relocation of obstetric gynaecology services from public to the private sector 52.

Tamil Nadu is the eleventh largest state in India by area, and the sixth most populous. When compared with All-India statistics, the health status of residents of Tamil Nadu is considerably above average and has seen significant improvement over the years 54. Infant mortality rates have declined from 37 per 1000 in 2005 to 31 per 1000 in 2005/2006 – considerably lower than the All-India rate of 57 per 1000. The state has also made dramatic improvements in maternal mortality, reducing MMR from 195 per 100,000 live births in 1996 to 71 per 100 000 live births in 2007 5455.

In contrast to Gujarat's almost exclusive reliance on PPP, Tamil Nadu has continued to champion a public primary health care model while still struggling with many of the same challenges plaguing other areas of India. For some years, the state has faced chronic shortages of surgeons, anaesthesiologists, obstetrician gynaecologists and laboratory technicians in the public health system 56. In spite of that, the government has continued to invest in health infrastructure, including new primary health centres (PHCs) and extended hours at existing centres 57. In order to deal with staff shortages, the state has successfully used a variety of HR strategies, including enhancement of the non-specialist physician and nursing roles, innovations such as the creation of Comprehensive Emergency Obstetric Newborn Care Centres (CEmONC) in 51 government hospitals 58, and the relocation of some health system functions to the private sector.

As part of its effort to change the skill mix of its workforce, the Government of Tamil Nadu has been aggressively enhancing the roles of non-specialist physicians and nurses. Doctors with MBBS degrees, the lowest qualification for an allopathic physician, are being trained in surgery, obstetrics, anaesthesia and radiology 5960 in order to cope with shortages of specialists. There has also been a concerted effort to upgrade the skills of staff nurses with training in first aid, use of Misoprostil to prevent postpartum haemorrhage, maternal administration of magnesium sulfate, and better birthing practices 61. Additionally, laboratory technicians are being credentialed as X-ray technicians to increase diagnostic capability at PHCs 61.

Tamil Nadu has also focused on reorganizing its public health care systems to ensure accessibility of emergency obstetric care. CEmONC have been established at two hospitals in each district and staffed with specialists in ob/gyn, paediatrics, anaesthesia and general surgery in order to provide referral emergency obstetric and newborn care services 24 hours a day, seven days a week. Each CEmONC has an operating theatre, blood bank, diagnostic laboratory and ambulance service. At current levels, mothers from any area in Tamil Nadu can gain access to emergency obstetric care within one hour 62.

Tamil Nadu has also been encouraging public-private partnerships to facilitate the provision of ancillary services. While the state continues to provide most medical care, it is experimenting with private sector collaborations for ambulance services, facility maintenance, medical equipment, sanitation and construction, to name just a few 6364. In addition, Tamil Nadu is establishing PPPs to provide health care access in tribal areas. Presently it has collaborations with the private companies and NGOs for mobile outreach clinical services, blood banks and provision of training and support for community health workers in remote areas 64.

While overall maternal mortality continues to decline in Tamil Nadu, there is a dearth of data on the impact of individual health system strengthening measures on maternal mortality. The state is currently developing an online monitoring and evaluation system to provide real-time data on health system inputs, outputs and impact 65. Once in place, the system should provide additional information on how various initiatives will affect population-based health indicators such as MMR. As part of these efforts, there is a compelling need for additional research into the contribution of human resource strategies in reducing maternal death in Tamil Nadu.