10 best resources on … health workers in developing countries

Introduction

Health systems cannot function without trained health workers,¹ yet until recently researchers and policymakers paid relatively little attention to their role in developing countries. This is due in part to the inherent complexities and limited availability of data—both of which have also held back research in the world's wealthier countries. But in low- and middle-income countries, these difficulties have been exacerbated by a tendency to focus on more visible issues.

In recent years, however, this has changed. The need to address health workers in public policy took on particular prominence after 2000 when increased foreign aid for health programmes confronted limited capacity in many developing countries to apply those funds—often for a lack of skilled personnel. Research was also spurred by concerns that emigration of health workers was exacerbating the scarcity of health workers in many low- and middle-income countries. Consequently, health workforce issues are now attracting a great deal of attention from politicians, donors, practitioners, advocates and researchers (see, for example, Global Health Workforce Alliance 2006).

As attention has increased, research has also improved. Until recently, most publications could be characterized as either planning studies, using modelling and simulations, or personnel management studies, using interviews and observation to assess health care delivery approaches to staffing. While these studies have their uses, they also miss an important part of the picture because they tend to focus only on employment within the public sector and miss many of the dynamic interactions that occur between health workers and the labour markets in which they move. By contrast, a new generation of research is explicitly addressing both the private sector and the broader labour market with increasingly sophisticated data and research tools.

Our goal in this essay is not so much to present the ‘10 best’ resources on health workers, which would be presumptuous to say the least, but rather to introduce this new generation of research by highlighting a number of good research articles that demonstrate this trend. These articles share a growing awareness that the number and quality of health workers engaged in health care services is influenced by more than government decisions about spending and deployment. They encompass the behaviour of private practitioners and non-governmental organizations; people who seek health care; and health workers themselves as they make choices about their training, employment, location and work effort, all within a context defined by the broader labour market, politics and culture. This is why recent research is often framed within the perspective of labour market analysis even if it emerges from fields like political science, sociology, anthropology, public administration and business management. The better economic studies are also enriched by contributions from these other fields.

We focus on three particularly prominent topics in the literature, namely human resource requirements in health, health worker behaviour, and migration. This leaves aside many important topics that have received relatively less attention, including demand analysis, the role of public sector unions, the use of health workers with different skill sets, the management of health workers, and the scale up of professional education. We also recognize that our choice of articles was influenced by having easier access to the English-language literature and greater familiarity with studies on Africa. Consequently, this essay is our suggested reading for people who are new to the topic, who are interested in how labour market approaches are enriching the analysis of health worker issues, or who are participating in the ongoing dialogue to further research and applied policy on this important topic. In short, it represents the beginning of reading on this topic not the end.

Setting the stage

Research on health workforce issues in developing countries has evolved rapidly in recent years. In 2000, the Rockefeller Foundation sponsored a workshop on health workers in developing countries, culminating with the publication of Towards a Global Health Workforce Strategy (Ferrinho and Dal Poz 2003). Later, the Joint Learning Initiative published Human Resources for Health: Overcoming the Crisis (2004). Many of those involved in these two initiatives also contributed to efforts at the World Health Organization (WHO), which led to its 2006 World Health Report (WHO 2006).

All three of these are good introductions to the breadth of research on health workforce issues in developing countries. The World Health Report 2006 is the most recent of these, providing an overview of the role health workers play in delivering care as well as current thinking about major challenges and policy responses. It also presents the most comprehensive effort to date to classify and count the number of health workers around the world (Dal Poz et al.2006).

While the 2006 World Health Report is at the cusp of the new trend in health worker research, it does not provide as cogent an introduction to thinking about health workers in a dynamic labour market as can be found in Vujicic and Zurn (2006). They write:

We argue that this discrepancy—between what ‘is’ and what ‘ought to be’—arises in large part because policy makers often fail to take into account the behavioural characteristics of individuals who produce health care services (i.e. providers), individuals who consume health care services (i.e. patients) and institutions that employ health care professionals. The actual HRH capacity in a country, we argue, is not determined by any one of these three parties, but results from their interaction within a well-defined HRH labour market.

Vujicic and Zurn apply such an approach to Malawi, showing how the analysis of supply and demand for health workers can generate a better understanding of the problem facing the country and a wide range of appropriate policy options.

Human resource requirements in health

Efforts to estimate the need for health workers have a relatively long history. For example, WHO convened an expert panel in 1971 that outlined four approaches to this problem: calculating the need for health workers based on ‘(1) health manpower population ratio; (2) service target-setting; (3) health demands; and (4) health needs’. It cited a number of studies in OECD countries, most of which could be considered planning simulations and none of which incorporated analysis of the labour market (WHO 1971).

A number of studies have sought to estimate overall need for health workers by estimating the relationship between their numbers and health outcomes (Kim and Moody 1992; Chen et al.2004; Anand and Baernighausen 2007). Anand and Baernighausen (2004) authored one of the better known of these efforts, estimating health worker requirements by measuring the relationship between health worker-population ratios and infant, child and maternal mortality. Such cross-country statistical approaches may seem to be the only alternative given our limited data. Nevertheless, studies like this one have to be interpreted cautiously since they do not control for important confounding factors and assume health workers have the same effect on health status across countries.

Kurowski and co-authors (2004) adopt a different approach by collecting detailed data in two specific countries, Tanzania and Chad. Using a list of priority health interventions and estimates of the time required to deliver them, these authors derive the number of health workers that would be needed to generalize these services to the entire population in each country (see also Kurowski et al. 2007). Thus, their estimates take into consideration health needs, estimates of service coverage, time weights developed from actual experiences, and estimates of staff productivity. This provides a persuasive argument that substantial gaps exist in these countries between the level of human resources available and the level required to deliver a basic set of health care services.

Jokhio and co-authors (2005) use one of the more innovative and compelling empirical approaches to estimating need for health workers. Similar to Anand and Baernighausen, Jokhio and co-authors estimate the relationship between health workers and health outcomes. However, they analyse district-level data on health status in Pakistan, comparing districts that participated in a pilot project to train traditional birth attendants, provide additional materials, and better coordinate care to districts that received usual care. They find a decline in perinatal mortality in the participating districts. The nature of the data and methodology make this one of the most rigorous studies for estimating the impact of health workers on health outcomes because it has strong internal validity, suggesting that future researchers would do well to exploit this kind of data which can be captured when new initiatives are rolled out.

Provider behaviour: financial and non-financial incentives

Many studies look at incentives faced by health care professionals in developed countries (see McGuire 2000 and Simoens and Hurst 2004 for references) but rather few do so in the developing world. Fortunately, this state of affairs is changing and studies are now showing the ways salaries, benefits, promotional opportunities, working conditions and intrinsic motivation influence health worker behaviour. Studies that look at these questions used to rely exclusively on qualitative analysis of interviews, but more recently these have been complemented by studies that either look at preferences as revealed by actual behaviour or use structured questionnaires to elicit preferences.

Chomitz and co-authors (1998) investigate the effect of incentives in Indonesia on newly graduated physicians who must choose where to practice. They use observational data on how doctors responded to a policy change in 1996, when the Indonesian government adopted a policy of giving health professionals who serve in remote areas a greater chance of subsequently getting good appointments in urban areas. But they also use a contingent valuation survey, asking health professionals to choose among hypothetical job opportunities that represent a wider range of pay and promotional opportunities than are offered by the government. Both methods demonstrate that promotional opportunities can serve as a strong inducement to serve in remote areas but the observational data could not judge the effectiveness of pay differentials since, in the observational data, such differences were relatively small. The contingent valuation survey, though, was able to estimate the compensation levels that would likely be needed to encourage doctors to serve in remote areas. By exploring the tradeoffs, Chomitz and co-authors provide valuable information to policymakers regarding their options and how health professionals would be likely to respond to different strategies.

Mangham and Hanson (2008) explore factors that affect the employment preferences of nurses in Malawi. Like Chomitz et al. (1998), they also use a contingent valuation methodology to elicit preferences, an approach that has become increasingly popular in health worker research (see also Serneels et al. 2004 and Mangham et al.2009). The authors find that many attributes influence the preferences of nurses but find that opportunities to upgrade professional qualifications, government housing, and higher monthly pay had the greatest impact on the preferences of the nurses. This study highlights that public sector employers have many options available to them to increase the satisfaction of health workers above and beyond increasing the salary of its workers.

One of the choices made by health workers that has a large effect on the availability of health care is whether or not they show up to work. Unfortunately, absenteeism is a serious issue in many countries (Chaudhury et al.2006). Such absenteeism often reflects decisions by health workers to both hold public sector jobs and work as private practitioners. Ferrinho and co-authors (2004) survey this phenomenon of ‘dual practice’ among physicians, which is common in many countries. They discuss evidence regarding the consequences of dual practice in terms of extracting more income from patients, reducing time served in public health posts, conflicts of interest, and even corruption. By considering the factors that lead to dual practice and evidence on how providers responded to different reforms, they argue that this practice is better understood as a coping strategy in difficult contexts. They argue that simply prohibiting dual practice would be ineffective. Rather the most promising strategies would address the financial gap between what providers expect and what they earn, improve working conditions, encourage professional values, and increase social pressure from their peers and patients.

Migration

One topic that has attracted significant attention—and controversy—is the international migration of health workers. Hagopian and co-authors (2005) provide a good example of studies that address this topic through interviewing service providers to elicit information about their motivations and likely decisions (see also Awases et al.2004). Through interviews and focus groups at six medical schools in Ghana and Nigeria, this study confirms that better pay and working conditions are a major motivation for emigration. However, the authors also find that among West Africa's medical practitioners, ‘training and practicing abroad is a marker of success’, leading to what they call a ‘culture of migration’. Once such norms are established, the levels of pay and working conditions required to retain or encourage the return of émigrés is substantially higher.

While this research provides insights regarding health worker motivation, it often misses the full picture. For example, WHO (2006), Chen et al. (2004) and Dovlo (2005) warn that international migration is a problem for source countries but these studies disregard how international opportunities may affect entry into the profession. By contrast, Clemens (2007) develops a full labour demand and supply model to capture the effects of international migration opportunities and suggests that restricting migration may cause more problems than it might solve. Using a new dataset on African-born physicians in nine major receiving countries, he shows that the stock of African-born physicians working abroad is positively correlated with the domestic stock of physicians in their source countries and inversely correlated with domestic mortality rates there. While this finding may seem counterintuitive, Clemens’ model shows the mechanisms that make this possible and his statistical analysis rejects a number of common hypotheses. He concludes that African countries that produce larger numbers of physicians may have larger numbers of emigrants because they are ‘ill-equipped to absorb new physicians domestically’ and have ‘national characteristics that raise production and emigration simultaneously (such as the quality of medical schools)’. Alternatively, African governments may have responded to emigration by training more physicians and nurses, or people may enter the medical field in greater numbers because they see the opportunity to migrate as attractive. In each case, he has identified direct and indirect effects, showing how the magnitudes vary and pointing in the direction of more helpful public policies.

These last two papers show how the literature on health workers in developing countries is increasingly empirical and sophisticated. Detailed and structured interviews are providing greater insights into motivations and behaviours, including the meaning that health workers ascribe to their occupation. At the same time, studies that fully utilize labour market perspectives are providing fuller understanding of the multiple affects policy changes can have when supply and demand are treated dynamically and individuals are treated as agents instead of passive objects. We are encouraged by the progress that has been made in recent years, but believe that there remains a great need for even more, and higher quality, research on this topic.

Resources

1. World Health Organization. 2006. The World Health Report 2006: Working Together For Health. Geneva: WHO.

2. Vujicic M, Zurn P. 2006. The dynamics of the health labour market. International Journal of Health Planning and Management21 (2): 101–15.

3. Anand S, Baernighausen T. 2004. Human resources and health outcomes: cross-country econometric study. The Lancet364 (9445): 1603–9.

4. Kurowski C, Wyss K, Abdulla S, Yemadji N, Mills A. 2004. Human Resources for Health: Requirements and availability in the context of scaling-up priority interventions in low-income countries. HEFP Working Paper 1/4, London School of Hygiene and Tropical Medicine. Unpublished. Online at: http://www.hefp.lshtm.ac.uk/publications/working_papers.php

5. Jokhio AH, Winter HR, Cheng KK. 2005. An intervention involving traditional birth attendants and perinatal and maternal mortality in Pakistan. New England Journal of Medicine352 (20): 2091–9.

6. Chomitz K, Setiadi G, Azwar A, Ismail N, Widiyarti. 1998. What do doctors want? Developing incentives for doctors to serve in Indonesia's rural and remote areas. Policy Research Working Paper 1888, World Bank. Unpublished.

7. Mangham LJ, Hanson K. 2008. Employment preferences of public sector nurses in Malawi: results from a discrete choice experiment. Tropical Medicine and International Health13 (12): 1–9.

8. Ferrinho P, van Lerberghe W, Fronteira I, Hipolito F, Biscaia A. 2004. Dual practice in the health sector: review of the evidence. Human Resources for Health2 (14).

9. Hagopian A, Ofosu A, Fatusi A, Biritwum R, Essel A, Hart LG, Watts C. 2005. The flight of physicians from West Africa: Views of African physicians and implications for policy. Social Science & Medicine61 (8): 1750–60.

10. Clemens M. 2007. Do visas kill? The effects of African health professional emigration. Center for Global Development: Washington, DC. Unpublished.

Acknowledgements

The authors would like to thank Marko Vujicic and Ruth Levine for comments on an earlier draft of this paper. All remaining errors and omissions are our own responsibility.

Endnote

References