Bulletin of the World Health Organization

The health worker shortage in Africa: are enough physicians and nurses being trained?

Yohannes Kinfu a, Mario R Dal Poz b, Hugo Mercer b & David B Evans c

a. School of Population Health, University of Queensland, Brisbane, Qld., Australia.
b. Department of Human Resources for Health, World Health Organization, Geneva, Switzerland.
c. Department of Health Financing, World Health Organization, Geneva, Switzerland.

Correspondence to Yohannes Kinfu (e-mail: y.kinfu@uq.edu.au).

(Submitted: 28 January 2008 – Revised version received: 11 June 2008 – Accepted: 14 July 2008 – Published online: 10 February 2009.)

Bulletin of the World Health Organization 2009;87:225-230. doi: 10.2471/BLT.08.051599


Recently, considerable attention has been focused on the apparent shortage of health workers in countries with the poorest health indicators, and the potential impact of the shortage on countries’ ability to fight diseases and provide essential, life-saving interventions.13 According to recent WHO estimates, the current workforce in some of the most affected countries in sub-Saharan Africa would need to be scaled up by as much as 140% to attain international health development targets such as those in the Millennium Declaration.4 The problem is so serious that in many instances there is simply not enough human capacity even to absorb, deploy and efficiently use the substantial additional funds that are considered necessary to improve health in these countries.

Health worker shortage in sub-Saharan Africa derives from many causes, including past investment shortfalls in pre-service training, international migration, career changes among health workers, premature retirement, morbidity and premature mortality.5,6 Yet the dynamics of entry into and exit from the health workforce in many of these countries remain poorly understood. This limits the capacity of national governments and their international development partners to design and implement appropriate intervention programmes. In this paper, we fill some of this information gap by providing the first systematic estimates of health worker inflow and outflow in selected sub-Saharan African countries.

For reasons of data availability, our analysis is restricted to two groups of health workers – nurses and midwives combined, and physicians – and to 12 countries for which the relevant data were available.


The analysis required information on the stock of health workers in each country, as well as annual inflows and outflows. Inflows are associated with the number of new workers hired each year, either graduates of training institutions, migrants or people re-entering the workforce. Outflows are caused by premature deaths among health workers, dismissals, resignations (e.g. to migrate or change career) and retirement. Much of this information is not available in many countries, so this study focuses on 12 African countries where information was available on the size, age and sex distribution of the health workforce as well as on graduations from training institutions: Central African Republic, Côte d’Ivoire, Democratic Republic of the Congo, Ethiopia, Kenya, Liberia, Madagascar, Rwanda, Sierra Leone, Uganda, the United Republic of Tanzania and Zambia.

Baseline health worker numbers by age and sex were obtained from a WHO survey of health workers conducted in African countries in 20057. This survey is also the source of the data on health worker training institutions, including the number of trainees by type of worker and year of graduation. We use the number of graduates as the maximum possible level of new domestic graduates hired each year. No data on in-migration of health workers or on the number rejoining the workforce annually were available for any of the countries, though we expect in-migration to be limited to those countries whose health workers are recruited by richer countries.

Information on the outflow of health workers is also difficult to obtain. There are patchy data available on the mobility and mortality of health workers as distinct from the rest of the population, but they are limited in scope and rigour.8,9 Thus, we preferred to use age- and sex-specific mortality rates for the population as a whole, as they are usually of good quality,10 and to assume these rates also applied to health workers. These data were also used to estimate the numbers of health workers retiring each year. In the absence of country-specific information on retirement ages in the public and private sectors, we applied an age of 60 years to all settings on the assumption that all health workers who survive to that age then retire.

Data on out-migration, resignation before retirement age and dismissals were also not available for most of the countries under study. Migration data, for example, are not collected routinely by occupation in either “supplier” or recipient countries and, even when pieces of the puzzle are available, they tend to be either incomplete or of indeterminate time scale.1113 As a result, for the present analysis we adopted rates obtained from two separate case studies in Mozambique and Zambia that provide time-specific data on spatial and career mobility.1416 Both studies focused on public sector health workers. In Mozambique, 2.3% of the workforce left service each year due to resignation (including for migration) or dismissal, while in Zambia only 1.5% left. Because we cannot say if these rates are typical of other countries, we report two sets of figures based on the two rates.

Finally, taking into account available information on inflows and outflows, we compared the estimated net growth rates of the health workforce to population growth rates estimated for the respective countries by the United Nations Population Division.17 This allowed us to assess two important outcomes given current trends. The first was whether the net growth rate of the health workforce is faster than that of the population, allowing health worker density to increase over time. The second was the extent to which the workforce would need to grow in each country to attain the minimum density of 2.28 health workers per 1000 population. This was the yardstick developed and reported as being necessary to achieve desired levels of coverage of key health interventions in the World Health Organization’s The world health report 2006.1


Table 1 presents the estimates of the density of physicians per 1000 population and the annual inflows and outflows per 1000 physicians currently employed. Rates of inflow (also known as workforce regeneration rates) were obtained by dividing the annual number of medical graduates by the total stock of physicians in each country. This is a useful way to understand the proportion of the current workforce that is being regenerated each year. The rate of outflow includes all causes. As explained, premature mortality and retirement rates are country-specific. Scenario I then adds the rate obtained from the Zambian study to calculate resignations and dismissals, while scenario II uses the higher rates observed in Mozambique. Table 2 reports the same estimates for nurses and midwives.

The results for all 12 countries combined show that, for every 1000 physicians practicing in these countries, 59 medical graduates are produced each year. The rate is slightly higher for nursing and midwifery, at 66 new graduates per 1000 practicing nurses and midwives.

The regional average, however, masks the diverse patterns in the study countries. For instance, in 9 of the 12 (the exceptions are the Democratic Republic of the Congo, Ethiopia and Sierra Leone) the rate at which new graduates enter the system is actually higher for physicians than for nurses and midwives. Moreover, countries that have a relatively high graduation rate for one type of health worker do not necessarily have a relatively high rate for the other. For example, Côte d’Ivoire has the highest graduation rate for physicians (14%) but ranks only 7th in the regeneration rate for nurses and midwives (2.7%).

Generally, in all countries outflows are slightly lower for nurses and midwives than for physicians because age-specific death rates are typically lower for women than men and the proportion of females is higher among nurses and midwives than among physicians. For the 12 countries as a whole, each year the health sector is expected to lose some 2.4% of its physicians and 2.1% of its nurses and midwives to premature mortality, and about 4–6% of both due to all causes combined.

Although the 12 countries as a whole are training sufficient physicians to replace outflows when inflows and outflows are considered together, this is not the case in at least one of the outflow scenarios for 6 countries. The situation is even worse when it comes to nurses and midwives, with only 3 countries (Ethiopia, Liberia and Sierra Leone) unequivocally training sufficient workers to replace those leaving the workforce.

However, even in the countries where training is above replacement rates, it is not clear that they will soon be in a position to meet current unmet needs or the increasing demands of an expanding population. Table 3 shows current density per 1000 population for physicians, nurses and midwives combined, with the net rates of increase (or decrease) under the two scenarios and the rate of population growth. Only 6 countries (Côte d’Ivoire, Ethiopia, Liberia, Madagascar, Sierra Leone and the United Republic of Tanzania) show net rates of increases under both scenarios. In the others, the absolute numbers of physicians, nurses and midwives seem to be declining. This decline is even more serious when considered alongside the relatively high rates of population growth in most of these countries.

Even among countries with positive net growth rates, only two (Côte d’Ivoire and Ethiopia) stand a chance of meeting some of the current unmet demands in the future by virtue of unequivocally having a faster-growing number of health workers than inhabitants. Nonetheless, the rate of health worker increase is much slower than that required to increase the density to the WHO target of 2.28 health workers per 1000 population in a relatively short time. The column on the right shows the rate of health workforce growth required for each country for the target to be achieved by 2015, the year set for the achievement of the United Nations’ Millennium Development Goals. Not even these 2 countries are expanding health worker supply fast enough to achieve this aim.


Previous work on health workers in sub-Saharan Africa has focused on the numbers available and on the numbers leaving the workforce at a particular point in time.2,5,6 The results have clearly shown that the current number of health workers is insufficient to meet population health needs at that point in time. This study, which was the first to examine whether current pre-service training can improve the situation, took into account population increases and attrition due to premature death among health workers, retirement, resignation and dismissal. Although each of these components requires separate and careful analysis, the larger picture of workforce dynamics emerges only when they are considered together.

Training capacity insufficient

Our analyses suggest that workforce shortages in the countries under study are even more alarming than suggested by the existing literature. Not only are current numbers insufficient to meet health needs but, in at least 6 of the 12 countries, pre-service training is insufficient to maintain absolute numbers even at their current levels. Current rates of training are sufficient to increase health worker densities in the other 6 countries but, in 4 of them, not enough to keep pace with population growth. This will lead to a drop in health worker availability per person in those countries. Even the 2 countries where current rates of training will increase health worker density will not be able to meet the target level of 2.28 physicians, nurses and midwives per 1000 population until well after 2015.

Future direction

Boosting pre-service training is clearly important but is a longer-term solution because putting in place the infrastructure (human as well as physical) that is needed in these countries will take a long time. Hence, a variety of complementary, shorter-term responses must be considered. For instance, shifting some tasks from people requiring longer-term training to those requiring less intensive training will enable more services to be made available in a shorter time.18,19 Aggressive retention policies, such as improving the remuneration and working conditions of health workers, addressing unemployment, using telemedicine, and encouraging short-term in-migration from surplus to deficit countries, may also be possible, perhaps with donor support.2022 Preventing AIDS will reduce premature mortality among health workers in the longer-term, while providing antiretroviral treatment for health workers who need it will enable them to work longer. The issue of workers resigning to migrate or to change careers is also vitally important, and several international efforts are under way to address this complex issue.23

While these shorter-term options should be considered, it is important not to ignore the more expensive, longerterm issue of pre-service training. Only by addressing all of these facets together can solutions be found to the current health worker crisis in Africa. While considering the policy implications, it is also necessary to be aware of the limitations of the study, the most important of which is the difficulty in obtaining accurate figures for the numbers of health workers in a country. Ours come from a questionnaire sent to WHO country offices. The questionnaires were completed with the help of any official records that were available, including professional registers of members, though these might not be totally accurate or up to date. In addition, estimates of the annual number of graduates from training institutions were sometimes obtained by contacting each of the known institutions. The figures on outflows associated with reasons other than death were taken from two in-depth country studies. Data limitations also prevented us from focusing on other workers besides physicians, nurses and midwives.

We acknowledge, therefore, that the estimates presented in the study might not be exact and highlight the need for more investment in collecting the basic data necessary for informed decision-making. The fact that outflow estimates for dismissal and resignation were derived from two case studies that may not be representative in themselves also calls for caution. However, to address these data problems, we have tried to make the most conservative assumptions possible. For example, we assumed that all graduates from training institutions would immediately enter the workforce. There will be some immediate loss of potential health workers at this stage, so our estimates probably overestimate the ability of current training institutions to replenish supply. It is also important to acknowledge that with current attention being focused on health worker shortages, some of the countries under study may already have scaled up training and taken other steps to alleviate them, and this would not be captured by our figures. It is, therefore, necessary for countries to take appropriate action to promote the collection and analysis of data on entry and exit from the health workforce. ■


Yohanes Kinfu worked at WHO at the time this research was conducted.

Competing interests: None declared.