2. How countries anticipate skill needs in the health workforce

Skills assessment and anticipation exercises are tools used to generate information about the current and future skills needs of the labour market, the available skill supply, and how the two diverge. Their purpose is to make markets work better by informing labour market and education and training participants (Wilson, 2011[1]). Skills anticipation enables a range of stakeholders, including training providers, policymakers, employers, young people and adults, to make better informed education and training choices, leading to improvement in the use of skills and human capital development (ILO, 2015[2]). It enables a more proactive response to factors shaping future skills demand and supply, such as those identified in Chapter 1. While it is not possible to predict the future, skills assessment and anticipation exercises can help actors to plan and prepare for it and minimise the gap between the supply and demand for skills.

Previous studies have scoped out the various skills assessment and anticipation exercises that countries and regions use to measure and anticipate skill and labour demand and supply in the general labour market (OECD, 2016[3]; ILO, 2017[4]; CEDEFOP, 2008[5]). These include employer surveys, quantitative forecasting, qualitative methods, collection and analysis of labour market statistics, surveys of workers or graduates, and sector studies. Each of these methods has advantages and disadvantages, and is best suited to different types of policy uses. Using a combination of methods is therefore viewed as best practice to leverage the advantages of each method (OECD, 2016[3]; ILO, 2017[4]; CEDEFOP, 2008[5]).

This report provides a review of country practices in assessing and anticipating skill needs in the health workforce. The focus is on anticipating future skill needs, rather than assessing current skill needs. That said, an assessment of current skill needs is sometimes a starting point to anticipating future skill needs. In some countries in this review, assessing current skill needs in the health workforce was a more pressing priority than anticipating future skill needs, given limited resources.

This chapter reviews the types of exercises that are used by countries to anticipate skill needs in the health workforce. It draws upon available literature as well as semi-structured interviews with representatives from institutions that carry out such exercises in Australia, Argentina, Bangladesh, Canada, Colombia, Ethiopia, Finland, Germany, Ghana, Ireland, Republic of Korea, the Netherlands, Norway, South Africa and Sweden. Exercises are characterised along five dimensions: how skill needs are defined, time horizon and frequency, scope, the method used, and how they deal with uncertainty in their design. Annex 1 summarizes the skills anticipation exercises covered in this study along these five dimensions.

The term “skill” is understood as having the ability to carry out mental or manual activities, acquired through learning and practice. Skill is an overarching term which includes knowledge, competency and experience as well as the ability to apply these in order to complete tasks and solve work-related problems. Generic skills (also called core skills) are valued in every job, occupation and sector, and include social skills, cognitive skills and basic digital skills (ILO, 2021[6]). Specific skills, by contrast, are not transferrable from one job/occupation/sector to another and refer, for example, to industry-specific knowledge, technical knowledge, or practical competencies that are specific to a particular sector. The OECD Skills for Jobs database finds that certain generic skills are becoming increasingly important for health occupations, such as judgment and decision making, while other generic skills such as psychomotor skills are decreasing in importance (see Chapter 1).

Exercises that focus on skill needs per se, rather than proxies for skills needs such as occupations, have several advantages. First, skill requirements within health occupations may shift over time due to technological change, and focusing on skills enables a better understanding of how these skill requirements are changing. Second, if health workforce skill requirements are changing, focusing on skills allows for a more dynamic policy response than focusing on qualifications or occupations, which typically assume a static bundle of skills. Finally, a focus on skills is relevant for planning task allocation between medical jobs – for instance, to address sudden increases in demand as occurred in the context of the COVID-19 pandemic.

A key challenge for skills anticipation exercises is that skills can be difficult or costly to measure directly. Individual skills are not always easily understandable or quantifiable: it can be unclear what a given skill entails and how it can be learned or assessed. In practice, skills needs are often approximated by estimating which occupations (such as doctors or nurses) will be in greater or lesser demand. Indeed, many exercises in this review of country approaches start by projecting which occupations in the health workforce will be in greater or lesser demand in the future, as an approximation of skill needs. Occupations have the advantage of being easily understandable and they allow for manpower planning within the health workforce. A limitation with exercises that focus on occupations, however, is that it is not always clear which skills or qualifications are required to perform a given occupation. They also have the potential to overlook the ways in which the requisite skills of a given occupation may evolve over time, for instance, to enable a professional to operate new and emerging technologies.

Other common approximations of skills are levels of educational qualifications (such as technical/vocational, university), and/or fields of study (such as medicine, psychiatry, nursing). As with occupations, qualifications are easily understandable and data on the number of new graduates by qualification and educational attainment of the labour force are often readily available. That said, individuals with the same qualification can have different skills, and the available education programmes in a given country do not necessarily teach all the skills needed for a given job. Further, mapping from occupations to qualification needs is not straightforward, though some countries have developed methodologies to do this. For instance, the Advisory Committee on Medical Manpower Planning (ACMMP) in the Netherlands uses a combination of quantitative forecasting, surveys and Delphi methods to anticipate labour demand for 80 health service occupations (Capaciteitsplan), before mapping to qualifications to make recommendations about desired intake in medical programmes.

Exercises that focus specifically on skills needs in the health workforce are rare. As noted above, skills are harder to measure than qualifications or occupations and are less easily understood. A key challenge is that there is not a common language for referring to skills that is widely understood and accepted. Interviewees provided various reasons why their exercise focused on proxies of skills (such as occupations or qualifications) rather than skills per se. Germany’s Federal Institute for Vocational Education and Training (BIBB) and Ministry of Labour attempted to forecast skills by mapping occupations to 16 competencies (Krebs and Maier, 2022[7]), but concluded that a reliable skills forecast would also require sound data on how skill needs are changing within occupations. Korea cited not yet having a National Competency Standard for the health sector as a limiting factor in developing skills forecasts. Another cited barrier was that it was difficult to identify a clear policy use for information about skill needs in the health workforce. This could reflect that strict occupational licensing requirements in health occupations limit the potential for task and labour reallocation and optimal use of skills within the workplace (a limitation that is discussed further in Chapter 3).

Though rare, this study nevertheless unearthed some initiatives that focus directly on future skill needs in the health workforce. For the most part, exercises that focused on skill needs relied on qualitative methods. For instance, in tandem with their quantitative occupational forecasts, Finland’s Skills Anticipation Forum uses qualitative methods to anticipate future skill needs in nine sectors, of which one is social, health and welfare services. Their definition of skills includes both generic skills and occupation-specific skills derived from O*NET. They use a variety of frameworks to define skills, including Austria’s AMS career information system (Arbeitsmarktservice Österreich); the European classification of skills, competences, qualifications and occupations (ESCO); O*NET; and DigComp 2.0, a digital competence framework developed by the European Commission to define digital skills. Australia’s Industry Skills Forecasts, too, take a direct approach to assessing skill needs. Industry skills committees for specific health occupations poll employers about which skills they are missing and which skills they expect will increase in demand, using a list of 12 generic skills and occupation-specific skills as a reference. Finally, while the focus of most initiatives in Norway is on occupations, the Directorate of Health carries out some initiatives that focus on skills in order to support their work in defining the learning outcomes for doctors and other health care personnel. Colombia’s project to update their occupational profiles of professional and transferable skills for health professionals, led by the Ministry of Health, focuses on both current skills and future skills needs that are likely to arise due to technological advances in the sector. In South Africa, in addition to assessing future demand at the occupation and qualification level, the Health and Welfare Sector Education and Training Authority (HWSETA) uses employer surveys to assess the skills needed on average in high, medium and low-skilled occupations, respectively.

Several exercises that anticipate future skill needs in the health workforce consider longer-term time horizons (10 years or more). This is most important for the use of skills intelligence for education and training policies and curricula (i.e. planning or updating curricula or determining training places), since it can take at least 7-10 years to train a health professional, once programmes and/or curricula are in place. In Germany, the QuBe-consortium, consisting of BIBB, the Institute for Employment Research (IAB) and Economics Structure Research (GWS) conduct 20-year quantitative forecasts to anticipate labour demand and supply of 63 sectors, including health, in addition to the Ministry of Labour’s five-year mid-term forecast. The Netherlands Institute for Health Services Research (NIVEL) and the Advisory Committee on Medical Manpower Planning (ACMMP) conduct forecasts for 12 to 18 years. Canada’s occupational forecast (Canadian Occupational Projection System) does 10-year projections for all occupations in the economy, including those in health. In Ethiopia, work to ensure the availability of a sufficient number of health professionals and appropriate skills mix falls under a 10-year National Human Resource for Health Strategic Plan, and labour market supply and demand projections have been carried out to 2030 to help inform education and training policy within the strategic plan.

Several countries consider shorter time horizons when anticipating skill needs in the health workforce. Shorter time horizons are appropriate for informing certain shorter-term policy responses: updating the content of upskilling and reskilling training programmes for the existing health workforce, regulating temporary migration flows, and reallocating tasks between medical jobs. Australia’s Industry Skill Forecasts for Enrolled Nursing carries out 5-year anticipation exercises. Statistics Norway’s projections have a long time horizon (15 years, and sometimes 30-40 years) while national and regional health planning exercises have a time frame of 4 years. In Finland, the focus used to be on longer term skill anticipation (10-15 years), but the approach has since changed to focus on medium-term skill anticipation (5-9 years). They made this change in response to an evaluation that concluded that a shorter time horizon could better facilitate matching skill supply to skill demand. In South Africa, the HWSETA has a five-year sector skills plan, which aligns with other mid- and longer-terms government strategic frameworks. The sector skills plans are supplemented by biannual studies to assess current sector skills development needs. Similarly, while rudimentary in terms of skills analysis, the health sector programme for planning human resources in Bangladesh follows five-year cycles.

Anticipation exercises for the health workforce are generally updated every two to four years. For instance, the Industry Skills Forecasts in Australia are carried out roughly every four years for each occupation. A technical report is produced on health sector skills assessment and analysis in South Africa every two years. In Korea, the frequency of exercises for the health workforce is set by law: under the 2019 Act on Providing Assistance to Health Professionals, the Ministry of Health and Welfare is now under a statutory obligation to conduct a survey and forecast every three years. There are no examples in this review of exercises that are updated annually.

The frequency of exercises sometimes varies upon their intended policy use. Colombia’s evaluations of job performance for clinical and administrative staff, which include skills assessments, have taken place every two years since 2004. Meanwhile, their exercises to update professional and transversal skills profiles of health professionals are targeted to take place every seven years to inform the renewal of health education programmes.

In a number of cases, exercises are carried out on a one-time basis with no intention of updating them regularly. For example, Norway’s new government-appointed Health Workforce Commission is charged with producing an assessment of the needs of personnel and skills in the health sector to 2040, taking into account ongoing trends in the sector and the need to make the health sector economically sustainable. The committee is made up of key stakeholders, including representatives from hospital administrations, health occupation bodies and academics familiar with health technology. The committee has one year to produce the one-off report on the skill needs of the health workforce. This can also be the case where studies are conducted by employers or workers organizations and are funded on an ad hoc basis. For instance, a one-off skill gap analysis of nurses to 2030 in the health care sector in South Africa is being carried out through the National Department of Health alongside the Public Private Growth Initiative.

When forecasting future skill needs of the health workforce, some exercises focus exclusively on the health workforce, whereas others take a broader perspective of the entire labour market, including health alongside non-health occupations and sectors. A whole-of-labour market approach can account for transitions between a wide range of occupations and qualifications within and outside of the health workforce. It also allows for internally consistent comparisons of labour and skill demand across different sectors. However, whole-of-labour market approaches may not be detailed enough to serve the purpose of health workforce planning for specific health service occupations, depending on the extent of focus on health within the broader multi-sector analysis. Exercises that focus specifically on the needs of the health workforce have several advantages. They can provide more detailed information about skill and qualification requirements in the health workforce and are able to define the health workforce at a more granular level (i.e. more detailed occupations) than if the health workforce is just one sector in the exercise. Given their clear focus, these exercises can also be more effective at promoting and facilitating social dialogue and quickly generating results on the future skills needs of the sector.

Several exercises included in this study adopt a whole-of-labour market approach. Many of these are multi-sector quantitative forecasts of future occupational or qualification needs. Some studies that adopt a whole-of-labour market approach have a thematic focus, such as digital skill needs. For instance, the government of South Africa, in collaboration with the ILO, ITU and UNDP, has designed a future skills strategy and implementation programme that aims to assess digital skill needs in education and training programmes throughout the labour market, including in the health sector.

Health workforce specific exercises often use qualitative methods and are often facilitated by sector skills councils or industry skills councils. Such councils have become a critical part of the overall national system of skills anticipation in many countries. For instance, Australia’s industry reference committees develop Industry Skills Forecasts for key sectors in the economy, including health, using foresight and survey methods. However, qualitative approaches are costly to implement for a large number of occupations and sectors because they require significant time and human resources.

Conducting both whole-of-labour market and health workforce specific exercises is a common practice in OECD countries, as the two complement each other. In the Netherlands, the Research Centre for Education and the Labour Market (ROA) forecasts future skill needs for 114 occupations across the entire labour market, of which 13 are health occupations. Also in the Netherlands, the Advisory Committee on Medical Manpower Planning (ACMMP) conducts a health workforce specific exercise that is able to cover 80 health occupations/qualifications, and these are grouped together into nine clusters of specialisations. In Finland, the Finnish National Agency for Education (OPH) combines the two approaches in the same exercise. As a first step, future drivers of change are identified, such as digitalisation or population ageing that apply to all occupations and sectors. This is done qualitatively using Delphi methods, as well as quantitatively using statistical forecasting models. The results from these whole-of-labour market exercises feed into nine sector-specific exercises, including one on social, health and welfare services. The sector-specific exercises identify future trends and skill needs that are specific to each sector.

In low- and middle-income countries (LMICs), conducting skills anticipation exercises is made challenging by weaker institutions, capacities and governance systems. Many developing countries have limited labour market information and require significant investment to develop robust information systems. In these cases, more limited sectoral skills anticipation exercises, such as sector surveys, can help provide useful information to fill data gaps and prepare the sector’s workforce for the future.

Practically all countries included in this study anticipate future skill needs of the health workforce at the national level, and many also produce sub-national results (Table 2.3). Within national studies, several countries studied had sub-national components, even if specific sub-national exercises had not been conducted. Australia and Canada were exceptions within the countries studied, in that specific sub-national exercises are conducted independently from national ones. In Australia, the National Skills Commission (NSC) conducts national-level quantitative forecasts by occupation and sector; the Department of Health conducts long-term, national workforce projections for doctors; and the states use qualitative methods to anticipate future skill needs by sector. In Canada, the provinces conduct quantitative labour market forecasts independently from those produced by Employment Social Development Canada (ESDC), though, the forecasting methods used by the provinces are similar to those of ESDC. In the LMICs studied, the majority of the countries implemented exercises at a national level, with some including regional or sub-regional components.

The decision to anticipate future skill needs of the health workforce at either the national or the sub-national level depends on the governance of health systems and the expected use of the skills intelligence. When regional governments are responsible for health services (such as in Sweden and Canada) or health education (as in Germany, where continuing education and training for nurses is a federal state responsibility) there is more demand for results at the sub-national level. In addition, when skills intelligence is used for career guidance or to plan for an adequate representation of health workers across the country (including in rural and remote areas), sub-national data are essential. In Ethiopia, labour market analysis was initially conducted only at the national level, but the ILO recently recommended that they start to develop sub-national level assessments, given that much of the management of the public health sector has been decentralized, creating challenges for policies, standards, and training programmes. The decision between national or sub-national coverage may also depend on the heterogeneity of the labour market across regions, and in particular the heterogeneity in demand for health services. Several countries indicate that they conduct sub-national forecasts because certain regions have a much higher demand for specific health services than others (for instance, due to different demographics), which implies that shortages in specific health occupations differ across regions. In LMICs, including Ethiopia, Ghana and South Africa, a sub-national component was either included or identified as a priority, due to an urban-rural divide. Healthcare in rural areas in these countries suffer disproportionately from a lack of skilled workers who are less attracted to working in these areas.

Limitations exist at both levels of assessment. On the one hand, national-level assessments may overlook specific skill needs that exist in some regions but not in others. On the other hand, sub-national assessments may be considered unnecessarily detailed when decisions about the content of curricula and the number of students entering health education programmes are made at the national level. Sufficient sample size and statistical infrastructure are particularly important for producing sub-national data, since disaggregating data for the health workforce by region requires robust and valid information at both the occupation and sub-national levels.

Quantitative skills anticipation exercises involve analysing various indicators of current and/or past demand for and supply of health workers and their skills or qualifications, in order to project future trends under given assumptions. Time series models make use of historical trends in the number of health workers, and extrapolate these trends to project the future supply of health workers. Regression models assume that the forecasted variable (such as the demand for health workers) is related to other variables in the environment (such as demographic changes), and create forecasts based on those associations. Other examples of quantitative forecasting models are optimisation models, generic mathematical models, stock-and-flow models, input-output models, social accounting matrices, and simulation models (Safarishahrbijari, 2018[8]). Computable general equilibrium (CGE) models are used to analyse the economy-wide effects of potential shocks and scenarios. For example, the Finnish National Agency for Education applies a CGE model that accounts for 100 industries (including health) and is able to assess the welfare impact of various policies.

Most countries included in this study use quantitative forecasting methods for at least one of their health workforce planning exercises. One of the main advantages of quantitative forecasting methods is that if the methodology is transparent, then the exercise is replicable, including in other countries or settings. Moreover, the method is consistent across a wide range of occupations (across or within sectors), which allows for comparisons between occupations. However, quantitative methods are also data demanding, requiring up to date and accurate labour market information, and people with sufficiently advanced skills in statistical and econometric analyses to develop and interpret them. Data and expertise were identified as barriers in several LMICs, leading to the external commissioning of work or to a greater reliance on qualitative data to inform skills and training needs decisions. Furthermore, it is important to take care when interpreting and presenting the results of quantitative forecast studies as they may give a false impression of precision and certainty.

A good practice in conducting quantitative forecasts is to make use of a variety of data sources, including qualitative data, because this improves the quality and precision of the output (OECD, 2016[3]). This section provides examples of potential indicators of labour demand and supply and of shortage and surplus indicators which are often included in quantitative forecasts. Some countries, particularly the LMICs included in this study, use these indicators independently of quantitative forecasts to assess current skill needs in the health workforce. When limited resources are available, the imperative is to focus on the immediate needs of the sector.

The examples in this section are mostly based on the in-depth interviews that were conducted for this study. For a more comprehensive overview of health workforce projection models, see Ono, Lafortune and Schoenstein (2013[9]).

Projections of the future size and structure of the population (i.e. demographic projections) are a commonly used indicator of demand for health services and health workers in quantitative forecasts. The most basic approach is to identify the share of workers by occupation (as well as sex and age, for instance), and to identify future demand by linking these employment shares with estimated changes in population size or structure. Demographic projections also indicate the demand for health services directly, as healthcare utilisation varies by age and gender and other subgroups. For instance, women between 20 and 40 years old are most likely to seek out midwives, and the ageing of the population increases the demand for geriatric nurses. The Netherlands Institute for Health Services Research (NIVEL) takes into account current utilisation of health services in their occupational forecasts, which, combined with other factors such as an estimation of unmet healthcare demand, expected demographic changes, epidemiological changes, or socio-cultural developments, leads to an estimation of future demand for health services workers.

Other demand indicators that are specific to the health workforce are the expected change in the prevalence of specific diseases (i.e. morbidity) and expected health care expenditures. Since morbidity patterns tend to differ by gender and age, they are often used in combination with demographic projections. For instance, the quantitative forecasts for the health workforce carried out by the Korean Ministry of Health and Welfare use demand indicators such as demographics, utilisation of health services, and the incidence and prevalence of diseases. The World Health Organisation (WHO) models future market-based demand for health workers for 165 countries around the world by using World Bank data on per capita out-of-pocket health expenditures, population aged 65+ and per capita gross domestic product (GDP) (WHO, 2016[10]).

Besides the variety of indicators discussed above, most OECD countries included in this study also use macroeconomic projections as indicators of demand for health services in their quantitative forecasts. GDP growth will generally influence the amount of public and private resources available to pay for health care and therefore the demand for health services and workers (Ono, Lafortune and Schoenstein, 2013[9]). Macroeconomic projections are typically provided by national statistics offices, national planning bureaus or other specialised research centres. For instance, in Canada, macroeconomic projections are contracted out to an external contractor, and include indicators such as GDP, oil prices and interest rates. These variables provide an indication of the future state of the economy at a broad scale and can affect both the demand for and supply of labour. In most LMICs interviewed, it was less common for macroeconomic projections to be factored into health workforce planning, potentially stemming from greater economic uncertainty.

Other factors may also influence the future demand for health workers, such as changes in health service delivery models (such as hospital-centred or primary care-centred systems); and changes in healthcare financing (such as the breadth and depth of health insurance coverage) (Ono, Lafortune and Schoenstein, 2013[9]).

Skill needs are not usually a direct input of quantitative forecasts, though some countries have developed methodologies to convert occupation needs to skill needs. When the skills dimension is taken into account, it is done so by linking occupational forecasts to the demand or importance of skills in occupations, using occupational skills frameworks. For instance, Employment and Social Development Canada is starting to map occupational projections to skill needs using O*NET, and the Dutch Research Centre for Education and the Labour Market (ROA) is similarly planning a research project to translate their occupational forecasts to skills forecasts using ESCO or O*NET. In Australia, the National Skills Commission uses the Australian Skills Classification to transform their occupational forecasts into skills forecasts. However, one limitation with using these types of occupational skills frameworks is that they assume that skill needs required for a given occupation are static over time and across countries. Occupational skills frameworks must be regularly updated so that the skill needs for a given occupation remain up to date, and even then they will often remain retrospective in terms of skill needs. Online job vacancy data can be useful in this regard, as they are available in real-time, and can take into account changing skill requirements within occupations. However, they may not be representative of all health occupations, since certain occupations may be less likely to use online vacancies to recruit new hires. For instance, Cammeraat and Squicciarini (2021[11]) show that online job vacancy data for the United States in 2019 represents about half of total employment of health professionals, and about a third of total employment of health associate professionals, as suggested by the United States Occupational Employment Statistics Survey.

The future supply of the health workforce is a function of the behaviour of the current stock of health workers, as well as projected inflows and outflows. Sources of inflows include education and immigration, while outflows include retirements, attrition, and emigration.

One of the most common measures of the current stock of workers used in quantitative projections is the number of employed individuals by occupation, often taken from population census data or labour force surveys. However, in certain circumstances, the available labour supply may significantly exceed the number of workers employed. For instance, in Ghana, government ministries and unions reported that public resources were a major constraint on hiring health workers and many trained workers remained unemployed for lengthy periods after graduating.

Professional license registers are an additional common indicator of the current stock of health workers. In order to ensure that all health workers meet minimum requirements in terms of knowledge and skills, most countries put in place professional licenses (i.e. the legal requirement to have completed specific education or training to practice). Many countries have a national register to keep track of which health workers have a professional license, and this can be used as a data source for the (potential) supply of health workers in the labour market. For instance, the Advisory Committee on Medical Manpower Planning (ACMMP) in the Netherlands uses annual data on the number of registered medical specialists from the national registers of the Registration Committee Medical Specialists (RGS) as an indicator of supply by detailed occupation, as well as by other variables such as age and gender.

Other potential data sources on the stock of health workers are the number of hours worked or contractual working hours (often obtained through population censuses or labour force surveys) and employment data from health facilities, such as employee head counts, duty rosters, staffing records or payroll records (WHO, n.d.[12]).

Enrolments in and graduations from health education programmes represent a common source of data about inflows to the labour supply. Data on health education graduates provide an estimation of the inflow into the health workforce, depending on the amount of time health education graduates take to find employment and the share who find employment in the health workforce. Health education enrolments allow researchers to forecast the likely number of new entrants into the health workforce in 10 to 15 years, once they complete their programmes. Some countries also analyse job mobility and the number of graduates from non-health education programmes to estimate the potential inflow into the health workforce (Maier and Afentakis, 2013[13]). Typically, data on enrolment, attrition and graduation rates by study programme can be obtained through the Ministry of Education and sometimes the Ministry of Health, depending on which is in charge of health education programmes. Not all countries produce or make available this data, however. In Ghana, for instance, private hospitals did not have access to the number of graduates from public institutions, making it harder to forecast labour supply.

Another important inflow to the health workforce in high-income countries is immigration from other countries (Ono, Lafortune and Schoenstein, 2013[9]). For instance, Employment Social Development Canada (ESDC) includes immigration as an indicator of inflows to the labour supply in its labour market projections that include the health sector. However, since recruiting health workers from abroad may aggravate shortages of health workers in the countries of origin, the WHO encourages countries to refrain from recruiting health personnel from countries that are suffering from acute shortages, and improve the planning of their own health workforce requirements instead (World Health Assembly, 2010[14]) (Ono, Lafortune and Schoenstein, 2013[9]). This problem of ‘brain drain’ of healthcare workers to higher income countries, thus exacerbating shortages in the supply of skilled labour in LMICs, was identified by respondents in LMICs studied, with varying degrees of severity. Some respondents, including in Ghana and South Africa, noted that the current wave of outward emigration of healthcare workers pre-existed COVID-19 but had potentially been exacerbated by countries in Europe and the US relaxing immigration laws for healthcare workers during the crisis. This issue of brain drain, as well as difficulties recognising foreign qualifications and professional licenses, may be why most countries included in this study do not explicitly include immigration as a source of inflows in the quantitative forecasts for the health workforce.

Outflows from the health workforce may be due to retirement, job transitions, burnout, family reasons or emigration. In the LMICs studied, there were significant job exit rates due to emigration to higher income countries, usually for higher pay or lower unemployment. Some respondents also indicated that worker fatigue and burnout since the COVID-19 pandemic was contributing to job exit rates, in addition to the loss of life of workers during the crisis. Although it is hard to predict how many people may leave the health workforce for personal reasons, countries often predict exits due to retirement based on demographic projections of the number of people that will reach the legal retirement age at a given moment in the future. However, in many countries, doctors continue to practice beyond the legal retirement age (albeit often with reduced working hours), especially those in private practice. Taking doctors’ actual retirement patterns into account may require modelling retirement as a gradual reduction in working time rather than an abrupt end of service (Ono, Lafortune and Schoenstein, 2013[9]). The Dutch Advisory Committee on Medical Manpower Planning (ACMMP) model takes into account physicians’ actual retirement age by using national registration data, and estimates the size of outflows into retirement over the next 5, 10, 15 and 20 years. In South Africa, 70 per cent of the healthcare workforce are projected to retire within the next ten years, causing a potential labour supply shortage and a spike in the number of nurses to be trained. In Ethiopia, supply-side forecasting was carried out by forecasting inflows (training capacity and migration) and outflows (retirements, emigration, deaths, resignations and dismissals) and using a stock-and-flow approach to estimate the future supply of the health workforce to 2030.

Exit rates can also be measured irrespective of the reason for leaving the health workforce. The German QuBe-consortium and Employment Social Development Canada (ESDC) explicitly take job exits and job mobility into account in their quantitative projections. They use different mobility matrices, for instance by age or gender, to estimate future supply and demand by occupation. Data on job exit rates at the occupational level are often obtained through population censuses or labour force surveys that include respondents’ previous occupation, or through administrative records on retirees and their previous occupation (Maier et al., 2017[15]).

Most of the exercises identified in this study do not incorporate indicators of skills supply in their quantitative forecasts. Under the assumption that workers acquire certain skills and experience as they get older while other skills decline or become obsolete, demographic projections may provide an indication of the future supply of skills. Another way to estimate the skills supply directly is by using Big Data and Artificial Intelligence (AI). Using techniques such as Natural Language Processing, job and curricula descriptions can be analysed to extract the skills that people in certain occupations or with certain educational qualifications are likely to have. For instance, Headai, a Finnish company developing AI solutions to facilitate decision making, uses AI to categorise descriptions of education and training curricula into skill categories as a measure of skills supply (Verhagen, 2021[16]).

Directly testing the skills of health professionals is another way to obtain a measure of the skills supply. However, no examples of direct assessments of skills of health professionals were identified among the countries that participated in this study. Tools for direct assessment of skills take the form of examinations or tests performed by respondents in the presence of trained interviewers, and direct observations of actual performance by trained observers. The OECD Survey of Adult Skills (PIAAC) is an international direct assessment that evaluates skills in numeracy, literacy and problem solving for the general working adult population. These results can be disaggregated for specific health occupations, down to 3-digit ISCO for some countries. Health Education England piloted a digital literacy self-assessment tool in one hospital in 2021, with plans to roll out nationally (Health Education England, n.d.[17]). The tool was designed to provide senior managers and leaders in the health workforce with anonymised data to help them understand the digital literacy skills and training needs of their staff (https://www.hee.nhs.uk/our-work/digital-literacy/digital-literacy-self-assessment-diagnostic-tool). A direct examination type of assessment instrument was also under development by the National Council of State Boards of Nursing (United States) for rigorously evaluating clinical reasoning skills among nurses (Maeda and Socha-Dietrich, 2021[18])

In addition to measures of skills supply and demand, some countries track indicators of shortages or surpluses as inputs to quantitative forecasts. One way to collect such information is to conduct employer surveys, in which employers are asked whether they experience any hiring difficulties. These are often undertaken by employers’ organisations, professional associations as well as by governments. While some data collected by employer surveys is purely qualitative (such as open-ended questions about skill needs), the responses to targeted questions can be used in quantitative forecasts. For instance, in Norway, municipalities and health facilities are asked how many people they need to hire for the health workforce in a given year. Although employer surveys are a good indicator of perceived shortages, these subjective assessments may, in some cases, reflect hiring difficulties for reasons other than shortages, such as poor working conditions, inadequate remuneration or poor human resources policies (OECD, 2017[19]). In South Africa, the HWSETA conducted employer surveys, but a lack of motivation among employers to participate in these surveys was considered a constraint on effectively collecting labour market information for skills analysis in the healthcare sector.

Hard-to-fill job vacancies (defined as job vacancies that remain open over a given reference period) are another indicator of shortages, because prolonged unfilled vacancies or high job vacancy rates can signal that employers are facing difficulties in finding enough people with the right skills to fill a position at a given wage (OECD, 2017[19]). For instance, in the Netherlands, the Advisory Committee on Medical Manpower Planning (ACMMP) uses vacancy rates by detailed occupation in their quantitative projections. In Germany, the QuBe-consortium uses occupation and skill specific search durations from an employer survey as an indicator of how hard it will be for employers to fill a vacancy. In Ghana, vacancies were established to be the main mechanism for informing training policies, with only a nascent focus on the future skills needs that the industry may require. In most countries, national statistics offices or public employment services (PES) provide vacancy data by detailed occupation and/or by sector. Real-time online job vacancy data (‘Big Data’) can also be used, although these data may not be representative of the whole labour market (OECD, 2017[19]). Member countries of the European Union can also use the Job Vacancy Statistics (JVS), which are available by country, region, year, sector (including Health and social services), and 1-digit ISCO occupation. While widely used, the use of hard-to-fill job vacancies as an indicator of shortages is primarily reactive and focuses on the current demand of professionals as a proxy for existing skills needs. Furthermore, vacancies may be hard to fill due to a variety of reasons including poor working conditions or non-competitive wages.

Another indicator of shortages in the health workforce that is used in quantitative projections is patients’ waiting time before seeing a doctor. For instance, the Advisory Committee on Medical Manpower Planning (ACMMP) in the Netherlands uses time series of waiting time (in weeks) for first outpatient clinic visits as an indicator of shortages of health services workers. They also use data on health education graduates’ search time for finding a postgraduate training programme. The assumption is that shorter time intervals indicate higher demand or even shortages of health services workers. They obtain these data from surveying medical doctors about how long they searched for a training place and administrative data on the time interval between final degree examination in medicine and intake in a postgraduate programme (ACMMP, 2020[20]).

The WHO provides an example of how shortfalls from a minimum requirement of supply might serve as an indicator of shortages of health services workers. Based on several analyses they calculate an indicative threshold of 4.45 physicians, nurses and midwives per 1 000 people, and use that in their estimations of health workforce needs and needs-based shortages by 2030. In Ghana, for instance, the WHO tool was being used as the primary tool to determine the number of workers that needed to be trained for each specialisation. The WHO emphasises the limitations of this approach in that this number only reflects a selected number of health service occupations, and it does not reflect the heterogeneity of countries in terms of baseline conditions, health system needs, optimal workforce composition and skill mix (WHO, 2016[10]).

As markets should assign a higher price to scarce skills or occupations, wage growth by occupation is a commonly used signal of skill shortages in many skill assessment and anticipation exercises (OECD, 2017[19]), though rarely used for such assessments in the health workforce. Wage data can, for instance, be obtained through administrative data (such as from tax authorities) or labour force surveys. Among private sector employers in Ghana, wage inflation was listed as a major indicator of skills shortages, and wage data was used as a tool for human resources planning. In this case, the data was coupled with information on the length of time vacancies were taking to fill. However, wage data are not used in the majority of exercises included in this study, nor in a previous OECD review of quantitative projection models in the health workforce (Ono, Lafortune and Schoenstein, 2013[9]). Among the 26 projection models from 18 OECD countries included in the previous study, wages (or other modes of provider payment) were only included in one study as a variable affecting the future supply and demand for health workers. A potential reason for the limited use of wage data in skills anticipation exercises for the health workforce is that such studies are often conducted at too detailed an occupation level to have sufficient and reliable data on wages. Another reason may be that in many countries, health workers predominantly work in the public sector, and wages in the public sector are not sufficiently sensitive to shortages and surpluses to serve as a useful indicator.

Qualitative skills anticipation exercises can consider a broader range of factors than those that can be easily quantified. This makes them helpful in skills anticipation, given that skills are difficult to quantify and skill needs are dynamic. Moreover, they can facilitate discussion about emerging skills needs associated with new technologies which may not yet show up in national qualification frameworks or occupational skills frameworks (such as O*NET, ESCO, or the UK Skills and Employment Survey). Qualitative methods are also generally easier to set up in contexts where financial resources or statistical expertise are more limited. This was reflected in LMICs being more likely to rely on qualitative methods. Disadvantages of qualitative methods are that they are subjective, non-systematic and potentially yield inconsistent responses.

Qualitative methods often involve gathering groups of experts and/or stakeholders to share their informed views on how the skill needs of the health workforce are likely to evolve. These include focus groups, stakeholder consultations, foresight methods, and Delphi methods. Other qualitative methods include surveys of managers, health workers, or graduates. While surveys are a qualitative method, their outputs can be both qualitative and quantitative.

Focus groups, stakeholder consultations, foresight methods and Delphi methods are similar in that they involve gathering stakeholders to share informed views on how skill needs in the health workforce will evolve. A focus group is a facilitated group discussion that is "focused" on a particular topic, such as which skills will be in higher demand in a given health occupation in the future (Brown, 2019[21]). The term ‘stakeholder consultation’ broadly refers to developing relationships with stakeholders with a range of aims: from a one-way relationship with the sole purpose of informing them, to a two-way relationship aimed at gathering feedback or even involving and empowering stakeholders in the decision-making process. The Delphi technique structures a group communication process by bringing together a panel of experts to formulate a prediction or set of priorities (Brown, 2019[21]). Foresight uses a range of methods, such as scanning the horizon for emerging changes, analysing megatrends and developing multiple scenarios to reveal and discuss ideas about the future (OECD, n.d.[22]).

In Australia, industry reference committees carry out Industry Skills Forecasts in non-academic health occupations using foresight methods. The purpose of the exercise is to update the competency framework (“training package”) for particular non-academic occupations, such as enrolled nursing1, ambulance and paramedics, and direct client care. Industry reference committees are comprised of key industry bodies related to the particular occupation. For instance, the Enrolled Nursing Industry Reference Committee includes representatives from Aged and Community Services Australia, the Australian College of Nursing, the Department of Health, and the Australian Private Hospitals Association, among others. As part of the exercise, the Department of Education and Training provides a list of 12 generic skills for industry representatives to rank in order of importance for the occupation. Each industry reference committee consists of about 20 members, and each state does their own exercise by sector. The industry reference committee approach depends on having well-established networks of industry experts who are actively engaged in improving the quality of skills and training in the sector, and willing to volunteer in lengthy consultations. The work of the industry reference committee is coordinated by support organisations (like SkillsIQ) that receive funding from government, and that are responsible for producing the final Industry Skills Forecast and submitting it to the Australian government for validation.

The qualitative component of Finland’s Skills Anticipation Forum is based on a series of foresight workshops. There are nine anticipation groups in total, and social, health and welfare services is one of them. Each of the five phases of the process involve a foresight workshop. Phase one and two build an understanding of the major trends affecting the sector and develop scenarios. Phase three looks deeper at what is happening within firms and organisations and how they are likely to cope under the different scenarios. Phase four anticipates skill and education needs of the sector under the various scenarios, and phase five focuses on developing the proposal for the Ministry of Education with recommendations about how qualification requirements should change, determining student numbers in health programmes and building training programmes. Quantitative forecasts serve as an input for discussion in these workshops, and particularly in the first two phases. Each anticipation group has about 25 members made up of employers, employees, entrepreneurs, technical and vocational education and training (TVET) and higher education institutions, and educational administrators. A network of experts is involved in validating the proposal. Finland is exploring ways to combine its qualitative and quantitative approaches in a more systematic way to anticipate skills needs, and has started incorporating the use of the “E-Delphi method,” an online discussion platform that allows participants to discuss different scenarios and megatrends.

The Irish Health Service also employed Delphi methods in its one-off 10-year projection of demand for workers in acute hospital services. They held workshops to develop scenarios based on grade mix and skill mix and consulted broadly with stakeholders in the acute hospital services. The findings from the exercise will be used to engage with the Department of Health to strategize about where to source the supply.

As part of their project “Accelerating adoption of AI in the health sector,” Canada’s Michener Institute of Education relies primarily upon stakeholder consultation to understand the readiness of health professionals to use new AI tools, and identify the skills that are needed to use these tools. The team built an extended network of partners from royal colleges and digital health groups who meet several times a year to share developments in their environments. A symposium also gathered 500 stakeholders to share experiences in the use of AI in the health sector.

Stakeholder consultations were also an important component of the approach taken by the WHO in developing the Global Competency Framework for Universal Health Coverage. The process involved iterative consultation and validation of the selected competencies and the underlying conceptual approach with a working group of education experts, as well as with a virtual community of practice that included academic institutions, individual experts, and agencies involved in health worker education. The approach was underpinned by a review of existing competency frameworks and competency-based curricula, as well as documentation about how the roles and responsibilities of health workers were likely to evolve.

In the United States, SEIU Pennsylvania (a sub-national trade union of healthcare workers) consults with health professionals and organises workshops and conferences, in order to assess future trends in the health sector and how this affects training needs of health workers. Part of this work is done in collaboration with the Consortium for Advancements in Health & Human Services, which offers an array of educational and consultancy-based services.

Surveys are used to collect data from targeted individuals and organisations about skills gaps and skills demand within the health workforce. Their output can be either qualitative (such as responses to open-ended questions about skill needs) or quantitative (such as the share of hospitals that have difficulty finding workers with required skills). In comparison with focus groups and other qualitative methods mentioned above, surveys are relatively more demanding in terms of cost and statistical expertise.

The qualitative output generated by surveys can provide a rich picture of skill needs. Surveys about skill needs in the health workforce are generally directed at employers and managers, health workers themselves and sometimes at recent graduates. The Netherlands’ Institute for Health Services Research (NIVEL) and the Advisory Committee on Manpower Planning (ACMMP) conduct surveys among both health workers and managers and ask them to evaluate how difficult it is to fill vacancies (quantitative output) and to indicate which tasks they expect will change (qualitative output). The Norwegian Committee on Skill Needs conducts an annual survey that polls county municipalities and employers in the health sector (as well as in other sectors) about how many people they need to hire. As part of Australia’s Industry Skill Forecasts, industry representatives who are part of industry reference committees are asked to complete a questionnaire to rank how important a list of skills are for the health workforce, and whether they expect their importance to increase over time. Korea’s Ministry of Health and Welfare conducts a workforce survey with health professionals themselves to identify health and medical personnel issues. Respondents can complete the survey in person, in writing, or over telephone. In Argentina, too, the trade union of health workers – Federación de Asociaciones de Trabajadores de la Sanidad Argentina (FATSA) – regularly conducts surveys among union members in hospitals, in order to assess current and future manpower needs and working conditions. The Netherlands’ Research Centre for Education and the Labour Market uses surveys of recent health graduates as an input into its sectoral forecasts.

The Norwegian Health Workforce Commission plans to use interviews to complement an analysis of hospital administrative data to meet its mandate to investigate the needs of personnel and skills in the health sector to 2040. To analyse if there are possibilities for different groups of health workers to start sharing tasks, or to change the way in which they are currently sharing tasks, the Commission is making use of administrative data from hospitals about which tasks are currently carried out by different occupations. They will likely complement this analysis with interviews with hospital managers and health workers, and may also physically observe how health workers carry out their tasks on the job.

Surveys can also be directed at employers and trainers to collect data on the training that workers are receiving. In South Africa, HWSETA conducted a survey among employers and skills development providers following interruption to training during COVID-19 to determine whether training had resumed. Seventy-five per cent of employers and 82 per cent of skills development providers indicated that training had resumed during lockdown periods, which was determined to be a positive development for the supply of skills.

Ideally, skills anticipation exercises adopt a holistic approach and combine various qualitative and quantitative methods in order to achieve robust and reliable results (CEDEFOP, 2008[5]). Indeed, most countries included in this study combine quantitative and qualitative methods, as part of the same exercise or as a set of exercises (Table 2.4). For instance, some countries (Finland, Germany, Korea, the Netherlands, Norway, and South Africa) employ qualitative stakeholder consultations to verify the validity of their quantitative forecasts. It is also common to include employer or graduate surveys as one of the inputs in quantitative forecasts, or alternatively, to use the results from quantitative forecasts as an input to stakeholder consultations. For instance, the Finnish National Agency for Education (OPH) and the Advisory Committee on Manpower Planning (ACCMP) in the Netherlands conduct quantitative forecasts that serve as an input for Delphi discussions. Colombia’s Ministries of Health and Education noted that the approach towards skills anticipation in the health workforce had been too qualitative, citing a lack of reliability and accuracy of previous exercises as a limitation to their policy application. This further highlights the need for a combination of quantitative and qualitative methods to produce skills intelligence that is fit for policy use.

Uncertainty is a key challenge when anticipating future skill needs: it is impossible to precisely predict the future, and therefore the required size of the future health workforce or the skills they will need to have for future scenarios. The longer the time horizon, the greater the uncertainty. Exercises with a shorter time horizon (less than 10 years) that are intended to inform shorter-term policy responses (such as upskilling the existing health workforce in digital skills, or temporary migration flows) tend to involve less uncertainty than those that project 10 years or more into the future, and that are intended to inform the number of training spaces in health education programmes or the curriculum of such programmes. Economic uncertainty can greatly affect the supply and demand of healthcare workers in LMICs, which are sometimes unable to absorb the healthcare workers that they have trained into the public healthcare sector due to budgetary limitations. Meanwhile, external factors, such as unexpected spikes in demand for migrant health workers abroad, can increase the brain drain of skilled workers in ways that can be difficult to plan for.

The costs of under- or overestimating future skill needs in the health workforce could be larger than in other sectors. Underestimates could lead to future health workforce shortages, which could in turn contribute towards higher morbidity and mortality rates in the population. However, overestimating future skill needs of the health workforce is costly too, given the large private and public investments needed to train specialised medical doctors (both in terms of time and financial resources).

Among the exercises included in this study, four different strategies for dealing with uncertainty in the skills forecasts could be identified. The first is to repeat the exercise frequently (such as every two or three years), in order to incorporate emerging insights about changing future trends. Another strategy for dealing with uncertainty in quantitative exercises is to provide ranges instead of exact numbers. For instance, the Advisory Committee on Medical Manpower Planning (ACMMP) publishes estimates of the minimum and maximum number of health workers that will be needed in the future. Employment Social Development Canada (ESDC) and the Research Centre for Education and the Labour Market (ROA) in the Netherlands transform the projected numbers of workers needed into categories (such as small, medium or large shortages/surpluses) and only report the category that each occupation falls into. Analysis in South Africa forecasted ranges for future nursing shortages, highlighting that an existing gap of 26,000-61,000 nurses was likely to increase to 131,000-166,000 by 2030. Representatives from South Africa acknowledged that the wide ranges were indicative of uncertainty and issues with data reliability. A third strategy is to communicate to the final user how reliable the projections are. The Netherland’s ROA provides a measure of an occupation’s historical business cycle sensitivity, where higher sensitivity to economic downturns or growth imply more uncertainty about that occupation’s labour demand projections.

Finally, several exercises deal with uncertainty in their design by investigating scenarios. When this is a quantitative exercise, the forecasting model is run under different assumptions or hypothetical future events, to show how they affect the forecasted numbers. A qualitative approach to scenario development involves asking field experts or policymakers about what potential future scenarios might look like, how likely these scenarios are to happen, and how they would affect the forecasts. One of the countries that incorporates scenario development into its qualitative exercises is Finland, where experts discuss how skill needs for the health sector (and other sectors) would change under different scenarios. While none of the scenarios analysed by the Finnish Skills Anticipation Forum has taken into account the possibility of a global health pandemic, efforts have been made to collect information ex-post about changing skill needs as a result of the COVID-19 pandemic. Members of the Skills Anticipation Forum participated in a survey to identify common skills and tasks between occupations, in order to understand which occupations could be an alternative in case of a shortage in another, as took place during the pandemic.

Validating findings is a necessary step in producing quality skills intelligence. Most countries validate findings from skills anticipation exercises with external experts before publishing them. This is an opportunity to discuss whether the results, and the assumptions they are based on, are considered plausible. Some countries (Australia and Canada) send the results to experts for feedback and conduct follow-up consultations in case there is any disagreement with the projected numbers. Other countries (the Netherlands, Finland, Norway) ask a mixed group of experts such as employers, education providers, researchers, health authorities, patient organisations and ministries to discuss the validity of the results together. In either case, the validation process leads to refining the forecasts. In Colombia, a variety of methods are used to validate findings from skills anticipation exercises in the health workforce, including expert focus groups, and cross-referencing/comparing data gathered from different sources in Colombia or from the same sector in other countries.


[20] ACMMP (2020), Recommendations 2021-2024, http://capaciteitsorgaan.nl/app/uploads/2020/04/2020_02_12-Capaciteitsplan-2021-2024-Hoofdrapport-DEFINITIEF-EN.pdf (accessed on 21 April 2022).

[21] Brown, J. (2019), “Interviews, focus groups, and Delphi techniques”, in Brough, P. (ed.), Advanced Research Methods for Applied Psychology: Design, Analysis and Reporting.

[11] Cammeraat, E. and M. Squicciarini (2021), “Burning Glass Technologies’ data use in policy-relevant analysis: An occupation-level assessment”, OECD Science, Technology and Industry Working Papers, No. 2021/05, https://doi.org/10.1787/cd75c3e7-en.

[5] CEDEFOP (2008), “Systems for anticipation of skill needs in the EU Member States”, CEDEFOP Working Papers, Vol. 1, https://doi.org/10.2801/24837.

[17] Health Education England (n.d.), Digital Literacy Self-Assessment Diagnostic Tool.

[6] ILO (2021), Global framework on core skills for life and work in the 21st century, International Labour Organization, https://www.ilo.org/skills/pubs/WCMS_813222/lang--en/index.htm.

[4] ILO (2017), Skill needs anticipation: Systems and approaches. Analysis of stakeholder survey on skill needs assessment and anticipation, International Labour Organization, https://www.ilo.org/skills/areas/skills-training-for-poverty-reduction/WCMS_616207/lang--en/index.htm.

[2] ILO (2015), Guidance Note: Anticipating and matching skills and jobs, International Labour Organization, https://www.ilo.org/skills/areas/skills-training-for-poverty-reduction/WCMS_534307/lang--en/index.htm.

[7] Krebs, B. and T. Maier (2022), Die QuBe-Kompetenzklassifikation als verdichtende Perspektive auf berufliche Anforderungen, Bundesinstitut für Berufsbildung (BIBB), Bonn.

[18] Maeda, A. and K. Socha-Dietrich (2021), “Skills for the future health workforce: Preparing health professionals for people-centred care”, OECD Health Working Papers, No. 124, OECD Publishing, Paris, https://doi.org/10.1787/68fb5f08-en.

[13] Maier, T. and A. Afentakis (2013), “Forecasting supply and demand in nursing professions: impacts of occupational flexibility and employment structure in Germany”, Human Resources for Health, Vol. 11/1, https://doi.org/10.1186/1478-4491-11-24.

[15] Maier, T. et al. (2017), “Modelling reallocation processes in long-term labour market projections”, Journal for Labour Market Research, Vol. 50/1, pp. 67-90, https://doi.org/10.1007/s12651-017-0220-x.

[19] OECD (2017), Getting Skills Right: Skills for Jobs Indicators, Getting Skills Right, OECD Publishing, Paris, https://doi.org/10.1787/9789264277878-en.

[3] OECD (2016), Getting Skills Right: Assessing and Anticipating Changing Skill Needs, Getting Skills Right, OECD Publishing, Paris, https://doi.org/10.1787/9789264252073-en.

[22] OECD (n.d.), What is Foresight?, https://www.oecd.org/strategic-foresight/whatisforesight/ (accessed on 12 May 2022).

[9] Ono, T., G. Lafortune and M. Schoenstein (2013), “Health Workforce Planning in OECD Countries: A Review of 26 Projection Models from 18 Countries”, OECD Health Working Papers, No. 62, OECD Publishing, Paris, https://doi.org/10.1787/5k44t787zcwb-en.

[8] Safarishahrbijari, A. (2018), “Workforce forecasting models: A systematic review”, Journal of Forecasting, Vol. 37/7, pp. 739-753, https://doi.org/10.1002/FOR.2541.

[16] Verhagen, A. (2021), “Opportunities and drawbacks of using artificial intelligence for training”, OECD Social, Employment and Migration Working Papers, No. 266, OECD Publications, Paris, https://doi.org/10.1787/22729bd6-en (accessed on 30 June 2022).

[10] WHO (2016), Global strategy on human resources for health: Workforce 2030, World Health Organization, https://apps.who.int/iris/handle/10665/250368.

[12] WHO (n.d.), Density of physicians (per 1 000 population), 2022, https://www.who.int/data/gho/indicator-metadata-registry/imr-details/112 (accessed on 19 April 2022).

[1] Wilson, R. (2011), Anticipating Changing Skill Needs: A Master Class, University of Warwick, New Skills Network event, https://www.ams-forschungsnetzwerk.at/downloadpub/Rob%20Wilson%20-%20Forecasting%20skills.pdf (accessed on 11 April 2022).

[14] World Health Assembly, 6. (2010), WHO Global Code of Practice on the International Recruitment of Health Personnel, World Health Organization, https://apps.who.int/iris/handle/10665/3090.


← 1. Enrolled nurses have completed their diploma qualification. They tend to work in a team and have less authority than registered nurses who have finished their degree.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD/ILO 2022

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.