An estimated one billion people worldwide experience some form of disability and are in need of health and rehabilitation services, the majority in low- and middle-income countries [1]. Despite the urgency of the issues at stake, prioritizing and monitoring of progress to improve health services for people with disabilities remains inadequate [2]. Notably, the Millennium Development Goals (MDGs) compact, meant to establish a unifying set of objectives on pressing health and development issues and encourage collaborative action among the global community, fails to explicitly mention rehabilitative health services or set service coverage targets for persons with disabilities [3].

Enhancing accessibility to health services means addressing the key constraints related to human resources for health (HRH). For one, efforts to implement the new and ambitious international guidelines for community-based rehabilitation (CBR) [4] are expected to place heavy demands on practitioners to work across disciplines and sectors to meet the medical and psychosocial needs of people with disabilities [5]. Yet despite their central role in services delivery, human resources for rehabilitation are an often neglected component of health systems development. Human resources for rehabilitation are often absent from national health sector plans and reviews or HRH development strategies [6].

Assessing the availability of rehabilitation health workers is a critical starting point for understanding the capacity of health systems to meet health-related rehabilitation service objectives in a country. A few studies have profiled the rehabilitation workforce and forecast gaps using different data sources and approaches, usually focusing on a single profession or practice modality and limited to a specific country or region [7–12]. Research on determinants of workforce supply and distribution among rehabilitation professionals is minimal [13].

One complexity in understanding the situation on human resources for health-related rehabilitation services is that there is no commonly adopted monitoring framework or universal "gold standard" for required human resources. In any health system, different categories of health workers may provide different forms of health and rehabilitation services. The specific mix of personnel needed in local contexts will vary depending on the circumstances of the area. For example, a country with large numbers of motor vehicle accidents may need more workers specialized to deal with cognitive and musculoskeletal impairments, whereas another country may need more workers skilled in providing services for disabilities associated with HIV/AIDS and other communicable diseases [1].

Not only do the settings for rehabilitation vary greatly from country to country, information on the availability of rehabilitation personnel to staff these settings is often only an estimate. Data remain fragmented and inadequate, especially in low- and middle-income countries. This is partly related to lack of common definitions and classifications, partly to poor availability and use of standard statistical sources for workforce monitoring, and partly to lack of political will to place monitoring of human resources for rehabilitation high on the health agenda - the latter itself may be related to the way societies often interpret and react to disability. For instance, Haig et al. facetiously concluded, taking into account the lack of documentation on physical and rehabilitation medicine in sub-Saharan Africa, the chance of a person with a disability in sub-Saharan Africa meeting a physician with specialist skills is about the same as that for an Antarctic penguin [14].

This study aims to strengthen the global information and evidence base on human resources for rehabilitation. Since the release by the World Health Organization (WHO) of its flagship publication on the health workforce, The world health report 2006: working together for health [15], an increasing number of studies have attempted to improve understanding empirically and methodologically of the global HRH situation in relation to selected health dimensions, notably those prioritized by the MDGs (see [16–18]); however, to the best of our knowledge none have examined the situation with regard to disability and rehabilitation. This article presents new cross-national findings of supply of and need for rehabilitative personnel within and across regions, as a basis for strategy development of the workforce in physical and rehabilitation medicine. It proposes a standardized approach for measuring and monitoring health workforce capacity to respond to population needs for rehabilitation services. The underlying objective is to encourage a greater number of countries and stakeholders to plan for an effective, sustainable rehabilitative health workforce and implement ongoing monitoring to inform decision-making for HRH policy and practice and enhanced accountability.

Data on the supply of rehabilitative personnel are primarily drawn from a custom extract of statistical information on health occupations from official national sources collated in the WHO's Global Atlas of the Health Workforce [19], the main international database on health workforce information. Depending on the organization of national health systems and means of monitoring, the database captures information from various administrative sources including health facility staffing records, civil service payroll records and registries of health professional regulatory bodies, as well as from population-based sources such as censuses and surveys with questions on labour force activity and occupation. In order to enhance cross-national comparability of results, data were mapped to the latest revision of the International Standard Classification of Occupations (ISCO), known as ISCO-08, a hierarchical framework of titles and codes for classifying and aggregating occupational information according to similarities in skill level and skill specialization required to fulfil the tasks and duties of jobs [20].

We further conducted web-based and bibliographic searches for additional data from official national sources on human resources supply published in health sector reviews, statistical bulletins and HRH strategic plans. We did not use data published in academic journals, books or other non-official sources. Some supplementary information was gathered from the website of an international non-governmental health professional association (World Confederation for Physical Therapy) of the voluntary associations among its member organizations [22]; follow-up electronic communications were sent directly to national correspondents for data validation, with a 52% response rate. All raw HRH data were translated into densities per 10,000 inhabitants in order to enable comparisons across populations and geographies.

Estimates of population need for rehabilitation services (albeit with a focus on medical need above other quality of life dimensions) were derived from the WHO Global Burden of Disease study [23], including data on cause-specific diseases, injuries and risk factors by country and region based on the best available evidence in 2008. Data from this source were mapped to the International Classification of Diseases to enhance comparability of findings [24]. Need was measured in terms of attributable years of life lost (YLL) as related to causes that are considered to require assistance of health professionals associated to rehabilitation. This included most types of non-communicable conditions and injuries as well as certain infectious diseases, maternal and perinatal conditions, and nutritional deficiencies. In the absence of an international standard for classifying health care procedures specific to rehabilitation, the determination was based on technical advice from WHO experts in disability and rehabilitation.

Datasets were merged and basic analyses were conducted using descriptive statistics and simple regression models to compare and contrast differences in terms of the two main variables -- that is, supply of and need for human resources for rehabilitation -- within and across countries and regions. Where appropriate, coefficients of determination were calculated to estimate the goodness of fit of the regression models.

The findings from this study offer the first global portrait of supply-need dynamics for human resources for rehabilitation. Overall, and sadly not surprisingly, lower supplies of rehabilitation health professionals were found among low- and middle-income countries, including many located in sub-Saharan Africa, where the disease burden related to causes requiring rehabilitation professional skills tends to be greatest. The negative supply-need link was found to generally hold for developing countries, but the strength of the relationship varied across regions of the world. No discernable relationship was teased among the subset of countries with developed and transitional economies with official HRH data available in the public domain.

Given the wide differences observed here across countries and regions in the numbers and distribution of rehabilitation personnel, it remains uncertain whether the currently available data are sufficient to allow analysts and decision makers to draw policy-relevant conclusions. One outstanding challenge in strengthening the global evidence base is the setting of common definitions and classifications of who are rehabilitation health workers. We attempted to enhance cross-national comparability through the mapping of data following internationally standardized classifications for social and economic statistics, notably the International Standard Classification of Occupations for workforce data. However differences in the nature of national economies, health systems and information systems often make it difficult to obtain comparable data. Important discrepancies may be attributed to the number of occupations included in the original information source (and whether health workers' current practice area was actually in rehabilitation services), timeliness of the available data (or lack thereof), and differences in coverage (e.g. whether the source covers health workers in all sectors: public, private, community based services, etc.).

At the same time, health professional density is not necessarily the most important factor in improving population health and welfare. For example, Indonesia and Zimbabwe have similarly low reported numbers of professionals associated to rehabilitation (fewer than 2 physicians and physiotherapists per 10,000 population) but highly divergent disease burdens attributable to associated causes (Figure 4). Both countries have implemented community-based rehabilitation programmes in order to attempt to address service deficiencies in areas with critical shortages of physicians, nurses, and other health and social care professionals [37]. The evidence base is growing for the effectiveness of some non-traditional or alternative cadres in delivering CBR services in lower income settings [38] - although such categories of health workers may not always be adequately captured in national HRH statistics. From a monitoring and evaluation perspective, significant challenges remain in defining and measuring the available workforce with a broader skill set across countries and over time within countries.

International calls are growing for improved collection, analysis and translation of information into evidence that can be used for purposes of HRH policy, planning, programming and accountability [15, 39]. This analysis was limited by partial data availability and by heterogeneity in the information sources accessed. In order to monitor trends in health workforce situation and performance, or for countries to share experiences and best practices, it is necessary to know how health workers are defined and classified in the original information source. For example, we found systematic differences in reported supply of physiotherapists according to the nature of the national data source.

Nevertheless, we believe this study provides much needed information on the current global status of human resources for rehabilitation, and hope that it will act as a catalyst for improving the future supply of and demand for quality evidence and research on this topic. For one, it is expected that possibilities for health workforce analyses will be strengthened in the current global series of censuses, known as the 2010 round, which will largely be able to exploit the new ISCO-08 revision [21]. Understanding and strengthening health systems capacity to meet population health and rehabilitation service needs requires better information and evidence on the range of human resources for rehabilitation at the local, national and international levels. All over the world, people with disabilities have many unmet health and rehabilitation needs, yet continue to face important barriers in accessing mainstream health care services and consequently have poorer health outcomes - a double burden in low- and middle-income countries [2]. Improving the availability and use of timely, comprehensive and reliable data on the different health occupations associated to rehabilitation is the first step towards evidence-informed workforce development strategies in rehabilitation.