Health interventions cannot be carried out without health workers. Personal health interventions involving patient contacts require the services of doctors, nurses or other types of health care providers, while the preservation and promotion of health require such additional health workers as public health specialists, health educationalists and media experts. The health system also needs the services of many other categories, including planners, health economists and accountants.The statistical association between human resources for health (HRH), intervention coverage and health outcomes has recently attracted considerable attention. For example, the relationship between availability of doctors, nurses and midwives across countries and intervention coverage (the percentage of deliveries with skilled birth attendance and the proportion of children fully immunized against measles) was explored by Chen et al. (2004) and Anand & Bärninghausen (2006), who also examined the relationship with maternal, infant and under-five mortality. These analyses show that health status and levels of coverage are positively associated with health worker density, here defined as the number of health workers per 1000 population. Chen et al. go a step further and suggest that countries with fewer than 2.5 health workers per 1000 population were very unlikely to achieve minimum desirable levels of coverage (80%) for skilled birth attendants and measles immunization. These studies build on an earlier literature that had produced contradictory results: some studies had found no association between health worker density and health outcomes and outputs, while others found the opposite (Anand & Bärninghausen, 2004). Part of the explanation lies in the fact that they used different sources of data, sets of explanatory variables, analytical methods and levels of analysis (facility versus geographical unit) (Anand & Bärninghausen, 2004).The results of these studies also differed in the number of countries included in the respective analysis. For example, Anand & Bärninghausen (2004) ran regressions for 117 countries in some models and 83 in others, omitting many of the poorest from their analysis because of data limitations – particularly the lack of data on the numbers of health workers. In their subsequent study (Anand & Bärninghausen, 2006) they included even fewer—only 49—countries, this time limiting attention to those for which data on immunization coverage were available from demographic and health surveys (DHS). Data availability and quality remain a serious limitation in cross-country studies. For this analysis we undertook intensive efforts to obtain access to available data on health workers, which allowed us to include many of the poorest nations and extend the analysis to 192 countries. The purpose of this paper, therefore, is to determine whether the relationships estimated previously are robust to the inclusion of many more countries in the data set. We also explore alternative functional forms and estimation procedures and investigate the impact of controlling for additional determinants of health outcomes and outputs.
|Place of Publication||Geneva|
|Publisher||World Health Organization|
|Number of pages||14|
|Publication status||Published - 2006|
|Name||The World Health Report 2006 - working together for health|
|Publisher||World Health Organisation|