The global distribution of physicians and nurses

  • Published on

  • View

  • Download

Embed Size (px)


<ul><li><p>The global distribution of physiciansand nurses</p><p>Heather Wharrad BSc PhDLecturer, Post-graduate Division of Nursing Studies,</p><p>The Medical School, University of Nottingham</p><p>and Jane Robinson FRCN MA PhD MIPD RGN ONC RHV HVTEmeritus Professor, Post-graduate Division of Nursing Studies,</p><p>The Medical School, University of Nottingham,</p><p>Queen's Medical Centre, Nottingham, England</p><p>Accepted for publication 23 September 1998</p><p>WHARRADWHARRAD HH. &amp; ROBINSON J. (1999)&amp; ROBINSON J. (1999) Journal of Advanced Nursing 30(1), 109120</p><p>The global distribution of physicians and nurses</p><p>Aim: To explore the global distribution of physicians and nurses and the</p><p>inuence of gross national product per capita on this distribution, using</p><p>available United Nations' (UN) sources. Objectives: to compare the international</p><p>distribution of physicians and nurses by country; to examine the inuence of</p><p>gross national product per capita (GNP) on the global distribution of physicians</p><p>and nurses; to explore the assumptions underlying the recommendations of The</p><p>World Development Report 1993 Investing in Health for health workforce</p><p>substitution; and to consider the implications for future studies of global health</p><p>labour distribution. Design: A database was compiled from various UN sources</p><p>on 147 countries. Using some of the variables from this database, a general linear</p><p>regression model for log GNP per capita on each of the two dependent variables</p><p>(log nurses and log physicians per 1000 population) was produced. Standard-</p><p>ized residuals obtained from these bivariate regressions were calculated and</p><p>plotted against each other to determine the relationship between the global</p><p>distribution of physicians to population and that of nurses. From this analysis</p><p>outlying countries could also be identied. Results: Ratios of physicians to</p><p>population by country varied from 002 to 44 per 1000 population (or from 1 to227 and 150,000 population), and nurses from 003 to 164 (or from 1 to 61 and133,000 population). There was a positive correlation (r 084, P &lt; 0001)between the number of physicians per 1000 population and the number of</p><p>nurses per 1000 population. GNP explained 49% of the variation in physicians</p><p>and 40% in nurses. Ranking of countries according to their standardized</p><p>deviation from mean regression lines for GNP against health personnel in</p><p>countries with both the lowest incomes and lowest numbers of health</p><p>personnel, resulted in little change from the original rankings of ratios of</p><p>physicians and nurses relative to population. For some of the wealthiest</p><p>countries, there was a marked fall in global ranking and for some middle income</p><p>Correspondence: Jane Robinson, 8 School Road, Coalbrookdale,</p><p>Telford TF8 7DY, England. E-mail:</p><p> 1999 Blackwell Science Ltd 109</p><p>Journal of Advanced Nursing, 1999, 30(1), 109120 Health and nursing policy issues</p></li><li><p>countries a marked improvement in ranking. Conclusion: 70% of the distribu-</p><p>tion of nurses globally can be explained by the distribution of physicians, and</p><p>the inuence of GNP per capita on the global distribution of physicians and</p><p>nurses appears to be substantial. In only a minority of the world's very poorest</p><p>countries is there evidence to suggest that higher numbers of nurses substitute</p><p>for low numbers of physicians. Standardization of the distributions by GNP</p><p>demonstrates that many countries (but not the poorest) regress to within one</p><p>standard deviation of the mean expected distribution. This suggests that</p><p>countries could set optimum levels of physicians and nurses within the limits of</p><p>their GNP. More realistically, the ndings suggest that recommendations for</p><p>modication of the structure of countries' health labour forces as a component</p><p>of health care reform may be more difcult to achieve than at rst appears. The</p><p>potential unreliability of the data sources used, and the implications for the</p><p>accuracy of the ndings, are discussed.</p><p>Keywords: `Giffen Goods', global labour distribution, health economics,</p><p>health workforce planning, nurses, physicians, The World Bank, United Nations</p><p>INTRODUCTION</p><p>Factors inuencing the global distribution of health labour</p><p>have been the subject of surprisingly little study given the</p><p>assumed ability of governments to manipulate the struc-</p><p>ture of health labour forces in many of the recommended</p><p>policies for contemporary health care reform. Changing</p><p>the proportions of skill mixes, substituting less expensive</p><p>staff for more costly, and encouraging health staff to work</p><p>in areas of greatest health need (such as primary health</p><p>care), all represent strategies associated with an assumed</p><p>potential to manipulate the health labour force (Journal of</p><p>the Royal Society of Medicine 1995, Cooper 1995). Yet</p><p>there is a notable absence of any theoretical development</p><p>to justify the recommendations and a dearth of evidence</p><p>as to whether they produce the intended effects. Human</p><p>resource development in health care apparently remains a</p><p>largely pragmatic art.</p><p>Gray (1993), a health economist, referring to the</p><p>Organization for Economic Co-operation and Develop-</p><p>ment (OECD, 1991) Health Data suggested that the level of</p><p>nursing provision is solely a function of a country's</p><p>wealth:</p><p> richer countries have more nurses, poorer countries havefewer. If the gure was to be extended to the left, in order to</p><p>include countries with very low GNP per head and correspond-</p><p>ingly high levels of mortality and morbidity, we would nd that</p><p>levels of nursing were generally very low indeed.</p><p>(Gray 1993 p. 157)</p><p>Hertz et al. (1994) observed that although increases</p><p>in medical and physician facilities alone were not an</p><p>absolute solution to poor health, in the case study countries</p><p>identied in their study, health care development had</p><p>been adapted to meet their own unique socio-economic</p><p>conditions. Curative and preventive services were direct-</p><p>ed towards existing patterns of morbidity and malnutri-</p><p>tion. Thus the provision and orientation of medical</p><p>facilities were seen to be important contributory factors</p><p>within an overall context of governmental attitudes</p><p>towards social and health development.</p><p>Hertz et al.'s conclusions are reected in the approach</p><p>of the The World Development Report 1993: Investing in</p><p>Health to the interplay between health, health policy and</p><p>economic development (The World Bank 1993). The</p><p>Report's main recommendations for the improvement of</p><p>health focus on poverty reduction measures, in particular</p><p>strengthening the health of households through recogni-</p><p>tion of the strong positive correlations between: per capita</p><p>income; women's education; public health infrastructure;</p><p>and improved life expectancy (The World Bank 1993</p><p>pp. 3751). Investing in Health also comments on the need</p><p>to address the fundamental problems of the imbalance of</p><p>health personnel in nearly all countries. It observes that:</p><p>there are too many specialists and not enough primary</p><p>care providers; health workers are concentrated in urban</p><p>areas; training in public health, health policy and health</p><p>management has been relatively neglected; and medical</p><p>training is subsidized even though many physicians work</p><p>in the private sector and earn high incomes (The World</p><p>Bank 1993 p. 139).</p><p>Investing in Health recommends a number of health</p><p>workforce strategies to rectify this situation including:</p><p>forms of labour substitution for improving the balance</p><p>between primary care providers and specialists; setting</p><p>recommended ratios of physicians to population (01 per1000) and graduate nurses to physicians (between 2 and 4</p><p>to 1 physician, or 0204 per 1000 population) for</p><p>H. Wharrad and J. Robinson</p><p>110 1999 Blackwell Science Ltd, Journal of Advanced Nursing, 30(1), 109120</p></li><li><p>minimum essential clinical services to the poor; the</p><p>introduction of measures to control physician oversupply;</p><p>curtailing specialist training; attracting primary care pro-</p><p>viders to under-served areas; reforming the nance of</p><p>health training; and redirecting the content to public</p><p>health, management, policy and planning (The World</p><p>Bank 1993 pp. 139144).</p><p>Investing in Health supports strongly the idea of labour</p><p>substitution, observing that ` the relatively high ratio ofnurses to physicians in Sub-Saharan Africa is a good sign'</p><p>and stating that whilst `There is no optimal level of</p><p>physicians per capita or optimal nurse-to-physician ratio, a</p><p>rule of thumb is that nurses should exceed physicians by at</p><p>least two to one' (The World Bank 1993 p. 139). The World</p><p>Bank's emphasis on the importance of maintaining an</p><p>overview of the ratio of nurses to physicians (`ve to one in</p><p>Africa but well under two to one in China, India, Latin</p><p>America, and the Middle Eastern Crescent' (The World</p><p>Bank 1993 p. 139) led us in this paper to include graphs</p><p>showing the relative numbers of nurses to physicians, in</p><p>addition to correlations for the single occupations.</p><p>In supporting an inverse relationship between physi-</p><p>cians and nursing staff The World Bank's recommenda-</p><p>tions discount the evidence that the supply of both</p><p>physicians and nurses is associated with a country's</p><p>wealth in terms of GNP per capita. Instead, Investing in</p><p>Health endorses a view that the supply of nurses behaves</p><p>(or ideally should behave) as a `Giffen good' (Whitehead</p><p>1970) that is: consumption of a cheaper commodity (i.e.</p><p>nurses) is high when material resources are low, and that</p><p>replacement with a more expensive commodity (i.e.</p><p>physicians) takes place as available resources increase.</p><p>In its recommendations for the manipulation of health</p><p>labour, The World Bank proposes that by adopting its</p><p>recommended measures, governments will be able to</p><p>inuence positively some of the most persistent health</p><p>problems affecting the poor, such as infant, under 5, and</p><p>maternal mortality. The World Bank's position on health</p><p>labour in the World Development Report 1993 thus</p><p>mirrors in several substantial ways that adopted in World</p><p>Health Organization reports of the 1980s and 1990s on</p><p>nursing and midwifery development (for a summary see</p><p>Robinson 1997).</p><p>The World Bank's recommended measures therefore</p><p>assume that governments have both the will and the</p><p>power to inuence such events. The study reported here</p><p>investigates an alternative premise: that it is the inuence</p><p>of socio-economic factors, as mediated by GNP per capita</p><p>rather than the intentional acts of governments, which</p><p>determine the inherent structure of the health labour force</p><p>in individual countries. The study tests this assumption</p><p>using data available from The World Bank and other</p><p>United Nations' sources. The focus is on the global</p><p>distribution of health personnel as shown by the relative</p><p>numbers of physicians and nurses in relation to GNP.</p><p>THE STUDY</p><p>Materials and methods</p><p>A data base was compiled from the following sources:</p><p> Nurse to doctor ratios (19881992) from The WorldBank (1993, Table A.8);</p><p> Ratios of physicians to population (19901995) andGNP per capita (US$ 1995, from The World Bank 1997,</p><p>Table A.6);</p><p> Missing data for GNP (US$ 1994) for Cuba andAfghanistan, from United Nations' Children's Fund</p><p>(1997, Table 1);</p><p> Gross domestic product (GDP, US$ 1994) for Malta andBelize from the United Nations' Human Development</p><p>Programme (1997, Table 1).</p><p>Complete data for ratios of physicians and nurses to</p><p>population (the latter inferred from the nurse to physician</p><p>ratios), plus GNP per capita, were established from these</p><p>combined sources for 147 countries (Tables 1 and 2).</p><p>Issues concerning the reliability of these data are discussed</p><p>under `Study limitations' at the conclusion of thispaper.</p><p>Copyright permission for the use of the above data was</p><p>granted by Oxford University Press and The World Bank.</p><p>Using these data, statistical methods were used to</p><p>analyse the relationship between GNP and the global</p><p>distribution of physicians and nurses. Prior to all regres-</p><p>sion analyses, exploratory analysis was carried out to</p><p>determine the linearity of the relationship between</p><p>variables. Non-linearity between variables (which often</p><p>arose for these data) was addressed by log transformation</p><p>(Bland &amp; Altman 19961 ). The purpose of the analysis was to</p><p>determine how much of the variation in the dependent</p><p>variables could be explained by the variation in the</p><p>independent variables and also to identify outlying</p><p>countries. The sample estimate of R2 tends to be an over-</p><p>estimate of the population parameter. The R2 values</p><p>presented are the adjusted R2 which compensates for the</p><p>optimistic bias of R2. This adjusted gure takes account of</p><p>the number of variables in the model and the sample size.</p><p>Standardized residuals obtained from the bivariate re-</p><p>gressions of GNP on physicians/1000 and GNP on nurses/</p><p>1000 were calculated. These residuals were plotted to</p><p>produce a scatter-gram (physicians/1000 on the abscissa</p><p>and nurses/1000 on the ordinate axis) which was divided</p><p>into nine quadrants representing standard deviations (SDSD)</p><p>from the mean (see results section, Figure 4). The central</p><p>quadrant contains countries which lie within 1 SDSD of the</p><p>two regression lines (those for physicians and for nurses) </p><p>hence the numbers of health personnel were as predicted</p><p>by the model. Countries lying in the top right hand</p><p>quadrant were those which have high numbers of health</p><p>personnel for their level of GNP. The lower left hand</p><p>quadrant contains those countries which have fewer health</p><p>Health and nursing policy issues Global distribution of physicians and nurses</p><p> 1999 Blackwell Science Ltd, Journal of Advanced Nursing, 30(1), 109120 111</p></li><li><p>personnel than expected for their level of GNP. This plot</p><p>therefore re-distributes the countries according to the</p><p>deviations from the expected numbers of health personnel</p><p>relative to GNP.</p><p>RESULTS</p><p>Tables 1 and 2 show that ratios of physicians to popula-</p><p>tion vary from 002 to 44 per 1000 population (or from 1physician to 227 and 150,000 population). Nurses range</p><p>from 003 to 164 (or from 1 to 61 and 133,000 popula-tion). Figure 1 is a scatter-gram showing the number of</p><p>physicians per 1000 population against the number of</p><p>nurses per 1000 population for 147 countries plotted on a</p><p>log scale. The size of the positive correlation coefcient for</p><p>these indices is 084, this is highly signicant (P &lt; 0001)thus indicating that countries having low numbers of</p><p>physicians also have low numbers of nurses (Table 3). The</p><p>dotted lines superimposed on the graph represent The</p><p>World Bank's (1993) minimum recommended ratios of</p><p>physicians to population (P1 to P2) and the lower recom-</p><p>mended ratio of nurses (N1 to N2) to population.</p><p>Countries in the lower left hand segment A (Figure 1)</p><p>with lower than recommended ratios of both physicians</p><p>and nurses to population are all from The World Bank's</p><p>(1997) low-income economy group (between $80 and $730</p><p>per capita GNP in 1995). Those in the upper left hand</p><p>segment (B) with similarly low ratios of physicians but</p><p>recommended ratios of nurses are also low-income group</p><p>economies. Countries on line P1 to P2 with the recom-</p><p>mended ratio of physicians and a recommended, or</p><p>higher, ratio of nurses include further low-income</p><p>economies and three from the lower-middle-income...</p></li></ul>


View more >