forecasting the global shortage of physicians: an economic- and needs-based approach

7/27/2019 Forecasting the global shortage of physicians: an economic- and needs-based approach

1/10

516 Bulletin of the World Health Organization |July 2008, 86 (7)

Forecasting the global shortage of physicians: an economic-

and needs-based approachRichard M Scheffler,aJenny X Liu,bYohannes Kinfu c& Mario R Dal Poz d

ObjectiveGlobal achievements in health may be limited by crit ical shortages of health-care workers. To help guide workforce policy,we estimate the future demand for, need for and supply of physicians, by WHO region, to determine where likely shortages will occurby 2015, the target date of the Millennium Development Goals.MethodsUsing World Bank and WHO data on physicians per capita from 1980 to 2001 for 158 countries, we employ two modellingapproaches for estimating the future global requirement for physicians. A needs-based model determines the number of physiciansper capita required to achieve 80% coverage of live births by a skilled health-care attendant. In contrast, our economic modelidentifies the number of physicians per capita that are likely to be demanded, given each countrys economic growth. These estimatesare compared to the future supply of physicians projected by extrapolating the historical rate of increase in physicians per capita foreach country.FindingsBy 2015, the global supply of physicians appears to be in balance with projected economic demand. Because our measureof need reflects the minimum level of workforce density required to provide a basic health service that is met in all but the least

developed countries, the needs-based estimates predict a global surplus of physicians. However, on a regional basis, both modelspredict shortages for many countries in the WHO African Region in 2015, with some countries experiencing a needs-based shortage,a demand-based shortage, or both.ConclusionThe type of policy intervention needed to alleviate projected shortages, such as increasing health-care training oradopting measures to discourage migration, depends on the type of shortage projected.

Bulletin of the World Health Organization 2008;86:516523.

Une traduction en franais de ce rsum figure la fin de larticle. Al final del artculo se facilita una traduccin al espaol.

a School of Public Health and the Goldman School of Public Policy, University of California, Berkeley, CA, United States of America.b Global Center for Health Economics and Public Policy, School of Public Health, University of California, Berkeley, CA, USA.c Department of Health Statistics and Informatics, World Health Organization, Geneva, Switzerland.d Department of Human Resources for Health, World Health Organization, Geneva, Switzerland.Correspondence to Richard Scheffler (e-mail: [email protected]).doi:10.2471/BLT.07.046474(Submitted: 2 August 2007 Revised version received: 12 November 2007 Accepted: 22 November 2007 Published online: 3 April 2008)

Introduction

Te world health report 2006: workingtogether for healthhas brought renewedattention to the global human resourcesrequired to produce health.1 It esti-mated that 57 countries have an abso-lute shortage of 2.3 million physicians,nurses and midwives. Tese shortagessuggest that many countries have insuf-ficient numbers of health professionalsto deliver essential health interventions,such as skilled attendance at birth andimmunization programmes. However,these estimates do not take into account

the ability of countries to recruit and re-tain these workers, nor are they specificenough to inform policy-makers abouthow, and to what extent, health work-force investment should be channelledinto training of different professions.

Tis paper focuses on physicians,who serve a key role in health-care pro-

vision. Using the most updated infor-mation on the supply of physicians overa 20-year period, we project the sizeof the future global need for, demandfor and supply of physicians to year2015, the target date for the Millen-nium Development Goals (MDGs).2Needs-based estimates use an exog-enous health benchmark to judge theadequacy of the number of physi-cians required to meet MDG targets.Demand estimates are based on a coun-trys economic growth and the increasein health-care spending that results

from it, which primarily goes towardsworker salaries. We then compare theneeds-based and demand-based esti-mates to the projected supply of physi-cians, extrapolated based on historicaltrends. Our results point to dramaticshortages of physicians in the WHO

African Region by 2015. We provide

estimates of shortages by country inAfrica and discuss their implicationsfor different workforce policy choices.

Methods

For illustrative purposes, we providea stylized version of the conceptualframework we employed for forecast-ing physician numbers in Fig. 1. First,

we project the supply in the per capitanumber of physicians (S) based on his-torical data on physician numbers foreach country; this serves as a baselineagainst which different forecasts can be

evaluated. We employ two forecastingmethods. Te forecast for the needs-based estimate (N) is determined bycalculating the number of physiciansthat would be required to reach Teworld health report 2006goal of having80% of live births attended by a skilledhealth worker.3Te second forecasting


2/10

Research

Forecasting the global shortage of physicians

517Bulletin of the World Health Organization |July 2008, 86 (7)

Richard Scheffler et al.

method reflects the demand for physi-cians in each country as determined byeconomic growth (D1 and D2). Withthese different estimates, shortages orsurpluses can be calculated. For exam-ple, by year 8, about 3.5 physicians per

1000 population will be needed com-pared to the projected supply of 3.0 per1000, producing a 0.5 per 1000 short-age. In comparison, 4.0 per 1000

wil l be demanded according to thescenario represented by D1, resultingin a demand-based shortage of about1.0 physicians per 1000. A differentscenario can arise if supply exceedsdemand, as represented by D2, result-ing in a surplus. We can then multiplythis estimated shortage by projectedpopulation numbers to calculate the

absolute deficit of the numbers ofphysicians. In this illustrative case,the needs-based shortage exceeds thedemand-based shortage. Tis frame-

work can be applied at the country,regional and global levels of analyses,depending on the level of aggregation ofphysician numbers.

We now describe our estimationprocedures more formally. First, base-line supply projections to the year2015 were estimated using the histori-cal growth rate of physician densities ineach country. Te following regressionequation was run for each country fortime t= {1980, , 2001}:

ln(physicians per 1000 populationt) =0+ 1 yeart+ t

where t is the random disturbanceterm, and

0 and

1 are unknown

parameters to be estimated from themodel. Tis exponential growth modelassumes that current trends in the growth

of physician numbers will continue asthey have historically for each country.Te needs-based approach is based

on an sin1-log model that relates phy-sician density with coverage of skilledbirth attendants, weighted by popula-tion size. Tis model is used to identifya level of physician density below whichvirtually no country has achieved 80%coverage. While physicians may notnecessarily be the same workers whoattend live births, we can determine theoverall required number of health-care

workers to achieve the goal and thusgauge the subset number of physiciansneeded to maintain the desired levelof service coverage. Tis approach as-

Fig. 1. Conceptual forecasting framework

Physicians(per1000population)

0.00

4.50

1

S: Projected supply extrapolated based on the historical growth in physicians per 1000 population

Year

2 3 4 5 6 7 8

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

D1: Forecasted demand based on economic growth (scenario with a projected shortage)

D2: Forecasted demand based on economic growth (scenario with a projected surplus)

N: Forecasted need based on reaching benchmark level of service provision

sumes that skills mix in health servicedelivery remains constant. Te needs-based model estimates the followingequation for all countries iat time t:

sin1(% coverageit) = 0+ 1ln(physicians per 1000 population it) +i+ t+ it

where i and t reflect country andtime fixed effects, respectively, i t

is a random error term, and 0and 1are unknown parameters to be estimatedfrom the model. Tis threshold densityfigure, along with population estimatesfor future years, was subsequently usedto calculate the number of physiciansthat would be needed in each country toattain the MDG of 80% coverage of livebirths. We opted for the sin1transforma-tion because it is more consistent withstatistical theory; the transformation ofthe dependent variable, which is a pro-portion, results in normally distributed

responses (asymptotically). Te sin1

-logmodel also achieved the highest R andbest goodness-of-fit to the data as mea-sured by the deviance.4Tis approachis similar to that followed in Teworldhealth report 2006.

Te demand-based approach uti-lizes gross national income (GNI) percapita as the predictor of demand forphysicians per 1000 population, along

with country fixed effects to accountfor unobservable heterogeneity acrosscountries, weighted by population size.

Previous research has shown that in-dicators of gross domestic product ornational income are the best predic-tors of health expenditures, of which,

labour is the principle component.59Tis method has also been employed inother forecasts of physician demand.10Tis approach estimates the followingrelationship for country iat time t:

ln(physicians per 1000 populationit) =0+ 1 ln(GNI per capitait5) + 2income leveli+ i+ it

whereireflects a vector of country fixed

effects, itis the disturbance term, and 0and 1are unknown parameters to be esti-

mated from the model. GNI per capita islagged 5 years to account for time requiredfor economic growth to affect health-care spending and, in turn, influencechanges in the health-care system and the

workforce. Because GNI data were onlyavailable until 2002 at the time of dataassembly, values for 20032010 werepredicted using the historical growth rate.For each country at time t, the growth ratein GNI per capita was calculated as:

exp(1) 1

from the equation:

ln(GNI per capitat) = 0+ 1yeart+ t

where t is the disturbance term, and

0 and

1are unknown parameters to

be estimated from the model. Clas-sification of countries by income level(low, medium and high) from the

World Bank was included in the de-mand equation as countries at differ-ent levels of development may exhibitstronger or weaker relationships be-


3/10


Research

Forecasting the global shortage of physicians Richard Scheffler et al.

tween economic growth and health-care spending, depending on the com-plexities of the health-care system andstructure of local markets.

Data

Data were compiled from severalsources. Te World Bank Health, nu-trition and population database con-tains physician numbers from 1980 to2001.11GNI per capita (adjusted forpurchasing power parity using the atlasmethod) for 19752002 was obtainedfrom the World development indicatorsdatabase.12Physician numbers were up-dated with the most recent figures fromthe World health statistics 2006databaseand from the OECD health data 2005database for OECD countries.13,14His-

toric and projected population num-bers were obtained from the UnitedNations Population Division.15 Weemploy physician density (per 1000population) in our analyses to account fordifferences in health-system size and

weight all regressions by population size.Our benchmark indicator for need isthe number of live births attended bya skilled health worker, also availablefrom World health statistics 2006. Weconstructed a panel of 158 countriesfrom 1980 to 2001, grouped according

to WHO regional classifications (availableat: http://www.who.int/about/regions).

Data were missing for some coun-tries, either due to unavailability ordifferences in reporting practices. Miss-ing data points occurring between twodata points were linearly interpolated;those that were not bounded by two realdata points were not interpolated. Forfigures on physicians per capita, miss-ing data for up to 6 consecutive years

were interpolated as some countriesonly reported periodically in the 22-year

span of our data compilation; interpo-lated values comprise 35.6% of the 2819data points for physician supply num-bers used in this analysis. Because GNIper capita data are more readily avail-able, only 4.7% of the 4238 GNI percapita data points used were interpo-lated. Tis method of treating missingdata points raises concern over therobustness of our estimation results.o address this, we systematically dropcountries from our sample for whichthe number of data points for physi-

cians per capita are fewer than six (24countries), fewer than seven (41 countries),and fewer than eight (63 countries). Esti-mated coefficients from these subsamples

may differ from the main sample due tothe fact that data reporting issues are morefrequently encountered for lower-incomecountries. We carry out formal Ftests ofour economic demand model on thesedifferent subsamples in our robustnesschecks.

Results

Current distributionWe first describe the current distribu-tion of physicians by level of healthexpenditures and burden of disease,displayed in Fig. 2. Te size of thecircle represents the proportion of

world health expenditures comprisedof the countries in a given region. In2004, the WHO Americas Region reg-istered the highest proportion of the

worlds health expenditures (over 50%),but had just over 20% of the worldssupply of physicians, and only 10% ofthe global burden of disease. With asimilar level of burden of disease, the

WHO European Region has over 35%of the worlds supply of physiciansalong with about 32% of the worldshealth expenditures. In contrast, coun-tries in the WHO South-East AsiaRegion suffer the highest proportionof the global burden of disease (29%)

with only 11% of the worlds supplyof physicians and just about 1% of

world health expenditures. Similarly,

the WHO African Region experiences24% of the global burden of disease,while having only 2% of the globalphysician supply and spending that is

Fig. 2. Physician distribution by burden of disease and health expenditure, by WHOregion, 2004

Globalburdenofdiseas

e(%)

0

35

0

Global physicians (%)

5

30

25

20

15

10

5

10 15 20 25 30 35 40 45

AfricanSouth-East Asia

Western Pacific

Americas

Eastern Mediterranean

European

Size of circle reflects regional share of global expenditure

Source: World health statistics 2006.13

less than 1% of global expenditures.Clearly, there is a dramatic imbalancein the global distribution of physicians,

with countries in the WHO Africanand South-East Asia Regions currentlyfacing the largest disparities.

Forecasting results

Te results of our analysis of the futureglobal distribution of physicians are

shown in able 1. If current trendscontinue, we will have 12.7 millionphysicians supplied by 2015 glob-ally. Given the standard error of ourestimated growth rates of physicians foreach country, this supply number canrange from 11.4 million to 14.3 mil-lion (95% confidence interval). Ourneeds-based model projects the ratioof physicians required to achieve 80%coverage of live births by a skilled at-tendant to be 0.55 per 1000 popula-tion, ranging from 0.41 to 0.61 basedon a 95% confidence interval (regres-sion results in able 2 and Fig. 3, avail-able at: http://www.who.int/bulletin/volumes/86/7/07-046474/en/index.html). Note that this criterion of needreflects a relatively low level of health-service provision that is already met inmany middle- and high-income coun-tries. Consequently, this needs-basedapproach estimates that the global re-quired number of physicians to achieve80% coverage is about 3.8 million,

ranging from 3.4 million to 4.2 million.Tis implies that there will be many morephysicians in 2015 than are needed toreach the benchmark outcome. However,


4/10

Research




the issue of equitable workforce distribu-

tion remains. Even though the globalsupply of physicians exceeds that of theneed, the WHO African Region islikely to experience a shortage of physi-cians (about 167 000 in 2015) accordingto the needs-based approach.

In contrast, based on projected GNIgrowth, there will be a global demand forabout 10.8 million physicians by 2015(regression results in able 2). However,

when we account for the 95% confi-dence interval, our demand-based globalestimates of the number of physicians

suggest a balance with the projectedsupply of physicians in 2015. Further-more, countries in the WHO African,

Americas and the Eastern MediterraneanRegions will likely experience surplusesof physicians of 111 000, 888 000 and557 000 respectively in 2015. Such aprojected surplus in excess of projecteddemand indicates that future economicgrowth based on the historical growthrate since 1975 may not sufficientlyincrease health expenditures in theseareas to retain newly trained physicians.

Tis has important implications forout-migration from shortage areas, asphysicians may be attracted to areas withhigher demand and higher salaries.

Table 1. Projected supply of, need for and demand for physicians, by WHO region, 2015

WHO region Supplya Needb Demandc

(1000s)d (1000s)d Surplus (+) orshortage ()d,e

(1000s) Surplus (+) orshortage ()d,f

African 255 422 167 144 111(208329) (377469) (213 122) (131159) (96124)

Americas 2773 538 2235 1885 888(2 5093 120) (481599) (2 1742 293) (1 6202 196) (5781 153)

Eastern Mediterranean 1077 308 769 520 557(9141 297) (275343) (734802) (442612) (465635)

European 3222 480 2742 2913 309(3 0153 449) (429534) (2 6872 793) (2 6153 259) (38607)

South-East Asia 1067 987 80 910 157(9311 225) (8821 097) (30186) (7181 155) (87350)

Western Pacific 4256 1016 3240 4432 176(3 8534 883) (9071 130) (3 1263 348) (3 1076 347) (2 0921 148)

World 12 650 3 752 8 898 10 804 1 846(11 43014 303) (3 3514 171) (8 4789 299) (8 63313 728) (1 0784 017)

a Supply is projected based on the historical growth of physicians per capita in each country.b Need is assessed based on reaching the goal of 80% of live births attended by a skilled health-care worker.c Demand is projected based on economic growth (both historical and projected) of each country.d Values in parentheses are 95% confidence intervals.e Surplus (shortage) is calculated as mean projected supply minus estimated need. High and low estimates reflect the 95% confidence interval of the needs-based

estimated 80% coverage attainment level.f Surplus (shortage) is calculated as mean projected supply minus estimated demand. High and low estimates reflect the 95% confidence interval of the demand-

based estimated coefficients.

We further investigate whether

the inclusion of countries with moremissing data, and thus relatively moreinterpolated observations, affects theregression coefficients. We systemati-cally drop countries with fewer than six,seven and eight actual data points fromour full sample and re-ran our demandregression equation. In all cases, the co-efficient for the elasticity of physiciansper capita with respect to lagged GNIfor the omitted category (high-incomecountries) is significant, does not appre-ciably vary across different subsamples

and is precisely estimated. However,coefficients on dummy variables formiddle- and low-income countries dodiffer across subsamples, reflecting thefact that missing data are more likely tooccur for these countries. Nevertheless,all coefficients on these dummy vari-ables are negative and significant, con-sistently indicating that the elasticityof physicians per capita is less sensitiveto GNI per capita in middle- and low-income countries than it is in higher-income countries. Results of F tests

comparing the full-sample specifica-tion to each of the restricted subsamplesfail to reject the null hypothesis of nodifference.

able 3 compares the number of

countries that are projected to experi-ence shortages in 2015 from the needs-and demand-based models. We definea country as having a shortage if theprojected supply of physicians meetsless than 80% of the projected demandor need. Overall, we find that 45 coun-tries will have a shortage in 2015 ac-cording to the needs-based approach,the overwhelming majority of whichare located in the WHO African Re-gion. According to the demand-basedmodel, 37 countries are likely to expe-rience a shortage in 2015; 15 of thesecountries are located in the WHO

African Region, 10 in the WHO EasternMediterranean Region, and 7 in the

WHO Western Pacific Region. Teseresults for the world are graphicallydisplayed in Fig. 4. Countries that areprojected to experience demand-basedshortages are also countries that willlikely experience strong economicgrowth in the near future. For ex-ample, the demand for physicians will

increase dramatically in China if therapid economic growth experiencedin recent history continues into thenear future. Other countries, such as


5/10


Research


Indonesia, are projected to experiencea needs-based shortage only, suggestingthat even though their economies maygrow strongly enough to support thesupply of physicians, the overall work-force number will not be adequate to

meet the needs-based benchmark of80% of live births covered by a skilledattendant.

Results for Africa

In the WHO African Region, somecountries will have both a needs-basedand demand-based shortage. Tis in-dicates that, even though there is acritical need for more physicians, andprojected economic growth will de-mand a larger physician workforce, theprojected supply of physicians will not

increase sufficiently by 2015 to achievebalance. Tese countries do not havethe required capacity to train the num-bers that will be demanded by 2015,and will likely need to depend on newlyrecruited workers from abroad, possiblyfrom neighbouring countries with pooreconomic performance. able 4 (avail-able at: http://www.who.int/bulletin/volumes/86/7/07-046474/en/index.html) displays the projected need for,demand for and supply of physicians inthe year 2015 for all African countries

included in our analysis. For example,

Table 3. Number of countries with physician shortages ain 2015

WHO region Needs-based model Demand-based model

African 32 15

Americas 1 3

Eastern Mediterranean 3 2European 0 10

South-East Asia 3 0

Western Pacific 6 7

World 45 37

a Shortage is defined as having a projected supply of physicians that meets less than 80% of the forecasteddemand or need, calculated at estimated means.

Fig. 4. Physician shortages in 2015 based on demand and need models

Demand-based shortage

Needs-based shortageDemand- and needs-based shortage

No shortage

Data not available for analysis

Kenya will need 24 000 physicians onaverage in 2015 (a shortage of about18 000 physicians) and the economy

will be strong enough to demand 7600

physicians on average, but the supplyin Kenya is only projected to reach anaverage of 6100 physicians. Algeria,on the other hand, will have an amplesupply of 97 000 physicians, a surplusof almost 77 000 physicians beyond

what is needed and a surplus of 67 000physicians beyond what will be de-manded. In Ethiopia, we see anotherscenario where 53 000 physicians willbe needed by 2015, but the economy

will only demand about 3000 physi-cians, even though the supply of physi-

cians is projected to reach about 5000.

Tese scenarios suggest that futuremigration of physicians could take anincreasingly regional dimension.

Discussion

Our projections suggest that, by theyear 2015, the global supply of physi-cians will be roughly in balance withdemand, while a significant surplus

will arise according to the needs-basedmodel. However, regional trends showthat distributional problems will likelypersist. More than any other regionof the world, Africa will likely experi-ence most of the physician shortages in2015. Given the disproportionate bur-

den of disease in this region, policies


6/10

Research




Rsum

Prvoir la pnurie mondiale de mdecins : une dmarche conomique fonde sur les besoins

ObjectifLes russites sanitaires dans le monde peuvent se heurter des pnuries critiques de personnel de sant. Pour guiderles politiques en matire de main doeuvre, nous avons estimla demande future, les besoins et la disponibilit en mdecins parrgion de lOMS afin de dterminer les endroits qui souffrirontprobablement dune pnurie de personnel dici 2015, datefixe pour la ralisation des objectifs du Millnaire pour ledveloppement.MthodesEn utilisant des donnes de la Banque mondiale et delOMS sur le nombre de mdecins par habitant de 1980 2001 dans158 pays, nous avons appliqu deux dmarches par modlisationpour estimer les besoins futurs en mdecins dans le monde. Un

modle ax sur les besoins indique le nombre de mdecins parhabitant ncessaire pour que 80 % des naissances vivantes soient

assistes par du personnel de sant qualifi. Par ailleurs, notremodle conomique dtermine le nombre de mdecins par habitantqui correspondra probablement la demande, compte tenu dela croissance conomique, pour chaque pays. Ces estimationssont compares loffre future de mdecins, projection obtenuepar extrapolation du taux historique daugmentation du nombrede mdecins par habitant dans chaque pays.Rsultats Dici 2015, il semble qu lchelle mondiale, loffrede mdecins parvienne quilibrer la demande conomiqueprojete. Notre mesure des besoins refltant la densit minimalede personnel de sant ncessaire pour assurer les services desant de base, densit atteinte dans tous les pays sauf les moins

dvelopps, les estimations fondes sur les besoins prdisent unsurplus mondial de mdecins. Nanmoins, au niveau rgional, les

for increasing the supply of physiciansare urgently needed to stem projectedshortages. According to our needs-based target of 80% coverage of livebirths by a skilled attendant, a 65%increase in the physician supply in

the WHO African Region will be re-quired compared to an increase of 44%of physicians to equilibrate demand

with supply. Tese enormous increaseswill requi re significant inc rea ses inhealth-care spending and active policyintervention.

Policy implications

Te type of policy intervention pur-sued will depend on the type of short-age likely to be experienced. Given thedifficulty of redistributing physicians

across country borders, countries thatmay face only a needs-based shortagemay want to consider expanding medi-cal training programmes. For countriesthat face a demand-based shortageonly, out-migration may be a particularconcern, suggesting policies gearedtowards retention. Countries that faceboth demand-based and needs-basedshortages may prefer a mixture of train-ing and recruitment policies. Govern-ment and donor organizations should

consider increasing financial support ofhealth-care workers as a means of im-proving recruitment and retention.

Te exact nature and extent of anypolicy intervention adopted will de-pend crucially on the characteristicsof each countrys health-care systemand institutions. While this analysisprovides a direction for where policies

should be targeted, such cross-countrycomparisons cannot fully account forthese complexities as well as otheraspects of distribution (e.g. physicianspecialty, race/ethnicity), practice styles(e.g. work hours) and trends in the de-

mographic characteristics of the work-force supply (e.g. ageing, gender mix),which have also been found to haveimportant effects on health-care ser-vice delivery and access.1618Moreover,health-worker mix is another criticalfactor in health-services production;further work is required in the areaof predicting the future numbers ofnurses, midwives and other ancillaryhealth-care workers who will be neededand demanded. As the WHO reportshows, nurses and other health work-

ers can help to make the clinical workmore productive, particularly in certainpatient-care services where there areskill overlaps.19Te use of telemedicinemay also have merit as a cost-effective

workforce model.20,21Limited resourcesclearly point to merits of these ap-proaches.

Limitations and future directions

While we have sought to provide someindication of the numbers of physi-

cians that will likely be required inthe future, some caution is warrantedin interpreting these figures. First, ourcriterion of need only reflects one aspectof health-care delivery; thus differentnumbers of physicians will be requiredto meet alternative normative criteriafor health services. Second, our projec-tions of demand and supply both rely

on trends in either economic growth orphysicians per capita, each of which arecontinually being affected by policy in-tervention. Comparable cross-countrydata along these dimensions are cur-rently unavailable and consequently

cannot be fully accounted for in thistype of forecasting. Moreover, projec-tions of demand will also likely beaffected by other factors other thaneconomic growth, suggesting direc-tions for further work in refining suchdemand models. Given these limita-tions, we nevertheless believe that thisexercise has been useful and informa-tive for providing an overall sense of

where physician workforce pol iciesshould be undertaken.

AcknowledgementsWe thank eh-wei Hu, Ray Catalanoand imothy Brown for their techni-cal expertise and advice. Steeve Ebenerand his team, including Yaniss Guigoz,at WHO also provided assistance

with mapping tools. We also thankthe numerous seminar participants atthe London School of Economics, theLondon School of Hygiene and ropi-cal Medicine, the International Health

Workforce Conference, the Advanced

Health Leadership Forum and theInternational Health Economics As-sociation for their comments and feed-back. Members of the Department forHuman Resources for Health at WHOprovided logistical, administrative andtechnical support for this study.

Competing interests:None declared.


7/10


Research


ResumenPrevisin de la escasez mundial de mdicos mediante un modelo econmico y basado en las necesidades

ObjetivoLos logros mundiales en materia de salud pueden verselimitados por situaciones crticas de escasez de trabajadoressanitarios. Como orientacin para la formulacin de polticasreferentes a la fuerza laboral, estimamos la demanda, lasnecesidades y la oferta futuras de mdicos por regiones de laOMS a fin de determinar dnde es probable que se produzcansituaciones de escasez en 2015, fecha clave de los Objetivos deDesarrollo del Milenio.Mtodos Utilizando datos del Banco Mundial y de la OMSsobre los mdicos por habitante entre 1980 y 2001 en 158pases, aplicamos dos tipos de modelizacin para estimar lasnecesidades mundiales de mdicos en el futuro. Un modelobasado en las necesidades determina el nmero de mdicospor habitante necesarios para lograr una cobertura del 80% delos nacidos vivos atendidos por parteras cualificadas. Nuestromodelo econmico, en cambio, determina la demanda probablede mdicos por habitante, teniendo en cuenta el crecimientoeconmico de cada pas. Esas estimaciones se comparan con

la oferta futura de mdicos prevista extrapolando la tendenciahistrica de aumento del nmero de mdicos por habitante paracada pas.ResultadosPara 2015, la oferta mundial de mdicos pareceestar en consonancia con la demanda econmica prevista. Comonuestra medida de la necesidad refleja la densidad mnima detrabajadores requerida para dispensar un servicio bsico desalud que slo los pases menos adelantados no consiguenproporcionar, las estimaciones basadas en las necesidadespermiten pronosticar un excedente mundial de mdicos. Sinembargo, a nivel regional, los dos modelos prevn escasecesen muchos pases de la Regin de frica de la OMS en 2015, yalgunos pases podran sufrir una escasez segn las necesidades,una escasez segn la demanda, o ambas cosas.Conclusin El tipo de intervencin normativa necesaria paramitigar esas escaseces, como una creciente formacin sanitaria ola adopcin de medidas para desincentivar la migracin, dependedel tipo de escasez prevista.

deux modles prdisent des pnuries pour de nombreux pays dela Rgion africaine de lOMS dici 2015, certains pays subissantune pnurie par rapport aux besoins, dautres par rapport lademande, voire les deux.

Conclusion La nature des interventions politiques ncessairespour attnuer les pnuries projetes, par exemple la formationdun plus grand nombre de personnes ou ladoption de mesurespour dcourager lmigration, dpend du type de pnurie prvu.

:

: . 2015

. : 158 2001 1980 : %80 .

. 2015 : . . 2015

. :

. References

The world health report 2006: working together for health1. . Geneva: WHO; 2006.The millennium development goals report 20052. . New York, NY: UnitedNations; 2005.Chen L, Evans T, Anand S, Boufford JI, Brown H, Chowdhury M, et al. Human3.resources for health: overcoming the crisis. Lancet2004;364:1984-90.PMID:15567015doi:10.1016/S0140-6736(04)17482-5Zar4. JH. Biostatistical analysis. Upper Saddle River, NJ: Prentice-Hall; 1996.

Cooper5. RA, Getzen TE, Laud P. Economic expansion is a major determinantof physician supply and utilization. Health Serv Res2003;38:675-96.PMID:12785567doi:10.1111/1475-6773.00139

Scheffler6. RM. Health expenditures and economic growth: an internationalperspective[Occasional Papers on Globalization]. University of South FloridaGlobalization Research Center. Nov 2004:1.10.Getzen7. TE. Macro forecasting of national health expenditures. In: Rossiter L,Scheffler RM, eds.Advances in health economics11:27-48. Greenwood, CN:JAI Press;1990.Newhouse8. JP. Medical care expenditure: a cross-national survey.

J Hum Resour1977;12:115-25 10.2307/145602. PMID:404354doi:10.2307/145602
http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15567015&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(04)17482-5http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=12785567&dopt=Abstracthttp://dx.doi.org/10.1111/1475-6773.00139http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=404354&dopt=Abstracthttp://dx.doi.org/10.2307/145602http://dx.doi.org/10.2307/145602http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=404354&dopt=Abstracthttp://dx.doi.org/10.1111/1475-6773.00139http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=12785567&dopt=Abstracthttp://dx.doi.org/10.1016/S0140-6736(04)17482-5http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=15567015&dopt=Abstract


8/10

Research




Pfaff9. M. Differences in health care spending across countries: statisticalevidence. [Medline]. J Health Polit Policy Law1990;15:1-67. PMID:2108198doi:10.1215/03616878-15-1-1Scheffler10. RM. Is there a doctor in the house? market signals and tomorrowssupply of doctors. Stanford, CA: Stanford University Press; 2008.Stats11. HNP [online database]. Washington, DC: World Bank; 2006. Availablefrom: http://devdata.worldbank.org/hnpstats[accessed on 3 April 2008].World Development Indicators (WDI) database12. [online database]. Washington,DC: World Bank; 2006. Available from: http://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICS[accessed on 3 April 2008].World health statistics 200613. . Geneva: WHO; 2006.Health data OECD, 200514. . Paris: OECD, 2005.World population prospects: the 2004 revision population database15. [onlinedatabase]. New York, NY: United Nations Population Division; 2005. Availablefrom: http://esa.un.org/unpp/[accessed on 3 April 2008].Brown TT, Coffman JM, Quinn BC, Scheffler RM, Schwalm DD. Do physicians16.always flee from HMOs? New results using dynamic panel estimationmethods. Health Serv Res2006;41:357-73. PMID:16584453doi:10.1111/

j.1475-6773.2005.00485.x

Brown17. TT, Scheffler RM, Tom SE, Schulman KA. Does the market value racial/ethnic concordance in physician-patient relationships? Health Serv Res2007;42:706-26. PMID:17362214doi:10.1111/j.1475-6773.2006.00634.xLiu18. JX, Brown TT, Scheffler RM.Are there enough minority physicians inCalifornia to go around?CPAC policy briefing, Health research innovation:CPAC reports. Sacramento, CA; 2006.Scheffler19. RM, Waitzman NJ, Hillman JM. The productivity of physicianassistants and nurse practitioners and health work force policy in the era ofmanaged health care. J Allied Health1996;25:207-17. PMID:8884433TeleHealth20. national program in support of the primary health care, Brazil.2007. Available from: http:/www.telessaudebrasil.org.br [accessed on 3 April2008].Vassallo DJ, Swinfen P, Swinfen R, Wootton R. Experience with a low-cost21.telemedicine system in three developing countries. J Telemed Telecare2001;7:56-8. PMID:11576493doi:10.1258/1357633011936732
http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=2108198&dopt=Abstracthttp://dx.doi.org/10.1215/03616878-15-1-1http://devdata.worldbank.org/hnpstatshttp://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICShttp://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICShttp://esa.un.org/unpp/http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16584453&dopt=Abstracthttp://dx.doi.org/10.1111/j.1475-6773.2005.00485.xhttp://dx.doi.org/10.1111/j.1475-6773.2005.00485.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17362214&dopt=Abstracthttp://dx.doi.org/10.1111/j.1475-6773.2006.00634.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=8884433&dopt=Abstracthttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=11576493&dopt=Abstracthttp://dx.doi.org/10.1258/1357633011936732http://dx.doi.org/10.1258/1357633011936732http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=11576493&dopt=Abstracthttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=8884433&dopt=Abstracthttp://dx.doi.org/10.1111/j.1475-6773.2006.00634.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=17362214&dopt=Abstracthttp://dx.doi.org/10.1111/j.1475-6773.2005.00485.xhttp://dx.doi.org/10.1111/j.1475-6773.2005.00485.xhttp://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=16584453&dopt=Abstracthttp://esa.un.org/unpp/http://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICShttp://web.worldbank.org/WBSITE/EXTERNAL/DATASTATISTICShttp://devdata.worldbank.org/hnpstatshttp://dx.doi.org/10.1215/03616878-15-1-1http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Retrieve&db=PubMed&list_uids=2108198&dopt=Abstract


9/10

Research


ABulletin of the World Health Organization |July 2008, 86 (7)


Table 2. Ordinary least squares regression results of needs-based and demand-based modelsa

Needs-based model:% live births attended by a skilled physicianb

Demand-based model:logged physicians per 1000

Estimated coefficient Robust standard error Estimated coefficient Robust standard error

Logged physicians per 1000 0.204c

0.010Logged GNP per capita (t5) 0.237 c 0.074Middle-income country 0.568 c 0.128Low-income country 1.565 c 0.182

Country fixed effects Yes Yes

N 409 2 513R 0.974Wald-stat (P> ) 446.52 0.000

F-stat (P-value) 74.15 0.000

GNP, gross national product.aFull sample n= 158 countries. Standard errors are clustered at the country level. Regressions are weighted by population size.b Sin1transformation applied.c Significant at 1% level.Data source: World development indicators database,12World health statistics 2006,13Health nutrition population statistics, OECD health data 2005,142004 revisionpopulation database.15

Fig. 3. Needs-based estimate of the number of physicians per capita required toachieve 80% coverage of live births by a skilled attendant

Coverage(%)

0

100

0

Density (physicians per 1000 population)

80

60

40

20

(0.49)Lower bound

(0.61)Upper bound

Threshold estimate (0.55)

1


10/10

B Bulletin of the World Health Organization |July 2008, 86 (7)

Research


Table 4. Projected supply of, need for and demand for physicians in 2015 for countries in the WHO African Region

Country Supply Needs-based model Demand-based model

(count) (count) Surplus (+) orshortage ()a

(count) Surplus (+) orshortage ()b

Algeria 97 739 20 896 76 843 30 682 67 057Angola 2 593 11 493 8 900 1 309 1 284Benin 490 6 154 5 665 905 415Botswana 975 927 48 521 454Burkina Faso 1 197 9 699 8 503 679 517Burundi 449 5 825 5 376 628 179Cameroon 822 10 447 9 625 1 539 717Cape Verde 105 345 240 242 137Central African Republic 201 2 550 2 349 210 10Comoros 322 559 237 142 180Congo 3 115 2 985 130 1 511 1 605Cte dIvoire 2 975 11 825 8 850 1 988 987Equatorial Guinea 86 344 258 296 209Ethiopia 4 916 53 306 48 390 2 967 1 949Gabon 188 881 692 647 459Gambia 23 1 036 1 014 76 54Ghana 3 414 14 574 11 159 2 097 1 317Guinea 1 552 6 524 4 972 1 703 151Guinea-Bissau 504 1 170 666 393 111Kenya 6 105 24 248 18 143 7 602 1 497Lesotho 51 957 906 118 67Madagascar 2 376 13 065 10 690 2 916 540Malawi 644 8 778 8 134 413 231Mali 890 9 927 9 037 1 038 148Mauritania 599 2 188 1 589 392 207

Mauritius 2 612 737 1 875 1 572 1 040Mozambique 277 12 901 12 624 683 407Namibia 2 623 1 233 1 389 756 1 867Niger 802 10 580 9 778 507 295Nigeria 72 123 88 298 16 176 36 580 35 543Rwanda 395 6 179 5 784 656 261Senegal 1 819 7 977 6 158 1 396 423Sierra Leone 759 3 784 3 025 424 335South Africa 35 882 26 282 9 599 33 543 2 338Swaziland 537 544 8 168 369Togo 433 4 305 3 873 720 288Uganda 2 282 22 999 20 717 2 315 33Zambia 433 7 594 7 162 1 518 1 086Zimbabwe 1 659 7 574 5 915 2 363 705

a Surplus (shortage) is calculated as supply minus need at means of predicted values.b Surplus (shortage) is calculated as supply minus demand at means of predicted values.

forecasting the global shortage of physicians: an economic- and needs-based approach

Documents