SciELO - Scientific Electronic Library Online

vol.43 issue3Technological profile assessment of voluntary HIV counseling and testing centers in BrazilEquity and provision of public dental services in the State of Paraná, Southern Brazil author indexsubject indexarticles search
Home Page  

Services on Demand




Related links


Revista de Saúde Pública

On-line version ISSN 1518-8787Print version ISSN 0034-8910

Rev. Saúde Pública vol.43 n.3 São Paulo May./Jun. 2009  Epub Apr 10, 2009 



Impact of the funding reform of teaching hospitals in Brazil


Impacto de la reforma del financiamiento de hospitales de enseñanza en Brasil



Lobo MSCI; Silva ACMII; Lins MPEII; Fiszman RI

IServiço de Epidemiologia e Avaliação. Hospital Universitário Clementino Fraga Filho. Universidade Federal do Rio de Janeiro (UFRJ). Rio de Janeiro, RJ, Brasil
IIInstituto Alberto Luiz Coimbra de Pós-Graduação e Pesquisa em Engenharia. COPPE. UFRJ. Rio de Janeiro, RJ, Brasil





OBJECTIVE: To assess the impact of funding reform on the productivity of teaching hospitals.
METHODS: Based on the Information System of Federal University Hospitals of Brazil, 2003 and 2006 efficiency and productivity were measured using frontier methods with a linear programming technique, data envelopment analysis, and input-oriented variable returns to scale model. The Malmquist index was calculated to detect changes during the study period: "technical efficiency change," or the relative variation of the efficiency of each unit; and "technological change" after frontier shift.
RESULTS: There was 51% mean budget increase and improvement of technical efficiency of teaching hospitals (previously 11, 17 hospitals reached the empirical efficiency frontier) but the same was not seen for the technology frontier. Data envelopment analysis set benchmark scores for each inefficient unit (before and after reform) and there was a positive correlation between technical efficiency and teaching intensity and dedication.
CONCLUSIONS: The reform promoted management improvements but there is a need of further follow-up to assess the effectiveness of funding changes.

Descriptors: Hospitals, Teaching, organization & administration. Costs and Cost Analysis. Rates, Ratios and Proportions, methods. Financial Management, Hospital. Economics, Hospital. Efficiency, Organizational.


OBJETIVO: Evaluar el impacto de la reforma de financiamiento en la productividad de hospitales de enseñanza.
MÉTODOS: A partir del Sistema de Informaciones de los Hospitales Universitarios Federales de Brasil, se construyeron fronteras de eficiencia y productividad en 2003 y 2006 con técnicas de programación linear, por medio de análisis envoltorio de datos, considerando retornos variables de escala y orientación a input. Se calculó el Índice de Malmquist para identificar cambios de desempeño a lo largo de los años con relación a la eficiencia técnica (cociente entre los puntajes de eficiencia en tiempos distintos) y eficiencia tecnológica (desplazamiento de la frontera en el período considerado).
RESULTADOS: Hubo aumento del aporte financiero en 51% y de la eficiencia técnica de los hospitales de enseñanza (de 11, pasaron a ser 17 en la frontera empírica de eficiencia), no ocurriendo el mismo con la frontera tecnológica. El uso del análisis envoltorio de datos estableció los benchmarks para las unidades ineficientes (antes y después de la reforma) y los puntajes de eficiencia mostraron una posible correlación entre la eficiencia técnica encontrada y la intensidad y dedicación de enseñanza.
CONCLUSIONES: La reforma permitió el desarrollo de mejoras gerenciales, pero es necesario mayor tiempo de acompañamiento para observar cambios más efectivos del modelo de financiamiento.

Descriptores: Hospitales Escuela, organización & administración. Costos y Análisis de Costo. Tasas, Razones y Proporciones, métodos. Administración Financiera de Hospitales. Economía Hospitalaria. Eficiencia Organizacional.




University and teaching hospitals (UTH) play a significant role in Brazil's health service, accounting for 10% of the beds, 12% of hospital admissions and outpatient treatment, 26% of the intensive care unit (ICU) beds and 38% of the highly complex procedures offered by the National Health System (SUS).11

Even so, over the last few years the crisis in Brazilian hospitals has been in the headlines and in particular the crisis in teaching hospitals, which involves elements of funding and management, implying a drop in the performance of their health care, teaching and research models.

As a first strategy for improving the financial situation an Inter-institutional Commission was set in 2003, with the participation of the Ministries of Health, Education, Science and Technology and Planning and presided over by the former. The objective of the Commission was to assess the situation of the UTHs in order to redirect national policy toward the sector.ª This assessment showed the need for greater integration between teaching hospitals and the local health systems. It also showed that hospitals were suffering from a shortage of funds and lacked improvements in their resource management capacity.

Following this initial diagnosis instruments were defined and requirements established for certifying Brazilian teaching hospitals, which was based on the integration of teaching and health care, participation in the SUS service network and maintenance of managerial and organizational mechanisms.b Once certified, the hospitals were included in SUS Teaching Hospital Restructuring Program and started the agreement process with the respective health managers of health service indicators and those related to teaching, research and the technological assessment of health system-related needs. The funding mechanism of the UTHs was contractually and definitively altered. It is now based on an overall budget for medium complexity procedures according to compliance with the agreed targets. Highly complex procedures are still paid based on production.9

The new funding that is based on a fixed budget and compliance with agreed targets had three guidelines: a) adjustment of the demand on the health system to the supply of heath services, the training and development of human resources and research; b) an increase in funds - the main demand of unit directors; c) a guarantee that efforts would be made to develop administrative capacity and local management - the general opinion of the federal bodies responsible for passing on funds.

Considering that one of the main reasons for the reform consisted in the need for increasing the efficiency of the health service provided by the teaching hospitals, the objective of this study was to assess the impact of financial reform on the productivity of teaching hospitals. An attempt was also made to discriminate the technological component of the change in the mechanism for transferring funds, as an instrument of national public policy, from the component that is inherent in the capacity of each unit for managing the new fund allocation method.



The measure of efficiency in data envelopment analysis (DEA), based on linear programming, is done by comparing a set of similar units, called decision-making units (DMU), which consume the same inputs for producing the same outputs. The only difference between them is the amount consumed and produced. So a DMU will be efficient if, when compared with the others, it has more output for a fixed amount of input and/or uses less input to generate a fixed amount of output.

By comparing DMUs with best practices, DEA constructs an efficient empirical production frontier. The DMUs that are at the frontier, and therefore efficient, have an efficiency measure of 100%, while DMUs located below the frontier are inefficient. The efficiency measure of inefficient DMUs can be calculated by the "radial" distance from the frontier, known as the efficiency technique. This indicates the proportion by which inputs must be reduced or outputs increased to reach the efficient frontier.5

The objective of the DEA is also to indicate where inefficiencies in the DMUs arise and present a benchmark so that inefficient units can reach the frontier of technical efficiency.

The production model used was the BCC DEA (named after the authors Banker, Charnes and Cooper1), oriented to input reduction for projection at the frontier. The BCC DEA linear programming model (equation 1) considers variable returns to scale, because some studies have already indicated the existence of economies of scale in hospital efficiency.13,16 The use of variable returns to scale allows an efficient unit to be compared only with production units of similar size on a similar scale.3 This aspect is relevant when it is intended to use benchmark knowledge in the agreement made between the manager and the health units.

Likewise, the choice of input orientation is due to the fact that hospital managers have greater control over inputs than outputs. In most countries the emphasis in health administration is on greater cost control, without compromising quality, rather than on prioritizing increase in demand. This explains why most of the studies that use DEA for measuring hospital efficiency are input-oriented, as we have observed in literature review.14

The Malmquist index measures the variation in productivity in different periods of time. This index was developed by Caves et al,2 inspired in the work of Malmquist.12 When two periods of time are compared, two different frontiers are observed: that of the initial period (t) and that of the final period (t + 1). It is possible, therefore, to obtain performance indices according to the reference technology (frontier) used (before and after).

The Malmquist-DEA method, proposed by Färe et al,4 has been the most widely used for calculating the measurement. The method applies the DEA linear programming algorithm for constructing the production frontier in both the periods analyzed. It considers the distance from each hospital or DMU that is being looked at (DMU0), before and after, the two different frontiers. The Malmquist-DEA index (Mo) is calculated from the geometric average of two indices, in which the first uses the period t frontier as a reference and the second the period t+1 frontier. A value for Mo greater than 1 indicates growth or evolution in the total productivity of the factors (TPF) between periods t and t+1, while a value less than 1 indicates a decline.

The input-oriented Malmquist index is given by:


corresponds to the technical efficiency measure of the DMU0 in period t, obtained by observing all DMUs in period t, in other words,

corresponds to the technical efficiency measure of the DMU0 in period t+1, obtained by observing all DMUs in the period t+1, in other words,

corresponds to the technical efficiency measure of the DMU0 obtained by substituting the data of the DMU0 in period t by those from period t+1, provided that the observations of the other DMUs was carried out in period t,

corresponds to the technical efficiency measure of the DMU0 obtained by substituting the data from the DMU0 in period t+1 by those from period t, provided that the observations of the other DMUs was carried out in period t+1.

In its decomposed form, the Malmquist index allows us to separate the sources of variation in productivity into two terms:

In this decomposition the ratio outside the brackets measures the change in the relative efficiency of the unit observed and is called catch-up. This ratio indicates if the unit's production is getting closer to or moving further away from the frontier. With regard to this, the geometric mean of the two ratios within brackets measures the change in the technological frontier in the two periods assessed in xt and xt+1 and is called the "frontier shift". These partial indices are important because they allow us to see if an increase in efficiency between one period and another should be attributed to an increase in productivity in the assessed unit or a contraction in the production frontier.

The technical change (catch-up) represents an increase in efficiency arising from a larger injection of financial resources (ratio between the efficiency score in 2006 and 2003, regardless of the behavior of the frontier). Technological change arises directly from the frontier shift, in this case the introduction of a new management model. There is, therefore, an equivalence in the terms of the index between the mathematical formula and the vision of those who are heavily involved with the reform, whether they are unit directors (who emphasize the funding crisis) or managers responsible for the funding (who are trying to face up to the crisis in management).

The data we analyzed come from the Ministry of Education's Federal University Hospitals Information System (SIHUF/MEC), and relate to the second half of 2003 and the second half of 2006, before and after the contracting reform, therefore. In order to guarantee that the group was homogenous group for comparison purposes, specialty and maternity hospitals were excluded. The 30 general hospitals of the MEC correspond to almost 20% of Brazil's teaching hospitals. They are represented by the initials of their respective universities.

Following the model proposed by Ozcan,15 as input variables we used: operating expenses (a figure based on the whole of the monthly income from SUS funds, before and after the target contract, because this is the main source of funding of these hospitals); labor force (total number of doctors and non-doctors), number of beds and service mix. According to the author, the two last variables are satisfactory proxies for the volume of capital of the hospital (Pearson correlation= 0.926). Service mix denotes the total number of diagnostic and therapeutic services offered by the unit and, therefore, the complexity of the services.

Outputs were considered to be: admissions, surgical procedures and outpatient consultations, all adjusted according to the hospital's complexity index. Also as a proxy measure of the case mix, the SIHUF includes a continuous variable called the Highly Complex Procedures Information System (SIPAC), based on the volume of funds necessary for carrying out highly complex procedures, as certified by the Ministry of Health. These include, among other things, neurosurgery, cardiac and orthopedic surgery and transplants. For the adjustment, each individual SIPAC value was divided by the mean of the amounts for all hospitals, thereby creating a SIPAC index, which was then multiplied by each output variable.10

Finally, to assess the teaching dimension of the hospitals the following indicators were used, in accordance with Grosskopf et al:7 teaching intensity (residents/bed ratio) and teaching dedication (residents/doctor ratio).



With the funding reform, the revenue for teaching hospitals coming from the SUS started to include: the contractual amounts of medium complexity, the amounts they received for carrying out highly complex procedures, and incentive amounts (such as those coming from contracting, inclusion in the emergency system, the offer of ICU, and others). This implied a 51% average increase in revenues (Table 1), without these particular hospitals having any corresponding increase in the number of admissions and adjusted surgical procedures (a drop of 6% and 4%, respectively). There was, however, an increase in consultations with external patients (44%). There were no significant changes in the other input variable during the period, such as the number of beds and employees or the service mix.

Table 2 shows the relative efficiencies of the hospitals in 2003 and 2006, according to the BCC-Input frontiers. In 2003 there were 11 efficient hospitals (five of them in the most productive region on the scale - MPRS). In 2006 this number reached 17 units (11 on the MPRS). In the same period the average efficiency increased from 89.6% to 91.7%. If we consider the frontiers separately, the relative efficiency only reduced in six units - FURG, UFBA, UFCE, UFRN, UnB and UNI-RIO - despite an increase in revenue in the period by 40%, 100%, 136%, 86%, 54% and 10% respectively. Of these units the only one without a drop in any of its production/output categories was UFBA (Table 1).



Only the UFGO had a Malmquist Index measure value that was greater than 1.0 (the number of surgical procedures increased by 23%, admissions by 10% and consultations by 89%). On the other hand, analysis of the decomposed Malmquist Index shows that there was an increase in technical efficiency (catch-up) or it remained the same in 23 of the hospitals. Among those that were efficient in 2003, only UnB moved away from the efficient frontier. With regard to technological efficiency, all the values were less than 1.0, which means the frontier contracted. The unviable aspects of the model for UFU and UNIFESP arose from the fact that the geometric location of 2006 was above the 2003 frontier, therefore, outside the production possibility set (PPS) for that period.

Table 3 shows the relationship of the reference hospitals for each inefficient unit for both the periods studied. Even when measured by mathematical modeling, i.e., by calculating the shortest distance to the frontier in which the reference hospitals become the vertices of a polyhedron of the empirical frontier, the model guaranteed similarities between the two: size, geographic location, characteristics and services provided. To reach the frontier, inefficient hospitals should study the absolute values of the variables of their benchmarks, so that possible changes can be made.



Considering the reduction in inputs necessary for all the hospitals to reach the frontier in 2003 this was 13% (7,299) of all employees, 19% (1,448) of all doctors, 19% (1,651) of all beds, 25% of the service mix and 8.3% or R$ 4,600,000 in the operating expenses. In 2006, the necessary decrease would be: 9% of all employees (5,060), 16% of all doctors (1,251), 9% of all beds (740) and 17% of the service mix on offer. As for operating expenses after the reform, the monthly amount of R$ 73.8 million could be reduced by 7.7% or R$ 5,700,000 without harming the performance of these hospitals. These data refer to the items to be considered in the target contracts, by hospital. They also show the funds to be liberated at the federal level.

Table 4 uses the mean and the median of the teaching intensity and dedication indicators for efficient and inefficient hospitals. It shows a trend in efficient hospitals, as indicated by the variables of the health service dimension. They were also more heavily involved with teaching activities.




The Malmquist Index, calculated using DEA, has been used to evaluate reforms in hospital funding systems, such as occurred in Finland and Austria in 1993 and 1997, respectively.8,17 One of the main advantages of applying the DEA for identifying the efficiency frontier lies in the fact that productivity change measures can be analyzed using two different components: "technology changes" (resulting from shifting the frontier in time) and "changes in technical efficiency" (comparative performance of the unit analyzed). In these studies, the results were monitored for at least four consecutive years. In both cases there was a significant improvement in the technological frontier in years following the reform, but without any major changes in the technical efficiency of the hospitals.

On the other hand, we observed an evolving improvement in the technical efficiency of teaching hospitals, with a 55% increase in the number of units located on the efficiency frontier after receiving the major injection of funds. But the same did not happen with technological change, which we understood as being the new model based on the fixed budgetary management of hospitals. It is too early, however, to draw any negative conclusions from the diagnosis of the proposed reform, since it is still being implemented. Likewise, the management contract, like the agreement instrument, is still in the development phase, specially for teaching and research indicators. This study, therefore, needs monitoring for a longer period so that a conclusion can be reached as to whether the new funding model is working well or not. Since the model only considered hospitals managed by the MEC, other studies must look at all the various types of teaching hospital to check if these results are also repeated in them.

Given the seriousness of the funding crisis in the period prior to the reform the increase in funds led to a minimization in the distance shift and/or a contraction in the frontier and financing relief for teaching hospitals, which even had different strategies for managing productivity (seen in the change in the number of admissions, surgical procedures and outpatient consultations in Table 1). The capabilities of the local management should be studied in those hospitals where despite an increase in funds there was still a reduction in technical efficiency. In such cases, an evaluation of the reference hospitals, as shown in Table 3, may be a viable way for attempting to improve their management. This information has been useful for the units themselves, since it provides them with guidance as to their own benchmarks. It is also useful for the regulatory body since it identifies possible targets to be agreed and because it creates adjustment parameters for the funds needed for the expected outputs.9

The efficiency of health service productivity may also be associated with the quality of the teaching activities carried out by the hospitals. Despite the fact that they are related to the increase in the cost of health services in hospitals,6 the development and excellence of teaching and research in hospitals should be considered a priority strategy for facing up to the funding crisis in teaching hospitals, by exploiting precisely those aspects that, in fact, differentiate them from the other hospitals in Brazil.



1. Banker RD, Charnes A, Cooper WW. Some models for estimating technical and scale inefficiencies in data envelopment analysis. Manage Sci. 1984;30(9):1078-92. DOI: 10.1287/mnsc.30.9.1078        [ Links ]

2. Caves DW, Christensen LR, Diewert WE. Multilateral comparisons of output, input and productivity using superlative index numbers. Econ J. 1982;92(365):73-86. DOI: 10.2307/2232257        [ Links ]

3. Coelli TJ, Prasada RDS, Battese GE. An introduction to efficiency and productivity analysis. Boston: Kluwer; 1998.         [ Links ]

4. Fare R, Grosskopf S, Norris M, Zhang Z. Productivity growth, technical progress, and efficiency change in industrialized countries. Am Econ Rev. 1994;84(1):66-83.         [ Links ]

5. Farrell MJ. The measurement of productive efficiency. J R Statl Soc (Ser A). 1957;120(3):253-81.         [ Links ]

6. Grosskopf S, Margaritis D, Valdmanis V. The effects of teaching on hospital productivity. Socioecon Plann Sci. 2001;35(3):189-204. DOI: 10.1016/S0038-0121(01)00006-4        [ Links ]

7. Grosskopf S, Margaritis D, Valdmanis V. Competitive effects on teaching hospitals. Eur J Oper Res. 2004;154(11):515-25. DOI: 10.1016/S0377-2217(03)00185-1        [ Links ]

8. Linna M. Health care financing reform and the productivity change in Finnish hospitals. J Health Care Finance. 2000;26(3):83-100.         [ Links ]

9. Lins MPE, Lobo MSC, Silva AM, Fiszman R, Ribeiro VJP. O uso da Análise Envoltória de Dados (DEA) para avaliação de hospitais universitários brasileiros. Cienc Saude Colet. 2007;12(4):985-98. DOI: 10.1590/S1413-8123200700040002010        [ Links ]

10. Lobo MSC, Bloch KV, Fiszman R, Oliveira MR, Ribeiro VJP. Sistema de Informações dos Hospitais Universitários (SIHUF/MEC): um banco de dados administrativo. Cad Saude Colet. 2006;14(1):149-62.         [ Links ]

11. Machado SP, Kruchenbecker R. Desafios e perspectivas futuras dos hospitais universitários do Brasil. Cienc Saude Colet. 2007;12(4):871-7. DOI: 10.1590/S1413-81232007000400009        [ Links ]

12. Malmquist S. Index numbers and indifference surfaces. Trab Estad. 1953;4:209-42.         [ Links ]

13. McCallion G, Glass JC, Jackson R, Kerr CA, McKillop DG. Investigating productivity change and hospital size: a nonparametric frontier approach. Applied Econ. 2000;32(2):161-74. DOI: 10.1080/000368400322859        [ Links ]

14. O´Neill LO, Rauner M, Heidenberger K, Kraus M. A cross-national comparison and taxonomy of DEA-based hospital efficiency studies. Socioecon Plann Scien. 2008;42(3):158-89. DOI: 10.1016/j.seps.2007.03.001        [ Links ]

15. Ozcan, Y. Sensitivity analysis of hospital efficiency under alternative output/input and peer group: a review. Knowledge and policy. Int J Knowl Transf Util. 1992;5(4):1-29.         [ Links ]

16. Sahin I, Ozcan YA. Public sector hospitals efficiency for provincial markets in Turkey. J Med Syst. 2000;24(6):307-20. DOI: 10.1023/A:1005576009257        [ Links ]

17. Sommersguter-Reichmann M. The impact of the Austrian hospital financing reform on hospital productivity: empirical evidence on efficiency and technology changes using a non-parametric input-based Malmquist approach. Health Care Manag Scien. 2000;3(4):309-21. DOI: 10.1023/A:1019022230731        [ Links ]



Maria Stella de Castro Lobo
Serviço de Epidemiologia e Avaliação
Hospital Universitário Clementino Fraga Filho
R. Professor Rodolfo Rocco, 255 - 5º andar
Cidade Universitária, Ilha do Fundão
21941-913 Rio de Janeiro, RJ, Brasil

Received: 11/26/2007
Revised: 7/22/2008
Approved: 10/24/2008



a Ministério da Educação, Ministério da Saúde, Ministério da Ciência e Tecnologia e Ministério de Planejamento, Orçamento e Gestão Portaria Interministerial 562 de 12 de maio 2003. Institui Comissão Interinstitucional com o objetivo de avaliar e diagnosticar a situação dos HUEs no Brasil, visando reformular e/ou reorientar a política nacional para o setor, considerando a necessidade de: melhoria da situação financeira, estrutural, organizativa e de gestão e definição do papel / inserção no SUS. Diário Oficial da Uniao. 28 maio 2003; Seção II: p.23.
b Ministério da Educação, Ministério da Saúde. Portaria Interministerial 1.000, de 15 de abril de 2004. Dispõe sobre os requisitos necessários para a certificação dos hospitais de ensino. Diário Oficial da Uniao. 15 abril 2004; Seção I: p.13-4.