Technical Efficiency of Zone Hospitals in Benin

Joses Muthuri Kirigia*
Omer A. Mensah*
Chris Mwikisa*
Eyob Zere Asbu*
Ali Emrouznejad**
Patrick Makoudode***
Athanase Hounnankan***

* World Health Organization Regional Offi ce for Africa
** Aston University, Birmingham, UK
*** Research, Training and Coaching Centre in Social
Sciences, Cotonou, Benin

Corresponding author
Joses Muthuri Kirigia
e-mail: kirigiaj@afro.who.int

Download PDF here


ABSTRACT

The objectives of this study were to measure the technical and scale efficiency of hospitals in Benin.

DEAP software was used to analyze the technical efficiency among a sample of 23 zonal hospitals in the Republic of Benin over a period of five years, i.e. 2003 to 2007. 

The yearly analysis revealed that 20 (87%), 20 (87%), 14 (61%), 12 (52%) and 8 (35%) of the hospitals were run inefficiently in 2003, 2004, 2005, 2006 and 2007 respectively; and they needed to either increase their outputs or reduce their inputs in order to become efficient. The average variable returns to scale (VRS) technical efficiency scores were 63%, 64%, 78%, 78% and 86% respectively during the years under consideration.

There is some scope for providing outpatient curative and preventive care and inpatient care to extra patients without additional investment into the abovementioned health services. This would entail leveraging of health promotion approaches and lowering of financial barriers to access to boost the consumption of underutilized health services, especially health promotion and disease prevention services.



RÉSUMÉ

Les objectifs de la présente étude consistaient à mesurer l'efficacité technique et d'échelle des hôpitaux au Bénin.

Le logiciel DEAP a été utilisé afin d'analyser l'efficacité technique sur un échantillon de 23 Hôpitaux de Zone de la République du Bénin sur une période de cinq ans, allant de 2003 à 2007. 

L'analyse annuelle a révélé que 20 (87%), 20 (87%), 14 (61%), 12 (52%) et 8 (35%) des hôpitaux étaient respectivement gérés de manière inefficace en 2003, 2004, 2005, 2006 et en 2007 et qu'ils avaient en outre besoin soit d'augmenter leurs recettes ou de diminuer leurs dépenses pour devenir efficaces. Les notations d'efficacité technique relatives aux Rendements d'Echelle Constant (REC) moyens étaient respectivement de 63%, 64%, 78%, 78% et de 86% pour les années considérées.

L est possible de fournir des soins curatifs et préventifs à des patients non-hospitalisés (en consultation externe) et des soins aux patients hospitalisés à d'autres patients sans investissement supplémentaire dans les services de santé ci-dessus mentionnés. Cette vision impliquerait d'accroître les démarches d'encouragement de la santé et de réduire les obstacles financiers en termes d'accès afin de stimuler la consommation des services de santé sous-utilisés, en particulier ceux de promotion de la santé et de prévention des maladies.

SUMÁRIO

Este estudo teve como objectivo medir a eficiência técnica e dimensional dos hospitais no Benin.

Foi utilizado software DEAP para analisar a eficiência técnica de uma amostra de 23 hospitais de zona na República de Benin durante um periodo de tempo de cinco anos, ou seja, entre 2003 e 2007. 

A análise anual revelou que 20 (87%), 20 (87%), 14 (61%), 12 (52%) e 8 (35%) dos hospitais funcionaram de modo ineficiente em 2003, 2004, 2005, 2006 e 2007 e necessitam ou de aumentar os seus outputs ou de reduzir os seus inputs para se tornarem eficientes. Os retornos médios variáveis da eficiência técnica foram 63%, 64%, 78%, 78% e 86% respectivamente durante os anos avaliados.

Existe alguma margem para fornecer mais cuidados terapêuticos e preventivos a doentes externos e internados sem investimento adicional nos serviços de saúde supra referidos. Isto implica a dinamização das abordagens de promoção da saúde e a redução das barreiras financeiras para aumentar a utilização de serviços de saúde não aproveitados, especialmente a promoção da saúde e os serviços de prevenção de doenças.


The Republic of Benin has a surface area of 112,622 km2 and is situated on the West African coast of Africa. It has a population of 8,439,000 people; 46% of whom live in urban areas(1). The human development index for Benin is 0.459, which gives the country a rank of 161 out of 179 countries with data (UNDP 2008). The human poverty index value of 44.5 % for Benin, ranks 125 among 135 developing countries for which the index has been calculated(2). The average life expectancy is 57 years(3).

The per capita total expenditure on health at average exchange rate was US$26 in Benin(4), which was two times lower than that of the Region, and US$8 lower than the bare minimum of US$34 per capita (which does not include costs of scaling up) recommended by the WHO Commission for Macroeconomics and Health(4). Approximately 50.2% of total expenditure on health came from government sources. Private spending on health constituted 49.8% of the total health expenditure; with about 94.9% of it coming from household out-of-pocket expenditures. Such high out-of-pockets expenditures constitute a barrier to efficient health service utilization.

The country has a total health workforce of 10,275 (1.485 health workers per 1000 people). About 68.1% of the total workforce is made up of technical health workers (e.g. physicians, nurses, dentists) and the remaining 31.9% are the health management and support workers(6). The overall health workforce density of Benin is lower than the Regional average health workforce density of 2.626 per 1000. Benin is one of the 57 countries in the world experiencing health workforce crisis(6). This implies that there is great need in Benin to utilize efficiently the available health workforce.

The Benin health system consists of three levels. First, the central level, organized around the ministry of health headquarters, whose mandate is to develop policies and norms and standards, mobilize resources, and oversee the overall management of the system. Second, the intermediate level includes six regional directorates of public health, whose mission is to translate national health policy into action and provide supervisory support to the peripheral level. Finally, the peripheral level is organized in 34 operational public health zones. Each zone covers a population of 100,000 to 200,000 inhabitants. Each zone has a hospital, health centres and village health posts/units. There is approximately a total of 491 public health centres; 34 zone public hospitals; 5 Department/provincial hospitals; 5 specialized public hospitals; 34 religious missions clinics; and 1,409 private-for-profit clinics(7).

Since the available health system inputs are limited, it is necessary to ensure that they are optimally used in providing health services to the greatest number of people possible. Unfortunately, to date no study in Benin has addressed the following questions: Are hospitals producing maximum outputs with the available inputs? Are hospitals operating at an optimal scale? Or are (dis)economies of scale rampant (i.e. inefficiency due to largeness or smallness of hospital size)? This study was meant to contribute to bridging that knowledge gap. Its specific objectives were to measure the technical and scale efficiency of hospitals in Benin over five years (2003-2007).

Method and Data

Economic efficiency is a product of technical efficiency and allocative efficiency. Due to dearth of data on health system input prices, the study reported in this paper was limited to measurement of hospital technical efficiency, comprising of both pure technical and scale efficiency components.

Efficiency analysts usually have a choice of employing either econometric(8) or mathematical programming methods, such as Data Envelopment Analysis (DEA)(9), to estimate technical and scale efficiency. In this study, we chose to use DEA due to its capability of estimating efficiency of hospitals that typically use multiple inputs to produce multiple outputs.

DEA has widely been used in measurement of technical efficiency of hospitals in Asia(10,11,12,13,14) Europe(15,16,17,18,19,20) and North America(21,22,23).However, applications of DEA among hospitals in the WHO African Region are few(25,26,27,28,29). This is the first study to attempt measurement of hospital efficiency in the Republic of Benin.

Under the assumption of variable returns to scale (VRS), DEA measures the technical efficiency (TE) of hospital z compared with n hospitals in a peer group as follows(9):

The above DEA model determined weights u and v from the data and assigned them to each input and output so as to maximize the efficiency rating - TEz - of the hospital being evaluated. The above model was run 23 times to obtain the efficiency scores for each hospital in the sample. The relative hospital efficiency scores are bounded between 0% (completely inefficient) and 100% (technically efficient). Therefore, any hospital that scores 100% is deemed technically efficient; and any hospital with a technical efficiency score of less than 100% is deemed technically inefficient.

Table 1. Health workforce in the Republic of Benin

Data

The sample consisted of 30 hospitals. However, complete inputs and outputs data was available for only 23 of those hospitals; i.e. 68% of the 34 total number of zone hospitals in Benin. Thus, the final analysis was based on the latter group of hospitals.

One of the researchers visited all the hospitals in the sample. At each of the hospitals he met with the medical officer in charge, explained the purpose of the study, and was given access to the relevant inputs and outputs records. He complemented the hard data from the hospital records with interactive interviews with the people in charge of different hospital departments. The inputs and outputs data were collected for 2003, 2004, 2005, 2006 and 2007.

The DEA was estimated with four inputs: total number of doctor/physician hours; total number of other staff (nurses, midwives, laboratory technicians, radiologists, anaesthetist, paramedical assistants) hours; non-salary running costs, which includes all non-personnel expenditures, e.g. expenditures on pharmaceutical, non-pharmaceutical supplies, fuel, utilities (water, electricity, telephone); and number of beds (a proxy of capital inputs). There were two outputs: (i) outpatients visits; and (ii) number of hospital admissions. The study reported in this paper used the DEA software developed by Coelli(30) to measure the yearly technical efficiency and yearly scale efficiency.

Results

Technical and scale efficiency

Table 2 presents the median, mean and standard deviations for the outputs and inputs of hospitals in Benin. Figure 1 summarizes the frequency distributions of technical and scale efficiency scores for hospitals in year 2003, 2004, 2005, 2006 and 2007. In 2003, 2004, 2005, 2006 and 2007, out of the 23 hospitals, approximately three (13%), three (13%), nine (39%), eleven (48%) and fifteen (65%) hospitals respectively had a variable returns to scale technical efficiency score of 100%. Hospitals with values of 100% are deemed technically efficient in the associated year.

Table 2. Descriptive statistics of outputs and inputs of Benin hospitals, 2003–2007

The individual hospitals' technical and scale efficiency scores during the five years are presented in Table 3. The yearly analysis has revealed that 20 (87%), 20 (87%), 14 (61%), 12 (52%) and 8 (35%) of the hospitals were run inefficiently in 2003, 2004, 2005, 2006 and 2007 respectively; and they needed to either increase their outputs (or reduce their inputs) in order to become efficient. The average VRS technical efficiency scores were 63%, 64%, 78%, 78% and 86% respectively during the years under consideration. This finding implies that if the hospitals were operating efficiently, they could have produced 37%, 36%, 22%, 22% and 14% more output (number of outpatient visits and admissions) using their current levels of input endowment. Alternatively, the hospitals could produce their current levels of health service output with 37%, 36%, 22%, 22% and 14% less of their existing health system input endowment.

Table 3. Hospital's technical and scale efficiency, 2003–2007

International partners should support country efforts by ensuring that external resources are increased, predictable, coordinated, and are aligned to country priorities and plans. Countries should make an extra effort to ensure that their plans at coordinating their knowledge processes at creation, acquisition, sharing and use should not be undermined by external pressures.

Scale efficiency

In 2003, 2004, 2005, 2006 and 2007, out of the 23 hospitals: four (17%), two (9%), two (9%), three (13%) and eight (35%) hospitals displayed constant returns to scale (CRS). These hospitals were operating at their most productive scale sizes.

Increasing returns to scale (IRS) were found during the five years in 18 (78%), 20 (87%), 16 (70%), 18 (78% and 13 (57%) hospitals respectively. If a hospital displays IRS, it should expand its scale of operation in order to become scale efficient. Three (13%), one (4%), five (22%), two (9%) and four (17%) hospitals manifested decreasing returns to scale (DRS). In order to operate at the most productive scale size, a hospital exhibiting DRS should scale down its scale of operation.

The average scale efficiency score in the sample was 51% in 2003, 46% in 2004, 52% in 2005, 59% in 2006 and 77% in 2007, implying that there was room for increasing total outputs by about 49% in 2003, 54% in 2004, 48% in 2005, 41% in 2006 and 23% in 2007. This can be accomplished through appropriate adjustment in the size of the scale-inefficient hospitals, where feasible. However, due to indivisibility of physical facilities this may not be possible.

Figure 1. Frequency distribution of technical and scale effciency scores

Scope for output increases and implications for policy

The inefficient hospitals in Benin could operate as efficiently as their peers on the efficiency frontier either by increasing their outputs or reducing utilization of their inputs. The total output increases and/or input reductions needed to make inefficient hospitals efficient are reported in Table 4. In 2007, for example, the inefficient hospitals combined would need to increase the outpatients visits by 260,066 (70%) and the number of admissions by 13,786 (12.4%) in order to become efficient.

Concerning hospitals with outputs falling short of the DEA targets, MoH policy makers could improve their efficiency by improving access to under-utilized health promotion, preventive and outpatient services, e.g. family planning services, antenatal and post natal care, hospital deliveries, child growth monitoring, immunization, HIV/AIDS education, Insecticide Treated Bed Nets, antimalaria treatment for fever, potable drinking water and clean sanitation. Utilization for underutilized preventive and curative services can be boosted through use of health promotion methods, e.g. health education, behaviour change communication, social marketing, information education communication (IEC), social mobilization, advocacy and lobbying ; and through implementation of prepaid national health financing schemes, either tax-funded health services or national social health insurance(32,33,34), which dramatically lower financial barriers to health services access when needed.

Alternatively, if it is very difficult to reduce inefficiencies by increasing utilization of currently underutilized essential health services, policy-makers could improve efficiency through transfer of human resources for health and beds to primary health level health facilities experiencing shortages. Savings of non-salary running costs could be invested in strengthening of primary level health facilities and community health out-reach activities. Those health service levels are critically important in the quest to achieve the health-related United National Millennium Development Goals.

Table 4. Output (input) increases (reductions) needed to make inefficient hospitals efficient, 2003–2007

Limitations of the study

First, DEA does not capture random noise (e.g. epidemics, natural and man-made disasters), and thus, it inadvertently attributes any deviation from frontier to inefficiency(35,36). Thus, by using DEA we may have over estimated the existing magnitudes of inefficiencies.

Second, it would be argued that the ultimate output of hospitals is the aggregate change in health status of the patients who received hospital outpatient and inpatient services. However, due to paucity of data on health status indices such as Quality Adjusted Life Years or health disability indicators such as Disability Adjusted Life Years, this study used intermediate outputs, i.e. number of outpatient visits and number of hospital admissions. On the other hand, even if it were possible to use health outcomes, there would be issues of attribution and consequently the need to adequately control for exogenous factors.

Third, it was not possible to adjust for the quality of both outputs (e.g. successful outpatient visits and inpatient admissions in terms of full recovery from illness, severity of disease differences) and inputs (e.g. more skilful and hard working health workers).

Fourthly, it was not possible to assess the extent to which observed efficiency variations are explained by differences across health zones in socioeconomic status, epidemiology, geographical and financial access to a hospital, variation in operationality of referral systems, and variation in complementary primary level facilities and quality of care differences.

Finally, unavailability of health system inputs prices hampered estimation of allocative efficiency, and hence, calculation of total economic efficiency of hospitals. Thus, the technical efficiency estimates reported in this paper should be viewed as an underestimate of the actual levels of waste prevailing in the hospitals of Benin.

Conclusion

This study has quantified both the technical and scale efficiency of 23 hospitals in Benin; identified the input reductions and/or output increases needed to make inefficient hospitals efficient; and magnitudes and sources of total factor productivity in each hospital.

The analysis revealed that 20 (87%), 20 (87%), 14 (61%), 12 (52%) and 8 (35%) of the hospitals were run inefficiently in 2003, 2004, 2005, 2006 and 2007 compared with the most efficient hospitals in the sample. The under-utilization of health services could be attributed to high out-of-pocket expenditures and quality of care issues. In 2007, all the hospitals manifesting variable returns to scale technical inefficiency would need to increase the number of outpatient visits by 260,066 and inpatient admissions by 13,786 so as to become technically efficient. Therefore, there is some scope for providing outpatient curative and preventive care and inpatient care to extra patients without additional investment into the abovementioned health services. This would entail leveraging of health promotion approaches and lowering of financial barriers to access to boost the consumption of underutilized health services, especially health promotion and disease prevention services.

Alternatively, depending on the decision of policy-makers, the hospital inefficiencies could be ameliorated transferring 55,823 doctors/physician hours; 15,781 hours of other staff (nurses, midwives, laboratory technicians, radiologists, anaesthetist, paramedical assistants); and US$1,584,069 of non-salary running funds to peripheral health facilities and community health programmes.

In their health financing strategy for the African Region(37), the Fifty-Sixth WHO Regional Committee for Africa recommended that member countries should institutionalize efficiency monitoring within national health management information systems (NHIS). Therefore, NHIS capacities ought to be enhanced to routinely capture the input, input prices and output data which could be used to monitor economic efficiency among hospitals and lower level facilities.

Annex

Concept and measurement of technical efficiency of zone hospitals

A hospital can manifest either constant returns to scale (CRS), increasing returns to scale (IRS) or decreasing returns to scale (DRS). Returns to scale inform health decision-makers what happens if, for example, they increase all hospital inputs by the same proportion. This could result in three scenarios: (i). CRS -doubling of all inputs results in doubling of outputs; (ii) IRS - doubling of all inputs may lead to more than a doubling of output; and (iii) doubling of all inputs leads to less than doubling of output. The implications for policy depend on which scenario prevails.

Figure A1 shows a production function where a hospital employs medical doctor hours to provide inpatient health services. It portrays two production frontiers. The first production frontier, which is a straight line 0GCH, assumes CRS. The second frontier, depicted by a concave line ABCD, assumes variable returns to scale (VRS).

Figure A1. Hospital technical efficiency

For example, if a hospital is producing at point E', using 0F medical doctor hours to attend to 0Y1 number of admissions, it is technically inefficient assuming either CRS or VRS. Under a CRS technology, hospital 'E' could have cared for a larger number of admissions (0Y3) with the same number of medical doctor hours (0F). If there are CRS, technical efficiency () of hospital 'E' is given by the ratio:

Similarly, under VRS technology, hospital 'E' could have attended to 0Y2 admissions employing the same number of medical doctor-hours 0F. Pure technical efficiency () assuming VRS is measured as:

A technically efficient hospital has a technical efficiency score of one (or 100%), whereas the inefficient ones have a score less than one (or less than 100%). For example, supposing that the pure technical efficiency of hospital 'E' was 75%. This implies that the hospital could have attended to 25% more admissions than it is currently attending to with the same number of doctor hours. Alternatively, hospital 'E' could reduce medical doctor hours by 25% and still attend to its current number of admissions.

In Figure A1, scale inefficiency is the difference between EH (technical efficiency under CRS) and ED (technical efficiency under VRS). Practically, scale efficiency (SE) is calculated as the ratio of technical efficiency under CRS and technical efficiency under VRS:

or .

Scale efficiency compares the average product of the hospital at 'E' to average product at the technically optimal point 'C'. This comparison tells us if the hospital has scale inefficiency due to being too small (in the IRS portion of the production function, e.g. E), or too large (in the decreasing returns to scale portion of the production function, e.g. D).

For example, if a hospital is producing at point 'E', using 0F medical doctor hours to attend to 0Y1 number of admissions, it is technically inefficient assuming either CRS or VRS. Under a CRS technology, hospital 'E' could have cared for a larger number of admissions (0Y3) with the same number of medical doctor hours (0F). If there are CRS, technical efficiency of hospital 'E' is given by the ratio.

Similarly, under VRS technology, hospital 'E' could have attended to 0Y2 admissions employing the same number of medical doctor-hours 0F. Pure technical efficiency assuming VRS is measured as.

A technically efficient hospital has a technical efficiency score of one (or 100%), whereas the inefficient ones have a score less than one (or less than 100%). For example, supposing that the pure technical efficiency of hospital 'E' was 75%. This implies that the hospital could have attended to 25% more admissions than it is currently attending to with the same number of doctor hours. Alternatively, hospital 'E' could reduce medical doctor hours by 25% and still attend to its current number of admissions.

In Figure A1, scale inefficiency is the difference between EH (technical efficiency under CRS) and ED (technical efficiency under VRS). Practically, scale efficiency (SE) is calculated as the ratio of technical efficiency under CRS and technical efficiency under VRS: or . Scale efficiency compares the average product of the hospital at 'E' to average product at the technically optimal point 'C'. This comparison tells us if the hospital has scale inefficiency due to being too small (in the IRS portion of the production function, e.g. E), or too large (in the decreasing returns to scale portion of the production function, e.g. D).

Scale efficient hospitals have a score of one (or 100%), whilst the inefficient ones have a score less than one (or less than 100%). For example, if the scale efficiency score for hospital 'E' was 85%, it means that about 15% of inefficiency is accounted for by unsuitable hospital size (inefficiently large or small). Since in Figure A1 a hospital producing at 'E' is operating in the IRS portion of the production function, it implies that the hospital is too large and there is scope of down-sizing by 15% while attending to its current level of admissions.


ACKNOWLEDGEMENTS

We gratefully acknowledge the contributions of all those who have actively participated in, and assisted, the preparations and conduct of the Algiers Ministerial Conference and the follow up regional consultation in Brazzaville that discussed the Framework for its implementation.

REFERENCES