Menu
Menu
Menu
Menu
Background: The Data Envelopment Analysis (DEA) method has been fruitfully used in many countries in Asia, Europe and North America to shed light on the efficiency of health facilities and programmes. There is, however, a dearth of such studies in countries in sub-Saharan Africa. Since hospitals and health centres are important instruments in the efforts to scale up pro-poor cost-effective interventions aimed at achieving the United Nations Millennium Development Goals, decision-makers need to ensure that these health facilities provide efficient services. The objective of this study was to measure the technical efficiency (TE) and scale efficiency (SE) of a sample of public peripheral health units (PHUs) in Sierra Leone. Methods: This study applied the Data Envelopment Analysis approach to investigate the TE and SE among a sample of 37 PHUs in Sierra Leone. Results: Twenty-two (59%) of the 37 health units analysed were found to be technically inefficient, with an average score of 63% (standard deviation = 18%). On the other hand, 24 (65%) health units were found to be scale inefficient, with an average scale efficiency score of 72% (standard deviation = 17%). Conclusion: It is concluded that with the existing high levels of pure technical and scale inefficiency, scaling up of interventions to achieve both global and regional targets such as the MDG and Abuja health targets becomes far-fetched. In a country with per capita expenditure on health of about US$7, and with only 30% of its population having access to health services, it is demonstrated that efficiency savings can significantly augment the government’s initiatives to cater for the unmet health care needs of the population. Therefore, we strongly recommend that Sierra Leone and all other countries in the Region should institutionalise health facility efficiency monitoring at the Ministry of Health headquarter (MoH/HQ) and at each health district headquarter. © 2005 Renner et al; licensee BioMed Central Ltd.
The Ministry of Health and Sanitation (MOHS) provides about 50% of health care services. The remainder is provided by the private sector (private-for-profit institutions and traditional healers) and national (e.g. Christian Health Association of Sierra Leone) and international (e.g. German Leprosy Rehabilitation Association and Medecins Sans Frontieres) NGOs [10]. The country has 13 health districts, each with a District Health Management Team responsible for the implementation, supervision and monitoring of health programmes in the district. Sierra Leone has a total of 31 government hospitals, 22 mission hospitals/clinics, 78 private hospitals/clinics and a network of 788 PHUs. As indicated in Table Table1,1, there are geographical inequities in the distribution of health facilities in the country [10]. Functioning PHUs and hospitals Source: WHO Regional Office for Africa [10] Table Table22 provides estimates of the number and ratio of human resources for health in 2002. Approximately 63% of the health workers were employed by the government and the remaining by NGOs and private-for-profit institutions. Estimated number and ratio of health personnel in 2002 Source: WHO Regional Office for Africa [10] Input and output data were analysed for the year 2000. Due to research resource constraints, the planning and information department at the MOHS decided to choose one health district for the study of PHUs. The choice of the study district was done using a simple random sampling technique. This process led to the choice of Pujehun District. Even though there are 46 PHUs in Pujehun today, in the year 2000 there were only 39 PHUs. The data were collected by Pujehun District Health Team using the primary health care facility efficiency analysis data collection instrument of the WHO Regional Office for Africa [11]. Turnock [12] developed a conceptual framework that ties together the mission and functions of public health to the inputs, processes, outputs and outcomes of the system (see Figure Figure1).1). He stated that health systems combine inputs (human, organizational, informational, financial and other resources) to produce outputs (programmes or services or interventions) intended to ultimately yield health or quality-of-life outcomes. In terms of measurability, the author posits that many inputs such as human, financial and organizational resources are easily counted or measured. He further explains that outputs (e.g. number of antenatal care visits, number of immunizations provided, number of people who receive health education and number of condoms distributed) are also generally easy to recognize and count. Following Turnock [12], a public health practice, such as a health centre, employs multiple inputs to produce multiple outputs. Relationship between inputs and the production process and resulting outputs. DEA (a non-parametric method) defines efficiency as the ratio of the weighted sum of outputs of a health centre to its weighted sum of inputs [13]. It is particularly useful in public sector organizations (e.g. health facilities) that lack the profit maximization motive and employ a multiple input, multiple output production process. The technical efficiency (TE) of PHUs was found by solving the following linear programming problem for each health unit in the sample: Max h0=∑r=1suryrj0 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaqabeGadaaakeaacqWGnbqtcqWGHbqycqWG4baEcaaMc8UaemiAaG2aaSbaaSqaaiabicdaWaqabaGccqGH9aqpdaaeWbqaaiabdwha1naaBaaaleaacqWGYbGCaeqaaOGaemyEaK3aaSbaaSqaaiabdkhaYjabdQgaQnaaBaaameaacqaIWaamaeqaaaWcbeaaaeaacqWGYbGCcqGH9aqpcqaIXaqmaeaacqWGZbWCa0GaeyyeIuoaaaa@4557@ Subject to: ∑i=1mvixij0=1∑r=1suryrj−∑vixij≤0,j=1,…,nur,vi≥0 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@6C1C@ Where: yrj = amount of output r from health centre j xij = amount of input i to health centre j ur = weight given to output r vi = weight given to input i n = number of hospitals s = number of outputs m = number of inputs This mathematical programming technique establishes a production possibilities frontier based on relatively efficient health centres and measures how far the inefficient health centres are from this ‘best’ practice frontier [14]. The efficient health centres lie on the frontier and are assigned a score of 1 or 100%. Inefficient health centres are allocated a score that is less than 1 (or 100%). The higher the score, the greater the efficiency, and vice versa. The variable returns to scale (VRS) model was estimated to facilitate the estimation of scale efficiency. It assumed that changes in inputs would lead to disproportionate changes in outputs. In other words, a percentage increase in input can yield less than a percentage change in output signifying diseconomies of scale, or more than a percentage increase of output implying existence of economies of scale. The scale efficiency (SE) is the ratio of constant returns to scale technical efficiency (TECRS) to variable returns to scale technical efficiency (TEVRS), i.e. SE = (TECRS)/(TEVRS) [15]. All the analysis was undertaken using DEAP, the software developed by Coelli [16]. The output-oriented DEA model was used for the analysis because the management of PHUs had no control over inputs, especially the deployment of human resources. However, given their public health orientation, PHU staff had a duty to induce demand (through health promotion strategies) for preventive health care services such as antenatal care, family planning services, immunizations, etc. Through their outreach public health work among communities, PHU staff were also supposed to mobilize community efforts and other resources to provide clean water and hygienic human waste disposal facilities, e.g. vented improved pit latrines, especially in rural areas and slums. As one can see in Table Table3,3, there is serious population under-coverage of the various interventions in Sierra Leone. This is mainly due to critical resource constraints, e.g. per capita total expenditure on health is only US$7 compared to the US$34 per person recommended by the WHO Commission for Macroeconomics and Health [8]. This implies that although there is a large unmet need for primary health care among communities, severe budgetary constraints make it difficult to increase inputs, even assuming that PHUs have control over inputs (which they do not have). Even where inputs (e.g. labour) might be under utilized, it is not within their power to dispose of excess inputs. We felt that output maximization is the most appropriate orientation for health centres which are given a fixed input and requested to produce as much output as possible. Thus, an output-oriented approach focused on the amount by which health unit outputs could be expanded with the same level of inputs. Manifestations of inaccessibility to basic health services in Sierra Leone Sources: UNICEF [19] and WHO/AFRO [20] Furthermore, the output- and input-oriented models will estimate exactly the same frontier, and therefore, by definition identify the same set of PHUs (firms) as being efficient. It is only the efficiency measures associated with the inefficient firms that may differ between the two methods [16]. In fact under the assumption of constant returns to scale, even the efficiency scores will not change. We, therefore, feel that the choice of model is not going to affect the results significantly. The DEA model was estimated with a total of eight variables: six outputs and two inputs. The six outputs for each individual PHU were: (i) number of antenatal plus post-natal visits; (ii) number of child deliveries; (iii) nutritional/child growth monitoring visits; (iv) number of family planning visits; (v) number of children under the age of 5 years immunized plus pregnant women immunized with tetanus toxoid (TT); and (vi) total number of health education sessions conducted through home visits, public meetings, school lectures and outpatient department. PHUs in Sierra Leone did not provide curative care; they were dedicated fully to the provision of health promotion and disease prevention services. The two inputs were: (i) technical staff (community health nurse, vaccinator and maternal and child health aide); and (ii) subordinate staff (including traditional birth attendants, porters and watchmen). The choice of inputs and outputs was guided by the public health conceptual framework and past studies.
N/A
