Determinants of Hospital Efficiency: insights from the literature

development goals entails curbing waste in health spending due to inefficiency. Inefficiency is a pervasive problem in health systems. The World Health Organization estimates that on average, 20-40% of the global total health expenditure is wasted. The proportion of total health expenditure attributed to hospitals is high, which implies that improving the efficiency of hospitals will lead to more efficient health systems. This study aims to synthetize the major determinants of hospital inefficiency and to develop a framework to identify causes of inefficiency and develop multi-factor interventions to address inefficiencies.

There is no one single magic formula or intervention that can be adopted by different hospitals and can be effective in improving hospital efficiencies. Multiple factors influence the efficiency of hospitals and to address hospital inefficiency multi-intervention packages focusing on the hospital and its environment should be developed.

Background
Inefficiency is a pervasive problem in health systems. The World Health Organization (WHO) estimates that on average, 20-40% of global total health spending is wasted [1]. In 2011, the annual cost of waste to the US health care system was estimated between a low of 21% and a high of 47% of the total health expenditure [2].
Improving efficiency of health systems has gained more attention due to rapid growth in health expenditure driven by factors that include demographic and epidemiological changes, growth in health technology and rising expectations of the population [3]. The 2018 Declaration of Astana on primary Health Care aptly states that the global community cannot afford waste in health care spending due to inefficiency if universal health coverage and other health and health-related targets of the Sustainable Development Goals are to be achieved [4].
Hospitals consume a large proportion of the total health expenditure. In the OECD member states, hospitals, on average absorbed about 38% of the total health expenditure in 2016. Figures for some iii.
Educational responsibilities -teaching hospital (TH) and non-teaching hospital (NTH); iv. Hierarchical classification -primary, secondary and tertiary hospitals; v. Degree of service specification: general and specialized hospitals; and vi. Employee status of their doctors: staff model and non-staff model hospitals.
A combination of the above criteria may be used in classifying hospitals. For example, private hospitals may be classified as private for profit or private not for profit.

Methods
The study is based on survey of the literature on hospital efficiency and its determinants. The studies include those that employ ratio methods of efficiency analysis, data envelopment analysis and stochastic frontier models and econometric models such as the tobit regression to assess determinants of technical efficiency. Data was extracted in a table format categorized as those that are within the hospital, outside the hospital but within the health system and those that are outside the hospital and health system in the broader macroeconomic system.

Search strategy
Relevant studies for this review were identified from PubMed, references cited in selected articles and grey literature that includes reports and guidelines from the World Health Organization (WHO),

Ownership
The evidence regarding the effect of hospital ownership on efficiency is mixed, and depends on a number of factors including the country and geographical location of the hospital under investigation, the hospital structure and its financing schemes, in addition to the methods and the comparative data sets used in measuring efficiency [15][16][17]. While some studies indicate that private FP hospitals are more efficient than private NFP hospitals [18][19][20], others, to the contrary, show that private NFP hospitals are more efficient than private FP ones [21][22][23]. As profit making is the key component of the mission of FP hospitals, increased profits could be achieved through increased efficiencies [24].
Other research concluded that government/public hospitals are more technically efficient than private NFP ones and that hospitals in general facing greater financial pressure tend to improve their efficiency [25]. To the contrary, a study in Germany examining changes in efficiency after privatization demonstrated that conversion from public to private for-profit status resulted in an increase in efficiency between 2.9% and 4.9%, which also appeared to be permanent rather than a transient change [9]. Still others have concluded that there is no clear evidence that private hospital ownership -non-profit and for profits -is associated with higher efficiency compared to public hospital ownership [26,27].

Size
Optimum hospital operational scale is a function of patient access, economies of scale and volume as a determinant of patient outcome [28]. Economies of scale characterize a situation in which fixed costs of production are higher relative to variable costs [28,29]. Economies of scale exist when longrun average costs decrease as the scale or volume of production increases. The existence of economies of scale implies that efficiency gains could be reaped by expanding hospital size. Optimum hospital size is seen when all economies of scale have already been exploited but have not yet reached a point of diseconomies of scale, where long-run average costs start to increase as the hospital size (scale of production) increases. Smaller hospitals might be inefficient because of their inability to spread fixed (overhead) and administrative costs across a greater number of cases/patients.
In a study assessing the technical and allocative efficiency of Greek public hospitals, it was found that small hospitals were the least efficient (technical efficiency score = 0.80) in comparison with mediumsized (0.86) and large (0.90) hospitals [30]. On the other hand, in a study of Turkish hospitals, small hospitals were found to be relatively more efficient compared to medium and large hospitals and had better patient satisfaction [31]. Research suggests that a hospital can take full advantage of economies of scale when its size ranges between 200-300 beds, while diseconomies of scale can be expected to occur when a hospital size is below 200 and above 600 beds [32]. A systematic review on the scale efficiency of hospitals has reported that there is scale inefficiency when the hospital size is below 200 beds and over 600 beds [33]. This implies that from the efficiency perspective, the optimal hospital bed capacity lies between 200 and 600 beds.

Specialization vs. Economies of Scope
The range of products offered by a hospital may affect efficiency. However, the direction of influence is unpredictable. Multi-output production sometimes enables achieving economies of scope and at other times results in diseconomies of scope. Scope economies occur when the joint production of two or more products (e.g. inpatient and outpatient services) can be achieved at lower costs than the combined cost of producing each output individually. In other words, a general hospital is more efficient than a specialized hospital. The opposite effect is seen when there are diseconomies of scope.
In their study of 133 Italian hospitals, Colombi et.al. found that hospitals specialized in single treatments have higher transient inefficiency than general hospitals, indicating economies of scope [15]. In the same vein, a study in three States of the USA using stochastic frontier cost function presented some evidence that general hospitals are more efficient than specialized ones [34]. A study [35] using an extensive panel dataset in England classified sources of heterogeneity in hospital services into two: i.
admission-type heterogeneity, which results from collocating the treatment of elective and emergency patients within the same hospital; and ii. Service-line heterogeneity, which is the result of collocating a number of clinical specializations within the same hospital.
The findings indicate the presence of negative economies of scope across the two types of admission -increased volume of elective admission to a hospital was associated with an increase in the cost of emergency care. Moreover, for emergency admissions, there was evidence of economies of scope across service lines, that is, increased emergency activity in one service line is associated with lower cost of emergency care in other service lines. No evidence of economies of scope was found across service lines for elective admissions.
In a developing country setting, a study from Vietnam reported evidence of modest economies of scale, which differed by category of hospital -economies of scope greater in provincial general than in central general hospitals. The question about economies of scope assessed whether it was less expensive to provide both inpatient and outpatient services at the same hospital than to have a separate facility for outpatient services [36].

Teaching Status
Some research indicates that THs were found to be less efficient than NTHs due to a number of factors including the use of medical residents instead of attending physicians (staff physicians) who are more experienced in caring for patients [37,38].Using DEA and tobit regression analysis, a study in Greece during the period of financial crisis 2009-2012 demonstrated that university hospitals are less efficient compared with non-teaching hospitals due to multiplicity of functions including teaching and research in addition to patient care [9]. Grosskopf et.al found that inefficiency attributed to congestion or the excess use of residents amounted to 20% of the total inefficiency score in teaching hospitals [39]. In the presence of congestion inefficiency, the increase of inputs over a given level results in a decrease of outputs. The teaching status of a hospital often requires conducting additional clinical tests and diagnostics for the benefit of the residents this imply that teaching hospitals are likely to use higher level of resources than nonteaching hospitals for producing the same level of output [40]. An empirical study of the determinants of hospital efficiency in Italy using four-randomcomponent stochastic frontier model did not find statistically significant effect of teaching hospitals on transient (short-term) and permanent (long-term) inefficiency [15].

Membership of Multihospital System
Studies have indicated that hospitals that do not belong to a multihospital system or are standalone have low efficiency scores [22]. It is argued that, when a hospital is part of a system, greater efficiency is achieved, as the production of multiple products in different hospitals within the system could only be achieved by the employment of highly specialized group of managers and technical experts, and the elimination of duplicative administrative functions [41]. In a recent research, comparing performance, operating characteristics, and market environments of low-and highefficiency hospitals in 37 states in the USA using data from the Healthcare Cost and Utilization Project, high-efficiency hospitals tended to be members of multihospital systems, non-teaching, and investor-owned [42]. The benefits of system membership depends greatly upon the characteristics of the system [43]. Moderately centralized systems were not found to have the most efficient hospitals.
Instead, highest levels of efficiency were found among hospitals members in centralized physician/insurance health systems and decentralized health systems.

Other hospital-level factors
A number of studies indicate that other internal hospital factors (including the case-mix index, occupancy rate, and outpatient to inpatient ratio), which are related to managerial style and performance may all significantly affect hospital efficiency.
A research of Japanese hospitals studying the link between managerial performance and hospital efficiency found that setting and monitoring financial parameters linked to managerial performance had a significant positive relationship with hospital efficiency [44].
Another factor, the case mix index (CMI) which is usually used as a categorization scheme, to assess the level of severity/complexity of cases treated by a hospital may affect the level of hospital efficiency. A study found that one of the main factors accountable for differences in cost per case between hospitals was the difference in case mix [28].
Hospitals with higher bed occupancy rates may be more efficient. A hospital occupancy rate is usually measured as total patient days divided by total bed days available. A study of referral hospitals in Uganda indicated that bed occupancy was among the significant factors explaining variations in hospital efficiency [45]. A similar study examining the efficiency of 112 Greek public hospitals, also found a positive relationship between occupancy rate and efficiency [46].
Shifting the delivery of healthcare from a hospital setting (inpatient) to a same-day ambulatory setting (outpatient) and reduction of the average length of hospital stay are other factors that have been generally adopted by policy makers to promote efficiency and control hospital spending. This trend has seen a decline in the number of hospital beds per capita in most European countries over the past two decades [47,48].

Geographic Location
According to location theory, an efficient location pattern is presumably one in which some societally predetermined level or volume of service is met at minimum total system costs of operation and travel, or alternatively, one that maximizes the volume of service within a predetermined budget constraint. [49]. A number of studies found that hospitals in relatively remote or rural areas were mostly less efficient than hospitals in urban areas. This is not surprising because the primary role of these hospitals is to provide basic healthcare services as the safety nets providers for the population under these geographical locations. As such, hospitals in rural and remote areas are not as busy as those in urban areas. This results in the presence of unused capacity including underutilized inputs such as doctors and other healthcare professionals leading to lower efficiency scores [40]. However, hospital efficiency scores may change when hospitals are evaluated in their own groups [50].

Competition
A number of studies investigated the effect of competition on hospital production and efficiency [51].
Analyzing the relationship between competition and efficiency is a difficult task due to the complexity of health systems and their structures. Furthermore, the potential role for competition in healthcare is often mixed [52]. Competition occurs when producers try to attract customers from their competitors by providing a more appealing combination of price and quality. In conventional markets, this may lead to greater efficiency; however, it is not always the case in the healthcare market [53]. A study in 2001-2004, assessing the effects of competition on efficiency among hospitals in Florida found that hospitals located in a less competitive market had lower technical efficiency scores than those in a more competitive market [54]. In contrast, research on hospital efficiency in Turkey indicates that efficiency is not significantly affected by the intensity of competition among hospitals [55].

Reimbursement Systems and Provider Incentives
The introduction of DRG-based prospective payment system was often found to have negative effects on hospitals' length of stay and positive effects on efficiency [56]. A number of studies conducted in the US and Europe indicate that the introduction of DRGs resulted in a positive shift in efficiency [57][58][59][60].

Discussion
Results of this study highlight that a host of factors influence the efficiency of hospitals. These range from hospital-level characteristics to other health system-level factors and those that are outside the health system in the broader economic formation. The hospital as an open system nested within the health system and other higher-level systems must interact with those systems (its environment) for survival, adaptation and growth [61]. Hence, assessment of the causes of hospital inefficiency must view the hospital in the context of its environment. Furthermore, the study underscores the fact that there is no single cause of hospital inefficiency; it is rather multifactorial. Therefore, solutions aimed at addressing hospital inefficiency should be multi-intervention packages focused on the hospital and its environment as seen fit.
The surveyed literature has not shown effect of the three types of ownership (public, PFP and PNFP) on hospital efficiency conclusively. Arguments based on theories including property rights theory, agency theory and public choice theory have concluded that private-for-profit hospitals are likely to achieve greater efficiency due to an assumed hospital behavioral objective of profit maximization attributed to the right to appropriate surplus finances, and non-interference by politicians whose objectives may be at odds with profit maximization [26]. In common usage, ownership implies the right to purchase or dispose of any assets of an organization including any financial surplus resulting from the process of production. This notion of ownership does not necessarily indicate the behavioral objective(s) of different forms of hospital ownership [62]. Hence, ascribing the profit maximization motive based on ownership and concluding that PFP hospitals tend to be more efficient compared to the public and PNFPs may not be necessarily true. Private for profit hospitals may also have other competing objectives such as patient welfare, quest for prestige/reputation, excellence in research and teaching. Moreover, market failures such as barriers to entry and asymmetry of information; the mix of providers and purchasers and how they interact; the prevailing contractual and regulatory arrangements; the incentives in place; and the extent of out-of-pocket payment or third party payment are among the contextual factors that may constrain efficiency related to private for-profit ownership [62,63]. This finding is important for policy makers, as it implies that private provision alone does not enhance efficiency in the hospital system and that a thorough analysis of enabling factors is crucial before a decision to privatize.
Studies have identified that scale economies exist in hospitals with bed-capacity less than 200 and diseconomies in those with bed capacity over 600, implying optimal capacity between 200 and 600 beds. The optimum size depends on a number of factors including location, population served and health system policies that they wish to implement. Scale inefficiency concerns have driven a number of countries to implement merger of hospitals. Cost savings were realized immediately after hospital merger [33]. The cost savings are attributed to: (i) reduced duplication and lower administrative costs; (ii) strengthened buying power of the hospital as a purchaser of goods and services; and (iii) spreading of the cost of common resources (e.g. theatres and diagnostic equipment) across a larger volume [64]. It should, however, be noted that in a private health care market, mergers may also be primarily focused on reducing competition and boosting profitability [28]. On the other hand, to improve the efficiency of hospitals that are too big and facing diseconomies of scale, downsizing of hospitals has been implemented [33].
Membership of multihospital system is regarded to operate via individual hospital-level scale and scope economies. This includes improved bulk purchase of inputs such as human resources and reduced interest rate in capital markets, elimination of duplicative administrative functions, economies in marketing a large organization and greater ability to control environmental factors [43].
The range of products offered by a hospital may affect efficiency. However, the direction of influence is unpredictable as can be inferred from the results section. Multi-output production sometimes enables achieving economies of scope and at other times results in diseconomies of scope. Decisions on whether or not to co-locate different functions such as emergency vs inpatient or outpatient vs inpatient care, and different clinical specialties should be based on a thorough analysis of the contextual factors that are likely to affect scope economies.
Enhancing managerial performance and hospital processes which could be achieved through different initiatives ranging from the use of bonuses as an instrument for regulating work incentives to reduce on labor costs, to merging hospitals or restructuring them into clinics; to replacing total budget accounting and introducing detailed accounting for various expenditures were all found to improve efficiency. Furthermore, the introduction and addition of DRG case mix index in a stochastic frontier efficiency analysis and factors influencing hospital efficiency resulted in a 50% reduction in inefficiency [65].

Developing a hospital efficiency framework
The preceding discussion revealed that various factors determine the level of hospital efficiency including hospital size, patient case-mix, ownership, location, teaching status, reimbursement system used and others [8, 33,63]. These factors can be classified broadly as factors that are internal and/or external to the hospital [15]. Furthermore, understanding the determinants of hospital efficiency entails a proper examination of hospital production function, which portrays the relationship between factor inputs (such as human resources, beds, medicines and technologies, etc.) and the outputs of a hospital (such as inpatient and outpatient services, number of health professionals trained etc.). Thus, the efficiency with which a hospital produces its valued outputs is influenced by factors that may be wholly or partially out of the control of the hospital. Figure 1 depicts this framework.
Thus, the technical and allocative efficiency of a hospital may be influenced at the level of the elements of the production function and their linkages (the main box - Figure 1) and/or factors that emanate outside the hospital both from within the wider health system and outside the purview of the health system. Accordingly, these factors may be clustered around two broad categories as: i.
Hospital-level factors, which are within reach and can be addressed by the hospital; and Extra-hospital factors, which are not within easy reach of the hospital and require interventions from higher levels -the wider health system or relevant sectors outside the health system.
ii. External influencing factors, which are not within easy reach of the hospital and require interventions from higher levels of the wider health system or other relevant sectors outside the health system.

Conclusions
The study shows that identification of specific determinants of hospital efficiency is a necessary step towards designing interventions to improve potential inefficiencies. There is no single standalone intervention that is effective in improving hospital efficiency; a multi-intervention package tuned to the context is rather required.
An abundance of research have appeared over the years intending to assess hospital inefficiencies, however, few have had any impact on corrective actions taken or new policies adopted by policy makers or hospital management [66,67]. Thus, one main lesson and recommendation that could be drawn from this literature review, is that there is no single magic formula or policy that can be adopted by different hospitals and can be effective in improving hospital performance and efficiencies. It is essential to identify the set of determinants of efficiency for each hospital so that targeted policies could be planned and implemented accordingly. The identification of specific determinants of hospital efficiency and establishing the efficiency performance framework (as in  Hospital production function and efficiency framework