Background

The recent global economic crisis has influenced the budgets of public organizations, including those of public hospitals and national health systems in general. According to the World Health Organization, for high-income and upper-middle-income countries, such as Italy, the main challenge relating to the provision of health services is to continue improving efficiency, quality, and equity [1]. Moreover, within the evolving social and economic environment, budget constraints have prompted a search for new ways of monitoring and controlling organizational finances that focus on the efficient and effective use of public resources [2]. Monitoring the performance of healthcare providers is a relevant issue worldwide, particularly in contexts such as hospitals given their significant effect on population health and the economy. The main problem facing hospitals has been inefficient use of existing resources rather than a lack of resources [3]. Therefore, hospital efficiency plays a strategic role in healthcare organizations. Assessing the determinants of efficiency allows managers to formulate appropriate organizational strategies to meet the challenges associated with continuous change. Focusing on the Italian National Health System, Guerrini et al. [4] note that in recent years, increasing attention has been paid to ensuring financial equilibrium and reducing the average annual growth rate of total health expenditure per capita. Italy has a regionally based National Health System that provides universal coverage free of charge [5]. Regional governments allocate resources to healthcare organizations and have a significant degree of autonomy in organizing the provision of healthcare planning and monitoring, as well as determining the number and vocation of healthcare providers [6], and the size of hospitals in terms of number of beds. One of the most important choices of Italy’s regional governments in relation to hospitals is the number of beds.

Literature review

Over the past 30 years, many research studies on hospital efficiency evaluation have been conducted in different countries. Färe et al. [7] published the first European study, and this was followed by many others [8,9,10,11,12,13]. Such studies have now been conducted in many other countries around the world, including in China, Iran, Brazil, Ukraine, and Angola [14,15,16,17,18].

Hospital efficiency has been evaluated in relation to many factors (see Table 1). Kohl et al. [19] consider capital investment of great importance in efficiency analysis. Chilingerian and Sherman [20] note the importance of labor to the service process in hospitals, identifying this factor as essential in assessing performance. In addition, other organizational choices, such as teaching status and the provision of first aid, are often considered relevant in assessing hospital efficiency in the literature [21]. A frequently investigated aspect of hospital efficiency is patient length of stay. Among previous studies there is consensus that an increase in the number of hospitalization days has a negative effect on hospital performance [22, 23]. Rebba and Rizzi [24] have found that a high number of beds per inhabitant is one of the major causes of hospital inefficiency because it increases overhead costs. In addition, Daidone and D’Amico [25] and Shahhoseini et al. [26] argue that efficiency is positively affected by hospital size. Nevertheless, this relationship remains controversial because research such as that of Nayar et al. [27] has found that small hospitals have higher efficiency and quality scores than do large hospitals. In addition, Chang [28] argues that the number of service types offered is negatively related to efficiency because a greater scope of service means a higher level of management complexity. This is supported by Campedelli et al. [29], who found that hospitals that provide a first-aid service incur higher costs than hospitals that do not provide this service.

Table 1 Factors affecting hospital efficiency
Full size table

Many studies on hospital efficiency assess pure technical efficiency (PTE) and scale efficiency [30, 31]. PTE represents managerial efficiency [32], which refers to management’s ability to save inputs to produce a certain amount of outputs or to produce more outputs with a given level of inputs [33, 34]. Scale efficiency indicates whether an organization operates at the most productive scale size [35].

However, despite the numerous studies on hospital efficiency conducted in many different countries, few studies have attempted to include quality measures [36], particularly those undertaken in the European context. One possible reason for this gap could be that the scientific community has not yet agreed on a common standard for addressing questions of quality in hospitals [19]. However, according to Chatfield [37], efficiency studies without quality considerations are neglecting a critical factor. Some research that does consider quality has been conducted in the United States [38, 39], and it identifies mortality rate as a good valuation of quality.

The purpose of this study was to assess the efficiency of public hospitals while including measures of quality. To fulfill this aim, the study attempted to answer the following research questions: (1) What are the main organizational factors that generate hospital inefficiency? (2) How do internal and external features affect hospital efficiency? (3) How has hospital efficiency changed over time? To answer these questions, the study employed multistage data envelopment analysis (DEA).

Data envelopment analysis

DEA is a nonparametric technique developed by Charnes et al. [40]. It is used to rank and compare the efficiency of various entities, defined as decision-making units (DMUs). DEA is grounded in an optimization algorithm that assigns a score between 0 and 1 to the DMUs given the input consumed and the output produced. DEA models allow assessment of the relative efficiency of DMUs by creating a production frontier using the best practice of the observed data. In addition, DEA can be considered an alternative to parametric frontier methods and financial ratio analysis. Rhodes [41] highlights that financial ratios allow benchmarking among a multitude of operating units, focusing on their financial results. Nayar et al. [27] note that the main flaw in measuring performance using this kind of methodology is the lack of technical indicators that enable evaluation of the efficiency of structures and the quality of services provided. According to Worthington [42] and O’Neill et al. [43], methods of nonparametric analysis such as DEA overcome the weaknesses of financial ratios and parametric analysis because they do not require any assumption related to the functional form of the relationship between outputs and inputs [44]. Further, DEA can not only identify inefficient units but can also assess the degree of inefficiency. DEA uses linear programming to construct a piecewise convex linear-segmented efficiency frontier, making it more flexible than econometric frontier analysis. Moreover, DEA can include multiple inputs and outputs. Despite these identified benefits, DEA presents a drawback: it attributes every deviation from the best practice frontier to inefficiency. However, such deviations might be due to statistical noise (e.g., measurement errors).

Methods

Study design

The study adopted a cross-sectional design to assess the efficiency of public hospitals in the Veneto region for the years 2018 and 2019. It used a longitudinal design to analyze the trend in technical efficiency in general hospitals from 2018 to 2019. More specifically, a Charnes, Cooper, and Rhodes (CCR) input-oriented model, decomposition of the obtained scores, and slack assessment were developed for 2018 and 2019. Subsequently, Tobit regression was applied to understand the internal and external sources of inefficiency. Finally, the Malmquist Productivity Index was used to assess how efficiency has changed over time.

Study population

To conduct the efficiency analysis, the Veneto region was selected as the case study site [45]. This was because of the region’s high level of interest in researching new ways to control the efficiency of public hospitals. Guerrini et al. [3] found that this region has specific characteristics (i.e., an increase in the elderly population and growing life expectancy of residents) that have led to a gradual increase in comorbidity and chronic diseases and a corresponding increase in demand for high-quality healthcare services. All the data used in this study were provided by Azienda Zero UOC Controllo di Gestione e Adempimenti LEA (Azienda Zero). A full dataset from 2018 to 2019 containing nonpublicly available technical data cost and revenue items was analyzed. The data included in this study are the operative costs and revenue for public hospitals, number of beds, number of FTEs, mortality rate, inappropriate admission rate, bed occupancy rate, length of stay, provision of first aid, and number of residents. During the period under analysis, the Veneto region had 53 public hospitals. One hospital closed in 2018, three hospitals treated only a particular type of patient and were therefore considered nonhomogeneous and noncomparable, and six hospitals presented missing data. Therefore, the final sample included 43 hospitals and 86 observations.

Selection of study variables

Selecting suitable inputs and outputs is crucial for ensuring meaningful efficiency analysis. To obtain the discriminative power of DEA, Dyson et al. [46] recommend being parsimonious with the number of inputs and outputs. According to these researchers, the number of DMUs should always be larger than two∙(#inputs + #outputs) to ensure sufficient discrimination between units. In this research, six inputs and seven outputs were chosen. Therefore, this rule was not violated.

Following Ozcan [47], we used three categories of inputs: capital investment, labor, and operating expenses. As suggested by Kohl et al. [19], a good proxy for capital investment is the number of beds (Beds). This variable is widely used in the literature [48,49,50,51]. As a proxy for labor, we used the number of hospital FTEs. Chilingerian and Sherman [20] note the relevance of labor to hospital efficiency. In addition, these researchers advise distinguishing between different types of personnel. Thus, we identified four categories of personnel: medical (FTE Med), nursing (FTE Nurse), administrative (FTE Admin), and technical (FTE Tech). As a proxy for operating expenses, we used the operating costs (Cost) of hospitals, which is a commonly used variable in the literature [52,53,54].

For the outputs, we used five variables to evaluate hospital efficiency and two variables to evaluate hospital quality. For efficiency-output variables for hospitals, Ozcan [47] advises including inpatient and outpatient visits. Thus, we used the number of outpatient visits (Outpatients) and the number of case-mix-adjusted inpatients (Adj. Inpatients). A commonly used output is operating revenue [55,56,57]. To measure this output, we used revenues from inpatient visits (Inpatient Revenue) and revenues from outpatient visits (Outpatient Revenue). Another variable commonly used to analyze hospital efficiency [89].

Limitations and further research

This study has some limitations. Beyond the typical limitations of basic DEA models [19], some issues are specific to this study. One is that we incorporated only a limited number of quality measures into the model, excluding measures such as hospital readmission rates because of the unavailability of data. In addition, an examination of demographic factors such as the age of population served could reveal significant information for the hospitals we analyzed. In addition, further research could extend this study by assessing hospital efficiency in other Italian regions.

Conclusions

This study contributes to the healthcare management literature because few previous studies in the European context have analyzed hospital efficiency using quality measures as outputs [36]. To fill this research gap, this research used hospitals in the Veneto region as a case study, identifying the organizational causes that affected efficiency, also considering quality measures, and how these changed over a period of two years. The results reveal that more than half of the hospitals under review were efficient. For the hospitals that were found to be inefficient, many had both input utilization and scale inefficiency. This study also provides empirical evidence for the main causes of inefficiency. Hospital size is one of the most important sources of inefficiency. In addition, administrative and technical staff often cause inefficiency. The contextual factors that most influence efficiency are the average patient length of stay and hospital size with respect to the population served. The role of technology is crucial to maintaining or increasing efficiency levels over time.

DEA is a good method for measuring hospital performance, in addition to the budgeting process that has traditionally been used in Italy. DEA is particularly useful for its ability to estimate the volume of inputs and outputs that can be optimized and its capacity to identify the main sources of inefficiency.

The results also underscore that a rethink of hospital size on the part of policymakers would be particularly valuable for increasing the current efficiency levels of many hospitals and maintaining constant and high service quality. Improving performance also depends on improving staff efficiency, particularly administrative and technical staff. This may be achieved by providing capacity building such as training staff members on efficient resource utilization. Finally, the results suggest that hospital managers should pay significant attention to advancements in technology and the skills needed to employ new technology in the best possible way.