Efficiency of Iranian Hospitals Before and After Health Sector Evolution Plan: A Systematic Review and Meta-Analysis Study

Purpose: Aging, chronic diseases, and development of expensive and advanced technologies has increased hospitals costs which have necessitated their efficiency in utilization of resources. This systematic review and meta-analysis study has assessed the efficiency of Iranian hospitals before and after the 2011 Health Sector Evolution Plan (HSEP). Methods: Internal and external databases were searched using specified keywords without considering time limitations. The retrieved articles were entered into EndNote considering inclusion and exclusion criteria, and the final analysis was performed after removing duplicates. Heterogeneity between the studies was assessed using Q and I2 tests. A forest plot with 95% confidence intervals (CI) was used to calculate different types of efficiency. The data were analyzed using STATA 14. Results: Random pooled estimation of hospitals technical, managerial, and scale efficiencies were 0.84 (95%CI = 0.78, 0.52), 0.9 (95%CI = 0.85, 0.94), and 0.88 (95%CI = 0.84, 0.91), respectively. Sub-group analysis on the basis of study year (before and after HSEP in 2011) indicated that random pool estimation of technical (0.86), managerial (0.91), and scale (0.90) efficiencies of Iranian hospitals for 2011 and before were better than technical (0.78), managerial (0.86), and scale (0.74) efficiencies after 2011. Conclusion: Type of hospital ownership was effective on hospital efficiency. However, HSEP has not improved hospital efficiency, so it is necessary for future national plans to consider all aspects.


INTRODUCTION
Hospitals have an undeniable role in providing healthcare services to society but their increasing costs have become an important challenge for many countries. In other words, utilization of technologies and new methods of diagnosis and treatment of diseases and also increasing numbers of elderly citizens, increasing chronic diseases, increasing demands for healthcare services and specialists, and hospital errors have increased health system costs (1,2). Because of these issues and problems, hospitals always encounter human and financial resource constraints which have necessitated efficiency in consuming resources more than ever (3).
The efficiency concept has been created from the combination of technical and allocative efficiencies. Technical efficiency means using the lowest amount of input to produce a specified amount Abbreviations: HSEP, Health Sector Evolution Plan; CI, confidence interval; DEA, data envelopment analysis; DMUs, decision making units; SSO, Social Security Organization; ISI, Institute for Scientific Information; SID, Scientific Information Database; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses. of output or using a specified amount of input to produce more output. Allocative efficiency means using the correct amount of input in terms of prices to produce a specified amount of output. Technical efficiency, on the other hand, was created by multiplying scale efficiency and managerial efficiency. Scale efficiency is the ability of an organization unit to perform in or near the most profitable scale to prevent loss in resources. Lastly, managerial efficiency means hard working, correct policymaking, application of the correct number of employees, and the correct combination of production factors (4).
One of the most widely used methods in assessment of different decision-making units (DMUs) such as hospitals and other organizations in terms of the components of efficiency (e.g., technical, scale, and managerial efficiency) is the data envelopment analysis (DEA) method. It is possible, through this method, to create a logical framework to distribute human and financial resources between different wards and sections of studied organizations (5). The DEA method, as a non-parametric programming technique, has been used since the mid 1980s to measure DMU efficiency (6). In other words, linear and multiple programming models are used in this method to assess the relative efficiency of a field, section, unit, or an organization, as a DMU, using multiple input and output indices (7). Numerous studies have assessed the efficiency of hospital efficiency using the DEA method. These studies can be divided into four categories. In the first category, the efficiency of university, teaching, and public hospitals, as the main providers of healthcare and therapeutic services, has been assessed in studies by Kalhor et al. (8) and Nabi lou et al. (9). In the second category, the efficiency of private hospitals has been studied and their efficiency has been compared with the firstcategory hospitals (10,11). The third category includes studies on hospitals affiliated with special entities such as Social Security Organization (12,13) and Armed Forces (14). The last category measures the efficiency of hospital wards such as radiology (15), dentistry (4), intensive care unit (16), and emergency (17) departments. Because the latter category studies wards of hospitals rather than the hospitals in their entirety and also have not assessed the technical, managerial, and scale efficiency of hospitals wholly, this category was excluded from the current study.
Although many studies have assessed the efficiency of hospitals using the DEA method in Iran, there has been no systematic review and meta-analysis study in this regard to present the final situation of hospital efficiency in Iran. By determining technical, managerial, and scale efficiency of Iranian hospitals, policymakers and planners can improve hospital efficiency through improving distribution and consumption of resources.
The extensive review of the literature by the authors of the current study has resulted in four systematic review and metaanalysis studies on Iranian hospital efficiency using the DEA method. The first study assessed studies in terms of the provinces where they were performed, whether they were input-or outputoriented, and whether they were fixed or variable return to scale models (18). The researchers in another two systematic and metaanalysis studies discussed the methods used to assess hospital efficiency (19,20). The last study only included a small number of studies on hospital efficiency and did not mention the efficiency subdimensions namely scale, managerial, and technical efficiency (21). As previous systematic review and meta-analysis studies have not assessed hospital efficiency using its subcategories, the current study assessed technical, managerial, and scale efficiency of hospitals through systematic review and meta-analysis.
Regarding PICOS framework or questions, the study included hospitals in Iran which had previously had their efficiency assessed and were entered into the study depending on the inclusion and exclusion criteria. The intervention framework was the assessment of the effect of HSEP on hospital efficiency, comparisons included comparing hospital efficiency before and after HSEP, outcomes included the amount of hospital efficiency, and finally the study design included assessment of hospital efficiency through systematic review and meta-analysis.

Search Strategy
The international databases of the Institute for Scientific Information (ISI), PubMed, Scopus, Google Scholar, and Persian databases of Scientific Information Database (SID), Magiran, and Barakat were searched using the combination of "efficiency, " "hospital, " "data envelopment analysis, " "DEA, " and "Iran" keywords in 2018. The references of the retrieved articles were searched to increase the study credibility and precision.

Inclusion and Exclusion Criteria
All published Persian and English language articles about hospital efficiency with a score between 8 and 12 were entered into the study without considering a time limit. If several formats of a research were published (such as a book, article, report, and so on), only one of them was entered into the study. Inputoriented studies were entered into the study. Short reports, letters to editors or editorial comments, one study that was available in two languages, studies on health care facilities other than hospitals, and studies on internal parts of hospitals were removed from the study. Two researchers assessed and extracted data from the studies independently and a third researcher resolved disagreements if they appeared.
This systematic review and meta-analysis utilized Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to minimize potential sources of bias (22).

Data Collection
A researcher-made checklist in an Excel spreadsheet was created to extract the studies' data including the first author name, year of data collection, place of study, language, sample size, and the score of technical, managerial, and scale efficiency. Another checklist designed previously, whose credibility has been proved by numerous studies, was used to assesses the studies' quality (19,21). This checklist includes 12 questions regarding the study aim, method, data collection, sample size, and study   population. Each question has the score between 0 and 1 and the score for each study is calculated by summing the scores of questions. So that the studies with scores between 8 and 12 were entered into the final analysis. The study protocol was approved by the Ethical Committee of Kerman University of Medical Sciences.

Data Analysis
Efficiency types were considered as a proportion in this study. Therefore, the numerator was the sum of technical, managerial, and efficiency scores and the denominator was the number of study hospitals. Heterogeneity between the studies was assessed using Q and I 2 tests. A P-value lower than 0.05 for the Q-test and an I 2 higher than 50% were considered as the measures of studies' heterogeneity. Because the studies were heterogeneous, the random effect model was used to estimate hospital efficiency. A forest plot with 95% confidence intervals (CI) was used to calculate different types of efficiency. Egger's and Begg's tests were used to assess publication bias. In order to assess the effect of the 2011 Iran Health Sector Evolution Plan (HSEP) (23) on hospital efficiency, the studies before and after it were compared. The data were entered into Excel 2016 to be edited and then transmitted and analyzed using STATA v.14.2.

RESULTS
Each one of the scientific databases were searched on the basis of a recommended search strategy by the databases themselves using defined keywords. For example, in the PubMed database, 23 articles were retrieved after placing the search query. Search query used for PubMed was: (((data envelopment analysis) OR DEA) AND hospital) AND Iran))).  Frontiers in Public Health | www.frontiersin.org articles remained. Also, the assessment of references of these articles resulted in two new articles. In this way, 49 articles were entered into the final step of the systematic review and metaanalysis. Twelve articles (24.48%) of these were in the Persian language and the remaining were in the English language. A PRISMA flow chart of the study retrieval and selection process with reasons for exclusion at each stage is provided in Figure 1. As mentioned before, each type of efficiency was entered into the meta-analysis separately, so that 50 studies for technical efficiency, 36 studies for managerial efficiency, and 41 studies for scale efficiency had entry requirements to the analysis.
The studies were performed from 1996 to 2016. After performing all the steps of study selection, 49 articles were entered into the final phase of the study. The number of hospitals assessed in these articles ranged from 4 to 122. The inputs considered in the studies included number of beds, number of operation rooms, physicians, nurses, support forces and other human resources, costs, education hours, and working days. The outputs were number of surgeries, outpatients, occupancy rate, bed days, admission, inpatients, surgeries, emergencies, bed turnover, mean patient stay, hospital income, bed occupancy rate, SERVQUAL score, number of clinical, paraclinical, and outpatient services, number of discharged patients, number of contracted insurance companies, access to emergency room, confronted with hospital infections, anesthesia problems, employee consent, active to fixed bed ratio, number of deaths, and patient-day. Two studies assessed charity hospitals, four studies assessed private hospitals, and five studies assessed Social Security Organization (SSO) hospitals. The remaining studies assessed hospitals affiliated with universities of medical sciences belonging to the Iran Ministry of Health ( Table 1).
The results indicated that there was heterogeneity in studies related to technical efficiency (heterogeneity chi 2 = 156, p < 0.001), managerial efficiency (heterogeneity chi 2 = 79.58, p < 0.001), and scale efficiency (heterogeneity chi 2 = 67.22, p < 0.001). I 2 index in technical and managerial efficiency was higher than 50%, which indicates high heterogeneity between the studies. This index was lower than 50% for scale efficiency.
Study publication error using Egger's test indicated that there was publication bias in technical and managerial efficiencies (P < 0.001), but there was no publication bias in scale efficiency (p = 0.19). Table 2 displays the results of Egger's testing for the three types of efficiencies. Begg's test indicated that there was no publication bias in the three types of efficiencies (P < 0.001).
The results indicated that technical efficiency of Iranian hospitals had high variation, so that it ranged from 0.34 in the Mahfoozpor et al. study to 1 in Raeisian et al. and Najafi et al. On the basis of random effects modeling, random pooled estimation of hospital technical efficiency was 0.84 (95% CI = 0.52, 0.78) (Table 3, Figure 2). The managerial efficiency of Iranian hospitals was between 0.59 in the Aboulhalaj et al. study and 1 in studies    Figure 4). The results of technical, managerial, and scale efficiencies are presented in Tables 2, 4, 5, respectively. In addition, the forest plots for technical, managerial, and scale efficiencies are presented in Figures 1-3, respectively. Sub-group analysis based on study year indicated that random pool estimation of technical efficiency of Iranian hospitals for 2011 and before and after 2011 was 0.86 (95% CI = 0.80, 0.91) and 0.78 (95%CI = 0.64, 0.89), respectively. The status of managerial efficiency for 2011 and before was better than after 2011 (random pool estimation equal to 0.91, compared to 0.86). Random pool estimation of scale efficiency for 2011 and before was 0.90 (95%CI = 0.86, 0.93), while random pool estimation of scale efficiency for after 2011 was 0.74 which is lower (95%CI = 0.61, 0.86) ( Table 6).

DISCUSSION
The assessment of hospital efficiency provides the groundwork to assess their performance and increase productivity when using limited resources. One of the ways of assessing allocated resources to obtain specified goals is efficiency studies. In summary, efficiency means using resources to their maximum to produce goods and services (61). This is the first systematic review and meta-analysis study regarding assessment of the efficiency of Iranian hospitals in terms of its subcategories namely technical, managerial, and scale efficiencies. Different methods have been used to assess Iranian hospital efficiency such as DEA, Pabon-Lasso, and Stochastic Frontier Analysis (SFA) in past decades (21). In this regard, as this study indicates, the DEA method is the most widely applied method to assess hospital efficiency (19).
Our findings showed that the random pool estimations of technical, managerial, and economics of scale efficiency were 0.87, 0.9, and 0.88, respectively. This finding indicates that the resources of the studied hospitals in Iran have been used in an inefficient way. One idea about hospital efficiency is that the expectation from hospitals to work efficiently is far from reality. The reasoning for this claim is the economic theory of firms that declare the hospitals cannot work at full efficiency because of uncertainty in costs and prices of services that they provide. In summary, lack of information on costs and prices is one of the main factors that has a negative effect on hospital efficiency (6,62).
Most of the studies were implemented in Tehran province (13 studies). Four studies investigated the efficiency of hospitals across all provinces of Iran. However, some provinces such as Sistan and Baluchistan had no individual reports about the efficiency of their hospitals. Therefore, there is an information gap for health policymakers and hospital managers in this field.
As the results indicated, most of the researchers tended to perform analyzes through the input-oriented method, because inputs are in the control of hospital managers, so that by creating changes in the inputs can change the rate of outputs to the desired extent. However, it is suggested that private and for-profit hospitals are excluded from this rule, because the managers of these type of hospitals want to maximize outputs and, as a result, hospital profits (63).
Human and capital resources such as the number of nurses and physicians and the number of beds were the main inputs in   (64). Furthermore, better management of human resources is associated with higher patient outcomes without significant increases in the cost of hospitals (65). The results indicated that most hospital efficiency studies suffer from some weak points. Therefore, the selection of inputs has been performed on the basis of resource review (e.g., previous published articles) not consideration of each hospital situation. Also, the inputs were not weighted, so that the resources with high specialty and expenditure receive the same weight as others. Hospital case mix has not been considered in this hospital efficiency assessment. This leads to low efficiency assessment in hospitals which have the most complicated cases. Lastly, some studies have not considered the data validity and the appropriate ratio of inputs and outputs with the number of hospitals precisely.
The study of Contor VJM and Poh Kl also provides some theoretical and methodological limitations of the DEA method in capturing the full view of efficiency of healthcare centers (66). However, with a suitable study design, the DEA method is among the most important and most applicable methods in the assessment of health system efficiency, especially hospitals (67).
The results indicated that technical, managerial, and scale efficiency of Iranian hospitals after performing HSEP decreased in comparison with before it. A study on Turkey hospitals from 2001 to 2006, which measured the effect of Turkey health sector reform on hospital efficiency to provide policy implications for policymakers, indicated that this reform had increased the efficiency of public hospitals but the efficiency of private hospitals had decreased (68).
As there was no hospital with full efficiency in the study and the increasing trend of health system costs and scarce resources, it is proposed to design and implement a system to monitor efficiency and consumption of resources especially in hospitals. This can help to identify inefficient hospitals and the causes of it. Health policymakers through cost management planning and increasing outputs can pave the way in this regard.

Strengths and Limitations
This is the first comprehensive systematic review and metaanalysis evaluating efficiency of Iranian hospitals which is applicable for comparison of the efficiency of hospitals before and after HSEP. The methodology adhered to the PRISMA statement (22).
The strength of the study is in performing meta-analysis after the systematic review which has specified the exact amount of technical, managerial, and scale efficiencies of Iranian hospitals. The Cochrane Consumers and Communication Review Group's data extraction template (69) was used to obtain the needed information about the studies included. Nevertheless, the retrieved studies were mainly administered in some easily accessible areas rather than in a balanced distribution all over the country. This limits the generalizability of the results.

CONCLUSION
This study indicated that many hospitals are inefficient. This implies that there is considerable room for efficiency improvement in hospitals. Hospital management has a unique role in this regard. Health systems have reformed in spite of increasing access and utilization of patients to the services, but have not considered efficiency improvement in hospitals. So, health policymakers and hospital managers should design and implement some related programs in order to monitor and improve the efficiency of hospitals.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.