Is aliation system successful in healthcare at Turkey?

The main objective of the study was to compare technical eciency between aliated and non-aliated hospitals of Turkey. There was six otput variables and four input variables to evaluate hospitals eciencies using with Ministry of Health (MoH) data from 2013 to 2017. Totally, 25 aliated hospitals and 53 non-aliated hospitals were analysed using input-oriented Data Envelopment Analysis (DEA) model under variable return to scale assumptions. The study results that, aliated hospitals eciency scores were higher than non-aliated training and research hospitals. These results showed us the application of the aliation system was successful in Turkey. Furthermore, this is the rst study which has been performing whole aliated and non-aliated hospital eciency measures together in Turkey and it offers an original contribution to literature with this direction.

Health System Reform program. Farzianpour et al. (2011) studied the assesment of performance of a liated hospitals of Tehran University using with Baldrige Excellence Model. Goudarzi et al. (2014) calculated the technical e ciency of a liated hospitals in Iran between 1999-2011 using with Stochastic Frontier Analysis. Farzianpour et al. (2012) analyzed performance of a liated hospitals in Tehran University with DEA. İlgün et al. (2019) worked on training and research hospital in Turkey. They examined only 47 units training and research hospitals, in spite of being 78 training research hospital in Turkey. Another study in Turkey is realized by Küçük et al. (2019). In this study, they did the provincial and regional analysis with the e ciency of 669 public hospitals of Turkey with DEA and present the relationship between health reform program and their analysis results.
In this study, technical e ciencies of a liated and non-a liated hospitals are examined by several dimensions using an input-oriented DEA model under the variable return-to-scale assumption. There are 78 training and research hospital in Turkey. The 25 of them are a liated and 53 of them are non-a liated hospitals. Therefore, this study using data collected from MoH Public Hospitals Directorate General from 2013 to 2017 for providing to evaluate the e ciency of 78 training and research hospital in Turkey with a liated and non-a liated discrimination.

Methods
One of the methods used to evaluate the performance of Decision Making Units (DMU) using with several outputs and inputs is DEA. In the literature, CCR model which was putforwarded by Charnes et al. (1978) and BCC model by Banker et al. (1984) are the common DEA models. DEA based on a non-parametric mathematical programming can measure the performance of DMUs. It calculates a score which is limited to 0 to 1 for each DMU. DMU indicates e cient with these scores is equivalent to 1. Otherwise, DMU is ine cient (Charnes et al., 1978). When the production process of a DMU is complex, due to imposes of main objective function of DEA is preferred rather than stochastic frontier, one of other approach without any limitations (Aigner et al., 1977).
Some of the important publications which is interested in hospital e ciencies are examined to understand hospital performance better. The country, number of hospitals and years information are reviewed owing to comprehend the time period, country coverage and working size of the studies. Also, method and orientation which is used for analysing the e ciency of the hospitals is presented in Table 1. Either developing or developed countries included from China to Spain, USA to Italy, Colombia to Taiwan, Iran to Greece in the summary of the systematic review process on hospital e ciency. According to the size of the hospital, the number of DMUs in these studies ranged from 16 to 669 DMUs. According to Input -Output oriented model status, the studies before 2010 were mostly oriented towards input-oriented DEA models, but in recent years number of outputoriented models was higher, for the concepts of cost ine ciency and especially resource e ciency became more important until the end of 2000s. The cases focusing on quality, such as patient satisfaction, are used in the studies. According to recent studies such as Gimenez et al. (2018), hospital e ciency has extended productivity analysis to include minimization of output ine ciency. One of the important issues of measuring e ciency in the hospitals is appropriate inputs and outputs. According to literature, inputs were commonly used by researchers in analysing hospital technical e ciency include the number of doctors, nurses, beds and total expenditure. The number of total patients and inpatients, average daily admission, the number of surgeries, bed occupancy rate and total revenue are commonly the outputs of the hospitals.
Several dimensions were used for measuring and assessing of hospital performance. Output and input variables identi cation process also the most important issue in DEA context. Output and input variables for hospitals in DEA have been determined with preceding empiric and theoretical studies which is given in Table 1 to capture the performance as accurately and comprehensively. There are six output and four input variables in this study are described below.
Output variables:

Results
Output and input variables summary data is shown at the Table 2. The inputs of number of doctors, number of beds, number of MRI units and scanners are increased 10-25% in these ve years period. This means that the physical, technical and labour opportunities of the hospitals are become better. However, total expenditures of the hospitals are increased more than these inputs with 74% increase. On the other hand, the total revenue of the hospitals are increased 50%. This means that, the hospitals costs are higher than their revenues.
In addition, despite the increase in human resources and physical infrastructure, it is seen that the expected increase in output is not realized with total inpatients variables in operation and hospitalization. This indicates with physicians concentrate on outpatient services rather than labour intensive services based on inpatient and surgery.
In 2017, while minimum number of bed was 85 as well as maximum number of bed was 1500 and the number of doctor was minimum 13, maximum 515. In this case, it is seen that a liation application is not a standard application and it cannot be transformed into a completely similar structure in terms of scale and capacity. Women's Maternity and Children' s hospitals have been e cient in 5 years. It is also observed that the nancial situation of public hospitals has been decreasing in parallel with the general increase in expenses since 2016.  The trend of training and research hospitals between a liated hospitals and non-a liated hospitals appears to be consistent at from 2013 to 2017 in Fig. 1. However, hospitals that are non-a liated; it has shown a lower performance than other training and research hospitals due to the fact that the structuring of educational staff, physical and infrastructural conditions take a long time, and the income-expense balance resulting from the increase in additional payment is deteriorated.
Average technical e ciency comparison of a liated and non-a liated hospitals and percentage difference between a liated and non-a liated hospitals were shown in Table 5. Average technical e ciency of a liated hospitals were increased 4,20% from 2013 to 2017. Average technical e ciency of nona liated hospitals were increased 17,19% from 2013 to 2017. These rising trends showed us the Turkish health system become more higher performance than before. Also, from 2014 to 2017 the two type of hospitals average technical e ciency difference were approximately 2% except 2013. E ciency results of a liated hospitals were higher than those of non-a liated training and research hospitals. Training and research hospitals which is a liated to have more organized structure in the production of services is an important point advantage, considering that the general hospital also they have with equivalent effectiveness analysis was performed to compare the general hospital. In this context, the hospital which inputs and outputs similar with the a liated hospitals, it is seen that the performance of general hospitals are higher than others. This shows us that the a liation model yields a positive result since these hospitals are predicted to have similar e ciency scores.

Discussion
In this study, at rst Turkish Health System and A liations System which was implemented as new system in Turkey was examined in the concept of Health Reform Program. Also, it is mentioned that the condition of the a liation system and why we need to use this system in Turkey. Then, literature review of measurement of a liated hospital e ciency studies conducted in various countries is examined. DEA was used as a methodology in the study. All training and research hospitals in Turkey which a liated and non-a liated have been analysed and special evaluations have been made for these hospitals between 2013 and 2017. So that, presenting more realistic results of the analysis overall training and research hospitals included in analysis.
While the previous system has more regular and organized, the a liation model is very new in Turkey. So that, we have to need a time for understand of the effect of the public. Nevertheless, with this system all patients are able to consult the academic staff of the university and owing to this reason the academician expertise more area with these all types of patient patterns. For all that, health service suppliers under pressure and the instruction cost of new hospital building are another invisible cost of this system. However, this situation needs more time and there is lack of any student education processes in these time period and also the opportunity cost become another important phenomenon.
The main limitation of the study is that before and after period of a liation was not examined separately. If such an examination is made in future studies, the bene ts of a liation system for university hospitals will be revealed in more detail. In addition, the two-stage Network DEA method is planned to use for analysing the "education and training" function and "service delivery" stages separately with their own parameters and to present a different perspective on the a liation system.

Conclusions
A liation model which began in 2009 as a pilot model in Turkey, expands to the opening of new medical schools with parallel policies has reached 25 in 2019. Thus, without any capital investments, the newly established medical faculties and public hospitals carry out health services and training and research activities together. According to a comprehensively analysis, the e ciency scores of the a liated hospitals were higher than those of non-a liated training and research hospitals which are included all Turkish training and research hospitals. Herewith, status of the a liated hospitals will be more clearly revealed. The desired results could be achieved from the gradual augmentation application.
Continuous changes in the a liation regulations, the legal infrastructure of the model is fully established, also academic and administrative problems are proof of this system di culties both for the university and the MoH hospitals. The e ciency analysis conducted within this framework con rms these ndings and results. All the results of this application positively affected. In general, it is possible to evaluate the effectiveness of the model in terms of e ciency scores.
Recently, there are some news about conjoint to several university hospitals to MoH hospitals in Turkish Government agenda. In addition, this model is expected to continue towards the merger of university medical faculties, which faced especially nancial problems, and MoH hospitals. For this reason, policy makers are advised to take precaution for regarding the function of model and structural problems. For example, the fact that hospitals remain between the "education and training" function and the "service delivery" function may cause con ict between university and hospital staff. Also, in order to eliminate this problem, the process of signing contracts with lecturers in hospitals and university staff is required to be initiated.
It is suggested to compare the a liation system with longer time series, and to make impact analysis studies that enable before and after period in which model is realized in two separate planes. And this study is the rst study that have been performing all a liated and non-a liated hospital e ciency measures in Turkey with an original contribution to the literature with this direction.

Declarations Ethical Approval and Consent to participate
This study don't need any ethical approval because no use of human and animal participant.

Consent for publication
Not applicable.

Availability of supporting data
Datasets that are restricted and not publicly available

Competing interests
The author declare that they have no competing interests.

Funding
This study was not funded by any private rm or public institution.

Authors' contributions
There is only one author in this paper. So that all part is designed by this author.