key: cord-1024461-xuwoqy18 authors: Ortiz-Barrios, Miguel; Gul, Muhammet; López-Meza, Pedro; Yucesan, Melih; Navarro-Jiménez, Eduardo title: Evaluation of hospital disaster preparedness by a multi-criteria decision making approach: The case of Turkish hospitals date: 2020-07-05 journal: Int J Disaster Risk Reduct DOI: 10.1016/j.ijdrr.2020.101748 sha: 849b8f5c31a062dba7a98c60a12153f4d4e68cde doc_id: 1024461 cord_uid: xuwoqy18 Considering the unexpected emergence of natural and man-made disasters over the world and Turkey, the importance of preparedness of hospitals, which are the first reference points for people to get healthcare services, becomes clear. Determining the level of disaster preparedness of hospitals is an important and necessary issue. This is because identifying hospitals with low level of preparedness is crucial for disaster preparedness planning. In this study, a hybrid fuzzy decision making model was proposed to evaluate the disaster preparedness of hospitals. This model was developed using fuzzy analytic hierarchy process (FAHP)-fuzzy decision making trial and evaluation laboratory (FDEMATEL)-technique for order preference by similarity to ideal solutions (TOPSIS) techniques and aimed to determine a ranking for hospital disaster preparedness. FAHP is used to determine weights of six main criteria (including hospital buildings, equipment, communication, transportation, personnel, flexibility) and a total of thirty-six sub-criteria regarding disaster preparedness. At the same time, FDEMATEL is applied to uncover the interdependence between criteria and sub-criteria. Finally, TOPSIS is used to obtain ranking of hospitals. To provide inputs for TOPSIS implementation, some key performance indicators are established and related data is gathered by the aid of experts from the assessed hospitals. A case study considering 4 hospitals from the Turkish healthcare sector was used to demonstrate the proposed approach. The results evidenced that Personnel is the most important factor (global weight = 0.280) when evaluating the hospital preparedness while Flexibility has the greatest prominence (c + r = 23.09). Disaster incidents are one of the most endangering events in human life and emergency solutions are needed because of their sudden occurrence (Hosseini et al. 2019) . As such events have devastating effects; they may cause some disruptions in the society to meet their health needs. Man-made and natural disasters can interfere with the activities of some organizations such as healthcare facilities. Sometimes the capacity of these facilities may not be sufficient to combat the physical and financial damage they may cause. Although the ability to be ready for disasters varies by country, it can be said that in the last century, when the frequency and devastating effects of disasters have gradually increased, no country is fully prepared and safe. Turkey, which has experienced many devastating man-made and natural disasters, faces with 8 disasters per year (natural and technological) on average. These resulted in 1043 death and 2937 injured annually on average according to the figures in the Emergency Events Database (EM-DAT). The disaster trends of Turkey are demonstrated in Figure 1 . According to these statistics, these disasters have caused a serious total damage in terms of financial loss. Turkey has passed a devastating earthquake in on 17th August, 1999 as can be easily inferred from the Figure 1 . Also, in the following years after 1999, earthquakes, mine accidents, floods, transport accidents, storms, epidemics, landslides, and miscellaneous industrial accidents have occurred in Turkey. All of these disasters, especially earthquakes due to their destructive effects on human loss and financial damage, cause various problems in the operation of hospital operations and health services. Nowadays, an overwhelming majority of the world is still fighting against a pandemic. Covid-19, a new type of coronavirus, is an infectious disease that first appeared on December 30, 2019 in Wuhan, China. The World Health Organization (WHO) declared COVID-19 later as an epidemic. In nearly 4 months, the infection, which first expanded with Iran and Italy, is spread all over the world. As of 22 April, 2020, it has caused nearly 2.5 million cases and 169,000 deaths in the world (WHO, 2020) . The world has faced an unusual number of infected people in hospitals. This surge made it mandatory for countries to make their hospitals, field hospitals, or specific pandemic hospitals prepared in a considerable and short time. Therefore, it is required to prepare the hospitals and all healthcare stakeholders for the disasters. In Hosseini et al. (2019) , preparedness is defined as the inclusion of activities set up to build a mechanism for rapid responses to limit the risks and effects. One of the crucial components of improving disaster preparedness is to evaluate the readiness of hospitals and then propose a ranking of them. Hospitals should be prepared for physical infrastructure and resource planning, as they are the only places to provide first care during a disaster. The postdisaster problems encountered by hospitals are stated in the literature as follows (Gul and Guneri, 2015; Dursun et al. 2012 ): A surge in patient arrivals, communication issues, lack of adequate treatment and care area, problems in patient transfer. Keskin and Kalemoğlu (2002) pointed out that the most important post-disaster problem faced by hospitals is the excessive patient admission. Similarly, the lack of coordination between disaster area and hospitals (Kurt et al. 2001) , lack of telecommunication (Ishii and Nakayama, 1996) , chaos, and triage difficulty due to patient surge (Yamauchi et al. 1996) are some of the other problems raised in the current literature. From these studies, it is inferred that disasters directly affect hospital activities. Since the occurrence of disaster events causes mass casualties, public panic and chaos. Most hospitals are unprepared to tackle thousands of patients arrived in case of a major disaster (Foxell and McCreight 2002) . Therefore, planning of hospital resources (treatment area, equipment and personnel) and determination of disaster preparedness levels of hospitals can be effective in reducing the damage caused by disasters. On the other hand, another issue related to disaster management is the importance of conducting risk analysis and loss assessment studies for urban areas (Pitilakis et al. 2011; Zuccaro and Cacace, 2011; Fawcett and Oliveira, 2000) . These studies provide insight into the accessibility of hospitals as well as the assessment of losses and thus how they can affect the needs of healthcare services after a disaster. In this study, a hybrid fuzzy decision making model was proposed to evaluate the disaster preparedness of Turkish hospitals. This model was developed using three well-known multicriteria decision-making (MCDM) methods named as FAHP, FDEMATEL and TOPSIS. The ultimate target of the study is to determine a ranking among hospitals in terms of disaster preparedness. FAHP is used to determine weights of six main criteria including hospital buildings, equipment, communication, transportation, personnel, flexibility and thirty-six sub-criteria under these six main criteria. AHP contains some important characteristics such as pair wise comparison, hierarchy, independency and consistency in decision making. Since, FAHP is frequently used as a weighting tool in the literature; we use it in this study with FDEMATEL which has ability to evaluate interdependence between all the sub-criteria of the model. Finally, TOPSIS is used to obtain the ranking of hospitals. To apply the hybrid decision making model, two different questionnaires are created and assessed by five experts related to the field of disaster management of healthcare facilities. To provide inputs for TOPSIS implementation, some key performance indicators are established and related data is gathered by the aid of experts from the assessed hospitals. The remaining of the study is organized as follows: Section 2 provides a review of the literature covering conceptual and research articles regarding the topic of hospital disaster preparedness evaluation. The third section gives the methodology used for the study. Initially, applied MCDM methods are described then, the proposed hybrid approach is identified. Fourth section concerns with a demonstration of the proposed approach. The evaluation criteria, the decision-making team, implementation of the questionnaires and results of each MCDM method are presented in this section. Final section concludes the study with some future research recommendations and limitations of the current work. Various conceptual-based and MCDM-based frameworks are proposed in the literature to assess the preparedness level of hospitals (Hosseini et al. 2019; Gul and Guneri, 2015; Top et al. 2020 ). In conceptual-based frameworks, researchers construct their disaster readiness frameworks without using any numerical tools. They directly focus on some dimensions such as structural, non-structural, functional, and human resources. On the other hand, in MCDM-based frameworks, researchers initially determine the main objective, disaster preparedness criteria and sub-criteria, alternative hospitals that will be assessed with respect to concerning these criteria/sub-criteria (the decisional hierarchy). Then, a decision matrix is built that includes a weight matrix of criteria/sub-criteria. Using this decision matrix, a final decision is made using an MCDM framework such as TOPSIS. In the literature, the number of conceptual-based frameworks and cross-sectional surveybased studies are more than MCDM-based frameworks. As an example, Hosseini et al. (2019) proposed a TOPSIS-based ranking model for eight selected Iranian hospitals. They ranked these hospitals in terms of disaster preparedness ability. Four crucial criteria of structural preparedness, non-structural preparedness, functional preparedness, and human resources are taken into consideration in their study. The weights of these main criteria are assumed as 40%, 25%, 10%, and 25%, respectively. No MCDM method has been followed to determine weights to these criteria. Results The results of the study suggest that any of the observed hospitals are not in on a well-prepared level. In another study, proposed an analytic decision-making preference model for the disaster preparedness of hospital emergency rooms in Colombia. The MCDM model includes three well-known methods of AHP, TOPSIS and DEMATEL. Gul and Guneri (2015) made a conceptual study regarding the preparedness capability of hospital emergency rooms in Istanbul, Turkey. They combined the data of reports, literature, and one-to-one interviews with experts who experienced the recent Istanbul earthquake. The interviews are analysed in terms of some earthquake key statements. The important issues faced during the earthquake from the viewpoint of hospital care are extracted. Top et al. (2020) applied a questionnaire about hospital disaster planning in a total of 430 Turkish hospitals. A number of 32 questions are included in the study. Focus The focus of the questions are is about having any written hospital disaster plan and whether performing an exercise on an annual basis or not. The characteristics of hospital disaster plans are queried with respect to concerning the hospital category (public, private, university and all). Unlike the above-mentioned studies, there exist many papers using a cross-sectional methodology, Delphi, and similar tools (Rezaei and Mohebbi-Dehnavi, 2019; Saeid et al. 2019) . For a broad literature review in disaster preparedness of hospitals, scholars refer to the paper of Alruwaili et al. (2019) . Tabatabaei and Abbasi (2016) performed a semi-quantitative cross-sectional study in some Iranian hospitals to assess risks during disasters based on the hospital safety index. Two different questionnaires were designed for collecting 145 metrics (structural, functional, and nonstructural factors) supporting the evaluation of hospital disaster readiness are the general information of hospitals and disaster ability byThis is not fully numerical or MCDM-based study. Otherwise, it benefits from a semi-quantitative disaster preparedness assessment method. Similarly, Naser et al. (2018) made a cross-sectional study to assess hospital disaster preparedness of 5 public and 5 private hospitals in South Yemen. are included to the study. Results The results of the study demonstrated that hospitals are in an unacceptable level of readiness. Samsuddin et al. (2018) investigated disaster preparedness attributes and hospital's resilience in Malaysia. A cross-sectional questionnaire has been performed among Malaysian hospitals' staff considering a total 243 preparedness attributes and 23 resilience indicators are used in the study. The results showed that human resources & training and the ability to adapt in a timely manner promptly are were ranked as the most critical attributes. The results can help hospital's stakeholders in Malaysia to improve the its preparedness capability. Marzaleh et al. (2019) developed a model for hospital emergency room preparedness against radiation and nuclear incidents as well as nuclear terrorism in Iran. By utilizing the Delphi method, 31 criteria are considered under three main classes, namely staff, stuff, and structure (system). In conclusion of the results, staff preparedness and stuff preparedness had the highest and lowest priority levels, respectively. Shabanikiya et al. (2019) designed a tool for hospital preparedness for surge capacity during disasters. They used the Delphi method in their developed toolkit as in Marzaleh et al. (2019) and assessed 64 components in five categories and 13 sub-categories. The developed tool was used in evaluating hospital preparedness for surge capacity in disasters and planning the future of hospitals against disasters. The current study is differentiated from both studies mentioned under the class of MCDMbased frameworks in some aspects. The first difference concerns the comprehensiveness of the hospital disaster preparedness criteria set. While Hosseini et al. (2019) considered four preparedness dimensions as mentioned above, constructed the hierarchy under seven criteria. In Ortiz- , the decision-making team identified 7 criteria and 23 sub-criteria to evaluate the readiness of emergency departments for a disaster situation. The criteria set of our study include hospital buildings, equipment, communication, transportation, personnel, and flexibility. These criteria are more inclusive than the two studies and can easily be adapted to potential models that can be developed later. Moreover, Ortiz-Barrios et al. (2017)'s study is developed for specific emergency departments. A main dimension similar to the one we use in this study and which we called "flexibility" is not mentioned in Hosseini et al. (2019) 's study. The second difference stems from methodological sides. In Hosseini et al. (2019) , an assumption and full subjectivity are preferred in weighting the main criteria. The assignment of criteria weights is not clear. Any weight assignment is not performed for sub-criteria. Unlike this, we follow a group-decision making procedure via an experienced decision team in weighting both criteria and sub-criteria using FAHP. Besides, none of the studies proposes a fuzzy-based MCDM approach. However, by integrating linguistic expressions and corresponding fuzzy numbers into our approach, we aim to eliminate the insufficiency and preciseness of the crisp pairwise comparison in classical AHP in capturing the right judgments of decision-makers. As discussed above, many studies are existed have been undertaken for assessing hospital disaster preparedness. Most of the papers contribute to the literature by proposing conceptualbased frameworks that we discussed. On the other hand, cross-sectional questionnaire-based models that suggest new attributes regarding hospital disaster readiness and review papers that provide a comprehensive overview to of the topic are also widespread. Also, our brief review in this section shows the importance of hospital disaster preparedness assessment from the MCDM viewpoint of MCDM throughout the literature. It can be observed that there exist is a wide ranges of MCDM methods used in the literature with applied application to various areas. So far, however, there have been limited papers regarding the use applications of FAHP, FDEMATEL, and TOPSIS in hospital disaster preparedness. By doing the current study, we aim to provide some contributions to the literature. These are as follows: 1. We developed a model specifically dealing with hospital disaster readiness assessment and ranking. This topic has been also addressed in other studies considering a number of service-quality criteria ; our model, however, incorporates On conclusion of the analysis of Upon analyzing the published papers, it is observed that our study covers a number six disaster readiness criteria (hospital buildings, equipment, communication, transportation, personnel, and flexibility) and thirty-six sub-criteria representing the entire context of hospital disaster management. Under six main headings of, thirty-six sub-criteria under these six main criteria are determined. Most of the criteria are proposed and defined for the first time in the literature. 2. We developed a three-MCDM-integrated approach that includes FAHP, FDEMATEL, and TOPSIS. FAHP is was firstly used In for determining the initial weights evaluation criteria of hospital disaster preparedness criteria. Furthermore, The FDEMATEL method can derive the weights considering was then applied for assessing the interrelations among criteria. Therefore, we also applied it to the determination of criteria weights. We have combined these two fuzzy MCDM methods with TOPSIS in this study to rank hospitals in terms of readiness. In view of the characteristics of all these methods either individually and in integrated style, the proposed approach can handle the problem in a systematic and analytical manner. Indeed, the hybrid approaches can tackle the limitations that single methods hold (Zavadskas et al., 2016; Ortíz-Barrios, 2017) . For instance, TOPSIS uses criteria weights that are usually defined randomly. FAHP, specialized in prioritizing decision elements (Saaty, 1988) , has been therefore proposed to address this drawback. On the other hand, FDEMATEL has been incorporated into this approach since FAHP is unable to evaluate interdependence among criteria. Although ANP can be also used to this aim, its application has been proved to be time-consuming and highly complex, especially in models with a significant number of criteria and sub-criteria (Kumar and Haleem, 2015) . On a different tack, TOPSIS was deemed as a suitable method for ranking the hospitals according to their disaster preparedness level. This technique was preferred over AHP since the latter method entails many pairwise comparisons given a large number of hospitals. Although a pilot application including 4 hospitals is presented in this study, it is noteworthy that the proposed approach has been projected to be used at a national level in Turkey where a significant number of alternatives needs to be considered. DEA can be also employed for this particular aim; it, however, assumes that inputs and outputs are known which cannot be fully ensured in all disaster readiness criteria (Velasquez and Hester, 2013) . In Table 1 , we present a comparison with the four aforementioned studies based on some specific aspects. It can be seen from Table 1 that the proposed approach meets all three important aspects (pairwise comparison criteria, fuzziness in determination of criteria weights, and interdependence evaluation between criteria), and is more satisfactory in terms of quantity of criteria and coverage of the criteria. The rest of the approaches only cover two aspects at the most. Based on the previous considerations, the novelty of this study is as follows: a) The inclusion of fuzziness in the calculation of criteria and sub-criteria weights using the FAHP method. This is motivated by the need for dealing with the imprecision and uncertainty of linguistic evaluations which makes the model more realistic and coherent with the real scenario. Besides, this aspect has not been addressed by the previous related studies as evidenced in Table 1 . b) The evaluation of fuzzy interdependence between disaster readiness criteria through FDEMATEL for facilitating the design of long-term improvement plans by the government and other stakeholders. Similar to the point a), this aspect has not been dealt in the reported related literature. c) The incorporation of several criteria that have not been considered in other related studies (i.e. flexibility and contingency staff). A six-step procedure (Fig. 2) is proposed to evaluate the hospital disaster preparedness and detect the weaknesses that should be tackled by each institution for upgrading their response to disaster incidents: Step 1: a decision-making team is chosen considering their expertise (related to disaster management, healthcare management, and MCDM) and experience (at least 10 years in the Turkish healthcare sector) on disaster management and emergency care services. The selected participants (N) will be asked to provide insights on the definition, importance, and influence of assessment decision elements (criteria/sub-criteria) through FAHP and FDEMATEL techniques respectively. In this regard, it is critical to define the number of experts participating in the decision-making process so that criteria weights, interdependence results, and TOPSIS scores can be calculated at a high confidence level (CL = 95% at a minimum) and low error level (e = 5% at a maximum). Step 2: the assessment criteria and sub-criteria are determined given the related scientific literature, government regulations and experts' considerations. Step 3: Fuzzy AHP is implemented to estimate the relative weights of criteria and sub-criteria under vagueness (see sub-section 3.1.1). Step 4: Fuzzy DEMATEL is applied to pinpoint the dispatchers and receivers per each cluster while estimating the strength of influence among criteria/sub-criteria (see sub-section 3.1.2). Step 5: FAHP and FDEMATEL are later integrated to calculate the final criteria and subcriteria weights with basis on interdependence (see sub-section 3.1.3). Step 6: TOPSIS is finally used for ranking the hospitals on the basis of disaster preparedness. In parallel, weaknesses are identified for propelling the design of focused improvement interventions in each institution (see sub-section 3.1.4). In this subsection, we identify the applied MCDM methods either used in fuzzy sets or in crisp environment. Therefore, prior to giving the details of FAHP and FDEMATEL method, an overview on the notations of fuzzy sets may be useful. Zadeh (1965) introduced fuzzy sets for better reflecting of the human judgments and assessment in decision making. Also, the usage of fuzzy sets is better for transforming linguistic decision of human judgment and reflecting uncertainty and ambiguity of the real world decision making processes. Hence, many problems have used the fuzzy sets. The inclusion of fuzziness, however, entails more complex calculations compared to the existing related approaches. To tackle this disadvantage, an Excel-based decision support system has been properly designed and adopted to accelerate the disaster preparedness evaluation in relation to: i) weighting and prioritizing disaster readiness criteria and sub-criteria, ii) identifying the dispatchers and receivers within the disaster management scenario, iii) ranking the hospitals according to their preparedness level, and iv) defining focused operational strategies for increasing the response of hospitals against outbreaks. One representation of fuzzy sets is the use of triangular fuzzy numbers. A triangular fuzzy number consists of comprises lower, medium, and upper numbers of the fuzzy as = ( , , ) where l, m and u which is crisp and real numbers ( ≤ ≤ ). The membership function of a triangular fuzzy number ( ) can be defined as follows. A triangular fuzzy number is presented in Figure 3 . Table 2 . Defuzzification style: graded mean integration representation (GMIR) "# $ % = ( $ + 4 $ + $ ) 6 ( FAHP is one of the commonly applied MCDM methods. Crisp AHP cannot mirror the subjectivity broadly. Although Saaty and Tran (2005) stated that AHP integration with fuzzy sets cannot be effective, there are many studies in the literature that integrate AHP with these fuzzy sets. Some arguments in these studies are as follows. Kahraman et al. (2003) stated that AHP cannot reflect the thinking style of human, therefore, the fuzzy extension of AHP has been developed. Chan et al. (2008) stated that with fuzzy cluster and AHP integration, mathematical uncertainty would be better expressed and could be therefore used in solving real-world problems. Also, Wang et al. (2008) expressed that AHP is integrated with fuzzy sets to better reflect uncertainty. Therefore, AHP is improved by fuzzy sets to demonstrate uncertainty and vagueness. Different improved versions of AHP by fuzzy sets are available in the literature (Buckley, 1985; Chang, 1996) . In this existing study, Buckley's (1985) method is applied to determine hospital disaster preparedness criteria. In some FAHP extensions, for example in Chang's extent analysis, a limitation is released. An irrational zero weight generation problem in criteria weighting (Chan and Wang, 2013) is detected in Chang's FAHP. Step 1-Pairwise comparison of each criterion: Linguistic terms are used in determining relative importance of each two criteria based on Eqs. 2 and 3. Although the Saaty natural scale (1: Equal importance; 3: Weak importance; 5: Strong importance; 7: Very strong importance; 9: Absolute importance; reciprocals) was initially proposed to denote the preferences between two elements either criteria or sub-criteria (Saaty, 1988) , a shorter and fuzzy version of this scale (1 : Equally important; 3 : More important; 5 : Much more important; 3 , : Less important; 5 , : Much less important) (Eq. 3) has been adopted to deal with the imprecision of linguistic evaluations whilst reducing some bias and confusion during the comparison process (Pecchia et al. 2011; IJzerman, Van Til, and Bridges, 2012; Ortíz-Barrios et al. 2017 ). 0 1 $8 = 9 1 , 3 , 5 , :;<=>;D E<=ℎ :;<=>;;D E<=ℎ :;<=>; 10) which are consistent with the interactive nature often found in healthcare scenarios (Leksono, Suparno, and Vanany, 2019) . The interrelations within each sub-criteria cluster were assessed through impact digraph maps (Figures 11 to 13 ). The influence diagram for criteria is presented in Figure 11a . In this cluster, the adopted threshold (p) was established as ‹ = OE .o • } = 2.01 after defuzzifying the respective total relation fuzzy matrix … . From the graph, it can be discriminated that all the criteria are dispatchers. Moreover, several interrelations are observed among these decision elements; some of them are of feedback nature (C2-C3; C2-C5, C3-C5; C3-C6). In light of the above-mentioned considerations, multidimensional emergency operation plans should be designed, socialized, disseminated, simulated, and deployed by disaster managers to upgrade the performance of hospitals upon facing disaster events. Simulation of such plans will provide further analysis on each criterion so that hospital weaknesses can be properly detected and tackled before the occurrence of a disastrous situation. SC1.7) . The influence of SC1.5, SC1.6, and SC1.7 on the rest of elements is based on the fact that these sub-criteria greatly restrict building, design, and location conditions of hospital infrastructure; aspects often considered by administrators when expanding, relocating, and adapting their facilities to the potential requirements derived from devastations. The influence diagram for "Communication" (C3) sub-factors is depicted in Figure 12a . In this configuration, the reference value was stated as ‹ = OE. o o } = 3.03. In this domain, a feedback interrelation was detected between SC3.2 and SC3.1. The need for constant communication support is critical for effectively underpinning the operation flows within ECNs. In the opposite direction, the ECN configuration affects the quantity and deployment of communication devices in each hospital. As also found in the previous clusters, interactions are present in each paired relation which entails high degree of complexity for managers when including these aspects in the emergency operation programmes. Finally, an influence map (Figure 12b ) was drawn to analize the interdependence among "Transportation" (C4) sub-criteria. The limit value considered in this cluster was set as ‹ = •.OEo • } = 1.05. Based on the diagram, the decision elements strongly interact with each other even in a double-direction form as presented in SC4.1-SC4.4, SC4.2-SC4.4, and SC4.3-SC4.4. Also, "Number of vehicles" (SC4.1), "Helipad space" (SC4.2), and "Accessibility (roads)" (SC4.4) were concluded to be deliverers whilst "Safety" (SC4.3) was classified as receiver. It is noteworthy that all transportation modes used in the healthcare system are insights required for the correct deployment of safety measures seeking for protecting victims in the wake of a disaster. Understanding the relationship between transportation conditions and patient safety will help disaster managers reduce potential adverse events during catastrophic events and establish guidelines for appropriate risk management. The global priorities of criteria (xLr $ ) and sub-criteria (xqr 8 ) on the basis of interdependence were estimated by implementing Eq. 16 and Eq.18 correspondingly. Likewise, local interdependence weights of sub-criteria (pqr 8 $ ) were calculated using Eq. 17. The results derived from these equations have been compiled in Table 13 . Based on FAHP-FDEMATEL outcomes, Personnel (C5) was found to be the most important factor (xLr ' = 0.280) when assessing the hospital preparedness when facing devastating situations. Little difference (0.015) was also detected between this criterion and Equipment (C2). Such results invite disaster management planners to fully consider these categories to effectively respond to the shock of disasters and return to stability. Indeed, effective management of Personnel (C5) and Equipment (C2) is critical for upgrading the performance of emergency care networks when facing catastrophic events. For instance, the availability of medical teams, administrative staff, and resources play a key role for ensuring successful onsite-rescue and within-hospital medical care. Moreover, sharing medical staff and other resources is one the main activities specified in the Memorandums of Understanding (MoUs) signed by hospitals. In particular, the presence of suitable and sufficient personnel and equipment supports hospital response, especially in the aftermath of a disaster. Considerable efforts should be then made on C2 and C5 to establish appropriate supplier agreements and train medical staff in disaster management so that Flexibility (C6) can be effectively pursued as highlighted in the previous section. Following this, we have stratified our analysis to look into the importance of each subcriterion in each cluster. For instance, in Hospital buildings (C1) category, the most important decision element was Physical infrastructure (SC1.1) (pqr = 0.354) whereas the second sub-criterion in the ranking was Location (SC1.2) (pqr = 0.203). Such aspects represent more than a half of importance (0.557) in C1 criterion and they should be therefore urgently focused for continuous monitoring and improvement in hospitals. In particular, disaster managers should analyse the current state of the division of internal hospital spaces, the external envelope, structure, services, and contents to grant the safety of medical staff and victims after the catastrophic event. Unfortunately, most hospitals fail to include built environment issues in their disaster management plans (Cimellaro, Reinhorn, and Bruneau, 2010; Loosemore and Chand, 2016) . In this regard, it is recommended to: i) check and modify (if needed) the hospital layout to facilitate flow throughout the hospital during a disaster, ii) perform maintenance activities to minimize infrastructure vulnerabilities, and iii) review past disaster experiences and their effects on physical hospital infrastructure. In relation to Location (SC1.2), it is necessary to lessen the average travel distance for strikes' victims over a range of potential disaster scenarios. Hospital location models (Arlym et al. 2019; Acar and Kaya, 2019) should be then used for supporting this decision so that timely medical care can be provided to patients. In Equipment (C2) criterion, Medical equipment for ES (SC2.4) and Medicine (SC2.1) were concluded to be the most relevant sub-factors with local weights of 0.188 and 0.171 respectively. Such results call for clearly determining the supplies and medical equipment needed for handling disasters effectively. In this regard, it is vital to partner with supply chains capable of providing the sufficient and appropriate resources during disasters without stockpiling within the hospitals' facilities. Prior to this, disaster managers should identify, plan, and coordinate the timely supply of medical equipment and supplies so that hospitals can deliver the appropriate medical care before and after the strike. Some other significant recommendations to properly manage these aspects can be found at de Jong and Benton (2019). Looking into the results within the Communication (C3) domain, it was concluded that the most critical aspect is Emergency network (SC3.1) (pqr o = 0.643). It is noteworthy that failure of hospitals during catastrophic events can highly affect public morale and increase needless deaths. In this sense, Emergency care networks (ECNs) may alleviate the burden faced by hospitals individually by co-ordinately sharing medical staff, emergency drugs, and other critical resources. Thereby, the service level for disaster resources can be meaningfully improved which consequently ensures timely disaster response. Notwithstanding the tremendous efforts made by governments in this particular aim, ECNs are still at the earlier stages. In this sense, inefficiency factors such as the lack of coordination among hospitals and the presence of non-value added activities should be properly tackled by disaster managers to grant their adequate performance when facing devastating community events. With regards to Transportation (C4) category, Accessibility (roads) (SC4.4) was identified as the most important sub-criterion with a local priority of 0.397. Road accessibility to hospital facilities is critical for granting rapid medical care to disaster victims. In this respect, ground failure, imploding buildings to road edges, and bridges collapse may occur and limit the patient flows during catastrophic situations. As accessibility changes after disastrous events (Ertugay, Argyroudis, and Düzgün, 2016) , it is suggested to: i) evaluate road closure probabilities during the different types of disaster, ii) elaborate a disaster management plan in which hospitals can be pre-allocated to improve accessibility, and iii) implement accessibility models for helping emergency managers to better define emergency routes for evacuation and medical care. In relation to Personnel (C5) area, the most critical aspect to be considered during disastrous situations is Education (SC5.1) whose local weight was calculated to be 0.341. In the presence of unanticipated strikes, disaster management preparedness is recognized as crucial for medical staff and nurses so that effective care can be provided to the multiple victims arriving to the hospitals. Indeed, continuing disaster management courses have become an important strategy for avoiding errors that may hinder the response of hospitals before and after devastating events (Al Khalaileh, Bond, and Alasad, 2012) ; hence SC5.1 has been also considered as the sub-criterion with the major interdependence weight in the hospital preparedness evaluation (xqr = 0.096). Some recommendations for effectively managing the disaster preparedness of staff include: i) continuously evaluate the core competencies, skills, and knowledge of physicians, nurses, and other support staff regarding the management of disastrous situations, ii) the inclusion of disaster preparedness in national curricula of medical staff so that worsening of events in mass causalty disasters can be effectively prevented, and iii) the implementation of disaster facility drills given the strong interrelationship found between this element and education. Ultimately, Blood bank (SC6.3) was found to be the most crucial aspect in the Flexibility (C6) domain with a local interdependence weight of 0.436. Undoubtedly, the use of blood products is vital for effectively addressing the diverse kinds of injuries emanating from disasters (either man-made or natural). In the wake in the past disaster experiences, it is then imperative to optimize the blood supply chain so that the healthcare system can navigate through events deviating from the normal day-to-day demands. In light of these considerations, it is recommended to i) determine the need for blood products depending on the potential disaster coverage, ii) ensure a seven-day supply of blood (Simonetti et al. 2018) , and iii) model and simulate transfusion services throughout the long and short run of a disaster. This chapter details the application of TOPSIS method whose main objective was to rank the hospitals based on their disaster preparedness whereas pinpointing the weaknesses that should be tackled by each institution so that better response can be expected when facing devastating events. Moreover, the sub-criteria most contributing to the PIS and NIS of each hospital can be discriminated so that focused enhancement strategies can be effectively deployed in the practical disaster scenario. Initially, a performance indicator (Table 14) was established per each sub-criterion. Following this, initial TOPSIS decision matrix X (Table 15 ) was arranged considering the hospital alternatives (" , " , " o , " • ), performance indicators, and sub-criteria. In particular, the performance indicators values were computed considering the mathematical formula depicted in PIS and NIS were also specified in Table 15 by employing Eq. 21 and Eq. 22 correspondingly. The normalized ratings are later calculated using Eq. 19 (Table 16) whereas the weighted normalized ratings were estimated by applying Eq.20 (Table 17) . The interdependence weights of sub-criteria were calculated through the integrated FAHP-FDEMATEL approach illustrated in the previous section. On the other hand, the Euclidean distance of each hospital (" , " , " o , " • ) from the PIS (q v n ) was calculated using Eq. 23 (Table 18 ). In a similar vein, the separation of each hospital from the NIS (q v , ) was estimated by applying Eq. 24 (Table 19 ). The ranking of hospitals and closeness coefficients L v * are shown in Figure 13 . Such coefficients were computed using Eq.25. The results revealed that the hospitals performed between 0.632 (H 2 ) and 0.742 (H 1 ); there is therefore much room for improvement and interventions from the stakeholders. It is then necessary to estimate the distances from PIS and NIS to identify the weaknesses of each hospital (sub-criteria whose Euclidean separation from PIS is over zero || Euclidean separation from NIS is equal to 0). In particular, it was found that H 2 evidences low availability of medical equipment -SC2.4 (21.3%; separation from PIS = 0.01813) which may cause delays in medical care provided during disaster and consequently increase the risk of mortality. Moreover, H 2 is the hospital with the farthest distance from the target community -SC1.2 (3 km; separation from NIS = 0), an aspect that may limit the timely medical care also considering the potential collapse of roads and adjacent buildings. Another weakness is the non-availability of a helipad space -SC4.2 -which highly restricts the patient transferring process from the disaster zone to hospital facilities (separation from NIS = 0). Additionally, there are no contingency staff -SC6.2 -for facing the disaster situation (separation from NIS = 0). Such disadvantage may compromise the hospital response mainly in large-scale catastrophic situations. On a different tack, similar to H 2 , H 3 concerns about the low availability of medical equipment -SC2.4 (16.7%; separation from PIS = 0.02034) and non-availability of helipad space -SC4.2 (separation from NIS = 0) and contingency staff -SC6.2 (separation from NIS = 0). Also, it has the lowest number of floors -SC1.3 (5 floors, separation from NIS = 0) and ambulances -SC4.1 (2 vehicles, separation from NIS = 0), an aspect limiting the capacity for addressing the peaks of demand that may arise from the disaster. Moreover, H 3 does not evidence the use of tents -SC2.7 (separation from NIS = 0) which also restricts its disaster preparedness. The major disadvantage is the number of disaster management programs organized by the hospital -SC5.1 (2 training programs; separation from NIS = 0), a fact that may trigger errors during on-site rescue, patient transportation, and medical care. Regarding H 4 , more investment and maintenance intervention is needed for increasing the availability of medical devices -SC2.4 (10.6%; separation from PIS = 0.02342). In addition, H 4 also has the farthest separation from the target community -SC1.2 (3 km; separation from NIS = 0) and the lowest number of floors -SC1.3 (5 floors, separation from NIS = 0). Apart from these findings, it is observed that H 4 presents the shortest % of correctly insulated ED rooms -SC1.6 (30.0%; separation from NIS = 0) and % of rooms with appropriate ventilation -SC1.7 (30.0%; separation from NIS = 0). Both conditions may hinder the correct deployment of disaster management plans in the wild; especially during medical intervention Similar to H 3 , H 4 does not either denote tent usage -SC2.7 (separation from NIS = 0). Also, the number of beds -SC2.9 -is limited compared to the rest of hospitals (200 beds; separation from NIS = 0). Shortage of beds has become a significant barrier for addressing large-scale outbreaks as those expected in the future. Moreover, H 4 only has one ambulance -SC4.1 (separation from NIS = 0) which may not be enough for facing the potential upcoming events (i.e. coronavirus (Huang et al. 2020) ). On the other hand, as all the hospitals participating in this study, no helipad space is available -SC4.2 (separation from NIS = 0). Moreover, H 4 has the lowest number of security guards -SC4.3 (18 guards; separation from NIS = 0) which may not facilitate the patient flow and medical staff protection during a disaster. The number of training programms in disaster management is also low -SC5.1 (2 programs; separation from NIS = 0) and new courses should be therefore implemented for increasing the competences, knowledge, and skills of H 4 workers. Lately, this hospital was not found to be flexible as revealed through poor outcomes in the associated sub-criteria. Finally, in relation to the leading hospital, the TOPSIS results evidenced the need for intervention in the next sub-criteria: SC1.4 (400; separation from NIS = 0), SC2.4 (1.5%; separation from PIS = 0.02842), SC2.7 (separation from NIS = 0), SC4.2 (separation from NIS = 0), SC5.1 (2 training programs; separation from NIS = 0), SC5.5 (30 trained employees; separation from NIS = 0), and SC6.2 (separation from NIS = 0). communication, transportation, personnel, flexibility and a total of thirty-six sub-criteria. FDEMATEL is hereafter applied to uncover the interdependence between criteria and subcriteria. As a ranking tool, TOPSIS is used to determine a ranking of hospitals. In application phase of the existed decision making model, two different questionnaires are created and assessed by seven experts related to the field of disaster management of healthcare facilities and academicians who are experienced for the disaster management topics. The numerical results demonstrated that "Personnel" is the most important factor (with a global weight value of 0.280) when evaluating the hospital preparedness while "Flexibility" has the greatest prominence (with a c+r value of 23.09). In the light of the numerical results obtained from this study, it is crucial that the observed and analysed hospitals design a better disaster preparedness plan in order to be more prepared against disasters. In relation to the scenario under study, the results revealed that the hospitals performed between 0.632 (H 2 ) and 0.742 (H 1 ); there is hence much room (Gap to target: 0.258 -0.368) for interventions increasing the current disaster readiness level. These interventions must address the main weaknesses detected in the cited set of hospitals: i) low availability of medical equipment, ii) lack of helipad spaces, iii) low availability of contingency staff, iv) lack of tents, and v) low number of disaster management training programs. Thereby, these hospitals will be better prepared for facing future outbreaks in terms of timeliness, quality, and efficiency. This approach is useful for the related research field considering that methods measuring the hospital disaster preparedness levels are lacking (Xin and Xu, 2012; Zhong et al. 2017 ). However, it has some limitations from both methodological and application viewpoints. We consider triangular fuzzy sets in both AHP and DEMATEL stages. Considering there exist various new versions of fuzzy set theory that reflect uncertainty and ambiguity of decision making process better, the current approach may be extended to integrate AHP and DEMATEL with some new versions of fuzzy set such as intuitionistic fuzzy sets, interval type-2 fuzzy sets, hesitant fuzzy sets, Pythagorean fuzzy sets and spherical fuzzy sets. From application viewpoint, we limit the current study with four Turkish hospitals. The proposed approach can be offered to health policy makers of Turkey as a model on a national scale. In this context, the authors intend to further improve the approach by means such as considering some new hospital disaster preparedness criteria that will be suggested by these policy makers. Also, as a second attempt, the authors contemplate adapt the approach to Colombian hospital emergency department disaster preparedness assessment. A healthcare network design model with mobile hospitals for disaster preparedness: A case study for istanbul earthquake A fuzzy DEMATEL method to evaluate critical operational hazards during gas freeing process in crude oil tankers Jordanian nurses' perceptions of their preparedness for disaster management Disaster Preparedness in Hospitals in the Middle East: An Integrative Literature Review. Disaster medicine and public health preparedness A policy model of preparedness the general hospital in reducing victims of earthquake and tsunami disasters in siberut mentawai island, indonesia Fuzzy hierarchical analysis, Fuzzy Sets and Systems Global supplier selection: a fuzzy-AHP approach Fuzzy extent analysis for food risk assessment Applications of the extent analysis method on fuzzy AHP Seismic resilience of a hospital system Dependence and power in healthcare equipment supply chains Disaster Plan of Hospital and Emergency Service in the Van Earthquake Accessibility modeling in earthquake case considering road closure probabilities: A case study of health and shelter service accessibility in thessaloniki, greece Casualty treatment after earthquake disasters: development of a regional simulation model US Disaster Recovery Readiness for a Biological Terrorist Incident: Part Four World problems, an invitation to further thought within the framework of DEMATEL Evaluation of service quality criteria for a private medical center by using SERVQUAL and DEMATEL methods Are emergency departments in Istanbul ready for the earthquakes? Past experience and suggestions for future preparedness from employees' viewpoint and the literature A fuzzy multi criteria risk assessment based on decision matrix technique: a case study for aluminum industry Emergency department ergonomic design evaluation: A case study using fuzzy DEMATEL-focused two-stage methodology Ranking hospitals based on the disasters preparedness using the TOPSIS technique in western Iran Clinical features of patients infected with 2019 novel coronavirus in wuhan, china. The Lancet Multiple Attribute Decision Making-Methods and Applications: A State of the Art Survey A comparison of analytic hierarchy process and conjoint analysis methods in assessing treatment alternatives for stroke rehabilitation. The Patient-Patient-Centered Outcomes Research Emergency medical care following the great Hanshin-Awaji earthquake: practices and proposals (a report from a university hospital located in the damaged region). The Kobe journal of medical sciences Multi-criteria supplier selection using fuzzy AHP. Logistics information management Earthquake and Trıage Evaluating bullwhip effect mitigation: an analytical network process (ANP) application Evaluation of the earthquake victims: August the 17th, 1999 our experience in the Marmara Earthquake Integration of a Balanced Scorecard, DEMATEL, and ANP for Measuring the Performance of a Sustainable Healthcare Supply Chain Barriers to building resilience to extreme weather events in australian hospitals Developing a Model for Hospitals' Emergency Department Preparedness in Radiation and Nuclear Incidents and Nuclear Terrorism in Iran A study of hospital disaster preparedness in South Yemen The analytic decision-making preference model to evaluate the disaster readiness in emergency departments: The ADT model Strategic hybrid approach for selecting suppliers of high-density polyethylene An integrated approach to evaluate the risk of adverse events in hospital sector On the compatibility between defuzzification and fuzzy arithmetic operations. Fuzzy Sets and Systems The use of analytic hierarchy process for the prioritization of factors affecting wellbeing in elderly User needs elicitation via analytic hierarchy process (AHP). A case study on a Computed Tomography (CT) scanner. BMC medical informatics and decision making Development of comprehensive earthquake loss scenarios for a Greek and a Turkish city: seismic hazard, geotechnical and lifeline aspects Evaluation of the readiness of hospitals affiliated to Isfahan University of Medical Sciences in unexpected events in 2017 Fuzzy logic with engineering applications What is the analytic hierarchy process On the invalidity of fuzzifying numerical judgments in the Analytic Hierarchy Process Investigating hospital preparedness in Ardabil Province against unexpected accidents Disaster preparedness attributes and hospital's resilience in Malaysia Developing a practical toolkit for evaluating hospital preparedness for surge capacity in disasters Identifying key performance indicators for holistic hospital management with a modified DEMATEL approach DEMATEL technique: A systematic review of the state-of-the-art literature on methodologies and applications An inter-regional US blood supply simulation model to evaluate blood availability to support planning for emergency preparedness and medical countermeasures Risk assessment in social security hospitals of Isfahan Province in case of disasters based on the hospital safety index An investigation of hospital disaster preparedness in Turkey Multiple attribute decision making: methods and applications An analysis of multi-criteria decision making methods Applying fuzzy linguistic preference relations to the improvement of consistency of fuzzy AHP Coronavirus disease 2019 (COVID-19) Situation Report-93 Developing global managers' competencies using the fuzzy DEMATEL method. Expert systems with applications Hospital emergency management research in China: trends and challenges Disaster-readiness of medical facilities in Aichi Prefecture Hospital service quality evaluation: an integrated model based on Pythagorean fuzzy A Fuzzy Decision-Making Model for the Key Performance Indicators of Hospital Service Quality Evaluation Fuzzy sets Hybrid multiple criteria decision-making methods: A review of applications for sustainability issues Identification of Key Performance Indicators for Hospital Management Using an Extended Hesitant Linguistic DEMATEL Approach Disaster resilience in tertiary hospitals: A cross-sectional survey in shandong province, china Seismic casualty evaluation: The Italian model, an application to the L'Aquila 2009 event Acknowledgments The authors would like to thank Prof. Heriberto Felizzola-Jiménez for their valuable contribution to this research. Also, many thanks to Giselle Paola Polifroni-Avendaño for her support during this project. ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: