A significant number of influencing factors hampers data collection and reproducibility of experiments in disaster settings, thus resulting in a weak level of evidence both in prospective and retrospective studies.Reference Bradt and Aitken1
After-action reports (AARs) are a well-known method to gather information and analyze real events or functional exercises.Reference Savoia, Agboola and Biddinger2 Originally, this term was developed in the last century by military forces as “Lessons Learned,” but, in the current literature, Koenig and colleagues suggested to discard this term because of its lack of scientific value and the low power of systematic dissemination of the data.Reference Koenig, Schultz, Gould Runnerstrom and Ogunseitan3 However, AARs of disaster events are a useful way to describe key actions and decisions taken in a multidisciplinary context such as disaster management but need standardized models to carry good quality evidence to the reader.Reference Bradt and Aitken1
In the last decade, several templates for disaster events reporting have been developed, but there is no consensus among these models,Reference Fattah, Rehn and Reierth4 and most are not validated. The proposed templates are mainly focused on prehospital management and field response and often address hospital disaster reaction only marginally.Reference Bradt and Aitken1, Reference Fattah, Rehn and Reierth4–Reference Lennquist8 Although hospitals are the cornerstone of health and medical systems during disasters, to our knowledge, there is not a specific model for reporting hospital disaster response. Anyway, the “Utstein-style template for uniform data reporting of acute medical response in disasters” contains several reproducible indicators of hospital reaction but was neither used nor validated before.Reference Debacker, Hubloue and Dhondt5
In this study, we intended to pilot the use of a modified version of the “Utstein-style template for medical response in disasters”Reference Debacker, Hubloue and Dhondt5 as a new assessment tool for hospital response to disasters.
METHODS
This study describes the use of a new data collection tool focused on hospital disaster response in case of disasters.
The study group included hospitals in the regions around the epicenter where the hospital disaster plans (HDPs) were fully activated after the first main shake of the 2016 Central Italy earthquake, in the morning of August 24, 2016. The report is limited to the first 24 hours from the first temblor.
Research Tool
A data collection tool focused on hospital disaster response was created modifying the “Utstein-style template for uniform data reporting of acute medical response in disasters.”Reference Debacker, Hubloue and Dhondt5 Items regarding hospital disaster response were extracted from this template and implemented with specific items from other validated forms regarding hospital disaster preparedness: the Hospital Safety Index: Guide for Evaluators, from Pan American Health Organization9 and the Hospital Emergency Response Toolkit developed by the Regional Office for Europe of the World Health Organization.10 Therefore, items such as personnel management and plans during disasters, number, and type of admitted patients were added, and the hospital treatment capacity was calculated according to the formula proposed by Takahashi and colleagues.Reference Takahashi, Ishii, Kawashima and Nakao11 A panel of experts composed of 3 senior faculty members from the Research Center in Emergency and Disaster Medicine (CRIMEDIM), Università del Piemonte Orientale, Novara, Italy, reviewed the instrument content for accuracy in December 2015 and provided appropriate final modifications to ensure the validity of the tool in March 2016. This template is articulated in 12 sections: background; hospital characteristics; HDP activation; personnel; emergency department (ED) - general aspects; ED - triage and admissions; department of surgery; radiology; intensive care; services; outcome; and standing down of HDP (Annex 1).
Data Collection Procedures
Data were collected from November 1, 2016, to May 1, 2017, after formal authorization given by each hospital’s boards. Site visits were organized, and relevant disaster planners of each facility – hospital managers, hospital disaster managers (HDMs), hospital incident commands members, and department directors – were interviewed to assess timing and actions undertaken.
Data and patient charts from hospital archives were analyzed. Patients were divided into 3 main groups, according to the following criteria:
Earthquake-related (ER) patients (at least 1 of the following criteria):
ˆ Earthquake casualties, clearly specified in the history
ˆ Patients without a clear history but from the disaster area
ˆ Patients with comorbidities, from the disaster area
ˆ Patients with trauma or wounds occurred during the first shake (eg, “escaping”), also distant from the disaster area
ˆ Non-traumatic, medical issues, with onset within 6 hours from the first shake.
Non-earthquake-related (NER) patients: none of the above criteria
Incomplete data: patients without complete or clearly understandable information and history
The collection tool was filled using a Microsoft Office Word 2010 sheet, Version 14.0.70 (Microsoft Corporation, Redmond, WA). Data were coded on a master sheet using a Microsoft Office Excel 2010 spreadsheet, Version 14.0.47 (Microsoft Corporation, Redmond, WA).
Data Analysis
Frequencies were used to describe respondents’ features; demographic characteristics of the patients were analyzed descriptively through their distribution frequency in the case of qualitative variables, with mean and standard deviation for quantitative variables (when possible).
Patients’ arrival times to EDs were stratified by intervals of 1 hour, starting from the time of the first temblor.
Data from the analyzed hospitals were not compared, due to extreme differences among geographical contexts, infrastructures, number, and type of available resources and casualties.
Ethical Aspects
The participation in the study was voluntary and independent. Confidentiality of information was ensured, and no financial incentive to participate in the study was offered. Because all data were collected such that individual subjects could not be identified or exposed to risks or liabilities, the evaluation was deemed exempt from institutional review approval by the local ethics committee (protocol No. 6/17).
RESULTS
The 2016–2017 Central Italy earthquakes consisted of a series of seismic events with epicenters along Central Italy’s Apennine Mountains. The first shock of 6.0 on the moment Richter magnitude scale stroke on August 24, 2016, at 03:36:32 AM (UTC + 2), and the epicenter located at the border of 4 contiguous regions (Marche, Abruzzo, Lazio, and Umbria; 42.706°N 13.223°E).12 Towns next to the epicenter reported significant damages to buildings, roads, and to the cultural heritage up to a radius of 100 km,13 and the death toll amounted to 299 people, mainly deceased under collapsed buildings.14–16
Four spoke hospitals around the epicenter fully activated their HDP and were included in the report: Ascoli Piceno (AP) and San Benedetto del Tronto (SBT) in Marche region, L’Aquila (AQ) in Abruzzo region, and Rieti (RI) in Lazio region. Hospitals’ characteristics are listed in Table 1. Two peripheral hospitals (Amatrice and Amandola) were severely damaged and promptly evacuated.13, 17
TABLE 1 Included Hospitals’ Answers to the Hospital Disaster Response Data Collection Tool
AQ – H: Hospital of L’Aquila; AP – H: Hospital of Ascoli Piceno; SBT – H: Hospital of San Benedetto del Tronto; RI – H: Hospital of Rieti; EMS: emergency medical services; EMS – OC: Emergency Medical Service Operation Center; ED: emergency department; HDP: hospital disaster plan; HDP – AO: Hospital Disaster Plan Activation Order; HIC: Hospital Incident Command; ER: earthquake-related; LP: low priority; M: mild; I: intermediate; S: severe; D: deceased; START: simple triage and rapid treatment; CT: computed tomography.
a In SBT – H there was not an independent CU. The Hospital Disaster Manager in charge on site depended on AP EMS – OC and on AP – H CU, as SBT – H served as a backup hospital.
b AQ – H CU chose not to create different areas because there was not a real patient’s surge.
Table 1 is the merged version of the 4 collection tools used in each hospital. The completion rate of the collection tool was of 97.1% (268 filled items, out of 276). All of the missing items belonged to the RI hospital (most of them in Section 6: Department of Surgery).
The RI hospital was the first to be alerted by the pertaining Emergency Medical Service – Operations Center (EMS–OC) and activated the HDP at 4:00 AM, whereas the other facilities were alerted by their respective EMS–OC about 30 minutes following the shake and activated the HDP within 4:30 AM. Patients’ influx started before HDP Activation Order (HDP–AO) only in AQ and AP hospitals. HDP flowcharts and action cards, in a paper or digital form, were available for all of the personnel involved in the activation only in AP and SBT hospitals.
HDP activation was completed with expected personnel and equipment fully deployed, between 7:00 and 7:30 AM in each hospital, even though each HDM reported that more personnel than expected responded to the activation. A significant number of ER patients arrived within 60 minutes after the HDP–AO (before EDs full activation) in AQ (8 patients, 11.9% of AQ total), AP (5 patients, 5.5% of AP total) and SBT (5 patients, 13.9% of SBT total) but not in RI hospital (only 1 patient, 0.7% of RI total). Furthermore, each ED reserved beds for incoming severe patients within 15 minutes from the HDP–AO, and only AP received 2 severe patients in the first 60 minutes.
Prehospital triage codes were not clearly detectable on each patient admitted in the included EDs. Moreover, a change in hospital triage systems was not reported except in RI where simple triage and rapid treatment (START)18 was used after the HDP–AO.
According to the results, 30 minutes after HDP–AO, 2 operating rooms were already available in AQ and SBT, whereas only 1 was ready in AP. Surgery departments were fully activated within 45 minutes from HDP–AO in SBT, and within 60 minutes in AQ and AP, but no emergency operations were performed till the morning (the first in AP began at 8:25 AM). The last emergency surgery of the day was performed in SBT at 6:55 PM. During the analyzed time interval, elective surgeries were cancelled and postponed to the next day. Data about the surgery department of RI were not collected due to the unavailability of records.
Among ER patients, those who had intermediate priority received more computed tomography (CT) and ultrasound scans. Radiography images were mostly performed on intermediate priority patients in AQ and AP, and on mild priority patients in SBT, while in RI, intermediate and mild priority patients were equally represented.
One intensive care unit (ICU) was available in each facility, and their complete activation required between 30 minutes in RI and 120 minutes in AQ and AP. Thirty minutes after HDP–AO, 4 beds in AP, SBT, and RI and 2 in AQ were already available, even though no patients were admitted at this time. Only 1 patient in AQ and 1 patient in AP needed mechanical ventilation.
Overall, 124 of ER patients (20.22% of the total ER patients) were hospitalized. SBT reported the highest discharge percentage (83.3%). In-hospital mortality of ER patients was of 1 in AQ and 1 in SBT hospitals. One ER patient was declared dead at the arrival at AP’s ED.
HDP was stood down during the afternoon of August 24, 2016, in AP and SBT hospitals (+11 h), while the activation lasted till the morning of August 25, 2016, in AQ (+27.5 h) and in RI (+28 h).
In Table 2, patients’ characteristics and fluxes are depicted. During the first 24 hours after the earthquake, a total of 613 patients entered the 4 EDs, with a mean age of 49.52 (24.34) years and a homogeneous distribution of gender. Pediatric patients were the least represented (Figure 1). Most of them were registered at RI hospital (178 out of 613; 29.04%), and the number of ER patients was greater than NER patients in every ED, except for SBT (see Table 2).
TABLE 2 Characteristics of Earthquake and Not-Earthquake-Related Patients
AQ – H: Hospital of L’Aquila; AP – H: Hospital of Ascoli Piceno; SBT – H: Hospital of San Benedetto del Tronto; RI – H: Hospital of Rieti; SD: standard deviation; H: hospital; ER: earthquake – related patients; NER: not earthquake – related patients; OBS – GYN: obstetric – gynecologic; ENT: ear nose throat.
a In the total amount, there are patients classified as “Incomplete Data” (not classifiable neither as ER nor NER).
FIGURE 1 Age of Patients Admitted to the Included Emergency Departments.
Out of the total, 344 patients (56.11%) reached the EDs with their own means, whereas the others were transported by EMS–OC ambulances or helicopters. Thirty-nine patients were triaged as severe, 268 as intermediate, 294 as mild, and 9 as a low priority complaint (see Table 2).
Three hundred thirty-six (54.81%) patients were included in the ER category, with a prevalence of trauma injuries (260; 77.38%). Medical complaints were the main presenting symptoms (179; 69.11%) for the NER category.
Figure 2 shows a peak of overall admissions within the first 6 hours from the tremblor. This pattern of arrival time was similarly depicted in AP and RI, while in the remaining the influx was diluted along the first part of the day.
FIGURE 2 Patients’ Arrival Time to the Included Emergency Departments.
DISCUSSION
This is the only study aimed to analyze and report the reaction of the hospitals involved in survivors’ management after the August 24, 2016, Central Italy earthquake. The high completion ratio of this modified template suggests its feasibility and utility in practical terms. Moreover, information was easy to collect on the field, both in electronic or paper archives and by interviewing the involved personnel. Therefore, this tool could enhance the standardized analysis of hospitals’ reactions to disasters, mainly to consolidate the body of knowledge of disaster medicine but also allowing debriefing and quality improvement processes within involved and non-involved facilities.
In Italy, since 1992, hospitals are required to adopt an HDP for external disasters, known as emergency plan for massive influx of casualties (PEMAF acronym in Italian), in order to face a sudden patients’ surge.19–21 According to the results, there was a significant delay in hospitals’ HDPs activation after the main shake, due in our opinion to 2 factors. First, according to Major Incident Medical Management and Support guidelines,Reference Mackaway-Jones22 EMS–OCs activated the major incident status only after the confirmation by ambulances on site, hence with a delay of about 20 to 30 minutes (except for RI, where an ambulance was already on the crash area and immediately confirmed the major incident status). Second, each facility spent about 30 minutes (after the respective EMS–OC activation call) to check the structural integrity of buildings before confirming HDP activation, as a prerequisite of efficiency and security after the earthquake.
Lennquist defines the action card as the most important component in the deployment of the HDP,Reference Lennquist23 and experts recommended its use since 1972.Reference Savage24 In AQ and RI hospitals, the plan and action cards were not available to all of the personnel for consultation, but only coordinators had a copy of HDPs. Lack of information about HDP activation is a well-known problem in the Italian setting,Reference Paganini, Borrelli and Cattani25 so the work of disaster culture diffusion should be continued and improved in the next future by teaching and practical activities among all Italian facilities.
The results show that EMS–OCs transported survivors mainly to AP and RI hospitals, both for their proximity to the epicenter and their easier access by road, whereas AQ and SBT acted as a backup in case of overcrowding of the former. According to literature, it is common to witness a disproportionate number of survivors being transported from the scene to the closest health care facilitiesReference Auf der Heide26 or a presentation of self-transported patients to the almost familiar hospital.Reference Auf der Heide27 In fact, patients’ surge is clearly depicted in the first hours after the main shake in AP and RI, whereas, in SBT and AQ, the influx rate is more variable and relied on several factors such as road accessibility, timing of extraction from the rubble, and admission capacity by the first 2 facilities (see Figure 2). The same factors could have affected the low presentation rate of patients at RI during the first hour after HDP–AO (see Table 1). Furthermore, in AP and RI, patients were mainly transported by EMS ambulances, whereas, in SBT and AQ, patients mainly used private transport (see Table 2).
Overall, patients had a good short-term outcome. In fact, most of the patients were discharged from the EDs within the first 24 hours, especially in SBT where the EMS–OC diverted mild priority codes. Only about one-fifth of ER patients were admitted to the wards, especially in surgery departments, and there were only 2 deaths among hospitalized patients (see Table 1).
Data in Figure 1 demonstrate a prevalence of adults among patients admitted to the EDs. This finding could be related with the time of the catastrophic shake, having occurred at 3:36:32 AM when entire families were sleeping at home and consequently were all involved in the tremblor. In fact, earthquakes occurring at night are associated with a higher mortality and risk injury than those occurring during the day,Reference Doocy, Daniels and Packer28 when the adult population is mostly at work.
During the afternoon of August 24, 2016, patients’ influx gradually diminished on all of the sides of the earthquake (see Figure 2). HDP was stood down in AP and SBT hospitals at 3:00 PM, whereas in AQ and RI the activation lasted till 8:00 AM on August 25, 2016. Hospital incident commands of these 2 last facilities waited for the closure of emergency activation state by the respective EMS–OCs, albeit this choice may have influenced hospital response capacity in the medium term. In fact, in the context of a disaster, a timely plan termination is important to re-establish normal activities, such as scheduled surgeries or clinical activity in the wards, and to allow staff rehabilitation, incident debriefing, psychological defusing, and performance evaluation.
As in line with previous reports,Reference Bartels and Vanrooyen29 traumatic injuries were more common among ER patients and accounted for their greatest percentage of admission to the wards (see Table 2). Furthermore, acute cardiovascular diseases are the predominant medical issue in the analyzed EDs as reported since the Northridge earthquake.Reference Bartels and Vanrooyen29 Among other medical diseases, an acute psychiatric illnesses incidence caused by strong emotional reactions is similar to other reports, whereas respiratory system issues are not represented as in literature.Reference Naghii30
Limitations
Because data were self-reported and retrospectively reconstructed, there could be a selective memory bias. Several survivors were not traced by the EMS–OCs involved in the management of this disaster. For example, according to the Italian Civil Protection, additional 40 patients were evacuated by helicopter to other than the included hospitals, mainly located in Rome (personal communication). This could have led to an underestimation of the total treated survivors’ amount.
Moreover, the included hospitals cannot be compared among them because of differences in logistics, geographical context, number or type of lesions, and unpredictable factors. Therefore, the results have to be interpreted carefully, and the conclusions are not valid for the whole management of this disaster.
CONCLUSIONS
According to the described findings, the Modified Utstein Template for Hospital Disaster Response Reporting is a valid instrument for hospital disaster management reporting. Most of the management reported by the included hospitals is concordant with the existing literature, with a good short-term outcome of the survivors.
Overall, the reaction of the included facilities followed the key points of disaster management, especially in the activation of HDPs. However, the results show that an improvement in HDPs implementation and application should be undertaken in the included facilities. The systematic use of this template in the future could lead to a better comprehension of hospital disaster reaction and to a comparison among responses to similar events. Moreover, hospital managers could use this instrument for debriefing activities, HDP revisions, and for an improvement of regional trauma networks and EMS–hospital joint disaster response protocols.
Acknowledgments
We wish to thank all of the personnel involved in the response to the earthquake. Moreover, for help with data collection, we are indebted to Massimo Loria (ED Director – Ascoli Piceno), Liliana Talamonti (ED – S. Benedetto del Tronto), Remo Appignanesi (Hospital Board – S. Benedetto del Tronto), Luigi Valenti (ED Director – L’Aquila), and Antonella Morgante (Hospital Board – Rieti).
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
List of abbreviations
- AAR
after-action reports
- AP
Ascoli Piceno
- AQ
L’Aquila
- CRIMEDIM
Research Center in Emergency and Disaster Medicine
- CT
computed tomography
- ED
emergency department
- EMS-OC
Emergency Medical Service – Operations Center
- ER
earthquake-related
- HDM
hospital disaster manager
- HDP
hospital disaster plan
- HDP-AO
hospital disaster plan – activation order
- ICU
intensive care unit
- NER
non-earthquake related
- RI
Rieti
- SBT
San Benedetto del Tronto
- START
simple triage and rapid treatment
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/dmp.2019.54
