Disasters, natural or human-made, can cause significant morbidity and mortality, leading to substantial economic and human loss.Reference Foster 1 According to the EM-DAT international disaster database maintained by the Centre for Research on the Epidemiology of Disasters, for the period from 2005 to 2014, an average of 367 country-level natural disasters were reported each year, including floods, storms, droughts, landslides, earthquakes, tsunamis, wildfires, volcanic activities, etc.Reference Singh, Sodickson and Abujudeh 2 In 2015 alone, natural disasters caused the deaths of 22,773 people, and an economic damage of 66.5 billion US dollars.Reference Singh, Sodickson and Abujudeh 2 Examples of common human-made disasters include armed conflicts, terrorist incidents, explosions, nuclear disasters, and transportation accidents.Reference Coleman 3 The 9/11 terrorist attacks on the United States in 2001, in particular, marked the beginning of the 21st century, in which there has been a heightened impact, both of terrorism and the war against it.Reference Forest 4
The definition of “disaster” provided by the United Nations Office for Disaster Risk Reduction emphasizes a level of disruption that “exceeds the ability of the affected community or society to cope using its own resources.” 5 Medical resources, a key component of the disaster preparedness and management,Reference Foster 1 must be properly allocated and triaged to deliver the best possible care for a potentially large number of victims.Reference Waeckerle 6 As a fast evolving field, diagnostic imaging has been increasingly relied upon to make rapid and non-invasive diagnoses and triage or care decisions for patients. 7 It is thus important to analyze how diagnostic imaging has been utilized in disaster management in the past, to augment future disaster management plans.
Bibliometric analysis involves the quantitative and qualitative analysis of research literature,Reference Godin 8 and is commonly used as an objective measurement of the productivity of individual researchers or organizations.Reference Abramo and D’Angelo 9 It can also be used to reveal the emergence of novel techniques and topics, and to guide future research.Reference Choudhri, Siddiqui, Khan and Cohen 10 - Reference Khan, Ullah and Riaz 12 Several specialties and journals have previously compiled and published lists of the most-cited articles in their field.Reference O’Keeffe, Hanna and Holmes 11 , Reference Dolan, Hanna, Warraich, Johnson and Khosa 13 - Reference Shahzeb Khan, Fatima, Bin Riaz, Butler, Manning and Khosa 16 According to our literature search, no current bibliometric analysis has focused specifically on the role of imaging in disaster management in an English language peer-reviewed indexed journal. To fill this void, we conducted a bibliometric analysis of articles published since 2000 that depicted the role of diagnostic imaging in clinical diagnosis, treatment and follow-up management of patients in response to disasters, for an understanding of the current utility of diagnostic imaging in disasters.
METHODS
This study did not require ethical approval because it was a retrospective evaluation of publicly available research literature.
Literature Search
MEDLINE indexes more than 5600 journals and contains more than 23 million references in life sciences and medicine. 17 All MEDLINE records were indexed with the National Library of Medicine Medical Subject Headings (MeSH) system, 17 which greatly facilitates comprehensive searches on complex topics. We conducted a MEDLINE search via OVID on January 9, 2017, with the detailed search steps listed in Table 1. In brief, MeSH terms related to various types of disasters were first queried (Steps 1-16). To ensure a comprehensive search, specific terms of 4 major outbreaks in the 21st century 18 (outbreaks of Ebola, Severe Acute Respiratory Syndrome [SARS], H1N1 influenza, and Middle East Respiratory Syndrome) were searched along with the MeSH term “epidemics” (Steps 12-16). Results from each individual search were combined using the Boolean operator “OR” (Step 17-18). Articles about diagnostic imaging were searched and combined in a similar fashion (Steps 19-22). The Boolean operator “AND” was then used to identify the overlapping set of articles in the two sets of combined results (Step 23). The results were then filtered to include only articles pertaining to “humans,” and published in or after 2000 (Steps 24-26). A total of 1262 unique articles were identified.
Steps Used in MEDLINE (OVID) Search to Identify Recent Original Research Articles Depicting the Roles of Diagnostic Imaging in Clinical Diagnosis, Treatment, and Follow-Up Management of Patients in Response to Disasters
Abbreviation: di, diagnosis; pc, prevention and control; ra, radiography; us, ultrasonography.
We restricted our analysis to only literature published since 2000, for two reasons. First, the rise of terrorism and the following anti-terrorism wars in the beginning of the 21st century marked a change in the landscape of human-made disasters.Reference Forest 4 It was therefore interesting to conduct an analysis in the context of this new era. Second, this approach allowed us to focus on the more recent disaster management experiences, in light of decade-long disaster reduction efforts led by the United Nations throughout the 1990s (the International Decade for Natural Disaster Reduction).Reference Kiv, Douthit, Shack and Biswas 19
Screening of Original Research Articles
Two board-certified, emergency radiology fellowship-trained radiologists reviewed both the title and abstract (as well as full-text when necessary) of all the articles according to the inclusion and exclusion criteria predetermined before the screening. In case of disagreement regarding inclusion decisions, a third radiologist was consulted to make a final decision.
Specifically, only original research articles depicting the role of diagnostic imaging in clinical diagnosis, treatment and follow-up management of patients in response to disasters were included. Articles excluded from this study were as follows: (1) literature reviews, pictorial reviews, meta-analyses, opinion pieces, and case reports; (2) studies in which diagnostic imaging was not used for the direct diagnosis, treatment, or follow-up management of patients; (3) studies that did not involve human patients; (4) studies concerned with postmortem victim identification; (5) studies not related to an actual disaster, such as articles about disaster preparedness or studies involving simulated training. After the screening, a total of 177 articles met our inclusion criteria.
With PubMed IDs as unique identifiers, bibliometric data of these 177 articles were then retrieved from Web of Science All Databases, including author names, author affiliations, title, journal, publication year, abstract, and times cited in Web of Science All Databases. The abstract and full-text of each article were again examined to collect the following additional information: number of authors, country of origin (defined as the country of affiliation of the first author), primary affiliated department (defined as the department that the first author was affiliated with), study design (retrospective or prospective), sample size, disaster type, specific disaster event, and utilized imaging modalities.
Ranking of Articles
The articles were ranked by the average yearly citation, calculated as follows: total citation count from Web of Science All Databases/number of years between publication and January 2017, accurate to month (1/12th of a year).
Only the top 100 most highly cited articles, as ranked by the average yearly citation, were included in our final bibliometric analysis.
RESULTS
We identified a total of 177 articles published since 2000 that met our inclusion criteria. It has been widely accepted that, to a certain degree, the citation level of an article could indicate its relative significance in its research field.Reference Garfield 20 We thus ranked these articles by average yearly citation, and conducted the final bibliometric analysis on the top 100 most highly cited articles, as listed in Table 2. The usage of average yearly citation, instead of total citation, for the ranking allowed for a fair assessment of more recently published articles.
The 100 Most Highly Cited Original Research Articles Published Since 2000 Depicting the Roles of Diagnostic Imaging in Clinical Diagnosis, Treatment and Follow-Up Management of Patients in Response to Disasters, Ranked by Yearly Average Citation
Abbreviations: SARS, Severe Acute Respiratory Syndrome; MERS, Middle East Respiratory Syndrome.
Overall, the average total citation count was 23.3 times (highest citation: 231, median: 11). A total of 85 articles had been cited at least 5 times, 62 articles at least 10 times, and 29 articles at least 20 times. The average citation count per year was 2.89 times/year (highest yearly citation: 18.1 times/year, median: 1.82 times/year).
Type of Disaster
These 100 articles studied a total of 8 types of disaster, encompassing human-made and natural disasters, as listed in Table 3. The most studied disaster types were disease outbreak (55 articles), armed conflict (23 articles), terrorist incident (10 articles), and earthquake (7 articles).
Types of Disaster Studied in the 100 Most Highly Cited Articles
Individual Disaster
Of the 100 articles, 92 articles provided sufficient information to identify a specific disaster event; 1 article discussed 2 types of disasters. The most studied individual disasters were the H1N1 influenza outbreak in 2009 (28 articles), the SARS outbreak in 2003 (24 articles), the War in Afghanistan, 2001-2014 (8 articles), the Iraq War, 2003-2011 (6 articles), and the Sichuan earthquake (China) in 2008 (6 articles). Table 4 lists the individual disasters and the number of articles in which they were studied.
Individual Disasters, When Specified, and the Number of Articles Where They Were Studied
Abbreviations: SARS, Severe Acute Respiratory Syndrome; MERS, Middle East Respiratory Syndrome.
Year of Publication
Figure 1 illustrates the distribution of the 100 most highly cited articles by their year of publication. The largest numbers of highly cited articles were published in 2010 (15 articles), followed by 2004 (14 articles) and 2011 (13 articles). The yearly trend shows two peaks for article publication: one from 2003 to 2005, another from 2010 to 2012.
Distribution of the 100 Most Highly Cited Articles Since 2000 Depicting the Roles of Diagnostic Imaging in Disaster Management by Year of Publication.
This yearly trend correlates well with the most reported disasters in our analysis. From 2003 to 2005, a total of 29 highly cited articles were published, and 23 (79.3%) of them were about the SARS outbreak in 2003. From 2010 to 2012, a total of 40 highly cited articles were published, and 26 (65.0%) of them were about the H1N1 influenza outbreak in 2009, with another 4 (10.0%) about the Sichuan earthquake (China) in 2008.
A time lag existed between the occurrence of a disaster event and the publication of articles. For example, in the case of the 2008 Sichuan earthquake in China, although the disaster occurred on May 12, 2008, among the 6 highly cited articles studying this earthquake included in our analysis, 2 articles were published in 2009, 3 in 2010, and 1 in 2011, with an average disaster-to-publication time of around 2 years.
The publication of articles regarding the SARS and H1N1 influenza outbreaks, on the other hand, was much more rapid. SARS was recognized at the end of February 2003. 21 The first highly cited research article on SARS was published electronically as early as May 8, 2003.Reference Wong, Antonio and Hui 22 Similarly, the H1N1 influenza outbreak in 2009 was recognized in April. 23 The first highly cited research article about this outbreak was accepted for publication in September 2009.Reference Agarwal, Cinti and Kazerooni 24
Country of Origin
The country of origin of an article was determined as the country of the primary affiliation of the first author. The 100 most highly cited articles originated from 23 countries/regions, with most articles being from the United States (25 articles), the People’s Republic of China (24 articles), and Canada (6 articles). Table 5 lists these 23 countries with the number of articles from each country.
Countries of Origin of the 100 Most Highly Cited Articles Since 2000 Depicting the Roles of Diagnostic Imaging in Disaster Management
The countries of origin of these articles reflect the geographic or national relationship with the investigated disasters. For example, out of the 25 highly cited articles that originated from the United States, 10 were about armed conflicts in Iraq or Afghanistan in which United States was involved, 7 were about terrorist incidents in New York and Boston, and 5 were about the H1N1 outbreak. Out of the 24 articles from the People’s Republic of China, 14 were about the SARS outbreak that affected Mainland China and Hong Kong the most, and 6 were about the Sichuan earthquake in China. For Canada, 5 of the 6 highly cited articles were about the SARS outbreak, in which Canada was the most affected area in the Western world.
Authors
The articles had 2-21 authors (median: 7). Among the first authors, 59 were primarily affiliated with the Radiology department. Other identified medical specialties of the first authors were Surgery (12 authors), Emergency Medicine (9), Infectious Disease (4), other Internal Medicine specialties (4), Respiratory Medicine (3), Military Medicine (3), Pediatrics (2), Atomic Bomb Disease (1), Chemical Injuries (1), and Ophthalmology (1). For 1 article, no departmental affiliation was given for the first author.
Journal
These 100 most highly cited articles were published in 56 journals, including 18 Radiology journals. Other than radiology, these journals covered a wide range of topics, such as emergency medicine, surgery, military medicine, critical care, and respirology. The journal Radiology published the most highly cited articles (11 articles), followed by the American Journal of Roentgenology (12 articles) Table 6 lists the journals in which at least 2 highly cited articles were published, ranked in descending order of the number of articles published, and their corresponding impact factors for the year 2015, as obtained from the InCites Journal Citation Reports.
Journals That Published At Least 2 of the 100 Most Highly Cited Articles Since 2000 Depicting the Roles of Diagnostic Imaging in Disaster Management
Abbreviation: JCR, Journal Citation Reports.
Study Design and Sample Size
Out of the 100 most highly cited articles, 81 studies were retrospective, and 19 studies were prospective. Many of the prospective articles were follow-up studies on the treatment outcomes. For example, 9 out of 24 articles (37.5%) about the SARS outbreak were prospective. Seven of these 9 articles were specifically about treatment outcomes.
The range of sample sizes was 7-4365, with an average of 197.8 and a median of 72. These results are summarized in Table 7.
Distribution of Study Design and Sample Size of the 100 Most Highly Cited Articles Since 2000 Depicting the Roles of Diagnostic Imaging in Disaster Management
Imaging Modality
Several major imaging modalities, such as conventional radiography, angiography, ultrasound, computed tomography (CT), and magnetic resonance imaging were utilized in these 100 highly cited studies. A total of 64 articles involved only 1 modality (on average 1.4 modalities per article). CT was the most reported modality (52.8%), followed by conventional radiography (33.3%) and ultrasound (9.7%). Table 8 shows the number and breakdown of the modalities discussed in these articles.
Number and Breakdown of the Imaging Modalities Investigated in the 100 Most Highly Cited Articles Since 2000 Depicting the Roles of Diagnostic Imaging in Disaster Management
Table 9 shows a further breakdown of the two most utilized imaging modalities in several major disaster types/events. Again, CT and conventional radiography were ranked consistently as the most utilized modalities in most cases, with the only exception of ultrasound being the second most investigated modality in armed conflicts.
Top 2 Most Investigated Imaging Modalities in Major Disaster Types or Events
Abbreviation: SARS, Severe Acute Respiratory Syndrome.
Non-Patient-Centric Articles
Our literature selection and bibliometric analysis focused on original research articles that included patients in the study design. For readers’ interest, we also compiled a list of 15 non-patient-centric articles (Table 10) that did not include patients in the study design, but nevertheless could provide important insights into the roles of imaging in disaster responses. These articles typically provide either a narrative description of the experience of imaging services in disasters, summaries of work-flow adjustments, or guidelines and recommendations based on real experience in disasters.
The 15 Most Highly Cited Non-Patient-Centric Articles Since 2000 Discussing Diagnostic Imaging in Disasters, Ranked by Yearly Average Citation
Abbreviation: SARS, Severe Acute Respiratory Syndrome.
DISCUSSION
In this study, we conducted a bibliometric analysis of the top 100 most highly cited articles published since 2000 that depicted roles of diagnostic imaging in clinical diagnosis, treatment, and follow-up management of patients in response to disasters. By illustrating the diversity of medical practitioners from different subspecialties as well as the varied disaster management strategies and research strategies depicted in these publications, we revealed the trends of utility of imaging in this field that may guide future research and aid disaster management planning.
In our analysis, disease outbreaks, armed conflicts, terrorist incidents, and earthquakes were the most commonly studied disasters. A few reasons could explain why certain disasters were more highly represented than others. First, disasters such as disease outbreaks, earthquakes, and armed conflicts typically cause morbidity and mortality of a larger number of people, and consequently result in a higher number of medical care encounters.Reference Guha-Sapir, Hoyois and Below 25 Second, research on certain types of disasters, such as armed conflicts or terrorist incidents, may have received prioritized support from national governments, in the form of research infrastructure and research funding. This may explain the large number of articles studying armed conflicts in our analysis.Reference McGeary and Hanna 26 Third, the research output in disaster medicine research may reflect the research capacity or interest of local hospitals or institutions that are in geographic proximity to the disaster location. For example, Wenchuan County, the epicenter of the 2008 Sichuan earthquake in China, was only 80 km away from Chengdu, a major city with a population over 10 million.Reference Dong, Yang and Chen 27 In our analysis, we identified 6 articles that studied this earthquake, 5 of which were conducted in the West China Hospital, the largest hospital in Chengdu. Fourth, the relative numbers of publications reflect the relative utility of diagnostic imaging in disasters. The recent Ebola outbreak in West Africa caused substantial mortality and generated public attention; 18 , Reference Bluemke and Meltzer 28 however, because of the limited role of diagnostic imaging in the direct diagnosis of Ebola virus disease,Reference Bluemke and Meltzer 28 our literature search only identified one highly cited research article on this topic.
It is reasonable to expect that these articles correlated with the disasters they studied, both spatially and temporally. Spatially, in our analysis, most articles originated from a country that was either the location of the disaster, or in the case of armed conflicts, a participant of the conflict, with a notable exception of an article discussing responses to the 2015 Guatemala landslideReference Dean, Ku and Zeserson 29 that was published by a group in United States. Temporally, articles about a particular disaster typically clustered in a short period following the disaster occurrence, as illustrated in the Results section: compared with the Sichuan earthquake in 2008, the disaster-to-publication delay in the SARS and H1N1 influenza outbreaks was much shorter. This could be explained by the fact that disease outbreaks, unlike earthquakes, have a longer hazardous duration. Thus, faster publications could have been driven by the motivation to help end the disease outbreaks, with the information discussed and dispersed through the publications.
Limitations
Our study has its share of limitations. In our literature search, we only used the MEDLINE database developed by the National Library of Medicine. Our intention was to make extensive use of the highly sophisticated MeSH system, 17 as shown in our search methods. Because a disaster is a complex concept, the preindexation of references using the MeSH keywords can ensure we receive a comprehensive set of results. This may not be possible in many other databases, notably in the Web of Science. In addition, MEDLINE indexes more than 5600 journals and contains more than 23 million references in life sciences and medicine. 17 Although simultaneous use of other major databases could theoretically supplement the search results in many topics, MEDLINE is considered highly comprehensive in medical fields.Reference Sun, Xiao and Miao 30 - Reference Dym, Burns, Freeman and Lipton 33 Further, because our focus was on the most highly cited articles, it was quite unlikely that a highly cited article would come from a journal that was not indexed by MEDLINE. It is our experience in bibliometric research that almost all highly cited articles come from an overlapping list of established journals that were generally indexed by most major literature search engines.Reference O’Keeffe, Hanna and Holmes 11 - Reference Dolan, Hanna, Warraich, Johnson and Khosa 13 , Reference Shahzeb Khan, Fatima, Bin Riaz, Butler, Manning and Khosa 16
The most utilized imaging modalities identified in our analysis are CT (52.8%), conventional radiography (33.3%), and ultrasound (9.7%). Because of potential selection bias based on the interests of researchers and journals, the utilization of these modalities in the published literature may not represent the actual utilization rate in the field. For example, the widespread use of chest X-ray and high-resolution CT in the SARS and H1N1 influenza outbreaks may have skewed our overall data. We have tried to mitigate this issue by providing a breakdown of the most investigated imaging modalities by the major disaster types/events. Further, the research literature may under-represent modalities that were more characterized and mature in disaster management. Nevertheless, the research interest in using these modalities as revealed in our analysis may, to some extent, reflect their particular value in disaster situations. For example, the rapid image acquisition and wealth of information generated by CT could explain its wide use in triaging patients, assessing and characterizing injuries, as well as in following up patients after the initial encounter.Reference Sedlic, Chingkoe, Tso, Galea-Soler and Nicolaou 34 , Reference O‘Keeffe M, Clark, Khosa, Mohammed, McLaughlin and Nicolaou 35 The rapid point-of-care access of radiography and ultrasound, on the other hand, make these modalities ideal where diagnosis and treatment within the golden hour is critical.Reference Cowley 36
CONCLUSIONS
In conclusion, our study identified and analyzed the 100 most highly cited original research articles by average yearly citation published since 2000 that depicted the utility of diagnostic imaging in clinical diagnosis, treatment, and follow-up management of patients in response to a variety of disasters. These results offer an important insight into the utility of diagnostic imaging in response to various disasters and could help guide future research in developing disaster management plans.
Disclosures
The Department of Radiology, Vancouver General Hospital has a Master Research Agreement with Siemens Healthcare, Forchheim, Germany. Dr. Khosa is the recipient of the Canadian Association of Radiologist/Canadian Radiological Foundation Leadership Scholarship (2017).
