Fatal earthquakes have become a frequent occurrence. Understanding the related factors associated with earthquake inpatient death will help reduce mortality among patients hospitalized after an earthquake.Reference Wen, Shi and Li1 Previous studies have discussed different related factors of earthquake inpatient mortalities. These studies each analyzed 1 or 2 different related factors, such as young age (≤15 years),Reference Dulski, Basavaraju and Hotz2 old age (≥65 years),Reference Cheng, Nieh and Hsu3, Reference Osaki and Minowa4 gender,Reference Peek-Asa, Ramirez and Seligson5 prehospital treatment,Reference Hai, Ya-rong and Xin-miao6 and admission to an intensive care unit (ICU).Reference Dulski, Basavaraju and Hotz2 Additional related factors include the presence of a severe traumatic brain injury (TBI),Reference Wen, Shi and Li1 trunk injury,Reference Hai, Ya-rong and Xin-miao6, Reference Hu, He and Zhang7 severe poly-trauma,Reference Gao, Yang and Yuan8 crush syndrome,Reference Bulut, Fedakar and Akkose9 multiple-system organ failure (MSOF),Reference Wen, Shi and Li1 infection,Reference Wen, Shi and Li1 or cardiac/respiratory disease.Reference Kloner10, Reference Tanaka, Oda and Iwai11 However, there are very few epidemiological studies that analyze multiple related factors of earthquake inpatient deaths after drawing from related factors in previous studies. We selected patients from the West China Earthquake Patients Database, which holds the prehospital, emergency, inpatient, and discharge information of inpatients from 4 earthquakes (see online Supplemental Appendix 1) that have taken place over the past 11 years in west China. This database is managed by Sichuan University’s emergency medical rescue base, and at the time of writing, held 36,604 patient records from 701 hospitals. The data set has previously been used to study single factors, such as age, crush syndrome, and infection for the prognostic evaluation of earthquake victims. As it is difficult to standardize the data in studies concerning earthquake casualties, the results of single factor analyses typically have many potential confounding factors. Therefore, we conducted an unprecedented multi-factor analysis study with the aim of showing the true effect of the related factors on inpatient deaths studied previously.
METHODS
The Institutional Review Board (IRB) of West China Hospital, Sichuan University approved this study. Taking informed consent was unnecessary as this study used an already existing data base as obtained through the informed consent of the IRB.
Study Design
This study used the West China Earthquake Patients Database’s patient characteristics and outcomes to screen suspected mortality related factors of hospitalized patients.
Selection of Participants
After excluding 3628 missing data (Figure 1), we analyzed the remaining 32,976 cases in our study. Of these, we analyzed the records of the 284 patients who died during hospitalization.
Flow Chart of Patient Exclusion.
Key Variables
Using the available information from the West China Earthquake Patients Database, we collected data on 12 dichotomous variables with reference to previous reports,Reference Wen, Shi and Li1–Reference Tanaka, Oda and Iwai11 which we labeled: pediatric patient, geriatric patient, gender, prehospital treatment, ICU admission, severe TBI, trunk injury, severe poly-trauma, crush syndrome, MSOF, infection, and cardiac/respiratory disease. We defined these dichotomous variables as follows: (1) Pediatric patient means a patient younger than or equal to 15 years old; (2) Geriatric patient means a patient older than or equal to 65 years old; (3) Gender means whether the patient was male or female; (4) Prehospital treatment means that the earthquake patient received primary medical treatment by health-care providers before being transferred to any hospital; (5) ICU admission means that the patient was admitted to the ICU, including adult ICU, neonatal ICU, pediatric ICU, or other intensive care wards, while hospitalized; (6) Severe TBI was defined as a single head injury resulting from an external mechanical force due to the earthquake, along with a Glasgow Coma Score of 8 or belowReference Wen, Shi and Li1; (7) Trunk injury was defined as a single chest or abdomen injury, judged by radiological examination (eg, a computed tomography scan or magnetic resonance imaging); (8) Severe poly-trauma means that more than 1 anatomical area or organ is severely traumatized, with an Injury Severe Score (ISS) greater than or equal to 15 points; (9) Crush syndrome refers to a severe shock-like condition that follows the release of a large body part, such as a limb, after a prolonged period of mechanical compression. Crush syndrome is characterized by hypovolemic shock, hyperkalemia, and renal failure; (10) MSOF refers to a clinical syndrome in which 2 or more organ failures occur simultaneously or sequentially. MSOF often follows trauma, major surgery, severe infection, and so on; (11) Infection refers to a microorganism of disease as diagnosed by clinicians, according to clinical symptoms, laboratory examination, or radiological examination; (12) Cardiac/respiratory disease refers to an inpatient’s diagnosis such as underlying cardiac/respiratory disease, pre-existing cardiac/respiratory conditions, and acute cardiac/respiratory disease, and is classified using a modified CDC Natural Disaster Morbidity Surveillance instrument.Reference Dulski, Basavaraju and Hotz2
The main outcome variable was in-hospital death, and included death taking place in a hospital’s emergency department, and death during hospitalization. As a result, mortality was defined as the ratio of death to the total number of patients in each of the 12 aforementioned variables.
Statistical Analysis
Measurement data with normal distributions were described by a mean ± standard deviation. Measurement data with nonnormal distributions were described by medians (25% quartiles, 75% quartiles). Enumeration data were constructed using composition ratios. Frequency counts and percentages were evaluated for all related factors (ie, variables). To reveal any association between a given variable and a case fatality, we performed a Chi-squared (χ2) test for trends. We constructed logistic regression analysis to determine the independent related factors of earthquake inpatient death. We calculated adjusted odds ratio (OR) estimates with 95% confidence intervals (CI) from the logistic regression analysis. We set the threshold for statistical significance at P < 0.05 for all analyses. This analytic process was conducted using SPSS statistics 17.0 software (SPSS Inc., Chicago, IL).
RESULTS
The mean (standard deviation) age was 45.03 (21.39) years for survival, and 54.95 (24.24) years for death. The median (25% quartile, 75% quartile) days before admission were 7 (2, 14) days for survival, and 4 (2, 12) days for death.
The frequency of factors used in the study is presented in Table 1. The death cases contained a greater percentage of severe TBI, MSOF, cardiac/respiratory disease, crush syndrome, ICU admission, prehospital treatment, and age (both ≤15 years and ≥65 years) compared with the survival cases (P-value < 0.05).
Frequency of Factors Used in This Study
Abbreviations: ICU, intensive care unit; MSOF, multiple-system organ failure; TBI, traumatic brain injury.
* P < 0.05.
Table 2 shows the 12 factors that predict earthquake inpatient death, as assessed using a univariate logistic regression analysis. MSOF was the greatest risk factor of inpatient death (OR, 80.862; 95% CI, 61.602-106.144), followed by severe TBI (OR, 32.248; 95% CI, 22.505-46.210). The significant variables in order of OR values were MSOF (OR, 80.862; 95% CI, 61.602-106.144), severe TBI (OR, 32.248; 95% CI, 22.505-46.210), ICU admission (OR, 16.128; 95% CI, 12.495-20.818), cardiac/respiratory disease (OR, 10.998; 95% CI, 8.657-13.973), crush syndrome (OR, 6.088; 95% CI, 4.412-8.399), old age (OR, 2.812; 95% CI, 2.220-3.562), infection (OR, 2.796; 95% CI, 2.172-3.600), prehospital treatment (OR, 2.125; 95% CI, 1.186-3.807), severe poly-trauma (OR, 2.090; 95% CI, 1.599-2.750), trunk injury (OR, 1.153; 95% CI, 0.856-1.553), female patients (OR, 0.844; 95% CI, 0.667-1.066), and young age (OR, 0.633; 95% CI, 0.405-0.988).
Univariate Analysis of Related Factors for Earthquake Inpatient Death
Abbreviations: 95% CI, 95% confidence intervals for odds ratio; ICU, intensive care unit; MSOF, multiple-system organ failure; OR, odds ratio; TBI, traumatic brain injury.
* P < 0.05.
Table 3 shows the related factors associated with earthquake inpatient death, as assessed using multivariate logistic regression analysis. Model 1 included 12 related factors: Severe TBI was the greatest risk factor of inpatient death (OR, 36.811; 95% CI, 23.205-58.395), followed by MSOF (OR, 30.760; 95% CI, 21.574-43.858). The next most significant variable in order of OR values were cardiac/respiratory disease (OR, 4.666; 95% CI, 3.311-6.576), crush syndrome (OR, 2.899; 95% CI, 1.885-4.461), ICU admission (OR, 2.382; 95% CI, 1.662-3.414), prehospital treatment (OR, 2.063; 95% CI, 1.046-4.066), old age (OR, 2.169; 95% CI, 1.626-2.893), severe poly-trauma (OR, 1.214; 95% CI, 0.859-1.715), trunk injury (OR, 0.998; 95% CI, 0.695-1.431), infection (OR, 0.835; 95% CI, 0.588-1.184), female patients (OR, 0.809; 95% CI, 0.619-1.057), and young age (OR, 0.624; 95% CI, 0.372-1.046). For cardiac/respiratory disease, old age is most likely to be a confounding factor, so we operated model 2, which deleted this variable. We compared the 2 models and found that the OR of cardiac/respiratory disease changed from 4.666 to 5.645 (change rate 20.98% > 10%),Reference Kernan, Viscoli and Brass12 suggesting that old age is part of the confounding factor of cardiac/respiratory disease. We also found there was a significant difference of the OR of the crush syndrome between model 1 and model 2. These results suggested that old age was also a confounding factor of crush syndrome.
Multivariate Analysis of Related Factors for Earthquake Inpatient Death
Abbreviations: 95% CI, 95% confidence intervals for odds ratio; ICU, intensive care unit; MSOF, multiple-system organ failure; OR, odds ratio; TBI, traumatic brain injury.
a Model 1 includes all the related factors; Model 2 deletes the variable of old age in Model 1.
According to the stepwise logistic regression analysis (backward procedure) in the last step model, there were 6 related factors associated with earthquake inpatient death (see online Supplemental Appendix 2) and another 6 related factors (young age, gender, severe poly-trauma, trunk injury, prehospital treatment, and infection) could be removed. Similar to the findings of previous studies, severe TBI was the greatest risk factor of inpatient death (OR, 31.913; 95% CI, 20.484-49.720), followed by MSOF (OR, 30.905; 95% CI, 21.733-43.947). The next most significant of the variables in order of OR values were cardiac/respiratory disease (OR, 4.427; 95% CI, 3.259-6.014), crush syndrome (OR, 2.880; 95% CI, 1.882-4.408), ICU admission (OR, 2.369; 95% CI, 1.656-3.389), and old age (OR, 2.169; 95% CI, 1.626-2.893). The Akaike information Criterion (AIC), Bayesian Information Criterion (BIC), Cox-Snell R2, and NagelkerkeR2 were not largely different between models (see online Supplemental Appendix 3), which indicates that both the first step and last step models were not significantly different from each other.
DISCUSSION
Our study found 6 factors associated with earthquake inpatient death risk: old age, ICU admission, the presence of severe TBI, crush syndrome, MSOF, and cardiac/respiratory disease.
Old age was a high-risk factor. The study by Pant and BanjadeReference Pant and Banjade13 on the Nepal earthquake suggests that the impact of the earthquake on the elderly population was not limited to personal injury, and that drug shortages and irregular inspections of post-earthquake care facilities increased the adverse consequences of chronic disease for these patients. Previous studies on the Wenchuan earthquake also reported a supply shortage of medicines and equipment in hospitals due to the destruction of infrastructure, such as road systems after the earthquake.Reference Wang and Chen14
It was not surprising that ICU admission was associated with earthquake inpatient death in this study, as ICU patients generally have higher mortality rates. Halpern et al.Reference Halpern, Rosen and Carasso15 inferred that factors associating ICU admission with death in earthquakes include the severity of the trauma, disease, the availability of blood, and appropriate antibiotics in resource-limited settings. Even so, there is limited evidence to reveal the relationship of intensive care resources and mortality in an earthquake. Further study on intensive care professionals and equipment should be done to illuminate the role of early recruitment and to help deploy critical medicine resources.
Severe TBI strongly predicted earthquake inpatient mortality. Previous studies,Reference Igarashi, Matsumoto and Kubo16 including the study by Zhang et al.Reference Zhang, Zhao and Fu17 on the Lushan earthquake and the study by Wen et al.Reference Wen, Shi and Li1 study on the Wenchuan earthquake, have emphasized that TBI plays an important role in predicting the death of earthquake patients. After reviewing 123 earthquake studies from 1990 to 2010, Bartels and vanRooyenReference Bartels and VanRooyen18 concluded that earthquake patients with severe TBI were “generally not saveable.” Presently, however, no research has proposed an effective way to solve this problem, let alone solve it in earthquake-stricken areas with scarce medical resources.
In our study, MOSF had a high OR value, which suggested that it was also an important factor in earthquake inpatient death. The review by Bartels and VanRooyenReference Bartels and VanRooyen18 lists MSOF as the most common cause of delayed mortality in earthquake inpatients. Even in normal situations, MSOF has a very high mortality rate given the usual lack of medical resources following an earthquake, and MSOF deaths are even more prominent in earthquake inpatients. Like severe TBI, there is currently no effective program to help reduce earthquake deaths caused by MSOF. After the Wenchuan earthquake, some Chinese health-care providers proposed that patients suffering from MSOF needed to be transferred to hospitals far away from the disaster area that had sufficient medical resources.Reference Zhu, Ji and Junxing19 This suggestion could be seen as a solution to reduce the mortality rate.
Crush syndrome was also strongly associated with earthquake inpatient death.Reference Najafi, Safari and Sharifi20 Previous studies have reported the high mortality of earthquake inpatients from crush syndrome as having a mortality rate of 13.4% after the Hanshin-Awaji earthquakeReference Tanaka, Oda and Atsushi21 and 21% after the Marmara earthquake.Reference Bulut, Fedakar and Akkose22 In our study, the mortality rate of earthquake inpatients from crush syndrome was 16.19%. The study by He et al.Reference He, Wang and Li23 on the Wenchuan earthquake reported that patients with crush syndrome had a higher mortality, which was probably related to severe complications, especially hyperkalemia. The present study also found that old age is a confounding factor of crush syndrome. Few previous studies have explored the relationship between old age and crush syndrome in earthquakes. The study by Ting et al.Reference Li, Jiang and Chen24 on the Yushu earthquake reported that 1 of 7 patients with crush syndrome was an elderly person, which may relate to the risk of the elderly being easily squashed by heavy objects. Further research can be conducted in the future to explore the relationship between these 2 factors.
In addition, cardiac and/or respiratory diseases were also frequently encountered in the hospital throughout the earthquake after-response. Previous reportsReference Kloner10,Reference Tanaka, Oda and Atsushi21 revealed an increased number of cardiac events and respiratory diseases following an earthquake. These cardiac events included fatal myocardial infarction, stress cardiomyopathy, heart failure, arrhythmias, hypertension, and sudden cardiac death.Reference Dulski, Basavaraju and Hotz2, Reference Kloner10 The major respiratory diseases in earthquake inpatients included pneumonia, acute exacerbation of chronic obstructive pulmonary disease (AECOPD), asthma attacks, progression of lung cancer, adult respiratory distress syndrome, and pulmonary embolism.Reference Dulski, Basavaraju and Hotz2, Reference Tanaka, Oda and Atsushi21 The incidence of respiratory disease may be affected by air pollution caused by landslide disaster and waste dust.Reference Xue, Yang and Sun25 Furthermore, the lack of medical personnel and equipment in the earthquake-stricken areas also led to an increase in mortality from these causes. As such, we suggest that hospitals in earthquake-prone areas should develop response plans based on realistic approaches to the problems (and limitations) they are likely to face regarding patients with cardiac and/or respiratory disease in earthquake scenarios. In addition, we also found that old age was a confounding factor of cardiac/respiratory disease. It is easy to understand that the incidence of chronic heart and lung disease and underlying heart and lung disease would be increased with advanced age. We found in this study that cardiac/respiratory disease is a related factor of inpatient death, regardless of whether old age is added to the model. Further study should be done to illuminate the degree of association between cardiac/respiratory disease and inpatient death.
Our study is subject to the following limitations. First, like some previous studies, our study dealt with incomplete records of inpatient data. We excluded 3628 cases with missing data, and, therefore, may be inaccurately estimating the importance of certain related factors of earthquake inpatient deaths. Second, we failed to consider injury mechanisms, structural factors, seismic features of earthquakes,Reference Chou, Huang and Lee26 and did not account for differences in search and rescue times. All of these could be relevant factors in predicting earthquake inpatient death. Third, the 4 earthquakes included in our database each occurred in rural areas. Compared with urban areas, rural areas are characterized by low population density and low variety and volumes of medical services. Thus, casualties and factors affecting inpatient deaths may be different in urban earthquakes. Finally, we only studied deaths during hospitalization and did not discuss the length of hospitalization in this study, because we were unable to gain enough information from the database. Further study on the relationship between length of stay and mortality is required.
Despite these limitations, this study is, to the best of our knowledge, among the first to explore related factors of inpatient deaths using data from multiple hospitals and multiple earthquakes.
CONCLUSIONS
To reduce the mortality of hospitalized patients after an earthquake, the related factors identified and analyzed in this study, such as old age, ICU admission, severe TBI, crush syndrome, MSOF, and cardiac/respiratory disease, should be the focal point of future earthquake response strategies regarding inpatient care.
Funding
This work was financially supported by the Science Foundation of Science and Technology Department of Sichuan Province (2015JPT0026) and the Science Foundation of Sichuan University (2013SCU19020).
Conflicts of Interest
The authors declare there are no conflicts of interest.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/dmp.2020.125.
