Risk factors of earthquake inpatient death: a case control study

At 2:28 p.m. on 12 May, 2008, a devastating earthquake hit Wenchuan County in southwest China's Sichuan Province. Measuring 8.0 on the Richter scale, the earthquake had a maximum intensity of φ (Modified Mercalli Intensity Scale) and a depth of 33 km. It led to 69,197 deaths, 374,176 injuries and 18,222 people being reported missing by 12:00 noon on 20 July. The severity and scale of this disaster has rarely been seen before, either in China or internationally.

The earthquake occurred in a rural and mountainous region in western Sichuan Province, about 50 miles (80 km) west-northwest of Chengdu (the location of the West China Hospital). West China Hospital of Sichuan University, as the only large-scale, state-level and general teaching hospital in the earthquake-affected area, received the largest number of injured people, treating a total of 2728 cases (872 were treated in the emergency department and 1856 were admitted) [1] from the disaster area. Until 11 October, only 36 of the injured had died in the hospital (fatality rate of 1.32%).

Understanding risk factors related to earthquake inpatient death is essential for effective patient management. Some previous studies have exposed the factors affecting casualties [2‐4] and the treatment of disaster trauma [5]. Few epidemiological studies, however, have been conducted on the determinants of mortality of patients hospitalised after an earthquake. This paper is aimed at identifying the contributing factors to mortality and providing future clinical reference for management of earthquake-related injuries.

The Institutional Review Board (IRB) of the West China Hospital in Sichuan University approved this study. No informed consent was necessary as this study used existing data, which was obtained with IRB approval and informed consent from IRB.

The male to female ratio for both groups ws 1 to 1.1. The mean (standard deviation (SD)) age and days before admission were 65.8 (21.0) years and 7.3 (5.9) days for cases, and 64.3 (20.9) years and 7.2 (6.0) days for controls. Less than 20% of patients were transferred to the West China Hospital within 72 hours after the earthquake for both cases and controls.

Table 1 presents the frequency of factors used in the case-control study. The case group contained a greater percentage of severe TBI, crush syndrome, ARF, thoracic organ injury, celiac organ injury, amputation, cranium fracture, infection, MSOF and prior major disease than the control group, although the control group constituted a large proportion of local hospital treatment, trunk fracture and extremity fracture.

Table 2 lists both univariate and multivariate effects of included variables. After entering all co-variables simultaneously in conditional logistic regression analysis, severe TBI, infection, MSOF and prior major disease were the significant determinants of earthquake-related inpatient death in the 2008 Wenchuan earthquake. Patients with a severe TBI had the greatest risk of death (adjusted OR = 253.3, 95% CI = 8.9 to 7208.6), followed by patients with MSOF (adjusted OR = 87.8, 95% CI = 3.9 to 1928.3). Having a prior major disease or being infected significantly increased the risks of earthquake-related death (adjusted OR = 14.9, 95% CI = 1.9 to 119.0; adjusted OR = 13.7, 95% CI = 1.8 to 103.7; respectively).

In univariate analysis, local hospital treatment lowered the risk of earthquake-related death (OR = 0.4, 95% CI = 0.2 to 0.9). Patients with crush syndrome, ARF and celiac organ injury had higher risk of death (OR = 6.7, 95% CI = 1.6 to 27.9; OR = 6.4, 95% CI = 2.1 to 19.5; OR = 4.8, 95% CI = 1.5 to 15.7, respectively). After adjustment for other variables, however, all four variables demonstrated no statistical significance.

Table 3 presents the case and control numbers by different risk factor numbers, along with case fatality. The chi-squared test for trend demonstrated that the case fatality increased as the factor numbers raised (x² = 21.864, p = 0.000).

The magnitude 8.0 earthquake that struck Sichuan Province on 12 May, 2008, was the strongest earthquake China has experienced in over 50 years. It caused the largest loss of life since the Tangshan earthquake in 1976. Identification of determinants for earthquake-related death is critical to triage, transfer, nursing and treatment, as well as for treating patients in shock. This paper revealed that severe TBI, infection, MSOF and prior major disease were the significant risk factors associated with earthquake-related inpatient death in the 2008 Wenchuan earthquake.

Over 60% of the dead were older than 60 years of age. Similar findings were published elsewhere [4, 10‐13]. Due to the fact that the epicentre was located in a mountainous region and there were lots of collapsed roads, the transfer of thousands of severely injured patients was delayed, which is the main reason why many patients were admitted to our hospital up to seven days after the earthquake.

People with severe TBI have a high risk of death. Bouras and colleagues [14] reported that the severe TBI case fatality rate among the senior population (65 years or older) was 86.9%, similar to the findings of a recent review [6]. This study demonstrated that earthquake inpatients with severe TBI have a large risk of death (adjusted OR = 253.3). This raises ethical questions of whether active treatment should be initiated once the severe TBI has been diagnosed, particularly among the older population. Infection, usually not a threatening factor, was a primary cause of earthquake inpatient death in our study, which means physicians need to pay more attention to testing, preventing and treating inpatients that might have an infection. Moreover, this finding was consistent with a previous study that revealed that the mortality rate was significantly higher in those patients with hospital-acquired infection than those without [15]. That the patient infection rate was notable in our hospital was high-probably related to: delayed transfer because of broken roads; heavy damage to the front-line hospital; and insufficient drugs, equipment and physicians. Although two articles [15, 16] have shown that MSOF is a primary cause of death, it was the conclusion of descriptive and qualitative studies. We presented the quantitative risk of MSOF to cause death using multivariable analysis, and demonstrated that it is responsible for inpatient mortality. Notably, MSOF may or may not result from infection. An previous article [17] reported that MSOF is primarily due to infection and the most common fatal expression of uncontrolled infection. If this was also true in our study, the OR estimates might be inaccurate. However, it is very difficult to determine the causality between MSOF and infection, and both are important predictors of death. Few articles have addressed the impact of prior major diseases or original diseases on the prognosis of earthquake traumatised patients. Multivariable analysis, however, revealed that the patient's prior major disease before the earthquake is responsible for the risk of death after the earthquake. Thus, physicians must pay attention to treating the prior diseases when treating new trauma in earthquake patient.

Univariate analysis revealed that local hospital treatment, crush syndrome, ARF and celiac organ injury were related to patient death. Although the four factors failed to be recognised as influential to mortality in multivariate analysis, they should be carefully managed when considering their clinical importance. The fact that local hospital treatment might reduce the risk of death suggests that rapid medical rescue plays an essential role in saving lives. One drawback of this study is that the information of patients' staying in a local hospital before transferring to our hospital was not available. It should be remembered that delay to definitive care is one of the most important predictors of outcome after trauma. Two previous papers [13, 16] reported that crush syndrome, ARF and vital organ injury account for mortality to a certain extent, similar to our findings.

Whatever the main risk factors for earthquake inpatient death are, the fact that those multi-factors (except local hospital treatment) co-existed might indeed increase the risk of patient death. The result of chi-squared tests for trend provided meaningful evidence that patients who were recorded with more risk factors were more vulnerable to death, which will benefit the earthquake patient management.

To the best of our knowledge, this paper is the first hospital-based case-control study on risk factors pertaining to earthquake inpatient death; however, there are some limitations to this study. To begin with, we failed to analyse the impact of age and sex on death because of matching. Secondly, we did not calculate sample size because this was an emergent natural disaster study, and nobody knows how many injured people will die in the hospital. This might lead to insufficient statistical power to disclose risk factors. Thirdly, the authors failed to consider the seismic and structural factors, along with time under the rubble, which could be associated with earthquake-related death. Nevertheless, compared with population-based case-control studies, this paper has provided information which directly relates to medical relief and earthquake inpatient management. Finally, a selection bias (e.g. admission rate bias or Berkson bias) should be noted. One of the most important exclusions from hospital-series case-control studies are at-scene deaths. As most earthquake deaths do occur outside the hospital, it is a critical exclusion for trauma death numbers. Therefore, the causes of death for hospitalised patients may dramatically differ from at-scene deaths. Moreover, although the discharged earthquake patients were at low risk of death, we failed to follow up with them and the post-discharge deaths (if any) were excluded. This might restrict the general applicability of our study results. However, what this paper demonstrated can be used when managing earthquake inpatients.

Based on the results of this case-control study along with the treatment experience in our hospital, some suggestions might be helpful when managing future earthquake patients as follows: older patients with earthquake trauma (especially those with prior major diseases) are the population at highest risk and more attention should be paid to them; an evidence-based system of triage should be drawn up to distinguish patients with different severities of injury; screening, treating and following up infections are vital to the patients' survival and once patients are infected, they should be isolated; and a team of multidisciplinary doctors should be organised as soon as possible to monitor, diagnose and treat critical cases, especially those in the intensive care unit.

The cause of earthquake-related death is multi-factoral and complicated. Better identification of the risk factors for death and their relation with each other can help medical providers identify vulnerable populations and take corresponding measures to reduce mortality. Future research with a large sample size, including macro- and micro-level factors, is needed.

Severe TBI, infection, MSOF and prior major disease were the significant risk factors associated with earthquake-related inpatient death, and should be given more attention when managing patients after an earthquake.