Epidemic hemorrhagic fever has been an ongoing threat to laboratory personnel involved in animal care and use. Laboratory transmissions and severe infections occurred over the past twenty years, even though the standards and regulations for laboratory biosafety have been issued, upgraded, and implemented in China. Therefore, there is an urgent need to identify risk factors and to seek effective preventive measures that can curb the incidences of epidemic hemorrhagic fever among laboratory personnel. In the present study, we reviewed literature that relevant to animals laboratory-acquired hemorrhagic fever infections reported from 1995 to 2015, and analyzed these incidences using fault tree analysis (FTA). The results of data analysis showed that purchasing of qualified animals and guarding against wild rats which could make sure the laboratory animals without hantaviruses, are the basic measures to prevent infections. During the process of daily management, the consciousness of personal protecting and the ability of personal protecting need to be further improved. Undoubtedly vaccination is the most direct and effective method, while it plays role after infection. So avoiding infections can’t rely entirely on vaccination.

Zoonoses, the inter-species transmitted infectious diseases, have become a serious public health problem. The World Health Organization (WHO) set reported that there are over 200 zoonotic diseases. Some of these diseases are spreading over the world, including epidemic hemorrhagic fever, which can be transmitted to humans from infected mice and rats; canine rabies, which is recognized as a fetal disease; tuberculosis and monkey B virus, which can both be transmitted from monkeys; and brucellosis, which comes mainly from cattle, sheep, and goats^[1]. Epidemic hemorrhagic fever is caused by hantaviruses, which can be directly transmitted to humans from mice and rats through contact with their fecal and urinal waste. The clinical manifestations of epidemic hemorrhagic fever include fever, bleeding, shock, and renal failure^[2]. It has been reported that patients with epidemic hemorrhagic fever are at high risk of death. Additionally, hantaviruses are originally carried by wild mice and rats, mainly by Apodemus agrarius and Rattus, and the hosts can spread the virus to humans from laboratory animals through several routes^[3].

In recent years, the standards and regulations for laboratory biosafety have been issued, upgraded, and implemented in China. Unfortunately, infectious diseases, in particularly zoonoses, continue to occur in laboratories, and these zoonoses can be caused by experimental animals. Of these infections, epidemic hemorrhagic fever has been the most frequently reported infection across the nation. In 2004, a serious epidemic hemorrhagic fever outbreak in China resulted in the infection of 9 laboratory personnel and one fatality due to renal failure^[4]. The laboratory-acquired hemorrhagic fever infections have also been reports in other countries^[5-7].

In the present study, literature that was relevant to the laboratory-associated infections reported in the period from 1995 to 2015 was reviewed and Fault Tree Analysis (FTA) was conducted. We aimed to identify and prioritize each basic fault for its importance to occurrence of epidemic hemorrhagic fever and propose the most effective preventive measure and other approaches to preventing and even eliminating future infections.

Literature Review

Wanfang Data, CQVIP database and PubMed were searched using the terms: animal laboratory, epidemic hemorrhagic fever. The publication date are from 1995 to 2015.

A total of 48 articles were acquired after the literature searches using Wanfang Data, 69 articles were acquired by CQVIP database, and only 1 article was acquired by PubMed. Then the iterative literatures were eliminated, and last a total of 71 articles were remained, among which 26 articles are related with epidemic hemorrhagic fever infections in laboratories. And 20 of 26 articles are about staff infections, 6 of 26 articles refer to laboratory animal infections (Figure 1). There were eleven incidences that occurred in seven different regions of China^{[4, 8-20]}. It has been well-documented that humans can become infected with Hantaviruses and develop epidemic hemorrhagic fever through contact with the urine, feces and saliva of rodents. Before the 1990s, epidemic hemorrhagic fever infections in humans were mainly caused by and transmitted through wild rodents (e.g. Apodemus and Rattus norvegicus), whereas after 1990s, hantavirus-infected laboratory mice and rats were considered the major source of infection to humans^[21]. In this study, we reviewed the events that occurred in China from 1995 to 2015, and found that infected laboratory mice and rats were the dominant cause of epidemic hemorrhagic fever among laboratory personnel, which was consistent with previous reports.

Fault Tree Construction    FTA is an important failure analytical tool. FTA was originally developed and introduced in 1962 by Watson at Bell Laboratory under the US Air Force Ballistics Systems Division to evaluate a missile launch control system^[22], and extended to a broad range of fields such as aviation, rail transportation, chemical industry and medicine^[23-25]. In FTA, an undesired top event is analyzed by using Boolean logic in order to combine a series of lower-level faults to determine not only the direct cause of the undesired event, but also indirect causes leading to the top event.

A fault tree, which is a logical diagram that allows the logical prioritization of contributors leading to the undesired top event, was constructed to identify basic faults or lowest level factors that may lead to epidemic hemorrhagic fever among laboratory personnel working with animals either directly or indirectly. We aimed to unveil the logical relationships of infection and to determine the most effective approach to preventing infection (Table 1). In this study, EasyDraw software was utilized to construct a fault tree diagram with event, gate and transfer symbols, from which the causes of infection were noted according to a review of the literature, and further revised and confirmed by two biosafety experts.

Upon the completion of the comprehensive literature review, we constructed a fault tree diagram. Laboratory epidemic hemorrhagic fever served as the top failure event that was first noted. Additionally, the possible reasons for infection, including pathogens harbored by experimental animals, exposure to animals (e.g. feeding or holding), unsuitable personal protective equipment (PPE), and lack of proper vaccinations, fell just below the top level of the fault tree as the second level of the intermediate events. With logic graphic symbols, the fault tree was broken down to lower levels and finalized by a total of 12 lower or basic elements or faults (X) which may lead to the undesired event, and probabilities were calculated (Table 2 and Figure 2).

Fault Tree Analysis: Minimal Cut Sets    After the fault tree was assembled, we conducted FTA. The premise was that if all the basic events occurred, then the top failure was inevitable. In most cases, however, a unique combination of basic events are necessary and sufficient to cause the undesired event, and the combination was defined as a minimal cut set and the number of minimal cut sets indicate the probability of the top failure event^[26-27]. Boolean algebra algorithms were used to calculate the number of the minimal cut sets, and a total of 32 minimal cut sets were obtained as follows:

{X1, X4, X5, X9}{X2, X3, X4, X5, X9}{X1, X4, X6, X9}{X2, X3, X4, X6, X9}{X1, X4, X7, X9}{X1, X4, X8, X9}{X2, X3, X4, X7, X9}{X2, X3, X4, X8, X9}{X1, X4, X5, X10}{X1, X4, X5, X11}{X2, X3, X4, X5, X10}{X2, X3, X4, X5, X11}{X1, X4, X6, X10}{X1, X4, X6, X11}{X2, X3, X4, X6, X10}{X2, X3, X4, X6, X11}{X1, X4, X7, X10}{X1, X4, X7, X11}{X1, X4, X8, X10}{X1, X4, X8, X11}{X2, X3, X4, X7, X10}{X2, X3, X4, X7, X11}{X2, X3, X4, X8, X10}{X2, X3, X4, X8, X11}{X1, X4, X5, X12}{X2, X3, X4, X5, X12}{X1, X4, X6, X12}{X2, X3, X4, X6, X12}{X1, X4, X7, X12}{X1, X4, X8, X12}{X2, X3, X4, X7, X12}{X2, X3, X4, X8, X12}.

As listed above, each minimal cut set contained 4-5 basic events. For instance, the cut set {X1, X4, X5, X9} was composed of 4 basic events: X1, animals purchased without complying with established standards; X4, lack of vaccination; X5, without PPE; and X9, injury by scratch, bite or edge tools. However, for laboratory workers who were not vaccinated against epidemic hemorrhagic fever nor wore PPE, and were injured by scratches and bites by infected animals, their risk of infection was much greater. Another minimal cut set {X2, X3, X4, X8, X12} consisted of five basic faults. If laboratory personnel worked in animal facilities within disease outbreak regions where no appropriate equipment to block wild rodents from entering animal facilities, and get exposed to wastes from the infected animals, but did not wear PPE or wore inappropriate PPE, they were likely to be infected by aerosols. In this study, there were 32 minimal cut sets, indicating at least 32 basic fault events that could occur and lead to epidemic hemorrhagic fever if the feasible precautions were not properly undertaken.

Fault Tree Analysis: Minimal Path Sets    If none of the basic events occurred, then the top event would not occur. Meanwhile some of the basic events could occur separately or non-simultaneously. The combinations of these basic events were not sufficient to cause the top event, and they are defined as path sets. The minimal path sets referred to the safe modes in which the top event would not occur if these basic events would be prevented simultaneously. Therefore, the fault tree was transformed into the successful tree and produced five minimal path sets which were as follows: {X1, X2} {X5, X6, X7, X8} {X9, X10, X11, X12} {X4} {X1, X3}.

As revealed, there were 1-4 basic events in a minimal path set, in which the undesired event would not occur if all the basic events in the minimal path sets would not happen. Each preventive measure, in turn, was identified from the minimal path sets that could be used to prevent epidemic hemorrhagic fever. In this study, a total of five minimal path sets were identified, representing at least five approaches that could prevent the undesired event (Table 3).

The five different preventive measures in terms of feasibility and efficiency were prioritized. As a result, being vaccinated was considered as the most effective approach to preventing epidemic hemorrhagic fever, and other preventive measures included improving training for laboratory personnel, compliance with standards and regulations, installation of adequate equipment in animal facilities and adornment of PPE. Among the preventive measures in the minimal path sets, however, avoiding the production of biological aerosols was the least feasible.

Structural Importance Analysis    Structural importance analysis was performed to prioritize each basic fault for its importance to the undesired top event, and the impact on the failure rates in the fault tree. According to the minimal cut sets, the order of basic faults for their importance to the top event was as follows: I [X4]> I [X1]> I [X2] = I [X3]> I [X5] = I [X6] = I [X7] = I [X8] = I [X9] = I [X10] = I [X11] = I [X12]. As displayed above, the basic fault, not being vaccinated against epidemic hemorrhagic fever affected the occurrence of the top event most, followed by the two basic faults: purchase of animals that did not comply with standards and regulations, and location of facilities within areas of disease outbreaks, which were more important to the top event than the following basic faults of wearing unsuitable PPE and exposure to the infected animals.

Main Findings and Suggestion

In this study, 12 basic elements were analyzed and the possibilities of each element were assessed using a fault tree diagram. Next, FTA was conducted, and 32 minimal cut sets and 5 minimal path sets were found to be associated with the epidemic hemorrhagic fever. Finally, the structural importance was analyzed, and results showed that not receiving a vaccination affected the occurrence of the top event more than any other factor did.

To the best of our knowledge, this is the first report that utilizes FTA to analyze the undesired infection in the field of biomedicine. Similar to FTA for safety issues occurring in different industries, all factors leading to the infection, such as facility, animals, handling procedures, skills and management were investigated and analyzed. Our main findings in this study are summarized and discussed below.

Firstly, laboratory management should be strengthened to ensure no pathogens are transmitted by laboratory animals. Having built upon our analysis in this study, we suggest the following preventive measures: (1) Laboratory animals should be purchased in compliance with the national standards and regulations as described in GB14922.2-2011^[28]. In particular, experimental mice and rats should not carry hantavirus. Although the strict regulations and standards have been issued and implemented^[29], unfortunately, in 2004 an infection event was reported as a direct consequence of a Wistar rat, that was believed to be infected at the time of purchase^[4]; (2) Animal facilities and laboratory management should be improved and preventive measures should be taken to block wild rodents within the disease-outbreak regions from entering animal facilities thus transmitting infection to laboratory animals. While it may seem impossible not to have any animal facilities in the areas of disease outbreaks, the improvement and regular inspection of animal facilities, along with self-inspection, can be made to effectively prevent infection; (3) Screening and detection of viruses should be regularly conducted in order to identify infected animals as early as possible.

Secondly, operating procedures should be optimized and direct contact with animals and their wastes should be avoided. We must strengthen the training programs for laboratory personnel to improve their experimental skills to minimize exposure. In the mean time, standard operating procedures for disposal of animals and their wastes should be implemented strictly in order to reduce risk of infection as much as possible.

Thirdly, the present study showed that being vaccinated against epidemic hemorrhagic fever was the most effective measure in preventing infections for laboratory personnel. Therefore, laboratory workers who handle experimental animals should be offered vaccines against epidemic hemorrhagic fever infection. In fact, the administration of vaccines against epidemic hemorrhagic fever provides the laboratory personnel with important advantage by greatly reducing the risk of epidemic hemorrhagic fever. Bivalent inactivated vaccines, which are available to individuals in disease-outbreak areas nationwide at no cost, function by stimulating the immune system to produce a specific antibody against both type Ⅰ and type Ⅱ epidemic hemorrhagic fever. In 2005, an infectious incidence of epidemic hemorrhagic fever resulted in four out of five individuals being infected, with the one who was not infected having been vaccinated against epidemic hemorrhagic fever^[14]. Until now, there has been no information about vaccination rates among laboratory personnel working with animals, but it is estimated that the rate may remain low across the nations. The incidence noted above, along with our analysis in this study, demonstrated that the administration of the vaccines may effectively prevent infection. Of course, vaccination is the last defense against infections. Thus, we cannot completely rely on vaccination ignoring other protections.

Finally, awareness of biosafety should be raised and personal protection should be enhanced. A wealth of previous studies have demonstrated that the aerosol has been the most common way to cause infection, but aerosol generation would be difficult to avoid. Since the proper use of PPE is very important, funds to purchase adequate PPE should be provided, and the training program for laboratory personnel on proper use of PPE should be offered.