Risk assessment is a powerful tool that provides a rational framework for designing and managing an OHSP at institutions that use nonhuman primates. The process of risk assessment requires a factual base to define the likelihood of adverse health effects of workplace-associated injuries and exposures, and it attempts to balance scientific knowledge with concerns of staff, investigators, administration, and the public at large. It involves a systematic approach to the identification and characterization of physical, chemical, and biologic hazards to individuals and populations in their environment. The consequences of such hazards can include severe illness or injury, an irreversible health consequence, an unfamiliar disease, and an undesirable situation that might have been avoided by use of an alternative approach or technology. Risk assessments typically require that attention be given first to the most important hazards, that is, the ones that can result in the worst health-related outcomes.
Successful risk assessment offers many advantages. For staff members, a well-defined assessment of risks in the workplace can provide a rational basis for safe practices and behavior. For institutional managers, a well-defined assessment of risks can provide clear targets for injury-prevention and exposure-prevention programs. For regulators and other oversight bodies, a well-defined assessment of risks helps in setting workplace health and safety standards and in monitoring compliance without the need for case-by-case judgments. For concerned citizens, a well-defined assessment of risks provides a concise focus for evaluating protection of the public welfare.
The purpose of risk assessment is to determine the probability of injury or illness due to specific hazards. Risk assessment also includes characterization of the uncertainties inherent in the process of inferring risk. The process in turn becomes the basis of risk management—courses of action to mitigate hazards at the national, regional, and local levels through the establishment and modification of regulatory standards and institutional occupational health and safety programs. Several key terms and concepts are used in risk assessment, including the following (NRC 1983; Osborne and others 1995):
Hazard: A source of risk, such as a substance or action that can cause harm.
Exposure: Contact with a hazard in such a manner that effective transmission of the agent or harmful effects of the agent may occur.
Dose-response relationship: A relationship in which a change in amount, intensity, or duration of exposure is associated with a change in the risk of the outcome.
Risk: The combination of the likelihood (probability) and magnitude (severity) of an adverse event.
Uncertainty: An instance of limited knowledge, false assumption, or statistical variability that contributes to a statement of confidence in conclusions drawn from a risk assessment.
Risk management: The process of formulating and implementing a course of action to mitigate hazards determined by risk assessment to be important.
THE PROCESS OF RISK ASSESSMENT
The process of risk assessment, as used by US regulatory agencies charged with protecting workers and the general public, involves four sequential steps (NRC 1983; Samet and Burke 1998): hazard identification, dose-response assessment, exposure assessment, and risk estimation and characterization. Multiple sources of data may be used to complete each step, including on-site review and investigation, epidemiologic investigation, surveillance, laboratory animal studies, and computer modeling (see Table 5-1). At the institutional level, risk assessments need not be formal endeavors led by recognized experts, but should focus on the same basic steps with most of the emphasis on hazard identification and exposure assessment. Often this will also reveal likely determinants of exposure to the hazards that should be addressed in the institution's occupational health and safety plan.
Assessment of Risk Associated with Animal-Related Research.
Research involving awake-behaving nonhuman primates requires special consideration during risk assessment. Nonhuman primates can weigh more than human beings, have considerable speed, strength, and manual dexterity, and can harbor zoonotic infectious agents. They can inflict serious physical injury and cause life-threatening illnesses to persons around them. One way to reduce the risks from these animals is by training them to perform certain movements. For example, macaques and squirrel monkeys can be trained to move voluntarily from the home cage into a restraint chair (Ator 1991). Another consideration when working with awake-behaving nonhuman primates is that they are often transported to testing facilities (e.g., laboratories or imaging facilities) outside of the animal quarters. The animals may traverse common use corridors and elevators, potentially exposing individuals not involved in the animal care program. Individuals in other areas of the building may also be exposed if the air exhausted from a testing facility is recycled into other building areas. For these reasons, procedures involving awake-behaving nonhuman primates must undergo additional hazard identification and risk assessment. It is also important that risk assessment of noninfectious hazards involve a qualified health and safety professional with training in the chemical and ergonomic hazards associated with their use. More detailed guidelines for working with awake-behaving NHP are forthcoming from the ILAR Committee on Guidelines for the Care and Use of Mammals in Neuroscience and Behavioral Research (NRC In press). This report will identify common research themes in contemporary neuroscience and behavioral research based on input from neuroscience and behavioral researchers most familiar with current standards of practice and veterinarian specialists in laboratory animal medicine; provide collective, professional judgment in applying current animal care and best use practices to procedures in these areas of research; provide information about new scientific and responsible use developments used to maintain animals during these experiments; and serve as an informational resource to assist researchers, laboratory animal medicine veterinarians and IACUC members in the interpretation and implementation of current standards of practice and promote the training of animal care specialists in this area.
The identification of hazards is typically a qualitative process, most often based on observation, experience, published reports and professional judgment. Hazards in the work environment can be identified by safety specialists using institutional logs, worker-compensation reports, and other information sources (NRC 1997) as well as direct observation of the animal facility. That process should be systematic and based on the principles of biologic, chemical, and physical safety; modes of transmission of infectious agents; understanding of the facility design, equipment, personal protection devices, and practices; and knowledge of applicable local, state, and federal regulations. Chapters 3 and 4 of this report provide an overview of the infectious and noninfectious hazards identified in the use of nonhuman primates in research. If the existence of a hazard cannot be definitively shown in the first step in the risk assessment process, the subsequent steps generally are not warranted.
A review of worker exposure and injury reports suggests that most workplace hazards found in nonhuman-primate research facilities are similar to those found in other laboratory animal research environments (bin Zakaria and others 1996; Poole and others 1998Poole and others 1999). Most common are animal-inflicted trauma from bites, cuts, and scratches; punctures from needle sticks and other sharps; musculoskeletal injuries, such as strains and sprains, especially involving the back; repetitive-motion injuries (ergonomic injuries); slips, trips, and falls; contact with allergens and chemicals; burns caused by contact with hot surfaces or steam; and various suspected exposures to materials that potentially contain infectious agents.
In addition to a review of worker injury and exposure logs, inspection of the facilities will assist in the identification of hazards. Various aspects of the facility design should be evaluated, such as air-exchange rates, air recirculation, and pressure differentials; use of high-pressure hoses, steam, or other cleaning methods; wastewater drainage; composition and uniformity of ceilings, walls, and floors; and laboratory access requirements and controls. Summaries of prescribed safety measures may be obtained from the IACUC and/or the environmental health and safety office responsible for review and approval of facilities. Inspections of the facility should also consider the daily flow of all materials relevant to potential hazards in the institution, such as the primates themselves and their tissues, caging equipment, environmental-enrichment devices and other husbandry items, animal waste, laboratory waste, cage-wash machinery and supplies, research and veterinary supplies and equipment, and facility-maintenance and janitorial equipment and supplies. For example, soiled primate caging may be moved along common use hallways of the animal quarters to the cage wash area, potentially exposing research staff not associated with the primate program. Each situation should be evaluated systematically with the biologic, chemical, and physical hazards associated with nonhuman primates and in conjunction with objective criteria and information sources.
The next step in risk assessment deals with the dose-response relationship. This action establishes the relationship between the quantitative level of a hazard and the probability of an adverse response to it (NRC 1983). Cumulative-dose effects over years of exposure are relevant to chemical toxicants, but single-event exposures are more often of primary concern in the case of pathogenic agents. In both cases, each potential exposure episode is regarded as increasing the cumulative risk, but the likelihood of disease after any single exposure episode is considered to be the same for everyone in the risk group (OSHA 1991). Threshold and non-threshold models of risk exist for various types of health conditions, predominantly with respect to noninfectious hazards, which adds controversy and complication to the risk assessment process (NRC 1983).
Once a hazard has been identified and the dose of the hazard that causes adverse consequences is determined, the next step in risk assessment is exposure assessment. This step estimates the exposure or contact between a hazard and a person (NRC 1983). Exposure assessment must take into account numerous modes of possible contact, such as splashes, bites, aerosols, and needle sticks. The extent to which people are in contact with potential hazards should be determined in conjunction with their job duties and the use of personal protective equipment. Exposure assessment must include evaluation of the experience and skill levels of people who are at risk for exposure. For example, in environments where exposure is associated with failure to comply with standard operating procedures or to use equipment properly (CDC-NIH 1999; NRC 1997), inexperienced personnel would have a greater risk of exposure than more experienced personnel.
Care should be taken when estimating exposure on the basis of injury and exposure logs. Under-reporting of occupational injuries and exposures to supervisors or health-care staff by nonhuman-primate research workers is common. A recent study (bin Zakaria and others 1996) found that 59% of animal-inflicted scratches, 50% of mucous-membrane exposures, and 20% of needle stick injuries went unreported. A variety of reasons were given by respondents in the survey, most commonly that reporting was “too much trouble,” that the injury was “not serious enough to report,” and that “the injury was accepted as a routine risk.” Others have suggested that only 45-68% of injuries are reported to supervisors and that for 4-8% of injuries, workers take no action (Sotir and others 1997). Those observations have important implications for effective communication of the hazards associated with exposure to nonhuman primates and their tissues, especially in light of the finding that occupational infections with B virus have not correlated with injury severity (Hilliard and Henkel 1998).
Risk Estimation and Characterization
The final step in risk assessment is risk estimation and characterization. In this step, the dose-response relationship and exposure assessment are combined to describe the risk to subject persons (NRC 1983). It is essential that persons responsible for conducting risk assessments be knowledgeable about the physical, biologic, and chemical hazards present in nonhuman-primate research, as outlined in Chapters 3 and 4. General principles of safety as they pertain to each hazard should be understood, including essential aspects of the laboratory, husbandry, and veterinary equipment in use; facility design elements, such as the systems for air handling and waste decontamination and disposal; systems of employee hygiene and medical surveillance; and how all these are integrated into the OHSP. Persons responsible for risk assessment must also have an appreciation of the flow of the typical workday activities of animal care, facility maintenance, and research as performed by the different members of the staff (and, if applicable, students and visitors). Knowledge of local, state, and federal regulations under which the facility operates is also important.
As noted throughout this report, the risk assessment process should initially focus on the greatest hazards, those with potential for important consequence for the greatest number of persons. For example, institutions that use macaques or their tissues should first ensure that the possibility of B virus exposure has been assessed. Some of the resulting safety measures will reduce potential injuries from other sources, as in the use of splash barriers, which protect the mucous membranes of the face against infectious agent exposures and chemical exposures during research and husbandry operations.
Once hazard identification is accomplished, other steps in risk assessment are aimed at estimating the risks associated with hazards identified in specific institutional settings. A wide variety of analytic tools are used in these efforts, including qualitative, semiquantitative, and quantitative methods to determine the likelihood of an event in a specified interval and the sources and magnitude of uncertainty and variability in the estimates (Hallenbeck 1993).
Most experts agree that risk assessments should be put into quantitative terms to the greatest extent possible (OSHA 1991). This is especially true when conducting risk assessments for the purposes of establishing regulatory safety standards. Limitations of the data available for use in quantifying the importance of specific hazards contribute to the uncertainty in estimates. Nevertheless, it is desirable in the development of OHSPs to add as much quantitative information to qualitative observations and institutional experience as possible, to arrive at the best possible evaluation of risks posed by specific hazards. It is not realistic to defer the process of risk assessment while waiting for data that may never become available.
Risk to workers is best measured through the use of incidence rate calculations, in which the numerator is the frequency (or number of new occurrences) of an event during a specified period and the denominator is the average size of the group considered at risk for the event:
For example, the incidence of needle stick accidents among veterinary staff involved in the care of nonhuman primates should use as the denominator the size of the veterinary workforce involved in venipuncture tasks in nonhuman primates at the institution. Incidence rates are useful whether the purpose is to compare trends as new safety-related equipment and policies are established in an institution or to compare the experience of different sites or institutions. Standardizing the average size of the group considered at risk for an event by using full time-equivalents, such as person workdays (pwd), in the calculation of incidence rates allows accurate comparisons. However, determination of the approximate size and nature of the group that should be considered at risk requires well-reasoned efforts. Rapid expansion of knowledge regarding the types and sources of hazards demands that this be a continuing commitment.
Understanding of uncertainty in risk assessment is important in conveying the likelihood of an adverse event or the magnitude of its consequences. Reductions in uncertainty do not change the risks, but they increase the mathematical precision of evaluation. Therefore, a clear understanding of the uncertainties included in risk estimates is essential for policy-making, lest misleading information and ineffective action plans result (Hallenbeck 1993). Data for use in quantifying important risk factors are often sparse, and this, combined with differences between information sources and inherent variability, contributes to uncertainty in estimates. By varying the assumptions that are used, a sensitivity analysis can help evaluate the ramifications of variables in risk estimates and yield better predictions of the impact of various management options.
Risk of Infectious Hazards Associated with Nonhuman Primates
Risk assessment of infectious hazards is particularly important in nonhuman-primate research facilities. As described in Chapter 3, nonhuman primates can harbor zoonotic agents—such as B virus, Mycobacterium tuberculosis, SIV, and enteric pathogens—some of which have dire consequences.
In performing a qualitative risk assessment, all risk factors are first identified and explored. For infectious hazards, the risk of becoming infected depends upon the likelihood of a relevant exposure to a source of infection and the likelihood of becoming infected if there is an exposure. The following elements may be evaluated when assessing the risk of infectious hazards in the conduct of laboratory animal work: animal contact, exposure intensity, exposure frequency, physical and biologic hazards present by the animals, hazardous properties of agents used in research protocols, susceptibility of employees, and occupational-health history of employees doing similar work (NRC 1997). Exposure intensity measures the estimated dose received among those exposed over some arbitrarily defined unit of time, whereas exposure frequency concerns the number of opportunities for any degree of exposure during the same period. For zoonotic diseases, both of there parameters are affected by the prevalence of the agent in the animals, its shedding pattern, environmental stability, and routes of transmission to humans. Exposure intensity values are often used when setting allowed safety standards to chemical or allergen hazards and are generally more applicable in those cases. Consideration of the importance of individual hazards identified in the workplace should include the size of the group at risk, the potential effects of the hazards, and the magnitude of the exposures (NRC 1997). Ranking of hazards based on their importance can include institutional experience regarding worker illness and injury rates, near-miss reports, reference information, and other documents (see Table 5-1).
As stated in the federal guidelines outlined in Biosafety in Microbiological and Biomedical Laboratories (CDC-NIH 1999), the primary role of risk assessment is to aid in the prevention of workplace-acquired infections, and the secondary role is to aid in the prevention of infections in the surrounding community (CDC-NIH 1999). In this context, risk assessment has led to the assignment of designated animal biosafety levels (ABSLs) 1-4 for experimental research activities with specified pathogens. ABSLs are described in terms of facilities, equipment, and practices, each being important in mitigating hazards or risks to workers and the public (see Table 5-2). This approach has a laudable record of contributing to overall workplace safety, despite the difficulty in assessing some variables, such as emerging infectious agents and genomic manipulations. A conservative approach is generally recommended when a lack of information forces subjective decision-making. For example, when infectious-disease risks are being considered, universal precautions are always advisable (CDC-NIH 1999).
Summary of Recommended Animal Biosafety Levels (ABSLs) for Activities in Which Experimentally or Naturally Infected Vertebrate Animals Are Used.
However, these federal guidelines do not address noninfectious laboratory hazards, nor were they intended to guide safety considerations outside the laboratory per se, such as in outdoor holding enclosures for nonhuman primates or in settings involving their wild capture and transportation. Thus, OHSPs for persons working with nonhuman primates in many situations must be based on general workplace safety considerations and analogous hazards in other industries without the benefit of the specific algorithms found in the national laboratory biosafety guidelines.
Important characteristics of most well-known infectious agents that are useful in risk assessment are readily found in Chapter 3 and the references cited at the end of this report. Such information is usually based on medical surveillance and epidemiologic studies and grounded in laboratory investigations of the agents themselves. Many agents known to have caused laboratory-acquired infections are listed in Section VII of Biosafety in Microbiological and Biomedical Laboratories (CDC-NIH 1999). The following characteristics predict risk, but the characteristics considered as a whole are more important than any one of them individually.
Stability in the environment
Concentration in specimens or in the environment
Origin (host, geographic location, or type of source)
Route of transmission
Availability of data from animal studies
Availability of effective prophylaxis or therapy
Experience and skill of personnel at risk
Selection of the appropriate ABSL for activities involving infectious material in nonhuman-primate research should be based on evaluation of these criteria, with modifications as needed in light of current scientific information. Answers to questions about the characteristics listed above often are not definitive, especially for newly described infectious agents and materials that contain recombinant DNA; in such cases, the risk-assessment process should include an institutional biosafety committee (NIH 1998).
Risk assessment leading to the requirement for an ABSL-2, 3, or 4 presupposes that the workforce is composed of immunocompetent people. Immunocompromised people are at increased risk in many cases when exposed to infectious agents. That is just one of the complexities that can enter into risk assessment, so other variables, as described in Section VII of Biosafety in Microbiological and Biomedical Laboratories (CDC-NIH 1999), must also be considered in evaluating risks to particular workers. In all research involving the use of nonhuman primates, the study director or principal investigator must work with the IACUC, the biosafety officer, and the primate center director to assess risks and set ABSLs in the context of the institutional administrative structure; ultimate authority rests with the senior institutional official (CDC-NIH 1999).
RISK OF OCCUPATIONAL INJURIES AND EXPOSURES AT NATIONAL PRIMATE RESEARCH CENTERS
Physical Hazard Risk Assessment
An epidemiologic investigation of work-related injuries and exposures among animal care, veterinary, and scientific staff at a US regional primate research center provided yearly estimates of incidence rates ranging from 44 to 65 animal-associated injuries per 100,000 pwd during the 5-year period of observation (bin Zakaria and others 1996). Animal-inflicted scratches and bites had the highest 5-year incidence rates (82.1 and 80.8 incidents per 100,000 pwd, respectively), together accounting for 51.7% of reported incidents. Cuts and mucous membrane exposures had the lowest 5-year incidence (45.0 and 17.6 incidents per 100,000 pwd, respectively). Fingers and thumbs were the most common anatomic sites of occupational bite injuries, and full-time workers were 3-4 times more likely to report injury episodes than part-time workers (those with less than 20 hours of animal contact per week). The injury-specific incidence rates differed with job category; veterinary residents in training had the highest overall injury rates. The frequency of all injury types decreased with increasing years of employment, and 33% of all reported injuries occurred in persons hired less than 6 months previously. Those findings have implications for risk assessment at individual institutions, where a workforce composed of many inexperienced people should be assessed at greater risk than institutions with an experienced workforce. The findings also offer an opportunity to compare some work-related injuries incurred during nonhuman-primate handling (such as needle sticks and mucous-membrane exposures) with injuries incurred by persons employed in human hospitals and other health-care settings, which could lead to improvements in safety-training and injury-prevention programs.
A quantitative risk-assessment study of primate-associated injuries and exposures at another US national primate research center has been reported (Weigler and Ponce 1999); it was based on analysis of institutional bite-scratch-splash exposure records over a 5-year period. In this case, a stochastic (random) simulation model was done to estimate the efficiencies of different B virus exposure prevention methods. Simulations were done using 2000 iterations with median latin hypercube sampling (Analytica Software, Decisioneering, Denver, CO) assuming that sources of risk were independent and that there was no threshold for exposure. The probability of B virus exposure among workers was estimated by including separate distributions for the prevalence of B virus among macaques in this setting (modeled as a normal distribution with a mean of 0.5 and standard deviation of 0.1) and the likelihood of shedding among infected animals (modeled as a triangular distribution with a mean of 0.02 and range 0 to 0.05). Each category of prevention method (protective eyewear, gloves, mask, and labcoat, laboratory procedures, and postexposure scrub) was included as a separate model for each type of exposure (bite/scratch, needle stick, cut, mucous membrane, and other), using two different statistical distributions (triangular and beta) for exploration. Empirical reasoning was used for parameter estimates of those distributions; for example, protective eyewear was given a mean of 0.75 and range 0 to 1 for protecting mucous membranes but no protection against bites, scratches, or other types of exposure. The actual institutional injury exposure record data were annualized for 8-hour person workdays at risk and stratified by type of worker (veterinary, husbandry, research, maintenance, student, other).The result of these simulation models was the expected incidence of B virus infections for the at-risk population of workers in the institution. That approach led to the prediction of one new human B virus exposure episode per 60 years in the institution, assuming a fixed population size of workers at risk. The study also included a sensitivity analysis of model predictions of the potential impact of different B virus risk-management strategies. Use of PPE that reduced scratch rates, improved laboratory procedures, and increased postexposure wound disinfecting efficiency was most influential in reducing the risk of B virus among workers in these models.
Simian Immunodeficiency Virus (SIV)
In a study involving a questionnaire-based survey of 550 persons working at 13 North American research institutions (Sotir and others 1997), a high frequency of needle sticks and mucocutaneous exposures (defined as animal-inflicted bites and scratches) was documented among persons working with nonhuman primates and their tissues. Over one-third of study participants were reported to have experienced needle sticks or mucocutaneous exposures while working with nonhuman primates, predominantly macaques but including at least six other genera. The study included serial serologic testing for SIV antibodies among study participants and considered whether there was exposure to SIV in the laboratory or to SIV-infected animals. Statistical methods were used to assess possible associations between workers' job categories, job tasks performed, length of employment, work with HIV-2 and SIV, work with nonhuman primates, and the frequency and types of injuries sustained in the workplace.
Persons working with monkeys that were SIV-negative or whose SIV status was unknown were more likely to have sustained (or to have reported) needle sticks or mucocutaneous exposure involving blood, body fluid, or unfixed tissue than were those working with SIV-infected animals. Those study results suggested that increased awareness led to improved safety practices or alternatively to different reporting rates. Some injury-specific frequencies differed with job category, but in contrast with the previously discussed study (bin Zakaria and others 1996), increasing years of employment increased the likelihood of injury occurrence.
Analysis of survey responses showed that persons responsible for more invasive tasks with animals (such as phlebotomy, dental work, surgery, necropsy, and experimental inoculation) were at greater risk for needle stick injuries than persons doing noninvasive work (such as husbandry, sanitation, and routine medication) even when years of experience were taken into account. However, bite and scratch rates did not differ with task type. Those findings parallel observations of health-care workers at risk for bloodborne pathogen exposure in occupational settings and highlight opportunities for focused preventive educational programs, especially for some occupational groups (such as husbandry staff) that might be less informed about work-related hazards (OSHA 1999).
Ranking or prioritizing hazards is one way to help determine which risk is the most serious and thus which to control first. Priority is usually established by taking into account the employee exposure and the potential for incident, injury or illness. By assigning a priority to the risks, you are creating a ranking or an action list.
There is no one simple or single way to determine the level of risk. Nor will a single technique apply in all situations. The organization has to determine which technique will work best for each situation. Ranking hazards requires the knowledge of the workplace activities, urgency of situations, and most importantly, objective judgement.
For simple or less complex situations, an assessment can literally be a discussion or brainstorming session based on knowledge and experience. In some cases, checklists or a probability matrix can be helpful. For more complex situations, a team of knowledgeable personnel who are familiar with the work is usually necessary.
As an example, consider this simple risk matrix. Table 1 shows the relationship between probability and severity.
Severity ratings in this example represent:
- High: major fracture, poisoning, significant loss of blood, serious head injury, or fatal disease
- Medium: sprain, strain, localized burn, dermatitis, asthma, injury requiring days off work
- Low: an injury that requires first aid only; short-term pain, irritation, or dizziness
Probability ratings in this example represent:
- High: likely to be experienced once or twice a year by an individual
- Medium: may be experienced once every five years by an individual
- Low: may occur once during a working lifetime
The cells in Table 1 correspond to a risk level, as shown in Table 2.
These risk ratings correspond to recommended actions such as:
- Immediately dangerous: stop the process and implement controls
- High risk: investigate the process and implement controls immediately
- Medium risk: keep the process going; however, a control plan must be developed and should be implemented as soon as possible
- Low risk: keep the process going, but monitor regularly. A control plan should also be investigated
- Very low risk: keep monitoring the process
Let’s use an example: When painting a room, a step stool must be used to reach higher areas. The individual will not be standing higher than 1 metre (3 feet) at any time. The assessment team reviewed the situation and agrees that working from a step stool at 1 m is likely to:
- Cause a short-term injury such as a strain or sprain if the individual falls. A severe sprain may require days off work. This outcome is similar to a medium severity rating.
- Occur once in a working lifetime as painting is an uncommon activity for this organization. This criterion is similar to a low probability rating.
When compared to the risk matrix chart (Table 1), these values correspond to a low risk.
The workplace decides to implement risk control measures, including the use of a stool with a large top that will allow the individual to maintain stability when standing on the stool. They also determined that while the floor surface is flat, they provided training to the individual on the importance of making sure the stool’s legs always rest on the flat surface. The training also included steps to avoid excess reaching while painting.