This report concerns pilot testing of a computer-assisted training course which was intended to build resilience to stress in healthcare workers facing an influenza pandemic. In particular, this study aimed to determine the optimal "dose" of such training by comparing three versions of the course (differing in duration) on pre-post changes in variables that were considered proximal determinants of resilience.
Well before the onset of the H1N1 influenza pandemic in 2009, pandemic preparedness plans recognized that psychological support of healthcare workers would be necessary during an influenza pandemic [1–4]. Observations made during and after the 2003 outbreak of severe acute respiratory syndrome (SARS) suggested that an emerging infectious disease causes stress in healthcare settings because of fear of contagion [5–7], concern for family health [6, 8, 9], job stress [6, 7], interpersonal isolation [6, 7], quarantine  and perceived stigma [7, 10, 11].
Moving beyond plans to support affected healthcare workers after exposure to a pandemic, evidence suggests that psychosocial support and training should be provided to healthcare workers prior to a pandemic to build resilience , thereby reducing the impact of stress after exposure. This may be important because predictive models often assume high absenteeism due to illness in healthcare workers and their families, and the stress of dealing with an extraordinary outbreak of infectious disease may further increase absenteeism . Stress-related absenteeism, which may be avoidable with intervention, provides a "second-hit" on a healthcare system which is compromised by other unavoidable losses of human resources. The importance of discretionary absenteeism was highlighted by a report that up to 53% of healthcare workers would refuse to attend work if multiple patients infected with pandemic influenza were admitted to their hospital .
We designed an educational intervention to improve resilience to pandemic-related stress. Resilience has been defined as overcoming stress or adversity or, more precisely, as having a good outcome after an adverse experience . The resilience literature is large and varied , ranging from studies of individual differences between children that promote healthy outcomes after early life adversity  to the factors that protect individuals exposed to extreme traumatic stressors [17, 18] or protect adults from mental illness . Nurses' resilience to workplace stresses may be bolstered by processes that enhance the quality of professional relationships and that foster the development of emotional insight, life balance, spirituality, and reflective thinking . Our intention was to design and implement a training intervention which healthcare workers could take prior to an influenza pandemic that would reduce the adverse stress-related effects of a subsequent pandemic exposure.
We first considered the educational goals. Outcomes of pandemic-related resilience, such as reduced event-related absenteeism and psychological stress, cannot be measured until exposure to a pandemic occurs. For this pilot we focused instead on proximal outcomes which could facilitate resilience. Studies of the individual and health-system variables which mediated the outcomes of stress related to SARS suggested that the best goals of training would be to (i) increase confidence in being well-supported by the hospital and well-prepared for the pandemic and (ii) enhance adaptive strategies of coping (increasing problem solving and seeking support and decreasing escape-avoidance) . Since the purpose of effective training is to change the behavior of healthcare workers in a pandemic situation, we also proposed, based on the theory of social learning , that enhancing self-efficacy (expectations about personal ability to respond adaptively to pandemic-related stresses) should be a goal of training because it is expected to be a proximal predictor of behavior . Finally, since many stressful aspects of an infectious disease outbreak are directly or indirectly of an interpersonal nature (e.g. concern for family health, interpersonal job stresses, interpersonal isolation and perceived stigma) and because interpersonal problems are associated with job stress in healthcare workers , reducing interpersonal problems was a goal. These pandemic related stressors could be expected to amplify the stress and strain which is commonly experienced by health care workers in the contemporary health care environment .
We next considered the optimal format of training. Since hundreds of thousands of healthcare workers around the world are affected by an influenza pandemic, resilience training should be available at reasonable cost on a very large scale. This is a challenge. Those modes of continuing professional education that demonstrate strong effectiveness regarding behavioral change are resource-intensive interventions designed for the training of individuals and small groups . On the other hand, computer-assisted and Internet-based learning technologies have been promoted as effective strategies to facilitate learning, which may overcome barriers related to timing and cost . Computer-assisted learning also has the advantages of standardization of the course material for all learners and enhanced opportunities for personalization of timing, pace and opportunities for review. It can be provided to large and widely distributed groups of people and, beyond the costs of developing the course, the incremental cost per learner may be modest.
Despite a growing body of literature on computer-assisted and Internet-based continuing professional education for health professionals, most studies are methodologically weak [27, 28]. Computer-assisted learning has been shown to be a feasible option for educating professionals when compared with printed and/or lecture based continuing education . Compared to no intervention, Internet-based training has a large effect on learner satisfaction, acquisition of knowledge, skills and behavioral change . Compared to other learning modalities the effects on these outcomes have been small, non-significant and inconsistent [28, 30–32]. Thus, there is no strong evidence base for the superiority of any particular modality. The application of computer-assisted learning to the goals and context of pandemic preparation is untested. Thus one purpose of our research study was to determine if the goals of pandemic resilience training could be accomplished with a computer-assisted educational course that could be widely distributed and self-administered.
Finally, we considered the content of the course and the types of learning experience that might achieve the identified goals. This consideration was guided by experience counseling healthcare workers during and after SARS , and by knowledge from other learning resources. We assumed that information about influenza and about stress and coping should be provided but would not, in itself, achieve the goals of training. Based on research in continuing professional education, we expected that the format of the course should address both objective and perceived learning needs, and include both cognitive interactivity and practicing of skills [25, 33]. The effect of these modes of training on educational outcomes of Internet-based training has been inconsistent, although practice of prescribed exercises more consistently contributes to skills acquisition . Based on the experience of counseling workers during and after exposure to SARS, we expected that exercises should be provided that not only involve cognitive interactivity but also exposure to affectively charged interpersonal events. The goal of these exercises is to enhance reflective, as opposed to immediately reactive responses to acute interpersonal stressors. The principle of increasing reflection had been valued by our colleagues during facilitated group discussion of SARS-related stressors . We also drew upon previous work on psychological first aid [35, 36], stress management , and coping .
The resulting course was named the Pandemic Influenza Stress Vaccine. We hypothesized that the course would lead to improvements in satisfaction with support and training, coping, pandemic-related self-efficacy and interpersonal problems. It was not known at the time of developing the course what the ideal "dose" of this intervention would be, in terms of the time spent on the course by learners and the comprehensiveness of its curriculum. Previous studies of Internet-based learning have shown that longer duration is positively associated with behavioural outcomes . We expected that a course that was too brief might not achieve the teaching goals while a course that was too long might be unnecessarily burdensome and lead to dropping out. The purpose of this study was to test three versions of the course (short, medium and long) with respect to improvements at the end of the course in the hypothesized outcomes and drop-out rates.