There are several studies assessing the practical use of mobile devices and handheld computers in the medical education setting, but what is lacking to date is the underpinning pedagogic basis for the use of such technology [6,7]. If limited resources are to be directed towards the provision of technology, it needs to be justifiable from a theoretical perspective and regarding outcomes of such an intervention. Research in e-learning to date has concentrated on demonstrating efficacy and comparing new technology with old [8]. What is needed to inform future use is further research directed towards 'when' and 'how' to use e-learning. Although this may be context specific, educators elsewhere can judge the transferability and relevance for their own institution.

Our model demonstrates how contextual factors can impact upon the learning process. Resources on the PDA were seen as a useful additional tool for them to have- a supplement rather than a replacement for their traditional learning strategies. This aligns with the theory of a learning ecology, an environment composed of a diverse variety of learning options, allowing each individual student to find the opportunity to access learning that addresses their own personal and immediate needs [20]. Appropriation in which users not only adjust the tool to best fit their activities, but the tool may also cause the user to change their behaviour to accommodate using the tool effectively, was also seen [21]. Home use of PDAs should also be seen as valid and part of the learning environment in mobile learning, with users able to access knowledge in different ways (PDA, book, personal computer) depending on the type and complexity of knowledge required. This finding also fits with the increasing popularity of tablet computers to access resources and the internet that are also available on home personal computers.

Both intrinsic mechanisms and external conditions may influence the learning possibilities available to a student [22]. As described in activity theory, the social context is important, with "social rules and conventions govern[ing] what is acceptable" and attitudes influenced by surrounding opinion [21]. Linked to this is the concept of acceptability of technology. The TAM model (Technology Acceptance Model) [23] suggested that usage of a system depended on several factors including perceived utility and ease of use that would determine the attitude towards technology and intention to use it. Holzinger [24] also suggested that previous exposure to technology and unobtrusiveness were also important factors that would affect acceptance and use of a technology. All respondents to our surveys had a personal computer and 38% already had another mobile device, suggesting that previous exposure was not a barrier to acceptance. The physical obtrusiveness of the PDA reflects the experiences of our students with respect to their concerns regarding teacher and patient opinion, and of a perceived failure to engage in the clinical moment, leading to potential inhibition of PDA use despite accepting its utility. This constrained their optimal use of the PDA. Alsos found that use of the device consumed the physician's attention, with poor action-transparency and inhibited patients from asking questions and raising issues [25]. Houston et al in contrast found that the majority of their patients had positive attitudes towards handheld computers, although the clinicians themselves had reservations [26].

With sufficient support these barriers can be overcome, but a need for change was paramount, from the student's own attitude and behaviours to that of the clinicians and patients. This was achieved through communication by the students of the authentic learning nature of the tool, and by encouragement by the clinicians. Guidance on etiquette was required, so-called "mobiquette" [5]. Institutional support helped both with encouragement and technical support, the importance of which has been highlighted by others [7,27]. Becoming a more formalised part of the curriculum may aid this.

There are some important learning theories that we were unable to demonstrate directly, but this does not mean that they do not contribute. Rather it is due to limitations of the research tools available. Experiential learning with reflection [28] is likely to have a key role as shown by the broken arrows in Figure ​Figure1,1, and further research to elicit this would be invaluable. We were also unable to demonstrate that learning that occurred affected the students' actual practice. Extending the methods of data capture along ethnographical lines may allow exploration of these but this is difficult given the very nature of mobile learning which is spread across many contexts and over long periods of time [29]. It makes it difficult to predict when a learning opportunity is likely to occur. Options include voice recordings or blogging, which may be more contemporaneous, or structured diaries have been used successfully [30]. An ethnographic approach would also allow assessment of the effect on student-teacher and student-patient interactions that a move to a smartphone platform may have, and to find how the students and patients find ways to manage this.

The strengths of this study are that a large cohort of UK medical students were engaged on whom both qualitative and quantitative data were collected on use of PDAs as learning tools in a clinical setting in the UK. Provision of the PDA and software enabled us to involve students beyond a small self selecting group keen on technology,so reducing bias.

There are limitations to the study. The surveys and focus groups were dependent on personal reporting and so subject to recall bias, and the quantitative tracking data was too limited to offset this. The focus groups may have had an inherent bias in that those who chose to participate had engaged more with the technology. Purposive sampling could have been used instead to reduce this. The survey response rates were lower than ideal leading to non-response bias, potentially limiting the generalisability of results. However triangulation of qualitative and quantitative results showed consistency and suggest the findings are valid. Information from the final year group would have been useful but were not surveyed as they were sitting their final examinations prior to graduation. The usage tracking developed for this study was limited in that it only showed the number of times a resource was accessed and relied on students synchronizing their PDA with a personal computer, which not all students did despite regular requests from the project team. Ideally, time of day, location and duration of resource use would have been useful data to collect, with automatic uploading of usage data through a wireless network, and this has now been developed for use in DrCompanion for smartphones. The project was pragmatic in that it gave the same device and resources to all students.