A new paper-based hospital observation chart has been developed using human factors design principles. This novel design, compared to previous charts, yielded fewer errors and faster response times in chart-users' detection of patient physiological deterioration compared to other Australian observation charts that were in use at the time. In recent clinical studies, the chart has also been associated with an 11% mortality reduction amongst intensive care unit admissions, as well as a 45% reduction in the incidence of cardiac arrests. However, there are a number of points of contention as to whether this design can be regarded as best practice. First, it is unclear whether the chart offers performance benefits to users highly experienced with alternative chart designs. Second, clinicians have questioned particular features that were designed to help users detect abnormal vital sign observations. For example, there is a dispute as to whether blood pressure and heart rate graphs should be presented as separate or overlapping plots. Third, disagreement surrounds the optimal design layout to facilitate users' calculation of summary scores that represent the physiological state of a patient. In the absence of expert consensus, this thesis sought to address each of these points of contention using behavioural experiments. In general, findings supported the design choices associated with the new observation chart. Specifically, in relation to the detection of abnormal observations, it was found that (1) even users experienced with alternative chart designs performed better with the new chart; (2) blood pressure and heart rate were better presented as separate graphs (even for chart-users who preferred plots that overlap); and (3) users' performance with drawn-dot observations, an integrated colour-based scoring-system, and grouped scoring-rows was consistent with apriori human factors design principles. One design aspect of the new chart was not supported: users were found to be less accurate calculating patient deterioration summary scores when the design involved recording interim steps in the calculation. Overall, it is argued that these experiments demonstrate the value of using behavioural experiments to assess best design practice, rather than relying solely on expert opinion. 2