RT Journal Article
SR Electronic
T1 Continuous measures of driving performance on an advanced office-based driving simulator can be used to predict simulator task failure in patients with obstructive sleep apnoea syndrome
JF Thorax
JO Thorax
FD BMJ Publishing Group Ltd and British Thoracic Society
SP 815
OP 821
DO 10.1136/thoraxjnl-2011-200699
VO 67
IS 9
A1 Ghosh, Dipansu
A1 Jamson, Samantha L
A1 Baxter, Paul D
A1 Elliott, Mark W
YR 2012
UL http://thorax.bmj.com/content/67/9/815.abstract
AB Introduction Some patients with obstructive sleep apnoea syndrome are at higher risk of being involved in road traffic accidents. It has not been possible to identify this group from clinical and polysomnographic information or using simple simulators. We explore the possibility of identifying this group from variables generated in an advanced PC-based driving simulator.Methods All patients performed a 90 km motorway driving simulation. Two events were programmed to trigger evasive actions, one subtle and an alert driver should not crash, while for the other, even a fully alert driver might crash. Simulator parameters including standard deviation of lane position (SDLP) and reaction times at the veer event (VeerRT) were recorded. There were three possible outcomes: ‘fail’, ‘indeterminate’ and ‘pass’. An exploratory study identified the simulator parameters predicting a ‘fail’ by regression analysis and this was then validated prospectively.Results 72 patients were included in the exploratory phase and 133 patients in the validation phase. 65 (32%) patients completed the run without any incidents, 45 (22%) failed, 95 (46%) were indeterminate. Prediction models using SDLP and VeerRT could predict ‘fails’ with a sensitivity of 82% and specificity of 96%. The models were subsequently confirmed in the validation phase.Conclusions Using continuously measured variables it has been possible to identify, with a high degree of accuracy, a subset of patients with obstructive sleep apnoea syndrome who fail a simulated driving test. This has the potential to identify at-risk drivers and improve the reliability of a clinician's decision-making.