
Improving the Human–Technology Interface | 235
appropriately to the alerts, modifying or cancelling 35% of the orders that trig-
gered them, designers had not found the right balance of helpful-to-interruptive
alerts. The system simply did not fit the clinicians’ workflow.
Such unintended consequences (Ash, Berg, & Coiera, 2004) or unpredictable
outcomes (Aarts, Doorewaard, & Berg, 2004) of healthcare information systems
may be attributed, in part, to a flawed implementation process; but there were
clearly also human–technology interaction issues. That is, the technology was
not well matched to the users and the context of care. In the pediatric case, ...