A hospital's systems are tested twice: on the night they fail, and on the morning a patient hands over a card the reception desk has never seen.
In the autumn of 1996 I spent three nights on a surgical ward in a hospital where my employer had just installed computerised order entry. I was twenty-five and responsible for the link between the ward screens and the pharmacy. At two in the morning a nurse with eleven patients and one colleague asked me why changing a dose took four screens. I had no good answer. She wrote the change on paper, telephoned the pharmacy and entered it at the end of her shift. The system had worked as specified. The ward had worked around it.
A system that a ward cannot operate at three in the morning, two nurses short, is not finished, however well it performed in the supplier's demonstration. I would rather spend money on the hours when the record is missing or unfamiliar than on another feature for the hours when everything works. The clinical teams decide how patients are cared for; my job is to make sure the tools still serve them when conditions are at their worst.