>http://www.computerworld.com/s/article/print/9141428/Harvard_study_Computers_don_t_save_hospitals_money?taxonomyName=Hardware&taxonomyId=12
>
>http://www.amjmed.com/webfiles/images/journals/ajm/AJM10662S200.pdf
>
>Choice quote: "Programmers of the successful systems told Himmelstein that they didn't write manuals or offer training. "If you need a manual, then the system doesn't work. If you need training, the system doesn't work," he said."

The guy is right. The system we wrote in 1994/5 (FPD2.6, Novell network) was intended for doctors (or nurses, whoever was in charge of datakeeping in a particular office), and was used as such. Any reporting that had to go to accounting, statistics and whatnots was derived from that.

Another system we developed for the system of local practices reduced their 17 monthly reports to one weekly report, basically a log. The totalling of it by various groupings and regroupings (sometimes on nested expressions and... awful, we almost wrote a language to deal with that) produced those 17. This saved between 10 and 15 hours a month for every doctor (and nurses would often have to help).

OTOH, we never got any other major hospital, because the money was too big, and nobody trusted little companies with solutions priced with a zero less. Whenever we said we're doing a hospital system and we got the polyclinic part done, someone would come up with a story of a university team somewhere, who spent a couple of $e6 and have already managed, in just a few years, to solve the laundry and were almost done with the kitchen.