Abstract
In the outpatient reception counter of the University of Tokyo Hospital, prescription data have been inputted by clerks at computer terminals for the medical accounting. The data will be useful for a survey of prescribed drugs, such as a drug history. Therefore, it is necessary to investigate the rate of errors on the inputted data. The inputted data were checked with the original prescribed sheets by pharmacists every day during 3 months. The detected errors were completely corrected by the clerks being informed by pharmacists. A total of 35192 prescribed sheets were checked for 3 months, and in 2053 sheets 2245 input errors were detected. In an intermittent preliminary survey before the check study, input errors were found in about 10% of prescribed sheets per day. At the beginning of the check study, the rate of input errors was abut 8%, and by the every-day check throughout 3 months, it decreased gradually, and finally became 5%. The corrected data mentioned above were checked again by pharmacists. Then, pharmacists found extra errors, which had been missed at the previous check, in about 1% of the data.
