Data are presented on the use of a browsing and encoding utility to im
prove coded data entry for an electronic patient record system. Tradit
ional and computerized discharge summaries were compared: during three
phases of coding ICD-9 diagnoses phase I, no coding; phase II, manual
coding, and phase III, computerized semiautomatic coding. Our data in
dicate that (1) only 50% of all diagnoses in a discharge summary are e
ncoded manually; (2) using a computerized browsing and encoding utilit
y this percentage may increase by 64%; (3) when forced to encode manua
lly, users may ''shift'' as much as 84% of relevant diagnoses from the
appropriate coding section to other sections thereby ''bypassing'' th
e need to encode, this was reduced by up to 41% with the computerized
approach, and (4) computerized encoding can improve completeness of da
ta encoding, from 46 to 100%. We conclude that the use of a computeriz
ed browsing and encoding tool can increase data quality and the percen
tage of documented data. Mechanisms bypassing the need to code can be
avoided.