Data processing in hospital bacteriology: experience of 18 months' trial.

The nature of bacteriological investigations creates data processing problems. Non-numerical results must be coded. This may be done retrospectively or at the bench, the former method involving extra work and staff, the latter implying simple codes, stringent checking, a nearby computer, and decoding programs. The use of a shared computer for producing daily reports diminishes departmental autonomy. Benefits occur in more readily available records and statistics, quality control, epidemiology, and diagnostic bacteriology.