Effect of discharge letter-linked diagnosis registration on data quality

Diagnostic data are essential for the assessment of medical practice: they are needed for retrieval of clinical cases and describing co-morbidity and complications. In most Western countries, diagnosis registration in hospital information systems is based mainly on completing forms after patient discharge. As this registration plays no role in patient care, data quality is usually unsatisfactory. To improve data quality, we redesigned the process of diagnosis registration at a paediatric department, and now paediatricians provide diagnoses with codes in a separate registration heading of the discharge letter. We compared the quality of this discharge letter-linked diagnosis registration with the quality of the previous form based registration.

Retrospective study with blinded before and after measurement. Re-abstracted diagnosis descriptions of the text of discharge letters were taken as gold standard.

A paediatric department in an academic medical centre.

From each registration period, 60 admissions were selected randomly. Mean age of the patients was 4.5 (SD +/- 5.5) and 5.2 (SD +/- 5.2) years for the old and new situation respectively. Mean length of stay was 8.8 (SD +/- 11.0) and 7.2 (SD +/- 12.4) days.

Discharge letter-linked diagnosis registration.

Completeness and accuracy, both at three-digit level of ICD-9-CM.

Completeness of form-based diagnosis registration was 51% (95% CI, 44-58%) and of discharge letter-linked diagnosis registration 54% (95% CI, 47-60%). Accuracy was 65% (95% CI, 58-72%) and 67% (95% CI, 60-74%) respectively.

The discharge letter-linked diagnosis registration does not provide a better basis for assessment of medical practice than the form based diagnosis registration.