Sök artiklar i SveMed+

Tidskrift

Sammanfattning

The Swedish health care system needs to make use of clinical data stored in the many different electronic record systems in use. Each has their own interface terminology. However, in order for the systems to communicate they must be able to translate/map information into a common terminology such as SNOMED CT. We have studied 5532 medical records from patients with diabetes stored in the Journalia patient record system, and mapped them into pre-coordinated SNOMED CT terms. The mapping was successful with 94% of nephropathy terms, 72% of neuropathy- and 63% of social terms. When the computer had generated ICD 10 codes we found that physicians had manually diagnosed 67% of the patients with retinopathy while the computer found 82%. Similarly, physicians coded only one third of the patients with microalbuminuria that the system had coded for this diagnosis. We conclude that electronic patient record systems using their own finely granulated interface terminologies can map into SNOMED CT or ICD10. Thus, information can be shared between similarly advanced systems and then be used to facilitate patient transfers, quality controls, and clinical research. In the next step such trials should be made for other diagnoses in order to gain experience and locate communication bottlenecks. What should be mapped in the end depends on the end-users' specifications. Not everything needs to be mapped.

MeSH