[Use and evaluation of the Belgian 3BT thesaurus adapted for Rwanda]

The Kigali university medical centre (CHUK) lacks visibility on its activities for many years. Institution management as well understanding the institution as the corner stone of the Rwandese health system and reinforcement of its visibility are all health policy major issues. Nevertheless, to this point no such tool has been developed nor implemented. The objective is to assess the feasibility of the use of the thesaurus 3BT as data collection tool in a Rwandese health institution. In august 2005, the thesaurus 3BT (Belgian, bilingual, biclassified ICD-10/ICPC-2) adapted to the CHUK has been implemented. Main issues: encoding quality and thesaurus operationality. Qualitative analysis has been performed on 899 coding of which 16 empty. Low occurrences (< or = 0.2%) of about 25% of codings of clinical diagnosis show the need for using a thesaurus but also to upgrade it. Near 45% of the empty codings could be fulfilled with a quick look to the original medical record. Some diagnoses are missing in the thesaurus 3BT. 66% of which have similar concepts in the thesaurus although not identifiable by a lay person. Finally, a clinical data blind coding test by a doctor used to classifications and by a lay person used to code clinical diagnosis from medical records of the hospitalised patients shows an exact similarity in 70% of the coding and a loss of coding precision in 20%. No coding error has been identified at this time. In conclusion, operationality of the thesaurus is quite acceptable in this study. The thesaurus makes easy the coding of clinical diagnosis even by lay people. Quality of data is enough to be able to interpret the quantitative results of the coding process. This study has to be repeated on a wider basis.