[Automated analysis of the quality of the minimum set of basic data. Implications for risk-adjusting systems]

Together with the age of the patient, the main diagnosis, secondary diagnosis (comorbility and complications) and the procedures performed are the critical variables for risk-adjusting. Therefore, its correct incorporation to CMBD is of great importance. However, several studies, especially in the United States, but also in Spain, have made evident the existence of important problems of quality in these data, difficulties for its improvement and the limitations which this has to assess the quality or the efficiency of hospitals. The objective of this study is to approach the quality of administrative and clinical collected in the CMBD of the Valencian Health Service (VHS) using an automatized process of analysis of data from the same CMBD, and discuss the implications for its management, as well as possible improvement strategies.

An automatized analysis of the quality of CMBD 1994 of the VHS (20 hospitals, 241,341 admissions) was performed, using indicators of valid fulfilling of field values, relationship between fields of the same episode, relationship between variables in different episodes and volume and specificity of clinical information.

The analysed CMBD contains few errors in management variables, with the exception of residence, but it shows important problems of volume and specificity of clinical information, as well as a high variability in its fulfilling and quality in different hospitals.

The quality of the clinical data of CMBD may be biased in its use with management aims or when assessing quality, as well as in epidemiological studies, evaluation of technology or use of services.