Abstract

MARQUEZ CID, Mirari et al. Validation of colorectal cancer diagnostic codes in a hospital administration data set. Gac Sanit [online]. 2006, vol.20, n.4, pp.266-272. ISSN 0213-9111.

Objectives: To validate the ability of a hospital administration data set (minimum data set [MDS]) to detect incident cases of colorectal cancer using the Murcia Cancer Registry (MCR) as the gold standard and to measure agreement between the MDS and registration of colorectal cancer. Material and method: A cross sectional validation study of the MDS of the main hospital in the region of Murcia (Spain) was conducted. The study population consisted of incident cases of colorectal cancer in 2000 obtained from the MCR and cases in the MDS of the above-mentioned hospital for the same year with an ICD-9 diagnostic code between 153.0 and 154.1, eliminating readmissions. During the process, two analyses were performed: one analysis with the principal diagnosis only and another with all the diagnostic codes. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and agreement was calculated with their 95% confidence intervals (CI). Results: With the first diagnosis only, the MDS detected 80% of the incident cases of colorectal cancer with a PPV of 75%. With all the diagnoses, the MDS detected 85% of the cases with a PPV of 64%. The agreement in codification was high at three digits (kappa 88% (95% CI, 0.79-0.97) first diagnosis, 89% (95% CI, 0.80-0.97) all diagnoses) as well as at four digits (kappa 77% (IC, 0.68-0.85) first diagnosis, 78% (95% CI, 0.70-0.86) all diagnoses) in both analyses. Conclusions: Because of its high sensitivity, the MDS is a good source for detecting incident cases of cancer. The high agreement found in the site of colorectal cancer helps to consolidate the MDS as a data source for cancer registration.

Keywords : Cancer registry; Hospital discharge database; Sensitivity; Agreement; Validation; Colorectal cancer.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )