Background: Quality of coding to identify cancers and comorbidities through the French hospital diagnosis database (Programme de médicalisation des systèmes d'information, PMSI) has been little investigated. Agreement between medical records and PMSI database was evaluated regarding metastatic colorectal cancer (mCRC) and comorbidities.
Methods: From 01/01/2013 to 06/30/2014, 74 patients aged≥65years at mCRC diagnosis were identified in Bordeaux teaching hospital. Data on mCRC and comorbidities were collected from medical records. All diagnosis codes (main, related and associated) registered into the PMSI were extracted. Agreement between sources was evaluated using the percent agreement for mCRC and the kappa (κ) statistic for comorbidities.
Results: Agreement for primary CRC and mCRC was higher using all types of diagnosis codes instead of the main one exclusively (respectively 95% vs. 53% for primary CRC and 91% vs. 24% for mCRC). Agreement was substantial (κ 0.65) for cardiovascular diseases, notably atrial fibrillation (κ 0.77) and hypertension (κ 0.68). It was moderate for psychiatric disorders (κ 0.49) and respiratory diseases (κ 0.48), although chronic obstructive pulmonary disease had a good agreement (κ 0.75). Within the class of endocrine, nutritional and metabolic diseases (κ 0.55), agreement was substantial for diabetes (κ 0.91), obesity (κ 0.82) and hypothyroidism (κ 0.72) and moderate for hypercholesterolemia (κ 0.51) and malnutrition (κ 0.42).
Conclusion: These results are reassuring with regard to detection through PMSI of mCRC if all types of diagnosis codes are considered and useful to better choose comorbidities in elderly mCRC patients that could be well identified through hospital diagnosis codes.
Keywords: Agreement; Cancer colorectal; Codes de la classification internationale des maladies (CIM-10); Colorectal neoplasm; Comorbidity; Comorbidités; Concordance; Dossiers médicaux; International Classification of Diseases (ICD-10) codes; Medical records; Métastases de cancer; Neoplasm metastasis.
Copyright © 2017 Elsevier Masson SAS. All rights reserved.