Related Experiment Video
Updated: May 19, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Hospital morbidity data for determining spread of disease at diagnosis for colorectal cancer: a validation study
Bridie Thompson1, John Lumley, Michael Coory
1School of Population Health, University of Queensland, Level 2, Public Health Building, Herston Road, Herston, QLD 4006, Australia. bridie.thompson@uqconnect.edu.au
Aims:
There is currently no routine collection of cancer stages in population-based data in Australia. This study evaluates the accuracy of International classification of diseases (ICD) codes for secondary neoplasms recorded in hospital morbidity data to assign spread of disease at diagnosis for colorectal cancer.
Methods:
The reference (gold) standard was the Australian clinicopathological stage (ACPS) documented by a treating colorectal surgeon and derived from histopathology and clinical findings. To allow comparison with stages derived from the hospital morbidity data (HMD), ACPS was mapped to the spread of disease (local, regional and distant). Sensitivity, specificity and positive-predictive values were calculated to compare the accuracy of stage derived from HMD.
Results:
Data from both the reference standard and HMD were available for 499 patients. HMD slightly overestimated patients with local disease (62.3 vs 56.9%). There was a corresponding underestimation of regional and distant spread of disease. While sensitivity for regional and distant disease was moderate (66.4 and 71.4%, respectively), specificity was high (92.7 and 96.6%, respectively).
Conclusion:
ICD codes for secondary neoplasms in HMD are limited in their utility for determining the spread of disease for colorectal cancer. Clinicians need to ensure that clinical coders are provided with enough information to accurately code for spread of disease. We recommend reporting histopathology in a synoptic format which includes background information on the presence or absence of distant metastasis and the tumor node metastasis stage.
