Development of an algorithm to support cancer registry registration by identifying patients with metachronous metastases after primary breast cancer
- 1Department of Research and Development, Netherlands Comprehensive Cancer Organisation, Boven Clarenburg, Utrecht, The Netherlands
- 2Department of Research and Development, Netherlands Comprehensive Cancer Organisation, Boven Clarenburg, Utrecht, The Netherlands; Department of Health Technology and Services Research, Technical Medical Centre, University of Twente, Drienerlolaan, Enschede, The Netherlands
Abstract
Background: Cancer registries do not have resources to follow-up disease statuses of all patients actively. To get information on metachronous metastatic breast cancer (mMBC), we developed an algorithm based on hospital data from, among other, inpatient admissions, outpatient treatments, diagnoses, and medication use to identify patients with a high probability of having developed mMBC, with the aim of reducing registration burden. Materials and methods: Nationwide hospital data from 2019 to 2020 were linked to data of patients with a primary breast cancer diagnosis between 1989 and 2020 in the Netherlands Cancer Registry (NCR). For 4395 patients, data on mMBC status (M1/M0) was available in the NCR and used as gold standard to develop a prediction algorithm. After validation its performance was evaluated with a pilot on 1056 patients with unknown M1/M0 status and an estimated prediction score of having metastasis in the range 0% to 100% which were manually checked by registrars. These results were used to retrain the algorithm and patients with a prediction score for mMBC of ≥0.6 were checked manually. Results: Predictive variables of the algorithm included specific imaging procedures and medication use, counselling, diagnosis for MBC, and the number of hospital visits. Manual review confirmed that higher predicted probabilities corresponded with higher proportions of mMBC found. Of 3379 patients with prediction score of ≥0.6, 3029 patients (90%) were indeed diagnosed with mMBC. Conclusions: We developed an algorithm based on hospital data to identify patients with a high probability of having developed mMBC. The algorithm will be updated continuously and implemented in cancer registry practice.
Related
- Development and external validation of an XGBoost-based prognostic model for small cell lung cancer with brain metastases
- Identifying Hemodialysis Patients With the Highest Risk of Staphylococcus aureus Endogenous Infection Through a Simple Nasal Sampling Algorithm
- Development of a concise objective biometric triage model for identifying myopia defined by cycloplegic refraction in school-aged children
- Eye-movement-based machine learning for identifying Parkinson’s disease dementia in a retrospective cohort: an interpretable model-development study
- Development of a nomogram for identifying prevalent medication discrepancy in maintenance hemodialysis patients: a multicenter cross-sectional study
BibTeX
@article{munck2026development,
title = {{Development of an algorithm to support cancer registry registration by identifying patients with metachronous metastases after primary breast cancer}},
author = {L. de Munck and H. Buisman and J. Verloop and S. Siesling},
journal = {ESMO Real World Data and Digital Oncology},
year = {2026},
month = nov,
doi = {10.1016/j.esmorw.2026.100811},
url = {https://doi.org/10.1016/j.esmorw.2026.100811},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Development of an algorithm to support cancer registry registration by identifying patients with metachronous metastases after primary breast cancer
AU - L. de Munck
AU - H. Buisman
AU - J. Verloop
AU - S. Siesling
PY - 2026
DA - 2026/11/01
JO - ESMO Real World Data and Digital Oncology
DO - 10.1016/j.esmorw.2026.100811
UR - https://doi.org/10.1016/j.esmorw.2026.100811
AB - Background: Cancer registries do not have resources to follow-up disease statuses of all patients actively. To get information on metachronous metastatic breast cancer (mMBC), we developed an algorithm based on hospital data from, among other, inpatient admissions, outpatient treatments, diagnoses, and medication use to identify patients with a high probability of having developed mMBC, with the aim of reducing registration burden. Materials and methods: Nationwide hospital data from 2019 to 2020 were linked to data of patients with a primary breast cancer diagnosis between 1989 and 2020 in the Netherlands Cancer Registry (NCR). For 4395 patients, data on mMBC status (M1/M0) was available in the NCR and used as gold standard to develop a prediction algorithm. After validation its performance was evaluated with a pilot on 1056 patients with unknown M1/M0 status and an estimated prediction score of having metastasis in the range 0% to 100% which were manually checked by registrars. These results were used to retrain the algorithm and patients with a prediction score for mMBC of ≥0.6 were checked manually. Results: Predictive variables of the algorithm included specific imaging procedures and medication use, counselling, diagnosis for MBC, and the number of hospital visits. Manual review confirmed that higher predicted probabilities corresponded with higher proportions of mMBC found. Of 3379 patients with prediction score of ≥0.6, 3029 patients (90%) were indeed diagnosed with mMBC. Conclusions: We developed an algorithm based on hospital data to identify patients with a high probability of having developed mMBC. The algorithm will be updated continuously and implemented in cancer registry practice.
SN - 2949-8201
ER -