Lung-Cancer Diagnostic Trajectories: A Nationwide Process-Mining Cluster Analysis of 18,569 French Patients
Author(s)
Arnaud Panes, PharmD, PhD1, Helene Denis, PharmD2, Lionel Bensimon, MSc3, Isabelle Durand-Zaleski, MPP, PhD, MD4, Laurent Greillier, MD5, Pascal-Alexandre Thomas, MD6, Marie Wislez, MD7, Alice Brouquet, MSc2, Marion Apert, MS3, Valérie Guimard, MD3, Christine Le Bihan Benjamin, MD, PhD8, Marion Narbeburu, PhD8, Christos CHOUAID, MD9.
1Artificial intelligence and cancers association, Paris, France, 2Heva, Lyon, France, 3MSD France, Puteaux, France, 4Assistance Publique Hopitaux de Paris URCEco, Paris, France, 5Assistance publique - Hôpitaux de Marseille, Marseille, France, 6Department of Thoracic Surgery and Oesophageal Diseases, Hopital-Nord-APHM, Aix-Marseille University, Marseille, France, 7Oncology Thoracic Unit Pulmonology Department, AP-HP, Hôpital Cochin, Paris, France, 8French National institute of Cancer, Boulogne-Billancourt, France, 9Service de Pneumologie, Centre Hospitalier Intercommunal de Créteil, Créteil, France.
1Artificial intelligence and cancers association, Paris, France, 2Heva, Lyon, France, 3MSD France, Puteaux, France, 4Assistance Publique Hopitaux de Paris URCEco, Paris, France, 5Assistance publique - Hôpitaux de Marseille, Marseille, France, 6Department of Thoracic Surgery and Oesophageal Diseases, Hopital-Nord-APHM, Aix-Marseille University, Marseille, France, 7Oncology Thoracic Unit Pulmonology Department, AP-HP, Hôpital Cochin, Paris, France, 8French National institute of Cancer, Boulogne-Billancourt, France, 9Service de Pneumologie, Centre Hospitalier Intercommunal de Créteil, Créteil, France.
OBJECTIVES: Treatment delay worsens lung-cancer prognosis, but French real-world trajectories are under-documented. This study aimed to identify pre-treatment care-pathway clusters in national claims data for lung cancer in 2018-2019.
METHODS: Stage-specific finite-mixture models were applied to the French Cancer Cohort, diagnosed with local/locally-advanced (LLC) or advanced/metastatic (AMC) tumours, using all reimbursed activity in the 12 months before primary lung cancer treatment (systemic or surgery). Age, sex, Charlson index (mean ± SD) and time to treatment (TTT, median [IQR]) were described.
RESULTS: Four pre-treatment trajectory-early, late, continuous, last-minute-were identified.Among LLC patients (n=6,964), the early cluster (n=3,444, 49%) had mean age of 66.9 ± 9.8, 61% male, Charlson 7.1 ± 3.8, TTT of 37 days [0-68]; imaging and specialist visits rose 6-3 months pre-therapy. The late cluster (n=2,617, 38%) showed 65.2 ± 11.3, 59%, 7.8 ± 4.5, TTT of 31 days [0-55], with escalation limited to the final 3 months. In the continuous cluster (n=378, 5%) demographics were 68.4 ± 10.1 years, 62% male, Charlson 7.2 ± 3.5, TTT of 34.5 days [0-77]. The last-minute cluster (n=525, 8%) recorded 64.1 ± 11.4, 59%, 10.4 ± 4.7, TTT of 2 days [0-25], with frequent emergencies and longer stays. Among AMC patients (n=11,605), no early care-consumption trajectory was observed. The standard cluster (n=9,491, 82%) recorded 64.9 ± 10.1, 67%, 10.7 ± 4.6, TTT of 35 days [20-53]. The continuously cluster (n=1,170, 10%) had 68.1 ± 9.3, men 73%, Charlson 9.9 ± 4.7, TTT of 58 days [33-87]. The last-minute cluster (n=944, 8%) showed 64.5 ± 10.5, 65%, 12.2 ± 4.0, TTT 0 days [0-4].
CONCLUSIONS: Pre-treatment care-consumption trajectories provide a quantitative baseline for benchmarking care pathways and guiding stage-adapted interventions. Large TTT disparities were unrelated to age, sex or comorbidity.
METHODS: Stage-specific finite-mixture models were applied to the French Cancer Cohort, diagnosed with local/locally-advanced (LLC) or advanced/metastatic (AMC) tumours, using all reimbursed activity in the 12 months before primary lung cancer treatment (systemic or surgery). Age, sex, Charlson index (mean ± SD) and time to treatment (TTT, median [IQR]) were described.
RESULTS: Four pre-treatment trajectory-early, late, continuous, last-minute-were identified.Among LLC patients (n=6,964), the early cluster (n=3,444, 49%) had mean age of 66.9 ± 9.8, 61% male, Charlson 7.1 ± 3.8, TTT of 37 days [0-68]; imaging and specialist visits rose 6-3 months pre-therapy. The late cluster (n=2,617, 38%) showed 65.2 ± 11.3, 59%, 7.8 ± 4.5, TTT of 31 days [0-55], with escalation limited to the final 3 months. In the continuous cluster (n=378, 5%) demographics were 68.4 ± 10.1 years, 62% male, Charlson 7.2 ± 3.5, TTT of 34.5 days [0-77]. The last-minute cluster (n=525, 8%) recorded 64.1 ± 11.4, 59%, 10.4 ± 4.7, TTT of 2 days [0-25], with frequent emergencies and longer stays. Among AMC patients (n=11,605), no early care-consumption trajectory was observed. The standard cluster (n=9,491, 82%) recorded 64.9 ± 10.1, 67%, 10.7 ± 4.6, TTT of 35 days [20-53]. The continuously cluster (n=1,170, 10%) had 68.1 ± 9.3, men 73%, Charlson 9.9 ± 4.7, TTT of 58 days [33-87]. The last-minute cluster (n=944, 8%) showed 64.5 ± 10.5, 65%, 12.2 ± 4.0, TTT 0 days [0-4].
CONCLUSIONS: Pre-treatment care-consumption trajectories provide a quantitative baseline for benchmarking care pathways and guiding stage-adapted interventions. Large TTT disparities were unrelated to age, sex or comorbidity.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
SA63
Topic
Health Service Delivery & Process of Care, Study Approaches
Disease
Oncology