Computational definition of medical exclusion and feasibility of excluding people not eligible for French population-based colorectal cancer screening from the French medico-administrative database

• Published in: BMC medical informatics and decision making Vol. 25; no. 1; pp. 334 - 12
• Main Authors: Koïvogui, Akoï, Balamou, Christian, Benamouzig, Robert, Duclos, Catherine
• Format: Journal Article
• Language: English
• Published: London BioMed Central 26.09.2025; BioMed Central Ltd; Springer Nature B.V; BMC
• Subjects: Age; Algorithms; Cancer; Cancer screening; Colorectal cancer; Colorectal cancer screening; Colorectal carcinoma; Colorectal disease phenotype; Colorectal Neoplasms - diagnosis; Computer applications; Crohn's disease; Databases, Factual - statistics & numerical data; Decision making; Diagnosis; Diagnostics; Early Detection of Cancer - statistics & numerical data; Epidemiology; Familial adenomatous polyposis; Feasibility; Feasibility Studies; France; Genetics; Health care disparities; Health Informatics; Health insurance; Humans; Inflammatory bowel disease; Inflammatory bowel diseases; Information systems; Information Systems and Communication Service; Life Sciences; Literature reviews; Management of Computing and Information Systems; Medical care; Medical exclusion rate; Medical screening; Medicine; Medicine & Public Health; Medico-administrative database; Oncology, Experimental; Participation rate; Patient Selection; Phenotypes; Polyps; Population-based studies; Search algorithms; Utilization; France; Colorectal cancer screening; Medico-administrative database; Colorectal disease phenotype; Participation rate; Medical exclusion rate
• ISSN: 1472-6947; 1472-6947
• DOI: 10.1186/s12911-025-03175-5

Background In the French population-based colorectal cancer screening program (CRCSP), the fact that the medical-exclusion rate was estimated only after a collection of voluntary statements from subjects could compromise an exhaustive collection of potential cases of medical-exclusion. The health insurance medico-administrative database (SNDS) that contains medical and healthcare consumption information have to date never been used to refine the target population of the CRCSP. Objective To identify in the SNDS, from published and disparate algorithms, the computational definitions of morbid situations that could justify medical exclusion from the CRCSP. Methods The non-systematic review of the literature synthetised an exhaustive list of algorithms targeting in SNDS, the morbid situations (CCR, colorectal adenoma/polyp, chronic inflammatory bowel disease, familial adenomatous polyposis, or Lynch syndrome colonoscopy, coloscanner, polypectomy) which may justify temporary or permanent medical exclusion from the CRCSP campaigns. Secondly, the discovered codes of morbid situations were searched on statistical reports to estimate their frequencies of use in SNDS (in 2021), and their interest in the computational phenotypes’ algorithm. Results The analysis of the literature (28 articles/studies) highlights the existence of diagnostic or therapeutic codes that can define in the SNDS database, the morbid situations justifying medical exclusion from the CRCSP. Except for personal or family history of CRC classifiable in the Z85.0 or Z80.0 codes of the ICD-10, almost all the morbid situations have a requestable definition in the SNDS. The target favoured by the search algorithms was the ICD-10 code (i.e., C18-C20, K50, K51). The definition codes listed were frequently used in SNDS in 2021, except for a few codes (D12.6 + 6, M07.5). From this definition of morbid situations by the only codes of the ICD-10 or the procedure codes emerges a feasibility and a decision-making algorithm for the choice of the person to be excluded from CRCSP campaign, using the SNDS. Age is the first discriminating variable in this decision-making algorithm because the CRCSP targeted people aged 50 to 74 years old and a restriction on age was made in several included SNDS’s studies. The second discrimination based on diagnostic evidence derives its relevance from the quasi-systematic search for ICD-10 diagnostic codes in SNDS’s studies. Conclusion In addition to being widely used in the context of medico-economic and epidemiological studies, the SNDS currently contains almost all the data essential for estimating the rate of medical-exclusion during colorectal cancer screening campaigns. While initiating the answer to the question of the choice of the most appropriate algorithm in each context, this review of the literature also emphasizes the need for validation studies because the quality of the algorithms used conditions the quality of the studies carried out in the medico-administrative databases. Summary points What was already known on the topic: • In the French population-based colorectal cancer screening program, the medical exclusion rate was estimated only after a collection of voluntary statements from subjects and GPs, which compromises an exhaustive collection of potential cases of medical exclusion. • The French medico-administrative database contains medical, and healthcare information have to date never been used to refine the target population of the screening program. What this study added to our knowledge: • This review showed that there was no unique algorithm that can identify in the French database, all morbid situations that can justify medical exclusion from CRCSP campaigns. • The use of a single phenotype model and the lack of validation of research phenotypes of the chronic colorectal diseases in the French database. • The French medico-administrative database currently contains almost all the data essential for estimating the rate of medical exclusion during screening campaigns.