Background. Perianal fistulising disease is a disabling complication of Crohn's disease that carries substantial morbidity and impairs quality of life; early identification of at-risk patients at the time of inflammatory bowel disease (IBD) diagnosis remains an unmet clinical need. Aim. To develop and internally validate a clinical prediction model for perianal fistulising Crohn's disease and derive a points-based risk score using variables available at the time of IBD diagnosis. Methods. Retrospective multi-centre cohort study of adult patients with Crohn's disease followed at nine Italian IBD referral centres (2006–2024). Baseline candidate predictors were age, sex, smoking status, Montreal location, L4 modifier, Montreal behaviour, rectal involvement, extra-intestinal manifestations and family history of IBD. Missing data were handled by multiple imputation (m=20) and pooled multivariable logistic regression with Rubin's rules. Bootstrap internal validation (1000 iterations) provided optimism-corrected performance metrics; decision curve analysis quantified clinical utility; ridge regression provided a robustness check. Reporting followed the TRIPOD+AI and STROBE statements. Results. Among 966 patients (323 cases, 33.4%), penetrating behaviour was the strongest predictor (adjusted OR 9.78, 95% CI 6.61–14.47; p<0.001), with current smoking (3.54, 2.38–5.25; p<0.001), former smoking (2.98, 1.89–4.69; p<0.001), colonic location (1.87, 1.15–3.05; p=0.012) and younger age at diagnosis (0.985 per year, 0.975–0.996; p=0.009) also independently associated; female sex was independently protective (0.64, 0.46–0.88; p=0.007). The optimism-corrected c-statistic was 0.77, with calibration slope 0.95 and calibration intercept –0.03. The points-based score stratified patients into five risk strata with observed event probabilities of 14.1%, 18.0%, 23.8%, 40.4% and 77.9%. Decision curve analysis showed greater net benefit than treat-all or treat-none strategies across clinically relevant threshold probabilities. Conclusions. A simple score based on baseline characteristics available at IBD diagnosis, including smoking status, stratifies newly diagnosed Crohn's disease patients into relative-risk groups for perianal fistulising disease. External validation in an independent cohort is the necessary next step before clinical use.
Background. La malattia di Crohn perianale fistolizzante è una manifestazione disabilitante della malattia di Crohn, gravata da significativa morbilità e da elevato impatto sulla qualità di vita; l'identificazione precoce dei pazienti a rischio al momento della diagnosi di malattia infiammatoria cronica intestinale (MICI) resta tuttavia un'esigenza clinica non soddisfatta. Scopo. Sviluppare e validare internamente un modello di predizione clinica per la malattia di Crohn perianale fistolizzante e derivare uno score di rischio basato su punti, utilizzando esclusivamente caratteristiche disponibili al momento della diagnosi di MICI. Metodi. Studio retrospettivo multicentrico di coorte su pazienti adulti con malattia di Crohn seguiti presso nove centri italiani di riferimento per le MICI (2006–2024). I predittori candidati al baseline sono stati età, sesso, status tabagico, localizzazione e comportamento secondo Montreal, modificatore L4, coinvolgimento rettale, manifestazioni extra-intestinali e familiarità per MICI. I dati mancanti sono stati gestiti con multiple imputation (m=20) e regressione logistica multivariabile aggregata secondo le regole di Rubin. La validazione interna mediante bootstrap (1000 iterazioni) ha fornito le metriche corrette per ottimismo; la decision curve analysis ha quantificato l'utilità clinica; la regressione ridge è stata utilizzata come controllo di stabilità contro l'overfitting. Il reporting ha seguito le checklist TRIPOD+AI e STROBE. Risultati. Tra 966 pazienti (323 casi, 33,4%), il comportamento penetrante è risultato il predittore più forte (OR aggiustato 9,78; IC 95% 6,61–14,47; p<0,001), insieme a fumo attivo (3,54; 2,38–5,25; p<0,001), ex fumo (2,98; 1,89–4,69; p<0,001), localizzazione colica (1,87; 1,15–3,05; p=0,012) ed età più giovane alla diagnosi (0,985 per anno; 0,975–0,996; p=0,009); il sesso femminile è risultato indipendentemente protettivo (0,64; 0,46–0,88; p=0,007). Il c-statistic corretto per ottimismo è stato 0,77, con calibration slope 0,95 e calibration intercept –0,03. Lo score basato su punti ha stratificato i pazienti in cinque strati di rischio con probabilità osservate dell'evento del 14,1%, 18,0%, 23,8%, 40,4% e 77,9%. La decision curve analysis ha evidenziato un net benefit superiore alle strategie treat-all e treat-none lungo soglie di probabilità clinicamente rilevanti. Conclusioni. Uno score semplice basato su caratteristiche disponibili al momento della diagnosi di MICI, inclusivo dello status tabagico, stratifica i pazienti con malattia di Crohn di nuova diagnosi in gruppi di rischio relativo per malattia perianale fistolizzante. La validazione esterna in una coorte indipendente è il passo necessario prima dell'utilizzo clinico.
Sviluppo e validazione interna di un modello di predizione clinica per la malattia di Crohn perianale fistolizzante: uno studio multicentrico italiano.
NARDIN, ELEONORA
2025/2026
Abstract
Background. Perianal fistulising disease is a disabling complication of Crohn's disease that carries substantial morbidity and impairs quality of life; early identification of at-risk patients at the time of inflammatory bowel disease (IBD) diagnosis remains an unmet clinical need. Aim. To develop and internally validate a clinical prediction model for perianal fistulising Crohn's disease and derive a points-based risk score using variables available at the time of IBD diagnosis. Methods. Retrospective multi-centre cohort study of adult patients with Crohn's disease followed at nine Italian IBD referral centres (2006–2024). Baseline candidate predictors were age, sex, smoking status, Montreal location, L4 modifier, Montreal behaviour, rectal involvement, extra-intestinal manifestations and family history of IBD. Missing data were handled by multiple imputation (m=20) and pooled multivariable logistic regression with Rubin's rules. Bootstrap internal validation (1000 iterations) provided optimism-corrected performance metrics; decision curve analysis quantified clinical utility; ridge regression provided a robustness check. Reporting followed the TRIPOD+AI and STROBE statements. Results. Among 966 patients (323 cases, 33.4%), penetrating behaviour was the strongest predictor (adjusted OR 9.78, 95% CI 6.61–14.47; p<0.001), with current smoking (3.54, 2.38–5.25; p<0.001), former smoking (2.98, 1.89–4.69; p<0.001), colonic location (1.87, 1.15–3.05; p=0.012) and younger age at diagnosis (0.985 per year, 0.975–0.996; p=0.009) also independently associated; female sex was independently protective (0.64, 0.46–0.88; p=0.007). The optimism-corrected c-statistic was 0.77, with calibration slope 0.95 and calibration intercept –0.03. The points-based score stratified patients into five risk strata with observed event probabilities of 14.1%, 18.0%, 23.8%, 40.4% and 77.9%. Decision curve analysis showed greater net benefit than treat-all or treat-none strategies across clinically relevant threshold probabilities. Conclusions. A simple score based on baseline characteristics available at IBD diagnosis, including smoking status, stratifies newly diagnosed Crohn's disease patients into relative-risk groups for perianal fistulising disease. External validation in an independent cohort is the necessary next step before clinical use.| File | Dimensione | Formato | |
|---|---|---|---|
|
Nardin_Eleonora_1205823.pdf
Accesso riservato
Dimensione
2.05 MB
Formato
Adobe PDF
|
2.05 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/109078