Paolo Dell'Oglio1,2, Zhe Tian3,4, Sami-Ramzi Leyh-Bannurah3,5, Alessandro Larcher6, Elio Mazzone6, Marco Moschini6, Vincent Trudeau3,7, Armando Stabile6, Andrea Gallina6, Nazareno Suardi6, Umberto Capitanio6, Alexandre Mottrie8,9, Alberto Briganti6, Francesco Montorsi6, Christian M Rochefort4,10, Pierre I Karakiewicz3,7. 1. Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, QC, Canada. paolo.delloglio@gmail.com. 2. Division of Oncology/Unit of Urology, Urological Research Institute, IRCCS Ospedale San Raffaele, Milan, Italy. paolo.delloglio@gmail.com. 3. Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, QC, Canada. 4. Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Canada. 5. Martini-Clinic, Prostate Cancer Center Hamburg-Eppendorf, Hamburg, Germany. 6. Division of Oncology/Unit of Urology, Urological Research Institute, IRCCS Ospedale San Raffaele, Milan, Italy. 7. Department of Urology, University of Montreal Health Center, Montreal, Canada. 8. ORSI, Academy, Melle, Belgium. 9. Department of Urology, Onze Lieve Vrouw Hospital, Aalst, Belgium. 10. Faculté de Médecine, École des Sciences Infirmières, Université de Sherbrooke, Sherbrooke, Canada.
Abstract
PURPOSE: The Deyo adaptation of the Charlson comorbidity index (DaCCI), which relies on 17 comorbid condition groupings defined with 200 ICD-9-CM diagnostic codes, lacks specificity in the context of radical cystectomy (RC) for bladder cancer (BCa). We attempted to develop a new comorbidity assessment tool based on individual comorbid conditions and/or BCa manifestations for specific prediction of perioperative mortality after RC. METHODS: We relied on 7076 T1-T4 nonmetastatic BCa patients treated with RC between 2000 and 2009 in the SEER-Medicare linked database. Within the development cohort (n = 6076), simulated annealing (SA) was used to identify (1) individual comorbid conditions, (2) individual BCa manifestations, and (3) the combination of both, that satisfy the criteria of maximal accuracy and parsimony for prediction of 90-day mortality after RC, after adjusting for several confounders. The accuracy of the newly identified groups of individual comorbid conditions and/or BCa manifestations and of the original DaCCI was tested in a 1000-patient external validation cohort. RESULTS: The combination of six individual comorbid conditions and two individual BCa disease manifestations (type II diabetes without complications, anemia, chronic obstructive pulmonary disease, congestive heart failure, aortocoronary bypass, cardiomegaly, urinary tract infection, and hydronephrosis), and seven individual comorbid conditions (type II diabetes without complications, anemia, chronic obstructive pulmonary disease, congestive heart failure, aortocoronary bypass, osteoarthrosis, and cardiomegaly) respectively showed 71.1 and 70.2% accuracy versus 68.0% for the original DaCCI. CONCLUSIONS: These new approaches are specific to contemporary RC patients and represent simpler methods compared with the original DaCCI, without any compromise in accuracy.
PURPOSE: The Deyo adaptation of the Charlson comorbidity index (DaCCI), which relies on 17 comorbid condition groupings defined with 200 ICD-9-CM diagnostic codes, lacks specificity in the context of radical cystectomy (RC) for bladder cancer (BCa). We attempted to develop a new comorbidity assessment tool based on individual comorbid conditions and/or BCa manifestations for specific prediction of perioperative mortality after RC. METHODS: We relied on 7076 T1-T4 nonmetastatic BCa patients treated with RC between 2000 and 2009 in the SEER-Medicare linked database. Within the development cohort (n = 6076), simulated annealing (SA) was used to identify (1) individual comorbid conditions, (2) individual BCa manifestations, and (3) the combination of both, that satisfy the criteria of maximal accuracy and parsimony for prediction of 90-day mortality after RC, after adjusting for several confounders. The accuracy of the newly identified groups of individual comorbid conditions and/or BCa manifestations and of the original DaCCI was tested in a 1000-patient external validation cohort. RESULTS: The combination of six individual comorbid conditions and two individual BCa disease manifestations (type II diabetes without complications, anemia, chronic obstructive pulmonary disease, congestive heart failure, aortocoronary bypass, cardiomegaly, urinary tract infection, and hydronephrosis), and seven individual comorbid conditions (type II diabetes without complications, anemia, chronic obstructive pulmonary disease, congestive heart failure, aortocoronary bypass, osteoarthrosis, and cardiomegaly) respectively showed 71.1 and 70.2% accuracy versus 68.0% for the original DaCCI. CONCLUSIONS: These new approaches are specific to contemporary RC patients and represent simpler methods compared with the original DaCCI, without any compromise in accuracy.