Comparisons of Propensity Score Methods for Assessing Racial Disparities in Multiple Myeloma Health Outcomes
Racial disparities have been observed in the health outcomes of multiple myeloma (MM) in the
US, but gaps remain in understanding these disparities with real-world data. This study investigates racial disparities in the MM overall survival (OS) and time from diagnosis to next treatment (TTNT) using the nationwide Flatiron Health electronic health record-derived deidentified database. We analyzed 11,997 MM patients from five race/ethnicity groups. Seven propensity score (PS) methods were applied to balance covariates, including nearest neighbor matching (with or without caliper), optimal matching, stabilized inverse probability of treatment weighting (IPTW), IPTW with trimming, stratification, and PS as a covariate. Cox proportional-hazards model was applied to assess associations between race/ethnicity and OS and TTNT. This study reveals racial disparities in OS among MM patients in the US and evaluates PS methods to assess covariate balance in observational studies.
US, but gaps remain in understanding these disparities with real-world data. This study investigates racial disparities in the MM overall survival (OS) and time from diagnosis to next treatment (TTNT) using the nationwide Flatiron Health electronic health record-derived deidentified database. We analyzed 11,997 MM patients from five race/ethnicity groups. Seven propensity score (PS) methods were applied to balance covariates, including nearest neighbor matching (with or without caliper), optimal matching, stabilized inverse probability of treatment weighting (IPTW), IPTW with trimming, stratification, and PS as a covariate. Cox proportional-hazards model was applied to assess associations between race/ethnicity and OS and TTNT. This study reveals racial disparities in OS among MM patients in the US and evaluates PS methods to assess covariate balance in observational studies.
Comparaisons des méthodes de score de propension pour l'évaluation des disparités raciales dans les résultats de santé du myélome multiple
Des disparités raciales dans les résultats de santé du myélome multiple (MM) ont été observées aux États-Unis, mais leur compréhension à l’aide de données réelles reste limitée. Cette étude examine ces disparités en analysant la survie globale (SG) et le temps jusqu’au prochain traitement (TTNT) à partir des données à l'échelle de la nation de santé électroniques de Flatiron Health. L’analyse a porté sur 11 997 patients de cinq groupes raciaux/ethniques. Sept méthodes de score de propension (PS) ont été appliquées pour calibrer les covariables, incluant l'appariement par plus proche voisin (avec/sans étrier), l'appariement optimal, la pondération par l'inverse de la probabilité stabilisée de traitement IPTW (stabilisée/élagage), la stratification et le PS comme covariable. Un modèle de Cox a été utilisé pour évaluer l’association entre race/ethnicité et SG/TTNT. Cette étude met en évidence des disparités raciales dans la SG et évalue l’efficacité des méthodes de PS pour calibrer les covariables dans les études observationnelles.
Date and Time
-
Language of Oral Presentation
English
Language of Visual Aids
English