Purpose Although standard-of-care has been defined for the treatment of glioblastoma patients, substantial practice variation exists in the day-to-day clinical management. This study aims to... Show morePurpose Although standard-of-care has been defined for the treatment of glioblastoma patients, substantial practice variation exists in the day-to-day clinical management. This study aims to compare the use of laboratory tests in the perioperative care of glioblastoma patients between two tertiary academic centers-Brigham and Women's Hospital (BWH), Boston, USA, and University Medical Center Utrecht (UMCU), Utrecht, the Netherlands. Methods All glioblastoma patients treated according to standard-of-care between 2005 and 2013 were included. We compared the number of blood drawings and laboratory tests performed during the 70-day perioperative period using a Poisson regression model, as well as the estimated laboratory costs per patient. Additionally, we compared the likelihood of an abnormal test result using a generalized linear mixed effects model. Results After correction for age, sex, IDH1 status, postoperative KPS score, length of stay, and survival status, the number of blood drawings and laboratory tests during the perioperative period were 3.7-fold (p < 0.001) and 4.7-fold (p < 0.001) higher, respectively, in BWH compared to UMCU patients. The estimated median laboratory costs per patient were 82 euros in UMCU and 256 euros in BWH. Furthermore, the likelihood of an abnormal test result was lower in BWH (odds ratio [OR] 0.75, p < 0.001), except when the prior test result was abnormal as well (OR 2.09, p < 0.001). Conclusions Our results suggest a substantially lower clinical threshold for ordering laboratory tests in BWH compared to UMCU. Further investigating the clinical consequences of laboratory testing could identify over and underuse, decrease healthcare costs, and reduce unnecessary discomfort that patients are exposed to. Show less
PURPOSE The aim of this study was to develop an open-source natural language processing (NLP) pipeline for text mining of medical information from clinical reports. We also aimed to provide insight... Show morePURPOSE The aim of this study was to develop an open-source natural language processing (NLP) pipeline for text mining of medical information from clinical reports. We also aimed to provide insight into why certain variables or reports are more suitable for clinical text mining than others.MATERIALS AND METHODS Various NLP models were developed to extract 15 radiologic characteristics from free-text radiology reports for patients with glioblastoma. Ten-fold cross-validation was used to optimize the hyperparameter settings and estimate model performance. We examined how model performance was associated with quantitative attributes of the radiologic characteristics and reports.RESULTS In total, 562 unique brain magnetic resonance imaging reports were retrieved. NLP extracted 15 radiologic characteristics with high to excellent discrimination (area under the curve, 0.82 to 0.98) and accuracy (78.6% to 96.6%). Model performance was correlated with the inter-rater agreement of the manually provided labels (rho = 0.904; P<.001) but not with the frequency distribution of the variables of interest (rho = 0.179; P =.52). All variables labeled with a near perfect inter-rater agreement were classified with excellent performance (area under the curve > 0.95). Excellent performance could be achieved for variables with only 50 to 100 observations in the minority group and class imbalances up to a 9:1 ratio. Report-level classification accuracy was not associated with the number of words or the vocabulary size in the distinct text documents.CONCLUSION This study provides an open-source NLP pipeline that allows for text mining of narratively written clinical reports. Small sample sizes and class imbalance should not be considered as absolute contraindications for text mining in clinical research. However, future studies should report measures of inter-rater agreement whenever ground truth is based on a consensus label and use this measure to identify clinical variables eligible for text mining. Show less
