An Empirical Examination of the Aspen Consensus Criteria’s Diagnostic Accuracy Using the Coma Recovery Scale-Revised Rating Scale Categories: A Rasch Co-Calibration Analysis

Patients with disorders of consciousness (DoC) are often misdiagnosed as being in a lower state of consciousness than they are, which can impede their access to care. Cut points for each state of consciousness on the Coma Recovery Scale-Revised (CRS-R) could support researchers in stratifying patients, facilitate interpretation of results, and enhance comparability across studies. The primary purpose of this study was to empirically examine the alignment between the CRS-R rating scale categories and states of consciousness on the same Rasch-transformed equal-interval ruler. States of consciousness included the unresponsive wakefulness syndrome (UWS), minimally conscious state (MCS), and emerged from MCS (eMCS). Our secondary objective was to generate and assess the sensitivity and specificity of the cut points for each DoC diagnosis on the equal-interval ruler. We hypothesize that empirically derived cut point thresholds will reclassify some records to a higher-level state of consciousness and provide enhanced precision at clinically relevant transitions. We used the Rasch measurement model to co-calibrate the CRS-R and two DoC diagnostic variables on a 0- to 100-unit equal-interval scale. One DoC diagnostic variable used the Aspen consensus criteria aligned to the CRS-R rating scale categories (Aspen-Based). The second DoC diagnostic variable was based on an examination of the CRS-R rating scale category difficulty (Rasch-Derived). We used CRS-R data from 262 participants with DoC (1,442 CRS-R records). The Aspen-Based DoC diagnostic variable indicated participants were UWS <41.57, 41.57 ≤ MCS < 73.16, and eMCS ≥73.16 units. Six CRS-R rating scale categories were not consistent with their Aspen-Based alignment to DoC state. Therefore, we realigned four of these six CRS-R rating scale categories to indicate their respective higher state of consciousness when we created the Rasch-Derived DoC diagnostic variable. After reclassifying the CRS-R records, 19% of the 1,442 CRS-R records reflected a higher state of consciousness. The Rasch-Derived DoC diagnostic variable indicated participants were UWS <34.40, 34.40 ≤ MCS < 67.07, and eMCS ≥67.07 units. We conducted a sensitivity and specificity analysis on the UWS to MCS and MCS to eMCS cut points using the Aspen-Based DoC and Rasch-Derived DoC person measures. We improved the existing sensitivity and specificity analyses by recommending a UWS to MCS cut point of 34.40 units and an MCS to eMCS cut point of 67.07 units on the 0- to 100-unit scale. Identifying additional rating scale categories that reflect higher DoC diagnoses (e.g., MCS or eMCS) and lower cut points for MCS and eMCS could have implications for improving the diagnostic accuracy of the CRS-R and ultimately, facilitating better access to medical and rehabilitation services. A future study should examine the external validation of the Rasch-Derived cut points using multimodal approaches such as neuroimaging or electrophysiologic evaluation.