Premium
Validity Evidence for the Use of Automated Neuropsychologic Assessment Metrics As a Screening Tool for Cognitive Impairment in Systemic Lupus Erythematosus
Author(s) -
TayerShifman Oshrat E.,
Green Robin,
Beaton Dorcas E.,
Ruttan Lesley,
Wither Joan E.,
Tartaglia Maria Carmela,
Kakvan Mahta,
Lombardi Sabrina,
Anderson Nicole,
Su Jiandong,
Bonilla Dennisse,
Zandy Moe,
Choi May Y.,
Fritzler Marvin J.,
Touma Zahi
Publication year - 2020
Publication title -
arthritis care and research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.032
H-Index - 163
eISSN - 2151-4658
pISSN - 2151-464X
DOI - 10.1002/acr.24096
Subject(s) - neuropsychological assessment , receiver operating characteristic , cognition , neuropsychology , logistic regression , psychology , discriminative model , cognitive impairment , medicine , physical therapy , audiology , psychiatry , machine learning , computer science
Objective Screening for cognitive impairment in systemic lupus erythematosus (SLE) conventionally relies on the American College of Rheumatology (ACR) neuropsychologic battery (NB), which is not universally available. To develop a more accessible screening approach, we assessed validity of the Automated Neuropsychological Assessment Metrics (ANAM). Using the ACR NB as the gold standard for cognitive impairment classification, the objectives were 1) to measure overall discriminative validity of the ANAM for cognitive impairment versus no cognitive impairment, 2) to identify ANAM subtests and scores that best differentiate patients with cognitive impairment from those with no cognitive impairment, and 3) to derive ANAM composite indices and cutoffs. Methods A total of 211 consecutive adult patients, female and male, with SLE were administered the ANAM and ACR NB. 1) For overall discriminative validity of the ANAM, we compared patients with cognitive impairment versus those with no cognitive impairment on 4 scores. 2) Six ANAM models using different scores were developed, and the most discriminatory subtests were selected using logistic regression analyses. The area under the receiver operating characteristic curve (AUC) was calculated to establish ANAM validity against the ACR NB. 3) ANAM composite indices and cutoffs were derived for the best models, and sensitivities and specificities were calculated. Results Patients with no cognitive impairment performed better on most ANAM subtests, supporting ANAM’s discriminative validity. Cognitive impairment could be accurately identified by selected ANAM subtests with top models, demonstrating excellent AUCs of 81% and 84%. Derived composite indices and cutoffs demonstrated sensitivity of 78–80% and specificity of 70%. Conclusion This study provides support for ANAM’s discriminative validity for cognitive impairment and utility for cognitive screening in adult SLE. Derived composite indices and cutoffs enhance clinical applicability.