Performance of Regression-Based Norms for Cognitive Functioning of Persons With Multiple Sclerosis in an Independent Sample

Abstract

Background: Cognitive impairment is common in multiple sclerosis (MS). Interpretation of neuropsychological tests requires the use of normative data. Traditionally, normative data have been reported for discrete categories such as age. More recently continuous norms have been developed using multivariable regression equations that account for multiple demographic factors. Regression-based norms have been developed for use in the Canadian population for tests included in the MACFIMS and BICAMS test batteries. Establishing the generalizability of these norms is essential for application in clinical and research settings.Objectives: We aimed to (i) test the performance of previously published Canadian regression-based norms in an independently collected sample of Canadian healthy controls; (ii) compare the ability of Canadian and non-Canadian regression-based norms to discriminate between healthy controls and persons with MS; and (iii) develop regression-based norms for several cognitive tests drawn from batteries commonly used in MS that incorporated race/ethnicity in addition to age, education, and sex.Methods: We included 93 adults with MS and 96 healthy adults in this study, with a replication sample of 104 (MS) and 39 (healthy adults). Participants reported their sociodemographic characteristics, and each was administered the oral Symbol Digit Modalities Test (SDMT), the California Verbal Learning Test (CVLT-II), and the Brief Visuospatial Memory Test-Revised (BVMT-R). From the healthy control data, we developed regression-based norms incorporating race, age, education and sex. We then applied existing discrete norms and regression-based norms for the cognitive tests to the healthy controls, and generated z-scores which were compared using Spearman rank and concordance coefficients. We also used receiver operating characteristic (ROC) curves to compare the ability of each set of norms to discriminate between participants with and without MS. Within the MS samples we compared the ability of each set of norms to discriminate between differing levels of disability and employment status using relative efficiency.Results: When we applied the published regression norms to our healthy sample, impairment classification rates often differed substantially from expectations (7%), even when the norms were derived from a Canadian (Ontario) population. Most, but not all of the Spearman correlations between z-scores based on different existing published norms for the same cognitive test exceeded 0.90. However, concordance coefficients were often lower. All of the norms for the SDMT reliably discriminated between the MS and healthy control groups. In contrast, none of the norms for the CVLT-II or BVMT-R discriminated between the MS and healthy control groups. Within the MS population, the norms varied in their ability to discriminate between disability levels or employment status; locally developed norms for the SDMT and CVLT-II had the highest relative efficiency.Conclusion: Our findings emphasize the value of local norms when interpreting the results of cognitive tests and demonstrate the need to consider and assess the performance of regression-based norms developed in other populations when applying them to local populations, even when they are from the same country. Our findings also strongly suggest that the development of regression-based norms should involve larger, more diverse samples to ensure broad generalizability.

Predicting disability progression and cognitive worsening in multiple sclerosis using patterns of grey matter volumes

“…We used conventional norms and a raw score change of 4 point (10% change in magnitude) as a clinically meaningful measure 39 to define SDMT progression. However, future studies should investigate whether regression-based norms for SDMT might better detect abnormal performance in patients with MS [58][59][60] and therefore improve the performance of ICA components.…”

Effects of Vascular Comorbidity on Cognition in Multiple Sclerosis Are Partially Mediated by Changes in Brain Structure

“…We also enrolled healthy controls who have been described in detail elsewhere ( 28 ). Briefly, healthy controls were aged 18 years or older.…”

“…Specifically, we used the SDMT ( 39 ) to assess information processing speed, the CVLT-II; Trial 1–5 total recall score ( 40 ) to assess verbal learning and memory, the BVMT-R (summed recall score for all three learning trials) ( 41 ) to assess visual learning and memory, and tests of verbal fluency (letter and animal categories) ( 42 ) to assess language and executive abilities. We converted raw test scores to age-, sex- and education-adjusted z-scores using local regression-based norms because we previously demonstrated that these performed better in our population than other published regression-based norms ( 28 ). Z -scores of ≤-1.5 were classified as impaired.…”

“…Using a healthy control population which was enrolled concurrently and underwent the same study procedures, we developed regression-based norms for each MRI measure that incorporated age and gender, similar to the approach used to develop norms for cognitive tests in this population ( 28 ). This allowed us to convert each MRI measure to a z-score, enhancing their comparability despite the differences in their value ranges, and normalizing them for subsequent regression analyses.…”

Neuropsychological and Structural Neuroimaging Outcomes in LGI1-Limbic Encephalitis: A Case Study

“…To evaluate LGI1-LE-associated neuroanatomical changes in the MTL, the patient’s T1-weighted brain MRI scans were compared intraindividually and qualitatively to the T1-weighted brain MRI scans of n = 12 healthy age-matched controls (i.e., aged 60 years or older) obtained from Marrie et al.’s (2021) Comorbidity, Cognition and Multiple Sclerosis (C-COMS) study. One healthy control was excluded from the volumetric analysis due to a corrupted image file.…”