Cognitive impairment affects ∼40–90% of people with multiple sclerosis (pwMS), contributing to diminished quality of life and functioning. ^{Reference Benedict, Amato, DeLuca and Geurts1} It is thus vital to accurately detect and measure cognition in research studies and clinical care. A common concern of pwMS and investigators is whether the “time-of-testing” affects neurological functioning, including cognitive performance. ^{Reference Mortensen, Theódórsdóttir, Sejbæk and Illes2} While this is understandable given diurnal increases in subjective fatigue, ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3} regarding objective cognitive measures, the literature is equivocal. ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3–Reference Beatty, Goretti, Siracusa, Zipoli, Portaccio and Amato5} These studies may be limited by small samples ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3–Reference Beatty, Goretti, Siracusa, Zipoli, Portaccio and Amato5} and failures to control for confounding variables. ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3–Reference Beatty, Goretti, Siracusa, Zipoli, Portaccio and Amato5}

In a large clinical sample, this study evaluated whether neuropsychological testing time independently predicts cognitive performance in pwMS. In addition to clarifying conflicting findings, an association between time-of-testing and cognition would have ramifications for occupational functioning and driving in pwMS, patient education, clinical interpretation of cognitive test results and future study design.

Data were obtained from a retrospective chart review of pwMS (diagnosed according to the McDonald criteria ^{Reference Thompson, Banwell and Barkhof6} ) who attended a tertiary neuropsychiatry clinic in Toronto, Canada, between 2020 and 2024 and had a documented time-of-testing. Neuropsychological testing was completed as part of routine care, scheduled to start continuously between 8:00 and 15:00, and was conducted in an office without window exposure. The Sunnybrook Research Ethics Board approved this study.

Testing time was organized in hourly increments. We collected demographic and disease-related data such as age, sex, educational years, employment status, disease duration, neurological disability (measured with the Expanded Disability Status Scale [EDSS]), illness subtype, disease-modifying therapy (DMT) use and Modified Fatigue Impact Scale (MFIS) scores to measure global fatigue.

Cognitive performance was measured using raw scores from the validated Minimal Assessment of Cognitive Function in Multiple Sclerosis (MACFIMS). ^{Reference Benedict, Cookfair and Gavett7} The MACFIMS neuropsychological battery includes indices of verbal fluency and executive function (the Controlled Oral Word Association Test [COWAT]), visuospatial function (Judgment of Line Orientation test [JOLO]), verbal learning and delayed recall (California Verbal Learning Test Second Edition [CVLT-II]), visual learning and delayed recall (Brief Visuospatial Memory Test-Revised [BVMT-R]), processing speed (Symbol Digit Modalities Test [SDMT]), 2- and 3-second versions of the Paced Auditory Serial Addition Test (PASAT) and executive function (Delis–Kaplan Executive Function System [D-KEFS]). The CVLT-II and BVMT-R have scores for total learning (CVLT-II-TL and BVMT-R-TL) and delayed recall (CVLT-II-DR and BVMT-R-DR). The D-KEFS has correct sorts (D-KEFS_corsor) and descriptive scores (D-KEFS_desc). This battery was always administered in the same order.

Preliminary point-biserial or Pearson correlations were computed between time-of-testing and all demographic, disease-related and MACFIMS neuropsychological battery measures. Thereafter, linear regression analyses were conducted to evaluate whether time-of-testing independently predicted cognitive raw scores, adjusting for age, sex, educational years, disease duration, EDSS scores and DMT use. The significance threshold was set at p < 0.01 to account for multiple testing.

Of 791 pwMS, the mean age was 43.76 years (SD = 11.30), 76.74% were female, 52.47% were employed and the mean years of education was 15.98 years (SD = 2.94). The mean disease duration was 9.95 years (SD = 8.86), and most participants had mild neurological disability (median = 2.00, IQR = 1.50–3.50) and relapsing illness (84.20%) and received a DMT (76.71%). The mean MFIS score was 50.08 (SD = 16.54). The median time of neuropsychological testing was 11:00 a.m. (IQR = 10:00–13:00). There were no significant associations between time-of-testing and any demographic or disease-related variables.

Later time-of-testing weakly correlated with reduced performance on JOLO (r = −0.10, p < 0.01), with trends toward significance for BVMT-R-TL (r = −0.08, p = 0.02), BVMT-R-DR (r = −0.08, p = 0.03) and SDMT (r = −0.09, p = 0.02). Supplementary Table 1 details mean cognitive raw scores and correlations with time-of-testing. As noted in Table 1, after adjusting for covariates, later time-of-testing independently predicted reduced performance on JOLO (β = −0.09, p < 0.01) with trends toward significance for reduced BVMT-R-TL (β = −0.07, p = 0.03), BVMT-R-DR (β = −0.07, p = 0.05) and SDMT (β = −0.06, p = 0.05) scores. Time-of-testing was not significantly associated with performance on the COWAT, CVLT-II (both tests), PASAT (both tests) or D-KEFS (both tests).

Table 1. Standardized beta coefficients from linear regression models to predict cognitive raw scores from time-of-testing, adjusted for covariates

Note: EDSS = Expanded Disability Status Scale; DMT use = disease-modifying therapy use; COWAT = Controlled Oral Word Association Test; JOLO = Judgment of Line Orientation; CVLT-II-TL= California Verbal Learning Test (Second Edition) total learning; CVLT-II-DR = California Verbal Learning Test (Second Edition) delayed recall; BVMT-R-TL = Brief Visuospatial Memory Test-Revised total learning; BVMT-R-DR = Brief Visuospatial Memory Test-Revised delayed recall; SDMT = Symbol Digit Modalities Test; PASAT_3 sec = Paced Auditory Serial Addition Test 3-second; PASAT_2 sec = Paced Auditory Serial Addition Test 2-second; D-KEFS_corsor = Delis–Kaplan Executive Function System correct sorts score; D-KEFS_desc = Delis–Kaplan Executive Function System descriptive score.

**if p < 0.01.

In pwMS, a later time of neuropsychological testing is associated with reduced visuospatial processing performance, even after accounting for the influences of demographic and disease-related variables. There is no evidence that time-of-testing affects verbal fluency, learning/memory, executive function or processing speed. Contributions of visuospatial processing to BVMT-R and SDMT performance may explain the observed trends of reduced performance on these tests. ^{Reference Patel, Walker and Feinstein8} While cognizant that the cross-sectional retrospective study design prevents determination of causality, these findings are further examined.

Previous research suggests reduced cognitive performance in pwMS at later times of day ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3} and after a 4-hour strenuous task; ^{Reference Krupp and Elkins4} however, these small studies failed to control for confounding variables. ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3,Reference Krupp and Elkins4} By addressing these limitations, we clarify that testing time does not influence performance in most cognitive domains, aligning with the null results of a separate small investigation. ^{Reference Beatty, Goretti, Siracusa, Zipoli, Portaccio and Amato5} This may reassure clinicians and pwMS about whether an early or late appointment influences the evaluation of these common areas of cognitive dysfunction. ^{Reference Benedict, Amato, DeLuca and Geurts1} Furthermore, these null results emphasize the specific influence of time-of-testing on JOLO test performance.

To our knowledge, no prior study has examined whether time-of-testing influences visual function or visuospatial cognition in pwMS. Nevertheless, it is known that pwMS experience diurnal increases in fatigue ^{Reference Claros-Salinas, Bratzke, Greitemann, Nickisch, Ochs and Schröter3} and that fatigue may be associated with reduced oculomotor function (especially saccadic inefficiency) ^{Reference Zangemeister, Heesen, Röhr and Gold9} and altered cortico-striato-thalamo-cortical network functional connectivity. ^{Reference Kampaite, Gustafsson and York10} What remains unknown is whether there is a diurnal decline in the visuospatial pathway in pwMS. Future studies that explore how time-of-testing influences visuospatial cognition should incorporate concomitant measures of eye movements and neuroimaging.

Study limitations include its retrospective cross-sectional exploratory design and no measures of chronotypes, sleep parameters or medications other than DMTs. Additionally, the test administrator and the participant typically determined the testing time in collaboration. While common in clinical testing, work schedules or driving capacity may influence scheduling. Nonetheless, we demonstrated reduced visuospatial cognitive test performance at a later time-of-testing, but it had no influence on other cognitive domains in pwMS. Although clinically relevant, it is unknown whether a diurnal decline in visuospatial cognitive performance translates to occupational or driving performance differences. Our finding supports studies to address these questions.