To apply the new IC-CoDE cognitive diagnostic taxonomy (Norman et al., 2020) to a large cohort of people with temporal lobe epilepsy (TLE) in India. The IC-CoDE taxonomy of cognitive diagnoses for 1,409 Englishspeaking adults with TLE from seven epilepsy centres in the U.S. has been published (McDonald et al., 2022). Initial results suggest that the IC-CoDE produces stable cognitive phenotypes across centres; however, its international applicability, including the suggested impairment cut-off needs to be considered across cultures and languages to avoid misclassification. The aim of this study was to apply the IC-CoDE to a population, outside of the U.S., diverse in language representation (i.e., bi/multi-lingual), assessment tools, normative data, and educational and cultural backgrounds to determine whether the same cognitive phenotypes and their relative frequencies would emerge.

Data from 549 adults with TLE (mean age=27.14 (8.04), 60.47% males) from a tertiary referral hospital in Mumbai, India who had undergone a comprehensive neuropsychological evaluation (minimum two tests in at least 4 of the 5 cognitive domains: memory, language, executive function, attention/processing speed and visuospatial) were analysed using the ICCoDE criteria. The base rate of impairment for individual tests was calculated using a cutoff of 1.5 standard deviations (S.D.) below the normative mean. The cognitive diagnostic criteria were applied, and the distribution and base rate of cognitive phenotypes was compared to the published taxonomy data from the U.S. (McDonald et al., 2022).

In comparison to the U.S. cohort, the India group was relatively younger, lower in the education level, had a younger age at seizure onset and a shorter duration of the epilepsy. Application of the IC-CoDE taxonomy using a 1.5 S.D. cutoff revealed an Intact cognitive profile in 48% of patients, Single Domain impairment in 32%, Bi Domain impairment in 15% and Generalised impairment in 5%. These findings were mostly comparable to percentages reported in the U.S. cohorts with Intact profile (47%; c2= 0.158, p=0.690), Single Domain (29%; c2= 46.26, p<0.01), Bi Domain (16%; c2= 0.298, p=0.585) and Generalised (8%; c2= 5.347, p=0.021) impairment. However, the most common impairment in the Single Domain group for the bi/multilingual India population was Memory (38%) followed by Attention (20%) and then Language (13%), diverging from the distribution in the U.S. data with maximum impairment in Language (49%) followed by Memory (32%) in the Single Domain Group.

These findings demonstrate that the IC-CoDE can be applied internationally, and the broad taxonomy of cognitive diagnosis holds even in a culturally, linguistically diverse population. Differences in rates of impairments across specific domains emerged with language relatively preserved in the India bi/multilingual population, and memory more frequently impaired than observed in the multi-centre U.S. sample. These findings may reflect differences in demographics, rates of bi/multilingualism, normative data, language tools, or underlying neuropathology, which should be further explored to determine their impact on cognitive profiles.