Creating cognitive normative data for rural Indian population

Introduction Population-specific normative data are essential for accurately interpreting cognitive test performance, particularly in low-resource settings characterized by low literacy rates. In India, normative values for the Indian Council of Medical Research-Neurocognitive Tool Box (ICMR-NCTB) among older adults in rural communities are limited. The objectives are to generate education-stratified normative values for ICMR-NCTB cognitive domains and to evaluate their cross-sectional classification ability for identifying early cognitive impairment among high-risk older adults in rural India. Methods Baseline data were derived from the Strategic Multimodal Intervention in at-risk elderly Indians for prevention of dementia (SMRUTHI) cohort, a large, multi-site community-based study conducted across four rural regions of India. Neuropsychological assessments were administered to adults aged 55 years or older using the ICMR-NCTB battery. Normative centiles were estimated using quantile regression and the Box-Cox power exponential (BCPE) method to account for skewness and kurtosis in the test score distributions and to adjust for site effects. Receiver operating characteristic (ROC) analyses assessed the ability of individual tests and composite scores to distinguish normal cognition from early cognitive impairment, as defined by the Clinical Dementia Rating-Sum of Boxes (CDR-SOB). Results Of 3,917 individuals pre-screened, 3,000 cognitively normal and 175 cognitively impaired participants were included. Education demonstrated moderate associations with performance, whereas the demonstrated of age and gender were minimal. The normative data demonstrate a steady pattern across older age strata, with stable, domain-specific age-related changes rather than an unexpected decline. Literates performed better than their low-literate counterparts. ROC analyses indicated acceptable to good classification performance for several tests, particularly episodic memory measures and the composite z-score (area under the ROC curves up to 0.84). Empirically derived cutoffs approximated the 25th percentile, striking a balance between sensitivity and specificity for screening purposes. Discussion The large-scale, community-based study generated, age-education-stratified normative data for the ICMR-NCTB among high-risk older adults in rural India. The findings underscore the importance of population-specific, percentile-based normative values for early detection of cognitive impairment in settings with low literacy and high cultural diversity, while highlighting limitations related to ceiling effects in certain measures.