Montreal Cognitive Assessment’s auditory items (MoCA-22): Normative data and reliable change indices

Our objective was to establish normative data and reliable change indices (RCI) for the Montreal Cognitive Assessment’s auditory items (MoCA-22). 4,935 cognitively unimpaired participants were administered the MoCA during an in-person visit to an Alzheimer’s Disease Research Center (<i>M_age</i> = 67.9, <i>M_education</i> = 16.2, 65.8% women, 75.9% non-Hispanic-White), with 2,319 unimpaired participants returning for follow-up. Normative values and cutoffs were developed using demographic predictions from ordinary and quantile regression. Test-retest reliability was calculated using Spearman and intraclass correlations. RCI values were calculated using Chelune and colleagues’ (1993) formula. Education, age, and sex were all statistically related to MoCA-22 scores, with education having the strongest relationship. Notably, these relationships were not consistent across MoCA-22 quantiles, with education becoming more important and sex becoming less important for predicting low scores. These models were integrated into a calculator for deriving normative scores for an individual case. Furthermore, there was adequate-to-good test-retest reliability (ϱ = 0.56 95% CI [.54, .59]; ICC = 0.75, 95% CI [.73, .77]) and changes of at least 2-3 points are necessary to identify reliable change at 1-3-year follow-up. These findings add to the literature regarding utility of the MoCA-22 in the cognitive screening of older adults.本研究旨在构建蒙特利尔认知评估量表（Montreal Cognitive Assessment）听觉条目（MoCA-22）的常模数据与可靠变化指数（Reliable Change Indices, RCI）。本研究共纳入4935名认知正常受试者，在阿尔茨海默病研究中心（Alzheimer’s Disease Research Center）进行线下访视时完成MoCA评估，其中平均年龄（M_age）为67.9岁，平均受教育年限（M_education）为16.2年，女性占比65.8%，非西班牙裔白人占比75.9%；另有2319名认知正常受试者完成了后续随访。本研究采用普通回归与分位数回归的人口学预测模型，构建了MoCA-22的常模值与截断值。重测信度通过斯皮尔曼相关系数与组内相关系数（intraclass correlation, ICC）进行计算。可靠变化指数的计算采用Chelune及其团队1993年提出的公式。受教育年限、年龄与性别均与MoCA-22得分存在统计学关联，其中受教育年限的关联强度最高。值得注意的是，上述关联在MoCA-22的不同分位数区间并不一致：在预测低得分时，受教育年限的预测权重逐渐升高，而性别的预测权重逐渐降低。本研究将上述模型整合为一款计算器，可用于推导个体受试者的MoCA-22常模得分。此外，本研究得到的重测信度处于尚可至良好水平（斯皮尔曼相关系数ρ=0.56，95%置信区间[0.54, 0.59]；组内相关系数ICC=0.75，95%置信区间[0.73, 0.77]）；在1至3年的随访周期中，受试者得分需至少变化2~3分，方可判定为存在可靠变化。本研究结果进一步丰富了关于MoCA-22在老年人群认知筛查中应用价值的相关文献。