To date, there is a lack of validation studies on the MoCA-BJ scale conducted with a large community-based Chinese sample from both urban and rural regions. The present study provides evidence with population-based sample to demonstrate that the most widely used version of the MoCA in mainland China (i.e., MoCA-BJ) has good internal consistency and criterion-related validity in general, and is fairly reliable to differentiate MCI from normal aging and dementia. The MoCA-BJ as a brief screening tool for fast detection of MCI could be used as a reference by Chinese frontline clinicians.
Consistent with the original MoCA report
], among all the seven cognitive domains, the delayed recall task was most impaired among individuals with MCI in comparison to NC, and is most sensitive in the differentiation of MCI from NC. On the other hand, the most sensitive domains in discriminating MCI from dementia were the orientation and visuospatial/executive domains. These results indicate that episodic memory seems to decline the most at an early stage of MCI. With the progression of the disease, more basic cognitive functions, such as orientation, visuospatial/executive functioning, show larger declines than memory, since that memory impairment might have reached the “floor” at certain stage of the disease.
Detailed analyses suggest that some items may need further modifications. First, the adapted trail making task in the MoCA-BJ requires switching between Arabic numbers and Chinese sequencing-meaning characters, which is more difficult than the original task that requires switching between numbers and alphabet letters. This is because Chinese sequencing-meaning characters are lower in familiarity in China in comparison to English sequencing-meaning letters (i.e., English alphabets) in English-speaking countries, and thus the performance on this task did not differ between MCI and dementia participants. The trail making task used in the Hong Kong/Changsha version of the MoCA
] requires switching between numbers presented against different background colors. This task is more user-friendly for the illiterate or low-education population, and thus is considered to be more suitable for MCI screening for Chinese older adults. Second, the pictures used in the naming task and the words used in the delayed recall task were directly taken from the original English version without any culture-specific adaptations. “Rhinoceros” in the picture naming task is rarely seen in China and “velvet” and “daisy” in the delayed recall are also unfamiliar to Chinese older adults. Therefore, the naming task showed a low internal consistency and these two items (i.e., velvet & daisy) showed low memory scores. Third, the performance of one item of the sentence repetition task (i.e. “I only know that Liang Zhang is the one to help today”) was so poor that only twenty percent of the population could successfully repeat it, because the literal translation of this sentence from English to Chinese has resulted in an awkward sentence structure in Chinese. There is a need for some adaption to this sentence repetition item.
Among the demographic variables, age showed a negative correlation whereas education level showed a positive correlation with the MoCA-BJ scores. Among the three regions, older adults in the New Town area scored highest on the MoCA-BJ, presumably driven by their significantly higher education levels. In contrast, older adults in the Old Town region had the poorest performance, presumably due to their older age. After adjusting for the demographic confounding variables (i.e., age & education), the correlations between the regions and the MoCA-BJ performances disappeared, suggesting an equivalent applicability of the MoCA-BJ in both urban and rural populations.
The MoCA-BJ has a high sensitivity in the detection of MCI using the recommended cut-off score 26, whereas the specificity is fair, which may increase the probability of false positive diagnoses. This finding is consistent with the results from the two surveys introduced earlier which were conducted in Chengdu
] and Xuzhou
] respectively. These studies also indicate a high sensitivity (Chengdu: 98.11%; Xuzhou: 94.70%) and a fair or even low specificity (Chengdu: 26.72%; Xuzhou: 20.10%) of the MoCA-BJ scale when using the recommended 26 cut-off score to detect community-based MCI. In addition, similar to what we found in the current study, researchers in the Chengdu study also recommended a cut-off score of 21/22, even though they did not report the corresponding optimal sensitivity and specificity. However, in contrast to Lu’s study
], our data did not support the MoCA-BJ as a very valid screening tool for MCI detection in the Chinese population. This discrepancy between the current study and Lu’s study may be due to the following reasons. First, the two studies used slightly different assessment tools. Although Lu et al. claimed that they administered the Beijing version of the MoCA in their study; the items used in their study were actually a combination of the items in the Beijing version and Changsha version of the MoCA. As noted in their description of the assessment, Lu and colleagues changed at least one item- velvet
in the delayed recall task to silk
(as in the Changsha version). This modification could improve the diagnostic accuracy due to this culture-specific modification. Second, the two studies involved different samples. The prevalence of MCI was about 20.0% (1687/8411) in Lu et al.’s study but only about 11.5% (115/1001) in the current study. Our sample is a closer match to the pooled MCI prevalence of 12.7% in the Chinese population as reported in a recent meta-analysis
]. This suggests that our sample could be more representative than Lu et al.’s. Third, the MCI diagnostic criteria in the two studies were different. Their diagnosis was mainly based on the CDR scores, whereas ours was made upon a more comprehensive and relatively stricter criteria, under which a diagnosis of MCI was determined not only based on the CDR scores, but also on the GDS scores, and supplemented by the NPI and HIS scores, and more importantly, on the psychiatrists’ clinical experiences. These differences in diagnostic criteria might also have contributed to the differences in the prevalence of MCI in the two studies. In summary, all these differences mentioned above could lead to the inconsistent conclusions between Lu et al.’s study and ours.
It should also be noted that the MoCA was not demonstrated to be superior to the MMSE in current study as other studies did
] in MCI identification, while the MoCA-BJ was indeed found to be a slightly more sensitive screening tool than the MMSE in a sub-sample with higher education (Figure
). The relatively low education level of Chinese older adults (mean education years
5.45) and the items with cultural and linguistic differences mentioned above could have contributed to these inconsistent findings. Due to the cultural and language differences and the low education level of Chinese elderly, some items that are reliably administered in Western countries to differentiate individuals from different diagnostic groups may not apply to the Chinese population
This study has several limitations. First, the given sample sizes of NC, MCI and dementia groups were different. Moreover, the number of participants in the dementia group was very small, which might have compromised the statistical ability of our between-group comparisons
]. In a similar vein, the small sample size in the rural area might also have weakened the conclusion drawn from this sample. In addition, the study was only conducted in Beijing. China is a large country with considerable regional variability in several sociocultural domains, such as economy, custom, climate, and diet, which could influence the performance of local residents on the neuropsychological tests. We recommend future studies to recruit participants from more diverse regions of the country.