Challenges in multimorbidity and cognitive function studies: A call for rigorous analysis and reporting

Abstract

This letter is in response to “Multimorbidity measures associated with cognitive function among community-dwelling older Chinese adults” by Wang and colleagues.¹

In their article, Wang et al. explored, using longitudinal survey data, the association of multimorbidity with cognitive function among older Chinese adults. We write to highlight key points for further exploration. First, exploratory factor analysis was used to examine the relationship among the 14 individual chronic conditions, but the Kaiser–Meyer–Olkin value of 0.67 can only indicate a mediocre pattern of correlations.² Addtionally, more than half of the factor loading were below 0.5, which is considerded as a threshold value with practical significance.

As the authors note, most of these conditions were self-reported by participants except for components of sensory pattern; this may weaken the correlation between other patterns and cognitive function and likely explain the strongest association with cognitive function among sensory pattern both at baseline and the first follow-up; only sensory pattern showed similar results in both sexes compared with the total population may also be attributed to this. Similarly, participants with more chronic conditions were more likely to live in urban areas, possible explanation is that urban areas have relatively abundant medical resources compared to rural areas. These information biases caused by different disease detection rates may lead to a deviation from reality.³

Lack of death information is another issue. Although the study is still ongoing, the mortality data in a longitudinal study can be gradually updated during the follow-up period and analyzed based on existing data. Completely ignoring the impact of mortality may lead to bias in the risk assessment of cognitive decline, especially in the elderly population where mortality is a very common and significant competitive risk. Besides, the proportion of participants with the three multimorbidity patterns was not provided in the article, which directly affects the explanatory power of statistical analysis.⁴

Overall, this study provided interesting information on the multimorbidity measures and their association with cognitive performance in Chinese community-dwelling older adults. Nonetheless, imperfect exploratory factor analysis, information bias, and incomplete reporting impact its reliability and validity.