Dyadic self- and informant-reported cognitive concern and subsequent MCI/AD diagnosis in ADNI: an exploratory prognostic modelling study
In an internally cross-validated ADNI cohort, adding a baseline dyadic mean of self- and informant-reported cognitive concerns to conventional predictors yielded only a small and statistically uncertain improvement in predicting progression to MCI or AD dementia, while the self-informant gap provided no additional predictive value.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
The human mind is a complex landscape, and for many older adults, the first sign that this landscape is shifting is not a dramatic failure, but a quiet worry. They might feel that their memory is not as sharp as it once was, or that finding the right word takes a little longer. This feeling, known as subjective cognitive decline, is a common experience. Sometimes, the person noticing these changes is the only one who sees them; other times, a spouse, child, or close friend notices the same struggles. When both the individual and someone close to them agree that something is off, it creates a powerful, shared picture of concern. Scientists have long wondered if this shared worry is a stronger warning sign of future memory loss than the individual's worry alone. Specifically, they wanted to know if comparing what a person says about their own mind with what a loved one says about that same person could help predict who will develop mild cognitive impairment or Alzheimer's disease dementia, even when standard medical tests look normal.
To answer this, researchers turned to a massive, long-running project called the Alzheimer's Disease Neuroimaging Initiative, which has gathered detailed health data from thousands of people over many years. They focused on a specific group: 1,762 adults who were considered cognitively normal at the start of the study. These participants did not have a diagnosis of memory loss or dementia. The researchers wanted to see if, at the very beginning of the study, the combination of self-reported and friend-reported concerns about daily thinking skills could predict who would later be diagnosed with a memory problem. They used a specific questionnaire called the Everyday Cognition scale, which asks about real-world tasks like managing finances, remembering appointments, or following a recipe. The team calculated two things from these answers: the average level of concern shared by the person and their friend, and the difference, or gap, between what the person thought and what the friend thought.
The study was designed with extreme care to avoid tricking itself. The researchers split the data into different groups, building their prediction models on some groups and testing them on others they had never seen before. This method ensures that the results are not just a lucky guess based on the specific people in the study. They compared a standard prediction model, which used well-known factors like age, education, genetic risk, and brain imaging results, against models that added the shared worry scores. The standard model was already quite good at predicting who would develop memory problems, correctly ranking the risk for about 68 out of 100 pairs of people where one developed a problem and the other did not. When the researchers added the average of the self and friend reports to this standard model, the prediction ability improved only slightly, moving to about 69 out of 100. This tiny improvement was so small that it could easily have happened by chance. Furthermore, adding the difference between the person's and the friend's answers provided no extra help at all.
The results suggest that while shared worry is indeed linked to future memory issues, simply adding a single snapshot of this worry to a model that already includes genetics and brain scans does not make the prediction significantly better. The researchers found that the data was missing for a large portion of the participants; only about 27% of the people in the study had both a self-report and a friend-report available. This missing data means the findings are based on a specific, perhaps more engaged, group of people and may not apply to everyone. The study also noted that the way the questionnaire was administered changed over the years of the project, which adds another layer of uncertainty. Ultimately, the author concludes that relying on this specific type of shared worry report, by itself, is not yet a reliable tool for doctors to use in deciding who will develop Alzheimer's disease. The small, uncertain gain in prediction power, combined with the high rate of missing information, means that this approach is not ready for clinical use. The study does not rule out the value of these reports entirely, but it does show that they do not offer a decisive advantage over the conventional tools already in use.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.