Public Interest in Cognitive Impairment: An Analysis of the Top 50 Articles on Cognitive Impairment on Altmetric

Background As the average life expectancy continues to increase, interest in cognitive impairment is increasing. Nowadays, as social media expands its reach, academic research is spreading through social media, changing the way and speed by which research is propagated and also who consumes this content. Therefore, using Altmetric, a new web-based set of metrics that analyzes the impact of content on social media platforms, we investigated the characteristics of influential research articles on the topic of cognitive impairment in social media. Methods An Altmetric Explorer search was performed on May 25, 2018, to extract the following information: (i) journal name, (ii) journal impact factor (IF), (iii) year of publication, (iv) article topic, (v) article type, and (vi) cognitive impairment subtype. Results The journal “Neurology” was the most cited journal for cognitive impairment articles shared on social media. Among the various types of cognitive impairment, most articles were related to dementia (all subtypes), Alzheimer's disease, and aging. The most common article type was original scientific paper, especially cohort study. The most popular topic was the identification of protective or risk factors for cognitive impairment. Conclusion The characteristics of articles with a high Altmetric Attention Score were somewhat different from those of articles with a high number of traditional citations. Social media had the disadvantage that it was difficult to verify the authenticity of the primary source in question, but the advantage was that it could immediately determine the trends regarding how information about that source was being shared and consumed. Therefore, it may be advisable to use Altmetric analysis in combination with traditional methods of evaluating the research articles to understand the dissemination of scientific research and to direct future research.


Introduction
As social media has evolved, academic research has spread rapidly to the public through this channel. As the life span of mankind has increased, interest in maintaining a healthy later life has increased, with a focus on cognitive function being central to this interest. Social media, therefore, is replete with information about cognitive function. Altmetrics is a new web-based set of metrics that analyzes the impact of social media platforms such as Twitter and Facebook [1].
e Altmetric Attention Score (AAS) is a weighted score, calculated using differential weights for multiple factors depending on the amount of visibility they promote [2]. Research articles with a high AAS meant that they got more attention from social media. Traditionally, citation-based metrics such as impact factor and h-index have been used to determine the impact of articles [3]. Traditional citations take a considerable amount of time to accumulate after publication, whereas AAS is updated as soon as attention occurs online, so AAS could be used as an ancillary tool to evaluate the influence of the research articles, immediately. erefore, we would like to investigate which of the cognitive impairment-related studies received high AAS and consider the impact of these trends.

Search Engine.
Altmetric Explorer (Altmetric, London, UK) is a web-based application that can be used to search the following sources of research output, to yield the mostrelevant and up-to-date picture of the following types of online activity and discussion from public policy documents, blogs, mainstream media, citations data, online reference managers, research highlights, postpublication peer-review platforms, social networks, Wikipedia, Open Syllabus Project, multimedia and other online platforms, and patents [4].
We followed the methods of Kim et al. [5]. e AAS and the Altmetric donut are designed to make it easy to identify how much and what type of attention a particular research output has received [6]. e AAS is automatically calculated using an automated algorithm created by the company Altmetric, based on the weighted count of all the attention a research output has received. It is based on three main factors: volume, sources, and authors. Each color of the Altmetric donut represents a different source of attention [6].

Search
Strategy. An Altmetric Explorer search was performed on May 25, 2018, for articles published in the 91 journals included in the 2015 InCites ™ Journal Citation Report ® grouped under the subject categories of clinical neurology, neuroimaging, neurosciences, radiology, nuclear medicine and medical imaging, surgery, and general and internal medicine.
Two researchers (Y. J. Kim and Y. Kim) read the full papers of the top 50 articles and extracted the following information by consensus: (i) journal name, (ii) journal impact factor (IF), (iii) year of publication, (iv) article topic, (v) article type, and (vi) cognitive impairment subtype.
To clarify the AAS for the cognitive impairment field, we excluded articles that suggested composite outcomes including psychiatric disorder. If the outcomes were treated separately, the article was included in this study.

Results
Altmetric Explorer identified 43,410 articles as being referred to online. e AAS, journal name, impact factor (IF) ranking, publication year, and number of citations of the top 50 articles are summarized in Table 1. Because Altmetrics started collecting data only during the second half of 2011, the articles before 2010 were rarely listed. e number of papers on this list peaked in 2017 (n � 20). Eight articles originated from the journal Neurology followed by JAMA Internal Medicine and JAMA, each with five articles. All but one of the articles were published in journals in the highest IF quartile (Table 2). Regarding cognitive impairment subtypes, 20 articles were on dementia, 14 were on Alzheimer's disease, 10 were on aging, and 2 were on chronic traumatic encephalopathy (Table 3). Regarding article types, 40 articles were original scientific papers, 4 were related to systematic reviews and meta-analyses, 2 were related to guidelines and advisory documents, 3 were related to reviews, and 1 was related to an editorial (Table 4). e most common topic was risk and protective factors for cognitive impairment (n � 27) ( Table 5). In particular, as reports of cohort studies and editorials were both popular in social media, people were interested in the result that sugarsweetened beverages increased the incidence of dementia (#3 and #41 in Table 1) [7,8]. However, green leafy vegetables and a Mediterranean diet had a protective effect against cognitive decline (#5 and #9) [9,10]. MIND diet was also associated with decreased risk of Alzheimer's disease [11]. Glucose and vitamin D levels were associated with the risk of cognitive decline (#17 and #18) [12,13]. Alcohol intake and marijuana use were associated with cognitive health. Marijuana decreased perfusion of hippocampus [14]. However, there were discrepancies in opinion on alcohol consumption. In a cohort study published in 2014, excessive alcohol consumption of more than the average daily amount of 36 g in midlife males was associated with faster cognitive decline (#28) [15], whereas in a study published in 2017, it was found that regular, moderate alcohol intake enhanced cognitively healthy longevity (#21) [16]. Leisure activity and midlife cardiovascular fitness were also associated with the risk of dementia. Some studies reported that comorbid conditions such as hearing loss (#19) [17], poor sleep (#26) [18], and chronic pain (#46) [19] deteriorated cognitive health. Environmental factors such as living near a major road (#4) [20] and lead exposure (#32) [21] affected cognitive health. Furthermore, medications such as proton pump inhibitors and anticholinergics were related to cognitive function. Over-the-counter supplement use was also effective against cognitive decline (#35) [22]. Bilingualism (#12, #44, and #47) [23][24][25] and marriage type (#50) [26] could also affect cognitive health.

Discussion
Here, we summarized the top 50 articles using AAS. e mean AAS was 1171.6, and the mean of the number of citations was 109.5. Articles were published the most in 2017, and the journal "Neurology" was the most popular originated journal. Among the types of cognitive impairments, most articles were related to dementia, Alzheimer's disease, and aging. e most common article type was original scientific paper, especially cohort study. e most popular topic was the protective or risk factor.
Most of the articles were published in 2017, likely because we performed the Altmetric Explorer search on May 25, 2018. e same method used in the article written by Kim et al., co-author of this article, was used here [5]. AAS was sensitive to the latest content, and recent publications tended to receive higher AAS. For example, the Altmetric top 20 list of Parkinson's disease research published in 2017 did not include any papers that were reported to be the most   [27]. However, in our report, two articles, about leisure activities and bilingualism as protective factors, were included in the top 50 list even though they were published before 2011. erefore, it was suggested that the influence of important articles through   social media might be weakened, but it did not disappear even after some time. e Altmetric score generally reflects the interest of the lay public, not the scholar, because it measures all social media sources. In the past, academic research articles were typically the exclusive domain of scholars, but nowadays, because of the development of media, research articles are consumed by the general public as well as scholars. So far, the research article was a means of communication for researchers and the degree of citation was an important indicator for evaluating the quality of the articles. However, due to the development of social media, expertise on topics such as medicine and disease was widely publicly available and research that received public attention could be more readily supported. Nonetheless, since public and professional interests differ, citation index and social media indicators were not always in agreement. is analysis showed that social media audiences were most interested in modifiable risk factors related to cognitive impairment. Diet and nutrients, social factors, alcohol and drugs, and medications investigated in articles with high AAS were risk factors that could be changed through behavioral modification. In addition, people were also interested in treatment and prevention, and in particular, they were more concerned with nonpharmacologic management than pharmacologic management among treatment. In the absence of clear pharmacological treatment for dementia, it seemed that there was a relatively greater interest in factors that could help improve cognitive function, as well as prevention and nonpharmacological treatment. e top-ranked article also reported about chronic traumatic encephalopathy, which could be avoided if exposure to trauma was minimized.
Disease groups with higher AAS were all subtypes of dementia, Alzheimer's disease, and aging. It was also worth noting that interest in aging without dementia was high. is could be related to the higher interest in prevention than treatment. As there was no disease-modifying treatment for dementia, it seemed that there was more interest in how to maintain cognitive function in healthy elderly individuals. Indeed, even among scholars, the development of preventive methods to reduce cognitive impairment through lifestyle changes has been emphasized over treatment. e paper by Kivipelto and collaborators was a randomized controlled study investigating whether an intervention of diet, exercise, cognitive training, and vascular risk monitoring could prevent cognitive decline of normal elderly individuals. is paper, published in 2012 in e Lancet, was one of the most cited in the Altmetric top 50 article list for cognitive impairment (939 AAS, 604 citations) [28].
On the other hand, there was less interest in professional contents such as pathomechanism or biomarkers of dementia. For example, in Alzheimer's disease, amyloid and tau proteins are related to the pathomechanisms of the disease. As a result, the technology of detecting amyloid and tau protein in CSF or visualization of amyloid and tau using brain imaging has been highly studied [29]. However, none of these studies were included in the Altmetric top 50 list. ese studies were not less important because of the low AAS of pathomechanisms or biomarker-related studies. Since the research that interests the public was not necessarily the most relevant and the Altmetric score based on social media is a reflection of the public's interest, it is difficult for AAS to accurately reflect the importance or quality of research. In addition, since social media lacks the ability to verify whether the information was real or fake [30], the possibility of misleading research spreading through social media could lead to a high AAS, despite lack of credibility or relevance. erefore, it could be dangerous to evaluate the value of academic publications by AAS alone.   Nevertheless, the world is rapidly evolving to use the web to share information rather than more traditional media, and because many journals are converting to open access, the information inherent in social media and other web platforms cannot be ignored. As an additional tool for evaluating a paper or topic, using AAS in combination with traditional methods can help better understand the impact of scientific findings.

Conclusions
We reviewed popular articles on cognitive impairment using Altmetric analysis. In the analysis, the most salient characteristics of top articles of interest were those that were most recently published, cohort studies, and those published in the journal Neurology. e protective or risk factors associated with cognitive impairment was the topic of greatest interest. In order to understand the flow of scientific research, we suggest using Altmetric analysis as an alternative tool, along with traditional tools for evaluating article impact.

Data Availability
e data used to support the findings of this study were supplied by "Advanced Search" in Altmetric Explorer (https://www.altmetric.com/explorer, Altmetric LLP, London, UK).

Conflicts of Interest
e authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.