Analysis of Trending Topics in Breast Cancer Articles From an Altmetric Perspective

Background It is widely known that social media has an impact on politics and the economy. The Altmetric Attention Score (AAS) is a new Web-based metric that was recently developed for use in the scientific field. The objective of this study was to assess which recent studies on the topic of breast cancer received the most attention from the general public. Methodology An Altmetric Explorer search was performed on January 7, 2022, to extract the following information: journal name, impact factor (IF), year of publication, article topic, article type, and level of evidence. Results The journal that published articles that received the most attention on social media was the New England Journal of Medicine (n = 8). All of the articles were published in journals in the highest IF quartile. The most frequent top three subjects in the top 50 articles were “treatment and management,” “risk factors for breast cancer,” and “breast cancer screening.” The number of articles with a level of evidence of 1, 2, 3, and 4 was 12, 17, 17, and 4, respectively. The correlation between AAS and citation was not significant. Conclusions The AAS seems to be a more reliable assessment of public perception of breast cancer. We propose that combining the AAS and traditional metrics may provide a more detailed description of scientific research output.


Introduction
The number of times published articles are cited is a common metric for determining a journal's impact. Researchers use citation analysis (bibliometric analysis) to identify the most valuable publications in their fields, such as oncology [1], ophthalmology [2], and cardiology [3]. Eugene Garfield, the founder of the Institute of Scientific Information, was the first to apply this method to scientific publications in the 1970s [4]. Although citation analysis is a frequently used tool, its most important disadvantage is that the length of time that has passed since publication may have an impact on the total number of citations (length time bias). Additionally, citations only reflect the impact on the scientific community but do not show the impact on politicians, patients, or the general public.
Over the last few years, the importance of social media platforms in the promotion, dissemination, and presentation of medical literature has been markedly enhanced. Fast and dynamic assessments of the influence on social media platforms have been facilitated by a new Web-based metric (Altmetrics). The Altmetric Attention Score (AAS) is a quantitative and qualitative measurement that complements standard citation-based measures. It analyzes the interactions of academics, scholars, and scientists as captured by reference management tools and social media platforms such as Facebook, Twitter, LinkedIn, and blogs. The purpose of this research was to determine which recent breast cancer papers had the highest AAS.

Search engine
Altmetric Explorer (London, UK) is an internet-based tool that searches various sources of research output (listed in Table 1) to produce the most relevant and up-to-date picture of the different types of online activity and discussion [5].  The AAS is generated automatically using a weighted count of all the attention given to research output. The Altmetric donut and the AAS are designed to make determining how much and what type of attention a particular research product has received easier [6]. Volume, sources, and authors are the three main components. The Altmetric donut is colored to represent a different source of attention ( Figure 1) [6].

Search strategy
An Altmetric Explorer search was performed on January 7, 2022. The term "breast cancer" was searched on the Altmetric Explorer ( Figure 2). To clarify the AAS for the breast cancer field, the top 50 articles with the highest AAS were identified by excluding the articles that were not directly related to breast cancer. The data were then further evaluated by examining the title, journal name, date of publication, study type, and topic. On the same day, the number of citations for each article was acquired from the Thomson Reuters Web of Science citation-indexing database. Because altmetrics started to obtain data in 2011, in the search list, publications dated before 2011 were not included. The level of evidence of the top 100 cited articles was detected in accordance with the Scottish Intercollegiate Guidelines Network (SIGN) criteria [7]. The level of evidence was assigned to studies based on the methodological quality of their design, validity, and applicability to patient care.

FIGURE 2: Flowchart illustrating the article allocation process.
Continuous variables were defined using median and interquartile range (IQR), whereas categorical variables were defined using percentages. The Kruskal-Wallis test was applied to compare three or more groups because the data were not normally distributed. The Spearman correlation coefficient was used for assessing the correlation between AAS, citations, number of years post-publication. P < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS Statistics for Windows, version 21.0 (IBM Corp., Armonk, NY, USA).

Results
We identified 163,028 articles regarding breast cancer using Altmetric Explorer. Table 2 summarizes the AAS, journal name, publication year, and citation count of the top 50 publications. Because altmetrics began collecting data in 2011, no articles prior to 2011 are listed in Table 2  The highest and lowest AASs in the top 50 article list were 3,619 and 879, respectively, while the median AAS was 1,063. The highest numbers of articles were published in 2017 (n = 9) and 2020 (n = 9). The top 50 articles were published in 28 different journals; eight articles were published in the New England Journal of Medicine, followed by the Journal of the American Medical Association (JAMA) with four articles ( Table 3).
All of the articles were published in journals in the highest impact factor (IF) quartile.

Rank Journal name
Number of articles Impact factor* Impact factor ranking** H-index**  The top-ranked article (#1 in Table 2) reported a deep learning model (artificial intelligence) for identifying breast cancer in screening mammograms. In that study, the authors presented an artificial intelligence system that outperforms radiologists on a clinically relevant breast cancer identification task. Classification according to subject categories is shown in Table 4. The most frequent top three subjects in the top 50 articles were "treatment and management" (n = 12), "risk factors for breast cancer" (n = 11), and "breast cancer screening" (n = 11). The most frequently mentioned subject in the treatment and management category (#2, #9, #22, and #49) was the role of multi-gene tests in identifying recurrence and adjuvant treatment in operated patients with early-stage breast cancer. There were two publications (#4 and #17) in the treatment and management category mentioning the role of tumor-infiltrating lymphocytes in the treatment of breast cancer. One of these articles (#4) was a case report which is the only case report included in the top 50 list. It is worth noting that half of the top 10 articles were about treatment and management. There were 11 articles concerning the risk factors of breast cancer, while the leading risk factors drawing interest were hair care product use (#3 and #30), menopausal hormone therapy (#12 and #27), and specific dairy foods (#37 and #42). In the top 50 list, 11 articles were noted regarding mammography which has been in use for many years for breast cancer screening. Mammography screening effectiveness (#7, #34, and #47) and impact on mortality (#15, #29, and #39) were investigated in more than half of the articles regarding screening. Only one article in the top 50 list was about coronavirus disease 2019 (COVID-19), and the subject in this article (#41) was the impact of COVID-19 on screening test volume. In April 2020, a 96% decrease was reported in breast cancer screening tests. The other subject categories included in the top 50 list were the pathogenesis of breast cancer (n = 7), breast cancer prevention (n = 3), quality of life (n = 2), diagnosis (n = 2), psycho-oncology (n = 1), and breast cancer statistics (n = 1) ( Table 4).  The designs and evidence levels of the articles in the top 50 list are shown in Table 5. As shown in the table, there were 43 articles in the clinical research category, four in the experimental animal studies category, and three in the tumor cell culture studies category. In the top 50 list, the number of articles with a level of evidence of 1, 2, 3, and 4 was 12, 17, 17, and 4, respectively. The top five articles with the highest number of citations are ranked at #28, #40, #35, #22, and #47 in Table 2. With 1,562 citations, the most highly cited publication (#28) was an article titled "Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer" published in the New England Journal of Medicine in 2018.

Clinical research
Meta-analysis 1 1 Randomized controlled trial 1 11 Prospective cohort study 2 11 Observational descriptive study 3 1 Cross-sectional study 3 5 Retrospective cohort study 3 7 Retrospective comparative study 3 2 Case report 3 1 Expert opinion (editorial or letter) 4 3 Expert committee report 4 1 Experimental animal study Prospective comparative study 2 3 Observational study 3 1 Tumor cell culture study Prospective comparative study 2 3 According to the top 50 list, the correlation between AAS and citation was not significant (r = 0.11, p = 0.244). Additionally, there was no correlation between AAS and the number of years post-publication (r = -0.09, p = 0.506). As expected, there was a significant positive correlation between citation and the number of years post-publication (r = 0.558, p < 0.001). According to the stratification of the top 50 list articles based on the level of evidence, the median AAS was 1,112.5 (IQR = 929.5-1,444) for level 1, 1,052 (IQR = 929.5-1,456) for level 2, 1,012 (IQR = 932-1,475) for level 3, and 1,167 (IQR = 921.5-1,334.5) for level 4, without any significant difference between the groups (p = 0.994).

Discussion
Cancer is a major disease burden worldwide. Breast cancer is the most common cancer diagnosed among US women (excluding skin cancers) and is the second leading cause of cancer death among women after lung cancer [8]. Breast cancer mortality has declined significantly in recent years. This is in line with increased early detection rates as a result of the implementation of national population-based mammography screening programs and more effective adjuvant therapy [9]. This study revealed the aspects regarding which the academic world and society interacted more frequently on social platforms about breast cancer while uncovering the trending publications. To our knowledge, this study is the first to assess the online attention that articles published in the field of breast cancer received.
The journals that published articles receiving the most attention on social media were the New England Journal of Medicine (n = 8) and JAMA (n = 4). These two journals are not specifically oncology journals. Table  3 shows that only 34% ( Table 4 demonstrates that many different aspects of breast cancer have been discussed in the top 50 list. A review of this table reveals that articles on diet and cosmetics in breast cancer come into prominence. While the association between diet and the effectiveness and toxicity of chemotherapy was investigated in two articles [11,12], diet and the risk of breast cancer or the prevention of breast cancer were investigated in eight articles. Two articles discussed the association between hair care product use and the risk of breast cancer [13,14]. Moreover, two articles evaluated the efficiency of the use of the scalp cooling device in chemotherapy for breast cancer in reducing alopecia [15,16]. As a result, the association between breast cancer and diet and cosmetics attracts public attention on social media. The top 50 list did not include any articles regarding breast cancer surgery which is the primary treatment for early-stage breast cancer and has been improving with more effective techniques over the years [17,18]. Furthermore, there were no scientific articles in the top 50 list involving major treatment agents such as CDK4/6 inhibitors or HER-2 blockage (trastuzumab, pertuzumab), which are among the revolutionary treatments that significantly prolonged survival in metastatic breast cancer [19,20]. This suggests that there are some perspective differences between academic papers and general social media. Clinicians are interested in various topics such as the pathogenesis of the disease, treatment guidelines, new diagnostic tools, surgery techniques, and new drugs that members of the general public are not interested in.
The data in Table 5 provide notable information about the level of evidence of articles on breast cancer that attracted more attention on social media. In the top 50 list, the number of articles with the lowest level of evidence (SIGN Level 4) was only four. Articles with levels of evidence of 1 and 2 which were analytical trials constituted 58% of the entire group. Although the level of evidence is a significant parameter for scientists, society may not have sufficient knowledge about the level of evidence or design of scientific publications. This may cause the information with a low level of evidence to become popular and get disseminated on social media. A classification of the articles in the top 50 list according to the four levels of evidence did not produce any statistically significant difference between the groups in terms of median AASs. Additionally, no significant correlation was noted between the numbers of citations and AASs. Similar to our study, a study conducted by Celik et al. involving a correlation analysis of citations and AAS for the top 50 articles on "cancer" did not demonstrate any correlation [21]. Therefore, it can be concluded that the top 50 list includes higher number of articles on breast cancer with higher levels of evidence. The lack of any correlation between the number of citations and AAS once again emphasizes the fact that altmetric analysis assesses different aspects of articles compared to traditional citation analysis and that the interest in scientific studies may differ among the academy and the social media.
The limitation of this study is that the AAS is a relatively new tool. When assessing study results, the absence of AAS in studies published prior to 2011 should be taken into account. The study's inclusion of 50 articles is another limitation. Instead of 50 articles, choosing 100 or 200 would have enhanced the study's power.

Conclusions
One of the novel measures of citations in social media is AAS analysis. This study provides useful information about the impact of the top 50 breast cancer articles in both academia and social media. The AAS seems to be a more reliable assessment of public perception of breast cancer. Finally, we propose that combining the AAS and traditional metrics may provide a more detailed description of scientific research output.

Additional Information Disclosures
Human subjects: All authors have confirmed that this study did not involve human participants or tissue. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue.

Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.