博文
健康素养研究的概念结构和主题趋势（英文全文）
|||
健康素养研究的概念结构和主题趋势
一种多视角文档共被引分析"（The Conceptual Structures and Thematic Trends of Health Literacy Research：A Multiple-Perspective of Document Co-Citation Analysis）。这一思路主要来源于陈超美老师2010年刚刚在JASIST（Journal of the American Society for Information Science and Technology）上发表的一篇论文，引入了“多视角共被引分析”这一概念。由于之前的共被引分析在标识聚类的时候总是以高被引文献（cited items）为来源，而“ 多视角共被引分析 ”强调不仅以高被引文献为来源，还要同时考虑这一共被引聚类的施引文献（citing items to each cluster）。因为施引文献在更大程度上表征了研究前沿。在该论文中，以“健康素养”领域的文献为对象，同时应用这两种视角进行了聚类分析，并确定了类的标识（前者通过人工阅读、分析高被引文献，后者通过citespace软件自动生成，从施引文献的title中用LLR算法抽取），结果发现两种视角得到的聚类标识基本上一致，但是后者往往比前者更具体，更细化，这一点和陈超美的研究结论相一致。最后，在研究健康素养这一领域的概念结构的时候，发现“精神健康素养”和“功能性健康素养”是两大相独立的研究领域。
The Conceptual Structures and Thematic Trends of Health Literacy Research:A Multiple-Perspective of Document Co-Citation Analysis
Du Jian1, Xu Peiyang1, Zhang Shijing2, Zhang Bin1
Du Jian, Xu Peiyang, Zhang Bin
Institute of Medical Information & Library
Chinese Academy of Medical Science & Peking Union Medical College
Beijing,China
e-mail:windowsdujian@163.com
xupeiyang@vip.163.com
binzhang@imicams.ac.cn
Zhang Shijing
Department of Medical Informatics
Tongji Medical College of Huazhong University of Science and Technology
Wuhan,China
e-mail:zhangsj9999@163.com
Abstract—The conceptual structures and thematic trends of health literacy research(1995-2010) are identified using a multiple-perspective of document co-citation analysis method, namely, labeling co-citation clusters from both cited and citing items. Comparative analysis shows that authors tend to choose broader-term labels when manually reading and analyzing cited references to a cluster, whereas algorithmically chosen terms from citing items tend to be more specific and narrowed. Combined with author co-citation analysis using WoS’ data, co-authorship analysis using data from PubMed and co-words analysis of both WoS’ descriptors and identifiers using CiteSpace, we discover that mental health literacy and functional health literacy are two separated research area in health literacy research and practice. Another two most active areas of research in recent years are also indentified.
Keywords—a multiple-perspective of co-citation analysis; health literacy; cluster labels; conceptual structures; thematic trends
I. INTRODUCTION
According to U. S. Institute of Medicine (IOM)[1] and Healthy People 2010[2], health literacy is defined as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions”. The term health literacy was first used in the article of Health education as social policy published by Simonds in 1974[3]. And since then, research has shown that individuals with inadequate health literacy have less knowledge of their diseases and treatments and a lack of skills needed to negotiate the healthcare system[4-8]. As such, health literacy has gained prominence amongst governments, health professionals and researchers. In order to reveal the conceptual structures and thematic trends of health literacy research and practice over the past 36 years(1974-2010), we used a multiple-perspective of co-citation analysis method recently proposed by Chaomei Chen[9] and focused on Document Co-Citation Analysis(DCA) in this article. The term multiple- perspective refers to the analysis of structural, temporal, and semantic patterns as well as the use of both citing and cited items for interpreting the nature of co-citation clusters. Hoping it can benefit to the research and practice of health literacy in the future.
II. Methods
2.1 data collection
The analysis is based on bibliographic records and citation information retrieved from Web of Science. Our search was conducted 4/22/2010 using strategy “Topic= (‘health literacy’ OR ‘healthcare literacy’) AND Document Type= (Article). Timespan=All Years. Databases =SCI-E, SSCI, A&HCI”. Finally we got 968 items. The first publication year occurred is 1995 with only 4 articles. In the meantime, an added search of PubMed conducted 4/22/2010 using strategy “‘health literacy’[Title/Abstract] OR ‘Health Literacy’[Mesh]” showed almost the same increasing trend of articles from 1985 to 2010.
Figure 1. The growing number of “health literacy” publications in the Web of Science and PubMed
Figure 1 shows the growing number of “health literacy” publications in the Web of Science and PubMed. Although the term “health literacy” was firstly proposed in 1974, however, until mid 1990s, it began to receive and attract extensive research. Especially in the 21st century, literature concerned is almost demonstrating an exponential growth trend.
2.2 A Multiple-Perspective of Co-Citation Analysis
Co-citation studies are among the most common used methods in quantitative studies of science, mainly including Author Co-citation Analysis (ACA) and Document Co-citation Analysis (DCA). Co-citation relations serve a fundamental grouping mechanism in co-citation studies. Researchers typically identify research domains or hot topics in terms of clusters of co-cited individual items. Since the traditional co-citation analysis typically focuses on cited members of clusters as a primary source of evidence for interpretation, Chaomei Chen innovatively introduced a multiple-perspective of co-citation analysis method considering both cited and citing items in 2010[9]. In fact, focusing on citers can improve our understanding of the nature of a research front and its intellectual base. In this paper, we will compare the cluster labels from cited and citing items. We use CiteSpace[10] to perform this part of study.
2.3 Strategic Diagram
In 1988, Law et al put forward the concept of "strategic diagram" to describe the relationship within a research area and the impact one has on others [11]. Till now, the strategic diagram is mostly used for co-word analysis to demonstrate the structure of a field or sub-field by calculating the degree of density and the centrality of each cluster. It places each hot topic of an area on the four quadrant of a coordinate, in order to describe the development of each theme. Using this principle, we will draw the Strategic Diagram innovatively based on the results of co-citation cluster analysis. According to the publication year of each article, we calculate the average publication year of each cluster so as to reflect the mean age of this topic, namely the degree of novelty. Based on cited times of each paper in this set of data, we calculate the average cited frequencies in order to describe the degree of attention. Finally, we developed the strategic diagram map taking the mean age and the average cited times of all top-30 highly cited references as the origin, the degree of novelty to each cluster as horizontal coordinate and the degree of attention as vertical coordinate.
III. RESULTS
3.1 Cluster labels from cited items
Co-citation networks are constructed with the top-30 most cited documents, with cited times at least 43 and accumulated percentage of cited frequencies nearly 10%. It means that we take out about top 1/10 of the highly cited items to conduct document co-citation analysis, and develop hierarchical cluster diagram using SPSS 11.5 as Fig. 2. The top-30 most cited documents are divided into seven clusters in terms of manually reading, recognizing and analyzing each one of documents to each cluster, namely,
C1: mental health literacy,
C2: health literacy of HIV patients,
C3: health literacy:a public health perspective,
C4: health literacy of patients with chronic diseases,
C5: health literacy survey,
C6: measurement tools of functional health literacy, and
C7:patients’ reading ability and readability of materials. Then we draw the strategic diagram of the seven document co-citation clusters as Fig. 3.
We can conclude from Fig.2 and Fig.3 that Cluster 1 “mental health literacy”, taking Jorm AF as the representative, is a very independent research field. The distance between it and other clusters is the largest(25). In fact, other clusters can be labeled as “functional health literacy” in terms of content analysis. It seems that mental health literacy is an important field in health literacy research and practice, yet has no direct and intensive connection with other fields. This issue will be discussed and validated in the next part of this paper. Meanwhile, as is shown in Fig. 3, its novelty is at a relatively high level, but its attention is relatively low.
Figure 2. Hierarchical cluster diagram of document co-citation analysis
Figure 3. Strategic Diagram of the document co-citation clusters
Cluster 6, namely “measurement tools of health literacy”, with the highest degree of attention but relatively older mean age. Why researchers still devote the most efforts to it? We predict that measurement tools of health literacy will be a sustainable research hot theme in the future. In fact, several articles concerned published most recently have provide sufficient evidence. For example, Jordan JE et al pointed out that an apparent gap exists between current definitions and measures of health literacy. The predominant approach to assessment has been direct testing of individual literacy(reading comprehension) and numeracy abilities. However these constructs do not reflect the range of attributes implied in existing definitions related to an individual’s ability to seek, understand and utilize health information[12]. Future research directions in the health literacy field have consistently identified the need to develop broader measures [13].
Cluster 3 “health literacy: a public health perspective”, with the youngest mean age but a relatively lower attention now, will be one of the most active areas of research in recent years. Our analysis will demonstrate that in the next part of this article.
Cluster 7, that is “patients’ reading ability and readability of materials” with the lowest degree of novelty and attention, is the initial and traditional health literacy.
3.2 Cluster labels from citing items
We use CiteSpace to conduct the document co-citation networks with the top-100 most cited references in each of the 16 one-year time slices between 1995 and 2010. Then, these networks were merged into a network of 850 co-cited references. The label of a document co-citation cluster was characterized by terms extracted from the citers of the cluster. In CiteSpace, nine methods of ranking extracted terms were implemented by choosing terms from three sources – titles, abstracts, and index terms of the citers of each cluster – and three ranking algorithms, namely, tf*idf weighting, log-likelihood ratio tests (LLR), and mutual information (MI). Top-ranked terms became candidate cluster labels. Some empirical studies by Chaomei Chen have proved that labels selected by log-likelihood ratio test appear to characterize the nature of clusters with finer-grained concepts than labels selected by tf*idf, and is more useful for differentiating clusters[14]. Among the nine methods, title terms with LLR is the best[9]. So, we select title terms with LLR to label each cluster. Table 1 summarizes candidate labels chosen by title terms with LLR.
Table 1 Candidate labels chosen by title terms with LLR sorted by size
|
ID
|
Size
|
Silhouette
|
mean(Year)
|
Label (LLR)
|
|
4
|
48
|
0.545
|
1998
|
health literacy (45.75); health-related knowledge (19.3); people (19.3)
|
|
1
|
22
|
0.682
|
1999
|
mental health literacy (30.12); treatment (20.09); belief (16.05)
|
|
5
|
21
|
0.821
|
1993
|
high readability level (21.66); anticoagulant patient information material (21.66); colorectal cancer (17.21)
|
|
2
|
5
|
0.64
|
2005
|
health inequality (11.22); healthcare system (11.22); health outcome (7.53)
|
|
3
|
2
|
0.715
|
2005
|
parent-provider relationship (7.26); community health center (7.26); quality (7.26)
|
|
6
|
2
|
0.911
|
2006
|
patient comprehension (10.12); need (6.91); consumer (6.91)
|
We use top-ranked terms as candidate cluster labels. There are four largest clusters, namely health literacy, mental health literacy, high readability level, and health inequality. The last two clusters, with both only 2 citers, will not be analyzed in this paper.
The silhouette metric[15] is useful in estimating the uncertainty involved in identifying the nature of a cluster. The silhouette value of a cluster, ranging from -1 to 1, indicates the uncertainty that one needs to take into account when interpreting the nature of the cluster. The value of 1 represents a perfect separation from other clusters. In this study, among the largest 4 clusters, “health literacy”, with silhouette score at 0.545, indicates that the nature of this cluster is relatively uncertain. Therefore, according to the second and third candidate cluster labels, namely, “health-related knowledge” and “people”, we can manually label this cluster as Functional Health Literacy.
The cluster of “Mental health literacy”, “high readability level” or “reading ability and the readability of materials” are labeled simultaneously both from cited and citing items. Moreover, “health inequality” from citing items is coincided with “health literacy: a public health perspective” form cited items. But the former is a more specific concept than the latter.
Meanwhile, Cluster 5 with labels “high readability level” and “anticoagulant patient information material” is one aspect of factors for improving patients’ reading ability which is also labeled as a cluster from the cited items’ perspective. Therefore, cluster labels chosen from citers of a cluster tend to be more specific and narrowed terms than those chosen by human experts.
From the aspect of mean(Year) in Table 1, “high readability level” is the oldest(1993), “health inequality” is the newest(2005), and “mental health literacy” is in the mid. This is also coincided with what is shown in Strategic Diagram of the document co-citation clusters. These findings suggest that the multiple perspective method has the potential to provide additional insights in complementary to existing methods and provide an intermediate level of support for interpreting the nature of research field.
3.3 Conceptual structures of health literacy research
In the previous analysis, mental health literacy has no direct and intensive connection with other fields. But this point is concluded from the perspective of document co-citation analysis. In order to invalidate it furtherly, we conducted author co-citation analysis using WoS’ data and co-authorship analysis using data from PubMed. Fig.4 shows top16 co-authorship network(a) from PubMed and top30 author co-citation network(b) from WoS. We can see mental health literacy represented by Jorm AF and functional health literacy represented by Baker DW, David TC et al indeed has no direct connections. They are two separated research fields.
In addition, we also conducted co-words analysis of both WoS’ descriptors (DE for short) and identifiers (ID for short) using CiteSpace. As is shown in Fig.5, at the terms’ level, health literacy research was still clearly divided into two areas, namely “mental health literacy” on the left and “functional health literacy” on the right. The topic “mental health literacy” was characterized by mental-health literacy, mental illness, mental health, depression, major depression, and so on. While the topic “functional health literacy” was represented by knowledge, numeracy, comprehension, communication, readability, discharge instructions, etc. The abilities acting on health information that the concept of health literacy focusing on are represented by such terms as information and Internet. However, they are on the edge of the network.
IV. DISCUSSION AND CONCLUSION
We have analyzed the conceptual structures and thematic trends of health literacy research (1995-2010) using a multiple-perspective of document co-citation analysis. The study reveals that authors tend to choose broader-term labels when reading and analyzing cited references, whereas algorithmically chosen terms from citing items’ titles tend to be more specific and narrowed. We also have shown that cluster membership and citer-focused labeling provide complementary information to form a comprehensive landscape of the field. Note that one may reach different insights into the nature of a co-citation cluster if different sources of information are used. The cited members of a cluster define its intellectual base, whereas citers to the cluster form a research front. The major advantage of our approach is that it enables analysts to consider multiple aspects of the citation relationship from multiple perspectives.
We discover that mental health literacy and functional health literacy are two separated research area in health literacy research and practice. Health literacy: a public health perspective is one of the most active areas of research in recent years, and Measurement tools of health literacy will be a sustainable research hot theme in the future.
References
[1] Institute of Medicine (IOM). 2004. Health literacy: A prescription to end confusion. Washington, DC: The National Academies Press, http://www.iom.edu/CMS/3775/3827/19723.aspx (accessed Aug 1, 2010).
[2] U.S. Department of Health and Human Services (DHHS). 2000. Healthy People 2010. Washington, DC: U.S. Government Printing Office, http://www.healthypeople.gov/Document/pdf/uih/2010uih.pdf (accessed Aug 1, 2010).
[3] Simonds, S. K. Health education as social policy[J]. Health Education Monograph,1974,2:1-25.
[4] Baker DW, Parker RM, Williams MV,et al. Health literacy and the risk of hospital admission[J]. Journal of General Internal Medicine,1998,13(12):791-798.
[5] Levin-Zamir D, Peterburg Y. Health literacy in health systems: perspectives on patient self-management in Israel[J]. Health Promotion International,2001,16(1):87-94.
[6] Williams MV, Baker DW, Parker RM,et al. Relationship of functional health literacy to patients’ knowledge of their chronic disease. A study of patients with hypertension and diabetes. Archives of Internal Medicine,1998,158(2):166-172.
[7] van Servellen G, Nyamathi A, Carpio F,et al. Effects of a treatment adherence enhancement program on health literacy, patient–provider relationships, and adherence to HAART among lowincome HIV-positive Spanish-speaking Latinos[J]. AIDS Patient Care and STDs,2005,19(11):745-759.
[8] Mancuso JM. Health literacy: a concept/dimensional analysis[J]. Nursing and Health Sciences,2008,10(3):248-255.
[9] Chen C, SanJuan FI, Hou J. The Structure and Dynamics of Co-Citation Clusters: A Multiple-Perspective Co-Citation Analysis[J]. Journal of the American Society for Information Science and Technology,2010,61(7):1386-1409.
[10] CiteSpace[EB/OL]. http://cluster.cis.drexel.edu/ ~cchen/citespace/
[11] Law J,Bauin S,Courtial JP,et al.Policy and the mapping of scientific change:A co-word analysis of research into environmental acidification[J].Scientometrics,1988,14(3-4):251-264
[12] Jordan JE, Buchbinder R, Osborne RH. Conceptualising health literacy from the patient perspective[J]. Patient Education and Counseling,2010,79(1):36-42.
[13] Nutbeam D. The evolving concept of health literacy[J]. Social Science & Medicine,2008,67(12):2072–2078.
[14] Chen CM,Zhang J,Vogeley MS. Making sense of the evolution of a scientific domain:a visual analytic study of the Sloan Digital Sky Survey research[J]. Scientometrics,2010,83:669–688.
[15] Rousseeuw PJ. Silhouettes:A graphical aid to the interpretation and validation of cluster analysis[J]. Journal of Computational and Applied Mathematics,1987,20:53-65.
Figure 4. Top16 co-authorship from PubMed and top 30 author co-citation from WoS
Figure 5. Co-words analysis of DE and ID using CiteSpace. 1995-2010,slice=4.(3,2,20;4,3,20;5,5,25)
https://blog.sciencenet.cn/blog-280034-382429.html
上一篇：参加第六届科学计量学与大学评价国际研讨会的收获（杜建来信）
下一篇：一个人的晚餐
