Volume 36, Issue 6, Pages 538-548 (June 2009)

12 of 16

ABSTRACT

FULL TEXT

FULL-TEXT PDF (641 KB)

SUPPLEMENTARY DATA

Forty Years of Secondhand Smoke Research: The Gap Between Discovery and Delivery

published online 16 April 2009.

Context

Public health initiatives often focus on the discovery of risk factors associated with disease and death. Although this is an important step in protecting public health, recently the field has recognized that it is critical to move along the continuum from discovery of risk factors to delivery of interventions, and to improve the quality and speed of translating scientific discoveries into practice.

Evidence acquisition

To understand how public health problems move from discovery to delivery, citation network analysis was used to examine 1877 articles on secondhand smoke (SHS) published between 1965 and 2005. Data were collected and analyzed in 2006–2007.

Evidence synthesis

Citation patterns showed discovery and delivery to be distinct areas of SHS research. There was little cross-citation between discovery and delivery research, including only nine citation connections between the main paths. A discovery article was 83.5% less likely to cite a delivery article than to cite another discovery article (OR=0.165 [95% CI=0.139, 0.197]), and a delivery article was 64.3% less likely (OR=0.357 [95% CI=0.330, 0.386]) to cite a discovery article than to cite another delivery article. Research summaries, such as Surgeon General reports, were cited frequently and appear to bridge the discovery–delivery gap.

Conclusions

There was a lack of cross-citation between discovery and delivery, even though they share the goal of understanding and reducing the impact of SHS. Reliance on research summaries, although they provide an important bridge between discovery and delivery, may slow the development of a field.

Article Outline

• Abstract

• Introduction

• Secondhand Smoke Exposure

• Secondhand Smoke Research

• Study Goals and Approach

• Evidence Acquisition

• Data Collection

• Selecting the data source

• Inclusion and exclusion criteria

• Article selection and classification

• Data Analysis

• Evidence Synthesis

• What Are the Characteristics of the SHS Citation Network?

• General network characteristics

• Highly cited articles

• How Does Information Flow Between Discovery and Delivery Research Within the SHS Citation Network?

• Identification of main citation paths

• Discovery main path

• Delivery main path

• Information flow between the discovery main path and the delivery main path

• What Are the Predictors of Citation Patterns Within the SHS Citation Network?

• Stochastic network modeling

• Are Research Summaries Filling the Gap?

• Discussion

• Conclusion

• Acknowledgment

• Supplementary data

• References

• Copyright

Introduction

Efforts in public health often focus on discovery of risk factors associated with disease and death. However, discovery is only half the story; it is critical to move along the continuum from discovery of health information to delivery of interventions that improve public health.1, 2, 3, 4 For example, although understanding the nature and extent of a new virus is important, vaccine development and other interventions to combat the virus are equally important, as is the delivery of these interventions to appropriate populations.

The path from discovery to delivery is a fundamental process in public health; however, it is not well understood. A few models showing the relationship between discovery and delivery have recently emerged (e.g., the National Cancer Institute Discovery-Development-Delivery model3). For example, in relation to cancer science, the “… research process spans a continuum from discovery of new knowledge about the process of cancer, to development of new interventions, to the ultimate delivery of new, more effective, and safer interventions to all who need them.”3

One helpful framework for understanding the path from discovery to delivery is the Diffusion of Innovations (DOI) theory.5 The DOI theory consists of three major steps: become aware of and learn about the innovation, develop a positive or negative attitude toward the innovation, and put the innovation to use.5 This theory has often been used to describe how innovations travel through social networks. For example, one exemplar study examined how interpersonal communication among physicians facilitated adoption of a new drug.6 Using DOI theory, scientific progress can be seen as a highly formalized process in which scientific discoveries are the innovations that are communicated (at least partially) through peer-reviewed studies.

Recent work in systems thinking and transdisciplinarity related to public health further elucidates how information and innovation move along the discovery-to-delivery continuum.7, 8 In public health, discovery research is typically conducted by researchers in different disciplines than those conducting delivery research; therefore, information flow between discovery and delivery research necessarily crosses disciplinary boundaries. A study on inter- and trans-disciplinary research in tobacco-harm reduction found that connections among researchers across disciplines were more likely to lead to synergistic outcomes (e.g., journal articles and research proposals) than were connections among researchers in the same discipline.8 The lack of transdisciplinary research in public health may contribute to the slow pace of scientific progress and of application of scientific discoveries to public health problems.9 For example, moving clinical discoveries from bench to bedside typically takes 17 years.10

Researchers and funders in public health increasingly advocate cross-disciplinary collaboration.9, 11, 12 Although there has been skepticism related to cross-disciplinary research,11 such as its high costs in terms of human and financial resources, cross-disciplinary collaborations have been described as having the potential to facilitate dissemination of different theories, methods, and approaches among researchers, ultimately resulting in better science.8, 12, 13 However, “… very little has been done to understand the extent to which research in tobacco control or other scientific entities is cross-disciplinary, and if so, what this process looks like.”8 Having a better understanding of cross-disciplinary research may allow scientists to be strategic in their collaborations, facilitating better science and faster transitions from discovery to delivery.

Behavioral theories, such as cultural contingency theory14, 15 and social learning theory,16 may also assist in understanding the slow process of translation from discovery to delivery and the absence of cross-disciplinary collaboration among scientists. In simple terms, these theories propose that individual behavior is influenced by the observation of the behaviors of others. When applied to the question of how science moves along the discovery-to-delivery continuum, these theories imply that scientists are observing and adopting the behaviors or strategies prevalent in their professional culture.

In short, the mechanisms by which information is transmitted among scientists and scientific areas are complex and influenced by many factors. For example, scientists may be constrained from cross-disciplinary communication and collaboration by a lack of funding opportunities or time to integrate multiple scientific areas,11 or they may simply be adopting observed behaviors that do not support or value cross-disciplinary work. As a first step in better understanding the path from discovery to delivery, the current study examines citation patterns, a formal marker13 of scientific communication, for the public health issue of secondhand smoke (SHS) exposure.

Secondhand Smoke Exposure

Research on SHS exposure forms one of the most important subdisciplines in tobacco control. As of 2006, over 126 million nonsmoking Americans were exposed to SHS in homes, vehicles, workplaces, and public places.17 This total includes 22 million, or about 60%, of children aged 3–11 years.17 Secondhand smoke contains at least 250 toxic chemicals, including more than 50 that can cause cancer.18 The 2006 Surgeon General's report summarized 4 decades of SHS research and concluded that there is “no risk-free level” of exposure.17

Secondhand Smoke Research

Secondhand smoke research has a 40-year history of publication in peer-reviewed journals. Within the field of SHS research (as in many public health areas), discovery and delivery can be viewed as distinct subfields. Discovery research focuses on identification and explanation of the health effects of SHS exposure, whereas delivery research evaluates policy and practice interventions that reduce or mitigate exposure. Discovery research first appeared in peer-reviewed journals in the mid-1960s.19 Over time, discovery research has associated SHS exposure with many poor health outcomes,20, 21, 22 providing the foundation for delivery of interventions to reduce exposure.23, 24, 25 The earliest delivery research examining SHS interventions was published in 1980.

Study Goals and Approach

The goal of this study was to better understand how discoveries related to SHS exposure were connected to delivery of interventions to reduce SHS exposure. To address this goal, citation network analysis (CNA) was used to examine the flow of information among SHS publications. By examining networks of who cites whom, CNA can determine roles and contributions of articles, authors, journals, and topic areas within a field and show how the field has developed.26, 27, 28, 29, 30, 31 Although CNA has been used to examine scientific development, it has not been used to trace the flow of information between discovery and delivery within a scientific area. In this study, CNA was used to understand the connections between discovery and delivery in SHS research.

This exploratory study focused on three research questions: (1) What are the characteristics of the SHS citation network?, (2) How does information flow between discovery and delivery research within the SHS citation network?, and (3) What are the predictors of citation patterns within the SHS citation network? By examining how information flows through the field of SHS research, some of the factors related to the translation of discovery into delivery may be better understood.

Evidence Acquisition

Data Collection

Selecting the data source

To develop a comprehensive list of published SHS research, journal-indexing databases were examined, and Web of Science (WOS) was selected for its breadth of coverage. Database completeness was assessed by conducting a pilot search for 63 exemplar SHS articles from three sources: the 2006 Surgeon General's report17; the Americans for Nonsmokers' Rights 2005 SHS bibliography32; and the 100 most-cited articles on smoking, as identified in a recent review.33 Of the 63 articles, WOS included 58 (92.1%).

Inclusion and exclusion criteria

To capture the majority of SHS articles, a list of 15 keywords was developed by selecting terms from the 63 articles, conducting a keyword search for secondhand smoke in MEDLINE and WOS, and informally surveying SHS researchers. Articles were limited to those with English titles and abstracts, published prior to 2006, and empirical articles (i.e., those including data analysis). The English limitation was minimal because non-English publications often translate titles and abstracts (www.thomsonreuters.com/business_units/scientific/free/essays/journalselection/). The search, conducted in 2006, produced a database of 4921 articles.

Article selection and classification

Initial data included many articles that were not relevant; for example, many did not have findings related to SHS. To narrow the data set and classify articles in each subfield, empirical articles were identified and classified using the following definitions. Articles classified as discovery research were those that included discovery or confirmation of the physical effects of SHS exposure, with SHS findings being the primary purpose or among the primary findings, such as Environmental Tobacco-Smoke and Lung-Cancer in Nonsmoking Women: A Multicenter Study.34 Articles classified as delivery research were those that included testing or evaluation of interventions designed to reduce exposure to SHS or mitigate the health effects of SHS exposure, such as Bartender's Respiratory Health After Establishment of Smoke-Free Bars and Taverns.35

Abstracts were used for classification purposes. For articles without abstracts (n=573, 11.6%), titles were used. Two researchers classified 250 articles with 89% agreement (kappa=0.78).36 The disagreements were discussed until consensus was reached. Each researcher coded half the remaining articles independently, yielding a total of 1877 empirical SHS articles included in the analyses. Nonrelevant articles were excluded from analyses. For additional information on the article classification process, see Appendix A, available online at www.ajpm-online.net.

Data Analysis

Citation network analysis was used to produce network graphs and statistics to examine the flow of information between SHS scientists and studies. Citation networks are directed, with links showing the direction of information flow. If article A is cited by article B, information flows from A to B, and the citation link is shown as A→ B (A sends information to B). Table 1 shows CNA vocabulary and concepts.

Table 1.

Network analytic methods and definitions used in the SHS research article citation network26, 27


	Network term	Description or definition
	Arc	An arc is an arrow representing the flow of information between two nodes. In citation networks, arcs go from cited nodes to citing nodes, so if A is cited by B, then the arc would look like this: A→ B, showing that A is sending information to B.
	Degree	Degree is the number of citation links a node has. In directed networks, degree is divided into two types: •indegree—the number of incoming links, or how many times an article cites others in the network •outdegree—the number of outgoing links, or how many times an article is cited by others in the network
	Degree	Degree is often used as a measure of network centrality. The more connected a node is (higher degree), the more central the node is to the network. Visually, indegree is the number of arrows pointing to a node, whereas outdegree is the number of arrows originating from a node.
	Density	The number of links in a network as a proportion of the total possible links; range: 0–1
	Source	An article not citing any others in the network (indegree=0)
	Sink	An article not cited by any others in the network (outdegree=0)
	Traversal count or traversal weight	The proportion of all paths between source and sink articles that contain a particular link or article (i.e., the extent to which a particular article or link is needed for keeping the network connected)
	Main path	The path from a source article to a sink article that has the highest traversal weights on its arcs

Several steps were followed in developing and analyzing the citation network in 2006–2007. Degree was examined first. Degree is a count of the incoming and outgoing citations of each article, providing a measure of the importance or prominence of an article (Table 1). Second, main paths for the discovery and delivery articles were developed. Main paths go from a source node to a sink node. Main paths include nodes and links with the highest traversal weights, or proportions of paths from all sources to all sinks in the entire network. Main path articles are seen as central in a field because they integrate information from previous articles and add substantial new knowledge to an area of research; they serve as the structural backbone of a body of knowledge.37 Third, stochastic network methods were used to formally test whether subfield (discovery or delivery) was a significant predictor of citation patterns. Pajek 1.13 was used38 for network measures and visualization, SPSS version 13.0 for other descriptive statistics, and R-statnet version 2.0 for stochastic network modeling.

Evidence Synthesis

What Are the Characteristics of the SHS Citation Network?

General network characteristics

The SHS citation network data included 1877 articles published between 1965 and 2005. Three networks were examined: the overall network (N=1877 articles); the discovery network (n=1497 articles); and the delivery network (n=380 articles). The descriptive statistics in Table 2 show the discovery network as larger, with a longer history, whereas the delivery network was more densely connected, perhaps because of its smaller size.39

Table 2.

Characteristics of SHS articles in the citation network


Characteristic	Overall network	Discovery network	Delivery network
Number of articles	1877	1497(80%)	380(20%)
Number of unlinked articles (degree=0)	222	175	47
Year of first publication	1965	1965	1980
Number of journals	418	342	140
Average number of times an article cited another article (SD)	3.9(4.2)	3.9(4.2)	3.9(4.5)
Average number of times an article was cited (SD)	3.9(7.8)	4.2(8.3)	2.9(5.3)
Number of citation links	7370	5748(78%)	945(13%)
Discovery cites discovery	5748(78%)	—	—
Discovery cites delivery	140(2%)	—	—
Delivery cites discovery	537(7%)	—	—
Delivery cites delivery	945(13%)	—	—
Average years between cited and citing article publication	6.2
Discovery cites discovery	6.5	—	—
Discovery cites delivery	4.5	—	—
Delivery cites discovery	5.9	—	—
Delivery cites delivery	5.0	—	—
Number of articles with median impact factor (2002–2006)
Impact factor <3	1103(59%)	807(54%)	296(78%)
Impact factor 3–5	529(28%)	461(31%)	68(18%)
Impact factor >5	245(13%)	229(15%)	16(4%)
Network density	0.0021	0.0026	0.0065a

The higher density in the delivery network may be an artifact of network size; larger networks are typically less dense than smaller ones.39

Highly cited articles

How often an article cited other articles (indegree) and was cited by other articles (outdegree) was measured for all articles, and for discovery and delivery articles separately. Discovery articles had higher average outdegrees, indicating that they were cited more often. The ten most-cited articles were discovery articles (Table 3). Seven of the top ten articles were about the exposure of children to SHS. Overall, a few articles were highly cited, whereas most articles were rarely or never cited. This finding suggests a “rich-get-richer” or scale-free phenomenon40 in which everybody in the field cites a few superstar articles.41, 42, 43, 44, 45, 46, 47, 48, 49, 50

Table 3.

The ten most-cited articles in the citation network for SHS exposure research


Study/Title	Number of times cited	Journal	2006 journal impact factor
Ware (1984)41	79	American Review of Respiratory Diseasea	9.1
Passive Smoking, Gas Cooking, and Respiratory Health of Children Living in 6 Cities	79	American Review of Respiratory Diseasea	9.1
Pirkle (1996)42	78	Journal of the American Medical Association	23.2
Exposure of the US Population to Environmental Tobacco Smoke—The Third National Health and Nutrition Examination Survey, 1988 to 1991	78	Journal of the American Medical Association	23.2
Chilmoncyzk (1993)43	78	New England Journal of Medicine	51.3
Association Between Exposure to Environmental Tobacco-Smoke and Exacerbations of Asthma in Children	78	New England Journal of Medicine	51.3
Colley (1974)44	74	Lancet	25.8
Influence of Passive Smoking and Parental Phlegm on Pneumonia and Bronchitis in Early-Childhood	74	Lancet	25.8
Martinez (1988)45	64	American Review of Respiratory Diseasea	9.1
Parental Smoking Enhances Bronchial Responsiveness in 9-Year-Old Children	64	American Review of Respiratory Diseasea	9.1
Martinez (1992)46	63	Pediatrics	5.0
Increased Incidence of Asthma in Children of Smoking Mothers	63	Pediatrics	5.0
Burchfiel (1986)47	61	American Review of Respiratory Diseasea	9.1
Passive Smoking in Childhood— Respiratory Conditions and Pulmonary-Function in Tecumseh, Michigan	61	American Review of Respiratory Diseasea	9.1
Jarvis (1984)48	55	Journal of Epidemiology and Community Health	2.8
Biochemical Markers of Smoke Absorption and Self Reported Exposure to Passive Smoking	55	Journal of Epidemiology and Community Health	2.8
Benowitz (1996)49	52	Epidemiologic Reviews	8.3
Cotinine as a Biomarker of Environmental Tobacco Smoke Exposure	52	Epidemiologic Reviews	8.3
Hanrahan (1992)50	51	American Review of Respiratory Diseasea	9.1
The Effect of Maternal Smoking During Pregnancy on Early Infant Lung-Function	51	American Review of Respiratory Diseasea	9.1

Now called American Journal of Respiratory and Critical Care Medicine

How Does Information Flow Between Discovery and Delivery Research Within the SHS Citation Network?

Identification of main citation paths

Main paths were identified for the delivery and discovery networks in order to understand the development of science37 in these two areas (Figure 1). These paths represent the primary flow of information through these subfields (see Appendix A, available online at www.ajpm-online.net, for methodologic details on the identification of the main paths).

View Large Image
Download to PowerPoint

Figure 1. Main citation paths through discovery (top; n=54) and delivery (bottom; n=27) research articles related to SHS exposure, and citation links between the two paths; see Appendix C, available online at www.ajpm-online.net, for list of articles.

Discovery main path

The 54 articles in the discovery main path were published between 1972 and 2001 (Figure 1, top). The link with the highest traversal weight (0.46) in the discovery main path was between a 1986 article by Burchfiel and colleagues47 and a 1987 article by O'Connor and colleagues,51 both focusing on pulmonary and respiratory conditions in children exposed to SHS. This traversal weight indicates that 46% of paths through the discovery network included the link between these two articles.

Delivery main path

The delivery main path included 27 articles published between 1985 and 2003 (Figure 1, bottom). Two connections had the highest traversal weight (0.35) in the delivery main path. One was between a 1985 article by Kottke and colleagues52 about the attitudes of patients, employees, and faculty at a smoke-free hospital and a 1990 article by Stillman and colleagues53 about Johns Hopkins Medical Institutions going smoke-free. The other highly weighted link was between the 1990 Stillman and colleagues53 article and a 1990 article by Borland and colleagues23 on workplace smoking bans.

Information flow between the discovery main path and the delivery main path

To determine whether there were connections between the backbone of discovery research and the backbone of delivery research, citation links between articles in the discovery main path (Figure 1, top) and articles in the delivery main path (Figure 1, bottom) were identified. The dashed links between the main paths show the citations resulting in information flow between the two. All nine links between the main paths consisted of delivery articles citing discovery articles. So, of 81 articles spanning >30 years, which constitute the backbones of these two subfields, only nine citation links existed between the areas. This finding shows little direct discovery-to-delivery information flow among the articles most critical in the development of SHS discovery and delivery science. Table 4 lists the articles linked by these nine connections.20, 34, 35, 54, 55, 56, 57, 58, 59, 60

Table 4.

Cross-citation links between the discovery and delivery main paths


	Cited (discovery articles only)	Cited by (delivery articles only)
	Hole (1989)54	Eisner (1998)35
	Passive Smoking and Cardiorespiratory Health in a General-Population in the West of Scotland	Bartenders' Respiratory Health After Establishment of Smoke-Free Bars and Taverns
	Fontham (1994)34	Longo (1996)55
	Environmental Tobacco-Smoke and Lung-Cancer in Nonsmoking Women: A Multicenter Study	Hospital Smoking Bans and Employee Smoking Behavior—Results of a National Survey
	Eisner (1998)56	Longo (1996)55
	Environmental Tobacco Smoke and Adult Asthma—The Impact of Changing Exposure Status on Health Outcomes	Hospital Smoking Bans and Employee Smoking Behavior—Results of a National Survey
	Svendsen (1987)20	Eisner (1998)35
	Effects of Passive Smoking in the Multiple Risk Factor Intervention Trial	Bartenders' Respiratory Health After Establishment of Smoke-Free Bars and Taverns
		Okah (2002)57
		Effect of Children on Home Smoking Restriction by Inner-City Smokers
		Stave (1991)58
		Effect of a Total Work-Site Smoking Ban on Employee Smoking and Attitudes
	Garland (1985)59	Okah (2002)57
	Effects of Passive Smoking on Ischemic Heart-Disease Mortality of Nonsmokers—A Prospective Study	Effect of Children on Home Smoking Restriction by Inner-City Smokers
		Stave (1991)58
		Effect of a Total Work-Site Smoking Ban on Employee Smoking and Attitudes
	Aronow (1978)60	Stave (1991)58
	Effect of Passive Smoking on Angina Pectoris	Effect of a Total Work-Site Smoking Ban on Employee Smoking And Attitudes

What Are the Predictors of Citation Patterns Within the SHS Citation Network?

Stochastic network modeling

Figure 1 suggests a gap between empirical SHS discovery and delivery research. Further evidence for this gap comes from the descriptive statistics in Table 2, which indicate that only 9% of all citation links in the overall network are between empirical discovery and delivery articles. Recent developments in stochastic network modeling allow us to look at patterns such as this one more formally.61 An exponential-family random graph model (ERGM), specifically a logit model, was developed to determine the probability of a citation link between two articles in the overall network, based on article characteristics.61 This model was built to formally test the hypothesis that the type of SHS subfield (i.e., discovery versus delivery) is significantly related to the probability of a cross-field citation link.

The model controlled for the number of ties in the network, year of publication, and journal impact factor (Table 5). The Edges/Arcs term equals the number of ties in the network and is analogous to the intercept in linear regression.62, 63 The variables represented by the Year term controlled for the publication year. Journal impact factor controlled for the median impact factor of each journal between 2002 and 2006 (the most recent 5-year period of impact factor availability). To determine the influence of subfield, terms for each possible mix of subfields were included. These variables represented change in the probability of a citation link when two articles were of the specified subfields (i.e., discovery or delivery).

Table 5.

Logit model predicting citation linkages among articles in the entire SHS citation network (N=1877)


Coefficient	Model 1				Model 2
Coefficient	Logit	SE	OR	95% CI	Logit	SE	OR	95% CI
Edges/arcs	−5.27	0.10	0.005	0.004–0.006	−5.14	0.10	0.006	0.005–0.007
Year citation patterns
Cites articles by year (indegree)	−0.10	0.001	0.905	0.903–0.907	−0.10	0.001	0.905	0.903–0.907
Cited by year (outdegree)	0.04	0.002	1.041	1.037–1.045	0.04	0.002	1.041	1.037–1.045
JOURNAL CITATION PATTERN
Cites articles (indegree)
Impact factor <3	ref	—	—	—	ref	—	—	—
Impact factor 3–5	0.79	0.03	2.203	2.077–2.336	0.80	0.03	2.226	2.098–2.360
Impact factor >5	1.58	0.03	4.855	4.578–5.149	1.61	0.03	5.003	4.717–5.306
Cited (outdegree)
Impact factor <3	ref	—	—	—	ref	—	—	—
Impact factor 3–5	0.23	0.03	1.259	1.187–1.335	0.22	0.03	1.246	1.176–1.322
Impact factor >5	0.32	0.04	1.377	1.273–1.489	0.31	0.04	1.363	1.261–1.475
Subfield citation patterns
Discovery cites discovery					ref	—	—	—
Discovery cites delivery					−1.80	0.09	0.165	0.139–0.197
Delivery cites discovery					−1.03	0.04	0.357	0.330–0.386
Delivery cites delivery					1.47	0.04	4.349	4.021–4.704
Model fit	−2LL	AIC		df	−2LL	AIC		df
	98,520	98,533		7	95,587	95,607		10

AIC, Akaike Information Criterion; −2LL, −2 times the log-likelihood

Table 5 shows model results. Model 1 shows the probability of a citation link based on the control variables. Model 2 adds subfield as a predictor of a citation link and is a significantly better-fitting model than Model 1 (χ2=2.93, df=3, p<0.001). Odds ratios show that, when controlling for year and impact factor, a discovery article was 83.5% less likely to cite a delivery article (OR=0.165 [95% CI=0.139, 0.197]), and a delivery article was 64.3% less likely (OR=0.357 [95% CI=0.330, 0.386]) to cite a discovery article than was a discovery article to cite another discovery article. A link between two delivery articles was more than four times as likely (OR=4.349 [95% CI=4.021, 4.704]) as a link between two discovery articles.

Are Research Summaries Filling the Gap?

Although there was little direct information flow between empirical discovery and empirical delivery articles, the 2546 citations of the 81 main path articles were examined, and a number of cited research summaries (e.g., Surgeon General's reports) were identified that bridge this gap. The 15 most-cited research summaries are shown in Table 6, as well as how often each was cited by discovery and delivery main path articles.64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78 Figure 2 shows the two main paths and the citation links from main path articles to the research summaries.64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78 Note that the research summaries were cited a total of 120 times by the 81 main path articles. So, although the main path delivery articles cited the empirical discovery articles a total of nine times, the main path delivery articles cited the research summaries a total of 55 times. This finding indicates that research summaries are playing an important bridging role in the path between discovery and delivery.

Table 6.

The 15 research summaries most often cited by main path articles (Figure 2)


Study	Cited by discovery	Cited by delivery	Total cited
EPA (1992)64	13	11	24
Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders	13	11	24
USDHHS (1986)65	8	12	20
Surgeon General's Report: The Health Consequences of Involuntary Smoking	8	12	20
National Research Council, Committee on Passive Smoking (1986)66	11	5	16
Environmental Tobacco Smoke: Measuring Exposures and Assessing Health Effects	11	5	16
USDHHS (1979)67	7	1	8
Surgeon General's Report: The Health Consequences of Smoking for Women	7	1	8
California EPA (1997)68	4	3	7
Health Effects of Exposure to Environmental Tobacco Smoke	4	3	7
USDHHS (1989)69	1	5	6
Surgeon General's Report: Reducing the Health Consequences of Smoking—25 Years of Progress	1	5	6
Australian National Health and Medical Research Council (1987)70	0	6	6
Effects of Passive Smoking on Health	0	6	6
Fielding (1988)71	2	3	5
Health Effects of Involuntary Smoking	2	3	5
Lee (1988)72	4	0	4
Misclassification of Smoking Habits and Passive Smoking	4	0	4
Lee (1992)73	4	0	4
Environmental Tobacco Smoke and Mortality	4	0	4
Lefcoe (1983)74	3	1	4
The Health Risks of Passive Smoking. The Growing Case for Control Measures in Enclosed Environments	3	1	4
Siegel (1993)75	0	4	4
Involuntary Smoking in the Restaurant Workplace. A Review of Employee Exposure and Health Effects	0	4	4
Weiss (1983)76	4	0	4
State of the Art: The Health Effects of Involuntary Smoking	4	0	4
Wynder (1983)77	4	0	4
Smoking and Lung Cancer: Some Unresolved Issues	4	0	4
Glantz (1991)78	0	4	4
Passive Smoking and Heart Disease. Epidemiology, Physiology, and Biochemistry	0	4	4

View Large Image
Download to PowerPoint

Figure 2. Main citation paths through discovery (top) and delivery (bottom) research articles related to SHS exposure, and the 120 citation links to the 15 research summaries cited most often by main path articles

Discussion

By analyzing the citation links among SHS publications, this study has: (1) provided evidence of a gap between empirical discovery and delivery research in public health; (2) showed how review articles, government reports, and other research summaries may be filling this gap; and (3) provided an example of an innovative approach to the assessment of scientific communication and development.

This analysis of SHS publications identified a 40-year history of research, beginning with an early focus on health effects and expanding over time to include research evaluating interventions. The subareas of SHS research, and discovery and delivery are fundamental in public health, as is the translation process between the two (i.e., translating research into practice). Despite sharing the common goal of understanding and reducing the impact of SHS, this study reveals a lack of direct connections between discovery research and delivery research.

Although citation links among articles are only one marker of scientific communication,13 the lack of citation linkages between original research articles may be an indicator of limited contact among scientists working on discovery and scientists working on delivery. Collaboration across disciplines has been identified as a way to increase the pace of discovery-to-delivery communication and to strengthen science.8, 9, 12, 79 Facilitation of collaboration among scientists has become a national priority9 with the development of transdisciplinary research centers and dedicated funding. Despite the potential and prioritization of transdisciplinary research, there has been a lack of methods and measures for understanding how scientists collaborate and how scientific information travels from discovery to delivery. This study used CNA as a new approach to examining the path from discovery to delivery through one of the most formal measures of scientific communication: study citations.

There are a few limitations to note about this study. Minor data-quality issues were encountered with WOS data, including occasional misspellings of author names or other identifying information. Mistakes found were corrected; however, it is probable that a small number of links were missed because of this issue. Another limitation was data comprehensiveness. The WOS database includes comprehensive but not all-inclusive journal coverage (www.thomsonreuters.com/business_units/scientific/free/essays/journalselection/), and data collection was limited to 15 keywords and to English articles. Although this strategy most likely identified the vast majority of SHS articles, some articles may have been missed.

In addition, the study focused primarily on published articles and examined only a small number of research summaries such as the 1986 Surgeon General's report65 and the 1992 Environmental Protection Agency report.64 Research summaries appear to act as intermediary documents, providing a bridge between empirical discovery and delivery publications, and they should be examined more closely.80 For example, there were 120 citation links from a discovery or delivery article to the 15 identified research summaries (Table 6; Figure 2). Also, delivery researchers might have been more likely to cite discovery studies to support intervention delivery, whereas similar support may not be necessary for discovery researchers. Finally, citation is a marker of collaboration among scientists but does not encompass all the ways or reasons collaboration occurs.9

Conclusion

This analysis examines a fundamental process in public health: the progression from scientific discovery to delivery. The findings show that discovery and delivery are distinct areas of empirical research bridged by systematic summary documents. Although research summaries may be easier to identify and utilize than original research articles, they do not typically contain the most recent scientific discoveries and may be out of date soon after they are published. Thus, one way to speed up the slow process of translation from discovery to delivery is to encourage scientists (and practitioners) not to wait for, or rely on, summaries alone, but to seek out current scientific discovery as it is reported in primary empirical scientific journals. Further applications of the CNA approach may provide insights into how scientific communication is defined and measured, and how patterns of scientific communication, such as the lack of cross-citation between discovery and delivery research, influence the delivery of public health science.

No financial disclosures were reported by the authors of this paper. The authors thank Hiie Silmere for her assistance in data collection.

Supplementary data

download text

Appendix.

References

1. 1Brownson RC, Ballew P, Brown KL, et al. The effect of disseminating evidence-based interventions that promote physical activity to health departments. Am J Public Health. 2007;97:1900–1907. CrossRef

2. 2Rabin BA, Brownson RC, Kerner JF, Glasgow RE. Methodologic challenges in disseminating evidence-based interventions to promote physical activity. Am J Prev Med. 2006;31:24–34. Abstract | Full Text | Full-Text PDF (238 KB) | CrossRef

3. 3NCI. The nation's progress in cancer research: an annual report for 2003. National Institutes of Health, USDHHS, NIH Publication No. 04-5536; 2004;.

4. 4Mueller NB, Burke RC, Luke DA, Harris JK. Getting the word out: a trial of multiple dissemination methods. J Public Health Manag Pract. 2008;14:170–176.

5. 5Rogers EM. Diffusion of innovations. 5th ed.. New York: Free Press Glencoe; 2003;.

6. 6Coleman J, Katz E, Menzel H. Medical innovation: a diffusion study. Indianapolis IN: Bobbs-Merrill; 1966;.

7. 7Leischow S, Milstein B. Systems thinking and modeling for public health practice. Am J Public Health. 2006;96:403–405. MEDLINE | CrossRef

8. 8Provan KG, Clark PI, Huerta T. Transdisciplinarity among tobacco harm-reduction researchers: a network analytic approach. Am J Prev Med. 2008;35(2S):S173–S181. Abstract | Full Text | Full-Text PDF (1354 KB) | CrossRef

9. 9Croyle RT. The National Cancer Institute's Transdisciplinary Centers initiatives and the need for building a science of team science. Am J Prev Med. 2008;35(2S):S90–S93. Full Text | Full-Text PDF (60 KB) | CrossRef

10. 10Balas EA, Boren SA. Managing clinical knowledge for health care improvement. In: Yearbook of medical informatics 2000: patient-centered systems. Stuttgart, Germany: Schattauer; 2000;p. 65–70.

11. 11Stokols D, Hall KL, Taylor BK, Moser RP. The science of team science: overview of the field and introduction to the supplement. Am J Prev Med. 2008;35(2S):S77–S89. Abstract | Full Text | Full-Text PDF (174 KB) | CrossRef

12. 12Syme SL. The science of team science: assessing the value of transdisciplinary research. Am J Prev Med. 2008;35(2S):S94–S95. Full Text | Full-Text PDF (43 KB) | CrossRef

13. 13Masse LC, Moser RP, Stokols D, et al. Measuring collaboration and transdisciplinary integration in team science. Am J Prev Med. 2008;35(2S):S151–S160. Abstract | Full Text | Full-Text PDF (389 KB) | CrossRef

14. 14Glens SS. Operant contingencies and the origin of cultures. In: Lattal KA, Chase PN editor. Behavior theory and philosophy. New York: Springer; 2003;.

15. 15Hovell MF, Wahlgren DR, Gehrman CA. The behavioral ecological model: integrating public health and behavioral science. In: DiClemente RJ, Crosby RA, Kegler MC editor. Emerging theories in health promotion practice and research. San Francisco: Jossey-Bass; 2002;.

16. 16Bandura A. Social learning theory. New York: General Learning Press; 1977;.

17. 17USDHHS. The health consequences of involuntary exposure to tobacco smoke: a report of the Surgeon General. USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2006;.

18. 18USDHHS. 11th report on carcinogens, 11th edition. USDHHS, National Toxicology Program. 2005;ntp.niehs.nih.gov/ntp/roc/toc11.html.

19. 19MacMahon B, Alpert M, Salber EJ. Infant weight and parental smoking habits. Am J Epidemiol. 1965;82:247–261. MEDLINE

20. 20Svendsen KH, Kuller LH, Martin MJ, Ockene JK. Effects of passive smoking in the Multiple Risk Factor Intervention Trial. Am J Epidemiol. 1987;126:783–795. MEDLINE

21. 21Kawachi I, Colditz GA, Speizer FE, et al. A prospective study of passive smoking and coronary heart disease. Circulation. 1997;95:2374–2379. MEDLINE

22. 22Lee CH, Ko YC, Goggins W, Huang JJ, Huang MS, Kao EL, et al. Lifetime environmental exposure to tobacco smoke and primary lung cancer of non-smoking Taiwanese women. Int J Epidemiol. 2000;29:224–231. MEDLINE | CrossRef

23. 23Borland R, Chapman S, Owen N, Hill D. Effects of workplace smoking bans on cigarette consumption. Am J Public Health. 1990;80:178–180. MEDLINE | CrossRef

24. 24Wakefield M, Roberts L, Miller C. Perceptions of the effect of an impending restaurant smoking ban on dining-out experience. Prev Med. 1999;29:53–56. MEDLINE | CrossRef

25. 25Emmons KM, Hammond SK, Fava JL, Velicer WF, Evans JL, Monroe AD. A randomized trial to reduce passive smoke exposure in low income households with young children. Pediatrics. 2001;108:18–24.

26. 26Hummon NP, Doreian P. Connectivity in a citation network: the development of DNA theory. Social Networks. 1989;11:39–63.

27. 27Wasserman S, Faust K. Social network analysis: methods and applications. In: Granovetter M editors. Structural analysis in the social sciences. Thousand Oaks CA: Sage; 1994;.

28. 28Lehmann S, Lautrup B, Jackson A. Citation networks in high energy physics. Phys Rev E Stat Nonlin Soft Matter Phys. 2003;68:0261131–0261138.

29. 29Moore S, Shiell A, Hawe P, Haines VA. The privileging of communitarian ideas: citation practices and the translation of social capital into public health research. Am J Public Health. 2005;96:1330–1337.

30. 30Nerur S, Sikora R, Mangalaraj G, Balijepally V. Assessing the relative influence of journals in a citation network. Commun ACM. 2005;48:71–74.

31. 31Cambrosio A, Keating P, Mercier S, Lewison G, Mogoutov A. Mapping the emergence and development of translational cancer research. Eur J Cancer. 2006;42:3140–3148. Abstract | Full Text | Full-Text PDF (2908 KB) | CrossRef

32. 32Americans for Nonsmokers' Rights Foundation (ANRF). In: Bibliography of secondhand smoke studies. Berkeley CA: ANRF; 2005;.

33. 33Byrne F, Chapman S. The most cited authors and papers in tobacco control. Tob Control. 2005;14:155–160. CrossRef

34. 34Fontham ET, Correa P, Reynolds P, et al. Environmental tobacco smoke and lung cancer in nonsmoking women: a multicenter study. J Am Med Assoc. 1994;271:1752–1759.

35. 35Eisner MD, Smith AK, Blanc PD. Bartenders' respiratory health after establishment of smoke-free bars and taverns. J Am Med Assoc. 1998;280:1909–1914.

36. 36Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. CrossRef

37. 37de Nooy W, Mrvar A, Batagelj V. Exploratory social network analysis with Pajek. In: Granovetter M editors. Structural analysis in the social sciences. New York: Cambridge University Press; 2005;.

38. 38Batagelj V, Mrvar A. Networks/Pajek: programs for large network analysis. vlado.fmf.uni-lj.si/pub/networks/pajek/.

39. 39Scott JP. Social network analysis: a handbook. 2nd edition. Thousand Oaks CA: Sage Publications; 2000;.

40. 40Barabasi A-L, Bonabeau E. Scale-free networks. Sci Am. 2003;288:60–69. MEDLINE | CrossRef

41. 41Ware JH, Dockery DW, Spiro A, Speizer FE, Ferris BG. Passive smoking, gas cooking, and respiratory health of children living in 6 cities. Am Rev Respir Dis. 1984;129:366–374. MEDLINE

42. 42Pirkle JL, Flegal KM, Bernert JT, Brody DJ, Etzel RA, Maurer KR. Exposure of the U.S. population to environmental tobacco smoke: the Third National Health and Nutrition Examination Survey, 1988 to 1991. J Am Med Assoc. 1996;275:1233–1240.

43. 43Chilmonczyk BA, Salmun LM, Megathlin KN, et al. Association between exposure to environmental tobacco-smoke and exacerbations of asthma in children. N Engl J Med. 1993;328:1665–1669. MEDLINE | CrossRef

44. 44Colley JR, Holland WW, Corkhill RT. Influence of passive smoking and parental phlegm on pneumonia and bronchitis in early-childhood. Lancet. 1974;2:1031–1034. MEDLINE

45. 45Martinez FD, Antognoni G, Macrı F, et al. Parental smoking enhances bronchial responsiveness in nine year old children. Am Rev Respir Dis. 1988;138:518–523. MEDLINE

46. 46Martinez FD, Cline M, Burrows B. Increased incidence of asthma in children of smoking mothers. Pediatrics. 1992;89:21–26.

47. 47Burchfiel CM, Higgins MW, Keller JB, Howatt WF, Butler WJ, Higgins IT. Passive smoking in childhood (Respiratory conditions and pulmonary function in Tecumseh, Michigan). Am Rev Respir Dis. 1986;133:966–973. MEDLINE

48. 48Jarvis M, Tunstall-Pedoe H, Feyerabend C, Vesey C, Salloojee Y. Biochemical markers of smoke absorption and self reported exposure to passive smoking. J Epidemiol Community Health. 1984;38:335–339. CrossRef

49. 49Benowitz NL. Cotinine as a biomarker of environmental tobacco smoke exposure. Epidemiol Rev. 1996;18:188–204.

50. 50Hanrahan JP, Tager IB, Segal MR, et al. The effect of maternal smoking during pregnancy on early infant lung-function. Am Rev Respir Dis. 1992;145:1129–1135. MEDLINE

51. 51O'Connor GT, Weiss ST, Tager IB, Speizer FE. The effect of passive smoking on pulmonary-function and nonspecific bronchial responsiveness in a population-based sample of children and young-adults. Am Rev Respir Dis. 1987;135:800–804. MEDLINE

52. 52Kottke TE, Hill C, Heitzig C, et al. Smoke-free hospitals: attitudes of patients, employees, and faculty. Minn Med. 1985;1:53–55.

53. 53Stillman FA, Becker DM, Swank RT, et al. Ending smoking at the Johns Hopkins Medical Institutions (An evaluation of smoking prevalence and indoor air pollution). J Am Med Assoc. 1990;264:1565–1569.

54. 54Hole DJ, Gillis CR, Chopra C, Hawthorne VM. Passive smoking and cardiorespiratory health in a general population in the west of Scotland. Br Med J. 1989;299:423–427.

55. 55Longo DR, Brownson RC, Johnson JC, et al. Hospital smoking bans and employee smoking behavior: results of a national survey. J Am Med Assoc. 1996;275:1252–1257.

56. 56Eisner MD, Yellin EH, Henke J, Shiboski SC, Blanc PD. Environmental tobacco smoke and adult asthma—the impact of changing exposure status on health outcomes. Am J Respir Crit Care Med. 1998;158:170–175.

57. 57Okah FA, Choi WS, Okuyemi KS, Ahluwalia JS. Effect of children on home smoking restriction by inner-city smokers. Pediatrics. 2002;109:244–249.

58. 58Stave GM, Jackson GW. Effect of a total work-site smoking ban on employee smoking and attitudes. J Occup Med. 1991;33:884–890. MEDLINE

59. 59Garland C, Barrett-Connor E, Suarez L, Criqui MH, Wingard DL. Effects of passive smoking on ischemic heart disease mortality of nonsmokers. Am J Epidemiol. 1985;121:645–649. MEDLINE | CrossRef

60. 60Aronow WS. Effect of passive-smoking on angina-pectoris. N Engl J Med. 1978;299:21–24. MEDLINE

61. 61Handcock MS, Hunter DR, Butts CT, Goodreau SM, Morris M. statnet: software tools for the representation, visualization, analysis and simulation of network data. J Stat Softw. 2008;24:1548–7660.

62. 62Morris M, Handcock M, Hunter DR. Specification of exponential-family random graph models: terms and computational aspects. J Stat Softw. 2008;24:1548–7660.

63. 63Hunter DR, Handcock MS, Butts CT, Goodreau SM, Morris M. ergm: a package to fit, simulate, and diagnose exponential-family models for networks. J Stat Softw. 2008;24(3):.

64. 64U.S. Environmental Protection Agency (EPA). Respiratory health effects of passive smoking. Washington DC: EPA, Office of Research and Development, Office of Health and Environmental Assessment; 1992;EPA/600/6-90/006F.

65. 65USDHHS. The health consequences of involuntary smoking: a report of the Surgeon General. USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1986;.

66. 66National Research Council Committee on Passive Smoking. Environmental tobacco smoke: measuring exposures and assessing health effects. Washington DC: National Academy Press; 1986;.

67. 67USDHHS. The health consequences of smoking for women: a report of the Surgeon General. USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1979;.

68. 68California Environmental Protection Agency. Health effects of exposure to environmental tobacco smoke. California EPA: Office of Environmental Health Hazard Assessment; 1997;.

69. 69USDHHS. Reducing the health consequences of smoking: 25 years of progress: a report of the surgeon general. USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1989;.

70. 70Australian National Health and Medical Research Council. Effects of passive smoking on health. Canberra, Australia: Australian government Publishing Service; 1987;.

71. 71Fielding JE, Phenow KJ. Health effects of involuntary smoking. N Engl J Med. 1988;319:1452–1460. MEDLINE

72. 72Lee PN. Misclassification of smoking habits and passive smoking: a review of the evidence. Berlin NY: Springer-Verlag; 1988;.

73. 73Lee PN. Environmental tobacco smoke and mortality. New York: Karger; 1992;.

74. 74Lefcoe NM, Ashley MJ, Pederson LL, Keays JJ. The health risks of passive smoking (The growing case for control measures in enclosed environments). Chest. 1983;84:90–95. MEDLINE | CrossRef

75. 75Siegel M. Involuntary smoking in the restaurant workplace: a review of employee exposure and health effects. J Am Med Assoc. 1993;270:490–493.

76. 76Weiss ST, Tager IB, Schenker M, Speizer FE. State of the art: the health effects of involuntary smoking. Am Rev Respir Dis. 1983;128:933–942. MEDLINE

77. 77Wynder EL, Goodman MT. Smoking and lung cancer: some unresolved issues. Epidemiol Rev. 1983;5:177–207.

78. 78Glantz SA, Parmley WW. Passive smoking and heart disease: epidemiology, physiology, and biochemistry. Circulation. 1991;83:1–12. MEDLINE

79. 79National Cancer Institute. Greater than the sum: systems thinking in tobacco control, Tobacco Control Monograph No. 18. Bethesda MD: USDHHS, NIH, National Cancer Institute; 2007;NIH Pub. No. 06-6085.

80. 80Economos D, Brownson RC, DeAngelis MA, et al. What lessons have been learned from other attempts to guide social change?. Nutr Rev. 2001;59:S40–S56. MEDLINE

a School of Public Health, Saint Louis University, Saint Louis, Missouri

b George Warren Brown School of Social Work, Washington University, Saint Louis, Missouri

c Center for the Study of Drug Development, Tufts University, Boston, Massachusetts

Address correspondence and reprint requests to: Jenine K. Harris, PhD, Saint Louis University School of Public Health, 3545 Lafayette Avenue, Suite 300, Saint Louis MO 63104

The full text of this article is available via AJPM Online at www.ajpm-online.net.

PII: S0749-3797(09)00154-8

doi:10.1016/j.amepre.2009.01.039

*Note: Add-ons open in a new browser window. Please disable any pop-up blocking software before viewing Add-ons.

12 of 16