American Journal of Infection Control 43 (2015) 121-6

Contents lists available at ScienceDirect

American Journal of Infection Control

American Journal of Infection Control

journal homepage: www.ajicjournal.org

Major article

Institutional origins of health careeassociated infection knowledge: Lessons from an analysis of articles about methicillin-resistant Staphylococcus aureus published in leading biomedical journals from 1960-2009 Fabio Rojas PhD a, *, W. Carson Byrd PhD b, Sanjay Saint MD, MPH c, d, e a

Department of Sociology, Indiana University, Bloomington, IL Department of Pan-African Studies, University of Louisville, Louisville, KY Hospital Outcomes Program of Excellence, Department of Veterans Affairs, Ann Arbor Healthcare System, Ann Arbor, MI d Department of Internal Medicine, University of Michigan Medical School, Ann Arbor, MI e Department of Veterans Affairs/University of Michigan Patient Safety Enhancement Program, Ann Arbor, MI b c

Key Words: Methicillin-resistant Staphylococcus aureus Citation analysis Biomedical research Databases History Organizations

Background: Biomedical research journals are important because peer reviewed research is viewed as more legitimate and trustworthy than non-peer reviewed work. Therefore, it is important to know how knowledge transmitted through academic biomedical journals is produced. This article asks if some organizations are more likely to produce research than others and if organizational setting is linked with an article’s impact, as measured by citation counts. Methods: Using research on methicillin-resistant Staphylococcus aureus (MRSA) as a case study, we examined the role that hospitals, universities, public health agencies, and other organizations have in shaping an emerging research area. We collected public data on the organizational affiliations of researchers who authored 1,721 articles in general interest and selected specialty journals. Results: MRSA research appears to have evolved in stages that require the participation of different types of organizations. Additionally, our analyses indicate that an author’s organizational affiliation predicts citation counts, even when controlling for other factors. Conclusion: Organizations vary greatly in their ability to produce research, and this should be taken into account by those who manage or award funds to research organizations. Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved.

* Address correspondence to Fabio Rojas, PhD, Department of Sociology, Indiana University, Ballantine Hall 744, 1020 E Kirkwood Ave, Bloomington, IN 47405. E-mail address: [email protected] (F. Rojas). Funding/support: This work was supported by the Robert Wood Johnson Foundation while the first author was a Robert Wood Johnson Foundation Scholar in Health Policy Research. We received financial support from the College of Arts and Sciences at Indiana University, The Robert Wood Johnson Foundation, and Department of Internal Medicine at the University of Michigan. Potential conflicts of interest: Dr. Rojas has received honoraria and speaking fees for lecturing at various universities and private organizations on the subject of organizational change. Dr. Byrd has no conflicts of interest to disclose. Dr. Saint has received numerous honoraria and speaking fees from academic medical centers, hospitals, group-purchasing organizations (eg, Veterans Health Administration, Premier), specialty societies, state-based hospital associations, and nonprofit foundations (eg, Michigan Health and Hospital Association, Institute for Healthcare Improvement) for lectures about catheter-associated urinary tract infection, leadership and followership, and implementation science.

Every year, public agencies, medical schools, and for-profit firms spend billions of dollars on biomedical research that creates the knowledge used in clinical practice. For this reason, health policy scholars examine the institutional underpinnings of medical knowledge. For example, researchers explore the knowledge produced at for-profit firms compared with nonprofit agencies to determine whether private funding of university biomedical research undermines the integrity of the research process.1-3 The historical and sociological literature on biomedical research asks how personal interests, laboratory conditions, and scientists’ social background shapes what they research.4-6 In contrast, less is known about whether some organizations have an advantage in producing biomedical research and publications. Little is known about how the setting of biomedical research (eg, university medical schools vs public health agencies) leads to systematic differences in the volume and impact of research.

0196-6553/$36.00 - Copyright Ó 2015 by the Association for Professionals in Infection Control and Epidemiology, Inc. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajic.2014.10.012

122

F. Rojas et al. / American Journal of Infection Control 43 (2015) 121-6

We address this issue by comparing the scientific output (in the form of publications) from organizations that produce biomedical research. Because studying the entirety of biomedical research is untenable, we focus on one area: the study of methicillin-resistant Staphylococcus aureus (MRSA). We chose MRSA research for both substantive and methodological reasons. Substantively, MRSA and other health careeassociated infections (HAIs) are a known threat to public health, with thousands of people injured or killed each year by MRSA and other antibiotic-resistant infections.7 Methodologically, examining MRSA research is beneficial because of the field’s relatively low barriers to entry; MRSA research does not always require the financial or bureaucratic resources associated with the most well-endowed research universities. In the remainder of this article, we briefly discuss the MRSA research area and the organizations that produce MRSA research. Then we present an analysis of the MRSA literature published in leading journals from 1960-2009. Through a 3-pronged analysis, we discovered that MRSA research evolved in distinct phases requiring organizations of varying research capacity. We also show that hospital- and university-based researchers produce more highly cited articles.

THE CASE: MRSA RESEARCH MRSA has long been recognized as a major public health issue.11,12 MRSA and other drug-resistant infections are responsible for thousands of patient deaths per year and billions of dollars in medical expenses.7,13 Starting in the 1990s, researchers grew concerned over recently discovered community-based MRSA, which suggested that antibiotic-resistant infections were no longer an idiosyncratic feature of the hospital.14,15 Research on the multiple strains of MRSA raised concerns among microbiologists, epidemiologists, and hospital infection specialists because some strains were resistant to even the most powerful antibiotics. Furthermore, there was a dramatic increase in the total number of MRSA infections in the United States. As a response to the proliferation of MRSA strains, community-based MRSA, and the resilience of MRSA to newer drugs, various public institutions and private foundations directed billions of dollars to the control, management, and prevention of the disease, leading to a dramatic increase in MRSA research.7,13,15,16 Although enormous resources have been expended to study the basic science of the bacteria and infectious process, the conditions that facilitate and inhibit transmission of the bacteria, and the development of treatments for infected patients, little research addresses the institutional context of where these projects are conducted.

FIELD OF BIOMEDICAL RESEARCH METHODS Biomedical research exists in a variety of settings ranging from solo medical practitioners who may report case studies to major government institutions, such as the Centers for Disease Control and Prevention (CDC), which conduct large, complex studies. In the United States, the rise of medical research is synonymous with the growth of institutions, such as medical schools; academic departments that specialize in human biology, biochemistry, and medicine; and government agencies (eg, National Institutes of Health) that fund, evaluate, and disseminate research findings. The for-profit sector generates a great deal of research as well, especially companies producing drugs and medical devices. As a result of for-profit sector growth and its influence on academia, the literature now documents differences between research generated in for-profit and that produced in nonprofit environments. Additionally, a robust debate exists about the effects of private funding on research findings.3,8 This brief description of the biomedical research field omits 2 important factors motivating the research presented in this article. First, an organization’s role in research changes over time because there is a research cycle.9,10 Those resources that are needed for a newly created specialty are different than those needed for longstanding mature fields. This suggests that the organizational context of research might change as a result. Young research specialties are more open to unaffiliated scientists, whereas researchers affiliated with institutions that can support large and ambitious research projects may dominate more mature specialties. Second, the content of the research also influences whether some organizations possess an advantage in conducting it. For example, highly complex medical procedures (eg, organ transplants) require enormous resources that only the largest universities and hospitals can provide. In contrast, other fields require minimal resources to study (eg, MRSA research), which often revolve around disease control strategies that require resources that are available to many researchers, not only those in leading medical centers or teaching hospitals. An independent researcher, for example, that wishes to conduct research on a hand hygiene intervention and subsequently publish the results does not require the formidable resources needed to develop pharmaceutical treatments and bring the results to publication.

Study overview and search strategy To understand which organizations participate in MRSA research, we conducted an automated Medical Subject Headings (MeSH) term search of PubMed and collected data on the organization listed by authors as their institutional affiliation. We wrote a program that gathered data on MRSA from 1960-2009 using the MeSH tree number B03.510.400.790.750.100.500. The exact phrase for this number is “methicillin-resistant Staphylococcus aureus.” Although there are broader Staphylococcus terms, our specific interest is in MRSA. The program was run through the Python language (Python Software Foundation, Beaverton, OR), and the data stored in Structured Query Language were then exported into Microsoft Excel (Microsoft Corp, Redmond, WA). We focused on MeSH terms because they indicate that the authors believed that the article substantially addressed MRSA. In contrast, we believe that searching the full text of an article could yield tangential mentions of MRSA. Our analysis focuses on articles that appeared in the following leading clinically oriented biomedical journals: American Journal of Infection Control, Annals of Internal Medicine, Archives of Internal Medicine, British Medical Journal, Clinical Infectious Diseases, Infection Control and Hospital Epidemiology, Journal of Hospital Infection, Journal of the American Medical Association, Lancet, and New England Journal of Medicine. These journals were selected by both expert guidance and impact scores. Our goal was to select journals that would have high impact and that would be read by professionals who deal with HAIs. Therefore, this is not a survey of all articles written on the topic of MRSA. Our search generated 1,721 articles. The results included all items in PubMed indexed under the search term (ie, editorials, letters, complete research articles). We then extracted the article title, abstract, page numbers, names of all authors, and journal title. We used the lead author affiliation as a representation of where the research had been conducted. To extract this information, we had a team of undergraduate research assistants read all 1,721 articles. Unfortunately this could not be automated because PubMed, as with most other databases, does not consistently report author affiliations. Furthermore, we found that many names, in particular those from outside the United States, were incorrectly or

F. Rojas et al. / American Journal of Infection Control 43 (2015) 121-6

incompletely extracted from the article’s PubMed citation when automated. For example, our program inaccurately extracted the names of European institutions that had non-English characters. Once we compiled the data, we coded each lead author’s affiliation into 4 possibly overlapping categories: hospitals; universities; public health agencies (eg, CDC and National Institutes of Health in the United States, National Health Service in the United Kingdom); and other. This last category was for lead authors without an organizational affiliation or who worked at for-profit agencies, think tanks, or public agencies that focused on non-health issues (eg, Department of Defense). As previously mentioned, these categories are not mutually exclusive. Many authors report appointments in multiple organizations (eg, school of medicine and a hospital), and they were assigned to both categories. We note that university denotes authors who reported an appointment (eg, professor, postdoctoral fellow, researcher, student) in a school of medicine, public health, biology, chemistry, or a related unit in a related social science, biological science, physical science, professional school, or research center. The category of university-affiliated researcher does not include authors who work in a hospital that is simply owned by a university system. We made this decision based on the knowledge that some hospitals associated with a university do not have medical education programs (eg, a residency), and especially for large urban hospitals, there is a great number of physicians who use the university title, despite no responsibilities to the university. Therefore, we limited this category (university) to authors who have a clear academic role in a degree granting unit or scientific research center. Statistical analyses Descriptive statistics were calculated to summarize the author affiliations, general article attributes, and appearance of MRSA articles in the 10 selected peer reviewed journals. To examine the factors contributing to the impact of MRSA articles, we used ordinary least-squares linear regression models to estimate an article’s total number of citations. The dependent variable was the total number of citations as reported in Google Scholar, the most comprehensive scholarly database in the world. We focused on Google Scholar citation counts because other databases report relatively few citations (eg, Scopus, PubMed) or have been criticized as being too restrictive because they do not include the widest range of scholarly publications, such as books or lower prestige journals (eg, Web of Science, formerly known as ISI Web of Knowledge).14,17 The primary independent variable was the lead author’s affiliation. We used the same codes for the author’s employer (hospital, university, government health agency, other), where 1 was assigned to each author affiliated with each organization type, and 0 was assigned if they were not. We included 2 control variables to account for heterogeneity. First, we included a dichotomous variable indicating that the author works in the United States. This helps address the possibility that an article may receive attention because of the nationality of the author or the national association of the institution. Second, it is possible that people may prefer to read or cite articles in general journals. Flagship journals in a field (eg, New England Journal of Medicine) may be seen as more prestigious or as having higher quality articles because they have lower acceptance rates. To account for this, we included a dichotomous variable to indicate that a journal is specifically focused on infectious diseases. To examine the relative impact of an article we used logistic regression to determine whether an article is highly cited. The dependent variable of highly cited refers to an article whose total citation count is 1 SD above the average citation count for all articles within the same time period (pre-1990, 1990s, 2000s) of

123

Table 1 Methicillin-resistant Staphylococcus aureus research characteristics and distribution in selected medical and biomedical journals Variable

Mean

Author affiliation University Hospital Public health agency United States based General article attributes Average year published Average years since publication Total citations Average number of citations Highly cited article Government funded

55% 27% 14% 36%

.50 .44 .34 .44

2001 8.01 30.69 3.87 23% 26%

6.74 6.74 32.65 5.41 .42 .44

Decade published*

Proportion of selected journals containing MRSA-specific articles American Journal of Infection Control Annals of Internal Medicine Archives of Internal Medicine British Medical Journal Clinical Infectious Diseases Infection Control and Hospital Epidemiology Journal of Hospital Infection Journal of the American Medical Association Lancet New England Journal of Medicine Total

SD

Pre-1990s 1990s 2000s

Total

08% .28

58

161

358

577

02% 02% 3% 17% 25%

.15 .14 .18 .37 .44

17 4 8 28 0

91 38 19 35 15

328 246 115 18 45

436 288 142 81 60

34% .47 02% .14

13 4

9 4

17 26

33 34

05% .21 02% .14

6 5 143

3 7 382

23 32 20 32 1,196 1,721

Note: Data compiled from 1,721 articles from selected leading clinically oriented biomedical journals. *Numbers represent article counts.

the article’s publication. We do this to account for citation pattern changes over time because norms for scientific research change. RESULTS General publication characteristics and trends Descriptive statistics were generated to identify various aspects of MRSA research (Table 1). Our analysis shows that 55% of the authors were university affiliated, 27% were hospital affiliated, and 14% were public health agency affiliated. Of the articles collected, 36% of all authors were based in the United States. Half of all articles in our sample were published in 2001 or later, indicating that the average article was published 8 years prior to the time of our data collection. The average article was cited approximately 31 times by others during the entire time period. Given that each article was in print for approximately 8 years, the average yearly citation is just under 4. Approximately 23% of articles were considered highly cited publications. Of note, 26% of all articles collected during this study were part of government-funded research projects. Of the 10 journals examined, 33% of all MRSA-related research articles published from 1960-2009 appeared in the Journal of Hospital Infection, 25% were published in Infection Control and Hospital Epidemiology, 17% were published in Clinical Infectious Diseases, 8% were published in American Journal of Infection Control, 5% were published in Lancet, 3% were published in British Medical Journal, and 2% were published in the 4 remaining journals. Figure 1 shows the yearly count of MRSA articles by author affiliation. A few trends are apparent. First, the overall frequency of articles was low before 1990. Second, for all time periods, the frequency of articles was highest among hospital-affiliated authors,

124

F. Rojas et al. / American Journal of Infection Control 43 (2015) 121-6

Fig 1. Yearly count of scientific articles addressing methicillin-resistant Staphylococcus aureus in selected leading biomedical journals, 1970-2009 (N ¼ 1,707). Data was compiled from 1,718 articles (3 articles from the 1960s were not included to enhance legibility) from selected leading clinically oriented biomedical journals. The institutional type indicates the affiliation of the lead author. Authors may have >1 affiliation. The residual category of others indicates medical practitioners affiliated with for-profit agencies, public agencies that address issues other than health, or no affiliation.

followed by those with an academic institution affiliation (eg, a medical school). The frequency of articles produced by authors affiliated with public health agencies or other institutions was comparatively lower; however, a slight increase is seen in the last few years for these groups of authors. Figure 2 shows the probability that an article is highly cited according to the first author’s affiliation. First, the likelihood that an article from the other category is highly cited drops greatly over 40 years. Although 30% of such articles are highly cited before 1990, by the 2000s only 5% of these articles were the most highly cited. Second, universities and hospitals were the greatest source of cited articles throughout the 1990s, but by 2000 public health agencies caught up. Article impact Table 2 shows the total article citation count for all years and specific time periods. When we looked at all years, we found that the organizations that were most likely to produce highly cited articles were public health agencies (11 more citations than the residual categories); hospital-based research received 8 more citations, and articles with university-affiliated authors received 6 more citations. American authors received 17 more citations on average when compared with non-American authors, and articles in specialty journals (eg, Clinical Infectious Diseases) received 11 fewer citations compared with articles in general journals. No statistically significant effects of author affiliation on total counts existed prior to the 1990s; however, university affiliation was associated with 3 more citations compared with other institutions during the 1990s. The trend shifted in the 2000s, with articles produced from authors affiliated with hospitals and public health agencies receiving 12 and 17 more citations, respectively, when compared with articles produced by authors from other institutions (eg, for-profits, think tanks). DISCUSSION Our analysis indicates that where MRSA research is produced matters in how it is cited. First, the establishment of specialty journals preceded enormous growth in MRSA research, but articles

Fig 2. Proportion of highly cited articles by institutional affiliation of the lead author. Data were compiled from 1,721 articles from selected leading clinically oriented biomedical journals. Highly cited is defined as an article whose total citation count is 1 SD above the mean for all articles written within its time period (eg, before 1990, the 1990s, after 2000). For articles published before the 1990s, during the 1990s, and the 2000s, mean and SDs are 38.4  35.3, 39.7  34.6, and 26.6  30.8, respectively.

in these journals were less cited and less likely to become the most cited, despite publishing the greatest number of MRSA articles. This may be because of the specialist audiences of these journals. On the other hand, public health agencies produced few articles, but these articles tended to garner many citations in comparison with articles produced by researchers at hospitals and universities. Third, an organization’s role in the production of highly cited articles changes greatly with the volume of the research literature. Prior to the 1990s, unaffiliated researchers were much more likely to have an impact. Throughout the 1990s, hospitals and universities produced more highly cited research examining MRSA. These findings suggest that time leads to the professionalization of research. When research fields begin, the scientific problems are accessible to anyone interested, irrespective of location. Little specialized knowledge is required to conduct research and publish in that area. As time passes and the field attracts more attention, scientists are likely to specialize, and publication becomes more competitive. Similarly, as the MRSA research field developed, physicians and scientists needed to acquire extra training and knowledge and ultimately work in an environment (eg, university) that possessed the resources needed for ambitious, large-scale projects to contribute to the field of MRSA research. Therefore, MRSA research has evolved into a field that is conducted by highly trained specialists and is based on a body of highly specialized knowledge.18-21 This MRSA-related knowledge developed along with a system of evaluation and communication of results and the socialization of new scientists to the field. The trends reported in this study are indicative of a profound change within the biomedical research community. Initially, MRSA and other antibiotic-resistant organisms were viewed as idiosyncratic health issues by most practitioners.15 Our results suggest that researchers outside the university system were most likely to conduct research on MRSA in the beginning. Once policymakers and health care practitioners viewed MRSA as a priority because of its rising incidence, more research was conducted in settings such as medical schools and microbiology departments. The increasing attention paid to MRSA by the biomedical research community is reflected in the growth of specialty journals. The major infectious diseases journals that we examined did not begin publication until nearly 30 years after the first report of MRSA in 1949. For example, Clinical Infectious Diseases, a frequent forum for HAI and MRSA articles, did not begin publication until 1979; other leading specialty journals were founded later. The lack of specialized outlets for research may have suppressed the dissemination of research results, especially those not deemed relevant to a general biomedical research audience. Once greater attention

F. Rojas et al. / American Journal of Infection Control 43 (2015) 121-6

125

Table 2 Linear regression models of an article’s total citation count and of being a highly cited article Linear regression model of an article’s total citation count All years Covariate Author affiliation other (reference) Hospital University Public health agency American Specialty journal Constant n R2

Pre-1990

1990s

2000s

B

P value

B

P value

B

P value

B

P value

e 8.4 6.2 11.3 16.8 11.6 23.7 1,721 0.08

e

Institutional origins of health care-associated infection knowledge: lessons from an analysis of articles about methicillin-resistant Staphylococcus aureus published in leading biomedical journals from 1960-2009.

Biomedical research journals are important because peer reviewed research is viewed as more legitimate and trustworthy than non-peer reviewed work. Th...
371KB Sizes 0 Downloads 3 Views