Publication Metrics and Record of Pharmacy Practice

Pharmacy Education

Publication Metrics and Record of Pharmacy Practice Chairs Dennis F Thompson, Erin C Callen, and Milap C Nahata

eaching, research, and service are the historic goals of institutions of BACKGROUND: Scholarship is an essential component of academic pharmacy. Department chairs are considered role models and mentors to junior faculty, but higher learning. Research, traditionally their publication record has not been documented. understood as the scholarship of discovOBJECTIVE: To quantify publication patterns of pharmacy practice chairs in ery, is 1 of 4 types of scholarship progeneral and, specifically, at health sciences center (HSC)-based versus 1 posed by Boyer. This model proposed non–HSC-based and public versus private colleges of pharmacy. equal consideration of the scholarship of METHODS: Pharmacy practice chairs were identified using the 2006–2007 roster discovery, integration, teaching, and apof faculty from the American Association of Colleges of Pharmacy. Of the 89 plication in faculty academic progrescolleges of pharmacy in the roster, 11 listed no pharmacy practice chair and 5 sion. Regardless of the type of scholarlisted more than one. Data were collected on the remaining 73 schools by searching each chair’s name on PubMed and Web of Science (WoS). Data on ship, however, peer-reviewed publicatotal publications, publications per year, total citations, citations per article, htions are an essential component of index, and m quotient were collected. 2-4 scholarship. Indeed, publications are RESULTS: A total of 2394 papers published by 73 pharmacy practice chairs were the archival record of a discipline and found in a search of PubMed. The mean number of total publications per chair was provide a historical record of its progres33 (95% CI 21 to 44). The mean number of publications per year was 1.4 using sion. PubMed and 1.6 using WoS. Mean h-index was 8.0 (95% CI 6.3 to 9.6). Mean Department chairs are usually responnumber of total citations was 410 (95% CI 252 to 568). Thirty-three percent (n = 24) had less than 10 lifetime publications and 18% had more than 50 lifetime sible for evaluation and mentoring of all publications. HSC-based chairs averaged 51.3 papers while non–HSC-based departmental faculty. Whether this is chairs averaged 19.1 (p < 0.01). Similar data were found for total citations (HSC = done personally or through delegation, 673 vs non-HSC = 216; p < 0.001). Public school chairs had an average of 41.5 chairs provide oversight, direction, and articles cited on PubMed, versus 15 for private school chairs (p < 0.01). Public support for junior faculty as they develschool chairs had an average h-index of 9.7 versus 4.4 for private school chairs (p < op their scholarly pursuits. They evaluate 0.001), and an average of 9.2 citations per article compared with 5.2 for private school chairs (p < 0.001). scholarship for annual performance reCONCLUSIONS: These data provide a normative pattern of publication metrics and views and for tenure and promotion conrecord for pharmacy practice chairs and demonstrate marked variability in siderations. In their role as academic leadscholarly productivity. ers and mentors, chairs should demonKEY WORDS: chair, publications, scholarship. strate skills in conceptualizing and Ann Pharmacother 2009;43:268-75. completing scholarly papers. The recent proliferation of new colleges of pharmacy Published Online, 20 Jan 2009, www.theannals.com, DOI 10.1345/aph.1L400 and the “graying” of the academy have increased the demand for pharmacy practice put added pressure on maintaining quality in the pharmacy faculty and qualified department chairs, which has potentially practice faculty and chair ranks. diluted the pool of qualified candidates. These factors have Bibliometric methods can be used to analyze the scholarly output of faculty and scientists.5,6 In the past, the disAuthor information provided at the end of the text. cussion of a faculty member’s scholarly output might have

T

268

n

The Annals of Pharmacotherapy

n

2009 February, Volume 43

Downloaded from aop.sagepub.com by guest on October 11, 2013

www.theannals.com

involved simply the number of papers published. As the academic world became more sophisticated, analysis of scholarly output grew to include such factors as publication types (eg, research article, review, editorial, letter, case report), distinction between peer-reviewed versus non–peer-reviewed papers, number of papers for which the faculty member is the first author, and citation analysis. Hirsch5 introduced the h-index (defined as the number of papers authored by a scholar with at least h citations each) as an easily computable index that gives a broad estimate of the importance, significance, and impact of a faculty member’s overall scholarly contributions. More recently, a series of new bibliometric measures have been described in the literature.7 The m quotient, g index, h(2) index, a index, m index, r index, ar index, and the hw index have all been added to the bibliometric vocabulary. As these new measures are added to the armamentaria of bibliometric analysis, it is important to develop discipline-specific normative values to bring context to the numbers. Knowledge from databases such as PubMed and Web of Science (WoS) about publications, citations, the hindex, and m quotient is sparse in pharmacy and limited data are available from surveys of self-reported publications of pharmacy practice chairs. Our goal was to document, using the PubMed and WoS databases, the publication metrics and record of pharmacy practice chairs in general and, specifically, at health sciences center (HSC)based versus non–HSC-based and public versus private colleges of pharmacy. Methods

Publication data were also obtained for each chair using the WoS database. Each chair name was searched using the author search feature on the menu. Using this feature, we narrowed these searches by both type and place of publication. The author index generated total citations to published works, h-index, and citations per article. We also calculated the m quotient by dividing the h-index by the number of years the individual scholar has published, according to the equation m = h/n, where h is the h-index and n is the number of years between the first published paper and the present. Searches were not restricted by language, type of article, or year of publication. Information from the AACP Web site, which lists colleges by various categories, determined the classification of colleges as public, private, HSC-based, or non–HSC-based institutions. Additionally, data were also collected on the number of publications for which the chair was the first author and the first year of publication. Publication rates (ie, publications per year) were calculated by dividing all publications identified in the search by the number of years the author had published. The number of years was determined by subtracting the year of first publication from the year 2007. Descriptive statistics such as means, medians, standard deviations, and confidence intervals were determined. Comparisons of data were performed by the Mann-Whitney U test with an a priori statistical significance level of p less than 0.01. Results A total of 2394 articles published by 73 pharmacy practice chairs were found on PubMed. Table 1 lists the descriptive statistical parameters for each of the main scholarship factors collected. Mean publication rates were 1.4/year with PubMed and 1.6/year with WoS. The mean lifetime citations per article and h-index were 7.9 and 8, re-

Chairs were identified using the 2006 –2007 roster of faculty from the American Association of Colleges of Pharmacy (AACP). Of the 89 colleges of pharmacy listed in the roster, 11 schools listed no pharmacy practice chair and 5 listed more than one. These 16 schools were eliminated from consideration, leaving 73 schools that were included in our study. All searches were completed during December Table 1. Publication Metrics for Pharmacy Practice Chairsa 2007 and January 2008. Searches were not re95% CI Publication Metric Median Mean on the Mean High stricted in any way (eg, year of publication, publication type, language). We collected all PubMed publications 19 33 21 to 44 355 (lifetime), n publication data on 73 pharmacy practice PubMed publications/year, n 0.8 1.4 0.98 to 1.8 11.5 chairs by searching their names on PubMed WoS publications (lifetime), n 20 37 23 to 51 437 (1965–January 2008). Most searches were unWoS publications/year, n 0.8 1.6 1.1 to 2.1 14 complicated, although some publication lists First author (lifetime), % 40 41.6 36 to 48 100 had multiple authors with the same name, parTotal citations (lifetime), n 128 410 252 to 568 3683 ticularly if the chair had a common name. DeCitations/article, n 6.7 7.9 6.2 to 9.6 37 cisions on inclusion were made using data h-Index 7.0 8.0 6.3 to 9.6 32 available from the abstracts such as subject m Quotient 0.3 0.36 0.3 to 0.4 1.1 matter of the article, co-authors, or author’s WoS = Web of Science. address. When necessary, we reviewed the a N = 73. original article to clarify the author(s). www.theannals.com


n


n

Low 0 0 0 0 0 0 0 0 0

269

DF Thompson et al.

spectively. Table 2 compares HSC-based and non–HSCbased schools, as well as public and private colleges of pharmacy. Pharmacy practice chairs from public schools had statistically significantly higher scholarship parameters than those from private schools in all categories except total citations to works published. Similarly, pharmacy practice chairs from HSC-based schools had significantly higher scholarship parameters than chairs from non–HSC-based schools in all categories except citations per article and h-index. Figure 1 illustrates the total lifetime PubMed publications for pharmacy practice chairs; 33% (n = 24) of the chairs published fewer than 10 lifetime publications, while 18% (n = 13) had published more than 50. Figure 2 shows the total lifetime citations from published works for pharmacy practice chairs; 41% (n = 30) had fewer than 100 lifetime citations to their published works, while 10 chairs had over 1000 lifetime citations of their articles. Both figures demonstrate a partial U distribution that illustrates the significant variability in these data. Figure 3 shows the hindex distribution for pharmacy practice chairs. The mean h-index was 8; almost half (44%) of the chairs had an h-index of 5 or less and 6 chairs had an h-index above 21.

The clinical pharmacy movement began in the 1960s. Concurrent with its development was the concern for the growing ranks of pharmacy practice faculty with little research training.3 Departments of pharmacy practice, while often the largest group within a college of pharmacy, have been challenged to meet the competing demands of substantial teaching, clinical practice, research, and service. Mentoring programs have been proposed as a method to improve junior faculty success as scholars and, at the same time, improve faculty retention.10 Whether done personally or through delegation, chairs have the responsibility to provide mentorship to departmental faculty. They must also evaluate the scholarly performance of faculty for annual reviews or for tenure and promotion purposes. Therefore, chairs must possess the skills to lead departmental faculty through the understanding of the research process and of publishing articles in peer-reviewed journals. A track record of successful scholarship would therefore seem to be essential for a department chair of pharmacy practice. Our data highlight the significant variability in scholarship metrics of current pharmacy practice chairs and suggest that not all chairs have been productive in publishing in the biomedical literature. Discussion HSC-based pharmacy practice chairs had significantly more publications than did non–HSC-based chairs. It Scholarship has been defined as the creation, discovery, would seem reasonable that HSC-based colleges may offer advancement, or transformation of knowledge. Research more resources and opportunities for, or place greater emand scholarship have been at the forefront of pharmacy edphasis on, conducting research and scholarship than do 2 ucation since the early part of the 20th century. However, 2-4 non–HSC-based colleges. Similarly, public colleges of many authors credit the Argus Commission report of 8 pharmacy are more likely to have a research and scholar1980 for linking scholarship to lifelong learning skills and ship mandate than private colleges, although this is not establishing the responsibility of colleges of pharmacy to universally true. Our data show that pharmacy practice conduct scholarly activity. This theme has been continued chairs from public colleges had significantly more publicaby the AACP’s ongoing Research and Graduate Affairs tions than did those from private colleges; these data are Committee. Further, the Accreditation Council for Pharconsistent with a previous study reporting all full-time macy Education guidelines now state that all pharmacy pharmacy faculty publications counted between 1976 and faculty should be scholars.9 1992.11 These authors found significant differences between HSC-based and non–HSCbased colleges of pharmacy and between pubTable 2. Average Publication Rates for Pharmacy Practice Chairs lic and private schools. Interestingly, publication Comparing HSC-based vs non–HSC-based and rates from pharmacy deans of HSC-based Public vs Private Colleges of Pharmacy schools were not different from those of pharPublication HSC non-HSC Public Private Metric (n = 31) (n = 42) (n = 49) (n = 24) macy deans of non–HSC-based schools, but publication rates from deans of public schools PubMed publications, n 51.3 19.1a 41.5 15a were significantly higher than those of deans 2.1 0.9a 0.76a PubMed publications/year, n 1.7 from private schools.12 WoS publications, n 61.3 19.6b 48.7 14.2b WoS publications/year, n 2.5 0.9b 2.0 0.7a We found that 42% of the total publications Total citations, n 673 216b 543.6 138 included in our data listed the pharmacy practice Citations/article, n 10.3 6.1 9.2 5.2b chairs as first author. Author order can have imh-Index 11.6 5.3 9.7 4.4b portant implications for tenure and promotion. In general, the first author usually contributes more HSC = health sciences center; WoS = Web of Science. a p < 0.01. than the second author, the second author usualb p < 0.001. ly contributes more than the third author, and so 270

n


n


www.theannals.com

Publication Metrics and Record of Pharmacy Practice Chairs

on. The last author is sometimes considered the senior author or principal investigator, which may count substantially to-

Figure 1. Lifetime PubMed publications of pharmacy practice chairs.

Figure 2. Total lifetime citations for pharmacy practice chairs.

Figure 3. h-Index for pharmacy practice chairs.

www.theannals.com

ward tenure and promotion. However, these are generalities and author order can be influenced by a wide variety of factors besides contributions.13 There are few data on author order in pharmacy publications, so it is difficult to interpret these data. Using quantitative and statistical methods to describe patterns of publications in a given field, bibliometric methods can be applied to an individual researcher, department, college, university, or geographic area.11,14-20 It can also be applied to scholarly journals or books to better characterize patterns of publications. Lotka’s law is one of the basic principles of bibliometrics.21 It states that, “…the number [of authors] making n contributions is about 1/n2 of those making one; and the proportion of all contributors, that make a single contribution, is about 60%.” Explained in simpler terms, of 100 authors each publishing 1 publication or more, 60 authors will have published 1 article, one-quarter of 60 (15 authors) will have published 2 articles, one-ninth of 60 (about 7 authors) will have published 3 articles, and so on. Lotka’s law expresses the skewed nature of publication production, which many refer to as the “80 –20” rule (ie, 20% of authors produce 80% of the publications). Some data are available to show the skewed nature of pharmacy and medical publications. Table 3 lists the percentage of pharmacy practice chairs, pharmacy deans, pharmaceutical sciences faculty, and medical faculty producing 50% of the publications found in this and previous studies.12,15,22-24 While illustrating the skewed nature of faculty publications in the biomedical field, these data do not conform strictly to Lotka’s law because Lotka’s law is based on authors, each of whom has published 1 or more paper. It is clear from Table 3 that not all faculty members have published papers in the biomedical literature. In fact, in some studies, a large proportion of faculty did not publish any papers, as reflected in the databases searched. A number of publication metrics have been used to measure scholarly output. The total number of papers published is commonly used to quantify a scholar’s work; however, mere counts may not measure the importance or significance of the work even if the individual publications are peer-reviewed. The total number of citations has been suggested as a measure of an individual’s scholarly impact. Citation counts generally reflect the relative useful-


n


n

271

DF Thompson et al.

ness of the scholar’s work as cited by others in that discipline. However, citation counts have been criticized by a number of authors as inaccurate and biased; this metric can be inflated by a few highly cited papers.25,26 Some papers may also be cited in a negative context (eg, inappropriate study design, incomplete results). The h-index was proposed by Hirsch5 in 2005 as a measure of scholarly output by a researcher. It is defined as “the number of papers with citation index > h.” For example, if Professor Smith has an h-index of 21, he has published 21 papers that have at least 21 citations. Hirsch found the h-index to be a superior metric for quantifying scientific achievement and predicting future success.16 For physicists at major research universities, an h-index of 10–12 has been proposed for awarding tenure, 18 for promotion to full professor, and 15–20 for selection as a Fellow of the American Physical Society.16 Many authors2729 agree with this positive assessment of the h-index, although critics have argued that the h-index is potentially biased and that it should be combined with other measures to provide context.6,30,31 Lehmann et al.6 also do not favor the use of the h-index, suggesting that the number of citations or the citations per article is the most reliable measure of scholarship and the best predictor of future scientific success. Primarily because citation numbers take time to develop, the h-index favors scientists who have been publishing for a number of years. In response to this, Hirsch has also introduced the m quotient (m = h/n, where h is the h-index and n is the number of years between the first published paper and the present), which normalizes the h-index for the number of years a scholar has been publishing.5 Hirsch has suggested that, for his field of scientific endeavor (physics), an m quotient of 1 characterizes successful scientists, an m quotient of 2 characterizes outstanding scholars at the top universities, and an m quotient of 3 characterizes truly rare scientists. In our study, the m quotient was 0.6–0.7 for 7 chairs, 0.8–0.9 for 6, and 1.0–1.1 for 2; the remaining chairs had an m quotient of less than 0.6. Table 4 is a compilation of faculty scholarship data from previous studies performed in both pharmacy and medi-

Table 3. Comparative Data on the Skewed Number of Publications

Authors

% of Group Publishing 50% of All Publications

% of Group Publishing 0 Papers

Pharmacy practice chairs

12

3

Pharmacy deans12

13

2

Pharmaceutical sciences faculty22

15

19

College of medicine faculty23

10

22

College of medicine faculty15

10 (produced 40%)

25

Family practice faculty24

272

n