OUP user menu

Chairpersons of Pathology in the United States
Limited Benchmarks for Publications

John Hedley-Whyte MD, Debra R. Milamed MS, David C. Hoaglin PhD
DOI: http://dx.doi.org/10.1309/AJCPMVAYD5V8DLZN 185-192 First published online: 1 August 2010

Abstract

Pathology chairpersons continue their preeminence in US medical schools. The Hirsch (h) index of citation was analyzed for the chairpersons of 8 specialty departments in 41 US medical schools. The current h indices of department chairs (302 of 328 positions filled) were tabulated by specialty and medical school. The proportion of women was noted, as was whether the chairs had completed the first decade of this millennium continuously in post in the same medical school. Of the 8 specialties, pathology chairs at the 41 schools are the most likely to have been in the same post in 1999 and 2009, and the pathologists who remained are exclusively male. In the 41 medical schools studied, women chairs constituted only 3.7% of all heads of departments in 1999 and 7% in 2009. The paucity of female chairs 3 decades after women became at least one third of the medical student body is unfortunate.

Key Words:
  • Advice
  • Governance
  • Hirsch index
  • Mentoring
  • Policy
  • Therapeutics
  • Women physicians

Almost a decade ago, Wick and Stoler1 and Domen,2 in this Journal, addressed the publication record of academic chairpersons of pathology in the United States. Surprisingly, no subsequent article has pursued this topic, despite the fact that in the 20th century, heads of pathology discovered, developed, and deployed therapeutic regimens that changed medicine and animal husbandry. Welch, Almroth Wright and Fleming, Florey, and Farber have worldwide fame.38 Domagk and Whitby introduced sulfonamides and proved their worth.9 We decided to see whether pathology chairpersons are continuing their intellectual preeminence in academic medicine into the first decade of the 21st century by investigating their Hirsch (h) indices,10,11 a limited form of citation analysis. We use the h index,10,11 together with a measure of turnover, for the current chairs of 8 specialty departments in 20 apparently first-sextile US medical schools and 21 US medical schools selected from the third sextile (upper half of the middle third) of the 126 Liaison Committee on Medical Education (LCME)-accredited US medical schools. We also made comparisons among specialties and between men and women as department heads for the period up to 2010.

Despite the oversimplifications inherent in reducing citation data on an author’s publications to a single number,12,13 the h index is widely used, in part because it is readily available. It has even been suggested that the h index constitute part of the vetting process of the European Research Academic Exercises1416 and be used as justification for allocation of research funding.1722

Calculation and verification of the published h index requires much attention to detail (see the “Materials and Methods” section). It would take months to verify the h indices for current chairs of 8 specialty departments in all 126 LCME-accredited US medical schools. Before the introduction of the h index, Wick and Stoler1 found that “On the initial NLM search, the largest number of citations per person was more than 4,000. It was obvious that this was because of non-specificity of common names that was difficult to resolve and that badly skewed the data.” They chose, therefore, to exclude 8 common names from their study.1

Wick and Stoler1 correctly state that “Chairpersons of pathology often are viewed as departmental role models in academic medical centers.” We compared the publication records of the chairpersons of pathology, summarized by this limited form of citation analysis, with those of the chairs of other clinical departments.

The ranking of medical schools1721 is controversial, but we have not analyzed ranking systems. We used ranking solely to compare 20 prominent medical schools with another less-prominently ranked group. Finally, reducing the contributions of a person’s publication record to a single number sacrifices much valuable information. Thus, the h index must always be used with caution, especially in the evaluation of promising young investigators.

Materials and Methods

We compiled a list of 20 apparently first-sextile US medical schools by comparing and consolidating several published rating systems.1721 We selected 21 US medical schools from the third sextile (upper half of the middle third) of the 126 LCME-accredited US medical schools as a comparison group.1721

For each of the 2 groups of US medical schools, we identified the chairs of academic departments of anesthesiology, medicine, neurology, pathology, pediatrics, psychiatry, radiology, and surgery using the Association of American Medical Colleges Directory of American Medical Education, 2008–2009, the most recent edition available.23 For department head positions without permanent appointees in the 2008–2009 edition, we consulted the Internet Web sites of the medical schools to obtain updated information. We noted interim chairmanships or position vacancies. For Harvard Medical School, Boston, MA, where the Directory of American Medical Education listed heads of academic departments at the various teaching hospitals, we recorded the departmental executive committee chairperson as head.23 We used the 1999–2000, 1989–1990, and 1979–1980 editions of the Directory of American Medical Education to examine the longevity in post of department chairpersons.2426

We used values of h 10,11 calculated from the electronic records of Thomson Reuters, ISI Web of Knowledge, Science Citation Index Expanded, during 2009.27 An author’s h index is the largest n for which he or she has n published articles, each of which has received at least n citations. For example, a person who has only 3 published articles in the ISI Web of Knowledge database, with 100, 3, and 1 citations, respectively, has an h, or h index, of 2. In contrast, a famous scientist who has 100 published articles, each cited at least 100 times (and other articles with fewer than 100 citations) has an h of 100. We used the online bibliographic services provided to an academic institutional subscriber with access to retrospective data from 1900 to the present.27

ISI Web of Knowledge, Science Citation Index Expanded, identifies authors by surnames and initials, rather than by surnames and full given names. It omits the designation “Junior” and roman numerals. Often articles by other authors with the same surname and initial(s) are erroneously included in the citation record of the target author. As did Wick and Stoler,1 we found this approach problematic because it is nearly impossible to separate articles by authors with common names. Thomson Reuters now suggests that researchers apply for a unique author identification number (researcher ID), but relatively few have registered, and the h index may not be directly calculated from the profiles accessible by author identification number. We used the following protocol to identify authors and calculate h in the ISI Web of Knowledge, Science Citation Index Expanded.

First, we obtained full name with initials, dates of degrees, approximate dates of career activity, scope of research, curriculum vitae, and a partial or full publication list from a public source such as a university Web site. We used the author index within the ISI database to identify authors and publications. We used ISI author identification aids within the ISI database to identify authors. We reviewed the ISI Web of Knowledge “Distinct Author Sets,” which are sets of publications with characteristics suggesting they are the work of the same author. The number of distinct author sets may vary from 0 to 100 or more and may combine name variations (eg, J. Smith and J.A. Smith in results of a search for J.A. Smith or J. Smith). Distinct author sets are intended only as finding tools, not comprehensive publications lists. We then limited the subject scope of publications to be attributed to the author. We obtained the list of publications and citation analysis report, which includes h.

We reviewed publications in a citation analysis report at least to the h number, to ensure that highly cited articles were a specific author’s work. One may select items from an author search believed to be the actual work of the author, to create a subset, but one cannot run a citation analysis on that subset. One can run a citation report from an author search, then delete individual articles that do not seem to be the work of the author, and run a subsequent citation analysis. If articles that are not the author’s work are included in the report, but have fewer than h citations, the total numbers of articles and citations will be elevated, but not h. PubMed28 was used to further define the body of work as that of a given sole author or coauthor. The time required for the calculation, review, and recording of an individual chairperson’s h index varied from 5 to 60 minutes, with the majority taking 10 to 20 minutes.

Thomson Reuters ISI Science Citation Index Expanded does not index textbooks; conference proceedings are covered by a separate conference proceedings index. Science Citation Index Expanded covers more than 6,500 journals in 150 scientific disciplines, including medicine.27 PubMed is linked to the National Library of Medicine MEDLINE database, which covers more than 5,000 medical journal titles, as well as some books, newsletters, and conference proceedings, from 1949 to the present, with some older material.28 PubMed does not include standalone proceedings abstracts, but Thomson Reuters ISI Science Citation Index Expanded often does include them.

Statistical Methods

We tabulated the h of chairpersons of academic departments in 20 leading medical schools and, separately, in institutions in our comparison group. Medians and ranges of h for department chairpersons were calculated for each medical school, and medians of h were calculated for each specialty, using Microsoft Excel, Microsoft Office Professional Edition, 2003, and updates (Microsoft, Redmond, WA).

We used 2-way analysis by median polish to separate the h data in each table into 4 parts: an overall value, the incremental contribution of the medical school, the incremental contribution of the specialty, and a table of residuals.29 The legends for Table 1 and Table 2 show, algebraically, the decomposition of the data into these 4 parts. The incremental contributions (usually referred to in statistics as “effects”) summarize, respectively, (1) the difference between the typical h index of the particular medical school’s department chairs and the typical h index of the department chairs in the group of medical schools and (2) the difference between the typical h index of the chairs in the particular specialty and the typical h index of the department chairs in the group of medical schools. An individual chair’s residual is the difference between that person’s h index and the value predicted (from the table as a whole) for that combination of medical school and specialty. Our aim was exploration and description rather than formal inference.

View this table:
Table 1
View this table:
Table 2

Results

Heads of pathology in the combined 41 medical schools studied have the best citation record, with a mean h index of 51.2 (median, 49.5; range, 28–83) for the first-sextile medical schools and a mean h index of 38.3 (median, 35; range, 23–60) for the comparison group. They have also had the longest aggregate survival in post. The overall typical h index of department chairs was 42 for the first-sextile medical schools (Table 1). Relative to that overall value, medicine, pathology, psychiatry, and neurology had higher typical levels, by 10, 9, 9, and 8 units, respectively. Radiology was lower by 12 units and anesthesiology by 22 units, with surgery and pediatrics in between. In the comparison group of third-sextile medical schools, the overall typical h index was 24 (Table 2). Pathology had the highest effect (+12), followed by medicine and neurology (both +8), surgery (+2), pediatrics (−3), psychiatry and radiology (both −6), and anesthesiology (−8).

Among the first-sextile medical schools, the medical schools’ effects ranged from +18 (Emory. Atlanta, GA) to −11 (University of Texas Southwestern, Dallas) (Table 1). For the comparison group of third-sextile medical schools, the medical schools’ effects ranged from +17 (school “O,” University of Miami, Miami, FL) to −12 (school “Q,” Tulane University, New Orleans, LA) (Table 2). Within the first-sextile medical schools, the ranking of the medical schools implied by their effects from the analysis of the chairs’ h indices has little relation to the rankings based on wider criteria.1721

An individual chairperson’s residual is simply the departure of the person’s h index from the value suggested by the analysis as typical for the particular combination of medical school and specialty. Taken together, the residuals in Table 1 or Table 2 show the extent of the variation in the h indices that is not accounted for by the contributions of the specialties and the contributions of the medical schools. Because the tables contain sizable positive and negative residuals, one common summary focuses on the interquartile range (IQR; the range of the middle half) of the data and the residuals. For the first-sextile medical schools, the middle half of the h indices Table 3 ranges from 26 to 51.5 (IQR = 25.5), and the middle half of the residuals ranges from −6 to +9 (IQR = 15) (Table 1). Similarly, for the third-sextile comparison group, the IQR is 20.5 for the h indices Table 4 and 13.5 for the residuals (Table 2). These comparisons of IQR between data and residuals indicate that variation in department chairpersons’ h indices has a substantial component that is unrelated to the specialty or the medical school.

View this table:
Table 3
View this table:
Table 4

During the first decade of the 21st century, the number of female chairs in the 41 medical schools studied increased from 11 of 300 filled positions to 22 of 302 filled positions in 2009.23,24 The h indices of the women were slightly lower than those of the men as judged by medians and means, with the exception of surgery, where the mean h index for women (n = 2) was 40 and for men (n = 35) was 33. Of the 22 residuals associated with female chairs, 14 were negative, 2 were zero, and 6 were positive. The survival in post during the decade was better for men than for women (Tables 3 and 4).

Where are the 11 chairs with an h index less than 18 (a suggested indicator of recognition10,16) in the first-sextile US medical schools? Ten of these chairs are anesthesiologists, and one is a radiologist; all are the sole department chair with an h index less than 18 in their respective medical schools. In 9 of the first-sextile schools, all chairs have an h index of 18 or more (table 3). The comparison group of third-sextile medical schools includes 43 chairpersons with an h index less than 18: 10 each in anesthesiology and psychiatry, 9 in radiology, 6 in pediatrics, 3 each in neurology and surgery, 2 in medicine, and 0 in pathology (Table 4). In the 41 US medical schools we studied, the h indices of chairs ranged from 6 to 92 in medicine and from 23 to 83 in pathology. The 36 neurology chairs had a range from 14 to 104. In these 3 specialties combined, only 5 chairs had an h index less than 18. In comparison, 3 of 38 chairs in surgery, 6 of 40 in pediatrics,10 of 36 in radiology, 10 of 38 in psychiatry, and 20 of 40 in anesthesiology, but none in pathology, had h indices less than 18 (Tables 3 and 4).

For the first-sextile US medical schools, the percentage of departmental chairpersons in post for the decade 1999 to 2009 ranged from 60% for pathology and 50% for psychiatry to 16% for medicine Table 5. For the third-sextile comparison group, this percentage ranged from 62% for pathology and 50% for pediatrics to 12% for medicine Table 6.

In the first-sextile US medical schools, a total of 7 department heads were in post for 2 decades, from 1989 to 2009, of 153 filled department head positions, or 4.6%: 2 in anesthesiology, 2 in pathology, 1 in psychiatry, and 2 in radiology. With the exception of 1 radiologist, all had h indices above the median for department chairs in their respective specialties. One 3-decade survivor, a radiologist, was in post from 1979 to 2009.

View this table:
Table 5

For the third-sextile comparison group, only 4 department heads were in post for 2 decades, from 1989 to 2009, of 147 filled department head positions, or 2.7%: 1 in neurology, 1 in pathology, 1 in pediatrics, and 1 in radiology. As in the first-sextile medical schools, only 1 chairperson was in post for 30 years—a radiologist (school “M,” Medical College of Wisconsin, Milwaukee). The denominators (ie, numbers of filled department head positions) are the numbers of filled positions in academic year 1999–2000; we did not count the positions filled in both 1999–2000 and 1989–1990. These numbers differ from the total number of filled positions shown in Tables 5 and 6 because these numbers represent the number of filled positions at both time points, 1999–2000 and 2008–2009.

Of the 7 female chairs in post in 1999 in the first-sextile US medical schools, only 3 were still in post in 2009—all anesthesiologists (Table 3). In the third-sextile comparison group, only 1 female chair—a pediatrician—remained in post for the decade (Table 4). The corresponding figures for men were 48 for the first-sextile schools, of whom the majority (30 of 48) were in the highest-ranked 8 schools (Table 3). In the third-sextile comparison group, 36 men survived in post from 1999 to 2009 (Table 4).

Discussion

Pathology chairpersons in the 41 US medical schools studied continued from the 20th century their intellectual preeminence, at least as judged by the h index: their summed h indices were overall more than 100 points higher than chairs of medicine (1,758 [n = 39] vs 1,655 [n = 37]) (Tables 3 and 4).

View this table:
Table 6

The paucity of female chairs 30 or 40 years after women became at least one third of the medical student body30 is unfortunate. We regret that the small sample for female chairs in this study does not permit a deeper analysis.

The United States seems to choose department chairs of its research medical schools differently from the rest of the world. With the exception of internal medicine, heads of academic departments in the specialties studied hardly ever move from a chair at one medical school to another. The reasons should be explored. For female chairs, especially those failing to withstand and pacify divisional unrest, a move to a more desirable medical school could have societal benefit.

In the complex environments of medical schools, the relationship of the chair to governing bodies is very important—support against the lack of vision of a de facto chief executive officer or chief financial officer can be crucial.31 Trustees may need reminding that in law they are ultimately responsible for providing the wherewithal to recruit and safely equip. In academic medical specialty departments at the moment, it is prudent for chairs to remember that an appropriately diverse divisional head structure is owed to the department and especially to the female half of the student body.

Critiques of the h index point out its limitations in the context of comparisons of individual and institutional academic performance.12,13,32 Necessary due diligence is time-consuming. We conclude that academic pathology does not suffer by comparison with any other medical discipline in the United States and that pathology chairs, as assessed by the h index, are highly productive among clinical chairs at select academic medical centers. Women are underrepresented among clinical department chairs. Pathology chairs tend to have longer tenures than chairs of other clinical departments.

Footnotes

  • Funded by the David S. Sheridan Professorship in Anaesthesia and Respiratory Therapy, Harvard University.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
View Abstract