Impact of clinical research as a confounder for medical school rankingsJENS PETER ANDERSEN, PHDMEDICAL LIBRARY, AALBORG UNIVERSITY HOSPITAL
Clinical and basic research
Definitions, as terminology varies: Basic research: experimental, basic medical research (e.g. medical
genetics)
Pre-clinical research: experimental, medical research usually not involving human subjects (e.g. animal studies)
Clinical research: experimental, medical research focused on patients
Medical schools and teaching hospitals
Generally, basic and pre-clinical research is carried out at universities, while clinical research is conducted at hospitals.
Medical industry may be involved in all stages.
Important differences
Generally: Distance from healthcare and –policy
Distance from public
For bibliometrics: Author-factors
Citation densities & speed
Example of field differences
Basic:Anatomy & morphology Average cited half-life: 7.6 years
Average JIF (top10): 3.3
Top JIF: 9.8
Clinical:Clinical neurology Average cited half-life: 6.3 years
Average JIF (top10): 11.7
Top JIF: 21.8
Hospital ranking
Leiden ranking: hospitals included as parts of universities
ARWU ranking: undocumented, but likely
Times Higher Education ranking: undocumented, but likely
U-Multirank: hospitals are included – also with regards to beds per student for teaching purposes
Consequences for ranking and assessment
Proper normalisation is always necessary (field, age etc.)
Impossible to discern between effects from universities and hospitals
Ranking tables may be misleading, e.g. for recruitment and student attraction, if research impact reflects hospital research
Clinical evidence and citations
From: Andersen, J. P., & Schneider, J. W. (2011). Influence of study design on the citation patterns of Danish, medical research. In Proceedings of the ISSI 2011 Conference (pp. 46–53).
Distribution of national and international publications for university and university hospitals
Research institutions
Univ Hosp national pub
Univ Hosp international pub
Univ national pub
Univ international pub
Rel
ativ
e ci
tatio
n im
pact
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Clinical evidence and citations
From: Andersen, J. P., & Schneider, J. W. (2011). Influence of study design on the citation patterns of Danish, medical research. In Proceedings of the ISSI 2011 Conference (pp. 46–53).
Meta-analysis
RCT
Clinical trialSystematic review
Nonsystematic review
CohortCase-
control
Case-report
0
0.5
1
1.5
2
2.5
3
0 200 400 600 800 1000 1200 1400
Mea
n no
rmal
ized
citati
on im
pact
Publications
Danish
International
Clinical evidence and social media
From: Andersen, J. P., & Haustein, S. (2015). Influence of study type on Twitter activity for medical research papers. In Proceedings of ISSI 2015 - 15th International Society for Scientometrics and Informetrics Conference (p. in press). Istanbul, Turkey.
Impact of universities and hospitals:Denmark as a case
Unpublished data.
Source: Web of Science, AD=Denmark AND PY=(2011-2013) AND DT=(Article OR Review)
Limited to medical research
Impact of universities and hospitals:Denmark as a case
Unpublished data.
Source: Web of Science, AD=Denmark AND PY=(2011-2013) AND DT=(Article OR Review)
Limited to medical research
Research profiles
Hospitals, teaching hospitals and universities are all dissimilar.
Profiling can illustrate varying types of institutions
Research profile I: King’s College London and King’s College Hospital King’s College London King’s College Hospital Collaboration
N c mc N c mc N c cc Coll-adv Hosp-adv
Obstetrics Gynecology 88 708 8,05 60 527 8,78 37 464 12,54 1,56 1,43
Surgery 171 721 4,22 57 225 3,95 28 127 4,54 1,08 1,15
Clinical neurology 241 1999 8,29 56 457 8,16 25 229 9,16 1,10 1,12
Gastroenterology hepatology
62 561 9,05 53 504 9,51 34 254 7,47 0,83 0,79
Hematology 89 982 11,03 35 348 9,94 16 170 10,63 0,96 1,07
Psychiatry 663 4572 6,90 12 85 7,08 9 77 8,56 1,24 1,21
Neurosciences 336 3034 9,03 25 188 7,52 15 86 5,73 0,63 0,76
Multidisciplinary sciences 255 2387 9,36 11 46 4,18 7 39 5,57 0,60 1,33
Clinical neurology 241 1999 8,29 56 457 8,16 25 229 9,16 1,10 1,12
Biochemistry molecular biology
191 1963 10,28 9 43 4,78 5 20 4,00 0,39 0,84
Research Profile II: Harvard University and Dana-Farber Cancer Institute* Dana-Farber Cancer
InstituteHarvard University Collaboration
N c mc N c mc N c mcDana-adv
Univ-adv
Oncology 704 8525 12,11 1458 15532 10,65 704 8525 12,11 1,00 1,14
Hematology 172 1660 9,65 447 4020 8,99 172 1660 9,65 1,00 1,07
Multidisciplinary sciences 144 3664 25,44 1518 26541 17,48 144 3664 25,44 1,00 1,46
Cell biology 137 3641 26,58 888 17056 19,21 137 3641 26,58 1,00 1,38
Biochemistry molecular biology
106 2816 26,57 976 16012 16,41 106 2816 26,57 1,00 1,62
Multidisciplinary sciences 144 3664 25,44 1518 26541 17,48 144 3664 25,44 1,00 1,46
Oncology 704 8525 12,11 1458 15532 10,65 704 8525 12,11 1,00 1,14
Surgery 36 181 5,03 1122 5193 4,63 36 181 5,03 1,00 1,09
Neurosciences 11 119 10,82 1020 8954 8,78 11 119 10,82 1,00 1,23
Biochemistry molecular biology
106 2816 26,57 976 16012 16,41 106 2816 26,57 1,00 1,62
*Unlike many other medical schools, Harvard does not own their affiliated hospitals.
Implications
Call for more complete details in rankings: Are hospitals counted?
Where does the impact come from?
Identification of basic and clinical research
Should we normalise based on study types? Or report profiles of study types?
Discussion points
Why does medical research require special attention?
1. It doesn’t – technical research institutes potentially provide similar problems
2. They differ from other university research areas, because of e.g. funding sources, employment types, research cultures, research goals – and there are vast geographical differences
What about translational research?
Thank you for your attentionIn picture: Aalborg University Hospital