Sysrevauthorship 2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

Title Can a Quality Systematic Review Have a Single Author?

Melissa L. Rethlefsen1, Tara J. Brigham2, Ann M. Farrell3, Leah C. Osterhaus Trzasko3


1Spencer S. Eccles Health Sciences Library, University of Utah; 2Mayo Clinic Libraries, Jacksonville, FL; 3Mayo Clinic Libraries, Rochester, MN

Reproducibility by Number of Screeners Low/Mean/High IOM and PRESS Scores by Number of Authors
100% 50
90%
80% 45

70%
Not Reproducible 40
60%
50% Reproducible
35
40% Reproducibility by Number of Authors
30% 30
100%
20% % Not Reproducible
90%

Score
10% There is a % Reproducible 25
significant % Mean PRESS Scores
0% 80%
difference Mean IOM Scores
Unknown 1 2
Number of Screeners
3 or more
between groups. R
70%
Mean PRESS Score by Number of 20
(p<.006) e
p Screeners 15 Error bars
60%
r 33 represent the
o low and high
50% 32 10
d Overall Mean scores for each
u Reproducibility 31 PRESS author group
40% Score
c % Line 5
30
i 30%
b 29
0
l 20% 1 2 3 4 5 6 7 8 9 10 11 12+
28
e Number of Authors
10% There is a 27
significant 0 1 2 3+
0% difference
Number of Screeners
1 2 3 4 5 6 7 8 9 10 11 12+ between groups.
Number of Authors
(p=.05)
There is no significant difference between groups.

Objective: To determine if systematic review quality is associated with the quality using a scored version of the PRESS instrument are published authors or number of screeners. Average PRESS scores peaked at 8
number of authors or number of screeners used to determine inclusion and elsewhere.[1] (mean=31.5) and 10 authors (mean=32.5). Articles acknowledging a
exclusion of studies. librarian or information specialist (LIS) averaged 1.1 author less than articles
Results: Number of authors ranged from 1-253, with an average of 9 with no LIS and 3.3 fewer than those with an LIS author.
Methods: Systematic reviews are generally considered a team undertaking, authors per systematic review (median=6, mode=5). Number of screeners
requiring sustained effort from many over time. However, published was unclear in 282 (44.8%) systematic reviews; remaining articles had Conclusions: Systematic reviews are still being published with solo
systematic reviews may have as few as one author, calling into question between 1 and 17 initial screeners (mean=2.2, median=2). Percentage of authors, though larger review teams are more common. Mid-size review
their methodological rigor. For this study, a dataset of 630 previously reproducible searches varied by number of authors, peaking at 11 authors teams (6-11 authors) appear to produce higher quality systematic review
identified systematic reviews from five high impact general and internal (65% reproducible) and lowest at 1 and more than 101 authors (p=.05). searches than smaller or very large authorship teams. A large proportion of
medicine journals from 2008-2012 was used.[1] The number of authors and Number of screeners was not significantly associated with reproducibility, systematic reviews do not clearly report their methodology for initial
the number of inclusion and exclusion criteria (or first pass) screeners were though articles reporting 1 or 3 or more screeners had higher mean scores title/abstract screening. Future research is needed to assess optimal
extracted from each article in duplicate. The number of authors and than the other two groups. Number of IOM standards met varies significantly systematic review team size and contributorship models.
screeners was statistically compared to compliance with Institute of by number of authors (p<.0001); articles with 7 and 8 authors have the 1. Rethlefsen ML, Farrell AM, Osterhaus Trzasko LC, Brigham TJ. Librarian co-authors correlated with higher quality reported search strategies in general internal medicine
systematic reviews. Journal of clinical epidemiology 2015 doi: 10.1016/j.jclinepi.2014.11.025

Medicine (IOM) recommended standards for finding and assessing highest mean number of standards met. There is also significant difference
individual studies (ANOVA) as well as overall reproducibility (chi square). between IOM scores by number of screeners (p<.0001). Post-hoc analysis
For those with reproducible searches, additional analysis was run to shows a significantly lower IOM score for articles with an unknown number Contact us:
compare number of authors and screeners to search quality (ANOVA). of screeners versus 2 screeners, but no significant difference between other mlrethlefsen@gmail.com or @mlrethlefsen
Details of methods for assessing IOM standard adherence and search groups. There is no statistical difference in PRESS scores by number of
MLA – Austin, TX– 5/17/15
More details/data online via the QR code:

You might also like