Introduction

The impact factor (IF) was introduced more than 50 years ago by Eugene Garfield to assist librarians in selecting source journals (Garfield 2006). Meanwhile, it is frequently used to evaluate departments, researchers, or even individual publications. For instance, some search or tenure committees look at the sum of IF points accumulated by a candidate and some universities allocate departmental budgets based upon the accumulated IF points. However, the IF has never been intended or validated for such purposes and is unfit for it (Garfield 2006). For instance, the IF of a given journal can change considerably over time (Ioannidis et al. 2010) and journals in some medical disciplines have a much higher median IF than those of other fields (Baethge 2012) despite lacking evidence that research in one field is inherently of greater quality than the other. Despite these limitations, many scientists select journals for submission of their manuscripts at least partly based on IF. Realizing such practice, journals in turn are interested in boosting their IF to attract the best possible manuscripts in their fields.

There is a general assumption among journal editors that review papers by average attract more citations and that inclusion of many review papers may be helpful to increase the IF of a journal. Only little data is available whether this assumption is true and how effective inclusion of review articles is to boost the IF of a journal. Against this background, we have empirically explored for five international pharmacology journals over a 5-year period whether review articles indeed accumulate more citations and how this affects the overall IF of a journal.

The IF is calculated based on their proprietary Web of Science® database (www.webofknowledge.com) using the equation

$$ \mathrm{IF}=\frac{\left(\mathrm{citations}\ \mathrm{in}\ \mathrm{year}\ \mathrm{x}\ \mathrm{to}\ \mathrm{source}\ \mathrm{items}\ \mathrm{published}\ \mathrm{in}\ \mathrm{year}\mathrm{s}\ x-1\ \mathrm{and}\ x-2\right)}{\left(\mathrm{number}\ \mathrm{of}\ \mathrm{source}\ \mathrm{items}\ \mathrm{in}\ \mathrm{year}\mathrm{s}\ x-1\ \mathrm{and}\ x-2\right)} $$

The Web of Science® website lists the number of total source items and classifies them, e.g., as “original articles” or “reviews.” Number of citations can be viewed separately for each category of source item and year of citation. This enables calculation of a pseudo-IF, i.e., the IF that would have resulted had the journal only published original articles or only reviews.

We have chosen to look at five general pharmacology journals for our analysis. These included journals owned by two commercial publishes (Elsevier for European Journal of Pharmacology (EJP) and Life Sciences (Life Sci) and Springer-Nature for Naunyn-Schmiedeberg’s Archives of Pharmacology (Naunyn)) and by academic societies (British Pharmacological Society for British Journal of Pharmacology (BJP) and American Society of Pharmacology and Experimental Therapeutics for Molecular Pharmacology (MP)). The five selected journals cover a range from those with high and with medium IFs, relative to all journals publishing original research within the field of pharmacology. Originally, we had planned to include the Journal of Pharmacology and Experimental Therapeutics, but it had published so few review articles during our analysis period that we did not consider the total number to be meaningful for the purpose of our analysis; therefore, we replaced it by MP, which is more specialized than the others but remains reasonably broad. To improve the robustness of our analysis, we performed all calculations and analyses for 5 consecutive years of IF assignment. Differences between total number of source items for a given journal and sum of original and review articles are caused by inclusion of some source items in the calculation of the official IF that are classified in other ways, e.g., as editorial material. All data underlying our calculations are shown in the Online Supplement.

As shown in Table 1, the five journals included in the analysis had mean official IFs in the 2009–2013 period ranging from 2.457 to 4.919. Within each journal and each year of analysis, the pseudo-IF based on reviews was about twice as high as that for original articles. However, the official IF was not much higher than the pseudo-IF calculated for original articles only, except for BJP (+ 32, 8, 12, 2 and 11% for BJP, EJP, Life Sci, MP, and Naunyn, respectively). To better understand this, we calculated percentage of source items classified as reviews on the Web of Science® website, which was 17.5, 4.8, 7.0, 3.1, and 8.3%, respectively for the five journals.

Table 1 Pseudo-impact factors for original and review papers as well as official impact factor of five major pharmacology journals. See main text for calculation methods
Full size table

These data confirm the widely held belief that review articles are cited more often than original articles, but the difference may be smaller than expected, i.e., represents factor of about 2. Interestingly, reviews published in a journal with a relatively high IF (BJP or MP) had a greater pseudo-IF than those published in journals with medium IF. On the other hand, pseudo-IFs from BJP and MP (9.640 and 10.041) were lower than those of top journals only publishing review articles within the same field such as Annual Reviews of Pharmacology (20.862) or Pharmacological Reviews (19.396) in the same period. We propose that this may at least in part be related to the fact that articles published in leading review journals typically are much longer than those published in journals primarily reporting original research and, accordingly, can be quoted to support a broader range of claims.

Our analysis also shows that the effect of including review articles on the official IF was only moderate for most journals. The exception is BJP, in which 17.5% of all articles published were reviews. Mathematically, this is not surprising if the pseudo-IF for reviews within a journal is only twice as high as that for original articles. We conclude that inclusion of review articles can increase the IF of a journal; however, this only has a moderate effect on the IF unless review articles constitute a relatively large fraction of articles within a journal.