The Nuances of the Impact Factor

I can’t tell if this article is Breaking News or an Explainer. Whatever it is, I found it interesting and enlightening. Dr. Bauchner’s article gives us insight into what an impact factor means, what our medical journals specialize in, and what journal editors consider when designing their journals and deciding what to publish. I think the article makes clear that, like everything else, the impact factor is affected by Goodhart’s law: when a measure becomes a target, it ceases to be a good measure.

Adam Cifu

Created in 1955 by Andrew Garfield, the impact factor (IF) is one measure – albeit an important one – of the quality of a journal. It is a relatively simple calculation, with numerous nuances. Calculated by Clarivate – there is a numerator – all citations to every article published in a journal over a 2-year period – divided by what Clarivate considers “citable” articles. Citable articles are research reports and long reviews. Brief article types, under 2-3 pages, with a limited number of references, are generally not considered citable. Specifically, editorials, research letters, perspectives, viewpoints, and other brief articles do not count in the denominator, but the citations they accumulate do count in the numerator.

Released every June, the IFs are already somewhat out of date. The newest IFs, released in June, are the 2025 IFs. They reflect data from 2023 and 2024. Recent impact factors offer insights into many of the nuances of this measure.

The 2025 IF of NEJM was 85 (51,607/611), whereas the IF for Lancet was higher at 109 (51,655/474). But as you can see from the denominator, Lancet has been very careful to “control” its denominator. It publishes far fewer “citable articles” than does NEJM. This has been true for many years. But that is not the entire story. Of the top 10 articles in Lancet, 5 were Global Burden of Disease papers, and 4 were worldwide estimates of various diseases. These 9 articles accounted for 24% (12,117/51,655) of the citations! Of the top 10 cited papers in NEJM, 8 were RCTs, and the other 2 were translational science papers, accounting for 10% (5386/51607) of citations. This highlights important differences in the journals. NEJM publishes what many would consider traditional research, RCTs, and other scientific reports, whereas Lancet publishes more epidemiological studies. From my standpoint, RCTs change practice, whereas epidemiological studies are more likely to influence health policy.

A similar difference is seen in the IF of the leading general cardiology journals, EHJ, Circulation, JAAC, and JAMA Cardiology. The 2025 IFs were 45 (22986/507), 41 (21872/529), 22 (13,095/588), and 15 (3470/228), respectfully. But again, a more detailed examination highlights differences in content. Of the 10 most cited papers in EHJ, 6 were guidelines, reflecting how often the European Heart Society produces them. Circulation published 3 guidelines and JAAC 2. The top 10 cited articles in EHJ accounted for 28% (6434/22986) of the citations! Although it should be noted that 2 of the top 10 papers in Circulation were reports from the American Heart Association. JACC published 3 consensus pathways from the ACC. JAMA Cardiology had no guidelines or society reports in its top 10 cited articles. Are guidelines, consensus statements, and reports from professional societies research? These publications certainly impact clinical care, but many would not consider them “traditional” research.

The major fully open-access journals offer more nuances. BMC (formerly Biomed Central), which began publishing in 2000, has an IF of 8.7 (9287/1062), and PLOS One, which began publishing in 2006, has an IF of 2.8 (88,549/31,104). These are among the oldest fully OA journals and represent this “new” publishing model. PLOS Medicine, which began publishing in 2004, has an IF of 9.9 (2838/287), and is meant to be more selective than PLOS One. BMJ Open debuted in 2011 and has an IF of 2.5 (19,449/7909), while JAMA Network Open, which debuted in 2018, has an IF of 11.7 (39,580/3391). These titles are representative of the thousands of OA journals that come in all shapes and sizes. In general, as fully OA journals publish more manuscripts, which drives their revenue, it is difficult to maintain a high IF – the exception has been JAMA Network Open, which, despite a relatively large denominator, has a healthy IF.

There are several other nuances that bear discussion.

• Should a journal release an important article in November, when it will count for fewer months in the IF than if it were released in January?

• Are journals adding more short article types – hoping to add citations from articles that don’t count in the denominator?

• Is there article creep, short article types that are longer with more references, but still not “citable?”

• Should journals add “big” epidemiology studies, knowing that they are well cited?

• Do journals focus on their denominator, “citable articles,” and at launch of a new journal keep it very limited, or if an established journal, over time, shrink the denominator, hoping to eliminate those articles that are not well cited?

Comparing journals from different fields is risky. Cardiology and oncology journals have higher IFs than journals in other fields simply because there are more journals in those fields – hence, more citations. Should editors of general medical journals give preference to those subject areas, knowing that those articles are likely to accumulate more citations?

The IF is not perfect – it can be manipulated – by “controlling” the denominator, by publishing certain types of articles that are known to be well cited, or by publishing more articles that might be cited, opinion pieces, or research letters, but do not count in the denominator. The IF does not reflect the quality of an individual article or the research work of an author. Yet, it has stood the test of time – and all the other journal metrics – CiteScore, Scimago Lab – have their own challenges. For example, both Lancet (92.4) and NEJM (74.8) have very high CiteScores, which is the number of citations accumulated over 4 years divided by all published reports. JAMA does not (26). This is in part because the CiteScore counts all authored reports in a journal, and JAMA has a very robust, popular authored news section, which adversely affects its CiteScore.

When selecting a journal to publish in authors should consider not only the IF, but whether their article is in scope for the journal (hint, read the journal before submitting to it!), the services provided (think turnaround time and quality of the editing), its audience, including print circulation and reach in social media, and increasingly whether the journal offers open access, which is often mandated by funders. In addition, cost is likely to become an increasingly important part of the discussion. With open-access fees exceeding $10,000 US for some journals, these will simply be out of reach for many investigators.

Howard Bauchner, MD, is a Professor of Pediatrics and Public Health at the Boston University Chobanian & Avedisian School of Medicine. He is also a Visiting Scholar at the National University of Singapore. Dr. Bauchner is also the former Editor in Chief of JAMA and the JAMA Network and former Editor in Chief of Archives of Disease in Childhood.