Bigger Is Not Always Stronger: Citation Concentration and the Illusion of Journal Impact

The journal impact factor (IF) is widely used as a shorthand for scientific influence 1. Yet its arithmetic construction assumes that the mean citation rate reflects the typical article—a premise that may not hold in strongly skewed citation distributions 2. Citation distributions in biomedical science are strongly right-skewed, raising the possibility that the IF primarily reflects the performance of a limited subset of publications rather than journal-wide impact. In highly skewed distributions, the mean is mathematically sensitive to extreme values. When a small fraction of articles accumulates disproportionately large citation counts, the average may substantially overestimate the central tendency of article-level influence. This structural property has direct implications for how journal metrics are interpreted in research evaluation. To evaluate this empirically, we analysed all articles and reviews published in 2022–2023 across nine major biomedical journals spanning general and specialty domains: The Lancet, New England Journal of Medicine (NEJM), Nature Medicine, Lancet Diabetes and Endocrinology, Diabetes Care, Diabetes, Diabetes Obesity and Metabolism, Obesity, and Nutrition Metabolism and Cardiovascular Diseases (NMCD). Citation data were extracted from the Web of Science Core Collection. For each journal, we quantified citation concentration using Lorenz curves and complementary inequality metrics 3. Articles were ordered from least to most cited, and cumulative citation share was plotted against cumulative article share. Deviation from the 45° equality line reflects concentration of citation influence. We also computed a citation concentration ratio (CCR), defined as the proportion of total citations generated by the top 10% of published articles. Across all journals, citation distributions were markedly asymmetric. The top decile of articles generated between approximately 40% and 50% of total citations in most journals, confirming that citation influence is consistently concentrated within a minority of publications. Importantly, the magnitude of concentration varied meaningfully across journals. General medical journals exhibited pronounced curvature of the Lorenz profile, reflecting dependence on highly cited landmark trials, consensus statements and guidelines. Specialty journals displayed similar but generally less extreme patterns. In highly concentrated settings, the SCS diverged substantially from the official IF. The largest proportional reduction was observed in The Lancet (−57%), followed by substantial reductions in NEJM (−43%) and Nature Medicine (−43%). Specialty journals showed heterogeneous effects, with reductions ranging from −20% to −56%. Among specialty journals, the magnitude of structural inflation varied substantially. Notably, Diabetes Obesity and Metabolism showed one of the smallest divergences between official impact factor and the structurally corrected score, suggesting a comparatively broader distribution of citation influence across published articles. Nature Medicine exhibited substantial citation concentration comparable to NEJM, but less extreme than The Lancet, indicating that even among top-tier general journals, the degree of structural inflation varies considerably. These findings indicate that high IF values do not arise from uniform distributional architectures. Journals may achieve similar apparent impact through different structural mechanisms—either through broad-based citation dispersion across many articles or through strong concentration within a limited subset of highly cited publications. Some specialty journals exhibited relatively limited divergence between official and distribution-adjusted metrics, indicating that citation influence may be more evenly distributed across publications. In such cases, the impact factor may better approximate the citation experience of the typical article. Because the IF is defined as an arithmetic mean, it is intrinsically sensitive to extreme values. In strongly skewed distributions, the mean may substantially exceed both the median and distribution-adjusted estimates of central tendency. When journal-level metrics are used to infer the quality of individual articles or investigators, concentration-driven inflation may propagate distortions into academic evaluation systems 4. The impact factor remains a useful descriptive indicator of citation activity. However, its interpretation requires explicit acknowledgment of underlying distributional structure. Routine reporting of dispersion-sensitive indicators—including median citation counts (MCC) and citation concentration ratios (CCR)—could enhance interpretability without replacing existing systems 5. From a practical perspective, improving bibliometric transparency does not necessarily require replacing the impact factor, but complementing it. Journals could routinely report distribution-sensitive indicators—such as MCC and CCR—alongside the impact factor on their official platforms. In parallel, researchers, institutions and evaluators should explicitly consider citation distribution characteristics when interpreting journal-level metrics or making publication and assessment decisions. In an era where journal-level metrics continue to influence research assessment, acknowledging citation concentration is not optional—it is necessary for bibliometric transparency. Figure 1 represents Lorenz curves of citation distributions across major biomedical journals (2022–2023 publications). Table 1 shows Official impact factor and structurally corrected citation metrics (2022–2023). The author has nothing to report. The author declares no conflicts of interest. The data underlying this article are publicly available from the Web of Science Core Collection. No new data were generated. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70733.