Introduction

Thomson Scientific, or the Institute of Scientific Information (ISI), publishes in June each year the Journal Citation Reports (JCR). Despite debate as to its usefulness, the journal impact factor (IF) remains the key statistic gleaned from the report. The role and limitations of impact factors have been previously described (1). An analysis of trends in nuclear medicine impact factors is yet to be reported in the literature.

The Impact Factor

A journals' impact factor is a measure of the ratio of recent citations to recent articles in a particular journal. The impact factor is the number of times the ‘average' article in a journal over the preceding two years was cited in journals in the subsequent year. That is, the 2006 impact factor is based on the number of citations in all ISI listed journals in 2006 of articles published in the specific journal in 2005 and 2004.

Figure 1

The ISI JCR offers 172 various categories for journals ranging from ‘acoustics' to ‘zoology'. Nuclear medicine journals generally fit the ‘radiology, nuclear medicine and medical imaging' category, however, a number of nuclear medicine journals may be found in other categories. For example, Clinical Physiology and Functional Imaging is the official journal of the Scandinavian Society of Nuclear Medicine but is located in the ‘physiology' ISI category. The number of specific journals comprising each category is quite variable. Grouping of journals within a category may represent a generic relationship or a more specialized relationship. Consequently, equitable comparison of the impact factor between ISI categories is not practical. Indeed, comparisons within many categories may be unreliable.

Within the ‘radiology, nuclear medicine and medical imaging' category in 2006 there were 85 journals of which just 11 are specifically nuclear medicine. The impact factor rating of nuclear medicine journals since 1998 has been summarized in table 1. A number of trends are evident:

1. The J Nucl Med has consistently ranked highest amongst nuclear medicine journals.

2. The impact factor for nuclear medicine journals has steadily increased consistent with the trend for the broader ISI category (figure 1).

3. A sharp upward spike in impact factor was generally seen amongst nuclear medicine journals in 2004. Most notably, Clin Nucl Med soured by 114.7% and Semin Nucl Med jumped 69.0%.

Those with a prominent 2004 spike tended to suffer a marked decrease in impact factor in 2005, generally below the figure of 2003. Decreases below 2003 scores were seen for Semin Nucl Med with a 48.3% decrease, Nucl Med Commun with a 31.5% decrease, Q J Nucl Med D Mol Im with a 28.6% decrease and the J Nucl Med with a 12.6% decrease.

The German language journal, Nuklearmed- Nucl Med, exhibited a trend with a 1 year delay from the wider nuclear medicine group with a 52.8% increase in impact factor for 2005 followed by a 33.2% decrease for 2006.

Figure 2

Table 1: Summary of impact factor ratings for nuclear medicine journals since 1998. The highest rating journal each year is highlighted in bold. Several anomalous patterns are highlighted by bold boxes.

Figure 3

Figure 1: The aggregate impact factors for nuclear medicine journals with an ISI listing since 2001 demonstrating a prominent 2004 ‘spike'. The dashed lines represent the ‘smoothed' impact factor trend for nuclear medicine journals (top) and the ‘radiology, nuclear medicine and medical imaging' category (bottom). It is evident that the impact factor has a strong upward trend for journals with a consistent ISI listing since 2001.

In 2006, the J Nucl Med was once again ranked highest amongst nuclear medicine journals with a ranking of fourth in the ISI category. The J Nucl Med increased its impact factor by 6.4% in 2006. Semin Nucl Med climbed back from a disastrous 2005 rating with a 49.1% improvement for 2006. Other significant improvers for 2006 were J Nucl Cardiol with a 17.3% improvement and Clin Nucl Med with a 23.2% improvement. Generally, the impact factor for the remaining nuclear medicine journals either maintained performance or improved slightly.

The 2004 general increase in impact factor followed by the subsequent 2005 decrease may represent penetration of innovation in nuclear medicine in other disciplines. That is, 2004 may represent a burst of referencing to nuclear medicine journals outside the discipline for new technology and techniques. An example might include the emergence of PET/CT or SPECT/CT articles in radiology journals and/or oncology journals. An inspection of the relatedness indices for 2003 and 2004 for Semin Nucl Med and J Nucl Med supported this notion. Not only did the number of related journals increase by 62.2% and 9.9% for Semin Nucl Med and J Nucl Med respectively, but more importantly, 2004 saw the emergence of oncology and radiology journals, particularly in the highest ranked 20 journals.

One of the obvious limitations of the impact factor is that it provides a defined 2 year period for calculations. The 2 year window is relatively arbitrary and perhaps would be better suited or be more equitable if a rolling 2 year window were employed. That is, the impact factor is determined for 2006 publications for citations to the years 2004 and 2005. Clearly a December 2006 article could cite a January 2004 article giving an effective currency window of virtually 3 years. This would be particularly advantageous to those journals with 12 or more editions annually.

More importantly, perhaps, is cycle time. Leading journals in respective discipline areas may be cited within the same year of publication. This perhaps provides an indicator of high impact and/or quality yet is excluded from the impact factor calculation. It may also be a function of the average peer review period, the average acceptance to publication period and/or the liquidity of innovation and development in the given discipline. The immediacy index attempts to capture rapid citation. A journals impact or quality might also be highlighted by prolonged citation of articles beyond the 2 year sampling window. The cited half life and citing half life attempt to capture this omission from the impact factor.

The Immediacy Index

The immediacy index (II) is a measure of the citation rate to current articles. The immediacy index is the number of times the ‘average' current article in a journal was cited in journals in the current year. That is, the 2006 immediacy index is based on the number of citations in all ISI listed journals in 2006 of articles published in the specific journal in 2006. Table 2 provides a summary of the immediacy indices for nuclear medicine journals in 2005 and 2006

Figure 4

The immediacy index should be interpreted as an adjunct to impact factor assessment. Figures 2 and 3, however, indicates that for both nuclear medicine journals and all journals included in the ISI JCR, the impact factor and immediacy index are strongly correlated. Thus, the impact factor also provides a reflection of immediacy index.

Figure 5

Table 2: Summary of immediacy indices for nuclear medicine journals in 2005 and 2006. The data provides a strong indication that the has the most immediate impact in industry practice and research. A number in excess of 1.0 indicates that, on average, each article in 2006 is cited at least once in 2006.

Figure 6

Figure 2: The 2005 and 2006 nuclear medicine journal impact factors versus corresponding immediacy indices. A moderately strong correlation is noted with a correlation coefficient of 0.875 and RSquare of 0.766.

Figure 7

Figure 3: The 2006 aggregated impact factors for each ISI category versus corresponding aggregated immediacy indices. A strong correlation is noted with a correlation coefficient of 0.915 and RSquare of 0.836.

Cited and Citing Half Lives

The cited half life is a measure of the median age of a journals cited articles in the current year. That is, half of the citations to articles of a journal in the current year will have been published within the cited half life. Table 3 provides a summary of the cited half lives for nuclear medicine journals in 2005 and 2006.

Cited half life (2006) = years for most recent 50% of citations received by a journal in all 2006 articles

The citing half life is a measure of the median age of an article a journal cited in the current year. That is, half of the cited articles included in a journal in the current year will have been published within the citing half life. Table 3 provides a summary of the citing half lives for nuclear medicine journals in 2005 and 2006

Citing half life (2006) = years for most recent 50% of all citations in 2006 articles of a journal

Figure 8

Table 3: Summary of cited half lives and citing half lives for nuclear medicine journals in 2005 and 2006. The data provides an indication that nuclear medicine journals have a longer cycle time from publication to impact on practice and research.

The cited and citing half lives are a simple calculation with complex implications. One perspective might be that a long cited half live indicates that the impact factor underestimates the longevity of published work in that field. The high cited half life for the J Nucl Med, for example, reflects its prominent position in the discipline and the ongoing citations its articles receive in subsequent years; an enduring impact. An alternative argument might be that a long cited half life simply reflects a long lag phase between completion of research to publication and/or an industry with stasis of innovation and development. The cited half life offers little insight into either the impact or quality of a journal because there is no account for the impact factor of journals making the citations. That is, a long citation half life would reflect very different processes depending on whether the contribution to prolonged cited half life came from high impact factor journals or low impact factor journals; the former providing information about the cited journal and the latter information about the citing journal. Moreover, prominence within a category or discipline area of journals focused on review articles would also prolong the cited half life. The prominence of Semin Nucl Med (100% of articles are reviews) in a relatively small discipline and the predominance of citations to other nuclear medicine journals contributes to prolonged cited half lives. Not surprisingly then, a positive correlation can be seen between nuclear medicine cited half lives and the relatedness index of each journal to Semin Nucl Med (figure 4). The relatedness index is a measure of the strength of cited relationships. No correlation is noted between cited half life and the impact factor of a journal (RSqaure = 0.014).

Figure 9

Figure 4: The 2006 nuclear medicine journal cited half lives versus corresponding relatedness index to . A moderate positive correlation is noted with a correlation coefficient of 0.595 and RSquare of 0.354.

The citing half life might provide a more useful tool for measuring journal quality. A long citing half life indicates that the articles in a journal in any given year predominantly use references to journal articles published some years earlier. Again, this may simply reflect a more sustained period of relevance of published work in that field. However, a negative correlation was noted between a journals impact factor and the citing half life (figure 5). Presumably, this reflects higher impact journals focusing on more ‘current' issues within the profession and, thus, citing more recent articles.

Figure 10

Figure 5: The 2006 nuclear medicine journal citing half lives versus the impact factor. A moderate negative correlation is noted with a correlation coefficient of 0.539 and RSquare of 0.393

Journal Relatedness

The JCR provides a breakdown for each journal of which journals cite articles from that journal (cited) and which journals are cited by that journal (citing). A journal relatedness index reflects the relationship between cited and citing behaviour of each journal. Not surprisingly, nuclear medicine journals generally show strong relatedness to other nuclear medicine journals. There is, however, variations in the strength of relationships that are not elucidated by the cited maximum relatedness index. For example, Ann Nucl Med has its strongest relationship with the J Nucl Med but the relationship is uni-directional. Table 4 provides a summary of the relationships between journals with a ratio of citing to cited relationships tabulated rather than the maximum relatedness. The data shows that the J Nucl Med has the strongest position in the discipline with all other journals citing the J Nucl Med more extensively than they are cited by the J Nucl Med. This is also reflected in the prominence of the J Nucl Med as the highest relatedness to nuclear medicine journals.

Figure 11

Table 4: Summary of cited to citing relationships for nuclear medicine journals in 2006. For each column, the journal with the highest relatedness index is bolded. Where that journal is itself, the second highest relatedness is also bolded. Strength of relationships and the power relationship should be read in columns. Numbers greater than 1.0 indicate that journal for the column is cited by the row journal more than the reciprocal. Only nuclear medicine journals in the top 20 for relatedness have been included.

Journal visibility might be highlighted by the number of journals contained in the relatedness index report. A greater number of journals included in the report would indicate a broader influence. Table 5 provides a breakdown of the number of journals included in the relatedness index for nuclear medicine journals. Not surprisingly, there was a general correlation between journal number and the impact factor. Lack of visibility was generally associated with specialized journal factors (review article, non English language or sub specialty focus).

Figure 12

Table 5: Summary of the number of journals included in the 2006 relatedness index report for nuclear medicine journals.

Number and Type of Articles

The impact factor makes no account of the number of articles a given journal contributes to industry. Indeed, high impact factors are often associated with journals publishing less than 20 articles over the 2 year sampling period. While the category of ‘radiology, nuclear medicine and medical imaging' includes a three journals offering less than 20 articles in the 2004 and 2005 period (Semin Musculoskel R 8, Ultrasonic Imaging 11 and Can Assoc Radiol J 15), nuclear medicine journals tend to offer article numbers commensurate with the number of editions offered annually and the journals profile.

Figure 13

Table 6: Overview of article numbers, types and citation numbers for nuclear medicine journals in 2006.

Self citation is the practice of referencing ones own previously published work in a manuscript (1). For the purposes of this analysis, self citation is the citation of a journal article by articles published in the same journal. Self referencing, on the other hand, is the citation by a journal article of articles published in the same journal. Self citation or self referencing may simply reflect the expertise or special interest of a journal in a particular domain (e.g. J Nucl Cardiol), the prominence of a particular journal within the discipline (e.g. J Nucl Med) or it may be a means by which journals attempt to boost their impact factor. Table 7 provides a breakdown of the 2007 self citation and self referencing rates for nuclear medicine journals. The low rates of Semin Nucl Med can be directly attributed to the review nature and selective themes of the journal. Similarly, the Q J Nucl Med D Mol Im has a large proportion of review articles. The high rates of the J Nucl Cardiol most likely reflects the specialized nature of the journal. The high rates of Nuklearmed- Nucl Med may reflect an active local industry or language barriers (German language journal). A ratio of self citation to self referencing might provide an indicator of the relative impact of a particular journal within the discipline area. This might offer a key indicator for comparing the quality or impact of journals between categories or even of unrelated journals grouped within the same category.

Figure 14

Table 7: Relationship between total citations and self citations, and between total references and self references for nuclear medicine journals in 2006.

The Quality Factor

Ironically, the impact factor does not, in fact, attempt to measure a journals impact but rather its quality. While the impact factor provides a useful rudimentary tool for comparing journal quality within a discipline area, it needs to be interpreted in conjunction with the full JCR. Still, the tool is imperfect in differentiating quality and impact amongst related journals. Moreover, the numerical differences in impact factors offers little insight into the relative impact or quality margin. There is, however, a tendency amongst users to rely on a single index of measure.

Conclusion

Nuclear medicine journals have demonstrated a trend towards increased impact factors in the 2006 JCR. Moreover, nuclear medicine journals have shown an aggregated impact factor increasing at a more rapid rate than that of the ‘radiology, nuclear medicine and medical imaging' category. The J Nucl Med has continued its dominance as the most highly ranked nuclear medicine journal. An analysis of the entire JCR provides universal evidence of the prominence of the J Nucl Med in the nuclear medicine discipline with the J Nucl Med ranking highest for each measure within the report. The impact factor does not, however, provide numeric delineation of the margin of impact or quality differences between individual journals.

Correspondence to

Geoff Currie School of Dentistry and Health Sciences Locked Bag 588 Charles Sturt University Wagga Wagga 2678 Australia Telephone: 61 2 69332822 Facsimile: 61 2 69332587 Email: gcurrie@csu.edu.au