More on Impact Factors

Abstract

To the Editor: We chuckled at the series of commentaries on impact factors in the May 2008 issue of Epidemiology.1 Miguel Hernan provides an entertaining parable on epidemiological concerns regarding the “brain irritability factor” (BIF); we feel there are other lessons regarding epidemiologic reasoning that can be gleaned from this story. Epidemiologists know that time series analyses can be useful in identifying why events occur. Why should Epidemiology publish a series of articles critical of impact factors commissioned in late 2007 when in 2000 the journal proudly declared, “We have grown steadily … in impact factor,”2 with no disclaimers attached? The Figure displays impact factors for the 3 journals discussed by Hernan; the first arrow indicates when pride was expressed by Epidemiology regarding its growing impact factor; the second arrow when serious concerns were raised. An epidemiology 101 question could ask: “What events may have generated the change of heart between 2000 and 2007?”FIGURE 1.: Impact factors for 3 journals available up to the time this letter was submitted.9Hernan points out that the proportion of seizures detected in the same hospital where patients were admitted was around 5% in Baltimore and Durham and 12% in Bristol. Again an elementary concern would arise that there are other levels of grouping that may throw light on such differences. Baltimore and Durham could be in the same overall administrative health district (let us call it “Empire,” for short), and Bristol could be in another, which, given the global hegemony of Empire, we may as well just label “Not Empire.” If the Empire hospitals show a strong bias to admitting Empire patients and excluding Not Empire patients, even when it is clear that the Not Empire patients have greater need, then of course all the hospitals within Empire will have much higher BIFs, without there being elevated readmission rates for any particular Empire hospital. In the nonhegemonic Not Empire there may be few hospitals that survive, and therefore if the increasingly globalized influence of Empire is resisted and admission is driven by patient need rather than origin, the beleaguered surviving hospitals will have raised readmission rates, even though on average Not Empire hospitals score markedly lower BIFs because of the inward-looking practices within Empire.3 However, Not Empire science, in one of its more organized forms, has indicated recently a need to keep up with Empire standards of behavior in this regard.4 Here in Bristol, we have a relaxed but amused relationship with impact factors. Accompanying our recent reprint of Eugene Garfield's classic paper on citation indexes,5,6 we pointed out that the median number of citations for articles published in the IJE in 2003 was, by the middle of 2006, zero.7 Richard Smith, the former editor of the (Not Empire) BMJ, was concerned that an obsession with impact factors could lead to journals becoming less readable and entertaining.8 We are unperturbed at the prospect of readers choosing which epidemiology journal they read on the grounds of readability and entertainment. However, we feel that when viewing the figure here and considering the timing of the change of heart at Epidemiology it is difficult not to raise a smile, and think that, on this occasion at least, impact factors have led to something amusing (if perhaps unconsciously so) appearing in an epidemiology journal. George Davey Smith Department of Social Medicine University of Bristol Bristol United Kingdom [email protected] Shah Ebrahim Department of Epidemiology and Population Health London School of Hygiene and Tropical Medicine London, UK [email protected] Despite our success in this year's impact-factor sweepstakes (data not shown in the figure), our editorial position is unchanged. We invite the editors of IJE to take more seriously the mounting abuses of impact factor. Read on. —The Editors