Problems and Progress regarding Sex Bias and Omission in Neuroscience Research

Abstract

Neuroscience research has historically ignored female animals. This neglect comes in two general forms. The first is sex bias, defined as favoring one sex over another; in this case, male over female. The second is sex omission, which is the lack of reporting sex. The recognition of this phenomenon has generated fierce debate across the sciences. Here we test whether sex bias and omission are still present in the neuroscience literature, whether studies employing both males and females neglect sex as an experimental variable, and whether sex bias and omission differs between animal models and journals. To accomplish this, we analyzed the largest-ever number of neuroscience articles for sex bias and omission: 6636 articles using mice or rats in 6 journals published from 2010 to 2014. Sex omission is declining, as increasing numbers of articles report sex. Sex bias remains present, as increasing numbers of articles report the sole use of males. Articles using both males and females are also increasing, but few report assessing sex as an experimental variable. Sex bias and omission varies substantially by animal model and journal. These findings are essential for understanding the complex status of sex bias and omission in neuroscience research and may inform effective decisions regarding policy action.

Keywords: Animal models; journals; neuroscience; sex bias; sex omission.

Graphical abstract

Figure 1.

Articles using mice and rats are a significant and stable proportion of the neuroscience literature. A, From 2010 to 2014, 13,857 neuroscience research articles were published by the J. Neurosci., J. Neurophysiol., Nat. Neurosci., Neuron, Science, and Nature (gray bar). Of these articles, 6,636 used rats or mice and were further analyzed (purple bar). The total number of articles using mice and rats was consistently distributed across years. B, The percentage of articles using rats or mice remained fairly constant across years.

Figure 2.

Sex omission is decreasing but sex bias remains present, with different patterns observed in articles using mice versus those using rats. Articles were categorized as either not reporting sex (orange) or reporting both males and females (red), only males (green), or only females (blue). A, All articles, using both mice and rats. Articles not reporting animal sex decreased from 2010 to 2014. Articles using only male animals increased from 2010 to 2014, comprising the largest proportion of articles by 2011. Articles reporting the use of both male and female animals also increased over time, nearing but not overtaking the percentage of articles using only males by 2013. Articles using only female animals remained stable and low. B, Mice. Articles not reporting mice sex decreased from 2010 to 2014. Articles reporting the use of both male and female mice increased over time and comprised the largest proportion of articles by 2012. Articles using only male mice increased from 2010 to 2014. Articles using only female mice remained stable and low. C, Rats. Articles not reporting rat sex decreased from 2010 to 2014. Article using only male rats increased from 2010 to 2014 and comprised the largest proportion of articles by 2011. Articles reporting the use of both male and female rats increased from 2010 to 2014, but were a much smaller proportion of the dataset than articles using only male rats. Articles using only female rats remained stable and low.

Figure 3.

The vast majority of articles using both male and female animals do not report analyzing sex as an experimental variable. Articles using both male and female animals were evaluated for any formal statistical test or statement that data from males and females were compared, regardless of outcome and whether or not data were reported. The overall percentage of articles incorporating sex as an experimental variable remained low and relatively stable from 2011 to 2014 (∼14%), after a noticeable decrease from the year 2010 (25%).

Figure 4.

Sex omission and bias differ by journal and change from 2010 to 2014. Articles were analyzed from the following journals: J. Neurosci., J. Neurophysiol., Nat. Neurosci., Neuron, Science, and Nature. Four of the six journals showed large decreases in sex omission. Of this group, Neuron showed the smallest decrease, beginning with 69% of articles not reporting sex in 2010, decreasing to 55% in 2014. In contrast, one journal, Science, showed an increase in the percentage of articles not reporting sex, rising from 51% in 2010% to 58% in 2014.

Figure 5.

Patterns of sex omission and bias markedly differ across years by journal. A, Articles not reporting sex. The percentage of articles not reporting sex decreased in five of six journals. The percentage of articles not reporting sex increased in the journal Science. The journals Science and Neuron showed high percentages of articles not reporting sex. B, Articles reporting both males and females. Most journals show increased percentages of articles reporting both males and females, although different patterns occur across time. C, Articles reporting only males. D, Articles reporting only females. The percentage of articles reporting the sole use of female animals remained stable and low in all journals. Green, J. Neurosci.; black, J. Neurophysiol.; blue, Nat. Neurosci.; red, Neuron; orange, Science; purple, Nature.