SOME THOUGHTS ON HOW EVERYONE ELSE IS BAD AT RESEARCH

This is a joke: MOGILab certainly does not believe that everyone else is bad at research. We do believe that research practices can always be improved, and that meta-analytic data on what the status quo actually is represents the first step. MOGILab is always looking for venues to encourage changes to common research practices both within the pain field and more generally.

The Pain Genes Database

This freely accessible web-based database, created in 2007 and updated continuously until 2016, was a compilation of the phenotype of all conventional transgenic knockout lines demonstrated to be significantly more or less sensitive to pain or analgesia. Information was provided about the gene itself, and the parameters of the pain tests and analgesics. A total of 434 knockout lines, and 1008 published papers were coded. The database has been discontinued, by the data contained within it can be downloaded in the Resources section.

 

Meta-Trend Analysis of PAIN Papers

In a project that took most of 2008, Dr. Mogil personally opened, inspected, and coded the features (purpose, subjects, stimuli, measures, and clinical conditions) of all primary research articles published in the journal, PAIN, from its inception in 1975 to 2007. Any number of interesting trends over time were identified, along with full bibliometric data. These data can be downloaded in the Resources section.

Measurement of Spontaneous Pain

In an editorial in PAIN published in 2004 entitled “What should we be measuring in behavioral studies of chronic pain in animals?”, MOGILab was one of the first to point out the disconnect between clinical symptoms of pain and preclinical measures in common use. That is to say, although the most common positive clinical symptom described by chronic pain patients is spontaneous pain (ongoing or episodic), the vast majority of preclinical studies measure experimenter-evoked hypersensitivity states that are comparatively rare in humans. Also see Animal Models for a more detailed discussion.

SABV

MOGILab tests both male and female animals in all experiments, and Dr. Mogil’s first observation of a qualitative sex difference was published in 1993. In 2005, we published an editorial in PAIN entitled “The case for the inclusion of female subjects in basic science studies of pain”. In it, we showed that in the decade from 1996 to 2005, a full 79% of preclinical studies published in PAIN were performed in male rodents only, with an additional 5% using both sexes but not mentioning whether sex differences were observed or not, and an additional 3% not even mentioning the sex of their subjects. Importantly, we provided data showing that in contrast to many people’s expectations female mice do not display higher (estrous cycle-related) variability in pain measurements; this provided the impetus for similar biomedicine-wide demonstrations of equal variability by sex in mice (Prendergast et al., Neurosci. Biobehav. Rev., 40:1-5, 2014) and rats (Becker et al., Biol. Sex Diff., 7:34, 2016). In 2016, Dr. Mogil published a Perspective in Nature entitled “Equality need not be painful”, making the scientific and ethical case for equal usage of female and male subjects in preclinical research.

Reproducibility Crisis

Biology and psychology are currently beset by what is variously termed the “reproducibility crisis” or “replication crisis”. MOGILab is skeptical about the extent to which this is actually a crisis, since many of the statistical and reporting practices claimed to be causing the crisis have actually greatly improved over the past few decades. That being said, it is critically important to address the issues being raised, lest the scientific research community lose the trust of the public. In collaboration with Drs. Jonathan Kimmelman and Uli Dirnagl, we proposed that preclinical papers should make a formal distinction between those experiments that are “exploratory” (in which statistical rigour should be relaxed) and those that are “confirmatory” (in which rigour should be enhanced). In collaboration with Dr. Malcolm Macleod, we published a Comment in Nature expanding on this idea, proposing a three-stage process of preclinical research: 1) exploratory studies, 2) confirmatory study (“preclinical trial”), and 3) generalizability study.

Meta-Analyses of Analgesic Papers

In collaboration with Dr. Jonathan Kimmelman, Dr. Mogil participated in the design of meta-analyses of preclinical research and post-approval trials of the blockbuster analgesic, pregabalin.

Use of Inbred Mice

The most common preclinical research subject in biomedical research (and the second-most common in pain research) is the C57BL/6 mouse. Beyond questions about the suitability of this particular strain for pain research (see Pain Genetics), it is a strange decision by the biomedical research community to have defaulted on an inbred animal. Although inbred mice present advantages for genetic and immunological purposes, they are, um, inbred, and subject to the vagaries of inbreeding depression and lack of heterozygosity. Why are they so popular? Probably because of the reasonable-sounding expectation that they would show less overall variance than outbred mice. We tested whether this expectation was actually true, by comparing the variability of inbred and outbred mice tested and reported together in published experiments across biomedicine. We found no significant differences in variability between inbred and outbred mice overall, or in any subcategory. We recommend that for general purpose studies outbred mice animals should be used.