Considering Sex as a Variable at a Research University: Knowledge, Attitudes, and Practices

Abstract

Biomedical research has a history of excluding females as research subjects, which threatens rigor, reproducibility, and inclusivity. In 2016, to redress this bias, the U.S. National Institutes of Health (NIH) implemented a policy requiring the consideration of sex as a biological variable (SABV) in all studies involving vertebrate animals, including humans. Unless strongly justified, females and males must be included in all studies and results reported disaggregated by sex. Recent evidence indicates, however, that misunderstandings of the policy and other significant barriers impede its implementation. To shed light on those barriers at our home institution, we conducted a study funded by the Emory University Specialized Center of Research Excellence on Sex Differences (SCORE). In semistructured interviews of Emory principal investigators in the biological sciences, we noted their knowledge of what the policy entails and why it was implemented, their attitudes toward it, and the extent to which it has or has not changed their research practices. Although attitudes toward SABV were generally positive, most researchers face challenges with respect to its implementation. We suggest interventions that can be mounted at the level of home institutions, such as raising awareness of locally available core facilities, to help address these challenges. More training is needed on what the policy asks of researchers, how sex is defined, the nonhormonal ways that sex differences can manifest, and best practices for statistical analysis of sex-based data. Home institutions may also want to explore ways to lessen the stress associated with rollout of SABV policy.

Introduction

Biomedical research in the U.S. has a history of being conducted primarily on men and males.¹ Beginning in the 1980s, the U.S. National Institutes of Health (NIH) has worked to redress this androcentrism by promoting the inclusion of women and females in the research they fund. In 1986, NIH began encouraging researchers to include women in study populations.² This action was followed by the NIH Revitalization Act of 1993, which mandated the inclusion of women and underrepresented minorities.³ Since then, although recruitment of women into clinical trials has been uneven across disciplines,^4,5 representation by women has increased dramatically overall.⁶

Although the 1993 NIH Revitalization Act increased the number of women recruited into clinical research, there has been less progress in the areas of basic and preclinical research.¹ In response to widespread calls to action,^7–9 NIH implemented a new policy¹⁰ in 2016 entitled “Sex as a Biological Variable” (SABV). This policy, which targets preclinical research specifically, requires proposals to describe plans to account for “sex” (see Barr and Temkin¹¹ for definitions) and to provide a strong rationale for any single-sex research design. By addressing the longstanding overrepresentation of male nonhuman animals and cells in biomedical research, SABV is intended to increase not only inclusivity but also rigor and reproducibility.¹²

Early assessments of the impact of the SABV policy suggest that although females are now included more often, they are still underrepresented.^13,14 Moreover, few researchers are including sex as a variable in their analyses.¹⁴ Thus, although research is more inclusive than before, the policy's full intent to increase parity and consider the influence of sex has yet to be realized.

In this study, our objective was to identify barriers to SABV implementation faced by preclinical and basic science researchers—particularly gaps that are amenable to interventions locally, within our home institution. We partnered with the Specialized Center on Research Excellence of Sex Differences (SCORE)¹⁵ at Emory University (Fig. 1), which is a center funded by the Office for Research on Women's Health (ORWH) at the NIH. Within the Emory SCORE, the Career Enhancement Core (CEC) develops new initiatives to help achieve the goals of SCORE, such as normalizing the inclusion of sex as a variable in all research domains at Emory and the Southeast region. Thus, our goal to identify barriers to SABV implementation were significantly aligned. We interviewed principal investigators (PIs) at Emory, focusing on gaps in knowledge and issues related to attitudes and practice. Our long-term goal is to use Emory as a model for overcoming the challenges we face while at the same time providing helpful insights relevant to other institutions.

FIG. 1.

The Emory SCORE comprises three main cores, which work together toward advancing the quality of women's health research at Emory University. The current study was conducted with funds from the CEC, the mission of which is to facilitate and normalize research that includes females and males.

Materials and Methods

The Emory University Institutional Review Board approved all procedures. Eligible participants were identified through purposive snowball sampling, focusing on basic scientists. Inclusion criteria were having a faculty appointment at Emory University and a record of funding as a PI. A total of 18 PIs were contacted by email for recruitment into the study. Three (17%) did not respond to the invitation and one (6%) responded but declined because their research direction had recently changed. We thus had an 83% response rate (n = 15/18) and 93% enrollment (n = 14/15).

We conducted semistructured interviews with the 14 enrolled PIs to explore their understanding of and responses to NIH's Policy on SABV. We asked questions focused on the following themes: (1) Knowledge about SABV policy, for example, “Why was SABV enacted? What does it require of researchers in your field?” (2) SABV implementation, for example, “Has SABV changed your research program at all? Do you perceive barriers?” (3) Data analysis and reporting, for example, “Do you account for sex in statistical models? Do you report data separately?” See Supplementary Table S1 for the complete interview guide. All interviews were conducted by Zoom during 2021; each lasted 60–75 minutes and was conducted by DLM and KK (n = 13) or DLM alone (n = 1). Verbal consent was obtained at the beginning of each interview.

All interviews were recorded. The audio files were submitted to a professional service (Temi or Rev) for transcription and then cleaned by DLM and two student assistants. The study team used NVivo software (v.12 for Mac) to codify the content of each transcript. Interpretive differences among the coders were discussed until an agreement on each was reached. The coded portions of the interviews were exported from NVivo and reviewed to identify emerging themes.

Approximately 1 year after the interview, PIs were contacted by email with two follow-up questions. First, because we had not asked about participant's gender in the interviews, we asked for that information using an open-ended question. Second, we asked whether participation in the interview itself had led to any changes in thinking about SABV or alterations of research methods. Thirteen of 14 PIs (93%) responded.

Results

Self-reported demographics are presented in Table 1. Faculty appointments were spread across eight Emory University departments, located in either the School of Medicine (71.4%) or College of Arts and Sciences (28.6%). Half (n = 7/14) of the PIs held advanced degrees in Neuroscience; the remainder held degrees in Physics, Biochemistry, Cell Biology, Evolutionary Biology, Physiology, or Psychology. Although most (n = 12/14; 85.7%) had NIH funding, two (14.3%) were funded only through non-NIH sources. Model organisms included humans, monkeys, birds, mice, and voles. Two PIs worked with SABV-exempt models (Drosophila, C. elegans). About two-thirds of the PIs (n = 9/14) worked with multiple animal species.

Table 1.

Information About Participants

Variable	Frequency	Percentage
Gender
Man	8	57.1
Woman	5	35.7
No response	1	7.1
Highest degree
PhD	13	92.9
MD/PhD	1	7.1
Year of degree
1980–1989	1	7.1
1990–9199	6	42.9
2000–2009	4	28.6
2010–2019	3	21.4
School affiliation
ECAS	4	28.6
SOM	10	71.4

ECAS, Emory College of Arts and Sciences; SOM, Emory University School of Medicine.

Knowledge

Knowledge of what SABV asks of researchers

All PIs had heard of the SABV policy put forth by NIH in 2016. When asked to describe details of the policy, however, some PIs could not, saying, “I'm aware there are policies and that it's a component of grants, but that's all,” or “I saw some things on Twitter.” For the other PIs, understandings of the SABV policy were wide ranging. Some PIs argued that including both males and females is not a requirement; rather, a PI needs only to consider including both sexes. A small number of PIs explained that unless there was justification to do otherwise, studies must have roughly equal numbers of males and females but neither disaggregation by sex nor comparing the sexes is required—the sexes can be pooled. More PIs, however, argued that when males and females are included, they must be separated and compared: the policy “requires actually building [sex] into your analysis to see if there are sex differences.”

Some descriptions of SABV policy were clearly misconceptions. These included the belief that the policy has not yet gone into effect: “I don't think there are any requirements. I think it will come, but I don't think there is at this moment,” that those studying basic biology are exempt: “You can justify [not using both sexes] by saying the type of question is so basic and fundamental to both sexes that it's unlikely to have any differences,” and that sample sizes must be doubled to include males and females: “You have to essentially ask the government for double the amount of money.”

We also noted the common misconception that accounting for sex requires consideration of steroid hormones; one PI argued, for example: “People need to say in their grants that if they find a sex difference then they will examine testosterone and estradiol.” Similarly, when asked “Do you test for sex differences?” one PI responded, “No, we have not measured hormones.” An apparent equivalence of “sex” and “hormones” was referred to by some PIs as an obstacle to incorporating SABV into research designs: “A barrier to comparing the sexes is that we don't have a good way to assay sex steroids” and “Most researchers just aren't trained to look for sex differences; estrogen signaling is really complex.” Others used arguments about hormones to justify not incorporating SABV: “We don't have to worry about sex differences because we are studying neonates before they have hormones and cycles.” Such comments seemed equally likely to be made regardless of whether the PI's research program focused on hormones or not.

A related misconception was that PIs must track estrous cycles in females: “So now [in addition to males] we've got two groups of females. Some in estrous, some in diestrus or metestrus. So, all of a sudden, the study has ballooned out… how do we do this?”

Knowledge of why SABV was implemented

All PIs were supportive of the SABV policy and spoke positively about the rationale for it. When asked why the policy was implemented, most pointed to a historical underrepresentation of women and an androcentric bias coming from “only looking at male model organism species…basically, ignoring any sex differences in the research.” Many were concerned about the applicability of research findings from men/males—one researcher specifically noted clinical trials had been biased toward white males—to all people/animals, noting that there might be important differences between men and women that influence disease vulnerability and treatments. As one researcher said, “there are incredibly huge biological differences between different sexes. And when we're doing studies it's really important to consider the physiology of both, the genetics of both. And if we're going to come to these conclusions it's pretty important to be inclusive with our samples, so that we can actually be accurate in what we say.”

Only two researchers mentioned rigor or reproducibility unprompted, for example: “This is part of the push toward rigor, kind of getting a better understanding of exactly what our data are trying to tell us, particularly given the variability.” While one PI said they had not heard about SABV being designed to address rigor, “I have not heard NIH use that so much in terms of describing sex as a biological variable,” others noted, “rigor and reproducibility was an obvious problem” and “we have to be careful about how we collect our data.”

Of the 14 PIs, only 4 (29%) described receiving training about SABV policy; 3 mentioned training from NIH in the context of reviewing grants, and 1 PI, who was affiliated with Emory SCORE, mentioned being aware of SCORE-related relevant materials. No PI described other kinds of training about the policy, including materials on the NIH website. A majority of the PIs had an editorial role at a journal; yet only one knew the journal's policy on SABV. A typical response was, “I don't know. I was supposed to read all that stuff, but I don't know.”

Attitudes

As noted above, PIs were generally supportive of SABV and agreed with the rationale for its implementation. One PI remarked, “It will help your research. It might actually reveal new biology.” Another commented, “I think it opens my mind a little bit in terms of understanding yet another important layer of variability.” When asked whether they had heard about disagreements or controversy surrounding SABV, several PIs responded they were aware of none. One noted, “I've never heard anything negative;” another remarked, “I think, in my field, it's well-accepted.”

Although most PIs were enthusiastic about the need for the SABV policy, attempting to comply with it was seen as a significant source of stress. The majority of PIs expressed some sort of negative feelings, for example coercion: “Congress is forcing this on us,” helplessness: “I'm really struggling right now,” hopelessness: “You're forced to consider it but, in the end, you can't do it,” exasperation: “another hurdle we have to jump over,” being overwhelmed: “It's been a frustrating and daunting task,” fear: “people feel that if they are not now doing male and female in everything they propose, they're going to be penalized,” resignation: “At some point, researchers get, … I don't want to use the word ‘tired,’ but… you have to pick your questions. You can't just do every experiment, every scientific question to address sex as a biological variable,” and abandonment: “The administration says: ‘We know it costs a crazy amount of money, just ask for more money to go over the cap.’ I'm like, ‘No, that's not how things go.’”

Despite these stressors, or perhaps because of them, almost all PIs spoke with a sense of exceptionalism. They stated that their field or their laboratory was special either because they were uniquely prepared for the policy's arrival (“My research community was aware that these changes were coming from NIH …. We knew it already”) or uniquely unable to accommodate the policy's requirements: “I think we're in a very special field… Are we actually capturing sex as biological variable? It's kind of an elephant in the room because I'm not sure it's even possible to really do it.” Some PIs seemed skeptical that SABV was being implemented by others, saying for example, “I do see attitudes that are hard to challenge. There're some more senior investigators sometimes that don't always implement [the policy].”

Practices

The majority of PIs indicated that they were already compliant with SABV policy at the time it was implemented. One PI commented, “No matter whether there's policy there, that's our design. We need to consider the sex… we want to know whether there's a difference. We did that before, I mean before 2016.” One PI insisted, “[we have] always balanced for sex. Always. Literally I just had the grant reviewed and study section was like, ‘[they're] all over the SABV thing, [they've] got it down,’ and it was the exact same language that I've been using probably for 10 years.” Even a PI working with fruit flies, which are SABV exempt, emphasized that including both sexes is better science: “We knew that if we [presented results] based [only] on males, then people probably shouldn't believe us… We were just trying to make sure that we said stuff that we could believe.” In at least one case, both sexes were already being used to mitigate costs: “It's really hard to just discard an entire sex… we just wanted to take advantage of our transgenic models and not just waste an entire group.”

A few PIs indicated that they have changed their research practices to comply with the policy. One PI commented that before the policy, “we were just as guilty as others… it was easier to develop the methods where we were controlling as much as possible. Yeah, [SABV is] something that we think a lot about now.” Another PI noted that in grant proposals, “Before [2016], I would simply say there's no reason to believe that there are any [sex] differences… I did not specifically monitor the sex. But now I do.” A PI working with cell lines explained, “We do pick our lines more carefully, right? At the beginning. Even before we start, we say like, ‘Let's use these four, two male, two female…’ And so I think that it actually has had a concrete influence on even designing the experiments.” PIs also noted changes in the way they respond to talks and articles in their field: “It comes up a lot question-wise whenever someone is presenting, if they weren't upfront whether it's males or females, people will always ask.”

PIs conducting single-sex studies reported a variety of justifications for including only one sex. Some explained that they were studying sex-specific or sex-biased conditions; others did not have access to equal numbers of males and females (see Barriers and Challenges, below). In some cases, the PI justified working with only one sex because they did not believe that there would be sex differences. “If I'm looking at a fundamental cell biological question, like how muscle contracts or how a motor moves on a microtubule and what is the role of a particular amino acid, probably those types of basic fundamental questions do not really warrant looking at sex as a biological variable,” argued a cell biologist. They went on, “If I'm asking a broad, circuit-based question about learning and memory, yes by all means I should analyze male mice and I should analyze female mice.” Interestingly, a PI who was studying memory had a different view: “If we're looking at working memory and… we're asking about general, like how do these systems respond to variables like delay or interference? I don't have a lot of reason to believe there's differences.”

Barriers and challenges

When asked if they perceive barriers to SABV implementation, only three PIs responded “no.” One qualified their answer, however, saying, “I, personally, haven't encountered specific barriers… I only see challenges.” All PIs did note factors that hinder compliance with the policy (Table 2). The high cost of increasing sample sizes, for example, was mentioned by 100% of PIs. Some PIs had access to only one sex of animals; one of these felt pressure to change their research program to investigate female-specific conditions (Table 2).

Table 2.

Challenges that Hinder Principal Investigator Compliance with Sex as a Biological Variable Policy

Challenge	Description	Example of PI comment
Financial cost	PI is working with expensive animals or cell lines	“The biggest [barrier] I can imagine is that you can implant in three animals. You could do two [of one sex] and one [of the other], but then you have n equals 1 for one [sex] and that just doesn't fly.” “There are many times where we think we can do this with two cell lines or three at most. And then it crosses our mind, what if we want to look for sex differences? Then we're in a pickle.”
	Housing per diems and other costs associated with care	“Let's say, for example, a female gives birth to four animals. Three are females and one is a male. Unfortunately, when we wean them, we will have one cage with three females and one cage with one male. If we were to separate the females to match the male, we couldn't afford our animal costs.”
	Supplies	“We have to ask for, basically, twice as many electrodes now.”
Time	Delays in career advancement	“What a lot of people think is, ‘Oh, that lab had a graduate student comparing males and females for a year. It took that person all this extra time, and they didn't find anything.’ I think you hear a lot of that.”
	Delays in completion of project	When asked why a male-only study was not scheduled to be replicated in females: “Because we need to finish the project. We are trying to finish it on time.”
	Publication delays	“If you have the same effect in both sexes, it's completely uninteresting… it's virtually unpublishable. If you've already published females, you can't go back and publish males… no one's going to look at it.” “What if there is a difference? Suddenly you're in a really bad position. You can't publish the work until you figured out why there's a difference and what's going on. Most people aren't equipped to do those kinds of studies.”
	Delays related to balancing for sex	“So maybe we were aiming for five males and five females of this genotype, but we're going to have to wait for another 3 months for the next litter to get one more female mouse that's the same genotype. So then we end up with six and four. I mean, in the end, is that really going to affect our experiment? So we go with the six and four because it costs us three more months [to balance the study].”
Disaggregation not feasible	Tissues are dissected from multiple animals (e.g., embryos) and cells are mixed in cultures	“We take the whole litter and dissect all the hippocampi and trypsinize it and plate these neurons… It's a network of male and female neurons [in culture].” “If I were to look at sex as a biological variable in that model… it would just not be feasible. It can be done, for example, genotyping an individual embryo, but you wouldn't get enough neurons to do an experiment.”
	Sex is unknown and not feasible to determine	“We studied [wild birds], and you can't tell [the sex] without physically opening them up and looking. For the most part you can't tell and so it just was ignored. So, unless there was some bias in trapping, you probably were getting a pretty good mix of the two sexes.”
Sex differences in natural history of model organism require nonidentical housing arrangements	Male mice must be housed singly to avoid aggression, whereas females can be housed in groups to save housing costs	“Female mice are often much easier to work with. They tend not to be as aggressive as the males. They're easier to house. Males you have to house together from a very early age. If you try and mix them later, they tear each other to pieces, so they have to house singly. It increases your costs and your number of cages. It creates complications with the IACUC. You're not allowed to house mice singly without having formal approval of that. It's certainly easier to work with a female model of mice.”
	At EPC, colonies of rhesus monkeys are housed in seminatural enclosures; because in their natural habitat males disperse, EPC removes most of the males, which go to laboratory-based studies. PIs typically have access to only one sex	“Primate research is already at a risk of just being so expensive it doesn't get done. If it starts being that you have to have an equal number of males and females… I think that would hurt the field.” “You can't study something in male monkeys that doesn't make ecological sense. So, you'll never find an instance where you have a large group of males hanging out to be able to study them… I can't study males [in that context] because they just don't exist.” “I'm really struggling with it… I don't think we can say any longer that females are more at risk [for the condition we study] and we've got to study this only in females… but females are the only monkeys that we have available. So how do you then shape the question so that you could focus it more on something female-specific?” “The center doesn't maintain animals in a way so that I can do the studies.”
Trade-offs between SABV and optimization of protocols	Balancing for sex can require choosing suboptimal samples for a study	“There've been times when we picked the lines we wanted based on other things… they're more reproducible, robust… working really well. And then we go and look, and they're all male… and then we were like, ‘Maybe we have to change it.’”

EPC, Emory Primate Center; IACUC, Institutional Animal Care and Use Committee; PIs, principal investigators; SABV, sex as a biological variable.

Although many of the barriers faced by PIs are not easily addressable (Table 2), we did identify some “low-hanging fruit”—factors that could be tackled with relatively simple interventions at the level of the home institution. First, we found that PIs were sometimes not aware of resources that were already available to them through Emory. One PI, whose interest in sex differences extends to hormonal mechanisms, commented, “It would be awesome if we had a core facility on campus where we could send a blood sample if we wanted to test for hormone level… But as of now, there's not a way that I could do it here. We would have to ship them off to a company.” Emory does, in fact, have a Biomarkers Core where steroid hormones can be sent for assay; thus, there was an easy solution to the PI's quandary. Another PI recalled their bewilderment when faced with the prospect of tracking estrous cycles and manipulating hormones, saying, “We had no idea where to start with this… most of [the relevant literature] was published in the 1950s. Good luck finding those papers.”

The PI solved these problems by consulting with the Emory Division of Animal Resources, which provided the PI's laboratory with training and protocols on vaginal cytology, ovariectomy, and orchidectomy in rodents. “They were incredibly helpful in getting us started,” said the PI, “They also had tips for, well, if you want to synchronize all your females, throw in male bedding three or four days before you want them to be in estrus and boom, they'll be in estrus. So that was a really easy way to learn fast from experts… that was our biggest barrier.” Although we emphasize that tracking cycles and measuring and manipulating hormones are not required by SABV policy, for researchers with interests in these areas, helpful resources may be available locally.

Second, we found a widespread lack of expertise in the statistical analysis of sex-based data. Of the PIs using factorial designs, that is, those testing for effects of a treatment in two sexes, only a minority (3/10, or 30%) reported testing for the statistical interactions that are required to claim that the sexes responded differently to the treatment. The rest used invalid approaches^16–24 to compare the responses across sexes. The most common of these invalid approaches was to test for effects of treatment separately within sex and to conclude a sex difference if the p-values were discordant (Table 3).

Table 3.

Many Principal Investigators Endorsed Invalid Statistical Approaches to Test for Sex-Specific Effects

Example #1
PI:	Let's say we have really large differences from baseline to 3 weeks in the females, but that's not there in the males. It's a possibility that, as a population, there still will be change. But when you look at the differences, you'll see, “Wow. The difference is really more in the females, and it's driving this.” So, not so much the males. Those things will come out in the numbers or in the test.
DLM:	Do you ever look for an interaction between the time points and sex?
PI:	Never sex, actually, not that I can remember. I don't think I have.

Example #2
DLM:	If you are testing for a sex-specific effect of treatment, would you look at the interaction? Or would it be enough if you saw an effect of treatment in the males and not in the females? If there were no interaction, would you call that a sex difference in the response to treatment?
PI:	Yes, because you would expect if there's a difference between males and females, if there's no interaction… the treatment may not actually be different. It may just be because they're males and females. So there doesn't have to be an interaction.
DLM:	So to say there's a sex difference, as long as you're able to detect that effect of treatment in one sex and not the other, that's enough?
PI:	Yeah.

Example #3
DLM:	Let's say that you've found a significant effect, meaning a p-value of <0.05 in one sex, but not the other. Is that enough to call that a sex difference? Would you say that's a sex difference?
PI:	Usually.

The examples illustrate a common statistical error known to produce inaccurate and biased results.^16–24

Effect of the interview itself on attitudes and compliance

During the interviews, several PIs mentioned that participation in our project heightened their interest in SABV and that they planned to make changes in their research protocols as a result. They made comments such as, “This is going to influence what I do in my position or in my interactions with trainees,” and “You're helping me write my grant!” To investigate the extent to which PIs actually did change their approaches, we followed up by email approximately 1 year later. Of the 13 PIs that responded to the question, 9 (69%) indicated that participation in our study caused them to pay more attention to SABV in their own work or that of others. For example: “The interview inspired me to dig deeper into the literature on sex differences in my field; the huge gaps were somewhat shocking” and “the interview stimulated discussions with lab members and colleagues.”

Four PIs reported that their research protocols changed as a result of the interview. For example, one PI said, “my whole lab has re-evaluated our statistical approaches to sex differences.” Another PI plans to determine the sex of samples and test for sex differences in their future work. One PI now pays closer attention to housing density: “In a lab meeting, I stressed the importance of group housing and social constructs and cage activity/grooming. If we have a single-housed male, they can be used for breeding purposes, but not experiments.” One PI, who is the director of a core, shared that as a result of the interview, “we have been more proactive to recommend that core facility users consider SABV… when we observe some skewing [toward one sex] we bring the importance of SABV to their attention.”

Discussion

In this study, we identified potential barriers to compliance with SABV policy at Emory University. Although our sample size was small, our findings may be used to ground the development of data-driven interventions that facilitate and promote sex-inclusive preclinical research. Although some of the challenges we identified will be difficult to address (Table 2), particularly those faced by researchers using nonhuman primate models, we identified issues that can be addressed by relatively simple interventions. These interventions include providing training on exactly what SABV asks of researchers and increasing awareness of locally available SABV-relevant resources, which may lessen stress and confusion associated with the implementation of SABV policy. We believe such interventions, described in more detail below, could be implemented not only at Emory but also at other institutions; in particular they could be integrated into future SCORE CECs or perhaps overall SCORE RFAs.

Further training is clearly warranted in two important areas. First, many PIs demonstrated slippery and confusing conceptualizations of sex as a variable. NIH defines sex as a multidimensional construct based on a cluster of independent but correlated traits, which include chromosomes, gonads, and hormonal milieu.¹¹ As such, sex must be defined explicitly and precisely in the context of each study and remain stable throughout.²⁵ Our data show room for improvement; while describing their research practices, PIs often shifted from “sex” to “hormones” somewhere between developing hypotheses and interpreting results, sometimes even using the terms interchangeably. Defining “sex” is critically important when considering SABV, and will become even more important as biomedicine moves toward foregrounding consideration of gender²⁶ and transgender health.²⁷

Second, training is urgently needed in the analysis of sex-based data. Garcia-Sifuentes and Maney²⁸ reported high rates of invalid statistical approaches to test for sex-specific effects of a treatment; appropriate statistical tests were employed in only three in 10 studies. Our findings were remarkably similar—only 3 out of 10 PIs who test for sex-specific effects described valid approaches. The rest reported approaches that increase rates of false positives to as high as 50%.¹⁶ This finding, which echoes many other calls to action,^16–24,29 is concerning because without rigorous statistical approaches, increasing the number of sex comparisons could falsely inflate the number of sex differences reported, hindering progress instead of promoting it.

One of the major practical challenges associated with these suggested interventions is how to disseminate information that is not actively being sought. In carrying out our study, we discovered that the mere act of participating in the interview increased awareness of SABV policy and caused PIs to more carefully consider their own implementation of it. Thus, institutions might consider a formalized process to conduct interviews or surveys, which would serve two purposes: to collect information about SABV compliance and to disseminate information about the policy and how to implement it effectively.

We focused here on the NIH's SABV policy, which was introduced to address bias in basic research on nonhuman animals and cells. Accordingly, we recruited PIs engaged in those types of research specifically. Only one researcher from the social sciences was interviewed, and none were conducting clinical trials on humans. Expanding the scope of recruiting participants for interviews, for example to clinical and social researchers, would permit assessment of sex-inclusive research beyond that targeted by the SABV policy.

Finally, our sample size was small and limited to a single institution. Nonetheless, our findings were consistent with a larger study in which PIs at multiple institutions were interviewed about SABV.³⁰ All findings reported in that study were replicated here, including largely supportive attitudes toward SABV, barriers related to cost of animals and housing (particularly nonhuman primates), and perceptions that the policy is unfunded. We plan to use the data from our interviews to generate questions for a survey that will reach hundreds of Emory researchers, which will permit a more quantitative analysis of responses.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported in part by a grant from the NIH SCORE at Emory University (U54AG062334), Pilot Project Grant “Identifying Barriers to Implementation of SABV at Emory University,” PIs DLM and KK, PO KBSH and by a grant from the Emory University Research Committee.

Supplementary Material

Supplementary Table S1