The Politics, Promises, and Perils of Data: Evidence-Driven Policy and Practice for Menstrual Health.

Menstruating bodies are political. This has deep implications for what we know and understand about menstruation, the menstrual cycle, and menstrual health ( Dan, 2004 ). Women’s bodily processes, menstruation and (peri-)menopause, are over-whelmingly viewed negatively ( Martin, 1988 ). Health conditions such as endometriosis are understudied, leading to delays in diagnosis and treatment, arguably due to a gendered dismissal of women’s pain and neglect of menstrual disorders in health care ( Anand et al., 2018 ; Dusenbery, 2018 ). The present momentum around menstruation comes with opportunities to shift some of these dynamics and to ensure that menstrual health becomes better understood and the needs of menstruating bodies are met. However, viewing menstrual health as a new and shiny “agenda item” in global development leads to new risks. We see a risk of rushing into policy and programmatic decisions without fully understanding the needs, interests, and priorities of the menstruators these programs are meant to serve. The sensationalization expressed in “1 in 10 girls in Africa miss school” demands quick fixes, and menstrual products are presented as the solution—yet without fully understanding the lived realities of girls and drivers of school absenteeism in vastly diverse countries across a large continent. Moreover, experience shows that rushing programmatic decisions often neglects difficult-to-reach marginalized populations ( Fukuda-Parr et al., 2014 ). The Sustainable Development Goals commit to inversing this trend through the commitment to “leave no one behind” and to reduce inequalities, but the actual collection of disaggregated data to monitor whether this commitment translates into practice is limited ( Winkler & Satterthwaite, 2017 ). This general trend also holds true in the context of menstruation; data on the needs and experiences of marginalized menstruators are limited. Over the course of our conversation, three key approaches have emerged that can help address some of these risks: contextualization, communication, and collaboration. First of all, context matters. Data must be fit for purpose. What kind of data we need depends on whether we work at the global, national, or local level, and we must pay attention to the particular contexts embedded therein. The objectives of data use also matter and must be transparent: Do we need data to improve our fundamental understanding of the menstrual cycle, to raise awareness, to determine which populations to prioritize, or to develop detailed programming? Second, we need to improve on communicating research results. The reason that the “1 in 10 girls” statistic continues to circulate is that there is a demand for easily accessible, simple, even simplified statistics. Yet, the reality is much more nuanced. For instance, with regard to school attendance, a multitude of factors are likely at play, and the role of menstrual stigma has thus far been underexplored. But we need to do better than simply acknowledging complexity. Using policy briefs, infographics, and other materials and developing clear policy recommendations can go a long way in communicating research findings that capture the rich potential of data-driven policy and practice. Third, and most important, we need to strengthen collaborations between researchers and practitioners. There are obvious challenges (e.g., related to the different timeframes of research and policy development), but the approaches we discussed based on participatory methods, mixed methods, and the use of data generated by apps bear significant potential. We all stand to benefit from such collaborations that drive evidence-based programming. At a time where there is enormous momentum around menstruation, the time to develop these collaborations among the NGO community, advocates, funders, and researchers is now.

At this point, reviews of the literature and the highest-quality studies reveal more about what we do not know than about what we do. Consider, for instance, Hennegan and Montgomery’s (2016) systematic review of quantitative studies on MHM in which they asserted that there is “insufficient evidence to establish the effectiveness of menstruation management interventions” (p. 17). More recently, Hennegan (2020) wrote: “despite enthusiasm and the best intentions, most [MHM interventions] are untested, and there is limited evidence to inform effective practice.” Even when carefully designed studies are conducted, the results are not encouraging. In a randomized controlled trial in western Kenya, researchers provided menstrual cups to schoolgirls to reduce absenteeism and found that the intervention was not statistically significant ( Benshaul-Tolonen et al., 2019 ). Regarding the causal link between traditional menstrual care (primarily the use of repurposed cloth as an absorbent) and poor health outcomes, a systematic review ( Sumpter & Torondel, 2013 ) showed that menstruation does present challenges for women in resource-poor settings, but that the relationship between traditional methods and health was unclear. One study ( Das et al., 2015 ) did show that some menstrual care practices, including type of menstrual absorbent, can raise the risk of urogenital symptoms, but wealth and degree of privacy were protective against some infections, such as bacterial vaginosis. Finally, Hennegan et al.’s (2019) systematic review and meta-synthesis of 76 qualitative studies led them to conclude that even our definition of MHM is too narrow: It is overly focused on hygiene practices, a lens that overshadows the psychosocial realities of menstruating. Lauren, can you expand on this? Where do you see a lack of data on menstruation?

My research follows a biocultural approach to breast cancer prevention. I explore both biological and cultural risk factors and focus on hormones as the mechanism that links what happens above the skin with what happens beneath it. Hormones inform both cancer etiology and risk assessment. I include cultural factors in my work because I started out as an anthropologist, so I see culture as fundamentally embedded in biology. I also see understanding the cultural context as an opportunity to accelerate the current 17-year time lag between the generation of research evidence and its translation into clinical practice ( Balas & Boren, 2000 ). I want to implement the findings from my etiologic and risk assessment studies, so that interventions and prevention tools are culturally compelling, which will ultimately make them more effective. I believe research is more likely to achieve impact through the integration of theories and methods from different disciplines—in my specific case, epidemiology and anthropology. In epidemiology, the foundation of the field is the two-by-two table. This is a simple four-cell grid with the number of those with and without exposure to a given disease on the vertical axis and the number of those with and without disease on the horizontal axis. For me qualitative data are everything outside the two-by-two table. They help us to incorporate the emic—a key concept from anthropology that denotes the on-the-ground perspective—to determine what are confounders, mediators, sources of bias, etc., which are all the aspects that go into strengthening our evidence from simple association to causation. Using the emic view to build a causal theory is more justified than researchers using their own perspective to determine the sources of bias. As an anthropologist, I believe that context is everything. As much as we are biological beings, we are also cultural beings. We know that culture gets beneath the skin, and I rely on quantitative biological data (e.g., biomarkers) to determine the mechanism of how this happens. Measurement error is reduced when biomarkers are used; if we link upstream predictors to specific biomarkers, then biomarkers can be reconceptualized as biocultural markers. One example of how I employed a biocultural, mixed-methods approach is in the Adolescence among Bangladeshi and British Youth (ABBY) study, a comparison of puberty in a sample of migrant Bangladeshi, British-Bangladeshi and white British girls ( Houghton, Cooper, Bentley, et al., 2014 ; Houghton, Cooper, Booth, et al., 2014 ). In that study, we explored pubertal outcomes as measured by a structured survey and hormones measured in saliva and urine samples. I also spent 2.5 years in the field conducting participant observation and focus groups. One of the key findings is that first-generation migrants reached puberty earlier than girls living in Bangladesh ( Houghton, Cooper, Booth, et al., 2014 ). My field notes allowed us to speculate that acculturative stress plays a major role in puberty timing. Thus, through the integration of quantitative and qualitative data, I have been able to generate new biocultural hypotheses that capture girls’ lived and embodied human experience. The distinguishing factor in my mixed-methods approach is the integration of qualitative and quantitative methods in the analysis phase of research ( Zhang & Creswell, 2013 ) and not just being adept in both methods and using them separately according to the research question. The integration of qualitative and quantitative data is not easy and requires going outside one’s comfort zone to learn new analytical techniques, but the results are rewarding. I get especially excited when my qualitative and quantitative results contradict each other. It is amid that apparent chaos that I usually find my next research question.

One of the biggest strengths of participatory research methods is that they enable researchers to capture the voiced experiences of the research participants. This can provide a powerful narrative, which can help to bridge the needs of different data users and ultimately make research more accessible. Chris Bobel referenced the “zombie” statistics that continue to circulate despite the fact that they are not supported by rigorous evidence. I will admit that I used some of these figures in my previous work, before I understood how unsubstantiated or methodologically deficient they actually are. I gravitated to these statistics because they portray a simple, compelling message. Not only can the average person easily understand them, but they provoke a visceral reaction, which is key when you are trying to raise awareness or funding for a cause. In my experience, participatory methods can provide similarly powerful messages. For example, my team collects menarche stories from adolescent girls around the world in which they detail the story of their first menstrual period—where they were, what happened, who they told, how they felt, and advice for other young girls who have not yet reached menarche ( Sommer et al., 2015 ; Sommer et al., 2020 ). These stories provide rich data on menarche, the type and quality of puberty education they had received (or not received), and the challenges girls face managing their menstruation, as well as a powerful narrative and personal element to share. We know that governments and donors like statistics, but in my experience, the most influential reports combine quantitative and qualitative data. Quantitative data can tell you if a program achieves its intended impact, whereas qualitative data can contextualize the numbers. Qualitative data can help to explain both the “why” and the “how.” For example, it can tell you why a program is not working, if funding has been allocated appropriately, or how to improve program design to better address the needs of girls and women. It can also allow you to explore more sensitive and complex topics, such as women’s and girls’ experiences with menstruation or their sexual health. In particular, my team utilizes participatory methods, which engage the research participants as key partners in the process of enquiry and emphasize co-learning. When used throughout program design and implementation, participatory methods allow researchers and implementers to better understand the specific interests and needs of program beneficiaries, including those who are often overlooked, and to design programs that better fit their needs. Lastly, participatory methods can capture the insights—both experiences and recommendations—of the participants and enable a more equal exchange. Statistics have an important role to play, but we also need to see and understand the full picture, including the complexities behind the numbers; participatory methods can help to illuminate that. Lauren, as an epidemiologist with a background in anthropology, what do you see as the advantages of using mixed methods for data collection?

There are a few perils inherent to meshing together technology and menstruation. The first one is political. Recently, for example, The Guardian reported that Femm, a menstrual tracker, was funded by anti-abortion campaigners ( Glenza, 2019 ). However, the users are not made aware of the political inclination of the app makers and the resulting content featured in the app. There is no explicit contract between the users of an app and the people who created the app. This is valid for all apps, but it is interesting to think about how it now comes into play within the whole ecosystem of menstruation. So people who create these tools may do so not only to produce a consumer device but also to further a political agenda. The second peril is privacy. Menstrual trackers report very intimate details: menstrual flow, type of symptoms, sexual activity. Again, there is no explicit contract between users and these apps. For example, the Flo app, which has several million users, shared all the menstruation data with Facebook, and users did not know about it ( Schechner & Secada, 2019 ). At best, some of this exchange of data is mentioned in the terms of an app, but it is unrealistic to assume that users actually read or even comprehend the content of these lengthy and complex terms. As for the political peril, uncertainty about privacy is pervasive to all apps out there, but, again, it becomes particularly interesting in the context of menstruation. We need to ask what can be done if knowledge about these intimate facts fall into the wrong hands. The last potential risk I see is related to my field of research in particular, so I feel very strongly about it. With all the talk about artificial intelligence and data science, we are active, but we are not there yet. And, as I mentioned, there is no explicit contract between users and these tools. When an app notifies a user that “your period is about to start in 3 days” or “tomorrow is ovulation,” there is no explanation provided to the user about how the app came up with these predictions. How accurate are these predictions? What is our degree of confidence about each of these single predictions? Recently, Apple created an app for menstrual tracking, which predicts ovulation and menstruation ( Jackson, 2019 ). But there is no information provided to users about how these predictions are made, to what extent they are valid, and under which tracking conditions they were established as accurate.

We at Freedom Cups know our strengths. We are good at being operational. We are good at getting things done efficiently. We hike, we teach, we distribute, we survey. We are good at getting into magazines and in front of people in positions to influence policy. But none of us are academics. Social enterprises and non-profits working in the space need support from researchers in the form of partnerships. We need support and help with forming a framework for inclusive language. We need support with crafting an effective curriculum. We need support with randomized, controlled trials and impact assessments. Caitlin, can you explain from your perspective what researchers can do to improve such collaborations?