Informational constraints

Often, girls may receive insufficient, inaccurate, or incomplete educations about the physiological process of menstruation and the hygienic management it requires. Much of the early literature on this topic focuses on the high number of girls who start menstruating without being informed why menstruation occurs or how to manage it [7, 28, 32]. For example, a survey of 25 districts in five countries in southeastern Sub-Saharan Africa estimates that 66% of girls did not know about menstruation before menarche [5]. In Ethiopia, a previous study found that 49% of girls did not learn about menstrual management before menarche [14]. Similarly, in the Kilimanjaro district of Tanzania many girls transition through puberty without adequate guidance on puberty or menses management [12].

Menarche without prior information is commonly associated with negative emotional reactions, including shock, fear, and lack of confidence to seek help [4]. Other risks of insufficient information may include unintended pregnancy and subsequent marriage [6] and urogenital infection [33].

While some interventions have incorporated puberty and menstrual education [34–36], few have examined the effect of such education on attendance. Examples include a non-randomized trial among 120 schoolgirls in Ghana whcih found that education-only or education-plus-pads increased girls’ attendance by the same amount, roughly 9%. In addition, the Nia Project, an ongoing large-scale intervention of 140 primary schools in Kenya, intends to evaluate the effects of puberty education, sanitary pad provision, and a combination of the two on school attendance, among a wide variety of outcomes [16, 37].

Socio-cultural constraints

We consider socio-cultural constraints as norms or interpersonal interactions within communities that can prevent girls from attending school or negatively affect their educational experience. A growing body of evidence suggests that girls commonly report feelings of embarrassment and shame related to menstruation, often these emotions are borne from both internalized and externally imposed constraints [1, 4, 7, 38]. These findings confirm the significant impact of menstruation-related socio-cultural constraints on girls’ schooling and educational outcomes.

We identify three interrelated but discrete phenomena in the literature: teasing, restrictions on menstruating women, and sexualization. First, in-school period teasing may cause girls to feel shame or discomfort, which, as a result can deteriorate girls’ mental well-being or compel them to avoid attending school. We cover period teasing extensively in Benshaul-Tolonen et al. (2020) [39]. Second, a multitude of qualitative studies have documented the existence of explicit cultural or religious restrictions on menstruating women [1, 40, 41]. For instance, a systematic review from Sub-Saharan Africa finds consistent evidence that menstruation is commonly associated with potential contamination, leading to restrictions on women cooking, touching water or livestock, and participating in community/religious gatherings [42]. We further explore the prevalence of restrictions in the Tanzanian case study in Benshaul-Tolonen et al. (2020) [39]. Third, members of the community may commonly associate menarche with sexual maturity, marriageability, and childbearing; thus menarche may be a determinant of premature school dropout. Several case studies [13, 43] in Sub-Saharan Africa and a review of the LMIC [44] find that early menarche is associated with early sexual debut, early marriage, pregnancy, and increased risk of sexually transmitted infections. Qualitative work in Tanzania has documented instances in which post-menarche girls are sexually propositioned or offered money for sex. As a result, these instances may lead to anxieties about sexual harassment and pregnancy or, in some circumstances, dropout [4, 12]. The prevalence of these instances and fears, however, is unknown.

Compared to the other constraint types discussed throughout this paper, there is a dearth of quantitative evaluations focused on socio-cultural interventions. However, a growing number of authors recognize the benefit of investigating socio-cultural interventions. Preliminary studies have investigated the importance of including boys and men, family, and communities in education to combat stigmas and taboos [7, 32, 39].

Biological constraints

A developing branch of menstruation-related literature has brought to light the significant impact of biological constraints on young women’s educational outcomes. Biological constraints can range from dysmenorrhea (menstrual cramps) to more severe disorders like endometriosis. In particular, dysmennorhea is a common gynecological complaint among adolescents in developing countries [8, 29]. A systemic review of 50 studies [45], and various single case studies [15, 31, 46], have reinforced the finding that cramps are a common cause of school absenteeism and reduced participation. From this, we are able to conclude that menstrual pain, without proper diagnosis and management, can prevent girls from attending school and inhibit their ability to focus or participate in class.

Educational interventions and the provision of pain management medication can be used to address biological constraints. However, no study, to date, has evaluated pain management as an isolated treatment. A multi-component pilot intervention among 232 girls in Uganda trained teachers in pain management strategies (stretching, using a hot-water bottle) and provided analgesics (painkillers) alongside other interventions [23]. After 9 months of treatment, the study found evidence that the proportion of girls using painkillers and those using at least one pain relief method increased as a result of the intervention [23].

Material constraints

Menstruation-related material constraints often fall into two categories: i) access to adequate menstrual hygiene management absorbents and ii) access to satisfactory WASH facilities––infrastructure that includes water, soap, privacy, and space to change. Correspondingly, a study completed in the Kamuli district of Uganda found that 90.5% of secondary school girls did not meet the criteria for adequate menstrual hygiene management [47].

The use of unsanitary and sub-standard menstrual absorbents is commonly observed, but is particularly prevalent among girls of low socio-economic class or in rural areas [8]. Documented accessibility issues include high prices for hygienic materials, little or no availability in local areas, or interpersonal deterrents from buying, such as embarrassment buying from male store owners [48]. As a result, in the southeastern African region, including Tanzania, common absorbent materials include natural materials (grass, leaves, cow dung, tree bark), cotton, animal skins, rags, and mattress stuffing [5]. However, if girls are provided with MHM absorbents such as pads or menstrual cups and proper support, they are likely to utilize the methods [17, 49].

Access to WASH facilities is a critical determinant of girls’ menstrual hygiene management since the body and reusable absorbents must be washed several times a day and then dried to meet adequate MHM standards according to current guidelines [3]. A case study from Sierra Leone found that girls without access to adequate WASH facilities during the school day may be forced to go home early, improperly dispose of their absorbents, or use latrines outside of school grounds leading to disruptions in their education [50].

The primary concern is that girls experiencing material constraints are at higher risk of infection, leaking, and smells. Not only does this pose a threat to girls’ physical health, but it has been found to encourage teasing behaviors from classmates [39]. Additional research from Odisha, India, establishes a link between women’s experiences with sanitation and their mental health [51]. Another, distinct area of concern is that girls may engage in transactional relationships or sex in exchange for pads and other unaffordable MHM materials [6, 9, 52]. Combined, these factors emphasize the importance of addressing menstruation-related material constraints to improve girls’ well-being and educational outcomes.

Accordingly, material policy solutions can also be broken down into two categories: the provision of sanitary materials and the improvement of WASH facilities. There is cross-sectional evidence suggesting that girls who use pads are less likely to be absent during their period [14, 31], but these studies fail to account for probable omitted variables, namely household income, parental education, and knowledge about menstruation. In addition, there are few randomized program evaluations that address a causal link between pad use and attendance or participation in school. Within the Sub-Saharan African context, a pilot non-randomized intervention in Ghana [53] and a quasi-randomized intervention in Uganda [21] found that girls in sanitary pad treatment groups had improved attendance, but they were unable to establish causality. A larger cluster randomized study from Kenya followed 644 girls over 10 months and found no evidence, or weak evidence, of reductions in absenteeism due to a menstrual cup and sanitary pad program [18, 37]. In Nepal, girls who randomly received menstrual cups and sanitary pads were no less likely to miss school during their period [19]. In addition, the provision of disposable sanitary pads can have a relatively low return on investment due to high unit cost, compared to reusable sanitary products such as menstrual cups [54], therefore, it is essential to determine the intervention’s relative efficacy in improving educational outcomes.

On the other hand, the study of WASH facilities on menstruation-related absenteeism is less developed and remains largely observational and qualitative (e.g. [15]). A notable exception is a cluster-randomized trial of roughly 6,000 students in Kenya. Studies on the intervention found that improvements in latrine infrastructure and cleaning reduced girls’ absenteeism, whereas latrine cleaning and hand-washing campaigns alone did not significantly affect attendance [24, 55].

Sampling

The sampled students were enrolled at four co-educational secondary schools in the Sengerema District (Mwanza Region) and Geita District (Geita Region), Tanzania. The bulk of the schools were rural, but one school was peri-urban. Three of the schools were day-schools, while one had a girls-only hostel. While the selection of four schools is not representative of the regions or of Tanzania as a whole, the in-school sample is representative due to randomization. The study used a random sample of all students in the target secondary school grades (1–3). Systematic sampling, within the schools, was done using the class registrar. Every third student on the registrar was selected to participate, provided they were present at the school. In total, 432 male and 524 female students participated in the study. Sample size was limited to 524 female students and four schools because including more schools and individuals was cost-prohibitive at the time of program development. Results from the boys’ surveys, and selected results from the girls’ surveys are published in Benshaul-Tolonen et al 2020 [39].

We undertook data collection in May of 2019 with one male Tanzanian lecturer (co-PI), two female US-based scholars, and two local research assistants (one male, one female). All research staff were present in the classrooms during surveying, providing a mixed-gender research group. Classroom teachers helped hand out and collect the surveys as well as keep order in the classrooms.

To avoid congestion of students in classrooms, special classrooms were assigned for the study––up to six rooms per school. Although boys were included, they were separated from the girls during surveying. We determined beforehand that a pen and paper survey would allow students the maximum amount of privacy and increase the likelihood of honest answers as compared to an enumerator-led survey. All female students received a fabric gift bag with a notepad, a pen, and a menstrual health pamphlet from the NGO AFRIpads upon completion of the survey and, at a later date, a set of three reusable sanitary pads.

The survey contained a knowledge assessment section, questions about menstruation-related experiences, attitudes, information sources, and teasing behaviors. The original Swahili questionnaires with English translations are included in the supplementary information section. The questionnaires were not validated on a student population prior to use since the study was considered a pilot study. In addition, field researchers made qualitative observations on each school’s latrine infrastructure.

Measurement of outcomes

A review of nationally representative data from 12 African countries, Loaiza and Lloyd (2008) found similar rates of absence between boys and girls [56], further corroborated by research in Kenya [37]. This may imply that girls partly miss school due to menstruation, but boys miss school at the same rate for separate reasons. In this study, we will measure absenteeism as days missed during a girl’s last period, rather than overall rates.

Four distinct educational outcomes are considered in this study: absenteeism, leaving school early, participation in the classroom, and concentration in the classroom. The survey asked students to recall their behavior during their last menstrual period. Specifically, each outcome behavior was linked directly to menstruation; e.g. “During your last period, were there days you could not attend school because of your period?”. In this study, we decided to separate absenteeism into two distinct outcomes: missing school entirely and leaving early. This decision was made as we believe that the two variables may have different effects, but also different origins. For instance, while absenteeism may be more likely due to feeling unprepared, leaving early may be more closely linked to being unable to manage your period during the school day which may result from unexpected timing of one’s period or inadequate materials and WASH facilities. However, it is possible that responses suffer from both recall bias and social desirability bias. Nevertheless, we highlight three strengths to our approach.

First, the paper’s primary strength is its unique quantitative data on reasons for absenteeism and other behaviors. Girls were asked to report why they had engaged in each of the four behaviors, choosing multiple answers from an extensive list. Second, we glean insight into the relative prevalence of different educational barriers. Third, by asking girls to report absenteeism or early departure caused directly by their period, we reduce risk of capturing absenteeism that is merely coincident with, rather than caused by, their periods.

Ethical considerations

Research permits for the study were received at the Bugando Medical Centre and Catholic University of Health and Allied Science Joint Ethics and Review Committee, and National Institute for Medical Research (NIMR) (Ref: MR/53/100/596) in Tanzania. IRB was approved at Barnard College, Columbia University (Ref: 1819–1110-010) in the USA. Permission to conduct the research were obtained from the Tanzania Commission for Research and Technology (COSTECH) and at Geita District Council (Ref: GDC/E.10/1/VOL.2/267).

The survey was conducted in four selected schools in the Senegrema and Geita Districts. Permission was obtained from district secondary school education officers and medical officers of health in order to collect data in the secondary schools. Likewise, written consent was obtained from each school’s headmaster or headmistress for their school’s participation in the study.

In addition, the students signed written assent forms in Kiswahili (the primary language of most of the population of Tanzania) prior to data collection. The forms explained the purpose, risks, and significance of the study, as well as participants’ right to participate or withdraw. Students’ parents were informed about research activities through their school management committees; however, written consent was not required from the parents according to the local ethics reviews.

Statistical analysis

All statistical analysis was performed in Stata v.15.1. To obtain the regression coefficients in Fig 4, we use a linear probability model. The model reports correlations between social, biological, and material constraints with the four educational outcomes of interest (absenteeism, leaving early, participating, and concentrating less). This identification controls for but does not report coefficients on respondent’s age, grade, and age at menarche, as well as school fixed effects.

To develop the three constraint indices, the 2–3 most applicable questions to each constraint were selected, scored, and then normalized to fall between zero and one. For each constraint type, individual students’ points are added together and then divided by the maximum amount of points to find the normalized score. The constraint indices were developed post-hoc, therefore, we are limited in our interpretation as they were not developed a-priori. Further information on the questions included in each index and the outcome variables can be found in Tables 1 and 2. Scoring information can be found in the supplementary information section along with score distributions for each constraint type.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics and menstruation-related constraints on educational outcomes.

https://doi.org/10.1371/journal.pgph.0000110.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographics and menstruation-related constraint indices.

https://doi.org/10.1371/journal.pgph.0000110.t002

For each educational outcome (dependent variables) the regression controls for all four constraint types (independent variables). Therefore, we are able to measure the relative impact of each constraint on girls’ educational outcomes. (1)

In the equation above, Y represents each of the four educational outcomes (missing school, leaving early, concentration, and participation). Every constraint index is represented in the equation which allows us to take into account each constraint type’s relative impact on the outcomes of interest. The equation includes school fixed effects, s, and controls for a vector of individual level characteristics, α_i (age, grade, and age at menarche).

In summary, the socio-cultural constraint index captures the prevalence of girls’ common menstruation-related fears, the biological constraint index captures experiences with physical pain and bleeding due to menstruation, and the material constraint index captures access to and experiences with proper MHM.

Although informational constraints are discussed throughout the paper, they are not included in the final model due to a lack of reasonable variation. We find that girls in the sample are generally well-informed about menstruation and that it is rare for them to experience any significant informational constraints. Therefore, the index is omitted to prevent the introduction of bias into the model and the other results. Moreover, the number of questions included in each of the remaining indices was limited by sample size considerations, including too many questions with differing rates of attrition caused sample sizes to fall drastically. Therefore, we only include the most relevant questions for each constraint type in the index. Distributions for each constraint index, including informational, can be found in S1 Fig. For transparency, the original regression that includes the first iteration of the material constraint index, the informational constraints index and the knowledge score are available in the supplementary information section, as well as S2 and S3 Figs.

Informational constraints

Contrary to much prior qualitative research in developing countries [7, 28, 32, 58], we find that overall, girls in the sample received detailed, high-quality information from multiple sources about the biological basis and management of menstruation. A clear majority of girls reported that someone had told them about periods and why they happen (90%) and that someone had taught them how to manage menstruation (95%) (Table 2, Panel A). Further, girls received information from an average of 4.5 sources, ranging from a list of relatives and friends to school, health, and religious officials (Table 2). We verify that the information was high quality, as 80% received information from health workers, and 76% from school curricula. This mitigates concerns that information from family and friends can propagate misconceptions or stigmas [59] (Table 2, Panel A). We find that girls also received practical information as more than 80% reported they’d received advice on how to use menstrual absorbents and how to wash themselves and stay clean (Table 2, Panel A). Thus, the cohort had a high level of informational preparedness.

While the overall results on informational constraints are optimistic, we find evidence that some girls slip through the cracks, and that proscriptions (restrictions on menstruating women) remain. Ten percent of girls in the sample (10%) had not been told about the causes of periods, yet further investigation revealed that 86% of the group were post-menarche. Therefore, it is possible that the girls who had yet to be informed about menstruation were not also awaiting their first period. Girls in this “uninformed” group were less likely to report receiving information on menstrual management (86%) or information from health workers (63%) and school curricula (54%). As a result, there exists a cohort of girls who have received below-average levels of information across many different measures. This indicates that there is still room for improvement in educating young women about menstruation, even in settings that have a fairly high level of informational preparedness.

Second, although girls received valuable advice on menstrual management, 77% also reported that they had been advised to avoid certain objects or activities during menstruation (Table 2, Panel A). In summary, while girls generally receive detailed practical information regarding absorbents and washing, the information is commonly accompanied by ideas that, whether harmfully or not, imply that menstruating girls should deviate from normal behavior. This practice can perpetuate stigmas and lead to socio-cultural anxieties related to menstruation. Consequently, it is essential that the impact of menstruation-related socio-cultural constraints on girls’ well-being and education is explored further.

Socio-cultural constraints

Socio-cultural constraints related to menstruation inhibit girls’ educational outcomes. Apart from cramps and pain, shame and fear are the two most prevalent reasons for reduced participation and concentration while in school. Roughly 134 girls in the sample cite fear and 86 girls cite shame as the reason for their change in participation or concentration (Fig 2). The question naturally follows: If shame and fear play such a large part in girls’ desires to hide their period in school, then what is the source of this shame?

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Reasons for absenteeism and reduced participation.

The sub-figures indicate the prevalence of each reason, as selected by girls in the sample, for absenteeism or reduced educational outcomes. Please note, the reasons listed here, although coded by constraint type, are not the questions used to build the corresponding constraint index. A) Reason for missing school during last period; B) Reason for leaving early during last period; C) Reason for not participating as normal during last period; D) Reason for not concentrating as normal during last period. Please note the underlying data in this figure was previously published in Benshaul-Tolonen et al. (2020) [39].

https://doi.org/10.1371/journal.pgph.0000110.g002

Through the survey, we investigate different types of teasing and harassment that have previously been linked to menstruation in qualitative research, but have rarely been quantified. We identify and study three overlapping types of negative social consequences for revealing one’s period: sexualization, a gendered sense of impropriety, and stigmas outside of peer interactions.

First, the vast majority of girls fear harassment borne from attitudes that associate menarche with physical and/or sexual maturity and marriageability. We asked girls about their fears regarding specific forms of harassment, if people knew they were post-menarche. Seventy percent (70%) of girls in the sample fear being touched against their will or harassed to go out with someone, while 87% fear unwanted pregnancies or dishonor and 64% fear the pressure to marry or take a boyfriend (Table 2, Panel B). These questions are used to develop our socio-cultural constraint index as they capture different forms of fear and shame that girls may associate with menstruation.

Second, the sample demonstrates a gendered understanding of appropriateness in revealing one’s period, both at school and in the home. While a substantial portion of girls (37%) report that they would feel shame if girls at school knew about their periods, a much larger portion (84%) would feel shame if male peers knew (Table 2, Panel B). We conjecture that menstruation-related shame is partly derived from this gendered sense of appropriate confidants for female-specific issues, and partly due to the fear of possible public shaming. Furthermore, 58% of girls fear being teased if people knew they were post-menarche. This supports our claim that fear of public shaming and perceived appropriateness are major factors in menstruation-related shame. In addition, we have previously reported that male peers are the most commonly feared perpetrators of period teasing among this sample of girls [39].

Third, girls are subject to socio-cultural constraints related to menstruation in their interactions with teachers and family members. For instance, twenty-six (26%) of girls fear teachers being unhelpful or not understanding if they knew a student was post-menarche. Likewise, 38% of girls report that menstruating women are restricted from certain daily activities in their homes (Table 2).

In conclusion, girls in the sample express both shame of revealing menstrual blood or their period status, and fear of harassment related to the sexualization of post-menarche girls. Given evidence that period-related teasing is common in the sampled schools, we conjecture that girls’ fear and shame are consequences of witnessing post-menarche girls suffer negative social interactions. Family, teachers, and friends may also instill fear by transmitting social norms, attitudes, and beliefs about real or imagined consequences of revealing menstruation, though our survey was not designed to trace the origin of these ideas. Accordingly, these aspects of shame and fear then predominantly contribute to girls’ absenteeism, decreased participation, and decreased concentration in school during menstruation. Thus, fear and shame are plausibly consequences of socio-cultural constraints on menstruating women, but in turn they also cause girls to constrain their own social behavior.

Biological constraints

In general, girls report that they have irregular periods which they find difficult to manage, signaling both biological and material constraints. In terms of biological constraints, 73% of girls reported that they keep track of their menstrual cycle, however, 62% of girls report that the time between their periods varies heavily (Table 2, Panel C). This indicates that even though girls may be familiar with cycle tracking techniques, the irregularity of menstruation can pose a threat to their ability to prepare for their period. Accordingly, 60% of girls, in the sample, reported that it is hard to anticipate and/or prepare for their next period (Table 2, Panel C). The sample reported an average degree of bleeding during menstruation of 2.6 (1 being no bleeding, 5 being very strong bleeding) (Table 2, Panel C and Fig 1). Additionally, for some, cramps (dysmenorrhea) and pain are prominent physical symptoms during menstruation.

When asked to judge their level of physical pain during menstruation (1 being no pain, 5 being very strong pain), the girls reported an average pain level of 3.1 (Table 2, Panel C and Fig 1). Correspondingly, “cramps and pain” is cited as the most or second-most common reason for all four educational outcomes (Fig 2). For instance, of those girls that missed a day of school during their last period, over 100 reported that they missed school because of “cramps and pain”—by far the most prevalent reason. Similar results are seen across educational outcomes––cramps and pain are the most or second most prevalent reason for leaving early, participating less, and concentrating less (Fig 2). Therefore, biological constraints related to menstruation can act as a serious impediment to girls’ educational outcomes. Furthermore, we find that 49% of girls in the sample always or sometimes use painkillers during menstruation (Table 2, Panel C). This finding emphasizes how debilitating symptoms of premenstrual syndrome can be to young women and that some may be reliant on pain medication to manage their period symptoms.

As a result, the questions regarding degree of physical pain and bleeding were used to develop the biological constraints index. The index aims to capture the degree at which each girl is affected by these biological constraints to determine their relative impact on each educational outcome.

Material constraints

Although prior research suggests that material constraints may largely affect young women’s ability to obtain their preferred MHM absorbents, the majority (73%) of girls in our sample report that they always use their most preferred method. As noted in Table 2, Panel D, 47% of girls most often use cloths as menstrual absorbents, though only 23% prefer cloth. Likewise, the majority of girls (54%) would prefer to use natural materials (which may include raw cotton but also grass, leaves, and cow dung), but only 43% use natural materials most often. In the survey, girls were not asked to specify what they meant as “natural material”, therefore, we are unable to provide a breakdown of this category. However, the largest disparity between use and preference is in relation to pads: 21% of girls would prefer to use pads, but only 7% use pads most often. We theorize that pad affordability is the primary cause of this deficit. Of the girls who did not use their most preferred method during their last period, 60% cited “not enough money” as one reason. Further, it is notable that the average girl spends more than two-thirds of her total monthly expenditure on pads (Table 2, Panel D). Fig 3 depicts average daily expenditure on sanitary pads during menstruation by girls’ monthly expenditure and compares spending to the national poverty line, international poverty line, and GDP per capita. The results show that average daily expenditure on sanitary pads is higher as girls’ monthly expenditure increases. Daily expenditure on pads was calculated using girls’ stated monthly expenditure on pads and the average menstrual cycle length in the sample (3.8 days). Through these calculations, we find that girls in the sample are spending anywhere from 12–70% of the national poverty line (6,247 TSH per day) on pads during their period (Fig 3). The second most prominent material constraint, with respect to using one’s preferred absorbent, is lack of availability. Over 25% of girls who did not use their most preferred method last month cited “method unavailable” as a reason (Table 2, Panel D). As a result, we developed the material constraint index using the question regarding using one’s preferred method and a separate question on one’s ability to wash their body at home.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Monthly expenditure and spending on sanitary pads.

This figure presents girls’ average daily expenditure on sanitary pads during menstruation (self-reported monthly expenditure on sanitary pads expressed in 1000 TSH, divided by the average length of menstrual period) by their monthly self-reported expenditure (ranked and expressed in 1000s TSH). In addition, the green bar shows the approximate daily value of Tanzania’s 2018 GDP per capita [60] (approximately equivalent to 2.9 USD per capita), the blue line presents Tanzania’s national poverty line [61], the red line indicates the international poverty line [62].

https://doi.org/10.1371/journal.pgph.0000110.g003

Aside from cramps and pain, material constraints such as accessing absorbents and school latrine infrastructure are the main reason why girls miss school or leave school early according to their survey responses (Fig 2). In the survey, we asked girls to choose every applicable reason for why they have missed school, left early, or experienced changes in participation and concentration. Prevalent reasons for missing school included “nothing to use” (75 girls) and “no place to change” (47 girls). Similarly, aside from cramps and pain, the most common reasons for leaving school early are “needed to change” (133 girls) and “needed to wash” (28 girls) (Fig 2). These results, indicate that a variety of material constraints prevent girls from attending school including insufficient access to absorbents or reliable WASH facilities. Correspondingly, the responses confirm a lack of sufficient infrastructure at school for changing and washing.

Furthermore, girls’ ability to afford or acquire pain medication may interact with their ability to manage menstrual pain which is commonly cited as a reason for menstruation-related absenteeism. Other biological constraints such as fatigue or sickness were less commonly reported. Put simply, access to pain medication, a material constraint, may impact girls’ educational outcomes. However, capturing the interaction between material constraints and biological constraints is beyond the scope of this paper and the relationship should be explored in future research.

Correlates of absenteeism and other impaired educational outcomes

In Fig 4, we investigate whether material, social, and biological constraints are significantly correlated with absenteeism, leaving early, decreased participation, and decreased concentration. The three categories of correlates tested—material, social, and biological—were chosen based on the four theories of constraints previously discussed. Ultimately, we form three conclusions of interest. First, under the logic of material constraints, girls who are unable to access or afford their preferred absorbent and/or are unable to wash their body at home suffer from inadequate MHM which may negatively impact educational outcomes. Second, according to the logic of socio-cultural constraints, girls who experience more menstruation-related shame or fear may be less comfortable in social settings during menstruation, and thus have more significant educational impediments. Finally, in the logic of biological constraints, girls who experience more physical pain or a greater degree of bleeding during menstruation may be worse equipped to manage or conceal their physical symptoms resulting in considerable obstacles to education.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Regression results: Effect of constraints on educational outcomes.

The figure above depicts plotted regression coefficients for each constraint type and its relationship with each educational outcome of interest. The model includes three constraint types: social, biological, and material and four educational outcomes: A) “During your last period, were there days when you could not come to school because of your period?” B) “During your last period, did you ever leave school early because of your period?” C) “During your last period, did you participate as much in class as you normally do (when you do not have your period)?” and D) “During your last period, did you concentrate as much in class as you normally do (when you do not have your period)?”. In addition, the figure includes p-value labels for each coefficient to indicate significance.

https://doi.org/10.1371/journal.pgph.0000110.g004

Comprehensive descriptions of each index and scoring guidelines can be found in the previous section, in Table 2, and in the supplementary information section.

We find substantial support for the socio-cultural constraints theory in relation to absenteeism and leaving school early. Girls who experience more socio-cultural constraints are significantly more likely to stay home or leave school early during their period and because of their period, (p < 0.10) and (p < 0.01) respectively. For example, in the fully identified model, girls who reported a socio-cultural constraint index score of 1 were 18.2 percentage points (56% of the sample mean) more likely to miss school (Fig 4) and 34.6 percentage points (74% of the sample mean) more likely to leave school early (Fig 4). However, the regression coefficient between the social constraint index and participating or concentrating in class is not statistically significant. Therefore, we conclude that shame, fear and the socio-cultural constraints that drive them, contribute to absenteeism more than the other educational outcomes.

Likewise, there is a significant association between biological constraints and our educational outcomes of interest. Girls who are more biologically constrained, those who experience more pain or bleeding during menstruation, are significantly more likely to stay home, leave school early, and concentrate less in class, (p < 0.01), (p < 0.01), and (p < 0.05) respectively. In the fully identified model, girls who report a biological constraint index score of 1 were 34.3% more likely to miss school (105.2% of the sample mean), 33% more likely to leave early (70.7% of the sample mean), and 14.7% more likely to concentrate less in class (44.5% of the sample mean) (Fig 4). These results indicate that biological constraints have a considerable impact on girls’ educational outcomes, especially on absenteeism.

Finally, the results demonstrate that material constraints play a similar key role in determining educational outcomes. In the sample, girls who experience more material constraints are significantly less likely to miss school and are more likely to concentrate less in class, (p < 0.10) and (p < 0.01) respectively. In the fully identified model, girls who report a material constraint index score of 1 were 14.9% less likely to miss school (45.7% of the sample mean) and 24.8% more likely to concentrate less in class (75.2% of the sample mean). These results indicate that material constraints may have a greater impact on educational outcomes like participating and concentrating in class than previously anticipated. Furthermore, it emphasizes the importance of access to MHM absorbents and WASH facilities.

Main results

Through this exercise, we found abundant evidence that menstrual hygiene management impacts educational outcomes. One third of girls stated that they missed at least one day of school during their last period, while nearly half of the girls voiced that they left school early because of their period. Moreover, we report significant rates of decreased participation and concentration in the classroom as menstruation-related educational impediments. Although absenteeism may represent more serious educational losses, our results indicate that limiting research to these outcomes can overlook significant barriers to educational access among adolescent girls. We investigate the causes of these educational impediments by breaking up potential constraints into four categories: informational, socio-cultural, biological, and material.

Three principal conclusions are derived from this study: i) interventions focused on the provision of menstrual absorbents should consider including participation and concentration as primary educational outcomes, ii) material constraint interventions should expand their focus to adequate WASH infrastructure, as it significantly impacts early departure, and iii) to date, socio-cultural constraints (in particular, fear and shame) and biological constraints (in particular, dysmenhorrea) are seriously understudied in quantitative research although they emerge as primary drivers of girls’ absenteeism from school.

These results present a relevant addition to the insights generated by previous correlational findings in the literature. For instance, the study conducted in Uttar Pradesh by Malhotra et al. (2016) [26] finds that two-thirds of the girls reported constraints in the management of menstruation. This figure represents a benchmark for this study, which revealed that 77% of girls had been restricted from doing certain activities during menstruation. Nevertheless, our study differs in many aspects, including regional focus and a different approach to household factors. While we focus on direct factors of MHM access, such as the ability to afford the products and washing infrastructure at home, their study looks at indirect correlations of MHM, including household economic and demographic characteristics. Overall, more research that encompasses all aspects of material constraints is needed to determine their relative importance.

Study weaknesses

The methodology of this study has several weaknesses. First, the results presented here cannot be interpreted as representative of Tanzania or the Geita and Sengerema Districts, because the four public secondary schools were selected for convenience. For the same reason, we were unable to intentionally target multiple types of schools, e.g. private, public, boarding or day schools. However, three schools were day schools and the fourth was a boarding school. All schools in the sample were mixed-gender.

Second, cost prohibited including a sample larger than roughly 500 girls, and the sample size was not calculated a priori to achieve specific statistical power. The exercise was intended as a pilot study to prepare for a larger randomized control trial, and is thereby limited in scope.

Third, our sample is restricted to girls currently enrolled in and attending school. As many qualitative studies have highlighted (e.g. [12, 14, 48, 78]), girls who have dropped out of school have different experiences with menstruation and are less likely to learn about it from school curricula. While school participation and absenteeism are our main outcomes of interest, we do not capture the more extreme consequence of school drop-out. Moreover, our study finds that age of menarche occurred slightly earlier in our sample relative to previous research. Therefore, we urge future studies or interventions to include younger girls to accurately capture the impact of access to menstrual hygiene on school dropout.

Fourth, we did not include the informational constraint index in the final linear analysis model. The constraint was omitted due to a lack of real variation in the sample. If it were included, the index would have introduced bias to the model and the remaining regression results. As a result, we are unable to consider how informational constraints impact girls’ educational outcomes in conjunction with the other constraint indices. However, alternative measures of information access, which we did not collect, could possibly yield meaningful variation in the data. Therefore, we encourage future research to consider applying this methodology to samples that naturally experience more informational constraints and expand on this finding. Nevertheless, we ran the analysis using two separate indexes for knowledge (previously defined in [39]) and information and present the results in S2 Fig. We find that the score on the knowledge index does not have a statistically significant correlation with our four educational outcomes.

Fifth, the regression results are limited in their interpretation due to the small sample size and inability to include extensive control variables due to survey formulation. In addition, the number of observations is affected because many of the girls in the sample have missing values for at least one of the question used in the indices. Therefore, it is possible that our regression outcomes are not fully representative of the study population. It may be that girls who perceive a particular question as socially sensitive are less likely to respond, thereby skewing the results. We cannot test for this but encourage future studies to bear this in mind when designing the questionnaire.

Sixth, as described in the background section, the domains of constraints on MHM –informational, socio-cultural, biological and material– were constructed following a review of the past literature. They map onto multiple existing frameworks for understanding menstrual health, are broadly comparable to UNICEF’s pillars of menstrual health interventions (2019) [11], and reflect many of the themes in the integrated model of menstrual experience [22] and the domains proposed by Plesons et al. (2021) [10]. However, the questions encompassed in each constraint were not chosen a priori to the data collection, but were included post-hoc. The post-hoc development of the indices, in some instances, constrains our ability to interpret the results, as the questions were not developed for this purpose. Post-hoc selection of questions can also allow researcher cherry picking of statistically significant results [79], an issue that we are aware but tried to avoid. For transparency, (1) we only selected questions that are clearly linked to each constraint type; (2) Table 2 shows the questions that were not selected due to the researchers’ priors; and (3) the exclusion of questions from the indices was based on sample size limitations.

Lastly, our main conclusion from this exploratory work is that menstruation-related absenteeism and early departure are caused primarily by socio-cultural constraints, while decreased participation/concentration are caused by material constraints. However, biological constraints, such as cramps and pain, are significant impediments to both absenteeism and participation/concentration outcomes. These conclusions are strongly borne out in our data and conceptually plausible. Although, it should be noted that as some reasons were not intuitively applicable to all outcomes, girls were not given identical lists of potential reasons for each outcome (See girls’ questionnaire in supplementary information). Further research testing this hypothesis would benefit from the forethought to create identical yet logical lists of reasons for worsening educational outcomes. Moreover, future studies should define a wider set of questions for the indices a priori, to capture more nuance in the constraints faced by girls.

Further research

There are several implications for policy. First, as other scholars have advocated, pain management should be considered as a dimension of menstrual management interventions. We call for program analysis on pain management as an intervention tool. We suggest the provision of analgesics, non-medicinal pain alleviating techniques, contraceptives and the integration of pain management into school and health curricula.

Second, research on menstrual-tracking should be completed to determine if it is a viable intervention for adolescent girls in low and middle-income countries. A clear majority of girls in this study experience variation in their menstrual cycle and, as a result, are unable to adequately prepare for their period. It is possible that improvements in menstrual-tracking technology could help ensure that girls are prepared for menstruation and ultimately improve their educational outcomes.

Third, interventions should specify which dimension of girls’ education they hope to improve (enrollment, attendance, or participation). However, comprehensive interventions may require both material and social components. Finally, the social dimension of menstruation and shame is complex. Girls express they would feel shame and fear sexualization or harassment as consequences of revealing they are post-menarche, but these reactions likely vary depending on the social sphere in question, i.e. the community, the family, with male peers, and with female peers.

Overall, we propose the development of interventions that simultaneously tackle all menstruation-related constraints: material, biological, informational, and social to create equitable educational environments and opportunities for young women.