Do Menstrual Hygiene Management Interventions Improve Education and Psychosocial Outcomes for Women and Girls in Low and Middle Income Countries? A Systematic Review

1The effective, hygienic management of menstruation is essential for women and girls to participate in society with dignity and comfort. Effective menstrual hygiene management (MHM) includes access to clean absorbents, with facilities to change, clean or dispose of these as needed, and access to soap and water for cleaning the body and absorbents. Studies across low and middle income countries (LMICs) have reported that more than 50% of girls have inadequate MHM, with higher proportions reported in rural areas.

2Health and social research has only recently sought to address the neglect of MHM as a significant development issue and barrier to achieving gender equality. Both qualitative and quantitative work has suggested that poor MHM results in school absenteeism, distraction, and disengagement. There is a shortage of literature addressing the impact of MHM on adult women, but poor MHM is also likely a barrier to occupational attendance and engagement. Focus groups from a factory in Pakistan suggested that women missed up to three days’ work per month due to menstruation.

3Other programs of qualitative work have reported psychosocial consequences of poor MHM (e.g.,). Feelings of shame, fear (and fear of stigma), anxiety, and distraction have been described across a range of contexts. Few studies have quantified the relationship between MHM and psychosocial consequences. Studies that have been conducted have often focused on menarche, reporting that high proportions of girls feel unprepared and afraid at this time.

4Interventions to address MHM have been categorised into two groups; hardware interventions designed to address material deprivations such as the provision of absorbents, or improved Water, Sanitation and Hygiene (WASH) facilities; and software interventions which address deficits in knowledge of menstruation and management by providing education.

5The cost and availability of sanitary products, and underwear in which to wear them, is a fundamental barrier to MHM. Commercial absorbents are frequently unavailable or too expensive. The provision of clean sanitary products (e.g., commercial or home-made pads) addresses this material deprivation and is hypothesised to reduce discomfort, and concerns regarding soiling outer garments. Improved management and comfort may also reduce associated stigma, ridicule, and embarrassment which deters women and girls from attending school or work. In Ghana, Dolan and colleagues found that over three quarters of schoolgirls surveyed reported soiling outer garments during their last menses, and found that school attendance improved by 9% after 5 months with the prevision of disposable sanitary pads.

6Inadequate WASH represents a barrier to MHM, particularly the ability to clean absorbents, and the body, in private. The design of toilets often fails to meet women’s physical and psychological needs. Latrines lacking doors or locks may threaten safety and cause embarrassment. Hygiene guidelines recommend changing absorbents every two to six hours dependent on blood flow, thus facilities are needed both at home and in schools or workplaces. Improvements to WASH access may enable women to clean reusable absorbents and genitals hygienically, as well as reduce discomfort and embarrassment. Open pit toilets or toilets without disposal facilities mean blood or used sanitary products reveal when a woman is menstruating, resulting in embarrassment and stigma. ‘Girl friendly’ changes such as gender-separate latrines, locks on toilets, discrete facilities for changing absorbents or washing have also been hypothesised to enable improved MHM.

7Adolescent girls have been found across a range of countries to lack knowledge of the physiology and management of menstrual bleeding. Ali and Rizvi reported that in Pakistan, less than 50% of girls surveyed received information about menstruation prior to menarche, only 15% information about management. Cultural beliefs and taboos may also contribute to poor MHM through the perpetuation of misinformation or unhygienic customs. Studies across contexts have reported taboos around the disposal of menstrual blood, and practices including restrictions to bathing and participation in social activities.

8Improving understanding of menstruation, such as knowledge of cycle length and requirements for hygiene, is hypothesised to improve management practices, self-efficacy, and reduce anxiety. In this way education interventions may facilitate school attendance and engagement in the classroom through improved ability to effectively manage menstruation, and confidence in management methods. This education may also reduce negative psychosocial consequences by normalising menses and dispelling myths.

9Sumpter and Torondel conducted the only systematic review of MHM interventions, focusing on health outcomes. The review included some education and social outcomes, but had a number of limitations. First, pre-post comparisons and observational studies were synthesised alongside randomised controlled trials despite representing different risks of bias. Second, the review searched only peer-reviewed journals thus missed the wider ‘grey’ literature. Third, only limited terms ‘school’ and ‘educat*’ were used to search for education outcomes. Only ‘social restrictions’ such as restricting diet or interactions during menstruation, rather than psychosocial consequences described in the literature, were included. The review did not discuss the inclusion or exclusion of WASH interventions. Finally, the review search was conducted in May 2012, with at least two trials published since this time.

10Two other relevant reviews have been conducted. Birdthistle and colleagues and Jasper and colleagues focused on WASH interventions. Neither identified all relevant studies, nor appraised the risk of bias in included studies.

11Hardware and software interventions have been employed by governments, international organisations and local charities seeking to address poor MHM and associated consequences. These programs have been enacted in the absence of evidence for their effectiveness. Given the weak evidence base and ongoing plans to implement MHM interventions, the present work will provide an essential appraisal of eligible studies and highlight implications for future research.

12This review will summarise and critically appraise evidence for the effectiveness of hardware and software interventions for MHM aimed at improving education, employment or psychosocial outcomes for females in LMICs.

13This review was conducted according to the requirements of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Review protocol was registered on PROSPERO (CRD42014010631) http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42014010631.

14Individually and cluster randomised controlled trials (RCTs/cRCTs) were eligible for inclusion, representing the most rigorous evidence with the ability to infer causality. Non-randomised controlled trials and controlled before-after studies were also included despite greater potential for bias, due to the lack of available trials and difficulty conducting RCTs in LMICs. Cross-sectional studies and pre-post designs which did not include a control group were ineligible.

15Studies were eligible if they reported on outcomes for menstruating females from LMICs as defined by the World Bank (http://data.worldbank.org/about/country-classifications).

16The provision of clean absorbents/sanitary products (disposable or reusable),

17Improved WASH or girl-friendly facilities. Key aspects of WASH interventions relevant to menstruation were derived from the WHO report on Water, Sanitation and Hygiene standards for schools in low-cost settings, the UN Gender, Water and Sanitation Policy Brief. Studies could include any number of these aspects where authors’ hypothesised that the intervention was sufficient to improve MHM practices: Improved clean water supply for menstruation management (e.g., access to water within latrine or private areas);Provision of soap or disinfectant for body and absorbent cleaning;Improved absorbent disposal facilities;Improvements to latrine privacy or safety.

18Improved clean water supply for menstruation management (e.g., access to water within latrine or private areas);

19Provision of soap or disinfectant for body and absorbent cleaning;

20Improved absorbent disposal facilities;

21Improvements to latrine privacy or safety.

22Interventions that delivered sufficient education to provide an understanding of the biological process of menstruation (e.g., the cyclic nature of menses, origin of menstrual blood), and which authors hypothesised was sufficient to improve either MHM practices or misconceptions and confusion. Information could be provided in person or via printed or electronic resources.

23Education. School attendance (full or partial days) from school records or self-reported.

24Employment. Self- or employer-reported attendance/absenteeism (full or partial days).

25Psychosocial Outcomes. Outcomes were derived from qualitative literature identifying psychosocial consequences of poor MHM and included; anxiety, confusion, depression, embarrassment, emotional distress, fear, powerlessness, self-confidence, self-efficacy, shame, stigma, worry, and self-imposed withdrawal from activities such as playing sport. Externally enforced restrictions (e.g., not attending religious ceremonies) were not eligible outcomes as interventions were not hypothesised to impact such traditions.

26Education. Academic achievement and school engagement.

27Employment. Measures of productivity, engagement, and worker satisfaction.

28Psychosocial outcomes. Other measures of attitudes related to menstruation.

29Menstrual knowledge and management. Knowledge of menstruation and improved management practices.

30Searching strategy was based on inclusion criteria and developed with reference to searches reported in previous reviews. Searches were conducted in English, but no restrictions were set for language, date, or publication type. Searching strategy for PsycINFO is displayed in Table 1.

31The following databases were searched from inception to present in January 2015:

32ASSIA (Applied Social Science Index and Abstracts)

33CENTRAL (Cochrane Central Register of Controlled Trials)

34CINAHL

35EMBASE

36Global Health

37IBSS (International Bibliography of the Social Sciences)

38MEDLINE

39OpenGrey

40POPLINE

41PsycINFO

42Science Citation Index

43Social Sciences Citation Index (SSCI)

44TRoPHI (Trials Register of Promoting Health Interventions)

45WHOLIS

46World Bank e-library

47Government departments and multilateral organisations were identified through the Menstrual Hygiene Matters report co-publishing agencies, and websites and databases were searched. Handsearching was conducted in two key journals, reference lists of included studies, and the Menstrual Hygiene Matters report. Subject experts were contacted to identify published, unpublished, or ongoing research.

48Two reviewers independently screened titles, abstracts and full text articles. Included articles were determined by consensus. Data were extracted regarding participants, methods, intervention, analyses, outcomes, and risk of bias using a piloted form.

49Risk of bias was assessed independently by both authors using the tool outlined in the Cochrane Handbook. Additionally, three domains recommended by the Cochrane EPOC group were used to assess threats to validity; imbalance of the outcome measures at baseline, comparability of intervention and control group characteristics at baseline, and protection against contamination. These were incorporated due to the inclusion of non-randomised studies and risk of contamination in software interventions.

50Narrative synthesis is presented for included studies with standardised measures of effect calculated to aid comparison. For studies with multiple intervention groups, data were extracted for all comparisons. Where interventions were comparable (e.g., education by different providers), groups were combined to generate an average intervention effect.

51Due to intervention and methodological heterogeniety data synthesis was possible for only one comparison, with two studies. Randomised and non-randomised studies were considered separately, as these designs represented different levels of risk of bias. Review Manager 5.3 was used to conduct meta-analysis using the generic inverse variance method to enable inflation of standard errors for clustering. Intra-cluster correlation coefficients (ICCs) were obtained from study reports or requested from authors. There were insufficient trials to conduct subgroup or sensitivity analyses.

52After duplicates were removed a total of 10,674 titles and abstracts were screened (see Fig 1).

53Eight trials were eligible for inclusion. Five evaluated education interventions, two the impact of providing sanitary products, and one included both interventions. One study assessed menstrual knowledge for both males and females. Authors were contacted, but unable to provide separated comparisons. To provide an overview of available interventions, this study was retained although no outcomes could be reported.

54Study characteristics are presented in Table 2. Six countries were represented; Iran (3), Saudi Arabia, Zimbabwe, Ghana, Nepal, and Kenya. Sample sizes ranged from 120 to 1823. Five studies included only girls post-menarche. Exclusion and non-participation rates were typically not reported, although the majority of papers implied that few invited girls declined to participate. Studies typically included school-aged girls, (approximately 11–18 years). Wilson and colleagues reported an age range of 11–26, although the study was set in a secondary school with a mean participant age of 16 years. One study focused on university students. This study included only students who reported experiencing dysmenorrhea (three menstrual cramps in the past six months).

55Limited information was provided regarding the content of education interventions, but all papers reported including key information such as the origin of menstrual bleeding and menstrual management. The amount of education varied, as did the timing of outcome assessment. The trial by Djalalinia and colleagues in particular failed to describe education content or duration. The three hardware intervention trials differed in the products tested. One provided disposable sanitary pads, one self-made re-usable pads, and one reusable menstrual cups. All trials compared intervention groups to no-intervention controls.

56The timing of outcome assessment varied, from immediately after a 120-minute intervention to two-years post intervention. Primary psychosocial outcomes were only assessed in one study. Most software trials reported only secondary outcomes of attitude towards menstruation and menstrual knowledge. One study used a previously validated menstrual attitude questionnaire (MAQ; see), however, this did not capture vital psychosocial constructs reflected in background literature. Knowledge of menstruation was assessed via study-specific questionnaires, for which comparability is unclear. Whilst Fakhri and colleagues assessed many relevant outcomes, including self-reported school attendance, authors reported only a single collapsed ‘menstrual health’ variable and did not report menstrual attitude outcomes according to intervention or control group.

57Two eligible ongoing study were identified, see S1 Table.

58Risk of bias assessments for included studies are summarised in Fig 2, and discussed below. Justification for included study risk of bias ratings are reported in S2 Table.

59Three studies were non-randomised, representing a high risk of selection bias. Positively, these studies assessed and reported good comparability of intervention and control group characteristics at baseline. Wilson and colleagues reported that all outcomes were comparable at baseline. The primary outcome of school attendance was balanced at baseline in the study by Montgomery and colleagues, however, psychosocial outcomes were not balanced, representing a high risk for these outcomes. Of randomised studies, only one adequately reported methods of randomisation and allocation concealment. Of the three individually allocated studies, two were vulnerable to contamination. Education provided to girls in the same university dormitories or school could have easily been passed to friends in the control condition.

60All studies had a high risk of bias due to blinding, as it would not be possible to conceal the intervention from those providing or receiving education or sanitary products. Similarly, the majority of outcomes were self-reported by participants and most included items (e.g., attitudes) vulnerable to self-report biases, and may have been influenced by the lack of blinding. Two studies were considered to have a low risk of detection bias as they used official school attendance records. Whilst attendance data may not be highly reliable in the context of LMICs, both studies reported triangulating school records with researcher visits or girls’ own attendance diaries and noted a high level of attendance record reliability. It should be noted that for attitude items reported in Dolan et al. a high risk of detection bias is present. Mbvizo and colleagues (1997) tested menstrual knowledge by assessing students’ answers to questions about menstruation; judged to represent low risk of bias as the ability to answer accurately was unlikely to be altered by lack of blinding. There was considerable attrition in Wilson et al. as authors were unable to follow-up one intervention school. Djalalinia and colleagues stated in text that 5% of the sample was lost to follow-up, however reported N’s deviate from this figure with 1231 participants of 1823 (67.5%) reported on at two-years post intervention, with even smaller numbers displayed in results tables. Abedian et al. reported modest attrition, but failed to employ an intention-to-treat analysis. Fetohy assessed outcomes immediately after the education session, resulting in no attrition. Despite a 15-month duration, Oster and Thornton reported negligible drop out (one participant), and employed an intention-to-treat analysis. Some missing outcome data was reported by Montgomery et al. but this was judged to be low risk in light of analytic approaches, including imputing missing values. Only one study reported trial registration. Of the remaining studies, five reported all outcomes included in methods. One study, reported only an aggregated outcome, and one did not include a list of measured outcomes in methods.

61No other sources of bias were identified, although the suitability of outcome measures and study quality varied and are highlighted further in discussion.

62Results of software intervention trials are summarised in Table 3.

63Only one study assessed the impact of education on school attendance. When comparing the education-only school (n = 25) to the control school (n = 35) a medium effect (SMD = 0.63 95%CI 0.11–1.16) was found. Although a broad confidence interval, and comparison of single clusters cautions against over-interpretation.

64No studies assessed secondary education measures of achievement, or engagement.

65No studies assessed the impact of MHM interventions on employment.

66Primary psychosocial outcomes were assessed in one study. Authors reported that outcomes of shame, lack of self-confidence, insecurity, and difficulty concentrating did not improve in the education-only arm of the study.

67Djalalinia and colleagues evaluated primary outcomes of fear and confusion specific to menarche. Authors reported that negative psychological effects of menarche were lower in the groups which received the intervention. However, insufficient data were reported to calculate an effect size, and findings should be interpreted with caution in light of unclear attrition and analyses.

68Most software trials reported only secondary outcomes of attitude towards menstruation, or menstrual knowledge and practices (Table 3). Abedian et al. found a large positive impact of education on menstrual knowledge. On average, scores on the 10-item test designed by study authors increased from 3.86 to 8.99. The study used the MAQ (Table 2) to assess attitudes and found medium to large effects on most sub-scales (Table 3). Fetohy found a large positive effect of education on menstrual attitudes compared to controls, however menstrual attitude was conceptualised an attitude towards healthy practices which fails to capture attitudes towards experiencing menstruation and is not comparable to other studies. Fetohy found a large difference in menstrual knowledge post-education, with intervention and control groups differing by approximately 10-points on a test scored from 0 to 33. Similarly there was a large standardised mean difference between intervention and control students in self-reported hygiene behaviours. However, this outcome was measured immediately after a single session of education, leaving students no time to have changed behaviours. Thus this result is likely to represent recall bias or social desirability effects, and demonstrates the biases inherent in self-reported behavioural outcomes, particularly where blinding to condition is not possible. Fakhri and colleagues found a significant impact of their intervention on the inappropriately collapsed ‘menstrual health’ variable which included a wide range of practices.

69School attendance was assessed in all three trials. In accordance with review protocol, randomised studies were not combined with non-randomised studies. Thus two studies were eligible for meta-analysis. Intervention heterogeneity was high as studies differed in the products provided and education was included alongside pads in one study. Random-effects meta-analysis with standard-errors corrected for clustering revealed a moderate non-significant pooled effect (SMD = 0.49, 95%CI -0.13, 1.11), see Fig 3. Statistical heterogeneity was negligible (Chi2 = 0.04, df = 1, p = 0.84, I2 = 0%).

70Oster and Thornton reported education outcomes using days rather than participants as the unit of analysis. For a 180-day school-year, authors report that provision of a menstrual cup resulted in a small but non-significant improvement (β = 0.14, p>.05). The interaction between receipt of a menstrual cup and days of school missed during menstrual periods was investigated and also found to be a small non-significant effect (β = 0.02, p>.05).

71One study reported on psychosocial outcomes. Psychosocial outcomes varied significantly between clusters at baseline. In addition, these outcomes were not evaluated in the control cluster at follow-up, and test-retest reliability of scales was not reported (or evaluated). Dolan et al. were only able to report pre-post intervention changes in psychosocial outcomes for each cluster. Authors reported that for both sites where pads and education were provided, shame, lack of confidence, insecurity, and difficulty concentrating were improved following the intervention, with many outcomes (including shame, insecurity and difficulty concentrating) reported by approximately 25% fewer girls in each site at follow-up. These findings represent a much higher risk of bias as they were not compared to a control group, and it may be the case that negative psychological consequences of menstruation decrease naturally over time.

72This review identified six software, and three hardware trials of MHM interventions. Whilst heterogeneous with regard to intervention, context and delivery, software interventions were generally found to improve knowledge of menstruation. Menstrual knowledge is hypothesised to improve girls’ MHM and reduce negative psychosocial consequences. Some supporting evidence for these outcomes was found, with management practices reported to have been improved by education in three studies. However, unclear measures and self-report biases mean results should be interpreted with caution. In terms of psychosocial outcomes, more positive attitudes towards menstruation were found for girls who had received education in two studies. One study used a previously validated menstrual attitude questionnaire (MAQ;). However, this measure was developed in a high-income, US-context. Survey items, for example, ‘menstruation is an obvious example of the rhythmicity which pervades all of life’ and ‘a woman who attributes her irritability to her approaching menstrual period is neurotic’ do not reflect the fear and stigma facing women in LMICs. Only two studies measured primary psychosocial outcomes. One RCT found a positive influence of education on feelings at menarche, in contrast to a non-randomised study which found no improvements in shame, self-confidence, or insecurity following education. This same software intervention found positive effects on school attendance. However, the non-randomised pilot study was small, with single clusters compared.

73School attendance was assessed in all three hardware trials. Moderate, but non-significant, improvements in school attendance were found for hardware interventions providing varied sanitary products. One iRCT found no significant impact of providing menstrual cups. Although context should be taken into account, with very high attendance reported at baseline a lack of results may reflect a ceiling effect. Further accurate uptake of menstrual cups may have been difficult to capture as these could be difficult to use for young girls. Meta-analysis of two non-randomised studies providing disposable sanitary pads and reusable home-made pads revealed a moderate non-significant effect. Given the small sample and cluster sizes of the studies, both pilots, this may be due to a lack of power. Larger trials are needed to determine the true effectiveness of MHM hardware interventions. In such trials, assessment of supplementary outcomes such as achievement or engagement would triangulate findings. Only one study assessed hardware intervention impact on psychological outcomes, with reductions in shame, lack of confidence, insecurity, and difficulty concentrating reported. However, as noted previously only pre-post analysis of individual clusters was possible.

74High risk of bias was present in the included studies. Three non-randomised studies were included. Positively, appraisal using additional items suggested by the EPOC group showed intervention and control groups in non-randomised studies to be comparable at baseline on characteristics assessed. Risk of contamination was an issue in two individually randomised studies. Studies suffered from inadequate reporting with many items in risk of bias assessments ‘unclear’. Further, few studies reported implementation, or analytic strategies, for appraisal.

75A lack of appropriate outcome assessment limits the conclusions which can be drawn currently. No software trials assessed academic outcomes, and only two studies assessed primary psychosocial outcomes. Whilst such outcome measurement is challenging in LMIC contexts, no studies evaluated potential harms. Stigma, harassment, and unwanted sexual attention arising from the disclosure of menstrual status have been documented. The identification of menstruating females through interventions may ‘out’ girls’ menstrual status, resulting in negative effects on attendance or psychological harm.

76This review has highlighted many weaknesses in the existing body of evidence for MHM interventions. The challenges of conducting rigorous evaluations in LMICs should not be overlooked when considering the value of these findings. Furthermore, large gaps in MHM research more broadly contribute to the difficulty of conducting intervention trials. As noted, the lack of measures specific to MHM limits the reliability of outcome measures. Similarly, accurately assessing education outcomes may be challenging due to the differing reliability of school attendance registers or comparability of achievement measures across contexts. Sparse data exists regarding optimal genital cleaning practices for women and girls during menstruation, or the contribution of menstrual absorbent types to genital infection, comfort or education or psychosocial outcomes. Age of menarche differs contextually, and targeting interventions to the correct age group may be difficult in absence of population-specific data. Finally, as qualitative research has revealed, menstruation represents a taboo topic for many, and accurately capturing practices, attitudes or disseminating education to communities may fail if strategies lack cultural sensitivity.

77The present review comprehensively searched peer-reviewed and grey literature. Searches identified a large number of studies, most of which were of poor quality. All papers included in past reviews were identified. In addition, many excluded studies were not captured in previous reviews although were likely to have met their broader inclusion criteria. Publication bias could not be assessed as the evidence base is still limited. This review did not include interventions to address dysmenorrhoea, painful menses in women with normal pelvic anatomy, which has been identified as an issue affecting attendance at work, school, and contributing to negative experiences of menstruation. However, background searching suggested interventions targeting dysmenorrhoea in LMICs are yet to be trialled.

78The use of school attendance, rather than achievement or engagement, as the primary education outcome could be criticised. Authors have argued that a singular focus on attendance is a limitation of present literature. However, no studies reported on academic achievement so this would not have affected the conclusions of the review.

79There is insufficient evidence to determine the effectiveness of MHM interventions, although some positive indicators are emerging. In seeking to implement programs, practitioners and policy-makers should consider comprehensive evaluations. This review has provided a summary and critical appraisal of interventions and assessment methods used to date, which may aid the development of such evaluations. Larger, randomised trials are needed to determine the impact of MHM interventions. In such trials, cRCTs are likely to be most appropriate for avoiding contamination issues. Trials must compare the various intervention strategies proposed; including both hardware and software interventions and the interaction between them. To aid this process, interventions should be based on a clear theory of change which includes the various individual and contextual factors which contribute to women and girls MHM. Such theories should be tested, and mediators and moderators of effects identified, for example the distribution of reusable sanitary pads may not be effective in the absence of improvements to WASH. Absorbent sustainability, acceptability, comfort, and risk of reproductive tract infections must all be considered when selecting sanitary products appropriate for interventions. A reliable and consistent assessment of all aspects of MHM in observational studies and trials may serve to draw attention to the various aspects of MHM contributing to outcomes.

80As noted above, there is a lack of research capturing MHM practices or quantifying the health, education, and psychosocial consequences of different MHM practices. Providing data on the hypothesised causes and consequences of poor MHM would help guide the development and evaluations in future. These trials should seek to use validated measures of psychosocial constructs and measures of women’s wellbeing specific to, or which include, MHM. This would improve comparability across studies, enabling researchers to grasp the severity of distress caused by menstrual poverty, and the influence of interventions.

81Studies are needed on the impact of MHM on adult women, and potential WASH interventions as no trials were identified. Future trials must evaluate potential harms of interventions, particularly consequences of ‘outing’ menstruating girls in contexts with high levels of stigma associated with menstruation.