The Menstrual Practice Needs Scale Short Form (MPNS-SF) and Rapid (MPNS-R): Development in Khulna, Bangladesh, and validation in cross-sectional surveys from Bangladesh and Uganda

The present study

We aimed to provide short and rapid versions of the MPNS to meet the needs of different users, and to compare the performance tools at shorter lengths. For the short form we aimed to halve the length of the MPNS-36, and to halve this again for a rapid version.

Guiding principles for item reduction were set a-priori based on past research and theory. First, we prioritised content validity and retaining the breadth of experiences assessed through the measure above structural validity. Item selection was not driven by item factor-loadings alone. As noted above, the experiences measured draw on systematic review of qualitative research and ensure that intervention evaluations capture holistic menstrual management experiences. Second, single items from the MPNS have been included as part of recommended indicators for national and global monitoring of menstrual health and hygiene.[17, 18] We prioritised retaining these in shorter forms to enable comparability of data collected using the scale with national data. Third, we decided a-priori to retain items in the MPNS that capture experiences separately relating to the home and school environments. Research has consistently highlighted differing experiences at home and at school (for adolescents) or work (for adults [19]), and studies using MPNS data have consistently shown differing experiences of menstruation in these settings.[6, 9] While duplicate items contribute to length of the scale, they are useful for policy and practice. In needs assessment, they highlight areas of greatest need, while in evaluations differences in items over time or between study sites can provide feedback on environments improved by the intervention. Fourth, we prioritised retaining a balance of positively and negatively orientated items in shortened versions. Assessments of MPNS dimensionality have consistently found that items capturing positive appraisals of experience such as satisfaction with the available changing facilities or having enough menstrual products, load on separate but correlated factors to those capturing insecurities such as worries about privacy or leaking.[3, 4, 6] As noted in the original development, including both positively and negatively framed experiences balances framing in administering the items with participants a offers a more nuanced assessment of experience. Combined scores across positively and negatively scored items have demonstrated stronger relationships with hypothesised correlates than these item sets alone across multiple studies.[3, 6, 15]

Study samples and data collection

Survey

A pilot survey was undertaken across 10 purposively selected schools with girls attending Classes 6-9 and aged 12-16 in October 2022. A target sample of 360 participants provided 10 participants per MPNS item. Female enumerators received 5 days of training and administered the survey verbally, entering responses into tablets and uploading to the BRAC James P Grant School of Public Health KoboToolbox server. Girls were provided with a written copy of the survey if they wished to follow along and read the questions for themselves. The printed survey included the visual response tool for the MPNS depicting the four response options.[3] Surveys lasted 45-60 minutes for post-menarche participants.

Cognitive interviews and the survey followed sensitization workshops at each school notifying teachers and parents about the study. Parents/Guardians provided written consent, and girls provided written assent to participate.

Sample demographics and menstrual practices

Participants self-reported their age, class level and the materials they used as menstrual absorbent. In all three data collections, girls self-reported if they washed and reused any menstrual materials during their last period. In both Bangladesh data collections, girls also self-reported if they attended school during their last menstrual period, and if they changed their menstrual materials at school. These were used as subsequent eligibility criteria for MPNS items.

The MPNS

The full MPNS-36 was administered to girls in the Khulna pilot. The subsequent cohort baseline used the short-form items only. Both Bangladesh surveys used the interview version of the MPNS, which presents each item in question format, for example “During your last menstrual period, were your menstrual materials comfortable?”. The full set of questions are displayed in tables in the results section. Participants provide responses on a 4-point Likert scale: never, some of the time, most of the time, always. These terms were determined to be appropriate in the Bangla translation, however training with enumerators also highlighted that the middle options could also be interpreted as ‘less than half of the time’ and ‘more than half of the time’. Prior to asking the MPNS items, in both the pilot and cohort surveys, enumerators administered a brief exercise to familiarise girls with the response options. This included asking girls about daily activities such as “Over the past week, how often did you have street food?”. Girls were also presented with a printed copy of the survey which displayed the response options and MPNS visual response tool [3]. Participants were asked to respond to MPNS items concerning the disposal (Items 12-15) of menstrual materials if they reported disposing of materials during their last period, including disposing of single-use or reusable products at the end of their life. They were asked to respond to questions about laundering materials if they reported washing and reusing any material during their last period (items 29-36). Girls who reported that the ‘never’ changed their materials at school during their last period were not asked questions about the experience of changing materials at school (items 25-28).

In Soroti, Uganda, as part of the original MPNS-36 development, girls completed a longer set of candidate scale items. These were delivered as statements which girls responded to on their own survey, for example “During my last menstrual period, my menstrual materials were comfortable” with response options: never, sometimes, often, always.

MPNS sub-scale and total score are calculated using the mean of included items. Positively worded items are scored from 0-3 and negatively worded items from 3-0. Higher MPNS scores represent more positive experiences of menstrual blood management.

Analyses

Quantitative analyses were undertaken in Stata 17 and in R version 4.3.1. Revalidation of the MPNS-36 drew on girls’ reflections in cognitive interviews used to refine translation and check interpretability of the questions. Quantitative revalidation was undertaken in the Khulna Pilot data. Descriptive statistics were used to summarise sample responses to each MPNS item and identify missing data. Confirmatory factor analysis (CFA) using the lavaan[25] package for R was undertaken using a diagonally weighted least squares estimator (DWLS). We considered RMSEA ≤0.08 as indicative of a fair fit, and Comparative Fix Index (CFI) and Trucker-Lewis Index (TLI) ≥0.90 as indicative of acceptable fit (with CFI and TLI ≥0.95 indicative of close fit).[26] Factor loadings ≥0.30 were considered acceptable. Scaled estimates are presented. Internal consistency was assessed using Cronbach’s alpha, and ordinal alpha calculated using the polychoric correlations given the four-point response scale.[27] Validity was assessed by exploring associations between the MPNS-36 total and sub-scales with constructs hypothesised to be related.

To develop the short and rapid forms, girls’ perception of item importance from cognitive interviews were integrated with a-priori defined priorities for item selection. Girls scoring of the importance of items in cognitive interviews were used qualitatively and in the context of interview discussion of these decisions.

The dimensionality, internal consistency and validity of the short form was tested in the Khulna Pilot survey following the same procedures used for the MPNS-36 revalidation described above. For tests of dimensionality (structural validity) we first undertook CFA, hypothesising the original sub-scales would be replicated in the short form. To supplement this assessment, we also undertook exploratory factor analyses (EFA) using the polychoric correlation matrix and oblique rotation (promax) to investigate alternative structures, however the original was the best fit for the data for the short form. To test the dimensionality of the rapid form, we undertook EFA, and CFA testing hypothesising one and two-factor structures.

Tests of dimensionality of the short and rapid forms using CFA were then replicated in the Khulna Cohort Baseline, Bangladesh and Soroti, Uganda datasets. We note that DWLS requires complete data and so analyses reflect the sample of participants with no missing items. In the dataset from Soroti, Uganda we used the first multiple imputation data generated using chained equations with the mice package[28] in R from the original study.[3] Internal consistency using Cronbach’s alpha and ordinal alpha, along with tests of associations with hypothesised correlates were undertaken. Correlation coefficients or binary logistic regressions were used as appropriate to the distribution of the outcome.

Patient and public involvement

NGO practitioners and researchers as the potential users and audience for the measure informed original development of the MPNS. Feedback from users shared with the development team informed considerations for short and rapid form development and cognitive interviews with girls were undertaken to support item selection. Community consultation for the broader AMEHC study was undertaken (see [20]) but this did not include focus on the MPNS.

Survey sample characteristics

Sample characteristics of the three quantitative datasets used in the study are presented in Table 1.

Re-validation of the MPNS-36 in Bangladesh

Twenty girls across classes 6-9 and aged 13-16 participated in cognitive interviews testing comprehension of MPNS items, translation, and informing the subsequent short-form development. Initial cognitive interviews highlighted easily understood questions, and those that required further amendments to translation. Most translation improvements were grammatical, relating to the ordering of sentence content. There were also modifications based on individual words. For example, “comfortable” had multiple translations in Bangla dependent on physical or emotional comfort. MPN1 refers to physical comfort, whereas items such as MPN10 referred to “feeling comfortable’” in English and the selected translation prioritised mental safety/peace comfort. By the conclusion of early translation modifications, all MPNS items were well-understood by respondents. In interviews, participants described varied circumstances and preferences that influenced their response selection. Quotations are presented in Supplementary Materials 1.

The MPNS-36 exhibited strong performance in the survey data collected in the Khulna Pilot in Bangladesh. The original MPNS-36 factor structure was an acceptable fit for the data (CFI=0.924, TLI=0.927, RMSEA=0.075, 90%CI=0.060-0.090). Sub-scales and total scale exhibited good internal consistency (total scale α=0.86) and the total and sub-scales showed multiple expected relationships with other constructs, including mental health, participation in school, and confidence managing menstruation at school. Full tables and text reporting the revalidation are presented in Supplementary Materials 2. This supplement also reports the responses to each survey item in this population.

Short and rapid form item selection

There was agreement across the cognitive interviews that the experiences captured in the MPNS items were important and relevant for girls. There was substantial inconsistency across the interviews in the items rated as most or less important. In discussions, girls described that items were likely to be more important for different individuals, depending on circumstances. For example, groups consistently rated safety at school (MPNS 28) as a less important item because they did not personally have concerns about their safety in school toilets. However, safety at home (MPNS21 and 22) were rated by multiple pairs as “very important”, with girls’ emphasising that if this was a concern for girls, it would be a top priority. Pairs who did not wash and reuse materials did not see the relevance of these items, while those with difficulties rated this as ‘very important’; particularly privacy for washing and drying absorbents. In the four sub-scales that apply to all participants, the most highly prioritised items were:

• Material and home environment needs: comfortable menstrual materials (MPNS1) and being able to wash hands (MPNS11). Hand washing was particularly relevant in the context where menstrual items are considered unclean, and participants described washing their hands after having any contact with items related to menstruation.

• Change and disposal insecurity: being worried others would see disposed menstrual materials (MPNS14) and privacy for changing menstrual materials at home (MPNS20).

• Material reliability concerns: being worried materials would leak (MPNS5).

• Transport and school environment needs: being able to change menstrual materials when desired at school (MPNS23).

The rationale underpinning item prioritisation and selection for the short and rapid forms are summarised in Table 2. Content validity was the highest priority when reducing the item set. While more similar items, related to more similar practices, often exhibited higher factor loadings and enhanced model fit, we prioritised selecting a single item that represented each practice experience and removed items that provided different perspectives on a similar aspect of blood management experience. Items were also prioritised based on (a) their selection as indicators for national and global monitoring, (b) item performance in past data collections, for example while items 8 and 9 capture unique experiences of transporting materials they have cross-loaded and created problems for scale factor structure,[6] and (c) girls’ perspectives on item importance in cognitive interviews.

Short form dimensionality

Dimensionality of the short form was assessed in the Khulna Pilot, Bangladesh. The original four-factor structure, and two reuse factors, remained an acceptable fit. In testing in the Khulna pilot sample, EFA was undertaken and further indicated that the original four-factor structure offered the best sub-scale solution. Single and two-factor structures did not offer better fit for the data and demonstrated poor factor loading for school-related items and material reliability concerns.

Structural validity was then replicated in the Khulna Cohort Baseline, Bangladesh and Soroti, Uganda populations. CFA findings for all three datasets are presented in Table 3. Despite the small sample size among the Bangladesh samples including girls who changed their menstrual materials at school, the four-factor structure was an acceptable fit, achieving a good fit in the original Soroti, Uganda data with a more substantial sample.

For girls who did not change their menstrual materials at school, and thus did not answer MPN27 and MPN28, the four factors were an acceptable fit for the data. However, we note that a Heywood case was observed in the cohort baseline data which excluded these items (see Table 3). In exploring this case, we found that both MPN23 and MPN24 have low bivariate correlations with other non-school related items in the measure (all polychoric correlations <.15 in both the pilot and cohort samples) including some 0 bivariate relationships. They have correlations with MPN27 and MPN28 (polychoric correlations .22-.32 across both samples) which in turn have meaningful correlations with other items across the scale. When removing MPN27 and MPN28 to test the factor structure in girls not changing menstrual materials at school, it is likely these zero-correlations (and the low correlation between the remaining school subscale and other factors (correlations=0.07-0.15) results in a Heywood case. It is consistent with our expectations of the measure that girls’ satisfaction with the available school facilities will not have a strong relationship to their satisfaction with their materials or home environment, while insecurities in the home and school environment have a closer relationship demonstrated by loading on the same factor. The sample excluding girls who change at school in this analysis also means that girls’ reports about their satisfaction with their disposal experience captures only the home environment, further minimising relationships between MPN23/24 and other scale items. As such we did not interpret the Heywood case as indicating model misspecification.[29] However, this finding suggests that the sub-scale structure is less stable when the full set of items is not included due to eligibility constraints.

Rapid form dimensionality

Items in the rapid form prioritised breadth. As a result, most factors were only represented by one or two items; with single item factors unable to be included in CFA for model fit. As we hypothesised based on findings from the full and short form MPNS, a single-factor solution was not an acceptable fit for the data (excluding reuse items) in the Khulna pilot dataset (CFI = 0.897, TLI = 0.828, RMSEA 0.131, 90%CI 0.058-0.205). A two-factor solution offered the best fit in the Khulna Pilot but did not reach thresholds of acceptable fit for all metrics, and this was replicated in the other two datasets (see Table 4). School-related items loaded poorly on the two factors. However, we note the small sample size to test the full set of rapid-form items in the Khulna pilot and cohort datasets given the limited number of girls who changed their menstrual materials at school. In the datasets including the larger number of participants who did not change their menstrual materials at school, remaining school-related items fit poorly with the factor structure and compromised stability of the model.

Validity and internal consistency

Mean, standard deviation, internal consistency and validity tests are presented in Table 5 for the respective MPNS-36, MPNS-SF and MPNS-R total scales. Findings for MPNS-SF sub-scales are presented in Supplementary Materials 3. The MPNS-SF total scales displayed adequate internal consistency across datasets, with sub-scales largely displaying strong reliability particularly when tested using α for ordinal items. Two-item sub-scales such as ‘material reliability concerns’ exhibited poorer reliability as expected for the small number of items. The MPNS-R had acceptable ordinal internal consistency in Bangladesh, but poorer reliability in the Soroti, Uganda sample.

Short and rapid forms of the measure were found to demonstrate hypothesised relationships with school participation, and confidence managing menstruation across datasets. For each reduction of the scale, the strength of these associations was attenuated.

Strengths and limitations

We triangulated insights from implementation of the MPNS, the perspectives of adolescent girls in Bangladesh, and advances in menstrual health research guidance on monitoring [17] to develop the MPNS short and rapid forms. A data-driven approach to shortening the scale would likely have yielded greater model fit of sub-scales but would reduce the breadth of practices and thus validity of the measure. A strength of the MPNS remains its development drawing on synthesis of qualitative research of women’s and girls’ experiences of menstruation across low-and-middle income country settings.[3, 16] The variety in menstrual practices that adolescent girls employ means not all MPNS items are relevant to all respondents. Non-applicable items reduce the sample sizes available for undertaking tests of dimensionality with complete data. In our Bangladesh samples where many girls do not change their materials at school meant restricted samples were available to test dimensionality for all items. While measures are emerging for capturing menstrual experiences in different populations,[32, 33] we did not have other measures of adolescent menstrual experiences against which to test convergent or divergent validity.

Implications for research and practice

Caring for the body during menstruation in a way that meets individual needs for hygiene, comfort, privacy, and safety is a requirement for menstrual health, and objective of many menstrual health policies and programs.[13, 34] The MPNS-SF and MPNS-R offer concise measures to capture this construct. Based on the findings of measure development, the MPNS-36 still offers the greatest nuance in describing population needs and explanatory power in investigating hypothesised correlates. The MPNS-36 may be best suited for needs assessment in a new population. The MPNS-SF offers a reliable and valid measure, with sub-scales matching the original, associations with hypothesised correlates, and good internal consistency. The measure maintains a balance in practices and captures both having needs satisfactorily met and insecurities related to menstrual management. It also offers minor modifications to improve the applicability of some items in the school environment. We recommend that the MPNS-SF be used when a shorter measure is needed such as in research and intervention evaluation where survey length is restricted, or multiple constructs must be accommodated. Finally, the MPNS-R provides a restricted set of questions with brief survey duration. In shortening the measure further, the MPNS-R no longer offers reliable sub-scales, with multiple original sub-scales represented by a single item. However, the tool retains breadth in capturing experiences of menstrual practices and environments and is preferable to selecting sub-scales of the original measure or ad-hoc item selection which results in incomparable data across studies. We recommend the rapid form is used where menstrual health is not the primary area of focus but included as part of studies of broader water, sanitation, and hygiene (WASH) or sexual and reproductive health (SRH).

Future longitudinal research using the MPNS measures will test their predictive validity and the impact of menstrual experiences on individuals’ lives. Subsequent waves of the AMEHC Bangladesh study will offer these opportunities,[20] and more cohort studies will be needed across contexts. Trials using the MPNS measures as primary or secondary outcomes will provide insights into the sensitivity of the measure to change and use for evaluation.[8, 12] Finally, future research should investigate the validity of the shorter MPNS forms in new cultural contexts, languages and age groups. As found previously,[6] the dimensionality of the MPNS-36 differs for adult women, and we would hypothesise similar differences for the short form. Improved availability of high-quality measures for menstrual health research and practice will strengthen the evidence base and aid comparability across studies.[1, 35, 36]

Ethical approvals