Women's toilet use behaviors and water use in a urine diversion squat pan toilet pilot

Sustainable Development Goal 6 of ‘Clean Water and Sanitation’ for all has focused investments in sanitation to expand global latrine coverage, reduce open defecation, and advance sanitation innovations with an explicit focus on the needs of women and girls (United Nations 2017). Investments in sanitation innovation have spurred the development of novel technologies that manage human waste safely, operate independently from sewered infrastructure, and generate renewable resources such as fertilizer from human waste (Nazari et al. 2020; Wald 2022). These non-sewered waste treatment technologies and their components require field testing and user engagement to ensure operational effectiveness and uptake among intended user populations (Murphy et al. 2009).

For the health and environmental benefits of these technologies to be realized, user adoption is essential, which requires that the systems reflect the needs (e.g. preferences, behaviors, and willingness to pay) of end users (Pattanayak & Pfaff 2009; Seymour et al. 2021). While studies documenting field-testing challenges in the published sanitation literature are limited, even fewer discuss user behavior alongside questions of technology adoption (Woltersdorf et al. 2016; Welling et al. 2020). As sanitation technologies become further specialized, the role of user behavior in technology-specific studies is important for understanding performance, so that prototypes can be adapted and iteratively optimized.

To further encourage adoption, the important role of gender in defining toileting behaviors and the social context of toilet use must be incorporated and well understood (Tilley et al. 2013; Schmitt et al. 2018). Innovations in sanitation should reflect gendered differences in sanitation and hygiene to promote adoption (Burt et al. 2016) and collect gender-sensitive and context-relevant data to advance technologies and interventions that meet the unique needs of women and girls (Fisher et al. 2017; Crawford 2020; Elledge et al. 2020; MacArthur et al. 2020).

Studies from the sanitation literature that examine user preferences typically focus on high-level design and perceived attributes of toilets to promote uptake and sustained use, such as cleanliness, water availability (Routray et al. 2015), maintenance, and convenience (Seymour & Hughes 2014; Morrison et al. 2018). Existing studies often lack an explicit focus on gender-specific findings. Sanitation behavior change literature focuses primarily on effective mechanisms for influencing built toilet use in place of open defecation, such as information-based educational campaigns (Evans et al. 2014), incentives (Pattanayak et al. 2009), and peer networks (Pakhtigian et al. 2022). Furthermore, few studies provide descriptive information on the prevalence of certain behaviors during toilet use. To assess system inputs, such as water use, some evidence is available to characterize the water volume used during toilet use (Kumar et al. 2021) and pour flushing (Welling et al. 2020), yet it leaves a gap in behavioral explanations for when and why water enters the system (e.g. for self-cleaning or for cleaning the stall).

NoMix toilets are designed to divert urine from the waste stream to facilitate nutrient removal in wastewater treatment systems and recover nutrients such as nitrogen (Yu et al. 2021; Wald 2022; Nguyen et al. 2024). Urine represents only 1% of wastewater volume but as much as 80% of wastewater nitrogen (Tarpeh et al. 2018), driving energy-intensive treatment processes that can be inadequately implemented in low-resource settings. Typically, NoMix toilets are comprised of two compartments, one for urine and one for feces. The toilets face barriers in consumer adoption due to behavioral requirements, such as the need to aim urine into a compartment separated from solid waste (Lienert & Larsen 2010; Mkhize et al. 2017; Larsen et al. 2021). The Urine Trap is a squat pan NoMix toilet that passively achieves urine diversion due to an opening that is not visible to the user (Gundlach et al. 2021). Developed with women in mind, the Urine Trap Indian squat pan is designed to limit undesirable urine splashing on one's feet while urinating in a squatting position (Schelbert et al. 2021). While the unique sanitation needs of women and girls are rarely accounted for in toilet design, the Urine Trap sanitation technology reflects design choices that consider gender-disaggregated data and women's unique physiology, including gender-specific body postures and urine stream angles.

A recent evaluation of the Urine Trap toilet's engineering performance during field testing in two women-only, shared bathrooms in India identified challenges in separation efficiency, despite very promising results in calibration testing (Kachoria et al. 2023). The technical assessment demonstrated a much-reduced nitrogen separation efficiency (36%) relative to the values expected based on system calibration conducted in situ (70–80%). This analysis motivated further exploration to consider the role of women's toilet use behavior in impacting the toilet's separation efficiency.

The hypothesis that user behaviors may affect urine separation efficiency was advanced based on initial testing. Behaviors such as facing backwards, not squatting down, and aiming urine at the back hole instead of onto the ceramic pan were suggested for further investigation. This study aimed to understand how user behavior may influence the performance of the urine diversion toilets in two women-only, shared toilet blocks in India. We aimed to determine users' knowledge, perceptions, and behaviors around this specific technology and to learn how their behaviors when using the toilet may be altering the toilet's intended function. A particular focus was given to liquid waste (i.e., urine, flush water, and wash water), specifically when and how liquid waste is introduced into the toilet as these are critical elements for urine diversion design.

The sanitation testing facilities provided unmonitored user access to bathroom stalls and hand washing sinks, and a separate space and entrance for engineering monitoring and sampling. All toilet stalls in these facilities had available water (i.e., a spigot, wash wand, bucket, and a small mug). The first site was located on a medical college campus where users were Tamil-speaking. The second site was in a privately owned textile mill with approximately 30 resident workers and 20 non-resident workers. The textile mill site users hailed from regions across India; the resident workers spoke Tamil, Hindi, and Odia, while non-resident workers predominantly spoke Tamil.

Finally, the survey tested several potential approaches for encouraging toilet use behaviors through visual and written messaging. The complete survey of instruments and visuals is included in Supplementary A.

Survey respondents represent a convenience sample drawn from regular toilet users at each of the two sites in Tamil Nadu. A series of three recruitment posters were created and printed in Tamil, Hindi, Odia, and English and placed at both sites to encourage participation in advance of the survey. Trained female enumerators from the PSG Institute of Medical Sciences & Research (PSG IMSR) delivered the survey to respondents between February 6 and March 30, 2023. All survey interviews were conducted privately near the toilet facility. Enumerators encouraged respondents to speak honestly, acknowledging that discussion of toileting behaviors can be considered socially taboo, and respondents' comfort was assessed following each interview (Rop 2010; IASC 2017; Elledge et al. 2020). The survey was designed, and data were analyzed by RTI International with support from the Duke University Center for WaSH-AID. The survey was translated from English to Tamil, Hindi, and Odia and administered accordingly based on respondent language. Survey responses were recorded on paper and entered into a spreadsheet and then translated into English.

At the testing site, the community of users was aware that the facilities were being used as part of sanitation research and that data were being collected for research purposes based on past communications. The survey instrument was reviewed and approved by the Institutional Human Ethics Committee, PSG IMSR, in conjunction with the IRBs of Duke University and RTI International. All respondents have used the toilets regularly in the past and provided written consent to engage in our voluntary survey. Survey respondents received a small gift of hygiene items for their participation.

Forty-one survey responses were collected and reviewed for quality control. Data were analyzed using thematic analysis (qualitative data) and statistical analysis (quantitative data) using STATA. We examined each of the four behaviors that were hypothesized to affect toilet performance (Figure 2). Other variables of interest that were analyzed were those indicating when participants used water (i.e., before, during, and after cleaning the floors). Descriptive statistics were obtained and each of these variables was cross-tabulated with a handful of demographic variables of interest.

Survey respondents included 41 women, all of whom were regular users of women-only bathroom facilities at two sites: a medical college campus (32%) and a textile mill (68%). All respondents were employed at or near the sites at the time of survey delivery. At the textile mill, 39% of respondents lived onsite as resident workers, while 61% lived nearby and worked onsite during the day (Table 1).

Respondent backgrounds varied across the two sites: respondents from the textile mill were, on average, younger, more familiar with open defecation than built toilets, and hailed from states outside the region, primarily from northern India. Respondents from the college campus were older and all from Tamil Nadu. At the textile mill, 46% of respondents were from Tamil Nadu, with the remaining sample originating in Jharkhand (39%), Odissha (10.7%), or Kerala (3.6%). Most respondents identified their hometown as a village (93%), as opposed to a city or a town. When asked what type of toilet respondents used most often, 15% of respondents from the college campus site and 32% of respondents from the textile mill selected open defecation.

No respondents identified a difference between the Urine Trap toilet's urine diversion squat pan and others they had used in the past or within the same facility. Respondents also stated that they did not alter their behaviors when using the stall where the Urine Trap toilet was located. When asked what was best about the toilet compared to others they had used previously, most respondents identified attributes that were unrelated to the Urine Trap toilet itself, such as the availability of water within the stall and the stall's cleanliness and comfort.

Figure 3(b) illustrates an analysis of urination positions and behaviors that do not favor efficient separation segmented by respondent characteristics. Respondents that are most familiar with open defecation as opposed to built toilets are more likely to aim urine directly into the hole of the squat pan, rather than onto the pan itself, suggesting that this behavior may stem from limited experience with built toilet use. Separately, users that squat only partially (i.e., do not squat down) draw from the oldest age group. When questioned further, respondents identified knee pain as a limiting factor. Given the low sample size, little can be effectively correlated between behavior and respondent background characteristics.

Separately, 88% of users clean the floor and foot pads of the toilet with each use; some or all of this water may enter the toilet instead of the floor drain depending on the construction of the bathroom.

Respondents who adopted toilet behaviors that negatively impacted urine separation efficiency (e.g. facing backwards) were asked how and if they might be influenced to change that behavior. Most respondents suggested that it would not be hard to change their practice, and many had not been aware that their behavior had an impact on the performance of the toilet. Several (23%) suggested that displaying a poster/image guiding the desired behavior might be helpful.

When respondents were shown sample visual and written guidance on urination positions and behaviors that favor efficient separation, few respondents found the cues clear (Supplementary B). Depending on the visual, between 3 and 37% of respondents could clearly decipher the behavior that was being encouraged.

Assuming that all respondents contributed an equal number of toilet uses, such that 32% of the total toilet uses reflected positions and behaviors that do not favor efficient separation, we calculate a nitrogen separation efficiency of η_TN = 43% for the Urine Trap based on calibrations reported in Kachoria et al. (Supplementary C). This estimated η_TN value is in line with the reported measurement of η_TN = 36% (Figure 5). Thus, it is conceivable that one-third of users adopting a position or behavior that does not favor efficient urine separation could nearly halve nitrogen separation efficiency. This result indicates that the outcome of the user survey described in this study helps to validate the prior engineering observations and the initial hypotheses. Supplementary C reports the details of the calculation, including literature values of nitrogen content in excreta and calibration measurements on the squat pan.

While this study features a focused line of inquiry around a specific toilet system, it brings to light important considerations around user behavior that may impact the design and engineering of toilet systems generally and NoMix urine diversion toilets more specifically. Few studies have examined the gender-specific physiological differences in toilet use practices, including urination, to which our study contributes valuable design input. Schelbert et al. (2021) found a wide variation in body postures and urine stream angles among men and women during a field test in Hyderabad, India. However, that study did not examine the direction users faced and water-use behaviors in the toilet stall, which introduce important toilet design considerations.

Another insight from this work is that targeted information and education materials inside toilet stalls may help to promote optimal use behaviors. However, these materials should be developed with input from the target population of toilet users and tested for effectiveness and context appropriateness. More broadly, the findings of this study reinforce the importance of field-based technology testing in determining user interaction with a technology. Such interactions may lead to the revision of the system design and/or help to determine whether behavioral nudges need to be integrated into its implementation.

Further, this study adds to the limited literature documenting water-use behaviors during toilet use (Kumar et al. 2021). The identification of multiple purposes and instances in which water is introduced into a toilet system during urination and defecation points to complex preferences and motivations for water usage. In some cases, diverse water uses were reported by survey respondents, while over 90% of the women reported wetting the toilet prior to use.

We also found that cleaning the stall floor around the toilet is common (88% of respondents), which has implications for water introduction into the system. Depending on the construction of the bathroom and the ease with which water may enter the toilet instead of the stall floor drain, the water used to clean the floor may end up diluting the blackwater effluent of a system. This is important to consider as it may impact the resulting blackwater strength and volume and may alter the treatment requirements of non-sewered sanitation systems.

In this study, we relied on self-reports of women's toileting behavior and such reports can introduce several kinds of bias and subjectivity to responses. The survey instrument was carefully reviewed to ensure that questions inquired objectively around behavior and did not indicate a ‘correct’ or socially optimal answer. Enumerators affirmed multiple times that there were no correct answers to the behavioral questions and reiterated the study's goal to improve technology based on how real users engage with the toilet. We also note that social desirability bias, should it have been present, would have attenuated the number of responses describing behaviors that do not conform to the expected norms of squat pan usage, the main finding of this work. The survey was additionally delivered by female enumerators to reduce the potential apprehension to discussing matters that are often not discussed or considered private, such as toilet use.

In conclusion, we found that urination positions and behaviors that do not favor efficient nitrogen separation performance in two urine diversion toilets used by women in India are common. The widespread nature of the behaviors suggests that the separation efficiency identified in previous studies may, in large part, be explained by these factors. This study additionally contributes to the limited literature on water-use behaviors during toilet use. These water-use behaviors hold useful implications for estimating blackwater strength and volume and for the design of non-sewered sanitation systems.

This study focused on the behavior and practices of women and the impact these personal choices have on toilet design and performance. It recognizes the importance of understanding gender-specific needs to ensure inclusive access to safe sanitation.

We acknowledge the Coimbatore field team members Viswa Barani and Pragadeesh Subramani for carrying out arrangements for the study, Sarani Sasidaran for facilitating the interviews at the textile mill, and Antony Raj and Prateek Kachoria. We acknowledge Claire Welling for reviewing the comparison to results from engineering studies. We thank Ms Kasthuri and Ms Suganthi for their assistance. This work was supported, in whole or in part, by the Bill & Melinda Gates Foundation grant INV001513 to Duke University. Under the grant conditions of the foundation, a Creative Commons Attribution 4.0 Generic License has already been assigned to the Author Accepted Manuscript version that might arise from this submission.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.