Study design, sampling frame, inclusion criteria, and sample size

To maximize generalizability, a random sample of rural ADP-owning households was drawn from iDE’s historical sales data in five Cambodian provinces where ADPs were sold at scale. From the random sample, a phone survey then identified households with filled pits to ensure that households included in the study had at least an opportunity to treat FS and switch pits. Households were only eligible for inclusion in the study if they owned their ADP for more than two years and had switched their pit connections from one full pit to the other at least once since their ADP was installed (full inclusion criteria listed in the Supplementary Information S2 File). With 8,410 ADPs used for at least two years per iDE’s sales database and an estimated 25% of ADPs filling after two years per the phone survey, 765 ADP-owning households were surveyed (S1 Table) [29,30]. Following standard practice described in the literature and internationally accepted protocols, the study design aimed to achieve a 5% precision when describing the target population [31]; however, practical restrictions due to accessing respondents that met all inclusion criteria, travel distances, and budget limited the survey’s precision to 6%. A description of the study population, including household demographics and details about household latrines is available in Supporting Information S3 File. This study design was approved in the Solutions IRB protocol 2021/10/2.

Data collection

A 40-minute in-person survey (see Supporting Information A) was developed and administered to 765 rural households in five of 25 Cambodian provinces where ADP installations were sold at scale. Based on a review of existing literature, including reports from WASH-sector organizations in Cambodia to ensure contextual relevance, the topics included in the survey were evidence-backed factors that affect sanitation practices and included:

• Demographics: household size, highest education level of any household member, primary occupation or livelihood activities of household members, poverty status (IDPoor), gender of respondent, household members with disabilities, and ethnicity. Household poverty levels were self-reported using the Cambodian National Government’s Identification of Poor Households (IDPoor) Programme metrics [32];
• Sanitation history and toilet use information: number of latrine pits, frequency of use, number of people using the latrine, sharing status, improvements, system modification (e.g., a pit is “pierced”; i.e., an overflow hole was installed in a pit) and maintenance issues (e.g., repairs);
• Knowledge of ADPs and SMS: understanding of FS dangers; pit-emptying and treatment processes that can improve safety; equipment and clothing used to empty pits safely; and methods of emptying pits;
• ADP experience: how, when, and why latrine was switched to a different pit; perceptions of expense, disgust, difficulty, and safety; disposal location; method of disposal; and maintenance issues (e.g., repairs); and
• Attitudes about ADP and SMS: perceptions of ADPs, including safety, ease of use, maintenance, disgust, and affordability.

All interviews were conducted at the site of the household-owned ADP using secure smartphones, and data was managed and stored on secure servers. The study design was approved in the Solutions IRB protocol 2021/10/2, and respondents provided oral informed consent to participate prior to beginning each interview. Written consent could not be obtained because most study participants were illiterate; Solutions IRB approved the use of oral consent; and oral consent was documented digitally, along with all other responses, for each household by recording the response of each study participant to the question “Do you consent to being a part of this study?”. Data collection occurred between Mar 31, 2022 and May 4, 2022.

Data analysis

Survey data were analyzed to describe 1) households’ knowledge and perceptions of recommended ADP practices and sanitation in general; 2) households’ practices and experiences with ADPs; and 3) common characteristics of households that use ADPs as recommended. Response frequencies were used to analyze these aspects of households, and descriptive statistics were disaggregated by province and geographic area (flood-prone vs. non-flood-prone), where relevant. To identify common characteristics of households that use ADPs as recommended (S2 Fig), five multi-factor indices were developed to describe households’ practices and perceptions related to recommended practices for ADP operation. These five ADP Indices are:

1. The Emptying Practices Index describes whether households emptied their ADP’s pits as recommended, including whether the household ever emptied a pit; whether the pit was emptied by a service provider; whether the pit was emptied using recommended practices; if the pit was pierced; whether the pit was left disconnected for a sufficient period of time before being emptied; and whether FS was disposed of in a recommended location and using recommended methods.
2. The Treatment Practices Index describes whether households treated FS stored in their ADP as recommended, including whether the household treated the new pit with lime before switching pits; whether lime treatment was performed by a service provider; and whether any product was used to treat waste in the pit before or after emptying.
3. The Switching Practices Index describes whether households switched their toilet from one pit to another as recommended, including whether the old pit was emptied after the household changed their connection to the new pit; whether the old pit was reconnected only after being disconnected for at least two years; whether the connection was changed by a service provider; and whether the old pit was disconnected from the toilet after switching pits.
4. The Sanitation Knowledge Index describes households’ understanding of recommended ADP practices, including whether the household understands that sanitation affects public health; and that disposing of FS into a body of water or onto a field is unsafe.
5. The Sanitation Attitudes Index describes households’ perceptions of the importance of following recommended ADP practices, including whether the household feels that treating FS is important; that safety is a priority when disposing of FS; and that it is appropriate for a household member to empty a pit with untreated FS.

The survey questions used to create each index are shown in S2 Table, and additional details are shown in S3 Table. Each survey question that was used to calculate each index was selected for that index based on decades of existing literature and iDE’s knowledge of the local context in rural Cambodia [21,27,33,34]. Considering the Emptying Index as an example, all survey questions used to calculate this index specifically addressed the emptying-related topics listed above, which had been reported in the literature to affect or be related to pit emptying. For example, the question “Did the household pierce their pit?” was included in the Emptying Index because many past studies have shown that pit piercing is a method commonly used by rural Cambodian households to keep their pits from filling and thereby delay emptying. Other indices were constructed similarly, and additional rationale for including each survey question in their respective index is included in S3 Table. Survey questions that had not been specifically related to an index in the literature was not included in the calculation of an index. Also, survey responses that showed correlation above 0.2 (e.g., poverty and education) were not included in any index to avoid collinearity.

To calculate each index, the responses to each question were scored on a three-point numeric scale that categorized the related practice, knowledge, or attitude as “contrary to recommended ADP practices” (0), “neither contrary to nor following recommended ADP practices” (1), and “following recommended ADP practices” (2). For questions with binary response options, responses were scored as “contrary to recommended ADP practices” (0) or “following recommended ADP practices” (2). The final score for each index was calculated as the average score for each response, which also standardized the indices. Possible scores on each index ranged from 0 to 2.

With no strong rationale or data-driven evidence to suggest that certain survey questions were inherently more important to their respective indices than others, the scores for each survey question were not weighted when creating each index to maintain transparency and objectivity of the resulting composite indices, as described in the literature [35–37].

Each index was used as a dependent variable in generalized linear regression models coded in R (version 4.2.3) to describe the relationship between households’ characteristics and these aspects of ADP operation. Household characteristics included in these models were geographic location, flood-proneness, education, poverty status, education, and pit overflow frequency (the number of times that FS exits a pit untreated) [32,38].

Study population

Approximately two thirds of households (540) were located in an area identified as flood-prone. The majority of study communities were located in close proximity to water bodies (e.g., lakes and rivers), which is common in rural Cambodia and highlights the ubiquity of flood-proneness for the majority of ADP customers [12]. Nearly two thirds of respondents (63%) were women, which likely occurred because men were more likely to be working outside of the household when surveying occurred [27,33]. Only 6% of households in the study population were identified as poor (either IDPoor 1 or IDPoor 2) [32], and two thirds of households reported that at least one household member had completed secondary school. More households were located in Svay Rieng and Prey Veng provinces, and the fewest were located in Kandal province.

The latrines of households in the study population were used by an average of 8.2 people daily. A majority of households did not share their latrine with anyone outside of their household (63%), and the average number of latrine users within these households was 6.2. The remaining 37% of households shared their latrine with, on average, 5.3 individuals from outside of the household. Prior sanitation ownership was high, with nearly all households having at least one pit prior to ADP installation (99%).

More details about the study population are available in Supplementary Information S3 File.

Household perceptions and knowledge of ADPs

While nearly four-in-five households valued the advantages provided by ADPs compared to a single-pit latrine (78%), which included their reduced cost and required land area over time (23%); their doubled FS storage capacity (16%); and continuous latrine use (10%), only 13% specifically valued the FS treatment provided by ADPs. This finding is supported by low overall levels of household knowledge about ADPs: only one in seven households understood that they had to treat each pit of FS with lime once it fills, and only one in eleven households understood that treated FS could be emptied safely after two years of storage treatment. Nevertheless, most households perceived emptying untreated FS to be unsafe (82%) and considered FS treatment and safe disposal locations to be important and to foster good relationships with neighbors (97%). Thus, there is a disconnect between households’ valuation of ADPs when it comes to their attitudes around SMS and how they actually operate their ADPs to achieve sustained SMS.

Household ADP practices compared to recommended ADP practices

No households followed all recommended ADP practices. With 64% emptying their pits themselves, no household waited the recommended two years to empty their pit undergoing treatment, risking direct exposure to untreated FS; only one in eleven waited longer than a year and a half; and six in ten waited less than one year. Regarding lime treatment, only 8% of pits were treated with lime by a trained service provider after switching pits. Although another 6% of pits were treated with lime by a household member, households were advised to contract a trained service provider to do this and therefore not trained to treat FS with lime and likely did not perform this process correctly. The most common reasons for not treating FS with lime were a lack of required materials, primarily lime (33%), and a lack of available service providers (23%).

Pit-switching practices overall did not follow recommended practices and varied widely: only 14% of households switched pits after emptying their disconnected pit; other households performed a variety of other practices that were either unsafe or did not take advantage of the ADP’s design, including emptying their connected pit immediately (not after storage treatment) themselves (31%) or hiring a service provider to empty it (24%); piercing their pit to maintain latrine functionality (15%); or switching pits without emptying the disconnected pit (14%) as the disconnected pit typically had additional capacity available given normal dewatering that occurred after it was disconnected. Switching pits without emptying the disconnected pit is unsafe because it exposes the FS undergoing treatment to untreated FS, effectively negating all treatment in the ADP [28].

Factors affecting households’ ADP practices and sanitation knowledge and attitudes

Poor sanitation knowledge often precludes performing safe sanitation practices [2,27], providing one potential explanation for the unsafe household behaviors observed in the sample. Households’ performance of recommended ADP practices and their sanitation knowledge and attitudes also vary due to many other factors. For example, the literature highlights that convenience and cost avoidance are strong motivators for households when managing FS [27,39,40]. For example, most households choosing to either 1) empty their connected pit immediately upon filling without any FS treatment, or 2) piercing their pit to prevent it from filling, exemplify practices of convenience and cost avoidance because they can be performed by households without a service provider. Lastly, households may not follow recommended practices because these practices are simply difficult to understand or implement.

Disaggregating household characteristics by ADP index allows us to identify what factors affect each step in managing FS with an ADP (Table 1 and S4–S8 Tables).

Download:

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

Table 1. Effects of household characteristics on ADP indices describing ADP practices and sanitation knowledge and attitudes^a.

https://doi.org/10.1371/journal.pone.0332118.t001

Completing formal education was associated with less frequent performance of recommended ADP emptying and pit switching as well as worse sanitation knowledge. Households with a member that had completed formal primary and secondary education surprisingly performed recommended emptying practices less frequently; this result disagrees with the literature, which reports that more education, particularly attaining at least a primary education, is associated with performing safer sanitation practices [34]. This discontinuity can likely be explained by the different methods used in this study (an index that considers multiple facets of SMS and removes confounders) compared to that in the literature (correlations between individual metrics). Also, households with no formal education may only learn about sanitation and associated topics (e.g., germs, public health) from the recommended ADP practices they are taught when their ADP is installed, reducing any possible confusion or conflict with other previous knowledge sources. The completion of formal education in rural Cambodia is not universal; those who do complete schooling may be more confident in thinking for themselves and thus inclined to perform sanitation process, like pit switching, themselves. This behavioral construct is described in the literature by the concept of perceived control [41] and may help explain why formal education and sanitation knowledge correlate with lower levels of recommended ADP emptying and pit-switching practices. Similarly, having vocational education was associated with a slight decrease in recommended switching practices.

Poverty was associated with more frequent performance of recommended treatment practices and better sanitation knowledge, but less frequent performance of recommended switching practices. Additional research is needed to explain these associations.

A higher frequency of pit-overflow events was associated with more frequent performance of recommended treatment practices but less frequent performance of recommended emptying practices and worse sanitation knowledge. First, the connection between more frequent pit-overflow events and worse sanitation knowledge is self-evident: households with poor sanitation knowledge are more likely to not empty their pit before it overflows because they likely do not understand the effects of such inaction [5,21]. Considering emptying practices, pits commonly overflow in rural Cambodia due to rainfall and flooding [20,42]; as a result, households frequently empty their pits to maintain latrine functionality by piercing them [43], which is not a recommended ADP emptying practice. Also, pit-overflow events require households to manage FS, regardless of the practice used; thus, as the literature states, households under these climatic conditions are likely to become more comfortable and familiar with managing FS, which can encourage them to perform whatever practices are most convenient or cost-avoidant, which are frequently unsafe [44]. Lastly, the increase in performance of recommended treatment practices with increasing pit-overflow events may occur due to households’ well-documented desire to “kill meruk” (meaning desire to reduce pathogen concentrations in FS) and prevent it from entering the environment, which likely leads them to treat FS following recommended practices given the number of times they must manage these events [27].

Flood proneness was associated with less frequent performance of recommended treatment practices and worse sanitation attitudes. These results support the literature, which states that flood proneness increases challenges related to latrine functionality and achieving SMS, and is typically associated with unsafe FSM practices [16].

Households in different provinces tended to operate their ADPs in different ways. For example, households in Prey Veng and Svay Rieng followed recommended emptying practices less frequently and had worse sanitation attitudes compared to the reference province Kampong Thom, while households in Siem Reap had more sanitation knowledge than the reference province Kampong Thom. Supporting the literature [21,45], these results highlight that regional variations in ADP practices and sanitation knowledge and attitudes are large and must be considered when developing ADP educational materials and targeting related behavior change campaigns.

Limitations

The results of this study are subject to some important limitations. Social desirability bias likely occurred during surveying. Particularly given questions about sensitive or taboo topics like sanitation, respondents tend to provide answers that they view as more socially acceptable but perhaps less accurate. This bias was mitigated by a brief discussion between enumerators and participants about why talking about such topics is important. All participants were also reminded that they did not have to give a response to any question.

Similarly, recall bias, which describes how respondents may remember past events inaccurately, likely affected responses in this study, particularly regarding whether and how lime was applied to a pit nearly two years prior. Recall bias was minimized by shortening recall periods, using cued recalls, and using landmark events (e.g., the rainy season) instead of calendar dates when possible throughout the survey.

The well-documented lack of service providers in rural Cambodia likely also affected responses to questions that were used to calculate the Emptying Practices index [21,33,46]. For example, households likely emptied their pits themselves (i.e., did not hire a trained service provider) with buckets or pumps (i.e., not with a safer vacuum truck) and pierced their pits more frequently due to a lack of ability to hire a trained service provider to empty their pit using safer methods. Thus, the availability of service providers, which varies across different districts and provinces, likely affected the results of this index and the results reported with this index [33].

Lastly, a systematic behavior bias was introduced into the study design by the sampling frame. The selected sample specifically included only ADP customers who had owned their ADPs between two and four years, and who had managed FS in some way (e.g., emptied their pit, switched their pit). Therefore, this study may describe a group of people that are different from the typical Cambodian rural population. Household demographics and an appropriate sample size and study design were used to ensure that the sample population matched the target population of all rural households in Cambodia; however, unmeasured or unknown data may have introduced a systematic bias that cannot be detected.