Study design and setting

We conducted a community-based cross-sectional study using structured interviews and non-participatory observations among households in Korail Slum, Dhaka, Bangladesh, during normal and water-logging situations. Data collection took place between 15–20 November 2024. Korail Slum is one of the largest informal settlements in Dhaka, spans approximately 100 acres in the middle of the city (Fig 1), and is home to around 16,500 households and 151,500 people [18]. The slum is characterized by overcrowded housing, narrow roads, and limited infrastructure. We selected the Korail Slum due to the geographic location in the middle of Dhaka which eases travel and access, the large population, and the vulnerability of the slum to water-logging.

Download:

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

Fig 1. Approximate outline of Korail Slum area within the city of Dhaka, Bangladesh – the yellow line designates the Slum’s boundaries (image modified from Google Map).

Footnotes: Dhaka city map modified from an image titled Dhaka City Map 2025 by Ahnaf Tahmid Manan – Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=163166332; Inset image with the Slum’s boundaries modified from Google Map by the authors.

https://doi.org/10.1371/journal.pone.0332534.g001

Study participants and sample size calculation

The study population included adult residents of Korail with the following inclusion criteria: 1) Age 18 years or older; 2) Having resided in Korail for one year or longer; 3) Able to communicate in Bengali. We excluded individuals who self-reported illnesses or other impairments that they deemed would prevent them from participating fully in the interview.

We calculated the sample size for the survey based on the primary objective of describing the proportion of Korail Slum residents with access to WASH facilities. As no previous data on access to basic WASH facilities in Korail slum existed, we assumed that the proportion of the urban poor with access to WASH facilities in Korail Slums was the same as that in a previous study among resource-poor population in Bangladesh (27%, or p = 0.27) [19]. At 95% confidence and 5% arbitrary margin of error, we performed sample size calculation using the epicalc package in R [20] and obtained the sample size of 303 residents of Korail slum. We assumed that an arbitrary 25% of potential participants would refuse to participate in the study, thus we adjusted the sample size to 404 residents.

Sampling method

We used systematic random sampling to select the households for the survey. The sampling frame for the study was derived from the Urban Health and Demographic Surveillance Systems (UHDSS) operated in selected slums of Dhaka and Gazipur, including Korail Slum. We used the 2024 version of the UHDSS, which included 16,536 households with contact information [18]. Our sample size was 404 residents, thus the sampling interval was 16536/404 ≈ 38 households. We approached the adult household members, provided information about the study, and invited the household member to participate. If a selected household had more than one eligible member, we used the “next birthday” method to identify the potential participant.

Study instruments

We collected data by face-to-face interviews and rapid observation of WASH infrastructures in the participants’ household premises. Our study instrument was a structured questionnaire hosted on the web-based KoboToolbox platform. We used smartphones with either a web browser or the KoboCollect application for data collection.

Initially, we developed an English questionnaire adapted from previous studies [21] and the WHO/UNICEF Joint Monitoring Programme (JMP) [22]. The lead author (MR) translated the questionnaire from English to Bengali, and another colleague in Bangladesh back-translated the questionnaire from Bengali to English. The lead author (MR) compared the original and the back-translated English versions, identified discrepancies, and made corrections to the corresponding parts in the Bengali translation. We subsequently finalized the draft questionnaire. We then pilot-tested the draft Bengali questionnaire in Mirpur Slum, Dhaka, to evaluate its phrasing, structure, length, and sequencing. We made revisions based on feedback from the pilot-testing session, and finalized the questionnaire for data collection.

Study variables: Access to water, access to sanitation facility, and hygiene behaviors at water-logging and normal periods

In this study, the term “normal period” refers to the period in which the participant resided in Korail Slum but did not experience water-logging, and the term “water-logging period” refers to the period in which the participant resided in Korail Slum and experienced water-logging. Our operational definition of the term “water-logging” was “inundation of at least 25 cm in depth lasting 10 hours or more.” We trained our research assistants to explain this definition in a uniform manner to all participants when conducting interviews about experiences during water-logging periods. We asked participants to describe their level of access to WASH during the most recent water-logging periods within the past 12 months that the participants experienced without giving the specific dates.

Measurement of socioeconomic status

We included questions regarding the participants’ education, income, and asset ownership to measure socioeconomic status among the study participants. The investigators dichotomized education into two categories (secondary education or higher vs. primary education or less), income (more than 20,000 BDT per month vs. 20,000 BDT or less), and availability of various household items. The household income of 20,000 BDT per month was approximately half of the average household income in urban areas of Bangladesh in 2022 [24], and we used the amount as a proxy cut-off point for being low-income in the slum setting. We then retained characteristics and assets with adequate heterogeneity of the distribution (i.e., more than 10% or less than 90% of all participants reported having the characteristic or asset) for Principal Component Analysis. The investigators then followed the instructions for Principal Component Analysis (PCA) [25] and obtained the factor scores. We then performed descriptive statistical analysis on the factor scores and used the 33rd and 67th percentiles as the cut-off points for the socioeconomic tertiles.

Data collection

We recruited research assistants for this study from social and health sciences students and graduates from local universities. The research assistants underwent three days of training on research protocols and interview techniques, and a field practice session in an area outside Korail Slum with similar demographic and socioeconomic characteristics.

After receiving ethical approval from the Bangladesh Medical Research Council (BMRC), we sought permission from community leaders, then we asked for assistance from local community health workers to escort the research assistants to the study households. Before recruiting each participant, the research assistants would inform the participant about the study’s purpose and procedures and ensure understanding regarding the voluntary nature of participation and confidentiality. We asked for verbal informed consent in lieu of written informed consent in order to accommodate to the context of Korail Slum where literacy level among the residents remained low.

After receiving verbal informed consent, research assistants then started the face-to-face interviews. After completing the face-to-face interview, research assistants then asked the participant for permission to observe WASH facilities used by the participant and their household members. All data were securely stored in an encrypted system with password protection to ensure confidentiality and data security. Data collection took place between 15–20 November 2024.

Data management

We did not enter any personally identifiable data into the system to ensure the confidentiality of the study participants. We conducted regular checks to identify and resolve data-related issues (e.g., missing values, duplicated entries). We also performed routine data cleaning and prepared the final dataset for statistical analysis.

Data analysis

We analyzed the study data using descriptive statistics to summarize self-reported hand hygiene behaviors and observed household wash conditions. Additionally, we performed cross-tabulations to compare self-reported handwashing behaviors with the observed availability of water and soap at household locations. In the cross-tabulation, we excluded all households where the participants reported no experience of water-logging within 12 months prior to the survey. We performed all statistical analyses using R software.

Human research ethics

We requested and received ethical approval for this study from the Bangladesh Medical Research Council (BMRC) (Ref: BMRC/NREC/2022–2025/561). Due to the low literacy levels among the residents of Korail Slum, we requested and received approval for a waiver of written informed consent and the use of only verbal informed consent.

We obtained verbal informed consent from all participants before initiating data collection. This process included reading a consent script aloud. The consent script included a summary of the purpose of the study, procedures, risks, benefits, and confidentiality. We gave the participants an opportunity to ask questions before verbally agreeing to participate or declining participation. Our research assistants documented the verbal consent directly in the electronic questionnaire form on KoboCollect. The form required an affirmative response regarding verbal consent to display the data collection questionnaire. During the information and informed consent process, our research assistants periodically reminded the participants that they were allowed to stop the interview or ask the investigation team to stop observing household conditions at any time. We instructed the research assistants to stop data collection (i.e., withdraw study participants) in case of emergency or when they deemed data collection to disrupt necessary activities within the household.

Changes in WASH access and behaviours during water-logging

All except one of our participants reported having access to Basic water facility. In other words, in Korail slum, there was practically universal access to a water facility at the Basic level according to the Joint Monitoring Program (JMP) classification. Similarly, with regard to sanitation, no participant reported using unimproved latrines or engaging in open defecation. The absence of open defecation suggested that the prevalence of open defecation in Bangladesh continued to decline after the reduction to one percent in 2015 [26]. These major shifts are possibly attributable to significant investments in water and sanitation infrastructures and behavior change campaigns. However, the validity of this self-reported absence of open defecation warrants scrutiny. Social desirability bias may have influenced responses, as participants may have been embarrassed to admit engaging in open defecation after being exposed to widespread sanitation-focused awareness campaigns [27]. The majority of households had limited sanitation during normal periods due to the act of sharing their improved latrine with people from other households. However, a small minority (approximately four percent) of participants from these households reported having access to basic (non-shared) sanitation facilities during water-logging periods. We do not know what accounted for this shift, although a previous study reported that slum residents would relocate housing within the same area during water-logging [28], such as with family members, and relocated residence might have non-shared toilet. Future studies should consider asking questions regarding temporary relocation during water-logging periods. One additional issue in our data is that we did not collect information regarding water and sanitation related behaviors (e.g., the frequency of drawing water or using sanitation facilities). Future studies should consider including behavioral information in order to understand water and sanitation access in a more complete manner.

The minimal variation across socioeconomic tertiles suggests that water access is robust to smaller disruptions, such as the common minor floods (i.e., water-logging). Our findings differed from previous studies, which found that slum residents resorted to open defecation during water logging as latrines became inaccessible [29], and that low socioeconomic status (SES) correlates with reduced resilience of the sanitation infrastructures [3]. There were no extreme weather events in Dhaka in the 12 months before the study period, thus water-logging events in our study were not severe disasters. In that regard, the stability of water and sanitation access during water-logging in our study findings does not imply resilience during catastrophic events or robustness against long-term climate impacts, which severely impact marginalized populations [30].

With regard to access to hygiene facilities, households with basic hygiene during normal periods showed robustness during water-logging periods, whereas some households with limited or no facilities during normal periods had interestingly higher levels of access during water-logging periods. The upward shift could be attributed in part to either relocation [28] or distribution of soap for hand hygiene by civil societies during events [31]. On a related note, hand hygiene behaviors may need improvement. Less than one-quarter of the participants reported always washing their hands with water and soap after toilet use, which implied that more than three-quarters did not. Those who did not always wash their hands during the normal period were unlikely to wash their hands frequently during water-logging periods. There was little upward shift and substantial downward shift (i.e., regression) in hand hygiene behavior during normal vs. water logging periods. The reason for the shift is unclear, but it is likely that household members may have felt the need to conserve water for other purposes during water-logging periods, hence the lower prevalence. The findings highlighted the difference between access to hand hygiene facilities and materials vs. hand hygiene behaviors [32]. Community-level behavioral interventions to promote handwashing [32,33] could offer a potential solution to increase handwashing with soap. The presence of water and soap at the majority of households was nonetheless a positive development, as members of households with available water and soap were more likely to report handwashing with water and soap [34].

Socioeconomic disparities in WASH access during climate stress

The extent to which differences in WASH access during normal vs. water-logging period varied by socioeconomic status was unclear. Access to water and sanitation was practically universal (as mentioned above), which precluded assessments in these domains. Those in the second tertile of wealth were most likely to report differences between normal vs. water-logging periods regarding hand hygiene behaviors. We do not have a clear explanation or theory regarding this pattern of variation. However, the prevalence of reporting regular handwashing with water and soap at both normal and water-logging periods was very low, thus we could not rule out chance as the best explanation for the observed heterogeneity. Future studies [35] should consider replication of the current study’s findings and use other study methods (e.g., structured observation and in-depth interviews) to understand the mechanism of such variations (if they indeed exist) more comprehensively.

Strengths and limitations

The strength of this study was that we used validated tools from JMP, thus minimizing issues with using unvalidated instruments and enabling comparison of the findings with those from other countries and settings. Additionally, rapid observation of water and soap availability at participants’ primary handwashing stations provided a more comprehensive understanding of hygiene in Korail Slum. However, our study also contained a number of limitations. Firstly, as we collected the study data during daytime, most of our participants were stay-at-home mothers and housewives. Women and caregivers are generally more aware of hygiene than men [36], and our sampling method might have limited the ability to generalize the study findings to male residents of Korail. Secondly, hand hygiene behaviors are known to be over-reported due to social desirability and self-serving tendencies [37]. Thus, the potential for information bias is potentially non-negligible in the study findings. Thirdly, we collected the study data in November 2024. Given the rapid on-going changes in Bangladesh socioeconomic and political condition, the findings of this study may not be generalizable to other periods and settings. Fourthly, we only observed the availability of handwashing materials but not handwashing behavior due to lack of time and resources, which limited the scope of our study findings.