Can improved handwashing infrastructure and soap change behavior? findings from a randomized control study in the Dadaab refugee camp in Kenya

Abstract

Background

Globally, around 40% of people lack access to handwashing facilities with soap and water, a proportion likely higher in crisis-affected settings such as refugee camps. This increases vulnerability to hygiene-related diseases including diarrhea and respiratory infections. This study assessed whether improved access to handwashing facilities (Jengu) increases handwashing with soap (HWWS) among refugees in Dadaab camp, Kenya. Jengu is a foot-operated handwashing facility which is designed to integrate water storage, soap placement, and drainage in a compact structure suitable for household use in resource-constrained environments.

Methods

A randomized control study was conducted in Dadaab refugee camp in Kenya between June 2023 to July 2024. The sample size was 300 households across nine camp sections. Households were randomly assigned to receive either an intervention group receiving a Jengu handwashing facility with a 20-L jerry can and regular soap supply (n = 150) or a control group receiving soap provision only (n = 150). Soap was distributed bi-monthly to both groups. Data were collected through structured observations and household surveys at baseline, one-month post-intervention, and endline. Surveys assessed sociodemographic factors, knowledge, attitudes, and practices, while observations captured HWWS behavior at predefined critical hygiene moments. Descriptive statistics summarized baseline data. Logistic regression and a difference-in-differences approach were used to evaluate predictors and intervention effects.

Results

Among respondents, 75.4% were female (mean age 42 ± 12.7 years), 62.3% of household heads had no formal education. Across all rounds, 4,323 critical handwashing events were observed, of which 38.5% involved HWWS. HWWS prevalence rose from 28.4% at baseline to 61.0% one-month post-intervention, then declined to 32.4% at endline, remaining above baseline. Total HWWS moments nearly doubled (350 to 597), with the largest relative increase before child feeding (sixfold rise). HWWS before eating was significantly greater in the intervention group at endline (OR = 1.77, 95% CI: 1.13–2.77, p = 0.013). Soap and water availability were key predictors, HWWS odds were 2.53 times higher with soap (95% CI: 2.08–3.06, p < 0.001) and 1.23 times higher with water (95% CI: 1.02–1.49, p = 0.027). Compared to baseline, HWWS odds tripled at one month(OR = 3.05, 95% CI: 2.47–3.73, p < 0.001) and remained elevated at endline (OR = 1.20, 95% CI: 1.01–1.42, p = 0.037).

Conclusions

Enhanced handwashing facilities, consistent soap provision, and improved water access substantially increased HWWS in this refugee setting. Soap availability was critical for sustaining behaviors, while the Jengu facility prompted short-term gains. Integrated WASH interventions combining infrastructure, supply continuity, and behavior change strategies are essential for lasting hygiene improvements in humanitarian contexts.

Background

Crisis-affected populations in refugee and displacement settings face a disproportionately high burden of communicable diseases, particularly acute respiratory infections (ARI) and gastrointestinal (GI) illnesses. Diarrhoeal disease and pneumonia remain leading causes of morbidity and mortality among children under five in humanitarian contexts. Globally, diarrhoeal diseases account for an estimated 1.7 billion cases annually among children under five [1], while respiratory infections are among the most frequently reported conditions in displaced populations and refugee camp health facilities [2]. These risks are exacerbated by overcrowding, inadequate ventilation, shared sanitation facilities, intermittent water supply, and fragile health systems conditions that facilitate the rapid transmission of respiratory and enteric pathogens. In protracted displacement settings, where such vulnerabilities persist for years, cumulative exposure risks further undermine disease prevention efforts [3–5].

Diarrhoeal disease and pneumonia remain among the leading causes of morbidity and mortality among children under five years in humanitarian settings. Evidence from low- and middle-income countries indicates that improved hand hygiene alone can reduce diarrhoeal incidence by up to 48% and respiratory infections by approximately 20–25% [6, 7]. More recently, Ross et al. (2023), confirmed through a large systematic review and meta-analysis that handwashing with soap significantly reduces the risk of acute respiratory infections, reinforcing its central role in infection prevention in resource-constrained settings [8].

To further justify the importance of improving hand hygiene in displacement settings, it is important to recognize the high burden of acute respiratory and gastrointestinal illnesses in refugee camps, including Dadaab. Humanitarian health assessments consistently identify ARI and diarrhoeal disease as leading causes of morbidity and mortality in refugee and camp settings, often accounting for a large share of clinic visits and hospitalizations alongside malnutrition and other epidemic-prone conditions. Historical analyses have shown that ARI, diarrhoea, measles, and malaria can collectively account for 60–95% of deaths among refugees when basic services are disrupted, underscoring the vulnerability of displaced populations to infectious disease transmission in camp environments [9].

Global analyses further demonstrate that enteric infections remain endemic in displacement camps: a systematic review of IDP and refugee settings reported a pooled diarrhoeal prevalence of about 29% (95% CI: 24–33%), with prevalence as high as 32% in refugee camps specifically [10]. These high prevalence figures for ARI and diarrhoea among displaced populations highlight the potential scale of preventable disease that could be addressed by effective hygiene interventions.

Promoting hygiene practices such as HWWS can substantially reduce the burden of these diseases [8]. In humanitarian contexts, improving access to handwashing facilities not only enables better hygiene but also fosters sustained behavior change [11]. Consequently, water, sanitation, and hygiene (WASH) interventions remain vital components of humanitarian response strategies [12].

In particular, improving hand hygiene behavior through HWWS is among the most effective and cost-efficient public health interventions for preventing major infectious diseases. It disrupts the transmission of pathogens responsible for diarrheal diseases, one of the leading causes of morbidity and mortality in refugee settings [13, 14]. Consistent HWWS can reduce the risk of diarrhea by up to 48% and respiratory illness by up to 23% [6, 15].

Dadaab refugee camp, one of the largest and longest-standing refugee settlements globally, exemplifies these challenges. Hosting over 380,000 refugees and asylum seekers, the camp has experienced decades of cyclical funding shortfalls, population flux, and infrastructure strain. Although hygiene promotion and soap distribution are routine components of humanitarian programming, observational evidence indicates that effective HWWS remains suboptimal [16]. One of the strongest determinants of handwashing behavior is the availability of a handwashing facility equipped with water, soap, and other cleansing agents [17]. These facilities serve as visual cues that encourage handwashing at critical moments. Individuals with access to designated handwashing stations are approximately twice as likely to wash their hands with soap than those without [18].

Globally, approximately 40% of the population lacks access to handwashing facilities with water and soap at home, and 18% have no facility at all [8]. In eastern and southern Africa, only 24% of households have a designated HWWS facility [19]. In humanitarian settings, access and use remain particularly limited. For example, a recent study conducted in refugee camps in Kenya (Dadaab) and Uganda (Kyangwali) reported that self-reported HWWS rates were 48.1% and 31.6%, respectively; however, when directly observed, the rates decreased to 33.4% in Dadaab and 12.1% in Kyangwali [20]. Similarly, across refugee camps in Thailand, Ethiopia, and Kenya, soap was used in only approximately 30% of observed handwashing events, with just 20% of post latrine handwashing involving both hands with soap [16]. In Maban County, South Sudan, despite high reported soap availability and regular hygiene promotion, the observed HWWS remained low at only 7% before eating, 23% before food preparation, and 46% after toilet use [21].

This study aims to assess whether providing refugee populations with increased access to desirable and durable handwashing facilities the Jengu handwashing facility would lead to increased handwashing behavior. Additional information on the design and functionality of the Jengu system can be found through the organization that developed the facility, Arup (https://jengu.org.uk/). The Jengu facility was developed through participatory design processes involving local communities, engineers, humanitarian actors, and researchers to ensure its technical feasibility and social acceptability. The collaborative approach prioritized the use of affordable, locally available materials and incorporated design features responsive to challenges in low-resource environments.

A defining feature of the Jengu facility is its hands-free operation, which is activated via a foot pump, minimizes the risk of recontamination and distinguishes it from conventional gravity-fed systems. Unlike earlier designs that rely on elevated or suspended containers, Jengu’s manual pump mechanism eliminates the need for height or complex setups, increasing the ease of maintenance and adaptability across settings. Additional innovations include a reflective steel surface, which is chosen to serve both as a mirror and a visual reminder for handwashing.

Methodology

Study settings and population

The study was conducted in Ifo refugee camp, one of four camps within the Dadaab refugee complex in Garissa County, northeastern Kenya. Established in 1992 to host refugees fleeing conflict in neighboring Somalia, Dadaab has evolved into one of the world’s largest and most protracted refugee settlements [22]. The complex currently hosts approximately 460,000 registered refugees and asylum seekers, the majority of whom are Somali and predominantly Muslim [23]. The camps are coordinated by the United Nations High Commissioner for Refugees (UNHCR), with humanitarian partners delivering essential services. For example, the Kenya Red Cross supports health services and disaster risk reduction, Peace Winds Japan oversees water, sanitation, and hygiene (WASH) activities, and the World Food Programme (WFP) manages food distribution. The camps are located in a semi-arid environment characterized by high temperatures, limited water resources, and densely populated residential blocks, where households often share sanitation facilities and rely on communal water points. These conditions, combined with constrained infrastructure and limited resources, create environments conducive to the transmission of communicable diseases such as diarrhoeal illnesses and acute respiratory infections. In such contexts, improved water, sanitation, and hygiene (WASH) practices particularly handwashing with soap are critical for reducing the spread of infectious diseases.

The Ifo refugee camp is home to over 90,000 residents, the majority of whom (96%) identify as Muslim and primarily rely on humanitarian aid for their basic needs [24]. The population of the camp fluctuates due to new periodic arrivals. Administratively, the camp is organized into nine sections (A, B, C, D, E, F, G, N, and S), and each section is further divided into blocks (e.g., A1, B1, C1), totaling approximately 111 blocks with varying numbers of households. Residents typically live in temporary structures such as mud-walled houses. Given the camp’s status as a temporary settlement, essential services and subsistence support are largely provided by humanitarian agencies.

At baseline, WASH conditions reflected typical infrastructure in protracted refugee camp settings. Households primarily accessed water from communal boreholes and tap stands managed by humanitarian agencies, with water collected in jerry cans for domestic use. Sanitation facilities consisted mainly of shared communal pit latrines located within residential blocks and used by multiple households.

Dedicated household handwashing facilities were uncommon. Handwashing was typically performed using water stored within household compounds or near cooking areas, and stations were rarely positioned near latrines. Soap was often stored inside the household rather than at a fixed handwashing point. This separation between sanitation facilities, water storage, and soap likely contributed to inconsistent handwashing with soap at critical moments prior to the intervention.

Study design, sampling and data collection

In this randomized controlled study the sample size was calculated via the formula N = Z²P(1 − P)/d², assuming a 23% prevalence of HWWS at critical moments [25], a 95% confidence level (Z = 1.96), and a 5% margin of error. This yielded a sample size of 272, which was adjusted by 10% to account for nonresponses, resulting in a final target of 300 households.

We applied proportionate allocation to select 300 households across the nine sections of the Ifo camp. A total of 20 blocks were purposely selected, with the number of blocks per section determined by the relative population size and number of households in each section. The selection of the blocks was conducted in consultation with local leaders to confirm both the availability of water and the presence of a mix of long-term and newly arrived refugees, thereby enhancing representativeness. Ensuring water access was essential in this study, as it aimed to evaluate the effectiveness of a fixed handwashing station (Jengu) with soap in improving and sustaining HWWS practices.

Proportionate sampling determined the number of households to be drawn from each block, whereas systematic random sampling guided household enrollment on the basis of spatial proximity. Within each selected block, every household had an equal probability of being chosen for participation. For the purposes of this study, a regular soap supply referred to the distribution of bar soap to participating households approximately every two months during the study period. Soap distribution was maintained for the entire duration of the study, from baseline through the 12-month follow-up, to support household hygiene practices and ensure that both groups had comparable access to soap while assessing the additional effect of the Jengu handwashing facility.

The selected households were then randomly assigned to either the intervention or control group via a lottery method, with 150 households allocated to each group. Households in the intervention group received a Jengu handwashing facility, a regular supply of soap throughout the study period, and a twenty-liter jerry can for water supply, whereas households in the control group received soap provision alone in the same quantity and distribution schedule. This study design therefore evaluated the incremental effect of the Jengu handwashing facility beyond soap provision alone.

Consenting households participated in both a household survey and structured observation sessions to assess hand washing practices. The survey, which lasted approximately 45 min per household, collected data on sociodemographic characteristics, handwashing knowledge, attitudes, and self-reported practices. Structured observations were typically conducted for 2–3 h, during morning hours, when several hygiene-related activities commonly occur within households. These included key critical handwashing moments such as after latrine use, before food preparation, and before eating. Conducting observations during this period increased the likelihood of capturing routine hygiene practices during daily household activities.

The data collection was carried out by trained enumerators recruited from the local community. The data collection team consisted of 18 enumerators (16 male and 2 female) who were trained on the study protocols and data collection procedures prior to field implementation. They work in pairs to increase data accuracy, ensure safety, and reduce observer bias and cognitive overload. During the structured observation sessions, a handwashing moment was defined as an observed opportunity for hand hygiene occurring at predefined critical times. Each occurrence of these activities was recorded as a handwashing opportunity, and observers documented whether handwashing occurred and whether it involved water only or water with soap, all members of the household were observed as engaged in their usual routines. The observers systematically recorded critical handwashing moments such as after using the toilet, before eating, after eating, before food preparation, before feeding a child, and after cleaning a child’s bottom. Focusing on whether handwashing occurred and how it was performed (e.g., with soap and water or water only), and the type of facility used.

To minimize the risk of reactivity (behavior change due to being observed), observers were positioned discreetly within the household compound, often in a corner or area where they could see daily activities without interrupting or engaging with the household members. Prior to the start of the observations, the participants were informed that the enumerators were there to better understand household routines, without specifying that handwashing behaviors were the main focus. This approach helped reduce social desirability bias and allowed for more natural behaviors to be recorded. Structured observations were conducted at three key time points: baseline (before the intervention), one month after the intervention, and at the end of the study (12 months post-intervention). The entire data collection phase spanned a 13-month period, from June 2023 to July 2024.

Data management and statistical analysis

Data were collected via the KoboCollect platform [26] and transmitted daily to a secure, cloud-based server to ensure data integrity and real-time monitoring. The data management procedures included regular data cleaning and consistency checks to ensure the completeness and accuracy of the dataset throughout the study period.

Data analysis was conducted via STATA version 17.1. Descriptive statistics were used to summarize the baseline sociodemographic characteristics of the respondents. These included variables such as marital status, educational attainment, religion, primary source of household income, duration of the respondent’s residence in the household, number of children under five years, and age distribution of the respondents. Frequencies and proportions were used to describe categorical variables.

The primary outcome of interest was the observed occurrence of handwashing with water and soap during critical hand hygiene moments. These moments were defined as follows: after using the toilet, before preparing food, before eating food, after eating, after cleaning a child’s bottom, and before feeding a child. Each observed handwashing opportunity constituted an event-level observation, which served as the main unit of analysis for estimating the prevalence of HWWS and examining associations with explanatory variables.

To explore associations between explanatory variables and the primary outcome, inferential statistical analyses were conducted. Specifically, random-effects logistic regression models were used to account for repeated observations from the same households across the three study periods (baseline, one-month post-intervention, and endline). In these models, the household was treated as the panel unit to account for potential correlation between repeated observations over time while examining the relationships between HWWS and key predictors such soap availability, water availability and study timepoints.

In addition, household-level summaries of handwashing frequency during structured observation periods were used for complementary analyses. These summaries aggregated the number of handwashing events observed within each household observation session.

To assess the effect of the intervention, a difference-in-differences (DiD) analytical approach was used to estimate changes in handwashing behavior over time between the intervention and control groups. For this analysis, the outcome variable was defined as the number of observed handwashing events during each household observation session, aggregated at the household level for each study period (baseline, one-month post-intervention, and endline). This variable, referred to as “HW times,” represented the household-level frequency of handwashing events observed during structured observation periods and was treated as a continuous outcome in the DiD regression model. The model therefore estimated the differential change in handwashing frequency between intervention and control households over time.

Although a baseline balance test was conducted to assess comparability between the two groups prior to the intervention, we acknowledge that baseline equivalence alone does not fully establish the assumptions underlying DiD models, such as parallel trends. The DiD results are therefore interpreted as complementary to the event-level regression analyses, which examined associations between handwashing with soap and key predictors across all observed handwashing opportunities.

To assess the effect of the intervention, a DiD analytical approach was employed to estimate changes in handwashing behavior between the intervention and control groups over time. This analysis used household-level summaries of observed handwashing events during each observation period, allowing comparison of trends between groups while accounting for repeated measurements collected from the same households throughout the study.

Although both groups received soap as part of ongoing hygiene promotion activities, the Jengu facility was introduced only at the intervention sites. The DiD model therefore accounted for shared exposures such as soap availability, enabling isolation of the incremental effect attributable specifically to the Jengu facility. Prior to the DiD analysis, a baseline balance test was conducted to assess the equivalence of key outcome variables between the two groups. The results indicated no significant differences at baseline, thereby strengthening the validity of the DiD estimates.

Outcome and explanatory variables

The main outcome variable of this study was the observed prevalence of handwashing with water and soap during critical hand hygiene moments. The explanatory variables influencing the handwashing time considered for this study were water availability, soap availability and the Jengu facility.

Ethical clearance

This study received ethical approval from the London School of Hygiene and Tropical Medicine Ethics Review Board and the AMREF Health Africa Ethics and Scientific Review Committee (AMREF ESRC P1267/2022). A national research permit was also obtained from the Kenya National Commission for Science, Technology and Innovation (NACOSTI/P/22/22140), alongside approvals from UNHCR and relevant county and sub-county authorities in Kenya. All participants were provided with information about the study’s purpose in a language they understood, and written informed consent was obtained from the household head or the next responsible adult on behalf of the household. To ensure confidentiality, no personal identifiable information was collected during data collection. Each household was assigned a unique identification code used throughout data collection and analysis. All electronic data were stored on password-protected devices and securely transferred to protected institutional servers accessible only to authorized members of the research team. The data was used solely for research purposes, and all study procedures adhered to established ethical guidelines for research involving human participants.

Results

Among the 300 households enrolled, 297 completed the study, with three lost to follow-up—two due to relocation from the camp and one due to withdrawal at the request of the household head. In total, 4,130 critical handwashing events and 894 surveys were conducted over the data collection period.

Across the study period, a total of 4,323 critical handwashing events were observed. Of these, 187 (4.3%) were missed handwashing moments where no handwashing occurred, whereas 4,130 (95.7%) were handwashing moments where hands were washed.

Baseline sociodemographic characteristics

The majority of the respondents (75.4%) were female. The age of the respondents ranged from 19 to 92 years, with a mean age of 42 years (SD = 12.71). Almost two-thirds of the respondents, heads of household (62.3%), had no formal education, with only 24.6% reporting having at least basic primary-level education. The majority (89.2%) practiced the Muslim religion. Over 90% of the respondents mainly relied on refugee rations, without their own reliable source of household income. The majority (94.7%) had lived in settlements for more than 3 years. Approximately one-third of the respondents (31.7%) reported having two children under 5 years of age (Table 1).

Table 1.

Baseline sociodemographic characteristics of the study population

Variable	Frequency (N = 297)	Percentage
Respondent gender
Female	224	75.42
Male	73	24.58
Age of the household head
19–29	47	15.82
30–39	103	34.68
40–49	71	23.91
50–59	43	14.48
60+	33	11.41
Household head education level
Not attended school	185	62.29
Basic primary education	73	24.58
Secondary education/O levels	33	11.11
College/University/Vocational training	6	2.02
Religion
Christianity	32	10.77
Muslim	265	89.23
Main household income source
Refugee rations/No income	289	97.31
Own Business (Merchant/Trader)	5	1.68
Casual Labor	2	0.67
Sale of Crops	1	0.34
Length of stay in the household
Less than one year	8	2.70
1 year- 3 years	8	2.69
Above 3 years	281	94.61
Number of households with children below 5 years
0	81	27.27
1	54	18.18
2	94	31.65
3	44	14.81
4 and above	24	8.09

Availability of soap and water at handwashing facilities

Overall, the availability of soap at or near handwashing facilities increased from 11.4% at baseline to 30.4% at endline. When the control and intervention groups were disaggregated, improvements were observed in both arms. In the intervention group, soap was present at 42.9% of handwashing stations at endline, up from 14.0% at baseline. In the control group, soap availability increased from 8.7% at baseline to 18.7% at endline.

Similarly, water availability at or near handwashing stations also improved over the study period, increasing from 13.0% at baseline to 34.6% at endline. In the intervention group, the proportion of handwashing stations with water increased from 18.4% at baseline to 47.3% at the end of the study. The control group also improved, with water availability increasing from 7.2% at baseline to 23.0% at the end of the study period (Table 2).

Table 2.

Soap and water availability across time points and groups

Soap availability	Baseline			One-month post intervention			Endline
	Control	Intervention	Total	Control	Intervention	Total	Control	Intervention	Total
No soap available in the household/compound	9 (4.6)	6 (2.9)	15 (3.7)	6 (1.7)	4 (0.9)	10 (1.8)	69 (18.3)	17 (4.9)	86 (11.8)
Soap available at or near the handwashing location	17 (8.7)	29 (14.0)	46 (11.4)	154 (43.6)	390 (91.3)	544 (69.0)	71 (18.71)	149 (42.9)	220 (30.4)
Soap available but not near the handwashing location	169 (86.7)	172 (83.1)	341 (84.9)	193 (54.7)	41 (9.6)	234 (29.7)	239 (63.1)	181 (52.2)	420 (57.8)
Total	195	207	402	353	435	788	379	347	726

Water availability	Baseline			1month post			Endline
	Control	Intervention	Total	Control	Intervention	Total	Control	Intervention	Total
No water available in the household/compound	2 (1.0)	6 (2.9)	8 (2.0)	1 (0.3)	2 (0.5)	3 (0.4)	17 (4.5)	8 (2.3)	25 (3.4)
Water available at or near the handwashing location	14 (7.2)	38 (18.4)	52 (13.0)	139 (39.4)	398 (91.5)	537 (68.2)	87 (23.0)	164 (47.3)	251 (34.6)
Water available but not near the handwashing location	179 (91.8)	163 (78.7)	342 (85.0)	213 (60.3)	35 (8.0)	248 (31.4)	275 (72.6)	175 (50.4)	450 (62)
Total	195	207	402	353	435	788	379	347	726

Observed and missed critical handwashing moments by study period

Across the study period, a total of 4,323 critical handwashing events were observed. Of these, 187 (4.3%) were missed handwashing moments where no handwashing occurred, whereas 4,130 (95.7%) were handwashing moments where hands were washed. The structured observation guide recorded whether handwashing occurred and how it was performed but did not systematically capture the reasons for missed handwashing events.

At baseline, 73 out of the 1,307 handwashing moments were missed, representing 5.6% missed opportunities. At one-month post-intervention, the number of missed handwashing events declined to 32 out of 1,086 (2.9% missed), indicating the lowest level of missed handwashing during the study. At the endline, the number of missed moments rose slightly to 88 out of 1,930 (4.6% missed), although this figure remained lower than that at the baseline (Table 3).

Table 3.

Observed and Missed Critical Handwashing Moments by Study Period

Critical handwashing moments	Baseline	One month	Endline	Total
Hand washing moments observed	1234	1054	1842	4130
Missed handwashing moments	73	32	88	187
Total	1307	1086	1930	4323

General handwashing moments across timepoints

Among all the handwashing moments observed, HWWS recorded 1,590 of these moments, representing a prevalence of 38.5% compared with 60.6% for handwashing with water alone. When the data were disaggregated by timepoint, the prevalence of hand rinsing with water only was 71.1% at baseline, 38.7% at one-month post-intervention and 66.1% at endline, whereas the prevalence of HWWS was 28.4% at baseline, 61.0% at one-month post-intervention, and 32.4% at endline.

Overall, the observed HWWS increased by 70% from baseline (350) to endline (597), with a peak at one-month post-intervention, followed by a slight decline that remained above baseline levels, showing sustained improvement (Table 4).

Table 4.

General handwashing times by timepoint

Action performed	Baseline	One-month post intervention	Endline	Total
Hands rinsed with water only	878 (71.1%)	408 (38.7%)	1218 (66.1%)	2504 (60.6%)
Hands washed with water and soap	350 (28.4%)	643 (61.0%)	597 (32.4%)	1590 (38.5%)
Hands washed with ash	4 (0.3%)	3 (0.3%)	12 (0.7%)	19 (0.5%)
Hands washed with sanitizer	2 (0.2%)	0 (0.0%)	15 (0.8%)	17 (0.4%)
Total	1234	1054	1842	4130

Handwashing with water and soap by time and study group

When the results are disaggregated by study group, a similar trend can be observed across both arms. In the intervention group, the number of observed handwashing moments at baseline was 180, which increased to 360 at one-month post-intervention, representing a 94.6% increase. This figure ultimately declined to 282, although it still reflected a 52.4% increase from baseline. In the control group, the number of baseline observations totaled 173 and increased to 283 at one-month post-intervention (a 63.6% increase). By end, the number had further risen to 315, representing an overall 82.1% increase from the baseline. These findings indicate a sharp initial rise in handwashing moments in both groups, particularly in the intervention group, followed by a subsequent decline in the intervention arm, whereas the control group continued to show gradual improvement over time.

At baseline, handwashing with water or soap was comparable between the control and intervention groups, accounting for 48.6% and 51.4% of the observed events, respectively (Table 5). One-month post-intervention, the proportion of handwashing events increased notably in the intervention group (56.0%) compared with the control group (44.0%), suggesting a short-term positive effect of the intervention. Overall, while both groups demonstrated fluctuations over time, the intervention group showed rapid improvement at the beginning of the intervention, followed by a modest decline by endline, whereas the performance of the control group improved steadily throughout the study period (Table 5).

Table 5.

Moments of handwashing with water and soap by time and group

Period	Baseline (n)	Baseline (%)	One-month Post (n)	One-month Post (%)	Endline (n)	Endline (%)
Control	170	48.6%	283	44.0%	315	52.8%
Intervention	180	51.4%	360	56.0%	282	47.2%
Total	350	100%	643	100%	597	100%

Critical handwashing with soap and water before and after intervention

From baseline to endline, the total number of observed HWWS moments increased across all six critical hand washing moments. The greatest relative increase was observed before a child was fed, when HWWS increased more than sixfold (11–68 events). Overall, the total HWWS nearly doubled, rising from 350 at baseline to 597 at the endline.

Both the intervention and control groups showed increases in HWWS from baseline to endline, with the greatest improvements observed after using the toilet and before feeding a child. At baseline, there were no significant differences between the two groups across any of the six critical moments, with odds ratios (ORs) close to unity (e.g., after using the toilet: OR = 1.01, 95% CI: 0.60–1.69, p = 0.970; before preparing food: OR = 0.50, 95% CI: 0.21–1.19, p = 0.116). At the end of the study, however, HWWS before eating was significantly greater in the intervention group than in the control group (OR = 1.77, 95% CI: 1.13–2.77, p = 0.013) (Table 6).

Table 6.

Critical handwashing with water and soap moments by period

Handwashing Moments	Baseline						Endline
	Control	Intervention	Total	OR	P value	95% CI	Control	Intervention	Total	OR	P value	95% CI
After cleaning a child’s bottom	22	29	51	1.3	0.402	[ 0.70–2.35]	41	38	79	1.04	0.869	[0.65–1.68]
After eating	9	14	23	1.36	0.593	[0.44–4.21]	22	13	35	0.66	0.272	[0 0.31- 1.39]
After using the toilet	60	64	124	1.01	0.97	[0.60–1.69]	113	101	214	0.99	0.989	[0 0.71 − 1.40]
Before eating	38	41	79	1.03	0.911	[0.59–1.82]	38	55	93	1.77	0.013	[1.13–2.77]
Before feeding a child	4	7	11	1.68	0.415	[0.49–5.84]	40	28	68	0.76	0.289	[0.45–1.26]
Before preparing food	37	25	62	0.5	0.116	[0.21–1.19]	61	47	108	0.83	0.404	[0 0.53 1.29]
Total	170	180	350				315	282	597

Association between soap and water availability and observed handwashing with soap

Our analysis demonstrated that both soap and water availability at or near the handwashing location were significantly associated with increased odds of observed HWWS in both the intervention and control groups. When soap was available, the odds of HWWS were 2.53 times greater than when soap was not available (OR = 2.53, 95% CI: 2.08–3.06, p < 0.001). Similarly, the availability of water was associated with a 1.23-fold greater odds of HWWS than when water was not available (OR = 1.23, 95% CI: 1.02–1.49, p = 0.027).

HWWS increased significantly over time. Compared with those at baseline, the odds of HWWS were 3.05 times greater at one-month post-intervention (OR = 3.05, 95% CI: 2.47–3.73, p < 0.001) and 1.20 times greater at the endline (OR = 1.20, 95% CI: 1.01–1.42, p = 0.037), indicating a sustained, although slightly diminished, effect over time. Similarly, with respect to water availability, the odds of HWWS were 4.52 times greater at one-month post-intervention (OR = 4.52, 95% CI: 3.68–5.56, p < 0.001) and 1.34 times greater at the endline (OR = 1.34, 95% CI: 1.12–1.59, p = 0.001), further supporting the sustained impact of the intervention on handwashing behavior Table 7.

Table 7.

Soap and water availability and observed handwashing with soap

Soap availability
Hands washed with water and soap		OR	Std.err	z	P value	95% CI
Soap available at or near the handwashing location	Soap not available	1	(base)
	Soap available	2.53	0.25	9.42	0.000	2.08–3.06
Baseline		1	(base)
One-month post intervention		3.05	0.32	10.57	0.000	2.47–3.73
Endline		1.20	0.10	2.09	0.037	1.01-1.42

Water availability
Hands washed with water and soap		OR	Std.err	z	P value	95% CI
Water available at or near the handwashing location	Water not available	1	(base)
	Water available	1.23	0.12	2.21	0.027	1.02–1.49
Baseline		1	(base)
One-month post intervention		4.52	0.48	14.38	0.000	3.68–5.56
Endline		1.34	0.12	3.29	0.001	1.12–1.59

Balance test of the hand washing moments (before intervention)

The balance test conducted before baseline revealed no significant difference in handwashing times between the treatment and control groups (t value = -0.55, p value = 0.578), suggesting that the groups were comparable prior to the intervention. The DiD approach is therefore validated, which implies that posttreatment differences are attributable to the intervention (Table 8).

Table 8.

Two-sample t test with equal variances

	obs1	obs2	Mean1	Mean2	dif	St Err	t value	p value
HW times by Treatm ~ 1	143	149	5.636	5.893	− 0.257	0.461	− 0.55	0.578

Difference in Difference Model Output

Table 9 below presents the DiD Model output. The DiD analysis, which evaluates the effect of the intervention over time, yields an estimate of -0.749 (p value = 0.212), indicating that the intervention did not significantly alter handwashing behavior between the control and treated groups.

Table 9.

Difference-in-Differences Model Output

Outcome variable	HW times	S D	|t|	P>|t|
Before
Control	5.636
Treated	5.893
Diff (T-C)	0.256	0.427	0.6	0.549
After
Control	8.447
Treated	7.954
Diff (T-C)	-0.49	0.421	1.17	0.242
Diff-in-Diff	-0.749	0.599	1.25	0.212

R-square: 0.10

* Means and standard errors are estimated via linear regression

**Inference: *** p < 0.01; ** p < 0.05; * p < 0.1

Discussion

This study assessed the effects of improving handwashing infrastructure and ensuring consistent soap availability on HWWS practices among refugees living in Dadaab, Kenya. Overall, HWWS moments increased substantially across all six critical handwashing occasions, with the most notable rise observed before a child was fed, where HWWS rose more than sixfold. Importantly, both the control and intervention groups demonstrated positive trends over time. These findings suggest that infrastructure improvements, coupled with the consistent provision of soap and safe water, can meaningfully influence hygiene practices in humanitarian contexts. The observed patterns also provide insight into how hygiene behaviors vary across daily activities; however, with reference to the moment-specific analyses we interpreted them as exploratory rather than causal estimates of intervention effects. Although handwashing with soap increased during the study period, improvements were observed in both the intervention and control groups, suggesting that soap availability played an important role in driving behavior change. Consistent with this pattern, the difference-in-differences analysis did not detect a statistically significant incremental effect of the Jengu facility beyond soap provision alone. Together, these findings highlight areas where hygiene interventions may influence behavior while underscoring the importance of enabling factors such as reliable soap availability, water access, and ongoing hygiene promotion. While improved handwashing infrastructure may enhance convenience and accessibility, consistent access to soap appears to remain a critical driver of hygiene behavior in humanitarian settings.

Another key observation across the study area was the persistent gap between HWWS and rinsing hands with water only. Although improvements were observed, rinsing with water alone remained more common than HWWS, which is consistent with the findings of previous studies across sub-Saharan Africa and globally [7, 18, 21, 25]. For example, Curtis et al. reported that rinsing hands with water was three times more common than HWWS in 11 countries [27]. This underscores the challenge of shifting behaviors from a widespread but less effective practice (rinsing with water) toward proper HWWS and highlights the need for interventions that emphasize both the health benefits and social value of using soap.

Soap availability was a decisive factor in sustaining HWWS behaviors. Our findings showed that soap availability was strongly associated with higher odds of handwashing with soap at critical moments. Households where soap was present at or near handwashing locations were more likely to demonstrate handwashing with soap during observed events, consistent with evidence from other studies highlighting soap access as an important enabling factor for HWWS [28, 29].

These findings have important implications for humanitarian WASH programming. Ensuring reliable access to soap may be one of the most effective strategies for improving handwashing behavior in resource-constrained settings. While hardware innovations such as improved handwashing facilities may enhance convenience and visibility, consistent soap availability appears to be a more critical determinant of sustained handwashing with soap.

In many low-income and crisis-affected settings, a limited or inconsistent soap supply has been reported as a major barrier, often resulting in the diversion of soap to other household uses, such as bathing and laundry [25, 30]. The consistent provision of soap throughout this intervention likely explains the observed improvements, underscoring the importance of supply-side interventions in contexts where affordability and access are constrained.

Water availability has also emerged as an important enabler of handwashing. Households with water at or near handwashing locations were more likely to practice HWWS, reflecting evidence from Bangladesh and other low-resource settings where proximity to reliable water sources was associated with increased handwashing frequency [31, 32]. Nonetheless, water alone was insufficient. Research has shown that improvements in water access yield greater impacts when combined with the presence of soap and functional handwashing infrastructure [33, 34]. This study’s findings reinforce the importance of integrated approaches that simultaneously address multiple enabling conditions for HWWS, particularly in fragile settings such as refugee camps [28, 29].

Another important finding relates to the sustainability of behavioral change. While HWWS rose sharply after the intervention was introduced, a partial decline was observed over time, although levels remained above baseline. This trajectory is consistent with other refugee camp studies where initial gains in HWWS diminished but did not revert to preintervention levels [21, 35]. Two temporal trends help explain this trajectory. The intervention group experienced a sharp early increase in HWWS, followed by a decline toward endline, possibly reflecting a novelty or early adoption effect in which newly introduced infrastructure temporarily boosts engagement before behaviors stabilize. In contrast, the control group showed a gradual but steady improvement, ultimately reaching levels comparable to the intervention group. As both groups received regular soap provision and were exposed to ongoing hygiene promotion activities, these factors likely contributed to sustained improvements in handwashing behavior across households. Together with the non-significant difference-in-differences findings, these patterns suggest that while infrastructure and supply may create an enabling foundation for change, consistent access to soap and continued behavioral reinforcement including health messaging, social norms, and motivational strategies may be more critical for sustaining handwashing behavior in humanitarian settings [11, 36].

Lastly, the timing of structured observations may also have influenced the observed patterns of handwashing behavior. Observations were conducted primarily during morning hours, when key hygiene activities such as after latrine use and before food preparation commonly occur. While this increased the likelihood of capturing critical handwashing moments, it may also have shaped the demographic composition of those observed. In many humanitarian settings, women are more likely to be present at home during daytime hours and often bear primary responsibility for childcare, food preparation, and household hygiene practices, which may partly explain the predominance of female respondents in this study. Similar gendered patterns in hygiene responsibilities have been documented in refugee and other low-resource contexts [17, 21].

Taken together, the findings of this study align with a broader body of literature demonstrating that structural factors such as access to water, soap, and functional handwashing facilities must be combined with behavioral and educational strategies to achieve lasting improvements in HWWS. In humanitarian contexts, where disease transmission risks are high, such integrated approaches are critical to protecting population health. Moreover, the results add to the evidence base showing that refugee and crisis-affected communities, despite resource limitations, can achieve and sustain improved hygiene behaviors when provided with enabling conditions and targeted support.

Limitations

This study has several limitations that should be considered when interpreting the findings. First, the use of structured observations to capture handwashing behavior may have introduced reactivity bias, whereby participants modified their actions in response to being observed (the Hawthorne effect) [37]. Although enumerators were trained to observe discreetly and participants were not informed that handwashing was the primary behavior of interest, it is possible that the frequency of handwashing with soap was overestimated.

Second, the study design did not include a passive control arm, which limits the ability to isolate the independent effects of the Jengu handwashing facility from those of soap provision and safe water. Both study groups received soap, and the observed increases in HWWS across groups suggest that soap availability played a strong role in influencing behavior. Future studies incorporating a third arm without any intervention would be useful in determining the unique contribution of infrastructure improvements versus supply-side interventions.

Third, while the follow-up period spanned approximately one year, this duration may not fully capture the long-term sustainability of behavior change. The slight decline in HWWS observed at the endline indicates that continuous reinforcement through regular soap provision and hygiene promotion may be necessary to sustain gains. Longer-term evaluations could provide further insights into the persistence of behavior change under routine humanitarian programming conditions.

Finally, the study was conducted within a single refugee camp (Ifo Camp in Dadaab, Kenya), and the findings may therefore not be generalizable to other humanitarian settings with different demographic, cultural, or environmental conditions. Nonetheless, the randomized design, high retention rate, and triangulation of survey and observational data strengthen the internal validity of the findings and provide robust evidence to inform WASH interventions in similar crisis-affected settings.

Conclusion

In conclusion, this study suggests that improvements in handwashing infrastructure, together with reliable access to soap and water, may support improved hygiene practices in refugee settings. While HWWS increased during the study period, improvements occurred in both groups, and the difference-in-differences analysis did not detect a statistically significant additional effect attributable to the Jengu facility. These findings suggest that sustained improvements in handwashing behavior are likely driven by a combination of enabling infrastructure, consistent soap supply, and ongoing behavior change efforts. In humanitarian contexts where disease transmission risks remain high, integrated WASH strategies that address both hardware and behavioral determinants remain essential for supporting improved hygiene practices. Future research should explore how such interventions can be adapted, scaled, and sustained in protracted crisis contexts.

Abbreviations

HWWS

Handwashing with soap

WASH

Water, sanitation, and hygiene

UNHCR

United Nations High Commissioner for Refugees

DiD

Difference-in-differences

ARI

Acute respiratory infections

GI

Gastrointestinal

IFRC

International Federation of Red Cross and Red Crescent Societies

Author contributions

Dennis Akeyo (DA), Innocent Kamara Tumwebaze (IKT), Vishna Shah (VS), Naomi Ng’ang’a (NN), George Kiragu (GK), Viviana Olivetto (VO), Stephen Philips (SP), Sheillah Simiyu (SS).DA, IKT, VS, NN, GK, VO, SP, and SS contributed to the conception and development of the manuscript. DA contributed to conceptualization, data curation, methodology, formal analysis, and drafting of the original manuscript. IKT contributed to the methodology and drafting of the original manuscript. VS contributed to the methodology, validation, and review and editing of the manuscript. SS contributed to the methodology, supervision, and review and editing of the manuscript. NN, GK, VO, and SP contributed to the review and editing of the manuscript.All the authors read and approved the final manuscript.

Funding

This study was funded by Elrha under Research for Health in Humanitarian Crises (r2hc) (funding grant number: 70171).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author or corresponding author organization on reasonable request.

Declarations

Ethics approval and consent to participate

This study received ethical approval internally from African Population and Health Research Center Scientific Research Committee, and the London School of Hygiene and Tropical Medicine Ethics Review Board. National approvals were also obtained from the AMREF Research and Ethics committee in Kenya and the National Commission for Science, Technology and Innovation (NACOSTI). Permits were also obtained from the UNHCR, and the county and the sub-county levels in Kenya. All study participants were provided with all the necessary information regarding the purpose of the study and information channels using the language they understood. Written informed consent was obtained from the household head or next in charge on behalf of the whole household.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.