Handwashing with soap is a cost-effective way of reducing diarrheal disease mortality in children under 5. Tracking this practice among child caretakers is a challenge, as the gold standard method structured observations is cumbersome, costly, and conducive to over-performance. The water, sanitation, and hygiene (WASH) field needs a valid, reliable proxy to track handwashing with soap in large surveys. This proxy is crucial as the new 2015–2030 Sustainable Development Goals (SDGs) may track hygiene. Using data from the Multiple Indicators Cluster Survey (MICS) and the Demographic Health Survey (DHS) from five countries, we conducted multivariate analyses to explore an association between the presence of functional handwashing stations (HWSs), (together with needed supplies) and the likelihood of lower reports of child diarrheal disease. A limited to moderate association exists in three of the five countries considered, characterized by comparable rates of childhood diarrhea: Malawi, Sierra Leone, and Zimbabwe. The relationship was detected when controlling for potential confounding factors (other WASH elements, socio-demographic factors, nutrition practices, and immunization status) and when accounting for cluster effects. The likelihood of reported diarrhea among children under 5 increases when there is no HWS, just a handwashing device with no supplies, only water or only soap. The relationship is moderate in Malawi and less strong in Sierra Leone and Zimbabwe. No relationship was found in Ethiopia and Ghana. Further exploration of the usefulness of this proxy in other African and non-African contexts is warranted.

INTRODUCTION

Diarrhea is considered the second most deadly disease for the world's poorest children (UNICEF 2012). According to the UN Interagency Group for Child Mortality Estimation (2013), 9% of child mortality worldwide in 2012 was caused by diarrheal disease. Pruss-Ustun et al. (2014) estimated that about 35% of diarrhea-related deaths among all age groups in low and middle income countries in 2012 were related to poor handwashing practices. Handwashing with soap at critical times before food handling and after contact with fecal matter can prevent diarrheal disease. Estimates of the number of deaths due to diarrheal disease (among different age groups, including children under 5) that can be averted with appropriate handwashing range from 23 (Freeman et al. 2014) to 48% (Cairncross et al. 2010), depending on studies analyzed and method of estimation used. However, the prevalence of handwashing with soap is low in many developing countries. Based on results from 42 studies, Freeman et al. (2014) estimated that: (a) in low and middle income countries, prevalence of handwashing with soap at critical times ranges from 13 to 17%; and (b) in countries representing all income levels, only 19% of people wash their hands after contact with excreta. The authors of that systematic review also argue that their findings may actually overestimate handwashing rates, given the methodology used to track this practice in the studies retained. In their own words, in observational studies, ‘it is not known whether the subject uses the latrine for defecation … handwashing after toilet use is a relevant proxy for handwashing after contact with excreta’.

Low handwashing rates worldwide, and especially in low and middle income countries, call for programs to promote good hygiene. In such programs it is crucial to track the effectiveness of efforts to change practices. This requires reliable measures of the practice. However, tracking handwashing practices is difficult; self-reports may not be reliable. Because handwashing is a socially acceptable practice (even where prevalence is low) study participants may over-report this practice, resulting in ‘respondent bias'. Manun'Ebo et al. (1997) detected over-reporting of handwashing before food preparation and eating, and Danquah (2010) found over-reporting of handwashing with soap after defecation. Thus, alternative methods for tracking handwashing practices are needed.

In studies of handwashing to date, the gold standard has been structured observations. However, Biran et al. (2008) concluded that structured observations are cumbersome, may have poor repeatability, and may be easier in some cultural contexts than in others. Furthermore, Ram et al. (2010) have argued that structured observations introduce reactivity: observed individuals may over-perform, leading to overestimates of actual behavior.

An alternative is to use a proxy indicator for handwashing. However, to date, evidence of the validity of any proxy indicator for this practice has been lacking. We propose that a possible proxy for handwashing includes spot checks/observations of handwashing stations (HWSs) defined here as devices (e.g., a tap) together with supplies for appropriate practice (i.e., water and soap). Spot checks to confirm the presence of such stations would not be able to identify who washes their hands (e.g., the caregiver or the child) or when the practice is performed (e.g., before handling food or after fecal contact). However, we propose that if results obtained from such spot checks can be shown to have an association with (a) results from structured observations (the gold standard) and (b) health outcomes, this proxy would acquire predictive validity.

The first association has already been shown. Biran et al. (2008) used two proxies to study handwashing: ‘observed soap beside a latrine’ and ‘observed soap in the yard’. They concluded that handwashing rates using these two measurement methods are both similar to those obtained when using structured observations. By the same token, Luby et al. (2009) reported that having water and soap available at the place to wash hands after using a sanitation facility is associated with washing hands with soap after fecal contact. Ram (2013) has suggested that observations of the availability of soap and water at dedicated HWSs, and inspections of hand cleanliness, are appropriate proxies for tracking handwashing because these indicators are both reliable and efficient.

The relationship between observed handwashing practices and reduction in diarrheal disease has been reported by Freeman et al. (2014), cited earlier, as well as in studies included in a systematic review by Curtis & Cairncross (2003). The question that remains to be answered is whether there is a relationship between the proposed proxy (spot checks of fully functional HWSs) and health outcomes. This paper focuses on the results of secondary data analysis conducted to study whether the presence of a functional HWS can predict lower diarrheal disease prevalence among children under 5 years of age a question not yet addressed by the literature. The hypothesis guiding this study is that HWSs that are fully functional are associated with reductions in diarrheal incidence, and that partially functional HWSs with only soap or only water available do not have the same effect.

METHODS

This analysis used Demographic Health Survey (DHS) and Multiple Indicator Cluster Survey (MICS) data from five sub-Saharan African countries: Ethiopia (DHS 2011), Ghana (MICS 2010), Malawi (DHS 2010), Sierra Leone (MICS 2010), and Zimbabwe (DHS 2010). These countries were chosen because they had similar rates of child diarrhea and recent survey data included questions related to handwashing. The total sample used in our analysis included 94.9% of all children under the age of 5 belonging to households surveyed, or a total of 47,827 children (Table 1). The only reason for excluding a case included in the data files of the household surveys was non-response to relevant questions. This large sample size provided sufficient power to detect small but statistically significant associations between the factors of interest.

Table 1

Sample size by country

  Total children under 5 from surveyed households Children included in the analysis 
Ethiopia 11,042 91.3% (10,081) 
Ghana 7,550 97.9% (7391) 
Malawi 18,013 99.0% (17,829) 
Sierra Leone 8,598 93.7% (8057) 
Zimbabwe 5,208 85.8% (4469) 
All countries 50,411 94.9% (47,827) 
  Total children under 5 from surveyed households Children included in the analysis 
Ethiopia 11,042 91.3% (10,081) 
Ghana 7,550 97.9% (7391) 
Malawi 18,013 99.0% (17,829) 
Sierra Leone 8,598 93.7% (8057) 
Zimbabwe 5,208 85.8% (4469) 
All countries 50,411 94.9% (47,827) 

We explored the potential relationship between HWSs and diarrheal disease via household surveys that included self-reported data by caregivers of children under 5 regarding diarrheal disease in this group during the previous 2 weeks. For purposes of analysis, we defined a ‘fully functional’ HWS as a location having a device that permits handwashing and which is supplied with both soap and water. Appropriate handwashing cannot occur in the absence of these two supplies. We assume that if a station only has water and no soap, individuals using that station engage in hand rinsing and not in handwashing per se. In that case we suggest that a weaker or no relationship would be found between the practice and prevention of diarrheal disease.

Our analysis used information collected through the DHS and the MICS, Version 4. The indicator tracked by these surveys is: ‘% of households with a dedicated place for handwashing where water and soap are present’. The instruments for both surveys require enumerators to request respondents to show them where family members most frequently wash their hands and, if permitted, to observe what device is used and whether water and soap are present at this location. If a faucet (tap) is available, enumerators are expected to open it to see if water is flowing.

We used logistic regression to examine associations between our handwashing proxy and our primary outcome diarrheal disease in children under the age of 5 reported by mothers or primary caretakers in the 2 weeks prior to the interview. Using a multivariate analysis approach, separate models were designed for each country. Models were run using the Statistical Package for the Social Sciences (SPSS) version 17 (SPSS Inc., Chicago, Illinois, USA).

Variables incorporated in the models are defined in Table 2 and included the outcome of interest plus indicators related to water, sanitation, and hygiene (WASH) as well as other potentially confounding factors related to demographics (household, primary caretaker, and child), child nutrition and feeding practices, and vaccinations. In the table, the reference category for each factor is italicized; in each case it indicates the condition for which the lowest rates of diarrhea would be expected. DHS and MICS sampling weights were used in the analysis to make the data representative for the entire target population. The models were also adjusted for effects due to the multi-stage sampling designs of the selected surveys.

Table 2

Definition of variables used in binary logistic regression model

Outcome of interest Definition 
Diarrhea 
  • Diarrhea: diarrhea self-reported by mothers/caretakers in the 2 weeks prior to the interview for children under the age of 5

 
Wash indicators Definition(reference categories italicized) 
Drinking water 
  • Improved1: source of drinking water is piped water (into dwelling, yard, plot, or to neighbor's yard); public tap or standpipe; tubewell or borehole; protected spring; protected well; rainwater collection; or bottled or sachet water with improved water source for other uses

 
  • Unimproved: source of drinking water is unprotected well; unprotected spring; river, dam, lake, pond, stream, canal, or irrigation channel; tanker truck; cart with small tank; other sources; or bottled or sachet water with an unimproved water source for other uses

 
Sanitation facility 
  • Improved: private facility of the following types: flush or pour-flush to piped sewer system, septic tank, or pit latrine; ventilated improved pit (VIP) latrine; pit latrine with slab; or composting toilet

 
  • Unimproved: shared or public facilities of any type; flush or pour-flush to elsewhere or unknown; pit latrine without slab, or open pit; bucket; hanging toilet or hanging latrine, or other

 
  • Open defecation: no facilities; bush or field

 
Distance to water source 
  • Close: water source located in dwelling, yard or plot; or less than 30 minutes to get to water source

 
  • Far: more than 30 minutes to get to water source

 
Household treatment of water 
  • Treated2: drinking water treated in the household by bleach/chlorine; cloth strainer; filter; solar disinfection; water tablets; or camphor

 
  • Untreated: water not treated in the household, or method was letting it stand and settle; other; or unknown

 
HWS 
  • No HWS: no designated place in the household for handwashing, or none that could be observed

 
  • HWS with no supplies: observed, designated place in the household for handwashing, but no soap or water present at the time of the interview

 
  • HWS with water only: observed, designated place in the household for handwashing, but only water was present at the time of the interview

 
  • HWS with soap only: observed, designated place in the household for handwashing, but only soap was present at the time of the interview

 
  • Fully functional HWS: observed, designated place in the household for handwashing, with both soap and water present at the time of the interview

 
Control variables Definition (reference categories italicized) 
Urban/rural 
  • Urban

 
  • Rural

 
Wealth quintile Composite measure of a household's relative wealth using ownership of selected assets: 
  • Poorest: 1st wealth quintile

 
  • Poorer: 2nd wealth quintile

 
  • Middle: 3rd wealth quintile

 
  • Richer: 4th wealth quintile

 
  • Richest: 5th wealth quintile

 
Household size Total number of household members 
Mother's education 
  • No education

 
  • Primary

 
  • Secondary3

 
  • Higher education

 
Mother's age Mother or primary caretaker's age (in years) 
Child's age Child's age (in years) from 0–4 
Child's sex 
  • Male

 
  • Female

 
Exclusive breastfeeding 
  • Exclusive breastfeeding: children younger than 6 months receiving breast milk and no other fluids or foods in the day or night before the interview, with the exception of oral rehydration solution, vitamins, mineral supplements, and medicines

 
  • Non-exclusive breastfeeding: children younger than 6 months not being exclusively breastfed according to the definition above – i.e., either partially breastfed or not breastfed at all

 
  • Older children: children 6 months and older

 
Age-appropriate complementary feeding 
  • Age-appropriate complementary feeding: children between 6 and 24 months who are breastfed in addition to being given solid, semi-solid, or soft foods in the day and night before the interview

 
  • Not age-appropriate complementary feeding: children between 6 and 24 months receiving all other diets

 
  • Older or younger children: children less than 6 months or older than 24 months

 
Vaccinations 
  • Any vaccination4: child has at least one vaccination recorded on immunization card, or mother reported child had received any vaccination including BCG; polio; pentavalent or diphtheria, pertussis and tetanus (DPT); measles; yellow fever; or hepatitis B

 
  • No vaccinations: no vaccinations recorded on immunization card, or mother did not report child receiving any vaccinations

 
Outcome of interest Definition 
Diarrhea 
  • Diarrhea: diarrhea self-reported by mothers/caretakers in the 2 weeks prior to the interview for children under the age of 5

 
Wash indicators Definition(reference categories italicized) 
Drinking water 
  • Improved1: source of drinking water is piped water (into dwelling, yard, plot, or to neighbor's yard); public tap or standpipe; tubewell or borehole; protected spring; protected well; rainwater collection; or bottled or sachet water with improved water source for other uses

 
  • Unimproved: source of drinking water is unprotected well; unprotected spring; river, dam, lake, pond, stream, canal, or irrigation channel; tanker truck; cart with small tank; other sources; or bottled or sachet water with an unimproved water source for other uses

 
Sanitation facility 
  • Improved: private facility of the following types: flush or pour-flush to piped sewer system, septic tank, or pit latrine; ventilated improved pit (VIP) latrine; pit latrine with slab; or composting toilet

 
  • Unimproved: shared or public facilities of any type; flush or pour-flush to elsewhere or unknown; pit latrine without slab, or open pit; bucket; hanging toilet or hanging latrine, or other

 
  • Open defecation: no facilities; bush or field

 
Distance to water source 
  • Close: water source located in dwelling, yard or plot; or less than 30 minutes to get to water source

 
  • Far: more than 30 minutes to get to water source

 
Household treatment of water 
  • Treated2: drinking water treated in the household by bleach/chlorine; cloth strainer; filter; solar disinfection; water tablets; or camphor

 
  • Untreated: water not treated in the household, or method was letting it stand and settle; other; or unknown

 
HWS 
  • No HWS: no designated place in the household for handwashing, or none that could be observed

 
  • HWS with no supplies: observed, designated place in the household for handwashing, but no soap or water present at the time of the interview

 
  • HWS with water only: observed, designated place in the household for handwashing, but only water was present at the time of the interview

 
  • HWS with soap only: observed, designated place in the household for handwashing, but only soap was present at the time of the interview

 
  • Fully functional HWS: observed, designated place in the household for handwashing, with both soap and water present at the time of the interview

 
Control variables Definition (reference categories italicized) 
Urban/rural 
  • Urban

 
  • Rural

 
Wealth quintile Composite measure of a household's relative wealth using ownership of selected assets: 
  • Poorest: 1st wealth quintile

 
  • Poorer: 2nd wealth quintile

 
  • Middle: 3rd wealth quintile

 
  • Richer: 4th wealth quintile

 
  • Richest: 5th wealth quintile

 
Household size Total number of household members 
Mother's education 
  • No education

 
  • Primary

 
  • Secondary3

 
  • Higher education

 
Mother's age Mother or primary caretaker's age (in years) 
Child's age Child's age (in years) from 0–4 
Child's sex 
  • Male

 
  • Female

 
Exclusive breastfeeding 
  • Exclusive breastfeeding: children younger than 6 months receiving breast milk and no other fluids or foods in the day or night before the interview, with the exception of oral rehydration solution, vitamins, mineral supplements, and medicines

 
  • Non-exclusive breastfeeding: children younger than 6 months not being exclusively breastfed according to the definition above – i.e., either partially breastfed or not breastfed at all

 
  • Older children: children 6 months and older

 
Age-appropriate complementary feeding 
  • Age-appropriate complementary feeding: children between 6 and 24 months who are breastfed in addition to being given solid, semi-solid, or soft foods in the day and night before the interview

 
  • Not age-appropriate complementary feeding: children between 6 and 24 months receiving all other diets

 
  • Older or younger children: children less than 6 months or older than 24 months

 
Vaccinations 
  • Any vaccination4: child has at least one vaccination recorded on immunization card, or mother reported child had received any vaccination including BCG; polio; pentavalent or diphtheria, pertussis and tetanus (DPT); measles; yellow fever; or hepatitis B

 
  • No vaccinations: no vaccinations recorded on immunization card, or mother did not report child receiving any vaccinations

 

1Piped water to a neighbor's yard was only a response option in Sierra Leone and Ghana. Sachet water was only a response option in Ghana. Only respondents in Ghana, Ethiopia, and Sierra Leone were asked about a water source for other uses when bottled water was the main source of drinking water. As a result, respondents in these countries could be classified based on whether or not their source of water for other uses was improved or unimproved. In Zimbabwe and Malawi all respondents who reported bottled water as their main source of drinking water were classified as having an unimproved water source.

2 Water tablets and camphor were the only response options in Ghana.

3 Reference category in Sierra Leone only, where higher education was not a response option.

4 Yellow fever was only a response option in Sierra Leone and Ghana. Hepatitis B was only a response option in Sierra Leone.

Using answers to questions about handwashing practices, households were categorized into five groups: (1) no HWS, (2) HWS with no supplies, (3) HWS with water only, (4) HWS with soap only, and (5) fully functional HWS. Other WASH indicators included in the analysis were: source of drinking water, distance to water source, type of sanitation facility, and household treatment of water based on relevant definitions from the WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation (WHO & UNICEF 2014). Since exclusive breastfeeding has been shown to reduce rates of diarrhea, variables related to diet and breastfeeding were added to the model. Receipt of vaccination was included as a proxy for general health-seeking behavior.

RESULTS AND DISCUSSION

Table 3 presents results by country regarding prevalence of child diarrhea in the previous 2 weeks. Rates ranged from 12.7% in Ghana to 17.5% in Malawi.

Table 3

Prevalence of diarrhea by country

Country Children with diarrhea in 2 weeks prior to survey 
Ethiopia 13.5% (1483) 
Ghana 12.7% (956) 
Malawi 17.5% (3158) 
Sierra Leone 15.5% (1331) 
Zimbabwe 13.2% (688) 
All countries 15.1% (7616) 
Country Children with diarrhea in 2 weeks prior to survey 
Ethiopia 13.5% (1483) 
Ghana 12.7% (956) 
Malawi 17.5% (3158) 
Sierra Leone 15.5% (1331) 
Zimbabwe 13.2% (688) 
All countries 15.1% (7616) 

Table 4 shows that the prevalence of HWSs in all households varied widely by country. Rates were lowest in Malawi and Ethiopia, where greater than 90% of children under 5 lived in households with no HWS; only 2.0% and 0.4% of children respectively lived in households with a fully functional HWS having both soap and water. The highest rates were in Zimbabwe, where 21.2% of children lived in households with a fully functional HWS. The percentage of children in households in that country with no HWS was 46.0%.

Table 4

Percentage of children under five in households with HWSs by country

  Ethiopia N = 11,042 Ghana N = 7,550 Malawi N = 18,013 Sierra Leone N = 8,598 Zimbabwe N = 5,208 
No HWS 98.5 80.3 92.7 34.3 46.0 
HWS with no supplies, only device 0.5 5.6 2.3 41.9 10.1 
HWS with water only 0.5 4.2 2.5 8.6 20.9 
HWS with soap only 0.0 1.9 0.3 4.0 1.8 
Fully functional HWS 0.4 8.1 2.0 10.9 21.2 
  Ethiopia N = 11,042 Ghana N = 7,550 Malawi N = 18,013 Sierra Leone N = 8,598 Zimbabwe N = 5,208 
No HWS 98.5 80.3 92.7 34.3 46.0 
HWS with no supplies, only device 0.5 5.6 2.3 41.9 10.1 
HWS with water only 0.5 4.2 2.5 8.6 20.9 
HWS with soap only 0.0 1.9 0.3 4.0 1.8 
Fully functional HWS 0.4 8.1 2.0 10.9 21.2 

A word of caution is in order. Seeking a relationship between functional HWSs and diarrheal disease by observing the data in Tables 3 and 4 will provide a wrong impression, since diarrhea is a complex phenomenon determined by many factors and handwashing with soap is only one of the possible determinants. The influence of other determinants has to be extracted in order to get a better sense of the net effect of handwashing.

Table 5 presents results (i.e., odds ratios (OR) and 95% confidence intervals (CI)) for other potential predictors of diarrhea including WASH indicators, demographic characteristics, child feeding practices, and vaccinations (as a proxy for general health-seeking behavior). Child's age was the most consistent predictor of diarrhea. As the age of the child increased from 1 to 60 months, the odds of diarrhea decreased in every study country. Worldwide, the incidence of diarrhea is highest from about 6 to 11 months of age when children are generally weaned and lose much of the protective benefits of breastfeeding, and declines thereafter (Keusch et al. 2006). An additional explanation may be provided by Hirve & Ganatra (1977), who suggest that the most vulnerable children die earliest and those who survive belong to a group that is better able to resist infections.

Table 5

Association between potential confounding factors with reports of child diarrhea in the 2 weeks prior to the survey (multiple logistic regression analysis)1

  Characteristic Ethiopia OR (95% CI) Ghana OR (95% CI) Malawi OR (95% CI) Sierra Leone OR (95% CI) Zimbabwe OR (95% CI) 
WASH Unimproved drinking water 1.2 (1.0–1.4)** 1.0 (0.8–1.2) 1.1 (1.0–1.2) 1.2 (1.0–1.4)* 1.4 (1.1–1.7)** 
Unimproved sanitation facility 1.1 (0.9–1.4) 0.9 (0.7–1.2) 1.1 (0.9–1.3) 1.4 (1.1–1.8)* 1.3 (1.0–1.6)* 
Far water source 0.9 (0.8–1.0)* 1.1 (0.9–1.4) 1.0 (0.9–1.1) 1.0 (0.8–1.3) 1.4 (1.1–1.8)** 
Untreated water 0.9 (0.8–1.1) 1.6 (1.1–2.2)** 0.8 (0.8–0.9)*** 0.9 (0.7–1.2) 0.8 (0.6–1.0)* 
SOCIO ECONOMIC Rural 1.1 (0.9–1.5) 1.0 (0.8–1.2) 0.9 (0.8–1.0) 1.0 (0.9–1.2) 0.5 (0.4–0.7)*** 
Poorest 1.0 (0.8–1.4) 2.2 (1.5–3.2)*** 1.0 (0.8–1.2) 1.0 (0.8–1.3) 1.1 (0.7–1.7) 
Poorer 0.9 (0.7–1.2) 2.3 (1.6–3.2)*** 0.9 (0.8–1.1) 1.0 (0.8–1.3) 1.1 (0.7–1.7) 
Middle 1.0 (0.7–1.3) 1.9 (1.4–2.6)*** 1.0 (0.8–1.2) 1.1 (0.9–1.4) 0.9 (0.6–1.3) 
Richer 1.2 (1.0–1.6) 2.0 (1.4–2.7)*** 0.9 (0.8–1.0) 0.9 (0.7–1.1) 1.1 (0.8–1.6) 
Mother's education: no education 1.2 (0.7–2.2) 1.5 (1.1–2.0)* 1.1 (0.7–2.0) 1.0 (0.8–1.2) 1.3 (0.4–3.8) 
Mother's education: primary 1.1 (0.6–2.0) 1.2 (0.9–1.6) 1.1 (0.6–1.8) 1.1 (0.8–1.4) 2.1 (1.0–4.3)* 
Mother's age 1.0 (1.0–1.0) 1.0 (1.0–1.0)* 1.0 (1.0–1.0)*** 1.0 (1.0–1.0) 1.0 (1.0–1.0)* 
Child's age 0.7 (0.6–0.7)*** 0.7 (0.7–0.8)*** 0.6 (0.6–0.7)*** 0.8 (0.8–0.9)*** 0.8 (0.7–0.9)*** 
Male 1.2 (1.1–1.4)*** 1.0 (0.9–1.2) 1.1 (1.0–1.2) 1.1 (1.0–1.2) 1.2 (1.0–1.5)* 
FEEDING Non-exclusive breastfeeding for child <6 months 2.1 (1.4–3.1)*** 1.2 (0.7–2.2) 2.7 (1.9–3.8)*** 1.7 (0.9–3.2) 2.2 (0.9–5.7) 
Not age-appropriate complementary feeding 0.9 (0.8–1.1) 1.6 (1.2–2.0)*** 0.8 (0.7–0.9)*** 0.7 (0.6–0.9)** 1.0 (0.7–1.3) 
VACCINES No vaccines 0.6 (0.5–0.7)*** 0.4 (0.1–0.9)* 0.7 (0.5–0.9)* 1.2 (0.8–1.6) 0.7 (0.5–1.0)* 
  Characteristic Ethiopia OR (95% CI) Ghana OR (95% CI) Malawi OR (95% CI) Sierra Leone OR (95% CI) Zimbabwe OR (95% CI) 
WASH Unimproved drinking water 1.2 (1.0–1.4)** 1.0 (0.8–1.2) 1.1 (1.0–1.2) 1.2 (1.0–1.4)* 1.4 (1.1–1.7)** 
Unimproved sanitation facility 1.1 (0.9–1.4) 0.9 (0.7–1.2) 1.1 (0.9–1.3) 1.4 (1.1–1.8)* 1.3 (1.0–1.6)* 
Far water source 0.9 (0.8–1.0)* 1.1 (0.9–1.4) 1.0 (0.9–1.1) 1.0 (0.8–1.3) 1.4 (1.1–1.8)** 
Untreated water 0.9 (0.8–1.1) 1.6 (1.1–2.2)** 0.8 (0.8–0.9)*** 0.9 (0.7–1.2) 0.8 (0.6–1.0)* 
SOCIO ECONOMIC Rural 1.1 (0.9–1.5) 1.0 (0.8–1.2) 0.9 (0.8–1.0) 1.0 (0.9–1.2) 0.5 (0.4–0.7)*** 
Poorest 1.0 (0.8–1.4) 2.2 (1.5–3.2)*** 1.0 (0.8–1.2) 1.0 (0.8–1.3) 1.1 (0.7–1.7) 
Poorer 0.9 (0.7–1.2) 2.3 (1.6–3.2)*** 0.9 (0.8–1.1) 1.0 (0.8–1.3) 1.1 (0.7–1.7) 
Middle 1.0 (0.7–1.3) 1.9 (1.4–2.6)*** 1.0 (0.8–1.2) 1.1 (0.9–1.4) 0.9 (0.6–1.3) 
Richer 1.2 (1.0–1.6) 2.0 (1.4–2.7)*** 0.9 (0.8–1.0) 0.9 (0.7–1.1) 1.1 (0.8–1.6) 
Mother's education: no education 1.2 (0.7–2.2) 1.5 (1.1–2.0)* 1.1 (0.7–2.0) 1.0 (0.8–1.2) 1.3 (0.4–3.8) 
Mother's education: primary 1.1 (0.6–2.0) 1.2 (0.9–1.6) 1.1 (0.6–1.8) 1.1 (0.8–1.4) 2.1 (1.0–4.3)* 
Mother's age 1.0 (1.0–1.0) 1.0 (1.0–1.0)* 1.0 (1.0–1.0)*** 1.0 (1.0–1.0) 1.0 (1.0–1.0)* 
Child's age 0.7 (0.6–0.7)*** 0.7 (0.7–0.8)*** 0.6 (0.6–0.7)*** 0.8 (0.8–0.9)*** 0.8 (0.7–0.9)*** 
Male 1.2 (1.1–1.4)*** 1.0 (0.9–1.2) 1.1 (1.0–1.2) 1.1 (1.0–1.2) 1.2 (1.0–1.5)* 
FEEDING Non-exclusive breastfeeding for child <6 months 2.1 (1.4–3.1)*** 1.2 (0.7–2.2) 2.7 (1.9–3.8)*** 1.7 (0.9–3.2) 2.2 (0.9–5.7) 
Not age-appropriate complementary feeding 0.9 (0.8–1.1) 1.6 (1.2–2.0)*** 0.8 (0.7–0.9)*** 0.7 (0.6–0.9)** 1.0 (0.7–1.3) 
VACCINES No vaccines 0.6 (0.5–0.7)*** 0.4 (0.1–0.9)* 0.7 (0.5–0.9)* 1.2 (0.8–1.6) 0.7 (0.5–1.0)* 

* P ≤ 0.05.

** P ≤ 0.01.

*** P ≤ 0.001.

1ORs adjusted for all characteristics in Table 2. Characteristics with no statistically significant associations with diarrhea excluded from the table include: open defecation, household size, and mother's secondary education.

In our findings, another important factor associated with child diarrhea was feeding practices, particularly for the youngest children. Those under 6 months who were not exclusively breastfed were 2.1 and 2.7 times more likely to have diarrhea in Ethiopia and Malawi, respectively, when compared to children who were. Other consistent predictors of diarrhea across countries included having an unimproved drinking water source (three of the study countries, with ORs ranging from 1.2 to 1.4), and having an unimproved sanitation facility (two of the countries, with ORs of 1.3 and 1.4).

Table 6 shows the results of multiple logistic regression for the proposed handwashing proxy considered in this analysis. When potentially confounding factors are held constant, caretakers of children living in households with a fully functional HWS were less likely to have reported an episode of diarrhea in the preceding 2 weeks than their counterparts in households with either no HWS or a HWS with incomplete or no supplies. In Malawi, caretakers of children in households with no HWS were 2.3 times more likely to report child diarrhea than those in households with a fully functioning HWS; in Zimbabwe, caretakers with no HWS were 1.8 times more likely to report a child with diarrhea.

Table 6

Association between type of HWS in the home and report of child diarrhea in the last 2 weeks (multiple logistic regression analysis)1

Type of HWS Ethiopia OR (95% CI) Ghana OR (95% CI) Malawi OR (95% CI) Sierra Leone OR (95% CI) Zimbabwe OR (95% CI) 
No HWS 1.1 (0.4–3.1) 0.8 (0.6–1.1) 2.3 (1.5–3.4)*** 1.2 (0.9–1.5) 1.8 (1.4–2.3)*** 
HWS with no supplies 3.2 (0.9–11.0) 1.1 (0.7–1.6) 2.5 (1.5–4.1)*** 1.5 (1.2–1.9)*** 1.1 (0.7–1.6) 
HWS with water only 0.7 (0.2–2.7) 1.3 (0.9–1.9) 1.7 (1.0–2.7)* 1.1 (0.8–1.5) 1.5 (1.1–2.0)** 
HWS with soap only n/a2 0.5 (0.2–1.1) 1.7 (0.7–4.2) 1.8 (1.2–2.6)** 1.0 (0.5–2.1) 
Full HWS Reference 
Type of HWS Ethiopia OR (95% CI) Ghana OR (95% CI) Malawi OR (95% CI) Sierra Leone OR (95% CI) Zimbabwe OR (95% CI) 
No HWS 1.1 (0.4–3.1) 0.8 (0.6–1.1) 2.3 (1.5–3.4)*** 1.2 (0.9–1.5) 1.8 (1.4–2.3)*** 
HWS with no supplies 3.2 (0.9–11.0) 1.1 (0.7–1.6) 2.5 (1.5–4.1)*** 1.5 (1.2–1.9)*** 1.1 (0.7–1.6) 
HWS with water only 0.7 (0.2–2.7) 1.3 (0.9–1.9) 1.7 (1.0–2.7)* 1.1 (0.8–1.5) 1.5 (1.1–2.0)** 
HWS with soap only n/a2 0.5 (0.2–1.1) 1.7 (0.7–4.2) 1.8 (1.2–2.6)** 1.0 (0.5–2.1) 
Full HWS Reference 

* P ≤ 0.05.

** P ≤ 0.01.

*** P ≤ 0.001.

1ORs adjusted for all characteristics in Table 2.

2No households in this category.

In addition, caretakers of children in households with a HWS but no supplies were 2.5 times more likely in Malawi and 1.5 times more likely in Sierra Leone to have reported diarrhea than counterparts in households with a fully functional HWS. Having a HWS with incomplete supplies was also a predictor of diarrhea, with ORs of 1.7 in Malawi and 1.5 in Zimbabwe, respectively, for children in households with a HWS with water but no soap. In Sierra Leone, children in households having a HWS with soap but no water were 1.8 times more likely to report diarrhea. There were no statistically significant associations between child diarrhea and type of HWS in Ethiopia or Ghana.

This analysis demonstrates an association in three out of five countries between presence and type of HWS in the household and likelihood of a child under 5 having diarrhea in the last 2 weeks. In the three countries (Malawi, Sierra Leone, and Zimbabwe) results are consistent with randomized controlled trial data showing a relationship between handwashing with soap and a reduction in diarrheal disease (Cairncross et al. 2010) suggesting that the confirmed presence of a fully functional HWS may provide a substitute for direct observations of handwashing. In addition, we found a stronger inverse association between having a fully functional HWS and reported diarrhea than for any other WASH-related factors. Wang (2011) suggested that in OR analysis, an OR between 1 and less than 2 indicates a limited association; an OR between 2 and 2.99 indicates a moderate association; and an OR of 3.0 and above indicates a strong association. Given our findings, using this rationale we may then say that the relationship found in Malawi is moderate for the most part and that the relationships elsewhere are limited.

A limitation of our study is the lack of consistency across countries in our results regarding the association between having a HWS and child diarrhea. Possible reasons for this inconsistency deserve further analysis. No relationship was found in Ethiopia, for example, the country with the lowest proportion of HWS among the five countries studied. In Ethiopia, the use of mobile handwashing devices (e.g., pails) is common. Relatively high use of HWS in a given country may be necessary for the proxy to be reliable.

Additional limitations in our analyses arise from the nature of the DHS and MICS survey questions. Our outcome of interest – child diarrhea in the 2 weeks preceding the survey – relied on caretaker reports. Finally, because child diarrhea varies by season, Schmidt et al. (2007) have pointed out the advantages of longitudinal over intermittent diarrhea prevalence studies. The importance of our proposed proxy would need to be established across seasons. There are two additional limitations of the analysis associated with the survey data collected through the DHS and the MICS. The analysis conducted here is a snapshot of what was observed in households at the time of the visit to conduct the interview and the recollection of diarrheal disease 2 weeks prior to the survey. There are no questions in the survey about the quality of water services that households may access to establish whether they are permanent or intermittent. By the same token, there is no information in the surveys about the location of functional HWSs to determine whether they are inside or near toilets and kitchens. Given the data limitations, we cannot explore if households with HWSs at any of these locations have less chance of reporting diarrhea for their children under 5. Finally, since the handwashing data does not come from structured observations, it is impossible to establish who was washing hands with soap: caretakers, children or both.

CONCLUSION

This study presents limited to moderate evidence supporting the potential usefulness of a proxy for tracking handwashing practices: the presence of a functional HWS with needed supplies.

Further analyses need to be conducted in additional countries to determine if the trend described here is present in other contexts, in Africa or elsewhere. Analyses should also be conducted in countries where there is a strong tradition of movable HWSs to see if movable devices can serve as a proxy in those contexts. In order to explore this question, we recommend that the DHS and the MICS include spot checks to report the type of device used to wash hands. Such information might help build a case for using the presence of a fully functional HWS (and possibly of a movable handwashing device) as a proxy for handwashing with soap an essential aspect of hygiene as part of the upcoming Sustainable Development Goals (SDGs).

The association demonstrated in three countries between the presence of a fully functional HWS and a decrease in reported child diarrhea also supports the argument for including promotion of HWS in hygiene programs. We postulate that fixed HWS may be preferable to movable devices because the formation of habits is strengthened by cues associated with specific physical contexts (Wood et al. 2005).

ACKNOWLEDGEMENTS AND DISCLAIMER

The research reported here was jointly funded by FHI 360′s internal research funds and by funds provided by USAID through Contract AID-OAA-A-10-00040 with FHI 360 to implement the WASHplus Project. The views expressed here are those of the authors and do not necessarily reflect those of USAID or FHI 360.

REFERENCES

REFERENCES
Biran
A.
Rabie
T.
Schmidt
W.
Juvekar
S.
Hirve
S.
Curtis
V.
2008
Comparing the performance of indicators of hand-washing practices in rural Indian households
.
Tropical Medicine and International Health.
13
(
2
),
278
285
.
Cairncross
S.
Hunt
C.
Boisson
S.
Bostoen
K.
Curtis
V.
Fung
I. C.
Schmidt
W. P.
2010
Water, sanitation and hygiene for the prevention of diarrhea
.
International Journal of Epidemiology
,
39
,
i193
i205
.
Curtis
V.
Cairncross
S.
2003
Effect of washing hands with soap on diarrhea risk in the community: a systematic review
.
The Lancet, Infectious Diseases
3
(
5
),
275
281
.
Danquah
L.
2010
Measuring hand washing behaviour: methodological and validity issues. South Asia Hygiene Practitioners Workshop.: http://www.ircwash.org/resources/measuring-hand-washing-behaviour-methodological-and-validity-issues-paper-presented-south
.
Freeman
M. C.
Stocks
M. E.
Cumming
O.
Jeandron
A.
Higgins
J. P. T.
Wolf
J.
Pruss-Ust
A.
Bonjour
S.
Hunter
P. R.
Fewtrell
L.
Curtis
V.
2014
Hygiene and health: systematic review of handwashing practices worldwide and update of health effects
.
Tropical Medicine and International Health
19
(
8
),
906
916
.
Hirve
S.
Ganatra
B.
1997
A prospective cohort study on the survival experience of under five children in rural western India
.
Indian Pediatr. 1997 Nov;
34
(
11
),
995
1001
.
Keusch
G. T.
Fontaine
O.
Bhargava
A.
Boschi-Pinto
C.
Bhutta
Z. A.
Gotuzzo
E.
Rivera
J.
Chow
J.
Shahid-Salles
S.
Laxminarayan
R.
2006
Chapter 19: Diarrheal disease
. In:
Disease Control Priorities in Developing Countries
.
2nd edition
. (
Jamison
D. T.
Breman
J. G.
Measham
A. R.
Alleyne
G.
Claeson
M.
Evans
D. B.
Jha
P.
Mills
A.
Musgrove
P.
, eds).
World Bank
,
Washington, DC
.
Luby
S. P.
Halder
A. K.
Tronchet
C.
Akhter
S.
Bhuiya
A.
Johnston
R. B.
2009
Household characteristics associated with handwashing with soap in rural Bangladesh
.
American Journal of Tropical Medicine and Hygiene.
81
(
5
),
882
887
.
Manun'Ebo
M.
Cousens
S.
Haggerty
P.
Kalengaie
M.
Ashworth
A.
Kirkwood
B.
1997
Measuring Hygiene Practices: a comparison of questionnaires with direct observations in rural Zaire
.
Trop Med Int Health.
2
(
11
),
1015
1021
.
Pruss-Ustun
A.
Bartram
J.
Clasen
T.
Colford
J. M.
Cumming
O.
Curtis
V.
Bonjour
S.
Dangour
A. D.
DeFrance
J.
Fewtrell
L.
Freeman
M. C.
Gordon
B.
Hunter
P. R.
Johnston
R. B.
Mathers
C.
Mausezahl
D.
Medlicott
K.
Neira
M.
Stocks
M.
Wolf
J.
Cairncross
S.
2014
Burden of disease from inadequate water, sanitation and hygiene in low- and middle-income settings: a retrospective analysis of data from 145 countries
.
Tropical Medicine and International Health.
19
(
9
),
894
905
.
Ram
P.
2013
Practical Guidance for Measuring Handwashing Behavior, 2013 Update. The Water and Sanitation Program. http://www.wsp.org/sites/wsp.org/files/publications/WSP-Practical-Guidance-Measuring-Handwashing-Behavior-2013-Update.pdf (accessed 2 October 2015)
.
UN Interagency Group for Child Mortality Estimation
.
2013
Levels and Trends in Child Mortality. http://www.childinfo.org/files/Child_Mortality_Report_2013.pdf (accessed 2 October 2015)
.
UNICEF
.
2012
Pneumonia and Diarrhoea: Tackling the deadliest diseases for the world's poorest children. http://www.childinfo.org/files/Pneumonia_Diarrhoea_2012.pdf(accessed 2 October 2015)
.
WHO, UNICEF
.
2014
Progress on Drinking Water and Sanitation. Annex 1. http://www.wssinfo.org/fileadmin/user_upload/resources/JMP_report_2014_webEng.pdf (accessed 2 October 2015)
.
Wood
W.
Tam
L.
Guerrero Witt
M.
2005
Changing circumstances, disrupting habits
.
Journal of Personality and Social Psychology
88
(
6
),
918
933
.