The impact of COVID-19 on households’ water use in Uganda

The unprecedented outbreak of COVID-19 necessitated the promotion of better hygiene practices to curb the spread of the virus. Better hygiene requires that households have a stable supply of water. However, little is known about the predictors of changes in water use in emergency situations such as COVID-19 in Uganda. This study uses data from a cross-sectional survey to examine the changes in the quantities of water used by 1,639 Ugandan households due to COVID19. This article also explores the factors that are associated with changes in water use. The month March 2020 is used in this study as a cut-off because this is the month in which the government implemented a lockdown to curb the spread of the virus. Results indicate that most households had an increase in the quantity of water used after March 2020 when compared with the period before March 2020. Household characteristics that were associated with a change in the quantity of water used were age, sex, education, main occupation of household head, household size and region of residence. The results can be used to inform the prediction and demand modelling of household water use for improved water interventions for equitable water supply during emergencies.

However, for proper handwashing to be effective, people need to have enough water that is readily available (Pickering et al. ). This is a challenge for people in remote areas especially in developing countries or even in areas where water supply is not guaranteed (Ekane et al. ). A higher demand for water than supply creates an avoidable water shortage (Sahin et al. ). A stable supply of water is crucial during pandemics when the demand for water to improve hygiene is at its highest (Kalbusch et al. ). The change in water demand impacts the performance of water distribution systems (Shang et al. ).
Access to clean water can be a challenge to the smooth functioning of the household (Rosinger et al. ). According to the WHO & UNICEF () report, about 2.1 billion people worldwide (mostly in developing countries) were estimated to lack access to clean water for home use (WHO & UNICEF ). While the quality of water used in homes is deteriorating in both developed and developing countries (Biswas & Tortajada ), developed countries are faring much better in terms of having clean and reliable water (Rosinger et al. ). In sub-Saharan Africa (SSA), limited access to clean water is attributed to poor institutions, bad leadership, poor service delivery and lack of political will address water scarcity (Dos Santos et al. ).
In sub-Saharan Africa, insufficient water supply is observed to be prevalent mostly in rural, informal or slum Mekuriaw & Gurmessa ). In Uganda, clean water supply remains a challenge in rural areas (Naiga et al. ). This problem is as a result of limited resources and skilled personnel, mismanagement and poor accountability of public funds (Calow et al. ). In urban areas, proper hygiene can be compromised in situations where households need to pay for the water they use (Naiga et al. ). This forces urban residents to restrict water use. Yet, this can facilitate the spread the viruses and bacteria (Burton et al. ). Moreover, limited quantities of water can negatively affect food availability (Quevauviller ). Transfer of water between households has been suggested to be a viable solution to help water-stressed households (Rosinger et al. ). Rainwater harvesting has also been suggested to solve water shortages in water-stressed areas (Zavala et al. ).
Unstable water supply compromises the fight against contagious diseases. While stable water supply is good for the prevention of the spread of diseases, there is limited knowledge on the utilisation of water in Uganda to aid in healthseeking behaviour such as handwashing (Musoke et al. ). Yet, understanding how people utilise water in their households is critical for building sustainable water systems that can be relied on in situations of a pandemic. This paper contributes to the growing empirical literature on COVID-19 and behavioural hygienic practices, by examining water utilisation due to COVID-19 in Ugandan households.
With the public threat created by the emergence of COVID-19, it is important to understand how prepared individuals, governments and relevant authorities are in terms of providing the necessary and needed services as well as designing sustainable interventions under emergencies.
That is, individuals should be able to use water as a means or solution to fight diseases or uphold an acceptable standard of hygiene during an outbreak of COVID-19.
As part of the interventions to promote handwashing during the COVID-19 outbreak, the president of Uganda gave a directive, at the height of the COVID-19 outbreak, to the water body (National Water and Sewerage Corporation) not to carry out any water disconnections during lockdown in order to avoid a failure to practice handwashing that might ensue as a result of limited water supply (NilePost ). While this directive was in good faith, water users would still need to clear their water bills at an appropriate time they get money. Regular handwashing would imply an increase in water bills to meet operational costs. These conditions call for citizens to adapt to new ways of life to minimise costs related to the use of water but also impacts on service delivery. Understanding new dynamics about water use before, during and after the COVID-19 pandemic can help shed light on strategies that can be adopted to improve water supply. Behavioural practices may be influenced by socio-economic factors. For example, higher educational attainment, being older and having better knowledge have been reported to be associated with better hygienic practices such as handwashing (Nteli et al. ).
While previous research (Staddon et al. ) has investigated the relationship between household socio-economic characteristics and water use in Uganda, little is known about the predictors of changes in water use in emergency situations such as COVID-19 particularly in developing countries such as Uganda (Brown et al. ; Branz et al. ). This study therefore addresses two main research questions: (1) How does utilisation of water vary under emergency situations (before and after March 2020)? (2) What are the household characteristics that are associated with either an increase, decrease or no change in water utilisation in emergency situations (before and after March 2020)? A cut-off period of March 2020 is used because that is the month the government of Uganda implemented a total lockdown for the first time due to an outbreak of COVID-19. Understanding the socio-economic predictors for changes in household water use can be used to inform efficient and effective prediction and demand modelling as a first step for the improved design of interventions for equi- to an individual's ability to own something as a measure of success in life (Chan & Prendergast ). Based on the theory of impulsive consumption, we hypothesise that water use in households could be influenced by impulsive consumption in many ways. For example, the COVID-19 outbreak could have influenced people's behaviour in terms of drinking water, washing, bathing, handwashing, and cleaning due to fear of death, thereby leading to an increase in demand for water. We contend total lockdown reflected a high severity of the COVID-19 pandemic which could have influenced high water use. Moreover, high demand and consumption for water use could be influenced by those who can afford it.

Source of data and sample size
The data used in this paper come from a household cross-

Data collection and ethical clearance
Data collection was conducted by well-trained research assistants using a paper questionnaire. Training of research assistants was carried out for two weeks prior to the start of the data collection exercise. The data collection exercise took place between 1 August 2020 and 31 August 2020. The response rate was 100% since the study had targeted 1,500 household heads. Ethical clearance to conduct the study was granted by the Makerere University Institutional Review Board. All participants had to consent to participate in the study.

Study sampling
Stratified stage random sampling was employed. A stratified random sampling approach was used to select study towns.
Umbrella Authorities for water and sanitation were considered as the strata in the sampling procedure so that all regions within the country were well represented. Stratified sampling for data collection considered regions, districts, from where small towns were sampled for the survey. Simple random sampling was used to select households for interviews in study areas.

Dependent variable
Household heads were asked to state the average quantity of water (in 20-litre jerrycans) they used in the household in a day before March 2020 and after March. Households whose quantity of water used after March 2020 was less than the quantity of water used before March 2020 were categorised as having had a decrease in the quantity of water used in the household due to COVID-19. Households for which the same quantity of water was used in the household before and after March 2020 were categorised 'same as before', and households whose quantity of water use after March 2020 was more than before March 2020 were categorised as having had an increase.

Independent variables
We created ten-year age groups for the age of the household head: 18-27, 28-37, 38-47, and 48þ. The survey collected information from both male-and female-headed households. Household heads were either currently married or not currently married. The 'not currently married' category comprised the widowed, separated, divorced and the never married. Households were categorised into two groups of household size: 1-4 or more than five household members. Educational attainment was grouped into four categories: none, primary, secondary, and tertiary or university. Household heads were asked to report on their main occupation. Responses were categorised into seven groups. For this study, household chores and all other employment are grouped together because of fewer cases.
Other categories of main occupation are none, farming, salaried employment, self-employment, casual labour and student.
We lumped together households whose main source of water was wells, spring, stream, or river because of the few cases. Other categories were piped water inside the household, piped water outside the household, harvested and borehole. Households were asked whether they pay for water to use in the household (Yes or No). Household heads were asked to state the average number of times they wash their hands, and households were categorised into two groups: 2-6 or 7-15 times. Household heads were asked to report on whether the quantity of water used in the household for cooking, washing dishes or clothes, cleaning the house, flushing the toilet, bathing, or washing hands after March 2020 increased, decreased or remained the same in relation to the quantity of water before March 2020.

Data analysis
Data analysis was performed using the STATA software. The distribution of household characteristics was presented at the univariate level of analysis. A Pearson-chi-square test was calculated at the bivariate level to test the association between selected household characteristics and a change in the quantity of water used in the household. A multinomial logistic regression model was fitted (because the outcome variable had more than two response categories) to examine the correlates of a change in water use among Ugandan households. Only variables that were significant at the bivariate level were included in the model.

Characteristics of households
The distribution of household characteristics in the study is shown in Table 1. The results show that most household heads were males (80%), currently married (75%), with secondary education (46%) and practiced farming as the main occupation (47%). About a third of household heads were in the age group 28-37 years, and most households had an average of 1-4 members (59%). Finally, most households in the sample were from the western region (29%).    The results in Table 2 are expected given the current state of the COVID-19 pandemic that calls for better hygiene.
Relationship between household characteristics and changes in quantity of water used since March 2020 Table 3 shows associations between selected household background characteristics and changes in the quantity of water used due to COVID-19 since March 2020. According to Table 3, most households (43%) reported an increase in the quantity of water used since March 2020 37% reported the, same as before March 2020 and 20% of households experienced a decrease. Table 3 shows that age, sex, current marital status, main occupation, and educational attainment of the household head, region of residence, household size, main water source and whether a household pays for water were all significantly associated with changes in the quantity of water used due to COVID-19 since March 2020. Table 4 shows an association between water utilisation and changes in the quantity of water used in the household due to COVID-19 since March 2020. Results from Table 4 indicate a significant association between handwashing, staying at home, cooking food, and washing dishes, cleaning the house, and washing dishes, bathing, and washing hands.
Correlates of changes in the quantity of water used in the household due to COVID-19 since March 2020    Given the results shown in Table 5, an interaction between current marital status and household size was introduced in the model (see Figure 1). The results shown in   Note: ** ¼ p < 0.05; *** ¼ p < 0.01.

DISCUSSION
The aim of this paper was to examine factors associated with water utilisation in Ugandan households before and after The results indicate that age, sex, education attainment, marital status, main water source, staying at home and main occupation of the household head, region of residence and average household size were significantly associated with water utilisation. The results in this study resonate with findings reported earlier by Nteli et al. (), who observed that higher educational attainment was associated with better handwashing practices. In this study, household heads with primary education were more likely to report a decrease in water utilisation than their counterparts with secondary education. While such a finding may imply that household heads with lower educational attainment may have poor hygienic practices, it is also possible that reduced water usage could have been associated with other underlying factors this study did not investigate. We contend that reduced use of water may imply water stress brought about disconnections, high costs, drought, or wastage through leakages. In situations where households are water-stressed, households can help out each other through water transfers Moreover, the increased water demand vis-à-vis a static system and water resources will have a ripple effect on the performance of a water supply system. In such situations it is appropriate to transition existing water supply networks to flexible systems which provide a good performance in the case of normal-situation supply as well as in emergency situations. Therefore, for regular water supply, a centralised supply system will have the option to be changed to a more decentralised system in cases of intermittent supply. Thus, the pressure differences in the network can be reduced and the equity of supply for all water users can be improved.
Flexible systems can provide tailored solutions to deal with water supply challenges during emergencies.
In addition, the paper provides a number of water supply intervention strategies. First, there is need to adapt water demand management strategies and innovative approaches to cope with the challenges for increased demand during emergencies. By applying the principles of water demand management strategies and innovative approaches such as integrated urban water management, impacts of increased demand in space and time to satisfy the water needs of a community at the lowest cost while minimising adverse environmental and social impacts can be met. However, technology innovations will have to be coupled with comprehensive system changes of the water supply systems.
Second, there is need to prioritise clean water supply in water-stressed areas such as low-resource environments, particularly in slums, congested areas, or rural areas. This is because most such areas do not have piped water and rely on seasonal harvested rain water or wells, or rivers, or streams. The implication is that in times of drought that may happen during an emergency, people may lack enough water, which may compromise hygiene. Adopting suggested approaches such as water transfer between households or harvesting rainwater can help water-stressed households and can boost economic development and reduce poverty.
Third, unlike salaried employees, seasonal workers such as casual labourers tend to be confined in low-resource settings that are water-stressed. During emergencies when they are not working, such people may not be able to afford to pay for watercalling for the government to either provide free water or water subsidies. Last, as a short-term strategy, the government can institutionalise the implementation and provision of boreholes in areas that do not have piped water to solve the problem of water scarcity. These water supply interventions can go a long way in ensuring that people have clean water to use during times of emergencies.

LIMITATIONS
Three main limitations emerge from this study. First, the dataset used in this study did not collect information that can be

RECOMMENDATIONS
Incorporating qualitative interviews to such quantitative data can provide hidden insights and help unmask reasons behind changes in water utilisation. Therefore, future studies can consider exploring reasons behind the patterns of water utilisation. Responses from such studies can shed light on