Beyond the faucet: social-geographic disparities and trends in intermittent water supply in Peru

Safe water consumption has become a major public health concern in recent decades due to environmental problems such as climate change, water stress, and freshwater pollution, which have led to water scarcity in various regions of the world (World Health Organization 2022; Organización de las Naciones Unidas para la Educación la Ciencia y la Cultura 2023). Although Sustainable Development Goal 6.1 seeks to ensure access to water and sanitation for the entire population in a safe, continuous (24 hours a day for 7 days a week), efficient, and quality manner (United Nations 2023), one of the most common problems that prevents the fulfillment of this goal is intermittent (<24 hours a day) piped water supply, known as intermittent water supply (IWS) (Simukonda et al. 2018). This problem has an unfavorable effect due to variations in water pressure, low quantity of water supplied, high costs, and adverse health effects due to the risk of microbiological contamination (Ghorpade et al. 2021). Globally, it is estimated that over one billion people rely on IWS, with the highest proportion and shortest daily supply in low- and middle-income countries. These regions face economic constraints that hinder the upgrading and maintenance of infrastructure, and unplanned urban growth that fosters disorganized expansion of drinking water distribution networks (Kumpel & Nelson 2016; Bivins et al. 2017; Beard & Mitlin 2021).

In Latin America and the Caribbean (LAC), the proportion of households supplied with piped water varies from 12.5% in the Dominican Republic to 85.9% in Panama (Deshpande et al. 2020). However, an estimated 28 million people in LAC have an IWS, with an average duration of 16 hours per day (Kumpel & Nelson 2016). Although several strategies have been proposed to improve access to safe and continuous water consumption in LAC, there are problems related to rapid urbanization, inadequate water management and governance, poor infrastructure, and contamination of water resources that promote water scarcity and generate an unequal distribution of this natural resource (Rodríguez et al. 2022). Previous studies conducted in this region report that IWS has negative effects on water quality and supply infrastructure (Erickson et al. 2017; Rubino et al. 2018; Molinero et al. 2021). In fact, IWS causes a higher concentration of heavy metals in the water due to corrosion of the pipes, a higher concentration of microorganisms such as Escherichia coli, and chlorine concentrations below acceptable levels (Erickson et al. 2017; Rubino et al. 2018; Molinero et al. 2021). Despite the negative consequences of IWS, the World Health Organization notes that LAC is one of the regions with one of the lowest investments in improving and restoring water supply infrastructure and improving access to water (World Health Organization 2022).

In Peru, the main cities belong to the coastal region and are located in desert areas characterized by semi-arid climates, where water supply comes from river basins that depend on glaciers and rainfall in the Andes region (Son et al. 2020). These characteristics increase the susceptibility of the region to the consequences of climate change and natural disasters, which has led to a decrease in water availability (Banco Mundial 2023). Although the proportion of piped water supply in Peru increased from 69.9% in 2000 to 81.6% in 2017 (Deshpande et al. 2020), on average, 25% of Peruvian households experienced an IWS with an average of 16 h of water supplied per day between 2010 and 2014 (Rawas et al. 2020). However, little is known on IWS in the last 5 years, and the main characteristics of households that most frequently have this type of supply. Therefore, we aimed to estimate the proportion of households with IWS in the national territory and by department during the last 5 years (2017–2022) and determine the association between sociogeographic characteristics of households and IWS using the most recent year available (2022).

We used data collected by the National Household Survey (ENAHO) during the years 2017, 2018, 2019, 2019, 2021, and 2022 to conduct a secondary data analysis. These years were selected because the question on the number of hours of water supply has been included in the ENAHO survey since 2017. The exclusion of 2020 was due to the use of an abbreviated questionnaire during the COVID-19 pandemic, which omitted the relevant questions.

The ENAHO is an annual nationwide survey designed to collect socioeconomic data in Peru (Instituto Nacional de Estadística e Informática 2022b). The study population includes all urban and rural households, and the survey uses a probabilistic, stratified, multistage, and area-independent sampling that is representative at the national, departmental, geographic, and urban/rural levels. The primary sampling units are rural population centers with <2,000 inhabitants and urban population centers with ≥2,000 inhabitants (Instituto Nacional de Estadística e Informática 2022b). The secondary sampling units are clusters of 120 households on average. Finally, the third sampling unit corresponds to private households (Instituto Nacional de Estadística e Informática 2022b).

The ENAHO includes four questionnaires: (i) living conditions and poverty; (ii) governance, democracy and corruption; (iii) income of agricultural producers; and (iv) income of the self-employed (Instituto Nacional de Estadística e Informática 2022b). These questionnaires generate several databases containing information related to housing and household characteristics, characteristics of household members, education, health, employment, access to social programs, citizen participation, and summary (containing variables related to household expenditure and poverty), among others. For this study, two household-level databases were used: ‘Sumaria’ (Summary) and ‘Características de la vivienda y hogar’ (Housing and household characteristics). These databases were merged for each year of the study, linking households through the variables: cluster, housing and household. After including households that completed the interview and with complete data on the variables of interest, the total sample was 29,018 households for 2017; 31,641 for 2018; 29,636 for 2019; 28,601 for 2021; and 29,705 for 2022.

While there is no consensus on the definition of IWS (Galaitsi et al. 2016), we define ‘Intermittent water supply’, when a household receives piped water supply service less than 24 hours a day, one or more days of the week (International Water Association 2023). This variable was constructed from the variable p110 (water supply in your household comes from), selecting households with a supply by public network inside or outside the household or public use pylon. Regarding the latter supply, it is defined as the use of a tap or standpipe by the household, located in the street or other public area, regardless of how the water is distributed or stored in the household (Instituto Nacional de Estadística e Informática 2023). Households with IWS were defined as those where it was reported that water was not supplied every day of the week or those where an affirmative response was given regarding the daily water availability (p110c) but with a supply duration of less than 24 hours per day (p110c1).

In addition, for descriptive purposes, we created the variable ‘Number of hours with water supply’, a quantitative variable from 1 to 24 from the variable p110c1 (‘How many hours per day?’ and in Spanish ‘Cuántas horas al día?’).

We included four exposure variables: area of residence, geographic region, district population density, and district human development index (HDI). These variables were selected on the basis of evidence of their association with water supply (Adams 2018; Rondinel-Oviedo & Sarmiento-Pastor 2020; Satpathy & Jha 2022), and on the basis of the differences of these socio-geographical characteristics with the dependent variable (Instituto Nacional de Estadística e Informática 2023).

The urban or rural area of residence was created from the variable ‘stratum’ which contains information on the number of inhabitants of the area where the household is located, being recategorized as urban when the area has 2,000 or more inhabitants and rural when the area has fewer than 2,000 inhabitants or corresponds to rural population centers.

For this study, the geographic region was divided into four categories: Metropolitan Lima, Coast without Metropolitan Lima, Highlands, and Jungle. Metropolitan Lima, which corresponds to the capital of Peru, is located in the coastal area of Peru bordering the Pacific Ocean and is made up mostly of desert areas crossed by rivers that come from the Peruvian Andes and flow into the Pacific Ocean (Ministerio del Ambiente 2021). Metropolitan Lima is home to over 10 million people. The Coast without Metropolitan Lima refers to the coastal region of Peru excluding the area of Metropolitan Lima area and is characterized by medium-sized cities. Highlands is a strip of land that is made up of mountain ranges known as the Andes and high-altitude areas. Jungle is contiguous to the Highlands region and is where the Amazon River begins. The Jungle is made up of forest areas and is characterized by its rich biodiversity and natural resources, including fauna, flora, minerals, and navigable rivers (Ministerio del Ambiente 2021).

The construction of the variable ‘population density of the district’ was based on data published by The National Center for Strategic Planning (CEPLAN), which considered the total population projected for the year 2020 and the territorial extension of the districts of Peru (Centro Nacional de Planeamiento Estratégico 2023). Based on these data, population density tertiles were created: low, medium, and high. The highest tertile groups, the districts with a high population density.

In relation to the district HDI 2019, the same source of population density was used (Centro Nacional de Planeamiento Estratégico 2023), which is constructed based on three indicators: life expectancy at birth, proportion of the population over 18 with secondary education, years of education, and per capita household income. Moreover, HDI values close to 1 indicate a better human development position in the territory. District HDI tertiles were created from these data and three categories were established: low, medium, and high. The highest tertile groups, the districts with the best HDI.

The household poverty level is calculated by the National Statistics Institute (INEI – acronym in Spanish) and reported in the database. It is characterized as non-poor, non-extremely poor, and extremely poor according to household expenditure. In our study, we generated a dichotomous variable considered as non-poor and poor (extreme poor and non-extreme poor groups).

Data analysis was performed using R software, version 4.2.3, and RStudio, version 2023.06.1. We included only complete cases for the variables of interest in the analysis and all analyses were weighted to account for the complex sample design of each year of the ENAHO.

To illustrate the relationships between water supply sources according to areas of residence (urban and rural), a chord diagram was created using RStudio. Descriptive analysis was performed to examine the variables studied. Line plots were generated to plot the number of hours of water supply per year in households with daily piped water supply, and the average number of hours of water supply in households with IWS for each year of study for each department. In addition, the proportion of households with IWS by year of study for each department was represented by a heat map.

Next, to evaluate the association between exposures and outcome, generalized linear regression analyses of the Poisson family with log link were used. For each exposure, a bivariate regression model was constructed, adjusting for household poverty level and housing location (area, geographic region, and department). Prevalence ratios (PR) and their confidence intervals (95% CI) are reported to estimate the association between exposures and outcome. Two-tailed p-values less than 0.05 were considered statistically significant.

Ethics committee approval was not required for this study, as it involved the analysis of secondary, publicly available data that does not permit the identification of participants. The databases used are available on the INEI web portal (https://proyectos.inei.gob.pe/microdatos/) by accessing the ‘Consulta por Encuestas’ tab and selecting the ‘ENAHO Metodología ACTUALIZADA’ data for the corresponding year.

Table 3 shows four crude and adjusted models considering the geographic (area of residence and geographic region) and district (population density and HDI) characteristics of households in the year 2022 as exposure variables. The models that evaluated the area of residence, population density, and HDI were adjusted for the poverty level of the household, the natural region where the household was located, and the department, while the model referring to the geographic region was adjusted for the poverty level of the household and the department. As for the geographical variables, households located in an urban area had an 83% higher prevalence of IWS (adjusted PR [aPR]: 1.83 [95% CI: 1.67–2.01]) compared with rural areas. Based on the administrative region, households in the Coast region had a 47% higher prevalence of IWS (aPR: 1.47 [95% CI: 1.37–1.59]), while the Highlands region and Metropolitan Lima had a 26% (aPR: 0.74 [95% CI: 0.67–0.82]) and a 62% (aPR: 0.38 [95% CI: 0.32–0.44]) lower prevalence of IWS compared with the Jungle region. On the other hand, within the district characteristics, households located in districts belonging to the middle tertile of population density had a 14% higher prevalence of IWS (aPR: 1.14 [95% CI: 1.07–1.21]), while households belonging to the highest tertile had a 42% lower prevalence of IWS (aPR: 0.58 [95% CI: 0.51–0.67]) compared with the low tertile. Finally, when it comes to HDI, households in districts with a middle HDI tertile had a 9% higher prevalence of IWS (aPR: 1.09 [95% CI: 1.02–1.16]), while households in the highest tertile had a 65% lower prevalence of IWS (aPR: 0.35 [95% CI: 0.29–0.42]) compared with the low tertile.

According to our findings, while the proportion of households with IWS changed between 2017 and 2022, it remained above 40% in 2022. In addition, in households with IWS, the average duration of water supply did not exceed 8 h in all the ENAHO years included. At the departmental level, the departments with the highest proportion of IWS were Tumbes, Ica, Piura, Loreto, La Libertad, Lambayeque, and Ucayali, while the departments with the shortest duration of water supply were La Libertad, Ica, Madre de Dios, Loreto, Puno, and Tumbes. Finally, households located in an urban area, the Coast region, and in districts that belonged to a middle tertile of population density and HDI had a higher prevalence of IWS compared with their counterparts in 2022.

In the last 5-year period, approximately 4 out of 10 Peruvian households supplied with piped water were found to have IWS, with an average duration of water supply that did not exceed 8 hours per day across study years. This result is lower than those reported in Asia (50%), LAC (60%), Southeast Asia (90%), and India (100%) (Zyoud 2022), but it is similar to that reported in a study conducted in South Africa (39%) (Loubser et al. 2021). Despite the differences between the proportions of IWS, our result on the average duration of water supply is different from that reported in East Asia and Pacific (16.7 h), Europe and Central Asia (13 h), LAC and North America (16 h), Sub-Saharan Africa (12.8 h), and the world average duration (12.5 h) (Kumpel & Nelson 2016), but it is similar to the average duration reported in South Asia. These differences could be due to the time horizon of the studies, the representativeness of the estimates, and the characteristics of low- and middle-income countries, where IWS is one of the most common forms of water supply due to a limited number of water networks and low supply coverage (Kaminsky & Kumpel 2018). In Peru, a previous study noted that 24.8% of households reported that water was not available for 24 h in the last 2 weeks between 2010 and 2014 (Rawas et al. 2020). This figure is lower than that reported in our study probably due to the dissimilar definition of IWS, and the time horizon included in the study. Our results could be attributed to the unequal distribution of water throughout the territory due to poor management and fractional water management, as well as climate change that has generated natural disasters in desert areas, such as floods, droughts, and pollution that promote water scarcity, especially in the coastal region (OECD 2021).

Tumbes, Ica, Piura, Loreto, La Libertad, Lambayeque, and Ucayali had the highest proportions of IWS, while the departments of La Libertad, Ica, Madre de Dios, Loreto, Puno, and Tumbes had the lowest average duration of water supply. The departments of Tumbes, Ica, Piura, La Libertad, and Lambayeque belong to the Coastal region of Peru, which is characterized by approximately 58% of the Peruvian population and semi-arid climates (Instituto Nacional de Estadística e Informática 2017; Son et al. 2020). In this region, it is hypothesized that the reduction in household water supply may be linked to the melting of glaciers in the Andes, and the consequent decline in meltwater supply due to climate change (Instituto Nacional de Estadística e Informática 2022a). However, this relationship warrants further investigation in future studies. Moreover, the Coastal departments present higher urbanization compared with the Highlands and Jungle regions, generating insufficient coverage of public water networks to supply the rapid urban growth, especially in regions located in the periphery of cities, as well as slums, and low-income households (Beard & Mitlin 2021). Furthermore, rapid urbanization in coastal areas leads to increased water use for economic activities and greater contamination of this resource that prevents its reuse (Dos Santos et al. 2017; Heidari et al. 2021). Thus, our study indicates that households located in the coastal region (with the exception of Metropolitan Lima) and in urban areas had a higher prevalence of IWS in 2022. On the other hand, the departments of Ucayali, Madre de Dios, Loreto, and Puno belong to the Jungle and Highlands regions of Peru, and have the lowest proportions of public water services compared with the Coastal region (Instituto Nacional de Estadística e Informática 2022a). According to the National Meteorological and Hydrological Service of Peru (SENAMHI), the Jungle and Highlands regions (particularly Puno) have a high probability of water flow deficit, which would lead to an increase in droughts and a decrease in water supply in these regions (Servicio Nacional de Meteorología e Hidrología del Perú 2023).

Additionally, our study shows that households located in districts belonging to a middle tertile of population density had a higher prevalence of IWS in 2022. Water availability is related to population size and growth, with regions with lower population density having lower water availability due to changes in freshwater resources, increased demand due to demographic changes, and lack of water supply facilities (Tzanakakis et al. 2020). On the other hand, households located in districts with a middle HDI tertile had a higher prevalence of IWS. There is a positive correlation between access to drinking water and HDI (Amorocho-Daza et al. 2023). Specifically, households located in districts with a low HDI tertile are more likely to have a lack of infrastructure and low quality of water supply, and low government capacity to cover water supply equitably (Kopp et al. 2021).

Although the National Water Resources Management System seeks to ensure adequate water management in Peru (Ministerio de Desarrollo Agrario y Riego 2014), our study indicates that 40% of Peruvian households have IWS, reflecting inadequate water management at different governmental levels. Therefore, water supply in a context of water scarcity in Peru should target the factors that increase the prevalence of IWS, above all, to decrease the inappropriate use of the natural resource. In addition, national and regional government institutions should redouble efforts to improve the infrastructure, quality, and continuity of water supply in the Coast and Jungle departments, which have the highest proportions of IWS and have experienced the worst consequences of climate change due to floods, droughts, and freshwater contamination. In addition, water supply conditions must be improved in order to comply with the United Nations Sustainable Development Goal 6.1, which seeks to ensure a safe and equitable water supply in the national territory (United Nations 2023).

The main strength of our study is the use of a nationally representative sample of households. In addition, the present study reflects, at the national and subnational levels, one of the main water supply problems worldwide. However, our study has limitations. First, the IWS variable was self-reported by respondents, which could be subject to recall bias because they involve specific events that have occurred in the past. This limitation may be further influenced by the presence of home water storage; households with storage, particularly those connected to the distribution network, might not even be aware of when water is actually being delivered. Moreover, this could lead to non-differential misclassification of the outcome, which may underestimate the effect estimates toward the null hypothesis. Second, all the ENAHO sample sizes vary over time; however, our results included sample weights, stratum, and sampling units specific to each year included. Third, the amount of missing data, especially in 2021, could lead to selection bias. However, the missing data represent only 4.2% of the sample included that year, which likely would not have a substantial impact on the study's results. Fourth, there could be imprecision in establishing how many hours households receive water, which could lead to an inadequate determination of outcome. Fifth, seasonal variability in river flows could affect water uptake at treatment plants and influence the presence of IWS. Furthermore, due to Peru's extensive territory and diverse geography, different seasons can be experienced across departments based on their geographic region. This geographical and seasonal variability might impact water supply differently across the country. Sixth, the occurrence of natural disasters and the absence of rainfall could have a significant impact on water supply in the coastal region that could introduce biases in the results. Finally, although the population density variable and HDI of the last available year were used, it is unlikely that these variables had significant variations in the study period. Additionally, the associations estimated in that year are cross-sectional in nature, so causality cannot be established due to the lack of temporality in the variables.

Our study showed a non-significant reduction in IWS from 42.5% in 2017 to 41.9% in 2022, which has remained unchanged over the study period. Furthermore, we found that the departments of Tumbes, Ica, Piura, Loreto, La Libertad, Lambayeque, and Ucayali presented the highest proportions of IWS, reflecting unequal water supply and the consequences of climate change in the Coast and Jungle regions. Additionally, there are geographic (urban area and Coast region) and district (population density and HDI) factors that increase the probability of IWS in households. In this sense, government institutions should reinforce strategies aimed at improving infrastructure and water supply to achieve a permanent and equitable supply throughout the country.

The data that support the findings of this study are available from the Instituto Nacional de Estadística e Informática of Peru (https://proyectos.inei.gob.pe/microdatos/).

The authors declare there is no conflict.