Identifying opportunities to improve piped water continuity and water system monitoring in Honduras, Nicaragua, and Panama: Evidence from Bayesian networks and regression analysis Ryan Cronk * , Jamie Bartram The Water Institute, The University of North Carolina, Chapel Hill, NC, United States article info Article history: Received 6 July 2017 Received in revised form 28 March 2018 Accepted 3 June 2018 Available online 4 June 2018 Keywords: Functionality Intermittent water supply (IWS) Safely managed drinking water Systems approach Sustainable development goals (SDGs) Sustainability abstract Many piped water systems in rural areas of Latin America and the Caribbean provide discontinuous service. In response to service delivery challenges, governments developed the Rural Water and Sani- tation Information System to monitor water service levels, infrastructure conditions, water committees, and technical assistance providers. Collected data are combined into metrics to represent water service sustainability. There is little analysis of these data to identify service delivery and system improvement opportunities and the sustainability metrics are unvalidated. Multivariable regression and Bayesian networks were used to analyze variables associated with the availability of 24-h water services using data from 5560 community-based piped water systems in Honduras, Nicaragua, and Panama. The regression models were compared to the sustainability metric. In Honduras and Nicaragua, the pro- portion of systems providing 24-h service spanned 71 percentage points between sub-national regions. Good condition infrastructure and year-round water source availability were associated with the avail- ability of 24-h service. The availability of support for system rehabilitation in Honduras and for pre- ventative maintenance in Nicaragua were associated with the availability of 24-h services. The Bayesian networks predicted that good condition infrastructure and year-round water source availability were more influential on the availability of 24-h service than management variables such as the availability of external technical support and funds to rehabilitate the system. In each country, insufficient household water tariffs were collected for 90% or more of systems to cover infrastructure, operations, and main- tenance costs. The r-squared values for the regression models ranged from 0.22 (Nicaragua) to 0.49 (Honduras) as compared to 0.05 (Nicaragua) to 0.03 (Honduras) for the sustainability metric e sug- gesting that regression models are better at predicting higher service levels. Rural water service oper- ators, technical assistance providers, local and national governments, and external support agencies could make better use of monitoring data by using interdisciplinary systems approaches to identify improvement opportunities to allocate technical and financial resources to systems with low service levels. © 2018 Elsevier Ltd. All rights reserved. 1. Introduction Continuous, sufficient, safe drinking water services are impor- tant for human health, human rights, well-being, and sustainable development (Bartram and Cairncross, 2010). They are urgently needed in rural areas of low- and middle-income countries (LMICs) of Latin American and the Caribbean (LAC) where water service levels are low (Bain et al., 2014a; b). More than 20 million people in rural areas of LAC (16% of the rural population) do not use an improved drinking water source and nearly 40.5 million people in rural areas of LAC (32% of the rural population of LAC) do not use piped drinking water at home (WHO/UNICEF, 2015). Many piped water systems in LMICs are discontinuous, providing less than 24-h of service per day (Kumpel and Nelson, 2016). Systems providing less than 24-h of service per day are more likely to contain fecal indicator bacteria (Kumpel and Nelson, 2013). An estimated 19% of water sources in LAC contain fecal contamination (Bain et al., 2014a,b). People with discontinuous * Corresponding author.148 Rosenau Hall, CB #7431.135Dauer Drive. Chapel Hill, NC, 27599-7431, United States. E-mail address: rcronk@live.unc.edu (R. Cronk). Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro https://doi.org/10.1016/j.jclepro.2018.06.017 0959-6526/© 2018 Elsevier Ltd. All rights reserved. Journal of Cleaner Production 196 (2018) 1e10