Assessment of risks to the quality of water supplied in Bushenyi-Uganda using the water safety plan approach

Sustainable Development Goal (SDG) Target 6.1 aims at achieving universal and equitable access to safe and affordable drinking water for all by 2030 (United Nations 2015). To achieve this target, drinking water supply systems for both rural and urban populations should be properly sited, operated, and maintained at all times. Recent studies showed that this is not always the case (Kigsirisin et al. 2016; Kanyesigye et al. 2019; String et al. 2020). In 2012, the Joint Monitoring Programme (JMP) estimated that at least 1.8 billion people globally used a source of drinking water that was contaminated with faecal matter (WHO/UNICEF 2021). According to the JMP report of 2021, about 26% of the global population was still lacking access to safely managed drinking water services, and only 30% in sub-Saharan Africa had access to safely managed drinking water services (WHO/UNICEF 2021). The aforementioned suggests that a great proportion of the population was still exposed to the risks of contaminated drinking water sources. Consequently, JMP recommended that monitoring of water safety should include both water quality testing and risk management which is in line with the WHO recommendation for the adoption and use of water safety plans (WSPs) (WHO/UNICEF 2021). WSPs are the most effective means of consistently ensuring the safety of drinking water supply through the use of a comprehensive risk assessment and management approach that encompasses all steps in water supply from catchment to consumer (World Health Organization 2017).

Systematic preventive management of water sources, water treatment and distribution systems is the key to sustained safe drinking water supply instead of relying solely on end-product testing (Serio et al. 2021). Preparedness planning against the major natural hazards for which water supply systems worldwide are vulnerable is critical to ensure sustained safe water supply. Such natural hazards include floods, avalanches and extreme dry weather. These may impact reservoirs, lakes, pools and water towers which are relied upon as surface raw water sources. Disaster preparedness plans that rely on geographical information systems, reliability analysis tools and database queries have to be put in place by water utility managers and reviewed periodically. For water supply networks, pressure management programmes such as pressure zoning, night operation analysis and pressure regulation aimed at water and energy saving have to be put in place (Tsakiris et al. 2011). Preventive management employs appropriate control measures that are designed to reduce, eliminate or prevent the risk of contamination of the water supply (Memon et al. 2023). When such preventive management is adopted, monitoring can focus on a few simple parameters that can be determined in situ or in the laboratory (Pal et al. 2018). The reasons for testing are aimed at rapidly identifying where control is compromised, so as to identify actions to be taken immediately to rectify the system before contaminated water is distributed and consumed (Tsitsifli & Kanakoudis 2020). Application of preventive management as prescribed under WSPs has been carried out in several Ugandan urban centres managed by National Water and Sewerage Corporation (NWSC), the Uganda national urban water utility agency, including the Bushenyi-Ishaka area. The water supply risks in the Bushenyi-Ishaka area include seasonal quality and quantity changes at the raw water source, infrastructure, and Operational and Maintenance (O&M) deficiencies of the treatment and distribution system (Muhangane et al. 2017). In response to these identified risks, appropriate control measures have been progressively put in place to improve water production and quality in the distribution system (NWSC 2019).

This study was carried out to assess the effects of environmental and operational hazardous events on the water quality at all the stages of water supply management, i.e. source, treatment, storage and distribution system and the risk control measures taken in Bushenyi-Ishaka Municipality. The specific objectives were to: (i) identify the hazardous events and risks affecting the water treatment and distribution system and the risk control measures taken between July 2017 and November 2020; (ii) establish the water quality status and its historic trends from July 2013 to November 2020 and (iii) determine the relationship between the identified risks, their control measures and water quality status.

In order to assess the water quality trends throughout the water supply system and the associated risks, secondary data for the period July 2013–December 2017 were obtained. The secondary data were obtained from daily and weekly reports from the NWSC Bushenyi-Ishaka Area laboratory, monthly reports from NWSC Mbarara Regional Laboratory and NWSC Central Laboratory at Bugolobi, Kampala. The secondary data comprised of the following parameters: pH, turbidity (NTU), total iron (mg/L), free residual chlorine (mg/L) and faecal coliforms (CFU/100 mL). Water quality monitoring was carried out by sampling and laboratory analysis on a monthly basis during the period from January 2018 to November 2020. The water quality parameters considered for sampling and analysis in this study were pH, turbidity, total iron, free residual chlorine and faecal coliforms. pH and free residual chlorine were measured in situ while the rest were analyzed at the NWSC Mbarara Regional water laboratory. Sampling was done between 8:00 am and 12:00 noon for each of the sampling days, using pre-washed and dried plastic screw-caped 1-l bottles for physicochemical parameters. The ambient temperature in the field during sampling ranged between 21 and 23 °C. For microbiological faecal coliform samples, pre-washed, sterilized half-liter glass bottles were used. After sampling, the samples were packed in clean-dry sample carriers containing frozen ice parks to maintain cool temperatures (below 8 °C) and transported to the laboratory within 2 h, and analysis commenced immediately. The temperature in the laboratory during analysis ranged between 22 and 23 °C. Regarding hazardous events and risks, data were generated through field assessments of the catchment and source, treatment, distribution network, consumer premises and through review of operational and maintenance reports for the water treatment and distribution processes. The hazardous events, risks and control measures data were generated by the NWSC Bushenyi-Ishaka area WSP team as part of the WSP development from July 2017 to July 2018 and WSP implementation, which was started in August 2018. The control measures and their monitoring schedules were progressively passed on to the management for implementation through training and the provision of revised procedures and work instructions.

The water quality analysis was done using standard methods for the examination of water and wastewater (APHA, AWWA and WEF 2012). For pH, analysis was done in situ by applying test method ISO 10523:2008 (Water quality-Determination of pH), using a glass electrode (SensION HACH Model, USA). For turbidity, analysis was done in the laboratory using the ISO 7027 method that elaborates on the measurement of diffuse radiation for water of low turbidity. The analysis was done using an electronic turbidity meter (2100Q HACH model, USA). The turbidity meter was pre-calibrated using a formazin standard solution each time a sample was inserted and the displayed reading was taken in NTU units. For total iron, analysis was done in the laboratory using a direct reading spectrophotometer DR6000-HACH model, USA (Spectrophotometric determination of iron in water). The procedure involved measuring the blank of which deionized water was used. This was followed by preparing a calibration curve by measuring iron standards and then the iron sample itself, reporting results in mg/L. For chlorine (free residual), analysis was done in situ by applying the colourimetric test method (International StandardISO 7393-2), using the portable SMART3 colourimeter (LaMotte Model, USA). The procedure involved the addition of DPD tablets into a vial of the sampled water, causing a colour change to pink, then inserting the vial into the meter, reading the intensity of the colour change (by emitting a wavelength of light) and automatically determining and displaying the colour intensity (free residual chlorine) digitally in mg/L. For faecal coliform determination, the membrane filtration method using membrane lauryl sulphate broth was used for colony enumeration −8074, incubating for 18–20 h at 44.5°C. An isotherm forced convection laboratory incubator (ESCO, Malaysia) was used and the results were quantified as colony-forming units in 100 mL (CFU/100 mL).

Water quality data, both secondary and primary, were analyzed to ascertain the central tendency and temporal trends for the raw, final and distribution system sampling points in Bushenyi-Ishaka Municipality from July 2013 to November 2020. The data were checked for format uniformity and completeness. Preliminary data analysis was done using Microsoft Excel 2016, and further analysis using SPSS Version 23 (IBM SPSS Statistics, USA). Regarding hazardous events and risks, qualitative description of both was recorded and tabulated according to how they applied to the water supply system stages of the source, treatment and distribution chain including a combination of these unit operations and processes: clarification, filtration, disinfection, storage and the distribution network. For each water quality parameter, the mean and standard deviation were calculated. One-way ANOVA tests were run to determine the effect of monthly changes on water quality (refer to supplementary information). Bivariate comparisons of raw and final water pH at the Nyaruzinga treatment plant were done using the student's t-test, while for the comparison of faecal coliform concentrations between 2013–2014 and 2015–2020, the Mann–Whitney U test was used.

During the WSP development (July 2017–July 2018) and implementation (effective August 2018), hazardous events were identified, corresponding risks assessed, and control measures were determined and implemented in the immediate catchment and source, treatment and distribution system of Bushenyi-Ishaka Municipality. These are presented in Table 1.

There was a significant monthly effect on the faecal coliforms concentration (F(11,77) = 2.38, p = 0.014) at the Nyaruzinga raw water source, but no significant monthly effect on the total iron concentration, turbidity and pH values. For the final treated water, there was no significant monthly variation for all the parameters (pH, total iron, turbidity, free residual chlorine and faecal coliforms). Similarly, for the Katungu reservoir, there was no significant monthly variation in terms of the mentioned parameters. Apart from total iron concentration (F(11,77) = 2.16), p = 0.025) at Bumbeire PSP, there was no significant monthly variation of pH, turbidity, free residual chlorine and faecal coliforms at all the distribution points sampled during the study period.

There was a significant annual variation in turbidity, residual free chlorine and pH, but no significant variation in total iron and faecal coliforms at the final water sampling point. The annual variation at Katungu tank was significant in terms of turbidity, residual free chlorine and total iron but not significant in terms of faecal coliforms and pH. At Kyeitembe P/S, there was a significant annual variation in turbidity, total iron, residual free chlorine and pH, but not in terms of faecal coliforms. At Rwentuha PSP, Bumbeire PSP and Market PSP-Bushenyi, there was a significant annual variation in turbidity, total iron, residual free chlorine and pH, but not in terms of faecal coliforms. At Kizinda PSP, Market PSP-Ishaka and Kashenyi TC, there was a significant annual variation in turbidity, residual free chlorine and pH, but not in terms of total iron and faecal coliforms. At Basajjabalala SS, there was a significant annual variation in turbidity, residual free chlorine and total iron but not in terms of pH and faecal coliforms. The levels of significance are given under each of the parameters in subsections from 3.2.4 to 3.2.8.

There was a significant variation in water quality between all the sampling points in terms of turbidity, total iron, residual free chlorine and pH apart from faecal coliforms. The significant differences are given under each of the following parameters.

The raw water turbidity was lowest in July 2016 (9.2 NTU) and highest in December 2013 (203 NTU), with an average of 45.7 ± 44 NTU. Turbidity for the final water and water in the distribution system largely complied with the National Standards for treated drinking water (≤5 NTU) (UNBS 2015). The percentage of samples that had turbidity complying with the national standard for treated drinking water was 93.1% (827/888). High turbidity for final water was observed in in March 2018 and March 2019 (18.0 NTU) and at Kashenyi Trading Centre in April 2018 (14.8 NTU). There was a significant annual improvement in turbidity for the sampled locations namely, Nyaruzinga final water (F(7,81) = 2.80, p = 0.012), Katungu reservoir (F(7,81) = 2.13, p = 0.049), Market PSP-Bushenyi (F(7,80) = 3.19, p = 0.005), Kyeitembe P/S (F(7,81) = 4.65, p < 0.001), Bumbeire PSP (F(7,81) = 3.22, p = 0.005), Rwentuha PSP (F(7,81) = 4.30, p < 0.001), Kashenyi PSP (F(7,81) = 2.29, p = 0.035), Market PSP-Ishaka (F(7,80) = 4.04, p < 0.001), Basajjabalaba SS (F(7,81) = 6.37, p < 0.001) and Kizinda kiosk (F(7,81) = 3.35, p = 0.004). There was no significant monthly variation in turbidity at all the sampling points during the assessment. There was however a significant difference among the sampling points (n = 890, p = 0.012) during the assessment period (2013–2020).

The total iron concentration in the final water and water in the distribution system was largely compliant with the national standards for treated drinking water, i.e. ≤0.3 mg/L (UNBS 2015). Only 27 out of 890 samples (0.03%) from the final water and all the sampled distribution points did not meet the standard. There was no significant annual change in total iron concentration for the final water during the study period. There was however, a significant annual change at Katungu reservoir (F(7,81) = 6.00, p < 0.0001), Market PSP-Bushenyi (F(7,81) = 4.36, p = 0.0004), Kyetembe P/S (F(7,81) = 2.30, p = 0.034), Bumbeire PSP (F(7,81) = 3.19, p = 0.004), Rwentuha PSP (F(7,81) = 2.72, p = 0.014) and Basajjabalaba SS (F(7,81) = 4.63, p = 0.0002). The annual change in total iron concentration was however not significant at Market PSP-Ishaka, Kashenyi PSP and Kizinda kiosk. There was no significant monthly variation in total iron concentration at all the sampling points during the assessment. There was however a significant difference among the sampling points (n = 890, p < 0.0001) during the assessment period (2013–2020).

During the study period, the majority of the samples (719 out of 801 = 89.8%) taken from the storage and distribution systems had free residual chlorine complying with the national standards for treated drinking water (0.2–0.5 mg/L) (UNBS 2015). There was a significant drop in the concentration of the free residual chlorine from the final water to the Katungu reservoir (F(9,880) = 118.39, p < 0.0001) and the sampled distribution points). There was no detectable free residual chlorine (<0.01 mg/L) at Ishaka Market PSP in February 2018, Rwentuha PSP in February 2018, Kyeitembe P/S in May 2018, Bumbeire PSP in January 2019, Kizinda Kiosk in May 2019 and, Kashenyi PSP in March 2019. At these sampling points, free residual chlorine was well below the national standard for drinking water of 0.2–0.5 mg/L (UNBS 2015). Conversely, the free residual chlorine concentration was above the upper limit of the national standard (0.5 mg/L) once at Kizinda Kiosk (0.6 mg/L) in August 2014. Similarly, 19 out of the 89 samples (21.3%) from the Katungu reservoir had free residual chlorine above the upper limit of the national standard of 0.5 mg/L. There was a significant annual change in free residual chlorine at all the sampling points during the assessment period, as follows: Final water (F(7,81) = 12.85, p < 0.0001), Katungu reservoir (F(7,81) = 2.46, p = 0.0245), Market PSP-Bushenyi (F(7,81) = 2.66, p = 0.0158), Kyetembe P/S (F(7,81) = 6.34, p < 0.0001), Bumbeire PSP (F(7,81) = 2.97, p = 0.008), Rwentuha PSP (F(7,81) = 5.15, p = 0.0001) and Basajjabalaba SS (F(7,81) = 5.83, p = 0.0002), Kizinda PSP (F(7,81) = 3.38, p = 0.0033, Market PSP-Ishaka (F(7,81) = 3.64, p = 0.0018, Kashenyi PSP (F(7,81) = 4.87, p = 0.0001). There was no significant monthly variation in free residual chlorine among all the sampling points during the assessment. There was however a significant difference among the sampling points (n = 890, p < 0.0001) during the assessment period (2013–2020).

For final water and water in the distribution, out of 890 samples collected, 857 samples (96.3%) complied with the national standard for treated drinking water (<1 CFU/100 mL), while 33 did not. Of the 33 samples that did not comply, most (61%) had a concentration of 1 CFU/100 mL. All but one of the remainder violations (15 CFU/100 mL at Kizinda Kiosk in December 2019) were in the range of 2–10 CFU/100 mL taking place 1–3 times per year from 2014 to 2020. The water quality in the distribution system therefore met the bacteriological national standard for drinking water that is, being absent in 95% of yearly samples (UNBS 2015).

The pH of raw water was low as reported by previous studies (NEMA 2019; Twesigye 2021). The pH of the treated water was significantly lower than that of raw water likely followed by the reactions of the chemical inputs utilized in the treatment processes that is, the aluminium-based coagulants and chlorine disinfectant (Saritha et al. 2020). The increase in pH in the distribution system apart from Katungu reservoir and Bumbeire PSP resulted from the introduction of supplementary supply of water from the Kitagata treatment plant that produces water, which exhibits a relatively higher pH (6.5–6.9).

The declining (improving) raw water turbidity trend (Figure 4) during the entire study period could be due to progressive adherence to environmental regulations particularly reduced wetland encroachment which is attributed to law enforcement by local authorities (Muhame 2019), as encouraged through the WSP control measures (Table 1). Turbidity in the distribution system largely complied with the national standards, which is important for ensuring that the concentration of pathogenic microbial organisms, whose survival tends to be proportional to the level of turbidity, is kept as low as possible (World Health Organization 2017). The annual reduction of turbidity in the final water and water in the distribution system was observed to occur with improved operation and maintenance of the treatment plant and the network over the years, during the development and implementation of WSP, as indicated in Table 1. The control measures included preparing a schedule for network pipe replacement which was gradually followed according to budget, and preparing and following a schedule for network flushing, especially for dead ends and valley points.

The treatment process namely aeration and clarification was able to remove iron to levels within the treated national standard (Podgórni & Rząsa 2014). The challenge of significant annual variation in iron concentration across sampling points in the distribution system may be attributed to deficiencies in operation and maintenance effort, staff capacity and availability of resources (Muinamia 2015).

The presented results show that the distribution system was adequately protected against regrowth or recontamination with pathogenic microorganisms, as evidenced by 89.8% of the samples meeting the national standard for free residual chlorine (World Health Organisation 2017). That this is the case may be attributed to the development and implementation of WSP, particularly the risk control measures (Table 1) namely; following a schedule for network flushing, tanks cleaning, and scheduled replacement of old pipes and undersized pipes that are prone to bursts and leaks. The significant drop in the concentration of free residual chlorine from the final water to the storage and distribution points is expected owing to chlorine decay following from reaction with organic and inorganic matter (mainly presented as turbidity) in the pipes over time (Karikari & Ampofo 2013). The incidental lack of free residual chlorine at several points in the distribution system could have led to exposure to pathogenic regrowth or recontamination through back syphoning during times of low pressure in the network (Karikari & Ampofo 2013).

The presented results show that there was a significant decrease in annual median faecal coliform concentration in raw water from 2015 to 2020. The observed reduction likely follows from progressive adherence to environmental regulations, particularly reduced wetland encroachment. This is due to law enforcement campaigns by local authorities (Muhame 2019). Additionally, the decrease in faecal coliform concentration may be due to wetland restoration through community engagement and training on alternative livelihood options (UNDP 2019). The community engagement and training were partly done under the WSP development and implementation effectively starting in July 2017 as indicated in Table 1. We also observed that most of the water samples collected (96.3% over the 7 years of the study) in the distribution system met the national standard for faecal coliform concentration. That this is the case is likely as a result of the effectiveness of treatment, adequacy of free residual chlorine in the distribution network and good maintenance of the pipe network, particularly timely response to bursts and leaks (Tsitsifli & Kanakoudis 2020). The effectiveness in treatment and network management was part of the control measures instituted during the development and implementation of WSP as indicated in Table 1.

This study shows that there were significant monthly changes in faecal coliform concentration in the Nyaruzinga raw water. That this is the case could be attributed to soil erosion during the rainy seasons from agricultural activities around and alongside the wetland (NEMA 2019). Municipal wastes and domestic sewage disposal into the wetland may also have led to faecal contamination, especially during the rainy seasons (Safari et al. 2012). As noted in Figure 6, the escalating faecal coliform trend observed from 2013 to 2014 did not recur thereafter. This is likely attributable to the WSP control measures (Table 1), that were put in place, particularly holding sensitization meetings with farmers and local authorities that discouraged illegal farming activities, cattle grazing and bricklaying in the wetland and close to the raw water source and recommended alternative livelihood activities.

For treated water (final and distribution system), there was no monthly variation, apart from a few incidences, owing to the fact that the treatment process and network management maintained the water quality within the national standards. The significant annual water quality variation at all the sampling points could be attributed to the various control measures instituted by management to improve the quality and quantity of supply (Serio et al. 2021). The apparent significant water quality variation among the sampling points in the distribution system could be due to hazardous events that applied differently to the sampling point locations. For example, bursts and leaks could happen at one location in the network affecting the water quality in the local area. The same applies to inadequate pressure in the network, delayed repairs, delayed cleaning and flushing of dead ends (Tsitsifli & Kanakoudis 2020; Karikari & Ampofo 2013).

• The main hazardous events and risks affecting the quality of water supplied in Bushenyi-Ishaka Municipality include extreme seasonal weather variations, source protection gaps, treatment deficiencies and distribution management gaps.

• Management of the water treatment and distribution system was sound, largely supplying water that was in compliance with the national standards for treated drinking water as follows: turbidity (93%), total iron (99%), free residual chlorine (90%) and faecal coliforms (96%). However, the pH of the treated water in the distribution system, range of 5.5 ± 1.0–6.1 ± 0.06, was below the national standard (6.5–8.5) from July 2013 to July 2018 until supplementary supply from the Kitagata water treatment plant was introduced in August 2018, attaining a range of 6.1 ± 0.07–6.7 ± 0.07.

• The quality of water in the distribution system largely complying with the national standards is attributed to adherence to control measures that were instituted against the risks in the treatment plant and the distribution network.

• For sustained system and quality assurance, it is recommended that stakeholder engagement is intensified to minimize activities leading to soil erosion and faecal contamination from the catchment.

• Efforts to improve treatment processes need to be emphasized, particularly clarifier and filter performance and chemical dose optimization. Provision of appropriate coagulant types, optimum dose determination and application is recommended. Regular resanding and periodic top-up of fine sand media is recommended for sustained treated water quality improvement.

• There is a need for additional investment in network management, particularly old pipe replacement, scheduled cleaning and timely response to bursts and leaks.

Further research is ongoing to evaluate the overall performance of the WSP of Bushenyi-Ishaka Municipality in terms of outcomes and short-term impacts.

The authors would like to acknowledge the contributions of the following who contributed towards the planning, data collection, data analysis and manuscript preparation. Eng. Francis Kateeba, General Manager, NWSC Mbarara, who by then was the Area Manager, NWC Busheyi, and rallied his section heads to carry out WSP development and implementation. Also acknowledged is Deo Mugume, Senior Quality Control Officer at NWSC Bushenyi, who helped with sampling and in situ analysis; John Owona, Philbert Bankunda, Deo Mugume and Dora Kukundakwe who were members of the Area WSP team that spearheaded the field exercise of hazard identification, risk assessment and determination of control measures. Acknowledgement also goes to Profilio Tebandeke, Principal Quality Control Officer at NWSC, Mbarara and head of the Regional water quality laboratory at Ruharo, together with Naboth Mwebaze and Rita Kamiti, who helped carry out some water quality analysis. The authors also acknowledge the contribution of Fred Muwanika who helped with statistical data analysis.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.