Physicochemical and microbial quality of bottled drinking water sold in Embakasi Central, Nairobi, Kenya Open Access

Ensuring access to safe drinking water for all is the primary aim of the 2030 Agenda for Sustainable Development. However, this goal is far from being achieved in most developing countries, including populations that rely on bottled water (WHO/UNICEF JMP, 2017). The rapid increase in the production of packaged drinking water introduces the possibility of distributing products that are unfit for human consumption (WHO 2016). For example, the consumption of contaminated packaged water in India has resulted in outbreaks of waterborne illnesses, such as cholera, typhoid, and hepatitis (Gangil et al. 2013; Joseph et al., 2018). A study conducted in Sri Lanka reported that out of the 22 brands of bottled drinking water sampled, 9 and 14% contained bacterial and fungal contaminants, respectively (Sasikaran et al. 2012). Sachet water often does not meet applicable standards for microbiological water safety (Fisher et al. 2015; Manjaya et al. 2019). Studies have also documented high levels of fecal contamination in bottled drinking water at the point of sale (Kassenga 2007; Keleb et al. 2022), including multidrug-resistant bacteria (Mohamed et al. 2020). These and other studies demonstrate that packaged drinking water fails to consistently meet international quality guidelines, especially pertaining to microbiological safety standards.

Inconsistent safety of bottled water supplies suggests that suppliers are failing to identify and address risks and hazards across the water supply chain from source to consumer (Peter et al. forthcoming). A study regarding bottled water in Dar es Salaam, Tanzania, found that improper storage conditions, an unhygienic environment, and high temperatures provided insufficient protection against bacterial regrowth and were likely the primary factors causing deterioration in bacteriological quality (Kassenga 2007). According to a study conducted in Kitale, Kenya, the presence of Escherichia coli in two brands of bottled water was attributed to the use of an ineffective purification method, improper implementation of quality control programs, and inadequate sanitary practices (Adaro et al. 2017). Thus, the absence of a multi-barrier approach and insufficient oversight led to the introduction of contamination into the bottled water supply chain.

Studies show that consumers' packaged water purchases are strongly influenced by cultural, social, personal, and psychological factors. Consumer perceptions that packaged water offers better safety, taste, and quality compared to tap water are the primary drivers of the growing demand for bottled water (Ward et al., 2009; Hu et al. 2011). In many developed countries, including the United States, health and safety concerns have significantly increased the consumption of bottled water (Napier & Kodner 2008). According to a study conducted in Surinam, demographic and psychological factors affect bottled water buying decisions (Durga 2010; Armstrong et al. 2014). In Iran, the growing demand for bottled water has given rise to numerous small-scale entrepreneurs engaged in production and distribution (Khaniki et al. 2010). Hence, multiple social factors drive the increased consumption of bottled drinking water globally, and this trend is likely to continue in the years ahead.

The ongoing expansion of packaged drinking water production and its unrestricted consumption pose concerns regarding public health. At the time of preparing this article, we were unaware of any published report on the quality of bottled drinking water in Embakasi Central sub-county. Disease outbreaks in the area have been attributed to contaminated bottled drinking water in the media (Modern Mom Kenya 2019). Due to the existing knowledge gaps on the effectiveness of local treatment processes, manufacturing, and handling practices, the extent of potential contamination hazards remains unknown and poses a unique threat due to the widespread distribution of packaged water in the county. Thus, due to concerns raised about the quality of bottled drinking water locally, the primary objective of this study was to carry out independent testing of the major bottled water brands sold in Embakasi Central. The sub-objectives of this study were to (a) assess microbial contamination levels (E. coli and heterotrophic bacterial concentrations), (b) assess physicochemical parameters of concern (including lead, nitrite, nitrate, fluoride, and pH), and (c) test associations between bottled water brands and contamination occurence.

In addition, 10% (14) more samples were collected to account for potentially lost or damaged samples or laboratory errors. Hence, the minimum sample size required was 153.

Samples were collected from a total of 8 brands, of which only 1 was listed among the 15 brands licensed to package and sell water in the study area. Hence, the study represents a subset of the brands available for sale in the study area, with most of the sampled brands likely originating from outside Embakasi Central. The production and supply process for all brands involved the following steps: reservoir or tank storage, treatment plant, packaging, storage in warehouses, transportation to vendors by truck, and finally storage and sale in shops and minimarts. The majority of these vendors were small with limited ventilation. Some samples were collected from street vendors who sold water from coolers or carts along the roadside. Under these conditions, the bottles were often exposed to sunlight, heat, dust, and debris.

A total of 158 water samples were collected in duplicate, for a total of 306 water samples. The bottles of water purchased for analysis had a capacity of 500 mL, and each duplicate had identical batch and lot numbers. The pro forma recorded the brand name, manufacturing and expiry dates, batch number, Kenya Bureau of Standards (KEBS) symbol, mineral content, water purification method, and manufacturing location. Each sample was preserved in its original sealed container, with eight brands anonymously labeled with letters A–H. Bottled water samples were stored in cooler boxes with ice packs and transported to the Kenya Medical Research Institute (KEMRI) Centre for Public Health Research for bacteriological analysis and the Government Chemist for physicochemical analysis. Samples were immediately analyzed for microbial and physiochemical parameters per the methods described in the American Public Health Association (Rice et al. 2012). In case of delay in processing, the samples were stored at 4 °C for no more than the prescribed holding time.

Microbial analysis: The membrane filtration method was used for bacteriological analysis using nutrient agar as the culture medium for heterotrophic plate count (HPC) and Sorbitol MacConkey Agar (Oxoid, Hampshire, UK) as the culture medium for E. coli. Bottled water samples of 100 mL each were filtrated through a membrane filter and incubated at 37 °C and 35 °C for 18–24 h for HPC and E. coli, respectively. To determine the reliability and validity of the laboratory methods, every 10th sample was analyzed in duplicate, and negative and positive controls were processed on each day of the analysis. In addition, to determine the sterility of the incubator and laboratory, an open and closed plate prepared with both Nutrient Agar and Sorbitol MacConkey Agar was left in the two incubators and a working bench for 18–24 h. After incubation, E. coli were identified based on the pink color and morphology of the colony, and HPC was enumerated by counting all colonies.

Samples were diluted based on the electrical conductivity and were aspirated to determine absorbance for sodium, potassium, calcium, and magnesium testing. The parameters were assessed and documented using a flame atomic absorption (AA) spectrometer, including an air acetylene flame along with a hollow cathode lamp specifically for magnesium and calcium. Sodium and potassium were assessed using the flame atomic-emission spectroscopy method. Samples were automatically sampled and processed, and readings were recorded.

To determine nitrate and nitrite concentrations, the content of one NitraVer 5 Nitrate reagent powder pillow was added to the sample cell using the Cadmium Reduction Method by Hach. The sample cell was swirled vigorously for 1 min to dissolve the powder. For nitrate and nitrite readings, the sample cell was left for 5 and 20 min, respectively, for the reaction to take place. Following a blank reading, the water sample was placed in the sample cell, and the reading was taken.

To determine sulfate concentration, the sample cell was filled with the bottled water sample to the 10 mL mark. The content of one SulfaVer 4 reagent powder pillow was added to the sample cell. The sample cell was swirled vigorously to dissolve the powder. The sample cell was left for 5 min for the reaction to take place. Following a blank reading, the water sample was placed in the sample cell, and the sulfate reading was taken.

Fluoride was determined using the colorimetric method. The fluoride level was determined based on decolorizing action, proportional to the quantity of fluoride on the complex of zirconium. Using color comparison, fluoride concentration was reported in mg/L. Lead was tested by an AA spectrophotometer equipped with a graphite furnace. The water sample was dispensed into the AA autosampler. Using the Perkin-Elmer Model 1100B, a Fisher element hollow cathode lead lamp was used for the determination of lead concentration.

All study data were compiled in Microsoft Excel Version 16.87 (Redmond, WA, USA), and water quality results were analyzed for basic descriptive statistics (mean, standard deviation, median, and range). Bivariate statistical analyses (Chi-square test of independence and Pearson's correlation) were performed in IBM SPSS Version 20 (Armonk, NY, USA) to test for associations between contamination frequencies and brand types.

Ethical approval for the research protocol was granted by the KEMRI Scientific and Ethics Review Unit (Protocol No.: KEMRI/RES/7/3/1) and from Eawag's internal ethical review committee (Protocol 1609_22072020). Permission to carry out the study was received from the National Commission for Science, Technology and Innovation, NACOSTI (REF: 164989) and the Nairobi County Government: (REF: EOP/NMS/HS/7/VOL.1/RS/12).

The quality standard for bottled drinking water is defined as the absence of detectable E. coli in a drinking water sample (WASREB 2008). The results of the analyses of HPC and E. coli contamination for the eight brands of bottled drinking water are summarized in Table 1. The maximum HPC was 1,000 colony forming units (CFU)/mL, and the maximum E. coli concentration was 250 CFU/mL. Across all brands, the mean (SD) HPC concentration was 503.0 (464.5) CFU/100 mL and the median was 395.0 CFU/100 mL (n = 158). The mean (SD) E. coli concentration was 2.4 (20.3) CFU/100 mL and the median was 0 CFU/mL (n = 158). In total, 19 of the 158 samples (12%) had detectable E. coli and were, therefore, in violation of national standards for microbial safety (WASREB 2008). Seven out of eight brands had at least one sample with E. coli contamination, with Brands 2 and 5 having a quarter or more samples each with detectable E. coli.

The results of the physicochemical analysis are summarized in Table 2. Among all samples, 13% had elevated levels of fluoride. The mean fluoride concentration for Brand 5 was 3.53 (SD = 0.57) mg/L, above the 1.5 mg/L threshold as defined by the national regulatory authority (WASREB 2008). Four of the eight brands had at least one sample with nitrate levels above the 10 mg/L limit, with a maximum of 43.0 mg/L (Brand 7). Our laboratory analysis did not reveal elevated levels for any other inorganic parameter of health or aesthetic concern per national guidelines (see the Supplementary Material for a summary of results for all inorganic parameters). However, 59 of the 158 samples (37%) had a pH level outside of the guideline range of 6.5–8.5, indicating a risk of operational ineffectiveness in the treatment process. For one brand (Brand 3), all samples were below the recommended range for pH.

A chi-square test of independence revealed that the proportion of samples impacted by contamination was statistically different by brand (χ² (7, N = 158) = 77.03, p < 0.001). However, there was no significant association between samples falling outside of the guideline range for pH and E. coli presence (χ² (1, N = 158) = 0.21, p = 0.65). Pearson correlation tests indicated that there was a significant positive association between pH level and both fluoride concentration (r(157) = 0.60, p < 0.001) and nitrate concentration (r(157) = 0.16, p = 0.039). Moreover, there was a significant positive association between HPC and fluoride concentration (r(157) = 0.22, p = 0.006).

This study is novel in being the first to investigate a range of quality indicators for popular bottled water brands sold in Kenya. These findings contribute to the limited evidence base on packaged drinking water quality, especially for physicochemical parameters that are not often assessed. The majority of parameters tested, including lead, conformed to Water Services Regulatory Board (WASREB) guideline values (Table S1). However, the main insight arising from this study is that nearly all brands investigated had at least one sample impacted by a contaminant of health concern. E. coli was detected in seven out of eight brands sampled in this study, similar to levels observed elsewhere (Kassenga 2007). A quarter of all samples were impacted by fluoride, nitrate, and/or E. coli at the point of sale. Even the top-selling bottled water brands in Kenya, which are widely perceived to be safe, had samples exceeding national guidelines for microbial safety (and some being characterized as high risk). The presence of chemical and fecal contamination in bottled water that is marketed as safe and wholesome should be an important concern for all, especially those tasked with the protection of public health.

Our results also revealed that certain brands were more heavily impacted than others, with fluoride being a contamination of particular concern. All 20 samples for Brand 5 exceeded the national guideline value for fluoride of 1.5 mg/L. Chronic exposure to elevated fluoride levels is linked to dental and skeletal fluorosis, among other conditions (Shaji et al. 2024). While investigating the source of contamination is not within the scope of the present study, it is likely that the fluoride detected originated from the natural bedrock. The recommended treatment technology for water containing high levels of fluoride is activated alumina or advanced treatment such as reverse osmosis (Peter et al. forthcoming). Brand 5 claims its bottled water undergoes treatment by reverse osmosis prior to packaging, yet our results point to a breakdown of this process.

Due to the varying quality of source water, a multi-barrier treatment process is recommended. The treatment processes reportedly used for the bottled water in this study included ultrafiltration, ultraviolet radiation, ozonation, and reverse osmosis. While these technologies are known to be effective and widely accepted, weak or lacking Hazard Analysis Critical Control Points (HACCP) systems may leave bottled water production plants vulnerable to contamination events (Kokkinakis et al. 2008). In addition, hygienic conditions must be consistently monitored and maintained throughout the handling, transport, storage, and sale of the finished product. Rapid deterioration of the chemical and microbiological quality may occur at any stage, particularly if bottles are subjected to direct sunlight or stored in unhygienic conditions (Dodoo et al. 2006; Ahimah & Ofosu 2012; Oyelude & Ahenkorah 2012; Danso-Boateng & Frimpong 2013; Duwiejuah et al. 2013; Addo et al. 2020). Hence, national regulatory agencies should put in place rigorous control measures such as the routine inspection of treatment and packaging plant processes, along with routine monitoring of the entire packaged drinking water industry.

Taken together, these results indicate a need for manufacturers to adhere to more stringent measures for the production of bottled water and for WASREB to better enforce these regulations. Regular operational monitoring and regulatory oversight of drinking water treatment processes is a necessary safeguard for public health. This includes evaluating the effectiveness of control measures within drinking water systems, establishing appropriate limits for these measures, regularly monitoring those limits, and taking prompt corrective actions if any deviations are detected before the water quality becomes compromised. Our results suggest a breakdown in the periodic audit of drinking water processing and packaging plants, which aim to ensure that operational controls are in place and functional and compliance measures are being maintained.

A comprehensive risk-based framework can improve the planning, operation, and management of water sources and supply. Assessment tools such as Quantitative Microbial Risk Assessment (QMRA) can be used to determine if the existing barriers against specific pathogens are sufficient. Potential interventions such as improving and maintaining a cold chain for packaged water and supporting small-scale operators to manage drinking water risks may improve the situation in Nairobi. These programs can be implemented by local or national governments, private companies, or nonprofit organizations. Supporting programs should include several activities: increasing awareness, knowledge building (education), stakeholder engagement, resource and training availability, research to identify adequate measures, and programs to protect water from contamination.

The study showed that, as a result of weak standards enforcement, the source to the consumer chain of the bottled water industry in Kenya remains vulnerable to contamination. The national regulatory bodies should hold the bottled water industry to standards of transparency in terms of where water comes from and how it is treated. Currently, consumers remain uninformed about potential quality issues because water companies are not obligated to disclose their water testing results. This assessment should take the perspective of multi-barrier protection, that is, more frequently sampling across the entire water supply chain from source to plant to distributor to sales point to consumer. Without an understanding of the entire chain, it becomes difficult for brands to identify and proactively address the causes of contamination. Questions that must be answered concurrently are: Who bears the cost of the additional assessments? And how to balance protection of public health with a need for a profitable water enterprise?

To bring bottled water up to standard, we recommend the following:

• (1) Assessments of water quality should encompass a broader array of parameters and additional indicators associated with drinking water services and their effects, especially those posing risks to vulnerable populations. Regulators must review the scope of existing water management and water safety plans to include priority chemical contaminants such as fluoride.

• (2) Regulators should improve sanitation surveillance activities and occupational hygiene protocols at water manufacturing plants. More importantly, water treatment facilities should be encouraged to implement HACCP programs to guarantee the safety and security of their supplies.

• (3) Stakeholders should discuss and implement steps to ensure the safety of bottled drinking water. The increased communication and collaboration of local water safety planning (WSP) teams is a vital step for establishing contextualized risk-based frameworks. Communication procedures and emergency response plans for nonroutine incidents or emergencies will ensure effective WSP implementation.

• (4) Increased training among bottling plant employees, plant operators, distributors, and vendors will improve the knowledge and understanding of hygienic practices in the packaging, transport, and handling process. Specific training programs using flyers, healthcare visits, community walks, songs or theater, community meetings, and radio or television messages may be implemented, especially for small-scale treatment operators and informal vendors.

• (5) Development and implementation of standard operating procedures will mark increased effectiveness of HACCP programs, as well as preventing recontamination during storage and handling. These monitoring procedures will lead to timely corrective actions where needed through water quality testing.

Our study had some limitations. Due to a limited laboratory budget, we were dependent on indicator organisms and were not able to test for specific waterborne pathogens that have the potential to contaminate the water, including the enterotoxigenic strains of E. coli. As our study was cross-sectional in nature and we did not use randomization techniques for selecting sale points or brands, the extent to which the findings can be generalized to the broader sub-county population is likely limited. In addition, our study design did not allow for making conclusive causal statements about the driving factors of water quality deterioration. Future research should explore causal relationships through longitudinal studies.

This cross-sectional study confirmed the presence of chemical and microbial contaminants in packaged drinking water at risk levels that require mitigation. These results provide evidence that the consumption of bottled drinking water in Embakasi Central may be a contributing factor to local disease outbreaks and background rates of diseases in non-outbreak situations. The risk of waterborne diseases associated with the consumption of bottled water in this setting is likely to remain in the absence of improved oversight and control measures.

This evidence serves as the basis for revisiting regulatory processes concerning bottled drinking water. At a global level, acknowledging that bottled drinking water is an integral and growing aspect of the water supply landscape will help one to focus attention on the aspects of the sector that need scrutiny. Furthermore, this study acknowledges the critical role that bottled water currently fulfills in offering access to improved water sources for low-income urban populations, a matter of particular urgency within the context of Kenya. Looking ahead, there is a need for further discussion and concerted effort in support of suppliers implementing effective WSP programs, and national governments properly regulating and monitoring bottled drinking water for the protection of public health.

The Centre for Public Health Research (CPHR), Kenya Medical Research Institute (KEMRI), and The Government Chemist, Kenya, provided access to laboratory facilities to complete this study. Daniel Boit, Grace Njenga, and Liz Sisenda assisted with field sampling and laboratory analysis. Erastus Muniu provided helpful feedback on earlier versions of the manuscript. Senteu Nchoe and Peter Ongalo contributed to the graphics in this paper.

GM contributed to conceptualization, methodology, data collection, formal analysis, investigation, and draft manuscript preparation. ZB contributed to conceptualization, methodology, formal analysis, investigation, reviewing and editing the writing, and supervision. GMK contributed to conceptualization, methodology, formal analysis, investigation, review and editing the writing, and supervision. PN contributed to the investigation. SJM contributed to project administration, reviewing and editing the writing, supervision, and funding acquisition. All authors reviewed the results and approved the final version of the manuscript.

This research was supported by the Eawag Partnership Programme (EPP) and the Sandec Department at Eawag. The views expressed and information contained in this manuscript are not necessarily those of or endorsed by these groups, which can accept no responsibility for such views or information or for any reliance placed upon them.

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

The authors declare there is no conflict.