Bottled water brands are contaminated with multidrug resistant bacteria in Nairobi, Kenya

Background: The demand for drinking water has necessitated the proliferation of bottled water companies in Kenya. This study evaluated if retailed bottled water in Nairobi Kenya complies with both local and international reference criteria. Methods: A total of 42 different water brands (25 approved by Kenya Revenue Authority (KRA) and 17 banned brands) were analyzed for both physicochemical and bacteriological quality. The spread plate method was used to obtain the total plate count of bacteria, while the membrane filter method was used to obtain total coliform count (TCC) and fecal coliform count (FCC). Structured interviews were used to gather company-related information. Results: Overall, 16% of KRA-approved and 35.3% of banned bottled water were contaminated with heterotrophic bacteria. Of the approved water brands, 4% were positive for total coliforms, compared with 17% of the banned brands. Similarly, 4% and 17% approved and banned water brands were positive for fecal coliforms, respectively. Escherichia coli (19.1%), Pseudomonas spp. (9.5%) and Klebsiella spp. (4.8%) were the most common bacterial types isolated from all water brands, most of which exhibited multidrug resistance. In multivariable analysis, water companies that cleaned pipework and bottles using chlorine-based disinfectants (OR 0.08, 95% CI 0.01 to 0.8), those that had food safety programs (OR 0.1, 95% CI 0.019 to 0.9), had standard operating procedures (SOP) for water sourcing (OR 0.1, 95% CI 0.012 to 0.9) and SOP for contamination protection (OR 0.1, 95% CI 0.02 to 0.9) remained independently associated with bottled water brands exceeding WHO TCC limits. Conclusions: A number of bottled water brands were contaminated with one or more types of indicator bacteria, some of which were multidrug-resistant. Water bottling companies’ processes contribute to contamination. Rigorous regulation and monitoring will improve water quality and safety.


Introduction
Water is essential for the human body and mental functions 1-3 as well as for chronic disease prevention 4 . Water is essential for thermoregulation, protection and cushioning of body vital organs, as well as for breathing and transporting nutrients and oxygen throughout the body 2 . It is not surprising therefore that water constitute 50-60% of the human body 2 . Inevitably therefore, adequate total water intake of between 2 to 2.5 liters per day is recommended 2 .
Achieving and maintaining good health requires the availability and consumption of clean, potable (drinkable) water. This requires that water must be devoid of pathogens, dissolved toxins, and disagreeable turbidity, odor, color and taste 5 . The current concerns about palatability and microbial and chemical contaminants in tap water 6 , have led to the proliferation in the consumption of bottled water reaching historical high accounting for billion gallons in consumption 6 . Bottled water offers a handy source of water for consumption both within and outside household settings. In developing countries such as Kenya, bottled water is habitually sold and consumed in hotel industries, markets places, streets, schools, and during mass gatherings such as wedding and spotting activities, workplaces, health care facilities, and emergency situations 7 . Unfortunately, bottled water is not always as sterile as perceived. Several reports are available showing contamination bottled water with heterotrophic bacteria and coliforms counts exceeding the national and international standards 8,9 . Studies have isolated various bacterial contamination from bottled water such as Vibrio cholera and Salmonella spp. 10,11 , Pseudomonas spp., Acinetobacter spp., Citrobacter spp. and C. violaceum 9 . As a result, several waterborne illnesses such as diarrhea account for significant morbidity and mortality among the young and the aged as well as immunocompromised populations 12, 13 .
The bottled drinking water in Kenya should meet the following minimum requirements: be free from pathogens and chemicals; clear (i.e. low turbidity); none saline and should not have offensive taste or smell 14 . The Kenyan Bureau of standard (KS EAS 153: 2014) reference criteria for packaged water requires the absence of total coliforms, Escherichia coli, Staphylococcus aureus, Pseudomonas aeruginosa, Streptococcus faecalis, Shigella and Salmonella in 100 ml of water 14 . Microbiological contamination of drinking water can have an immediate and significant impact on human health and must therefore be analyzed frequently. Among the factors reported to influence the microbiological quality of bottled water include; material of bottles, color of bottles and the length of storage 15 . This study investigated the bacteriological quality of bottled water and the association with the processes and handling practices of water bottling companies sold in Nairobi Kenya.

Ethical statement
This study was approved by Kenyatta National Hospital and University of Nairobi Ethical Review Committee (KNH-UoN ERC-P971/12/2016). Before recruitment to this study, all patients study participants written informed consent for study participation.  16 to meet the United States Environmental Protection Agency (USEPA) and World Health Organization (WHO) requirement for drinking water standard of 100 total coliforms/ml water. Applying the formula for estimating the population proportion with specified relative precision described by Lemeshow et al. 17 , setting the α at 0.05, a total of 288 bottled water samples were collected to achieve 0.90 power. This number of bottled water samples was divided equally among the 40 brands sold in Nairobi. Therefore, a total of seven bottles per brand were sampled.

Data collection.
At the time of the study, due to availability, 25 water brands approved by KRA were purchased from major retail outlets in Nairobi. The other 17 water brands non-approved by KRA were purchased by the roadside or from small retail shops in the streets of Nairobi. All the brands (25 KRA-approved and 17 banned bottled water brands) were sought without preferential treatment of any brands or retail outlets. Seven bottles of each bottled water brand from the same batch were purchased at different retail outlets and shipped in cool box to the laboratory for microbiological analysis within 6 hours of purchasing.

Structured interviews.
To investigate the role of manufacturing handling and packaging process on the microbiological quality of water, randomly, this study visited the premises of all the available 25 registered and 17 banned water bottle and packaging companies located in Nairobi. Those consenting (see Extended data for the consent form 18) underwent a structured 20-30-minute face-to-face discussion within the premises at secluded and secured offices to gather information including the following information: type of abstraction, pipe work materials, bottling process, staff training, policies and procedures and microbiological quality of bottled water in Nairobi adopted from the WHO/UNICEF Joint Monitoring Programme) (see Extended data for a blank copy of the survey 18 ).

Amendments from Version 1
We have incorporated the reviewers' suggestions and recommendations in the new version. The title, data analysis, results, and discussion section have been reviewed to read better in line with the reviewers' suggestions.
Any further responses from the reviewers can be found at the end of the article

REVISED
Microbial and physicochemical quality of water samples The water temperature and pH were measured immediately after purchase using the HACH Sensionþ MM150 Portable Multi-Parameter Meter (Hach Company, Loveland, CO), according to the manufacturer's instructions.
Each of the seven water samples per brand were analyzed separately. Bacterial contamination in these water samples were achieved using total plate count by the spread plate method and total coliform count and fecal coliform count by membrane filter method as described by WHO, 19 . Briefly, 100 mL of water samples were filtered through a 0.22-µm-pore-size membrane filter (Millipore Corp., Bedford, MA), and filters placed on membrane Fecal Coliform (m-FC) agar plates were incubated at 37 and 44°C for 18 to 24 h to determine total coliform (TC) and fecal coliform (FC) counts, respectively.

Bacterial identification
The bacteria isolates were subsequently cultured onto bile esculin agar, eosine methylene blue agar, m-endo agar les, and plate count agar. These were then identified using colony morphology, Gram's staining, biochemical tests and further characterized using the VITEK 2 system, version 0.8.01 (bioMerieux, Inc., Hazelwood, MO).

Data analysis
Frequency (%), mean and standard deviation, were used to describe the water physiochemical properties and bacterial colony count. Chi-square or Fisher's exact test were used to test for variation between variables. The association between the presence of TCC >100 CFU/ml contaminating bottled water and companies water handling and processing characteristics were calculated using Poisson regression. Manual backward elimination method was used to reach the most parsimonious model in multivariate analysis. This included factors that were associated with contamination with TCC >100 CFU/ml at the significance level of P≤0.05. All statistical analyses were performed using STATA v13 (StataCorp LP, College Station, TX, USA).
Company-related factors associated with acceptability of water samples In multivariable analysis, bottled water brands that used chlorine-based disinfectants for cleaning pipework/tankers and bottling equipment were less likely to exceed WHO TCC limits compared to those that did not use any detergent for cleaning (OR 0.08, 95% CI 0.007 to 0.8). Companies that had food safety programs (OR 0.1, 95% CI 0.019 to 0.9), procedures for water sourcing (OR 0.1, 95% CI 0.012 to 0.9) and procedures for contamination protection (OR 0.1, 95% CI 0.02 to 0.9) ( Table 4). Self-reported company details are available as Underlying data 18 .

Discussion
Evaluation of bacteriological quality of bottled drinking water is important and urgent in Kenya given the current upsurge of different brands of bottled water, most of which are not regulated. This study was unique and among the first in Kenya to evaluate the role of the practices used by water bottling companies in relation to the bacterial quality of water in line with the WHO acceptability criteria. This was compared between those bottled waters approved and banned brands by Kenya Revenue Authority (KRA). The bacteriological quality of bottled water from approved brands was found to be better than those of banned brands. The total coliforms and fecal coliform present in 100 ml of water were detected cumulatively in 9.5% of all brands, and in 4% of KRA-approved and 17.6% of banned bottled water brands. The proportion of bottle water brands with unacceptable in line with WHO limits were lower than the 50% reported in Bangladesh 21 , 37.5% in India 22 , 26% reported in Nigeria 16 and 25% in Nepal 9 . On the contrary, the proportion of unacceptable bottled water brands in our study was higher than the 4.6% reported in Tanzania 23 the 9% in Sri Lanka 24 and 0% reported in Saudi Arabia 25 . Although KRA approval is based on tax payment rather than on scientific basis, the high number of KRA-banned bottled water brands points to the possibilities of ineffectiveness of the disinfection processes used in these brands. In a process likely to be mainly for financial benefit by the bottled water manufacturers, studies have cited the improper practice of filling the bottle directly from tap water and sealing it without any prior treatment as among the reasons responsible for higher brands of bottled water beyond the acceptable limits of bacteriological quality 9 . Longer storage periods, especially of already-contaminated bottled water, have been shown to worsen the bacteriological quality. As in many developing countries, the laxity by the government body responsible for monitoring the quality of bottled water has been shown to account for higher levels of bottled water brands with unacceptable microbiological limits 9 .
With regards to total plate count or heterotrophic bacteria, in this study, a total of 23.8% of the bottled water brands (40%    KRA-approved and 60% banned brands) were contaminated. In other settings, higher percentages of between 20% to 100% of heterotrophic bacteria contamination of bottled drinking water have been reported 9,26,27 . Studies have associated long storage duration with high levels of bacterial concentration mainly due to larger surface area for growth, higher temperature, and the nutrients arising in the container 28 . This quantity of heterotrophic bacteria is shown to correlate with water pH. There were 35.5% of banned bottled water brands with a pH below the pH 6.5 minimum level recommended by WHO, which could account for the higher numbers of heterotrophic bacteria per milliliter detected in these brands. Similar results were also reported by Pant et al. 9 .
The presence of total coliforms and fecal coliforms in 4% KRA-approved and 17.6% KRA-banned bottled water brands exceeding WHO criteria is similar to that observed by other studies 23,29 . This is worrying and a pointer to either poor water processing, introducing flakes of human skin or indigenously acquired by filling the bottles directly from the natural sources or taps 23  Although all bacteria in our investigation were susceptible to ceftriaxone and ofloxacin, resistance to erythromycin, trimethoprim-sulfamethoxazole, doxycycline, tetracycline, gentamycin, chloramphenicol, kanamycin and ciprofloxacin were noted. Multidrug resistance (resistant to more than three drugs) in E. coli, Klebsiella spp. and Pseudomonas spp. from all bottle brands were also detected. The presence of different species of bacteria, including multidrug resistant strains, in supposedly bacteria-free bottled water is of important public health problem. Pathogenicity notwithstanding, their presence in bottled waters heavily consumed by those including the elderly, children and the immunocompromised, the hazards of contamination, and health risks to consumers should not be taken for granted 10, 30 .
This study incorporated a unique feature by investigating the manufacturing practices potentially associated with contamination of bottled water. In multivariable analysis, companies that used chlorine-based disinfectants for cleaning pipework/tankers and bottling equipment were less likely to have water brands exceeding WHO TCC limits compared to that that did not use any detergent for cleaning. Zamberlan et al. 31 showed the importance of disinfection processes used by the water bottling companies as playing a key determinant of bacterial concentration in bottled water. In Nepal, Pant et al. 9 showed that failure to disinfect water represents an important avenue for bacterial entry and colonization of water processing systems.
Our study further showed that companies that had food safety programs, procedures for water sourcing and procedures for contamination protection were less likely produce bottled waters with unacceptable microbiological limits. As expected, if companies are set up in line with guidelines set by regulatory authorities, then the end product will be devoid or have a reduced microbial contamination. In this study, although more of companies producing approved brands had recommended water collection and transportation systems, water treatment procedures (filtration, UV and chlorination and reverse osmosis), packaged water using polycarbonate containers, used machine during bottling process, had batch tracing system, routinely tested their products according to the WHO guidelines and having recommended handling standard procedures than banned ones, these were not factors associated with the contamination of bottled water.
Our study had some limitations. First, due to limited resources available, the study could not process large enough numbers of the samples to include all brands sold in the country. Second, owing to the limited laboratory methods used, we were not able to identify all the potential pathogens that contaminated the water, including other pathogenic bacteria, viruses, fungi, and parasites. Third, the cross-sectional nature of our study only allowed us to describe associations between water company processes and procedures and bacterial quality and not a causal conclusion. Such outcomes can be confirmed in a longitudinal study. These limitations notwithstanding, one of the key outcomes of this investigation is the capacity to show that the perceived safe bottled water brands, including the top-selling and most expensive brands in Kenya, could be contaminated with bacteria beyond the WHO recommended limits. Additionally, some of these bacteria associated with significant disease outbreaks were multidrug-resistant. The study also showed that water bottling companies' operations and processes are key avenue for bacterial water contamination. The Kenya Bureau of Standards is the Kenyan regulatory and monitoring authority for all water and packaged foods. Our results may suggest, however, that concerted efforts must be made to improve the ability of national governments to properly regulate and monitor these products which has been shown to improve product quality and safety 32,33 . This manuscript discusses Bottled water brands contaminated with multidrug resistant bacteria in Nairobi, Kenya. Water is essential for the human body and mental functions. This requires that water must be devoid of pathogens, dissolved toxins, and disagreeable turbidity, odor, color and taste. In my opinion, it is a good article to be accepted for indexing in F1000Research only after doing the following comments:

Data availability
Comments: The English language of the paper should be improved. 1.
For the paper title I propose use "Contamination of bottled water brands with multidrug resistant bacteria in Nairobi, Kenya" instead.