A study was undertaken in Njoro Township, Kenya to evaluate the extent to which drinking water was subjected to post-collection faecal contamination in low-income and high-income households. Boreholes were the main drinking water sources, accounting for roughly 70% singular access. The microbial quality of drinking water from the boreholes deteriorated from the point-of-collection through conveying containers of small-scale water vendors to household storage containers, irrespective of their income status. The densities of Escherichia coli (EC) were relatively low at the point-of-collection – median (M): 18 CFU/100 mL, range (R): 0–220, n = 60 – increasing considerably in the containers of water vendors (M: 290 CFU/100 mL, R: 30–350) and slightly (M: 360 CFU/100 mL, R: 0–520) between vendors and low-income households, many of whom used the services of vendors unlike high-income households who relied on a piped system on premises (M: 40 CFU/100 mL, R: 0–500). Post-collection contamination was high in low-income households compared to high-income households but differences were not significant between the two household categories with and without household water treatment (HWT). Different HWT methods in the two household categories significantly reduced faecal contamination, but unhygienic handling and poor storage practices afterwards caused recontamination. HWT and behavioural change measures need not selectively target household groups solely on the basis of their income status.

INTRODUCTION

In the last two decades, major milestones have been achieved on improving coverage for drinking water supply in both urban and rural settings in many developing countries. Over 70% of the global progress made on access to improved sources of drinking water has been achieved through gaining access to piped drinking water on premises (WHO/UNICEF 2013). However, more than three billion people still obtain water from non-piped systems and public standpipes for their domestic needs (WHO/UNICEF 2013). This means drinking water is collected in containers from a point-of-collection, transported manually, and in most cases, stored and handled for prolonged periods in households, thereby increasing the risk of microbiological contamination. Although there is considerable variability in the results of studies on post-collection faecal contamination, stored domestic water usually contains higher levels of microbial contamination than the same water at its source due to unhygienic handling and storage practices (Sobsey 2002; Wright et al. 2004; Levy et al. 2008).

Compared to piped systems, improved non-piped systems such as boreholes and protected springs have a huge potential for accelerating access to drinking water due to the ease with which they can be constructed and scaled-up. They are heavily promoted and represent a significant proportion of the systems providing drinking water in rural settings and low-income urban areas (Fewtrell et al. 2005). However, a large proportion of these so-called improved sources may provide water that is microbiologically or chemically contaminated at the point-of-collection or at the point-of-consumption (Bain et al. 2012; Onda et al. 2012). Under current international norms, drinking water from improved non-piped systems is normally not subjected to post-collection protective measures, unlike treated piped water which is normally required to contain prescribed levels of residual disinfectants (Clasen & Bastable 2003). For instance, guidelines for non-piped systems are usually expressed in terms of technical control measures at the point-of-collection while maintaining quality at the point-of-consumption is the sole responsibility of the beneficiary.

Guaranteeing the safety of drinking water at the point-of-consumption remains a major challenge, especially in communities where collection and household storage is required before use. Most interventions in this area have focused on measures which prevent or minimize post-collection contamination to ensure acceptable quality at the point-of-consumption. With respect to microbial water quality, two complementary approaches are promoted, namely household water treatment (HWT) and behavioural change on household hygiene and sanitation advancing the MDG No. 7c to halve, by 2015, the proportion of the 1990s total world population without access to safe drinking water and basic sanitation (Rosa & Clasen 2010). Concurrently, advocacy and educational training on behavioural change have targeted hygiene and sanitation to improve collection, transport and safe storage of drinking water. Unusually, HWT and behavioural change interventions selectively target low-income households with the notion that safe drinking water is generally more susceptible to post-collection faecal contamination in low-income households compared to high-income households.

Thus, the main objective of this study was to evaluate the extent to which drinking water was subjected to post-collection faecal contamination in low-income and high-income households in Njoro Township, a rural community in south-western Kenya.

MATERIALS AND METHODS

Description of the study area

Njoro Township is an agricultural settlement within Njoro River Catchment in south-western Kenya (Figure 1). It is the headquarter town of Njoro sub-county in Nakuru county, about 20 km west south-west of Nakuru Town at 0 ° 19′S and 35 ° 56′E. The population of Njoro Township is approximately 30,000, with approximately 6,000 households [www.majidata.go.ke (accessed March 2014)]. The main water sources are groundwater (boreholes), rain water and surface water from Njoro River, which flows through the township. Rainfall (600–1,800 mm/year) is harvested during the wet season, which occurs in May–September and November–December. The main dry season occurs in January–April, with a short dry spell in October–November.

Figure 1

Location of Njoro Township within Njoro River Catchment in south-western Kenya.

Figure 1

Location of Njoro Township within Njoro River Catchment in south-western Kenya.

During the dry season, residents of Njoro Township experience serious water constraints with long queues for water at vending outlets and a considerable increase in the number of people fetching water from Njoro River, a polluted stream (Yillia et al. 2008). Correspondingly, the peak incidence of waterborne diseases coincides with the dry season, with a prevalence rate of 1.7% and typhoid fever leading with 49% of all reported illness cases and gastroenteritis with 29% (Chabalala & Mamo 2001). Yet another serious water-related health concern in the area is the high concentration (10 mg/L) of natural fluoride in the groundwater, which occurs in excess of the WHO recommended minimum concentration (1.5 mg/L) in drinking water (WHO 2011).

Two types of on-site sanitation systems are common in Njoro Township; flush toilets with private septic tanks and pit latrines with or without a slab. Most high-income households usually have a pit latrine within the yard in addition to a flush toilet inside the house. Pit latrines form the most common type of sanitation system in the township. They can be shared or unshared facilities. A 2010 pilot survey in the township reported sharing among 76% of low-income households (Macharia 2011). Sharing is common among neighbouring low-income households, especially among households in the same yard. There are no sewer connections and no sewerage treatment plant in the township. Therefore, excreta and sewage disposal is privately managed by households.

Household survey

Out of 240 households that were targeted, 214 (85%) participated in the household survey. Participating households were categorized into high-income and low-income categories (Table 1) based on the sum of daily income of occupants in gainful employment, as well as access to water and sanitation as ranked by the drinking water and sanitation ladders of WHO/UNICEF Joint Monitoring Programme (JMP) (WHO/UNICEF 2013). This information was corroborated with site-specific observations to stratify the township along seven estates for the household survey – one exclusively high-income, three exclusively low-income and three with mixed income households.

Table 1

Household categorization based on household income and access to water and sanitation

ParametersHigh-income householdsLow-income households
Daily household income 21 US$ (range = 15–60) 1.3 US$ (range = 0.5–3.0) 
Occupants in gainful employment, are in well-paid jobs (mostly blue-collar jobs) Occupants in gainful employment, are in poorly paid jobs (mostly farm labourers) 
Includes proprietors of medium businesses Includes proprietors of mostly small businesses 
Drinking water supply Piped borne water supplied directly into household (yard and/or house) Households use public taps, some manual transport/delivery required 
None used water from the stream Households used stream water 
Rain water collected and stored in concrete tanks/overhead containers Rain water collected and stored in small containers and buckets 
Aspects of sanitation types Flush toilet with private septic tank No flush toilet 
Private toilets and latrines not shared Pit latrine with/without a slab 
Ventilated improved pit latrine Pit latrine shared with others 
ParametersHigh-income householdsLow-income households
Daily household income 21 US$ (range = 15–60) 1.3 US$ (range = 0.5–3.0) 
Occupants in gainful employment, are in well-paid jobs (mostly blue-collar jobs) Occupants in gainful employment, are in poorly paid jobs (mostly farm labourers) 
Includes proprietors of medium businesses Includes proprietors of mostly small businesses 
Drinking water supply Piped borne water supplied directly into household (yard and/or house) Households use public taps, some manual transport/delivery required 
None used water from the stream Households used stream water 
Rain water collected and stored in concrete tanks/overhead containers Rain water collected and stored in small containers and buckets 
Aspects of sanitation types Flush toilet with private septic tank No flush toilet 
Private toilets and latrines not shared Pit latrine with/without a slab 
Ventilated improved pit latrine Pit latrine shared with others 

The household survey was undertaken by public health personnel responsible for Njoro District. They were prepared by one of the authors (P.W.M.) at a workshop on practical matters for administering the questionnaire. The questionnaire with a list of pre-arranged questions was administered via face-to-face interviews with household representatives, 98% of them being women aged between 25 and 50 years. They were required to provide information on household demography, income from gainful employment and access to water and sanitation. Respondents were also required to provide information on HWT practices, as well as handling and storage practices – length of storage, types of storage containers used and the methods used for extraction.

Collection of water samples

Table 2 shows the total number of water samples targeted from various contamination points in the township. Sampling was undertaken over a period of 60 days from October to December in 2010. There were six sampling episodes in total, one episode every 6–10 days continuously. All water samples were collected in sterile containers (1,000 mL) and transported to the laboratory in a cooling box with ice cubes within 6 hours of collecting the first sample. After each water sample was collected, in situ measurements were made of physico-chemical parameters (dissolved oxygen, temperature, pH, conductivity and total dissolved solids) in a receiving container specifically set aside for this purpose. Water samples were obtained from: (i) water sources (point-of-collection); (ii) the conveying containers of small-scale water vendors; and (iii) the storage containers of low-income and high-income households (Figure 2).

Table 2

Total number of water samples targeted from various contamination points in the township

Water source (contamination point)Sampling episodesSampling pointsSample typeNumber of samples
Rain water 10 60 
River water 
Borehole water 2a 72 
Public standpipe 2a 12 
Vendors 6b 36 
High-income households 10 5c 300 
Low-income households 10 5c 300 
Total number of samples    810 
Water source (contamination point)Sampling episodesSampling pointsSample typeNumber of samples
Rain water 10 60 
River water 
Borehole water 2a 72 
Public standpipe 2a 12 
Vendors 6b 36 
High-income households 10 5c 300 
Low-income households 10 5c 300 
Total number of samples    810 

aWith/without hose attached to tap.

bSix different vendors.

cFive water types identified collectively for all the households.

Figure 2

An outline of the drinking water distribution chain in Njoro Township. Each box represents a potential contamination point from which samples were obtained; HWT: household water treatment.

Figure 2

An outline of the drinking water distribution chain in Njoro Township. Each box represents a potential contamination point from which samples were obtained; HWT: household water treatment.

Water sources (point-of-collection)

Three main water sources were sampled, i.e., boreholes, rain water and stream water (Figure 2). Six private boreholes and one public standpipe were sampled. Each borehole had a tap connected to an overhead storage reservoir into which raw groundwater from approximately 200 m had been pumped. The public standpipe received treated drinking water from Njoro Water Works, which also supplies water through piped connections directly to the yards of mostly high- and middle-income households. Each borehole tap and the standpipe had a hose attached to it. First, drinking water was sampled through the hose in the same way as it was collected by people at the vending outlets. Next, water samples were obtained after removing the hose and flaming the opening of the tap. Stream water was collected from Njoro River at Njoro Bridge, the main water abstraction point along the stream. Sampling was done from the left bank of the stream at the same location from which most residents fetch water for domestic use. Rain water samples were collected from a selection of participating households with overhead storage containers attached to the roof of their houses.

Conveying containers of small-scale water vendors

Six small-scale water vendors at different borehole vending outlets were selected randomly. Water samples were taken from one representative conveying container (usually 20 L jerry-cans). A different vendor was approached during each of the six sampling episodes to increase the number of vendors in the sampling matrix and boost the chances of capturing variations in the water quality data during collection, transport and delivery by vendors. This would be missed by targeting only one vendor repeatedly.

Household storage containers (point-of-consumption)

Ten low-income and ten high-income households which participated in the household survey were selected for microbial examination of the water stored in their homes. Samples were obtained with the dippers used in each household. As household water was stored according to the water source and whether or not drinking water was subjected to HWT, samples were separated into five household water types: (i) groundwater from boreholes, including water from Njoro Water Works; (ii) harvested rain water; (iii) surface water from the stream; (iv) mixed water (i.e., when water from more than one source was stored in the same container); and (v) household-treated water from any of the water types listed above. Characterization of household water types was based on consultation with household representatives and confirmation with measurements of conductivity and total dissolved solids, the two parameters being conservative markers that were measured on site.

Laboratory analysis of water samples

Laboratory analysis of water samples was carried out at Egerton University, Department of Biological Sciences following standard analytical procedures prescribed by American Public Health Association for water and wastewater analysis (APHA 1999). The membrane filtration method was used to estimate total coliform (TC), Escherichia coli (EC), intestinal Enterococci (IE) and Salmonella, and the pour-plate method was used for heterotrophic plate count bacteria (HPC). For each sample, appropriate dilutions (two to three dilution steps) were made in replicates. Replicate dilutions for TC, EC, IE and Salmonella were drained through sterile membrane filters (0.45 μm; 47 mm). Filters for EC and TC were placed on Chromocult Coliform Agar (MERCK) and incubated at 37 °C for 24 hours. Slanetz & Bartley medium (mEnterococcus Agar) (Oxiod CM0377) was used for IE with incubation at 37 °C for 48 hours, whereas Salmonella Chromogenic Agar Base (Oxoid CM 1007) was used to recover Salmonella at 37 °C incubation for 24 hours. Dark-blue to violet colonies were counted for EC, pink colonies for TC in addition to EC colonies, whereas reddish brown colonies were enumerated for IE and pink colonies for Salmonella. Enumerations for EC, TC, IE and Salmonella were reported as colony forming units (CFU) per 100 mL. For HPC, 1 mL each of each replicate dilution was mixed with liquefied Yeast Extract Agar (Oxoid CM0019) and allowed to solidify. The plates were then incubated at 37°C for 48 hours. Countable colonies for HPC were reported as CFU per mL. Clostridium perfringens (CP) was examined using Fluorocult-Tryptose Sulfite Cycloserine (F-TSC). To select for CP spores, water samples were pre-heated at 75 °C for 15 minutes in a water bath before filtration. After which, filters were placed on F-TSC agar plates, put in anaerobic jars containing Anaerocult A and an Anaero-strip anaerobic system, and incubated at 44 °C for 24 hours in a dry incubator. All black glowing colonies upon illumination under ultraviolet light at 360 nm were reported as CP per 100 mL.

Statistical analysis

All statistical analyses were performed using SPSS® (version 10.0), Sigma-Plot® (version 9.0) and Microsoft Excel® (version 2010). Non-parametric statistics was applied to account for the zero inflated and right censored distribution of bacteria densities. The Wilcoxon rank test was applied for dependent samples, i.e., comparisons of related water samples from: (i) source; (ii) vendors' containers; and (iii) household storage containers. In contrast, the Kruskal–Wallis test was applied for independent samples, i.e., comparisons between unrelated water samples from: (i) different sources; (ii) different household water types within a group of households; and (iii) different household water types in different groups of households.

RESULTS AND DISCUSSION

Water supply and microbial quality of water sources

The main water sources used in Njoro Township are: (i) harvested rain water from roof tops; (ii) surface water from a nearby stream (Njoro River); and (iii) groundwater from several boreholes. Boreholes were the main source of domestic water, accounting for roughly 70% singular access (borehole only). Over 92% of high-income households had an in-house piped-water system or at least a piped system on premises. However, drinking water was supplied only once or twice a week, thereby requiring prolonged storage before consumption. The average daily per capita water use was highly variable for low-income and high-income households, and significantly different (p < 0.05) for the two categories of households. Daily per capita water use ranged from 59 to 121 L/person/day (l.p.d) for high-income households and 10–34 l.p.d. for low-income households. The average per capita daily water use for low-income households was 19 l.p.d., which was, incidentally, slightly lower than the minimum daily water requirement (20 l.p.d.) recommended by the WHO (2011). Two-thirds of high-income households who harvested rain water stored it in concrete tanks and/or overhead containers attached to the collecting roof. The remaining high-income households and most low-income households harvested and stored rain water in smaller containers, usually 20 L jerry-cans, buckets or metal drums, which lasted for a short duration after collection. No high-income household used water from the stream, which was used instead by many low-income households.

Generally, water samples obtained from the boreholes had relatively lower bacteria counts compared to harvested rain water or stream water samples (Table 3). HPC, TC, EC and IE in stream samples were 1–2 log units higher compared to water samples from boreholes or harvested rain water. Except for CP, which was not recovered from any of the water sources, bacteria indicators in stream water samples were significantly higher (p < 0.05) than those collected from the two other water sources. In particular, whereas Salmonella was recovered in all stream samples (median = 4 CFU/100 mL; range = 2–19), it was undetected in samples collected from harvested rain water and samples from the standpipe operated by Njoro Water Works (Table 3). Similarly, no Salmonella was detected in any but one of the boreholes. Samples obtained from that borehole with a hose attached to the mouth of the tap were notoriously contaminated, with Salmonella detection (median = 0 CFU/100 mL; range = 0–3) in 50% of samples plus elevated levels in other bacteria parameters. However, no Salmonella was recovered from the same borehole after removing the hose and flaming the mouth of the tap, which indicated that Salmonella contamination was coming from the hose and not from the well.

Table 3

Median levels (range in parentheses) of bacteria densities in samples from different water sources including results of statistical tests (letters in superscript) comparing water sources

Water sourcesHPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)Salmonella (CFU/100 mL)
Public standpipe* 5,000 (4,000–9,000)a 40 (40–170)a 1 (0–1)a 1 (1–3)a 0 (0–0)a 
Borehole (without hose)Δ 1,900 (1,900–3,300)a 118 (118–850)b 16 (0–130)b 22 (10–130)b 0 (0–0)a 
Borehole (with hose) 9,500 (9,000–29,000)b 137 (137–1,410)b 34 (0–210)b 56 (0–280)b 0 (0–3)a 
Rain water 2,100 (2,000–14,000)a,b 338 (217–660)b 19 (0–30)a 80 (20–160)b 0 (0–0)a 
Stream water 18,000 (16,000–23,000)c 4,900 (2,900–9,800)c 900 (350–2,100)c 630 (410–920)c 4 (2–19)b 
Water sourcesHPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)Salmonella (CFU/100 mL)
Public standpipe* 5,000 (4,000–9,000)a 40 (40–170)a 1 (0–1)a 1 (1–3)a 0 (0–0)a 
Borehole (without hose)Δ 1,900 (1,900–3,300)a 118 (118–850)b 16 (0–130)b 22 (10–130)b 0 (0–0)a 
Borehole (with hose) 9,500 (9,000–29,000)b 137 (137–1,410)b 34 (0–210)b 56 (0–280)b 0 (0–3)a 
Rain water 2,100 (2,000–14,000)a,b 338 (217–660)b 19 (0–30)a 80 (20–160)b 0 (0–0)a 
Stream water 18,000 (16,000–23,000)c 4,900 (2,900–9,800)c 900 (350–2,100)c 630 (410–920)c 4 (2–19)b 

HPC: heterotrophic plate count; TC: total coliforms; EC: Escherichia coli; IE: intestinal Enterococci.

*Pipeborne water supplied by Njoro Water Works.

ΔAfter flaming.

Without flaming.

Boreholes have been classified as ‘improved’ water sources by the WHO/UNICEF Joint Monitoring Programme (JMP) classification scheme (WHO/UNICEF 2013). However, the JMP approach has been criticized because the assumptions about the microbial safety of water based on its source does not take into consideration several well-documented problems (Bain et al. 2012; Onda et al. 2012). It is well known that microbial quality of drinking water at the point-of-collection can be compromised by unhygienic handling practices during collection. In the present study, for example, hoses were attached to taps at the vending outlets to minimize wastage and also to keep the people collecting water from entering the premises of vending outlets. However, hoses were held by hand and inserted into the collecting containers when water was collected, after which they were left on the ground or dangled on the fence for the next user. Consistently, median bacteria densities in samples taken with a hose inserted onto the tap were typically higher compared to samples taken after the hose had been removed, being statistically different (p > 0.05) from some parameters, e.g., HPC counts (Table 3).

Post-collection faecal contamination

Table 4 shows the physico-chemical parameters of diverse household water types. Conductivity and total dissolved solids were used as conservative markers to differentiate household water types. Both parameters were relatively high in groundwater obtained from the boreholes and the public standpipe compared to water obtained from the steam or samples from harvested rain water. The other physico-chemical parameters were not conspicuously different among the diverse household water types. All households had borehole water stored in the home but no high-income household had stream water and no low-income households had rain water. As storage in low-income households was generally short (1–2 days), rain water was already used up by the time they were visited by the water quality team. Mixing water from different sources was a common practice especially in high-income households. Typically, they would mix harvested rain water and groundwater from a borehole by storing both in the same container. Storage containers were usually large concrete tanks (3–5 m3), which were partly submerged into the ground and connected to the piped system on the premises and the rain water harvesting system attached to the roof of the house. Other storage containers in high-income households included overhead plastic/metal containers (2–4 m3) that were used to store domestic water for long durations (4–6 weeks or even more). Mixing among low-income households was done mainly with water from different boreholes as they typically used small storage containers for a short duration.

Table 4

Physico-chemical parameters of water types stored in the households (mean and range in parentheses)

Water typeCond. (μS/cm)TDS (mg/L)DO (mg/L)Temp. (°C)pH
Low-income households 
 Public standpipea 151 (166–189) 97 (42–121) 6.8 (6.0–7.3) 20.1 (17.9–23.8) 7.4 (7.2–7.9) 
 Borehole 169 (144–200) 108 (93–128) 6.0 (4.9–5.8) 19.9 (17.1–23.4) 7.7 (7.2–8.1) 
 Stream 82 (63–98) 53 (41–63) 7.4 (6.7–7.9) 19.7 (16.0–22.6) 7.8 (7.2–7.9) 
 Mixed water 141 (54–200) 90.5 (35–128) 6.6 (4.9–8.8) 19.8 (16.0–23.8) 7.6 (7.1–8.1) 
High-income households 
 Standpipea 167 (162.4–200) 117 (16–141) 5.8 (3.6–7.3) 19.8 (13.0–24.0) 7.2 (7.5–9.3) 
 Borehole 181 (168–209) 121 (105–134) 6.6 (3.1–8.0) 19.7 (17.0–24.0) 7.5 (7.0–9.0) 
 Rain water 11.3 (6.3–25) 7.3 (4.1–16.2) 4.9 (2.7–7.0) 19.9 (17.0–22.2) 7.1 (5.2–8.2) 
 Mixed water 134.9 (6.3–220) 87 (4.1–141) 6.1 (2.7–8.0) 20.2 (13.0–29.3) 8.1 (5.2–9.3) 
Water typeCond. (μS/cm)TDS (mg/L)DO (mg/L)Temp. (°C)pH
Low-income households 
 Public standpipea 151 (166–189) 97 (42–121) 6.8 (6.0–7.3) 20.1 (17.9–23.8) 7.4 (7.2–7.9) 
 Borehole 169 (144–200) 108 (93–128) 6.0 (4.9–5.8) 19.9 (17.1–23.4) 7.7 (7.2–8.1) 
 Stream 82 (63–98) 53 (41–63) 7.4 (6.7–7.9) 19.7 (16.0–22.6) 7.8 (7.2–7.9) 
 Mixed water 141 (54–200) 90.5 (35–128) 6.6 (4.9–8.8) 19.8 (16.0–23.8) 7.6 (7.1–8.1) 
High-income households 
 Standpipea 167 (162.4–200) 117 (16–141) 5.8 (3.6–7.3) 19.8 (13.0–24.0) 7.2 (7.5–9.3) 
 Borehole 181 (168–209) 121 (105–134) 6.6 (3.1–8.0) 19.7 (17.0–24.0) 7.5 (7.0–9.0) 
 Rain water 11.3 (6.3–25) 7.3 (4.1–16.2) 4.9 (2.7–7.0) 19.9 (17.0–22.2) 7.1 (5.2–8.2) 
 Mixed water 134.9 (6.3–220) 87 (4.1–141) 6.1 (2.7–8.0) 20.2 (13.0–29.3) 8.1 (5.2–9.3) 

Cond.: conductivity; TDS: total dissolved solids; DO: dissolved oxygen; Temp.: temperature.

apipeborne water supplied by Njoro Water Works.

Post-collection faecal contamination was widespread in all the households as unhygienic handling and storage was prevalent, irrespective of household category. Faecal contamination levels increased from the point-of-collection to the point-of-consumption (Table 5). The increase in bacteria densities was particularly evident for water obtained from the vending outlets of boreholes and the public standpipe. Median bacteria densities were low at the vending outlets but they increased slightly in the carrying containers of water vendors and then drastically in household storage containers. The increase was significant (p ≤ 0.05) between the point-of-collection and point-of-consumption in both categories of households. EC densities, in particular, were significantly higher in household containers compared to the point-of-collection (e.g., p = 0.019) for EC and p = 0.003 for HPC). However, the difference was not significant between levels in household containers and those in vendors' containers.

Table 5

Median levels (range in parentheses) of bacteria densities in samples taken along the contamination chain with results of statistical tests (letters in superscript) comparing water sources and water types in the containers of vendors and the two household categories

HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Water sources 
 Public standpipe* 500 (500–900)a 0 (0–200)a 0 (0–1)a 0 (0–3)a 
 Boreholes (no hose)Δ 800 (750–2,000)a 120 (100–900)a,b 16 (0–400)b 220 (10–230)b 
 Boreholes (with hose) 1,900 (1,500–2,900)b 210 (200–1,100)b 18 (0–220)b 132 (10–190)b 
Vendors' containers 
 Boreholes 1,900 (1,400–3,100)b 700 (300–1,400)b 290 (30–350)b 350 (40–620)b 
 Low-income households 
 Borehole water 5,800 (2,500–9,800)c 1,400 (1,200–3,000)c 360 (10–520)b 510 (20–730)b 
 Stream water 32,000 (29,000–34,000)d 18,000 (13,000–21.000)d 1,400 (120–3,300)c 1,700 (200–3,600)c 
 Mixed water 6,400 (1,400–11,000)b,c,d 1.700 (200–3,900)b,c 490 (10–600)b 620 (40–700)b 
 Treated (HWT) 300 (0–400)a 0 (0–300)a 0 (0–10)a 0 (0–10)a 
High-income households 
 Borehole water** 2,700 (400–2,800)a,b 150 (40–600)a,b 40 (0–500)b 160 (0–700)b 
 Rain water 900 (50–1,400)a,b 400 (50–700)a,b 20 (0–300)b 20 (0–650)b 
 Mixed water 1,200 (1,000–8,000)b,c 400 (30–1,700)a,b 82 (0–400)b 300 (0–700)b 
 Treated (HWT) 700 (0–800)a 280 (70–600)a,b 0 (0–30)a 0 (0–50)a 
HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Water sources 
 Public standpipe* 500 (500–900)a 0 (0–200)a 0 (0–1)a 0 (0–3)a 
 Boreholes (no hose)Δ 800 (750–2,000)a 120 (100–900)a,b 16 (0–400)b 220 (10–230)b 
 Boreholes (with hose) 1,900 (1,500–2,900)b 210 (200–1,100)b 18 (0–220)b 132 (10–190)b 
Vendors' containers 
 Boreholes 1,900 (1,400–3,100)b 700 (300–1,400)b 290 (30–350)b 350 (40–620)b 
 Low-income households 
 Borehole water 5,800 (2,500–9,800)c 1,400 (1,200–3,000)c 360 (10–520)b 510 (20–730)b 
 Stream water 32,000 (29,000–34,000)d 18,000 (13,000–21.000)d 1,400 (120–3,300)c 1,700 (200–3,600)c 
 Mixed water 6,400 (1,400–11,000)b,c,d 1.700 (200–3,900)b,c 490 (10–600)b 620 (40–700)b 
 Treated (HWT) 300 (0–400)a 0 (0–300)a 0 (0–10)a 0 (0–10)a 
High-income households 
 Borehole water** 2,700 (400–2,800)a,b 150 (40–600)a,b 40 (0–500)b 160 (0–700)b 
 Rain water 900 (50–1,400)a,b 400 (50–700)a,b 20 (0–300)b 20 (0–650)b 
 Mixed water 1,200 (1,000–8,000)b,c 400 (30–1,700)a,b 82 (0–400)b 300 (0–700)b 
 Treated (HWT) 700 (0–800)a 280 (70–600)a,b 0 (0–30)a 0 (0–50)a 

HPC: heterotrophic plate count; HWT: household water treatment; TC: total coliforms; EC: Escherichia coli; IE: intestinal Enterococci.

*Running pipeborne water from boreholes supplied by Njoro Water Works.

**Stored pipeborne water from boreholes supplied by Njoro Water Works directly into the household.

ΔAfter flaming.

Without flaming.

Even though post-collection faecal contamination was generally more serious in low-income households compared to high-income households, contamination levels were not significantly different (p ≥ 0.05) between the two household categories. The concentration of EC in borehole water, for instance, was in the range 0–4 log units for low-income households and 0–2 log units for the high-income households, while IE was in the range 0–4 and 1–2 log units for low-income and high-income households, respectively (Table 5). Specific household and demographic characteristics were certainly different between the two household categories but these differences did not seem to strongly influence the microbial quality of domestic water stored in the households. This is most likely because similar post-collection unsanitary practices were observed in both categories of households. For instance, the decline in microbial quality of drinking water after collection could have been amplified through increased bacteria growth or regrowth in water already contaminated with faecal matter at the point-of-collection. Like vendors, several households did not clean their conveying containers, some of which were used to transport polluted water from the stream as well. An average increase of 1–2 log units in EC densities was observed in the conveying containers of water vendors when compared with the microbial densities in the same water at the point-of-collection.

Once contamination has occurred, the period of storage before consumption could also influence the survival and growth of bacteria in the storage containers (Roberts et al. 2001; Levy et al. 2008). A long period of storage was particularly common among high-income households. Storage containers were often left open and/or poorly covered while some storage containers, especially large containers, had not been cleaned since they were installed 3–6 years ago. Containers left unclean for a long period could favour regrowth of pathogenic microbes in the layer of biofilm that may develop in the walls of the container (Sobsey 2002). Similarly, the characteristics of storage containers have been associated with post-collection faecal contamination of domestic water. For instance, storage in open containers allows faecal contamination to occur as well as in those using containers with large openings and extracting water from them by dipping handheld utensils into them (Jensen et al. 2002; Schmidt & Cairncross 2009). Stored water in households using large containers with larger openings (50–100 cm in diameter) was often extracted by dipping handheld utensils/dippers into the container. Pouring was used for domestic water stored in smaller vessels such as 20 L jerry-cans with a small opening (6 cm in diameter). A few households with large storage vessels used a spigot attached to a closed container but this practice was not implemented exclusively as the spigot was occasionally functional.

Household water treatment

Table 6 shows the median densities of stored borehole water in the two household categories with and without HWT. Boiling was a very common HWT practice among high-income households, but it was rarely used in low-income households due to the associated fuel costs. Alternatively, most low-income households undertook HWT by chemical disinfection. They used a chemical disinfectant called Water Guard®, a chlorine-based product used for treating drinking water at the household level in several countries in sub-Saharan Africa. The product is widely available in Kenya, where an estimated 85% of the population is aware of the brand and as many as 16% of households use it regularly (PATH 2010). It is distributed mainly in liquid form to ensure dosing flexibility. In Njoro Township, a 20 mL bottle of Water Guard® is retailed for KES 20–30 (0.23–0.35 US$) and a cap-full (3.5 mL) could effectively treat 20 L of water in 30 minutes.

Table 6

Median levels (range in parentheses) of bacteria densities in stored borehole water with and without HWT including results of statistical tests (letters in superscript) comparing HWT methods in the two household categories

HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Low-income households 
 Borehole (without HWT) 5,800 (2,500–9800)a 1,400 (1,200–3,000)a 360 (0–520)a 510 (20–730)a 
 Borehole (with HWT) 300 (0–400)b 0 (0–300)b 0 (0–100)b 0 (0–100)b 
High-income households 
 Borehole (without HWT) 2,700 (400–2800)a 150 (40–600)c 40 (0–500)a 160 (0–700)a 
 Borehole (with HWT)β 700 (0–800)b 0 (70–600)b 0 (0–30)b 0 (0–50)b 
HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Low-income households 
 Borehole (without HWT) 5,800 (2,500–9800)a 1,400 (1,200–3,000)a 360 (0–520)a 510 (20–730)a 
 Borehole (with HWT) 300 (0–400)b 0 (0–300)b 0 (0–100)b 0 (0–100)b 
High-income households 
 Borehole (without HWT) 2,700 (400–2800)a 150 (40–600)c 40 (0–500)a 160 (0–700)a 
 Borehole (with HWT)β 700 (0–800)b 0 (70–600)b 0 (0–30)b 0 (0–50)b 

HPC: heterotrophic plate count; TC: total coliforms; EC: Escherichia coli; IE: intestinal Enterococci; HWT: household water treatment.

HWT by chemical disinfection.

βHWT by boiling.

Improvement in microbial water quality after HWT was achieved irrespective of the HWT method employed. There were no significant differences (p ≥ 0.05) between HWT by chemical disinfection in low-income households and HWT by boiling in the high-income households (Table 6). The microbial quality of stored water improved significantly with the two HWT methods. However, unhygienic handling and poor storage practices caused recontamination afterwards. Often, households added newly treated water to previously treated water without first emptying and cleaning their storage container. Whereas no EC or IE colonies were detected in some HWT samples that had been treated a few hours before sampling, growths of up to 2 log units were observed a few days after HWT (Table 7). Boiled water, in particular, was usually left uncovered to cool quickly. During cooling, it is possible that dust particles with attached bacteria will accumulate in the warm water, which naturally favours regrowth of heterotrophic bacteria. This may explain the notable increase in HPC densities in the high-income households which practised HWT by boiling compared to chemical disinfection in low-income households. In addition, the residual effects of chemical disinfection may have suppressed the regrowth of heterotrophic bacteria in HWT samples obtained from low-income households.

Table 7

Median levels (range in parentheses) of bacteria densities in stored borehole water after HWT, with results of statistical tests (letters in superscript) comparing storage periods

HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Low-income households 
  < 1 day after HWT 300 (10–400)a 10 (10–300)a 0 (0–10)a 0 (0–10)a 
 1–3 days after HWT 320 (120–380)a 100 (10–200)a 50 (10–180)a 20 (10–110)a 
 4–7 days after HWT 400 (200–560)a 80 (20–400)a 10 (10–120)a 20 (10–200)a 
High-income households 
  < 1 day after HWTβ 500 (10–700)a 10 (10–200)a 0 (0–30)a 0 (0–50)a 
 1–3 days after HWTβ 800 (300–900)a 50 (40–300)a 20 (10–100)b 20 (0–100)b 
 4–7 days after HWTβ 750 (200–800)a 90 (30–170)a 30 (20–100)b 40 (40–110)b 
HPC (CFU/mL)TC (CFU/100 mL)EC (CFU/100 mL)IE (CFU/100 mL)
Low-income households 
  < 1 day after HWT 300 (10–400)a 10 (10–300)a 0 (0–10)a 0 (0–10)a 
 1–3 days after HWT 320 (120–380)a 100 (10–200)a 50 (10–180)a 20 (10–110)a 
 4–7 days after HWT 400 (200–560)a 80 (20–400)a 10 (10–120)a 20 (10–200)a 
High-income households 
  < 1 day after HWTβ 500 (10–700)a 10 (10–200)a 0 (0–30)a 0 (0–50)a 
 1–3 days after HWTβ 800 (300–900)a 50 (40–300)a 20 (10–100)b 20 (0–100)b 
 4–7 days after HWTβ 750 (200–800)a 90 (30–170)a 30 (20–100)b 40 (40–110)b 

HPC: heterotrophic plate count; TC: total coliforms; EC: Escherichia coli; IE: intestinal Enterococci. HWT: household water treatment.

HWT by chemical disinfection.

βHWT by boiling.

CONCLUSION

The study demonstrated that drinking water quality deteriorated from an improved water source to the point-of-consumption, irrespective of household income category. Post-collection faecal contamination was generally high in low-income households but contamination occurred in both household categories through unhygienic handling during collection, transport and storage practices. Different HWT methods in the two household categories significantly reduced faecal contamination but unhygienic handling and poor storage practices afterwards caused recontamination. It was inferred that HWT and behavioural change measures need not selectively target household groups solely on the basis of their income status. Interventions aimed at improving access to safe drinking water at the point-of-consumption should aim ultimately at providing uninterrupted access to piped systems on premises to circumvent unhygienic handling during collection, transport and storage.

ACKNOWLEDGEMENTS

The study was largely financed by the Austrian Development Co-operation with a fellowship award to Pauline W. Macharia through the International Post-graduate Programme in Limnology. The authors are grateful to the residents of Njoro Township who participated in the study.

REFERENCES

APHA
1999
Standard Methods for the Examination of Water and Wastewater
, 20th edn.
American Public Health Association (APHA)
,
Washington DC
.
Chabalala
H. P.
Mamo
H.
2001
Prevalence of Water-Borne Diseases within the Health Facilities in Nakuru District, Kenya
.
MSc. thesis
,
University of Nairobi
,
Kenya
,
114
p.
Clasen
T. F.
Bastable
A.
2003
Faecal contamination of drinking water during collection and household storage: the need to extend protection to the point of use
.
J. Water Health
1
(
3
),
109
115
.
Jensen
P. K.
Ensink
J. H.
Jayasinghe
G.
van der Hoek
W.
Cairncross
S.
Dalsgaard
A.
2002
Domestic transmission routes of pathogens: the problem of in-house contamination of drinking water during storage in developing countries
.
Trop. Med. Int. Health
7
,
604
609
.
Levy
K.
Nelson
K. L.
Hubbard
A.
Eisenberg
J. N. S.
2008
Following the water: a controlled study of drinking water storage in northern coastal Ecuador
.
Environ. Health Persp.
116
(
11
),
1533
1540
.
Macharia
P. W.
2011
Microbial Quality of Drinking Water and its Effect on Health in Sub-Saharan Africa: A survey of Njoro Town, Kenya
.
MSc. Thesis
,
UNESCO-IHE Institute for Water Education
,
Netherlands
,
84
p.
PATH
2010
Market Assessment of Household Water Treatment Products in Eight African Countries
.
Project Brief. Safe Water Project, Program for Appropriate Technology in Health (PATH). http://www.path.org/publications/files/TS_safe_water_mkt_assess.pdf (retrieved March 2013)
.
Roberts
L.
Chartier
Y.
Chartier
O.
Malenga
G.
Toole
M.
Rodka
H.
2001
Keeping clean water clean in a Malawi refugee camp: a randomized intervention trial
.
Bull. World Health Org.
79
,
280
287
.
Sobsey
M.
2002
Managing Water in the Home: Accelerated Health Gains from Improved Water Supply
.
World Health Organization
,
Geneva, Switzerland
. .
WHO
2011
Guidelines for Drinking Water Quality
, 4th ed.
World Health Organization
,
Geneva, Switzerland
, pp.
1
564
. .
WHO/UNICEF
2013
Progress on sanitation and drinking water: 2013 update. WHO/UNICEF Joint Monitoring Programme (JMP) for water supply and sanitation. http://www.wssinfo.org/fileadmin/user_upload/resources/JMPreport2013.pdf (retrieved November 2013)
.