Analyses of drinking water quality during a protracted cholera epidemic in Malawi – a cross-sectional study of key physicochemical and microbiological parameters

Safe drinking water is defined as safe for drinking, preparing food, washing, and personal hygiene (Bos et al. 2016). De Zuane et al. and the World Health Organization (WHO) have set out the basic parameters for determining and monitoring the quality of safe water, namely, physical, chemical, microbiological, and radiological parameters (Table 1) and estimate that a person needs 50–100 l of water per day to meet basic requirements such as drinking, cooking, hygiene, and washing of clothes (De Zuane 1997; Howard et al. 2020). Safe water is critical in public health, particularly in preventing and controlling several diseases such as diarrhoea, schistosomiasis, trachoma, and soil-transmitted helminths (World Health Organization 2023). Recognizing the importance of safe drinking water to life and well-being, the United Nations declared it a fundamental human right; this is in the realization that clean drinking water is critical to achieving all human rights (United Nations General Assembly 2010). This is further affirmed by the Sustainable Development Goals (SDG), the sixth of which aims to ensure access to water and sanitation for all the world's population by 2030. Seven out of the eight targets of SDG 6 are water related, with target 6.1 particularly advocating for equitable and universal access to safe and affordable drinking water for all (The Global Goals 2023). Unfortunately, a significant proportion of the world's population can still not achieve these targets. As of 2020, only 30% of the African population had access to safely managed drinking water (World Health Organization & United Nations Children Fund 2021). This and other social, economic, cultural, and humanitarian factors continue to sustain the high incidence of diarrhoeal diseases such as cholera on the continent (Ajayi & Smith 2019).

Cholera is an acute diarrheal infection, which is acquired through the ingestion of water or food that are contaminated by a bacterium known as Vibrio cholerae (Clemens et al. 2017). The seventh cholera pandemic reached the shores of Africa in 1971 and has spread rapidly on the continent and is now endemic in several countries, including Malawi (Ali et al. 2015). Since 1998, the country has experienced yearly seasonal epidemics during the rainy season in the southern parts, which are flood-prone (Msyamboza et al. 2014). On 3 March 2022, through its Ministry of Health, the country declared a cholera epidemic following confirmation of a case in the Machinga district in the southern parts of the country (World Health Organization 2022a, 2022b). The epidemic occurred against the backdrop of tropical Storm Ana and Cyclone Gombe, which hit the country's southern parts in 2022 and resulted in massive flooding and population displacement (World Health Organization 2022a, 2022b). This epidemic proved to be one of the longest and deadliest in the history of cholera in the country. As of 16 July 2023, the country had recorded 58,941 cases and 1766 deaths in all its 28 districts, with a case fatality ratio of 3% (World Health Organization 2023). The urban districts of Lilongwe, Blantyre, and Mangochi, on Lake Malawi's shores, accounted for more than 50% of the cases. The epidemic was further worsened by tropical Cyclone Freddy, which hit the central and southern parts of the country in March 2023. The country, with the support of its national and international partners, instituted wide-ranging epidemic response interventions, including the establishment of cholera treatment centres to manage the cases, implementation of emergency Water, Sanitation and Hygiene (WASH) interventions, including water quality testing, monitoring, and surveillance, and cholera Risk Communication and Community Engagement (RCCE) in all the affected districts.

Anecdotal evidence indicated that contaminated water played a major role in the propagation of the epidemic. Available literature also indicated that a sizable percentage of the Malawian population obtains their drinking water from sources which could be potentially contaminated. Mumba et al. (2021) estimated that 60% of the population has inadequate access to safe drinking water, while Mkandawire & Banda (2009) show that 21, 37, 16, and 26% of the Malawian population, respectively, use piped water, boreholes, rivers and lakes, and unprotected wells. While these studies have previously documented the status of the quality of drinking water in Malawi, a recent literature search showed no empirical data on drinking water quality during the cholera epidemic. Such empirical data are critical for the planning, implementation, monitoring, supervision, and evaluation of cholera epidemic control interventions, particularly the WASH and RCCE components (Blanton et al. 2015). This study, therefore, sought to establish the situation of drinking water quality in Malawi to provide empirical information for evidence-based decision-making, planning, implementation, and evaluation of the preparedness and response interventions during the current and future cholera epidemics in the country.

We conducted a cross-sectional study to assess drinking water quality in 17 cholera-affected Malawi districts from February to April 2023. Malawi, a country in southeastern Africa, borders Tanzania to the north and northeast, Mozambique to the east, south and southwest, and Zambia to the west. In 2021, the country's population was estimated at 19.4 million people. Administratively, it is divided into three regions, northern, central, and southern regions, which are subdivided into 28 districts. Seven of the districts are located on the shores of Lake Malawi, while one district is an island on the Lake. It is one of the least developed countries globally, with an estimated 85% of the population living in rural areas. The country's main water sources are surface water, such as lakes, rivers, and streams, and underground water, such as boreholes and shallow wells. Only 24.2% of the population has access to safely managed sanitation services. About half (50.7%) of the Malawian population is estimated to fall below the poverty line; this rises to 56.6% in rural areas. Life expectancy at birth is 65.6 years as of 2019. The southern and central regions of the country are prone to tropical storms and cyclones which often result in flooding, population displacement, and epidemics of waterborne infections.

Seventeen of the 28 districts of the country were enrolled in the study. The selection criteria were districts with large cholera caseloads and affected by tropical cyclone Freddy, mostly in the country's central and southern regions during the study period. In each district, the data collection teams visited the major cholera treatment centres to assess the cholera line lists to determine where the cholera caseloads were large. In each locality, a purposive sampling method was used to select households, water kiosks, boreholes, and wells from where samples of drinking water were collected. The definitions of various water sources in Malawi are described in Box 1. Attempts were made to sample all and, if not possible, as many drinking water sources in the selected locality as possible.

Three research teams, each comprising five persons, namely, a water chemist, water microbiologist, water quality technician, health surveillance assistant and driver, collected the water samples and data. The Ministry of Health, Ministry of Water and Sanitation, and WHO WASH experts trained the teams on sample collection techniques, calibration, and use of the equipment and recording of the data before the study. The team conducted pilot tests in Lilongwe in households and water sources not included in the study to familiarize themselves with the equipment, water sampling, and testing and data collection procedures. The first team covered Nkhotakota, Salima, Mchinji, Machinga, Mangochi, Balaka, and Neno districts. The second team surveyed Lilongwe, Dedza, Ntcheu, Zomba, Phalombe, and Mulanje, while the third covered Nsanje, Chikwawa, Blantyre, and Thyolo districts. In each locality, the research team collected the water samples from various points where the population collected their drinking water: boreholes, shallow wells, municipal water systems, water kiosks, and drinking storage containers in homes. The data collection team collected approximately 200 ml of water at each sample collection site using the sterilized glass collection vessels or bags included in the testing kit. The water samples were tested for eight water quality parameters, namely, pH, temperature, turbidity, total dissolved substances (TDS), free residual chlorine (FRC), electricity conductivity (EC), thermotolerant coliform (TTC), and Escherichia coli. The acceptable ranges and levels for these parameters are outlined in Table 1.

All tests except TTC and E. coli counts were conducted onsite to avoid cross-contamination. The tests were performed using the various modules of the Aquasafe^® WSL25 Pro water testing kit. pH was measured with Hanna HI9125 digital pH/ORP meter. EC and TDS were measured with a rugged, waterproof, handheld Hanna HI99300 digital EC/TDS meter. Turbidity was measured with a handheld Hanna HI98703 field meter, while FRC was measured with Lovibond MD100T. TTC was measured with a Membrane Filtration kit (M EZFITBASE1), while E. coli count was done with the same equipment but using chromogenic E. coli agar. For TTC and E. coli counts, 100 ml of the water sample was passed through water filter papers which were then put into a sterilized petri dish containing transport media and labelled. In locations where the water filter papers could not be transported to the laboratory within 2–4 h, they were incubated in the field using millipore mobile incubators, while the other samples were transported to the National Central Water Laboratory for incubation. In the laboratory, the filtered water samples were incubated between 36 and 44 °C for 18–24 h. The filtrates in Millipore Petri dishes were then taken out, and the colonies were read using a magnifying glass. TTC colonies have a pinkish colour, while E. coli colonies have a greenish-to-bluish colour on the filter paper. The counting of the TTC and E. coli colonies on each filter paper was stopped when the count reached 100. The results were entered into forms with 16 variables: name of compound, district, area, date and time of sample collection and testing, type of water source, and the eight water quality parameters.

A clerk entered the data into Microsoft Excel. The Microsoft Excel database was regularly reviewed by a supervisor to ensure the completeness of the data. A data manager cleaned and exported the data from Microsoft Excel to Stata version 15 (College Station, TX, USA) and R (version 4.2.2) on R Studio (version 2022.12.0-353) where all the analyses were conducted. Continuous variables were summarized as medians and interquartile range (IQR), and categorical variables as proportions. We conducted between-group comparisons using a chi-square or Fisher's exact test whenever the expected frequencies were less than 5. We conducted three separate univariable logistic regression analyses. The dependent variable was either TTC positive or negative, E. coli positive or negative, or both TTC or E. coli positive or negative. For each of the regression models, the independent variable was the water source (household versus non-household). The results are presented in the form of odds ratios (ORs) and their corresponding 95% confidence intervals (CIs). These ORs represent the risk of contamination of household water sources with TTC, E. coli, or either of the two. We also tested the correlation between the presence of FRC and the presence of TTC and E. coli in the water source. For all statistical tests, two-tailed p < 0.05 was considered statistically significant. We presented the analysed data as tables and figures.

For E. coli and TTC, the percentage of contaminated drinking water sources was significantly higher in shallow unprotected wells (80.0% for E. coli and 95.0% for TTC) and in households (55.8% for E. coli and 86.0% for TTC). For the physicochemical parameters, the percentage of the abnormal rate was significantly higher in shallow unprotected wells for the turbidity (95.0%) and pH (35.0%), while 90.9% of the household water samples had sub-optimal FRC. A total of 66.4% of all the water samples had NTU of above 1, which is the nationally acceptable limit for Malawi. When computed using an NTU limit of 5, which is acceptable in emergency situations, the result was 16.8%. A total of 87.6% of the water samples that were expected to have FRC had sub-optimal levels of the chemical which is below the Malawi acceptable level of 0.2 mg/l. When a cutoff of 0.5 mg/l of FRC, which is acceptable during cholera epidemics, was used, only 12.3% of the water samples had the required dosage of chlorine, which is expected during a cholera epidemic (Table 4). Overall, the proportion of water contaminated with E. coli was 45.5% (95% CI, 41.64–49.41). The proportion of water contaminated with TTC was significantly higher, i.e., 80.3% (95% CI, 76.96–83.18; p < 0.000) (Table 4).

Our study shows that most of the drinking water sources in Malawi were contaminated with TTC and E. coli, turbid, and lacked the required dosage of chlorine to render water safe for drinking in the context of a cholera epidemic. Our results also confirmed that regardless of the type of contamination, drinking water sources at the household level were more likely to be contaminated than all other types of water sources in Malawi, suggesting ongoing contamination at the household level. These findings confirm the anecdotal evidence that the prolonged cholera epidemic in Malawi could have been due to widespread contamination of drinking water sources.

The fact that most drinking water sources were contaminated cast doubt on the effectiveness of the ongoing WASH and RCCE interventions which commenced implementation almost a year before the study. We posit that these interventions are ineffective due to a few reasons. First, while intensive RCCE interventions were implemented in the country, our observations showed that the messages did not adequately communicate the technicalities involved in the safe handling and storage of household water and hand hygiene. Second, Malawi's high poverty level may be a hindrance to households' compliance with the conditions required to purify and safely store household water. Observations during the study period showed that even when communities were aware of water purification and storage practices, lack of resources often hinder their ability to adopt such practices. Third, the mass water chlorination which was ongoing at the time of the study may have been inadequate in terms of its coverage, intensity, and methodology; this is corroborated by the sub-optimal FRC levels in most drinking water sources in the study area.

Chlorine is a commonly used agent for microbiological quality control of drinking water in water distribution systems. FRC is required in sufficient quantities in drinking water to guarantee its safety throughout the water chain by inhibiting microbiological growth (World Health Organization 2013). The effectiveness of chlorine depends on the water's turbidity level, storage method, and pH of the water. Storage of water away from direct sunlight (i.e., in shaded areas) and in tightly sealed water containers reduces the chances of chlorine evaporation (Rosalam et al. 2007). In addition, FRC optimally disinfects water at a pH of 6.8–7.2 (World Health Organization 2013). In an everyday context, the acceptable level for FRC is 0.2 mg/l. However, during a threat or epidemic of cholera, the acceptable limit is increased to 0.5 mg/l (World Health Organization 1993). A significant percentage of the water samples in this study had sub-optimal FRC when subjected to the standard acceptable limits, while only about 12.5% of the water samples met the acceptable level of FRC for an epidemic. This may have contributed to the widespread contamination of household drinking water reported in this study. This observation is supported by the finding that water samples with sub-optimal FRC were more likely to be contaminated, which underscores the importance of intensive, effective, and sustained water chlorination during cholera epidemics. These findings may be due to a few reasons. First, there may have been inadequate chlorination from the waterworks and at the household level. Second, the drinking water sources in the country are largely from untreated sources that are likely to be contaminated, such as boreholes, shallow wells, and other unprotected water sources (Mkandawire & Banda 2009). Third, this finding could have been due to how the water is stored and handled at home. Our observations showed that most household drinking water is stored in open buckets, water drums, and uncovered jerry cans, which could have facilitated rapid evaporation of FRC where present.

The finding that most household drinking water was contaminated is corroborated by other studies (Clasen & Bastable 2003; Copeland et al. 2009). This finding is not surprising given the fact that a significant amount of the drinking water sources sampled in this study were contaminated. However, it is of significant concern for controlling the current epidemic as the household, which should be at the centre of the cholera epidemic prevention and control efforts, perpetuates it. We attribute this to two main reasons. First, the sub-optimal level of FRC observed in the majority of drinking water indicates the ineffectiveness of the ongoing chlorination efforts. The second plausible reason could be due to poor storage and handling of household drinking. Several studies have described the factors associated with household water contamination (Gasana et al. 2002; Quick et al. 2002; Oswald et al. 2007). These include the type and design of the storage vessels, the method of drawing drinking water from the storage vessels, the pattern of use of the drinking vessels and water storage conditions. Our observations showed that a combination of these three factors could have also contributed to the contamination of household drinking water.

The high level of TTC may also be associated with the abnormal levels of turbidity and temperature of the water found in this study. Turbidity influences safe drinking water in two ways. First, turbid water provides a suitable medium for pathogens to survive, and second, it reduces the effectiveness of chlorine as a water disinfectant (Kelley et al. 2014; Bwire et al. 2020). The higher dosage of chlorine, which is required for turbid water, could increase the cost of chlorination and result in chlorine odour and taste in the water, a situation that may result in the rejection of such water. The median turbidity of 1.61 NTU reported in this study is higher than acceptable national and international (WHO) limits but falls within the acceptable limits for emergency situations. Moreover, the turbidity of more than 60% and more than 16.8% of all the water samples were, respectively, above the nationally and internationally acceptable limits, pointing to a critical problem with water turbidity in Malawi. This could contribute to an increased risk of water contamination in the study area. Water temperature also contributes to the quality of drinking water. Temperatures above 17 °C have been demonstrated to be favourable for the survival of pathogens in water (Bwire et al. 2020). The median temperature of 26.5 °C, reported in this study, could have contributed to the widespread contamination of the water sources.

Unsurprisingly, many of the unprotected wells and boreholes were contaminated with TTC and E. coli given that several studies have documented the risk associated with the consumption of untreated boreholes and well water (O'Dwyer 2015; Taonameso et al. 2019; Masindi & Foteinis 2021). However, the process leading to this contamination in the Malawian context requires further elucidation as to whether this is through seepage through the walls, contamination by humans, or surface sewage water (Graham & Polizzotto 2013).

The findings of this study should be interpreted within the context of three limitations. First, the study was conducted in the districts that had active transmission of cholera; thus, its findings may not reflect the situation in the districts that were not reporting cases of the disease at the time of the study. Second, this is a cross-sectional study which tested only one sample from the water sources. The collection of several water samples at various times would have been more representative and shed light on the variations in water quality over time. Nevertheless, we believe that the study's large sample size has compensated for these limitations. Third, most of the water samples were collected from households (stored water), and we could not link contaminated household drinking water to their sources of collection to understand the exact point and mode of contamination. This limitation is, however, mitigated by the fact that most of the drinking water sources in the study were contaminated.

This study demonstrated widespread contamination of drinking water during a cholera epidemic in Malawi, which may be the plausible reason for the protracted nature of the epidemic. The fact that most of the drinking water at the household level was contaminated supports this conclusion. The high level of household water contamination is likely due to ineffective chlorination, poor personal and household hygiene and poor water collection, transportation, storage, and consumption practices. The widespread turbidity and sub-optimal FRC found in this study could have also contributed to the widespread contamination of the drinking water. Addressing the root causes of this widespread contamination of drinking water sources will play a critical role in breaking the transmission chain of the current epidemic and preventing future ones in Malawi and other African countries.

On the basis of our findings, we propose five key recommendations. First, ongoing surveillance and analyses of drinking water's physicochemical and microbiological parameters and rapid sanitary surveys should be implemented during cholera epidemics. The data from the surveillance system should be used to define the variations in water quality over time. Furthermore, the data should be used to plan, implement, monitor, and evaluate evidence-based WASH and RCCE interventions, particularly those aimed at ensuring safe drinking water. To ensure sustainability, the water quality surveillance should be integrated into the Integrated Disease Surveillance and Response (IDSR) system in cholera-endemic settings. Second, we recommend intensive promotion of sustained and effective hand washing and chlorination of household and other drinking water sources as part of RCCE/WASH interventions to prevent and control the current and future outbreaks. Third, focused RCCE interventions that communicate the technicalities associated with safe handling, storage, and consumption of drinking water in simple language should be implemented to address the widespread contamination of household water during the current epidemic. These interventions should include anthropological methods to understand the local practices fuelling contamination. Fourth, further epidemiological studies are required to identify the other risk factors associated with this protracted epidemic. Finally, we call on the Government of Malawi and its health and WASH partners to implement comprehensive and long-term interventions to strengthen the water and sanitation policies, institutions, and systems in Malawi towards the attainment of the SDGs water targets in the country. This in our view is the key strategy for effective and sustainable prevention and control of cholera and other diarrhoeal diseases in the country in the long term.

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

The authors declare there is no conflict.