Assessment of water, sanitation and hygiene interventions in response to an outbreak of typhoid fever in Neno District, Malawi

On May 2, 2009 an outbreak of typhoid fever began in rural villages along the Malawi-Mozambique border resulting in 748 illnesses and 44 deaths by September 2010. Despite numerous interventions, including distribution of WaterGuard (WG) for in-home water treatment and education on its use, cases of typhoid fever continued. To inform response activities during the ongoing Typhoid outbreak information on knowledge, attitudes, and practices surrounding typhoid fever, safe water, and hygiene were necessary to plan future outbreak interventions. In September 2010, a survey was administered to female heads in randomly selected households in 17 villages in Neno District, Malawi. Stored household drinking water was tested for free chlorine residual (FCR) levels using the N,N diethyl-p-phenylene diamine colorimetric method (HACH Company, Loveland, CO, USA). Attendance at community-wide educational meetings was reported by 56% of household respondents. Respondents reported that typhoid fever is caused by poor hygiene (77%), drinking unsafe water (49%), and consuming unsafe food (25%), and that treating drinking water can prevent it (68%). WaterGuard, a chlorination solution for drinking water treatment, was observed in 112 (56%) households, among which 34% reported treating drinking water. FCR levels were adequate (FCR ≥ 0.2 mg/L) in 29 (76%) of the 38 households who reported treatment of stored water and had stored water available for testing and an observed bottle of WaterGuard in the home. Soap was observed in 154 (77%) households, among which 51% reported using soap for hand washing. Educational interventions did not reach almost one-half of target households and knowledge remains low. Despite distribution and promotion of WaterGuard and soap during the outbreak response, usage was low. Future interventions should focus on improving water, sanitation and hygiene knowledge, practices, and infrastructure. Typhoid vaccination should be considered.

Introduction organization staff. Key messages were reinforced using a locally produced DVD about the typhoid outbreak, dramatic performances, posters and flyers, and live demonstrations of water treatment using recommended treatment products, hand washing, and construction of hand washing stations. All community members were encouraged to attend, and time was allocated for answering questions posed by attendees. Six of 17 villages also received improved infrastructure, including construction of borehole wells and pit latrines (Table 1). An estimate of the number of households in each village was provided by village leadership. The female head of household was selected as the target respondent for the survey as she would be expected to be the most knowledgeable about household water, sanitation and hygiene practices.

Questionnaire design
A household questionnaire (S1 Fig) was developed by the Centers for Disease Control and Prevention (CDC), the Neno District Health Office, and non-governmental partner organizations involved in post-outbreak response activities. It was designed to assess knowledge regarding the causes, treatment, and methods of preventing typhoid fever and household knowledge, attitudes and practices on water, sanitation, and hygiene. Household water treatment practices were reported for the 2 weeks before administration of the household questionnaire (i.e., in 2010) and during the previous year (i.e., in 2009). The final survey was administered by trained enumerators in the local language, Chichewa.

Household surveys
In each of the 17 villages, households were randomly selected. Enumerators located the central point of each village and determined a random direction for household sampling by spinning

Household stored drinking water testing
In all households in which drinking water was stored and available for testing at the time of the visit, samples were collected and tested for free chlorine residual (FCR) using the N,N diethyl-p-phenylene diamine (DPD) colorimetric method (HACH Company, Loveland, CO, USA). FCR levels were considered adequate at ! 0.2 milligrams per liter [19].

Water source testing
Two water samples were collected from improved drinking water sources in 8 of 17 survey villages, including 9 boreholes and 3 taps. Water samples were tested for total coliform bacteria and Escherichia coli using presence-absence broth with 4-methylumbelliferyl-β-D-glucuronide (HACH method 8364, HACH Company, Loveland, CO, USA).

Data management and analysis
Data were entered into a Microsoft Access 2007 database (Redmond, WA, USA) and analyzed using SAS version 9.3 (Cary, NC, USA). Attendance at a "typhoid talk" was not known in advance of the administered household survey and so, random samples could not be independently drawn from households that attended a "typhoid talk" and households that did not attend. However, response frequencies were computed for all household respondents and stratified by those who reported attending a "typhoid talk" and those who did not. To identify differences between these two groups, comparisons of proportions were conducted using the Rao-Scott design-adjusted chi-square test accounting for village clusters. Differences were evaluated for statistical significance at the alpha = 0.05 level.

Ethics
This survey was initiated in the setting of an ongoing typhoid fever outbreak in an effort to guide additional interventions. Human subjects research designees at CDC and on the Malawi National Human Subjects Review Committee determined that this activity constituted public health response and program evaluation rather than research. Verbal permission to enroll households within a village and collect water from improved water sources was obtained from village leadership. Female heads of household provided verbal consent for their participation in the survey and testing of household stored water. All consents were obtained in Chichewa.

Household survey
A total of 393 households were visited and 202 (51%) were enrolled in 17 villages; 187 (48%) female heads of household were unavailable and 4 (1%) refused participation. Among enrolled households (n = 202), the median age of respondents was 30 years (range 18-83 years) ( Table 2). The median household size was 5 persons (range 1-12) and the median number of children under 5 years of age was 1 (range 0-4). Some formal education (i.e., any attendance in a primary or secondary school) was reported by 78% of respondents and 57% reported being able to read. Among the household assets included in the survey (bicycle, motorcycle, car, radio, television, refrigerator, solar panel), ownership of a radio was most common (69%); 26% of households reported owning none of the household assets. One hundred and eleven (56%) of 198 household respondents in 17 villages reported attending a "typhoid talk"; four respondents in 4 villages were unsure. Reported attendance ranged from 33% of respondents in six villages, including the two villages where no talks were given, to 92% in Kweneza. Formal education, self-reported literacy, and household assets were not significantly different among those who reported attending a "typhoid talk" compared with those that did not. Respondents who attended a "typhoid talk" (n = 111) reported that talks were led by community health workers (67%), clinicians (29%), and non-governmental organizations (14%). Of 111 respondents who attended a "typhoid talk", 85 (77%) reported receiving free products at the talk. Of these, 86% received WaterGuard and 78% received soap.
Among all household respondents (n = 202), the most commonly reported causes of typhoid fever were poor hygiene (77%), drinking unsafe water (49%), and consuming unsafe food (25%) ( Table 3). Boiling or treating drinking water (68%), hand washing (52%), and cleaning cooking utensils and vessels (38%) were the most commonly reported methods for preventing typhoid fever. Among those that reported attending a "typhoid talk" (n = 111), reported causes of typhoid fever were poor hygiene (86%), drinking unsafe water (54%), and consuming unsafe food (30%). Also, these respondents reported that boiling or treating drinking water (73%), hand washing (52%), and cleaning cooking utensils and vessels (45%) were methods for preventing typhoid fever. Cleaning cooking utensils and vessels was more Ã Self-reported literacy, any formal education, no formal education, and household assets were similar between the two groups using the P < 0.05 by Rao-Scott designadjusted chi-square test accounting for clustering by village (significance considered at P < 0.05). https://doi.org/10.1371/journal.pone.0193348.t002 commonly reported among respondents who attended a "typhoid talk" compared with those who did not (P = 0.0261). Most respondents (98%), regardless of their reported "typhoid talk" attendance (n = 202), indicated that they would seek treatment for typhoid fever at a hospital or clinic, while only 2% reported use of a home remedy. Among all households (n = 202), the primary sources of household drinking water were unimproved wells (45%), boreholes (42%), rivers (7%) and taps (5%); respondents who reported using a borehole as their primary water source (n = 84) also reported drinking water from unimproved wells (54%) and rivers (6%). Nearly all respondents (198, 98%) reported having ever treated their drinking water; among these, only 30% reported always treating their drinking water (Table 4). Among households who reported ever treating their drinking water (n = 198), WaterGuard was the most popular treatment method (93%) in 2010; additional treatment methods included use of homemade chlorine solution (25%) and boiling (15%). Unwashed fruits and vegetables 12 (6) 6 (5) 6 (7) Person-to-person spread

Methods of preventing typhoid fever
Traditional healer 0 (0) 0 (0) 0 (0)  , belief that their current water source is safe (11%), and affordability (8%); only 3% of respondents reported bad taste or smell associated with the use chlorine-based treatment products as a barrier for regularly treating their water. Overall (n = 202), 89% of household respondents reported that they received free Water-Guard (Table 4). Few respondents (8%) reported having ever purchased WaterGuard, even though 39% reported an insufficient household supply of WaterGuard in the past month. Fifty-six percent of households were observed to have a bottle of WaterGuard at the time of interview; among these (n = 112), 34% reported that the water stored in their home was treated ( Table 5). In households with drinking water storage vessels available for observation (n = 176), water was stored in a combination of vessels with mouths wide enough to allow a hand to pass through the opening and touch the stored water (75% of households) and narrow-mouthed vessels (26%), and 35% of these households had at least one uncovered drinking water storage vessel (Table 4). Seventy-five percent reported using a cup or ladle to scoop water from observed storage vessels.
Seventy-seven percent of all households (n = 202) were observed to have soap; among households with soap (n = 154), 51% reported using it to wash hands ( Table 4). Use of soap for washing hands was more commonly reported by household respondents who did not attend a "typhoid talk" (62%) compared with those that did (42%) (P = 0.0070). Respondents (n = 202) reported that hand washing should be performed after using the toilet (92%), before eating (73%), after washing and cleaning babies (50%), and before cooking (40%). Washing hands before eating was more commonly reported by respondents who attended a "typhoid talk" (81%) compared with those that did not (63%) (P = 0.0077). Privately owned (56%) and shared pit latrines (38%) were the most common reported sites of defecation; 7% reported open defecation.

Laboratory investigation
Free chlorine residual levels were adequate (FCR ! 0.2 mg/L) in 29 (76%) of the 38 households who reported treatment of stored water and had stored water available for testing and an observed bottle of WaterGuard (Table 5). Among the 38 households, FCR levels were adequate in 80% of 25 that reported attending a "typhoid talk", and in 69% of 13 that reported not attending a "typhoid talk". Tests for bacterial coliforms and E. coli were positive in samples from 5 of 9 boreholes and all 3 public taps tested in survey villages.

Discussion
Sixteen months following the onset of a major typhoid fever outbreak in Neno District, Malawi and after targeted education and prevention interventions by Ministry of Health and partner organizations, household knowledge of the causes and methods of preventing typhoid fever, and adoption of safe water, sanitation, and hygiene practices at the household level, were suboptimal. Educational activities reportedly did not reach almost one-half of the target population. Even among household respondents who did attend a community-wide educational activity, knowledge of the relationship between drinking unsafe water, poor hygiene and typhoid fever was less than ideal. Despite the distribution of free WaterGuard and soap, few households adopted point-of-use water treatment and improved hygiene practices into their regular household routines. Qualitative research conducted at the same time as this investigation revealed persistent underlying skepticism about waterborne transmission of typhoid fever and the effectiveness of water, sanitation, and hygiene interventions to prevent further disease transmission [20]. Beliefs that the outbreak started and spread widely because of an ancestral curse, witchcraft, and 'bad air' in combination with an unusual illness associated with rapidly progressive disease and often fatal outcomes and failure by healthcare clinicians to diagnose and treat the illness likely contribute to this skepticism [20]. These results highlight the need for more effective interventions to improve household knowledge of typhoid fever transmission and prevention, and increase uptake and maintenance of preventive behaviors including regular water treatment and hand washing.
Most households reported getting their household drinking water from unimproved sources, and all stored water in their homes for extended periods. In this setting, household treatment and safe storage of drinking water are universally applicable, but were practiced by fewer than half of all households; therefore, most households remained at high risk for waterborne diseases, including typhoid fever. One possible explanation for the low rate of household drinking water treatment was the lack of regular access to household chlorination products, including WaterGuard; while many households reported having received free WaterGuard, many also reported an insufficient free supply to treat all of their water and few purchased water treatment products. Efforts to distribute free WaterGuard reached most households. However, free WaterGuard supplies were limited, and in most households, free distribution was not sufficient to promote sustained behavior change and regular treatment of all stored drinking water. Uptake in the use of WaterGuard or other household water treatment products may also be hampered by low familiarity with these products before the outbreak and by underlying beliefs regarding other modes of transmission [20][21]. Free chlorine residual levels were adequate in most households who reported using WaterGuard and had a bottle in the home at the time of the visit, suggesting adherence to recommended product use instructions.
Knowledge and adoption of recommended hand washing practices was also limited. Only half of households with soap reported using it to wash hands. Hand washing behaviors at key times, including hand washing after cleaning a child who has defecated and before preparing meals, were reported by less than 50% of respondents. Latrine use, either of a shared or privately owned latrine, was high; however, open defecation was still reported by some households.
Over the past several years, outbreaks of typhoid fever have been documented in other sub-Saharan African nations where access to safe water and sanitation facilities remains limited [21][22][23][24][25][26][27]. The emergence of antimicrobial resistance in these outbreaks complicates case management and outbreak control, highlighting the need for effective, practical interventions to reduce the risk of typhoid fever transmission [17,[22][23]27]. Household stored drinking water treatment, hand washing with soap, and latrine use are all recommended to reduce transmission of enteric illnesses, including typhoid fever. However, encouraging adoption of these practices in every household is challenging and published research describing the behavioral factors that influence adoption and sustained use of these prevention interventions is scarce [28][29]. Several factors could have contributed to poor uptake of recommended prevention methods. Reported barriers included an insufficient supply of water treatment products for regular use and time needed to treat water; unlike in other areas affected by typhoid fever outbreaks, few reported bad taste associated with water treatment products [21]. Other studies have reported similar findings, but also hypothesize that the cost of the recommended products, reluctance to purchase products previously received for free, beliefs about the safety of current water sources and the underlying mechanism of disease transmission, and the lack of ongoing campaigns to promote adoption of recommended behaviors hamper the uptake of recommended prevention practices [20,[29][30][31][32]. Qualitative research conducted in the study area, including focus group discussions and in-depth interviews, found that communities perceived typhoid fever to be dangerous and highly contagious, yet widely-held beliefs about typhoid transmission through curses, witchcraft, and 'bad air' were incompatible with prevention recommendations that focused on water treatment, hygiene and sanitation [20]. Future efforts to change household behaviors will require addressing community concepts about typhoid fever disease causation and transmission and improving routine access to recommended household water treatment products and soap through creative, cost-effective approaches that leverage existing commercial channels [33]. Behavior change is a complex process that is more effective when it is based upon tested and accepted theories that enhance behavioral change and includes repeated interventions to increase knowledge and promote new practices [28]. In this case, efforts to improve safe water, hygiene, and sanitation practices in the affected population could be enhanced through repeated promotional efforts to increase familiarity with available products by community leaders and peers rather than through "oneoff" village-level meetings and product distributions led by "outsiders" [34][35][36]. Furthermore, prevention efforts that occurred in neighboring affected villages in Mozambique, where water treatment products are marketed under a different name, were reported by the community as less intense. Better coordination by health agencies in both countries might have improved adoption of recommended prevention practices by all.
Limitations of this study include a high rate of target household respondent unavailability, the lack of a pre-intervention baseline evaluation for comparison, ongoing prevention activities during the evaluation period, and limited surveillance for typhoid fever. Respondents who were not available for enrollment because of work in the fields or travel outside the village may have responded differently to survey questions than those who were available for participation. The lack of a pre-intervention survey or prior knowledge of whether a household benefited from post-outbreak interventions made it impossible to randomly sample households from populations that either attended or did not attend these talks and to measure the impact that intervention efforts had on changing knowledge or altering practices within the home. Therefore, interpretation of tests of association should be done with caution. Ongoing prevention activities during the evaluation, including the distribution of WaterGuard, may have influenced household responses. Limited surveillance data from Neno District makes it difficult to associate the impact interventions may have had on knowledge, attitudes and practices in the home with the apparent decrease in typhoid fever illnesses and deaths.
In summary, despite ongoing outbreak interventions, including community-wide educational campaigns and distribution of WaterGuard and soap, knowledge regarding the causes and prevention of typhoid fever and ownership and use of products that help reduce disease transmission remained low, even among household respondents who reported benefiting from these interventions. Future efforts to improve household water quality and sanitation and hygiene practices need to be more forceful and sustained until deficiencies in improved water supply and sanitation infrastructure can be fully addressed. In the interim, the option to complement these efforts with a targeted typhoid fever vaccination campaign should be strongly considered [20].