3.1. Baseline Utilization Study
We enrolled 117 persons who met the definition of flocculent-disinfectant user and 193 control-persons who had never used the flocculent-disinfectant (Table 1
). Flocculent-disinfectant users were more likely to use a turbid water source (Odds Ratio [OR] = 19.7, 95% Confidence Interval [CI] = 3.1–812) and to attribute diarrhea to their drinking water (OR = 2.5, CI = 1.4–4.6). Users were less likely to express the belief that diarrhea is a serious problem in the community (OR = 0.4, CI = 0.3–0.7). Mean spending on soap and toothpaste was significantly higher for users versus
non-users (46.5 versus
37.2 Ksh, p
= 0.02). The mean socioeconomic status index was significantly higher for users than non-users (p
After adjustment for economic status index, spending on soap and toothpaste, and knowledge of the previous CDC/KEMRI study, two factors remained significantly associated with flocculent-disinfectant use. Use of turbid water sources was strongly associated with flocculent-disinfectant use (Adjusted Odds Ratio [AOR] = 19.7 CI = 2.5–153; p = 0.004). Those who used flocculent-disinfectant remained less likely to express the belief that diarrhea is a serious problem in the community (AOR = 0.4, CI = 0.3–0.7; p = 0.001).
3.2. Follow-up Utilization Study
Of the 117 users in the baseline utilization study, 104 (89%) completed questionnaires for the follow-up study. (Table 2
) Of those interviewed, eight (8%) reported using flocculent-disinfectant in the past 7 days. Twenty-six (25%) had not used the flocculent-disinfectant since the time of the baseline study. Overall, 50 (48%) reported treating their water by some method in the past 7 days. Of those who did not use the flocculent-disinfectant consistently, the most commonly cited reasons were lack of availability (66%) and expense (20%). Of the 78 (75%) respondents reporting flocculent-disinfectant use since the study period in December 2003, 65 (83%) purchased the flocculent-disinfectant directly from a SWAP representative and only 11 (14%) reported purchase from a duka
(small shop). In contrast, of the 74 (71%) respondents who reported use of sodium hypochlorite in that time same period, 37 (47%) reported purchase from a duka
and 20 (26%) reported purchase from a market.
Although 18 (17%) respondents reported daily use of either the flocculent-disinfectant or the sodium hypochlorite solution on the questionnaire, only 14 reported chlorinating the water stored in their home at the time of the interview; 11 of these 14 (11% of 104 total respondents) had free chlorine present in their stored water.
On bivariate analysis, drinking water from turbid sources at least 4 months per year was reported by 96 (92%) respondents. Socioeconomic status was not associated with reported consistent use (OR = 0.9, CI = 0.7–1.3; p = 0.6).
On multivariate analysis, after adjusting for economic status and awareness of the previous CDC/KEMRI study, respondents who reported consistent use were less likely than reported sporadic users to express the belief that their drinking water made their family sick (AOR = 0.34, CI = 0.1–0.9; p = 0.03). Socioeconomic status was not significantly associated with reported consistent use after adjustment for these other factors. These associations were essentially unchanged regardless of whether reported consistent use was defined by reported volume of flocculent-disinfectant used or by confirmation of presence of free chlorine in the household water at the time of the interview.
3.3. Use-Prevalence Survey
Of the 1,530 compounds listed in the most recent DSS census of Asembo, a total of 1,452 (95%) were included in the survey. Five-hundred-thirty-one (37%) compounds reported ever using the sodium hypochlorite solution compared with 105 (7%) who reported ever using the flocculent-disinfectant. Two-hundred-twenty-four (15%) reported use of the sodium hypochlorite in the past 7 days while 14 (1%) reported using the flocculent-disinfectant in that time period. Overall, 549 (38%) compounds reported ever using some form of household-level water treatment and 231 (16%) reported household-level water treatment in the past 7 days. Village-specific rates for ever using the flocculent-disinfectant varied from 0.7% to 16%. Rates for use of flocculent-disinfectant the past 7 days ranged from 0% (6 villages) to 13%. Reports of ever using the sodium hypochlorite solution ranged from 21% to 59%, while rates of use in the past 7 days ranged from 7% to 27%.
These studies demonstrate a complex array of issues contributing to use of household-level water treatment products in western Kenya. While initial use of the flocculent-disinfectant was strongly associated with having turbid drinking water, this association did not persist in the study of reported consistent use. Although cost is often cited anecdotally as a reason for lack of use of household-level water treatment products, in our study economic status was not associated with reported consistent use among early users. Improvements in health do not seem to definitively influence use either: Luby et al.
have demonstrated that even the experience of decreased diarrheal disease burden among residents of rural Guatemala was not adequate to motivate consistent use [10
Dependence on a turbid water source emerged as the strongest motivator for flocculent-disinfectant use in this setting. The association with turbidity persisted after adjusting for socioeconomic status, spending on personal care items, and beliefs about the relationship between water and health. This result supports the hypothesis that the ability of flocculent-disinfectant to visibly clear turbid water is a compelling impetus to initial use. However, the allure of clearer water was not associated with reported consistent use based on the data from the follow-up survey. In this cohort with prior experience with flocculent-disinfectant and a high dependence on turbid water, sporadic use of sodium hypochlorite solution was comparable to use of the flocculent-disinfectant (71% versus 78%). The relatively high use of sodium hypochlorite despite the turbid water burden may be a reflection of familiarity with the sodium hypochlorite solution. Lower cost or greater ease of use for sodium hypochlorite may also have been determinants of use despite the advantages of flocculent-disinfectant for those using tubid water. Our data suggest that consumers often tried both locally available products, but reported using sodium hypochlorite more consistently than flocculent-disinfectant, for both the past year and the past week. Seventy-five percent of those who used the flocculent-disinfectant since 2003 also used sodium hypochlorite solution in that time period. In both the community as a whole and among those who used flocculent-disinfectant at baseline, the prevalence of sodium hypochlorite use eventually surpassed flocculent-disinfectant use. Thus, although dependence on turbid water correlated with trying flocculent-disinfectant, other factors appear to influence the decision to treat household water consistently and what product to use for this treatment. Since the time of the study, flocculent-disinfectant has expanded to national distribution networks in Kenya; this expansion may increase use by addressing the issues of availability that we found in our study.
Economic factors clearly influenced usage patterns. The choice of sodium hypochlorite over flocculent-disinfectant may largely be a function of the difference in retail cost as sodium hypochlorite solution cost less than 1 US cent per 20 L of water treated while flocculent-disinfectant cost 12 US cents per 20 L treated. Use did in fact decline remarkably in the follow-up survey with 25% of initial users reporting they never used the flocculent-disinfectant product again; however, the lack of a statistically significant relationship between reported consistent use and socioeconomic status in the follow-up survey suggests that something besides finances also affects usage patterns. The manufacturer is undertaking price-reduction studies in rural Kenya for the flocculent-disinfectant; these may provide a sense of how much affordability ultimately impacts use.
Lack of availability emerged as an important determinant of flocculent-disinfectant use based on data from the cohort of prior users. Based on qualitative data from interviews, problems with flocculent-disinfectant distribution caused gaps in availability that in turn pre-empted use. Availability of flocculent-disinfectant in the local market decreased dramatically after the change in credit policy at SWAP. Rural community groups who served as vendors during the initial phase of sales did not have adequate cash resources to purchase flocculent-disinfectant in bulk. Without income generation from the wholesale purchases, these groups could not sustain the retail market. Those who reported buying sodium hypochlorite reported purchases from multiple sources including dukas, markets and chemist shops. These locales are part of the indigenous consumer culture, and availability there made sodium hypochlorite much more accessible than flocculent-disinfectant, which had minimal penetration into these venues. Further penetration into the conventional retail sector may contribute to increased use of the flocculent-disinfectant through more consistent availability.
The documented prevalence of nearly 40% for ever using household-level water treatment products in this rural Kenyan setting demonstrates their potential as a way for even severely economically disadvantaged persons to benefit from safe water. The challenge lies in getting households to adopt this proven intervention. Behavior change communication can help; teaching safe water handling in elementary schools and clinics has demonstrated increased household use of water treatment products in pilot studies [15
]. These factors may not be sufficient motivation if prices are too high. In our context, micro-credit programs through an NGO made it possible for communities to purchase stock at wholesale prices thus making the products accessible to more people. In another study in western Kenya, Freeman et al.
found that although awareness of household-level water treatment products was high across wealth quintiles, use dropped precipitously in the lowest quintile [17
]. In the poorest segments of the population, where morbidity and mortality from waterborne diseases are highest, consistent use of either household-level water treatment product may require subsidies outside of the retail market for the foreseeable future.
Inspiring sustained use will require consistent availability, affordability in the local context, and a more comprehensive understanding of the factors that motivate those who consistently treat their water. This understanding will require further research and data-driven implementation strategies that address the behavioral and economic issues along with the public health issues. Such strategies could be informed by more in-depth behavioral research to further explain the behaviors and choices documented in our studies and specifically assess the relationships between use and social marketing activities. The World Health Organization’s International Network to Promote Household Water Treatment and Safe Storage, a collaboration of UN agencies, bilateral development agencies, international non-governmental organizations, research institutions, international professional associations, the private sector, and industry associations provides an integrated forum for identifying research needs on household-level water treatment and informing policies and programs [18
The study was limited by several factors. Low prevalence of usage in the community made it difficult to determine robust statistical associations for factors affecting flocculent-disinfectant use. Small sample size also prevented comparisons between those who used various combinations of water treatments, and analysis of seasonality of use; however, the sample sizes were sufficient to identify major factors associated with use.
Courtesy bias likely resulted in some over-reporting of use, based on the results of the utilization study in which 16% of those reporting chlorination did not test positive for residual free chlorine. In addition, cross-sectional studies do not permit an objective assessment of consistent use.