AJPH -- eLetters for Eisenberg et al., 97 (5) 846-852

Using stochastic modelling, Eisenberg and colleagues show how multiple transmission pathways at the community and household levels conspire to infect vulnerable populations with the agents responsible for the enormous burden of diarrhoeal disease.¹ They suggest that as a result, an integrated approach is required in order for environmental interventions to reduce the burden of diarrhoeal disease. They assert, for example, that �[w]hen sanitation conditions are poor, water quality improvements may have minimal impact regardless of the amount of water contamination.�

As the authors note, this conclusion was the same as that reached by Esrey² and VanDerslice and Briscoe³. Esrey drew his conclusions from an analysis of cross-sectional data from the Demographic and Health Surveys (DHS), a methodology that he acknowledged to be inferior to randomized controlled trials and that was criticized on other grounds⁴. VanDerslice and Briscoe reported on an observational study of conventional source-based water supplies (protected wells, boreholes, communal tap stands), an intervention that has been shown to leave water subject to recontamination.⁵ The authors might have also cited Briscoe�s work from a decade earlier in which he used transmission modelling to show the impact of multiple exposure pathways.⁶ However, even Briscoe acknowledged that a single environmental intervention such as improvements in water quality could significantly reduce diarrhoea if it were the �dominant transmission route,� citing John Snow�s celebrated work in London. Nevertheless, policy makers and programme implementers have relied on this evidence to question the value of water quality interventions in settings were sanitation has not been addressed.

These conclusions, however, are not supported by a large number of randomized and quasi-randomized controlled field trials (RCTs), usually considered the gold standard for epidemiological evidence. Many of these intervention studies examine interventions to improve water at the point of consumption in order to minimize the risk of recontamination. We recently conducted a Cochrane review of 38 trials of water quality interventions for preventing diarrhoea involving more than 53,000 persons from 19 countries over 20 years.⁷ Pooled estimates of effect showed that water quality interventions were effective in settings without �improved� sanitation (defined in accordance with the WHO/UNICEF Joint Monitoring Programme): among four trials reporting rate ratios, the pooled estimate of effect (random effects model) was 0.78 (95%CI 0.64 to 0.95); two trials reporting risk ratios had an even larger protective effect, with a pooled estimate of 0.55 (95%CI 0.47 to 0.65). While pooled estimates of effect were even greater in settings with improved rather than unimproved sanitation, the difference was not statistically significant.

Eisenberg and colleagues acknowledge that intervention trials have shown the protective effect of water quality interventions, but argue that �the range of efficacies seen in water intervention trials illustrates that when sanitation levels are poor, water quality projects may have minimal effect on public health.� In fact, there was substantial heterogeneity in results from all water quality interventions, and sub-group analysis based on level of sanitation did not reduce such heterogeneity. However, a careful review of the individual trials that reported on sanitation conditions shows that most water quality interventions implemented in settings without improved sanitation were nevertheless effective in preventing diarrhoea (Table 1). These results do not include large and rigorous RCTs in outbreak settings where water quality interventions have also been shown to be effective in preventing diarrhoea in settings with minimal sanitation coverage.⁸

Reference	Description	Relative Risk (and 95% CI)
Chiller 2005	RCT in 12 rural villages in Guatemala	0.62 (0.47, 0.82)
Clasen 2004	RCT in rural village in Bolivia	0.30 (0.19, 0.47)
Clasen 2005	RCT in three rural settlements in Colombia	0.40 (0.21, 0.76)
Clasen 2006	RCT in rural village in Bolivia	0.47 (0.24, 0.92)
Garret 2004	Quasi-RCT in rural Kenya	0.44 (0.28, 0.69)
Gasana 2002	Quasi-RCT in rural Rwanda	1.00 (0.89, 1.12)
Handzel 1998	RCT in slum in urban Bangladesh	0.67 (0.54, 0.84)
Kirchoff 1985	Blinded, cross-over RCT in rural Brazil	1.07 (0.88, 1.30)
Quick 1999	RCT in 2 peri-urban communities in Bolivia	0.57 (0.39, 0.84)

Table 1: Randomised and quasi-randomised controlled trials of water quality interventions in settings without improved sanitation (adapted from Clasen⁷)

As Eisenberg and colleagues note, interventions to improve water quality, hygiene, and sanitation have each been independently shown to be effective in preventing diarrhoea.^9,10 If the transmission pathways are at least partially distinct, as the F-diagram they cite shows, then intuitively, their protective effect might be somewhat synergistic. If so, this would support the advantage of the integrated approach that Eisenberg and colleagues advocate based on the results of their stochastic model. Once again, however, the evidence from field trials does not support such a conclusion. In meta-analysis, interventions that combine improvements in water quality with improvements in sanitation or even hygiene instruction do not appear to be more effective than interventions to improve water quality alone.^6,9 This surprising result was observed even by Esrey.¹¹ Fewtrell and colleagues offer a number of possible reasons, but stress the importance of this conclusion on policy: �The lack of evidence of success of multiple interventions is important, because many large-scale, publicly funded interventions in less developed countries follow a model in which water supply, sanitation facilities, and hygiene education are provided even when recipients are primarily motivated by the desire to obtain a more convenient or reliable water supply.�⁹

The above-cited systematic reviews both cite a variety of shortcomings with the intervention studies included in therein that diminish the strength of the conclusions that can thus be drawn therefrom.^7,9 Important differences in study design, setting, intervention, and the method of collecting, measuring, and reporting morbidity limited the potential to meta-analyse results and contributed, perhaps, to the heterogeneity in most pooled estimates. There was also evidence of publication bias. As Eisenberg and colleagues observe, few of the trials were blinded. And as we emphasised in our review, these blinded trials showed no statistically significant protective effect, possibly for other reasons we noted. Nevertheless, as our analysis of the methodological quality of all 38 trials demonstrated, the considerable protective effect of water quality interventions that was reported by most RCTs cannot be dismissed for lack of overall rigor.

The transmission model used by Eisenberg and colleagues is sound and their results are biologically plausible. They acknowledge the need to focus on the critical pathway under certain circumstances. However, their suggestion that an integrated approach may be necessary in order to achieve real progress in preventing diarrhoeal disease must be reconciled with the growing number intervention studies showing the effectiveness of discrete environmental interventions. By definition, integrated approaches are more complex and costly, and deferring the introduction of simple water quality interventions—such as point-of-use chlorination or solar disinfection—until we can deal with the vast challenge of improving sanitation may deprive millions from real, substantial, and cost-effective health gains. Rigorous RCTs in settings with varying sanitation coverage and employing new methods designed to minimize the biases that tend to accompany assessments of diarrhoeal disease will help clarify the potential for water quality interventions to prevent diarrhoea even in the absence of sanitation.¹² In the meantime, while simulation studies provide a valuable contribution to our understanding of environmental interventions to prevent diarrhoea, their conclusions should be read in the context of the entire body of epidemiological evidence and weighed by policy makers and programme implementers in accordance with their relative probative value.

1. Eisenberg JNS, Scott JC, Porco T. Integrating public health control strategies: balancing water sanitation andhygiene interventions to reduce diarrheal disease burden. Am. J. Pub. Health 2007. doi 10.21.2105/AJPH.2006.086207

2. Esrey SA. Water, waste, and well-being : a multi-coutry study. Am J. Epidemiol. 1996; 143:608-623

3. VanDerslice J, Briscoe J. Environmental interventions in developing countries: interactions and their implications. Am J Epidemiol 1995; 41:135-144.

4. Cairncross S, Kolsky PJ. Re: Water, Waste and Well-Being: a Multicountry Study. Am J. Epidemiol 1997;146:359-360.

5. Wright J, Gundry S, Conroy. Household drinking water in developing countries: a systematic review of microbiological contamination between source and point-of-use. Trop. Med. & Int�l Health 2003;9(1):106-17.

6. Briscoe J. Intervention studies and the definition of dominant transmission routes. Am. J. Epidemiol. 1984;120(3): 449-55.

7. Clasen T, Roberts I, Rabie T, Schmidt W, Cairncross S. Interventions to improve water quality for preventing diarrhoea. (A Cochrane Review). In: The Cochrane Library, Issue 3, 2006. Oxford: Update Software

8. Colwell RR, Huq A, Islam MS, Aziz KMA, Yunus M, Khan NH, Mahmud A, Sack RB, Nair GB, Chakraborty J, Sack DA, Russek-Cohen E. Reduction of cholera in Bangladeshi villages by simple filtration. Proc. Nat. Acad. Sci. 2003;100(3): 1051-5

9. Fewtrell L, Kaufmann R, Kay D, Enanoria W, Haller L , Colford J. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infect Dis 2005;5:42-52

10. Curtis V, Cairncross S. Effect of washing hands with soap on diarrhoea risk in the community: a systematic review. Lancet Infect Dis 2003; 3:275-81

11. Esrey SA, Potash JB, Roberts L & Shiff, C. Effects of improved water supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bull. WHO 1991;69: 609-21

12. Schmidt WP, Luby S, Genser B, Barreto M, Clasen T. Estimating the longitudinal prevalence of diarrhoea: an alternative to continuous surveillance. Epidemiology (in press)