To a greater degree than with water supply and sanitation, lamentably little reliable evidence exists on the cost or the effectiveness of interventions to change hygiene behavior and still less on the relative cost-effectiveness of different approaches to the design of such interventions.
The Shortage of Evidence
With regard to effectiveness, Loevinsohn (1990) reviewed health education interventions in developing countries and applied four relatively modest criteria of scientific rigor to the 67 published studies he found:
a description of the intervention in sufficient detail to allow its replication
an objective outcome measure, based either on health status or on behavior change
a control group and a sample size greater than two clusters or 60 individuals
a description of the target population (in terms of their level of education and other factors) adequate to permit a judgment of the relevance of the study to other contexts.
Only three studies were found to meet all four criteria. One (Stanton and Clemens 1987) dealt with environmental hygiene promotion and raises some doubts—although the hygiene behavior of the intervention group was better than the control, both were significantly worse than they had been before the intervention.
A subsequent review of 31 studies (Cave and Curtis 1999) found 5 more studies that could be considered methodologically sound, but none showed a clear effect on behavior. Of a further 11 studies of "reasonable" rigor, only two showed a major effect on behavior.
Shortcomings also exist in the cost data. Many costings are based on budget forecasts and not on real expenditures. Even when actual expenditures are used, major difficulties exist in apportioning the overhead costs that make up a significant proportion of the total. Health educators and the resources they use (such as vehicles) are rarely dedicated exclusively to health education. A further problem in the derivation of unit costs is agreeing on the denominator, which can be the number of people attending health education sessions, the number of members in their households, or the number of people in the target catchment area. For those reasons, different analysts are likely to derive different unit costs from the same data; indeed, the same authors have on occasion arrived at widely differing unit cost figures from the same data.
Time adds a further dimension to this discussion. Do interventions to promote hygiene behavior change have to be implemented continuously, or at least annually, if their effect is to be sustained, or are such changes self-sustaining?
We will take the last question first. Wilson and Chandler (1993) returned after two years to a population in which a four-month intervention to promote hand washing with soap had included provision of free soap. They found that 79 percent of mothers, the original target group, had continued the practice despite the fact that they now had to buy the soap.
Further evidence of the sustainability of new hygiene behaviors was found by Cairncross and Shordt (2003) in a collaborative study with partner organizations in six developing countries in Africa and South Asia. Target populations of previous hygiene promotion projects were visited at 12-month intervals, and various indicators of hygiene behavior were assessed and compared. In four of the six countries, indicators for populations in which the intervention had ended relatively recently were also compared with those in areas where the last intervention had ended several years previously. Those two types of comparison, with the various indicators assessed in each country, allowed a total of 46 comparisons to be made. Only in three such comparisons was there any indication of a falling-off of hygiene with time since the intervention ended; in one case, the falling-off was attributable to the deteriorating condition of the latrines from wear and tear rather than to a decline in compliance.
In some cases, new hygiene practices have become stronger or more prevalent after the ending of external intervention to promote them, as they become self-propagating and consolidated in the community's material culture (Allan 2003).
It is likely that hygiene promotion activities need to be repeated from time to time—say, every five years—but are not required on a continuous basis. It follows from this observation that calculations of cost-effectiveness should take into account the morbidity and mortality averted not only during the implementation of the intervention, but also for a number of years—perhaps five—thereafter.
Cases in which the costs as well as the effectiveness of hygiene promotion programs have been documented objectively are few indeed. In the absence of suitable data, Varley, Tarvid, and Chao (1998) calculated a costing for a typical program from first principles, arriving at a cost of US$3 (range US$2 to US$3) per household per year, or US$0.60 per capita.
One of the few cases in which data exist is a program in urban Burkina Faso described by Borghi and others (2002). Their data show that the total cost to the provider of the three-year intervention was US$0.65 per capita, or US$4.54 per seven-person household, after deducting the cost of the international research component. Of this total, 63 percent is composed of administration and undifferentiated start-up costs of the project. Most of the remaining costs were accounted for in roughly equal measure by house-to-house visits, discussions in health centers, hygiene lessons in schools, and street theater presentations.
Additional costs were incurred by the 18.5 percent of households that complied, practicing improved hygiene as a result of the program, amounting to US$8 per household per year. More than 90 percent of that sum was the cost of soap for hand washing.
However, on the basis of the observed increase in prevalence of hand washing with soap, the intervention was estimated to have averted sufficient diarrhea morbidity and mortality to save US$2.80 per household per year (US$15 per compliant household per year) in direct costs of medical care and indirect costs attributable to lost productivity. Of this total, 93 percent represented the lost future productivity associated with the deaths of young children.
Waterkeyn (2003) provides an example from rural Zimbabwe. In the two districts in which the Community Health Clubs approach was examined, it was successful in increasing the prevalence of hand washing with soap among the club members by 6 percent and 37 percent, respectively, and it was successful in reducing the prevalence of open defecation by 29 percent and 98 percent, respectively. The marginal cost of the intervention, which used existing health staff, was US$4.00 per club member, or an average of US$0.67 per member of an affected household. Including the salaries of staff members would roughly double the figure to about US$1.40 per capita.
Those figures can be compared with an estimate of US$5.00 per mother (in 1982 dollars) by Phillips and others (1987) based on a review of several programs. Assuming that roughly 1 in 10 members of the population are mothers of young children, this cost is equivalent to about US$0.50 per capita. For cost-effectiveness analysis, a nominal cost of US$1.00 per capita is, therefore, taken because it is roughly the midpoint of the range of recent estimates.
Effect on Diarrhea
Esrey and others (1991) found only six studies of the effect of hygiene promotion interventions on diarrhea morbidity, with a median reduction of 33 percent. A subsequent review by Huttly, Morriss, and Pisani (1997) arrived at a similar result—a median reduction of 35 percent.
The interventions promoting the single hygiene practice of washing one's hands with soap tended to achieve greater reductions in disease than those that promoted several different behaviors. That finding was confirmed by a systematic review of the literature on hand washing (Curtis and Cairncross 2003), which concluded that hand washing with soap—and interventions to promote it—could reduce diarrhea morbidity by 43 percent and life-threatening diarrhea by 48 percent. Because the effect of diarrhea prevention in DALYs is mainly attributable to the prevention of diarrhea deaths, the higher of these two figures is more appropriate for calculating the effect of hygiene promotion on the burden of disease.
It is not surprising that interventions advocating more behavior changes should have less effect, because numerous messages dilute one another in the minds of the target audience. Because some of the interventions in the systematic review were planned without an adequate prior program of formative research, it is possible that they could have had a still greater effect if they were better conceived.
Effect on Respiratory Infections
Reasons exist to believe that hand washing with soap could be a cost-effective intervention not only against diarrheal diseases, but also for the prevention of acute respiratory infections (ARIs). The intervention is plausible, given what is known about the transmission routes of ARIs, and there is also epidemiological evidence, in that all six published studies of the effect of hand washing on ARIs show a significant reduction (Cairncross 2003b).
These two disease groups are the most important causes of child mortality worldwide, and respiratory infections also cause significant adult mortality, for which no alternative preventive intervention is yet available, field-tested, and ready for implementation. A randomized, controlled trial of the efficacy of hand-washing promotion on an ARI outcome is an urgent priority for future research.
Interactions with Water Supply and Sanitation
It can be argued that there is little point in encouraging people to wash their hands if they do not have access to water or to use a latrine if they do not have one.
The argument has only limited validity where sanitation is concerned; an important role for any hygiene promotion is to promote sanitation itself. With regard to water, in the studies reviewed by Curtis and Cairncross (2003), the reductions in disease achieved by hand washing in settings with indoor piped water supply were not significantly different from those achieved elsewhere. Given that the rationale is ambivalent and the evidence inconclusive, the simplest plausible assumption is that the effects of water supply, sanitation, and hygiene promotion on diarrhea are independent and additive to one another.