The Value of Time

The saving in time and drudgery of carrying water home from the source is substantial, and several reasons exist to attribute a money value to it. The most powerful argument for the money value of poor women's time is that households often pay others to deliver their water, or pay to collect from nearby rather than from more distant sources that are free of charge. Thompson and others (2001) found that, of urban East African households lacking a piped supply, the proportion paying for water had increased from 53 percent to 80 percent over 30 years. In a survey of 12 sites in 10 countries, Zaroff and Okun (1984) found that households were spending a median of over 20 percent of their income on the purchase of water from vendors. The prices charged by vendors are typically more than 10 times—and can be up to 50 times—the normal tariff charged by the formal water supply utility.

Cairncross and Kinnear (1992) found that vendor prices increased with the time required to collect the water, showing that households pay more as the alternative of collecting water themselves becomes more burdensome. If the amount paid to the vendor for bringing the water is divided by the time saved from collecting it, the implicit value that people ascribe to their time can be calculated. Whittington, Mu, and Roche (1990), working in rural Kenya, showed in this way that the implicit value of the time saved was roughly US$0.38 per hour, very close to the average imputed wage rate for such households of US$0.35 per hour.

Because the poorest urban households typically spend more than 90 percent of their household budget on food, the money they spend on water is sacrificed from their food budget (Cairncross and Kinnear 1992). The provision of water more cheaply thus offers a substantial nutritional benefit to the poorest.

Assessing the Time Saved

The cost of water collection in rural areas is usually in time and effort rather than in money paid to vendors. The saving in time and drudgery underlies many social benefits. Given the relevance of the time-saving benefit to water supply policy and the fact that the benefit is usually uppermost in the mind of the consumer, it is remarkable how few data have been collected on the amounts of time spent collecting water.

Working in 334 study sites in Kenya, Tanzania, and Uganda, Thompson and others (2001) found a mean distance from rural unpiped households to their water sources of 622 meters. In urban areas, the distance was only 204 meters, but queuing at the tap meant that a water collection journey took almost as long.

Feachem and others (1978) found in 10 villages of the densely populated lowlands of Lesotho that the installation of a water supply had saved the average adult woman 30 minutes per day. In one-third of the villages, the saving per woman was more than an hour a day. Lesotho has many springs, so that time saving is likely to be on the low side compared with Africa as a whole.

These time-saving benefits are confirmed by the Multi-Indicator Cluster Surveys of the United Nations Children's Fund (UNICEF). A recent analysis of the responses in 23 African countries has produced a more representative account of water collection journey times in that continent (G. Keast, UNICEF, personal communication 2003). Nearly half the households interviewed (44 percent) required a journey of more than 30 minutes to collect water, implying that the women in such households spent an hour or more each day in water collection. At almost any reasonable level of service, most of that time would be saved by an improved water supply.

In Asia, an Indian national survey for UNICEF found that women spent an average of 2.2 hours per day collecting water from rural wells (Mukherjee 1990). A study in Sri Lanka, which is generally considered to be well provided with water sources, found that 10 percent of women had to travel more than 1 kilometer to their nearest source (Mertens and others 1990).

Valuation of the Time-Saving Benefit

Putting a precise figure on the money value of the time of poor people is a tricky task, even for the most self-confident economist. In 1987, Churchill and others took US$0.125 per hour as an illustrative but not unrealistic figure. To take the same figure today could hardly be described as extravagant. Assuming this valuation of an hour of time—and that a water supply bestows a mean saving of only 15 minutes per person per day—yields a conservative estimate of the value of the time-saving benefit of US$11.40 per year. The data presented earlier indicate that, at least in Africa, the true figure is nearer to double that amount, enough to justify the full construction cost of a dug well or borehole supply in a single year. In Latin America and the Caribbean, costs are higher, and time savings may be less, but rural incomes are also higher—and so, therefore, is the value of people's time. Little doubt exists that, in all three regions of the developing world, the value of time saved is sufficient on its own to justify both the investment costs (at any reasonable rate of amortization) and the operation and maintenance costs of water supplies.

Even in settings where water vending is not common, contingent valuation surveys have widely demonstrated a willingness to pay for water supplies, particularly at the level of service of house connections (World Bank Water Demand Research Team 1993). In general, such measured willingness to pay has exceeded the cost of providing the supplies, and payment to vendors often exceeds it by many times.

Policy Implications

Whether the consumers actually pay for the full value of the time-saving benefit, it is what makes water supplies popular and largely it is what motivates politicians to invest in them. More than half the total annual investment in water supply in the developing countries of Africa, Asia, and Latin America and the Caribbean is from domestic sources (WHO and UNICEF 2000). Most of the investment is from the public sector. In general, investments in water supply—whether by the governments of developing countries or by external support agencies—do not come from health sector budgets and are not compared with other health interventions when investment decisions are made, even though health benefits do arise from water supply improvements.

Water supply is thus a health-related intervention that comes without cost to the budgets of the health sector. Although it undoubtedly offers health benefits, it has a sufficient economic and political rationale in other social benefits associated with time saving. The health benefits are a positive externality to this rationale. However, this fact does not mean that the authorities responsible for public health should ignore the water sector. The function of the health sector is one of regulation, advocacy, and provision of supplementary inputs, as appropriate, to ensure that potential health benefits of water supply are realized to the optimal extent.

For example, the regulatory role of the health sector in quality surveillance of drinking water is well known and widely accepted. Substantial and largely unexploited additional potential is present in this role if quality is interpreted in the wider sense of quality of service rendered by the water supply utility, in terms not only of water quality but also of quantity, continuity, coverage, control of sanitary hazards, and cost. Those other aspects, as will be argued in the following sections, are no less important for health.

Where a regulatory role is not available to the health sector or agencies concerned with public health, advocacy can be no less cost-effective. For example, connection charges are a major barrier to house connections for low-income groups. In many cities of the developing world, the individual connection charge is about a month's basic wage. Advocacy of lower connection charges, with the amount recovered from the monthly water tariffs, can therefore help achieve an increase in the number of people who have house connections and who can benefit from the corresponding health gain at no cost to the public purse. Finally, the health sector can provide important complementary services, such as hygiene promotion and promotion of low-cost sanitation to increase coverage; because of the nature of such services, the water sector, with its focus on technology, is ill-equipped to offer them.

The unit costs of such regulation and advocacy are minimal. One example is the case of UNICEF's participation over the past 30 years in India's rural water supply program. UNICEF's investment has represented no more than 1 percent of the total, but its influence has played a central part in the evolution of the technical and institutional model of the program that supplies water to 1 in 10 members of the human race.

An example of the effectiveness of such measures is provided by the interventions of the Mexican Ministry of Health in June 1991. Fostered by fear of the devastating effects of cholera, these measures included the chlorination of water supplied for human consumption and the prohibition of sewage irrigation of fruit and vegetables. As a result, the incidence of diarrhea in children under five years of age fell from 4.5 to 2.2 episodes per child-year, and the corresponding mortality rate fell from 101.6 to 62.9 per 100,000 children (Gutiérrez and others 1996).

The current rate of annual investment per capita in water supply and sanitation, including both national investment and external aid funds, is reportedly US$2.25 in Asia, US$7.53 in Africa, and US$8.87 in Latin America and the Caribbean (WHO and UNICEF 2000). One percent of the water sector's investment would, therefore, be US$0.02 to US$0.10 per capita. If each ministry of health in the developing world were to invest such a sum in public health advocacy and regulation related to water supply, the sector's performance, at least where low-income groups are concerned, could be transformed. It is hard to put a figure on the health effects of such investment, but the Mexican example suggests that they would be substantial. For the sake of cost-effectiveness estimation, such spending is arbitrarily assumed to have the effect of ensuring improved water supplies for an additional 10 percent of the population to which it refers.

Classification and Burden of Water-related Diseases

Before Bradley's classification can be applied to diseases (rather than transmission routes), it requires a small adjustment (Cairncross and Feachem 1993) to allow for the fact that practically all potentially waterborne infections that are transmitted by the feco-oral route can potentially be transmitted by other means (contamination of fingers, food, fomites, field crops, other fluids, flies, and so on) all of which are water-washed routes. In addition to the feco-oral infections, a number of infections of the skin and eyes can be considered water washed but not waterborne. The final classification is shown in table 41.1.

Table 41.1

The Bradley Classification of Water-related Infections.

The classification can now be used to assess how the disease burden prevented by water supply is distributed among disease groups. Bradley himself did this, a time long before the disability-adjusted life year (DALY) had been invented as a unit of benefit measurement (White, Bradley, and White 1972, 191). He used official statistics on the number of cases of each disease diagnosed and treated by health services in East Africa and combined them with notional percentages by which morbidity and mortality caused by each condition could be expected to fall if water supply were "excellent."

Those notional reductions were based on subjective assessments of the literature available at the time and were described by their author as "little more than guesses," but it is hard to prove many of them seriously at fault, even today. A selection is presented in table 41.2.

Table 41.2

Percentage Reductions in Disease Rates Assumed by Bradley.

The result of these calculations was that the feco-oral disease group accounted for 91 percent of the deaths preventable by water supply, 50 percent of inpatient bed nights, and 33 percent of outpatient consultations. Rosen and Vincent (2001) have made a similar calculation for the whole of Africa in 1990 and found that the feco-oral group accounted for 85 percent of the preventable DALYs. When measured in terms of deaths or DALYs, feco-oral infections account for the vast majority of the impact, because of the high mortality caused by diarrheal diseases among young children. Most deaths from diarrheal diseases are of children younger than age five, and most of those are among children younger than two. A child death averted is worth 30 DALYs. Varley, Tarvid, and Chao (1998) have calculated that for diarrhea morbidity reduction to have the same effect in DALYs as averting one such death, it would have to prevent 115,000 child-days of diarrhea. After the diarrheal diseases, the next most important category in terms of DALYs (12 percent of the total) is the water-based group, primarily schistosomiasis. The purely water-washed diseases, mainly skin infections, represent a more conspicuous portion only when compared in terms of the burden placed on health services by inpatients or outpatients.

How representative is this African breakdown of the developing world as a whole? Diarrheal disease among poor communities is cosmopolitan. A global review of studies of the incidence of diarrhea morbidity could find no clear geographic or climatic trend (Bern and others 1992), so the burden of disease is no doubt similar around the developing world. The second most important disease group is represented by schistosomiasis, which is absent from much of Asia and Latin America. The relative importance of feco-oral disease is, therefore, likely to be still greater in the poor communities of Asia and the Western Hemisphere than it is in Africa.

Epidemiological Questions and Problems

The predominant contribution of feco-oral diseases to the burden of disease attributable to water supply raises an important question, because this group can be transmitted by both waterborne and water-washed routes. It is important for the water engineer to know whether scarce funding should be spent on improved water treatment and measures to protect water quality or instead on providing a limitless supply of water at a high level of access and convenience and encouraging its use for improved hygiene practices. We need to know, that is, whether the feco-oral infections endemic in poor communities are mainly waterborne or mainly water washed.

Moreover, the fact that some diarrheal diseases are still prevalent in communities with a high level of water supply service indicates that water supply alone cannot completely prevent these diseases. A further question then, is this: by how much do water supply improvements reduce diarrheal diseases?

Numerous studies have sought to answer these questions, but they are hard to answer rigorously, for several reasons. First, it is almost impossible, ethically and politically, to randomize the intervention. Where the intervention is an improvement in the level of access to water, it cannot be blinded; no placebo exists for a standpost. Where quasi-experimental studies have been used—opportunistically exploiting an intervention allocated by political or technical means—significant confounding has frequently been found (Briscoe, Feachem, and Rahaman 1985).

Confounding has been especially intractable in studies in which the allocation of facilities has been on a household basis, so that the exposure groups are self-selected—for instance, studies in which individual households that have chosen to install a private tap are compared with others that have chosen not to do so. The former households are likely to be wealthier, better educated, and more conscious of hygiene than their neighbors, so it would not be surprising if they were also more likely do many other things that protect their families from feco-oral disease. The more sophisticated studies have used multivariate models to control for confounding, but where relative risks are low and the exposure groups are self-selected, even those models do not guarantee that confounding is eliminated (Cairncross 1990).

A further difficulty arises from the fact that cases of feco-oral disease in a given community cannot be considered independent events, because such diseases are infectious. The sample size, it can be argued, is the number of such villages rather than the number of individuals enrolled in the study. Yet a number of important studies in the literature compare a single intervention area with only one control area.

Other epidemiological weaknesses exist in the data. Blum and Feachem (1983) reviewed 50 studies of the health effect of water supply and sanitation projects and noted that every one contained one or more of these basic errors of methodology. A further weakness in the evidence for the effect of water supply on diarrheal disease burden is that most of it relates to diarrheal disease morbidity, and significant assumptions are needed to extrapolate such evidence to an effect on diarrheal mortality.

Effect on Diarrheal Disease

Esrey and Habicht (1985) and Esrey and others (1991) reviewed the same literature from a different perspective. Though conscious of the methodological shortcomings of most studies, they sought to assess the overall reductions in diarrheal disease that water supply could be expected to cause. They applied a number of criteria of epidemiological rigor and took the median reduction in morbidity reported from each type of intervention. Their conclusions are summarized in table 41.3.

Table 41.3

Median Reductions in Diarrhea Morbidity Reported from Different Water Supply and Sanitation Interventions.

For more than a decade, this review has remained the most authoritative on the subject. However, the small reductions in disease that it reports for water supply conceal an important heterogeneity. Though these overall results are frequently quoted, the following remark by Esrey and others (1991, 613) has usually been overlooked:

In the studies reporting a health benefit, the water supply was piped into or near the home, whereas in those studies reporting no benefit, the improved water supplies were protected wells, tubewells, and standpipes.

In the studies in the two reviews by Esrey and Habicht (1985) and Esrey and others (1991) in which the water supply was provided in the home, the median reduction in diarrheal disease is 49 percent (from 12 studies), and the reduction from the two better studies is 63 percent. Those reductions are several times greater than the overall median impacts in table 41.3. The 63 percent figure will be used in the burden of disease calculations that follow. In the two better studies, the members of the comparison group were using not an unimproved water supply, but a protected water source away from the home. The reductions they found are, therefore, in addition to those resulting from a public standpost level of service.

Some subsequent studies have confirmed this pattern. For example, Bukenya and Nwokolo (1991) showed in Papua New Guinea that use of a household tap was associated with 56 percent less diarrhea than use of public standposts providing water of good quality.

Conditions for Health Effect

Providing a public water point appears to have little effect on health, even where the water provided is of good quality and replaces a traditional source that was heavily contaminated with fecal material. By contrast, moving the same tap from the street corner to the yard produces a substantial reduction in diarrheal morbidity. How is this pattern to be understood?

The first step to an explanation is an understanding that most endemic diarrheal disease is transmitted by water-washed routes and is not waterborne. Although waterborne epidemics of diarrheal diseases such as cholera and typhoid have been notorious in the history of public health, the endemic pattern of transmission seems to be different, particularly in poor communities. Five types of evidence support this view:

• Negative health impact studies. As mentioned earlier, Esrey and Habicht (1985) and Esrey and others (1991) cite a number of studies of the health impact of water supplies in which water quality improvements have failed to have a significant effect on diarrheal disease incidence.

• Food microbiology. Studies of the microbiology of foods in developing countries—particularly the weaning foods fed to children in the age group most susceptible to diarrheal disease—have shown such food to be far more heavily contaminated with fecal bacteria than is drinking water (Lanata 2003), even when the water has been stored in open pots.

• Seasonality of diarrhea. In countries with a seasonal variation in temperature, bacterial diarrheas peak in the warmer season, whereas viral diarrheas peak in the winter. This pattern suggests that the bacterial pathogens show environmental regrowth at some stage in their transmission route, which means that they must have a nutritional substrate. Water is, thus, a less likely vehicle than food.

• Fly-control studies. Trials in rural Asia and Africa have shown that fly control can reduce diarrheal disease incidence by 23 percent (Chavasse and others 1999).

• Hand-washing studies. A recent systematic review of the effect of hand washing with soap has shown that this simple measure is associated with a reduction of 43 percent in diarrheal disease and 48 percent in diarrheas with the more life-threatening etiologies (Curtis and Cairncross 2003).

Those five types of evidence suggest that domestic hygiene—particularly food and hand hygiene—is the principal determinant of endemic diarrheal disease rates and not drinking water quality.

The second step is an understanding of how the level of service and convenience of a water supply influence such hygiene practices in the home. Taking the amount of water used per capita as an indicator of hygiene changes, other things being equal, one finds that providing a source of water closer to the home—and therefore more convenient to use—has very little effect on water consumption unless the old source was more than 1 kilometer (30 minutes' roundtrip journey) away from the user's dwelling (Feachem and others 1978).

However, water consumption doubles or triples when house connections are provided (White, Bradley, and White 1972), and reason exists to believe that much of the additional consumption is used for hygiene purposes. For example, Curtis and others (1995) found that provision of a yard tap nearly doubled the odds of a mother washing her hands after cleaning her child's anus and more than doubled the odds that she would wash any fecally soiled linen immediately.

In conclusion, water supplies are likely to have an effect on diarrheal disease when they lead to hygiene behavior change—that is, when the old source of water was more than 30 minutes' roundtrip away or when house connections are provided.

By a happy coincidence, then, the health benefits of water supply are most likely to be realized in exactly those cases in which the time-saving benefit is greatest—when the old source of water is farthest away, and when the new one is on the plot of the individual household. Though water supplies offering house connections are more expensive, the additional time savings offered by this level of service mean that people are willing to pay more for them. Moreover, collecting revenue from households with private connections is far simpler than collecting it from public taps because the sanction of disconnection can be used against households that default on payment of the tariff.

Calculating the burden of disease associated with inadequate water supply requires a figure for the reduction associated with the levels of service for which coverage statistics are available. The following burden of disease calculations are based on a reduction of 17 percent from an improved public water supply (table 41.3) and of a further 63 percent from house connections.

The effect of water supply improvements (and of hygiene practices such as hand washing) on diarrhea mortality can be expected to be at least as great as—and probably greater than—their effect on morbidity for several reasons. A theoretical argument for this improvement pattern is given by Esrey, Feachem, and Hughes (1985) in terms of infectious doses. Esrey and others (1991) also reported a median reduction of 65 percent in diarrhea mortality attributable to water supply, sanitation, or both in three studies, compared with 22 percent from 49 studies of morbidity. The effect of hand washing on life-threatening diarrheas—shigellosis, typhoid, cholera, and hospitalized cases—is greater than that on diarrhea morbidity as a whole (Curtis and Cairncross 2003). Finally, the two known direct studies in the literature of the effect of house connections on diarrhea mortality ("Serviço Especial da Saúde Pública," an unpublished study in Palmares, Pernambuco, Brazil, cited by Wagner and Lanoix 1959; Victora and others 1988) found reductions of 65 percent (relative to a public standpost) and 80 percent (relative to various communal sources, some polluted), respectively.

Effect on Other Disease Categories

Water supplies have a beneficial effect on a number of disease groups other than diarrhea, although the corresponding burden of disease is far less. The median reductions in morbidity from other water-related conditions, reported by Esrey and others (1990), are shown in table 41.4.

Table 41.4

Median Reductions in Morbidity Associated with Improved Water Supply and Sanitation: Conditions Other Than Diarrhea, Related Most Closely to Water Supply.

To be effective in controlling schistosomiasis, the water supply must be so convenient as to discourage water contact for laundry and bathing. It is unlikely that this level of convenience can be achieved without house connections.

Evidence suggests that water availability and hygiene can produce substantial reductions in trachoma (Emerson and others 2000). Because the reductions come from hygiene improvements such as hand and face washing, they are also likely to be greatest with house connections. Dracunculiasis is affected by water quality, but the simplest improved water supply is adequate to prevent transmission.

Conflicting evidence exists about whether water supply or improved water-washed hygiene affects the transmission of intestinal helminths. On one hand, Henry (1981) found in an intervention study in St. Lucia that piped water supplies were associated with a 30 percent reduction in ascariasis among children under age three over a two-year period. On the other hand, Han and others (1988) showed in Burma that an intervention to promote hand washing with soap had no effect on prevalence or intensity of infection with Ascaris spp. However, the potential contribution of water supply to reducing the burden of disease through its effect on these other infections is relatively minor when compared with its effect on diarrheal disease.

Willingness to Pay

The governments of developing countries cannot afford to provide heavily subsidized sanitation to all—or even to the majority—of their populations. The 2.6 billion people in Africa, Asia, and Latin America who do have adequate sanitation—53 percent of the population of those regions—have paid most of the cost themselves. Even those of the urban poor who do not have sanitation have expressed a willingness to pay for its full cost—or at least the local cost (excluding major interceptor sewers and treatment works, if required)—in a number of surveys, as long as credit is available on reasonable terms to smooth the cash flow (Altaf 1994). With regard to the rural poor, the success of well-conceived sanitation promotion programs in achieving coverage close to 100 percent, without a substantial subsidy, in some of the poorest rural communities in the world (Allan 2003) shows that people are willing to pay for sanitation if a suitable product is offered to them on suitable terms.

Why then do 2.4 billion people still lack sanitation? Several factors constrain the expression of the existing demand.

The constraint most frequently mentioned by unserved householders is cost, but this factor is usually more a perceived constraint than an objective one, for several reasons. First, many households are unaware of the true cost of latrines in their area, or the lower-cost models are not offered because local suppliers and artisans do not know about them or are attracted by the greater margins to be made on the more expensive technologies. Second, the high cost of capital to the poor rules out their borrowing the cost of a latrine, which to them would be a substantial investment. Third, they may be wary of investing in a property that belongs to their landlord, lest it be used as an excuse for a rent increase or even eviction. They may also feel, with some reason, that it is for the landlord to make the investment, rather than themselves, and they may be waiting for the landlord to do so. This belief has a similar effect to the common misapprehension of citizens, often encouraged by politicians, that the local government is responsible for sanitation and will eventually come to their aid; in either case, the outcome is inaction.

Other constraints include lack of ready access to necessary techniques and skills or to specific building materials and components. Where the skills exist locally, residents may lack confidence in the quality of work and value for money offered by the local artisans, or they may not know how to contact the right artisans. In many urban areas, local building regulations make low-cost sanitation technologies illegal.

Those constraints are compounded by the fragmentation of governmental responsibility for sanitation. Often it is devolved to local governments with little capacity to implement sanitation improvements. At the national level, one ministry may be responsible for sewerage and another for low-cost technologies; one may be responsible for construction, another for promotion, and a third for enforcing building codes and planning regulations.

Policy Implications

There are important externalities to households' investment in sanitation. Households are protected from their own feces by their sanitation facilities, but so, too, are their neighbors, and this factor is probably more important in epidemiological terms. If households are not fully aware of the health benefit—or if much of it accrues to others—a case exists for public intervention to increase coverage because these externalities exist.

This public intervention need not be in the form of subsidy. Strong arguments can be marshaled against a subsidy for low-cost sanitation (Cairncross 2003a). Subsidy limits the number of facilities that are built to the size of the subsidy budget; it encourages the design and marketing of unaffordable sanitation systems; it frequently leads to capture by the better-off, who install expensive toilets while the poor go without; and it distorts the market, diverting the efforts of latrine builders who would otherwise be seeking to meet the needs of low-income groups.

The intervention can be by regulation. National and local governments have substantial regulatory powers that can be used to increase sanitation coverage without significantly increasing costs or public expenditure. For example, more than 90 percent of households in the town of Bobo Dioulasso, Burkina Faso, have their own latrine (Traoré and others 1994) as a direct result of the local administration's practice in the past of withdrawing rights of land tenure from owners who did not build a latrine on their plot within a specified time. Another regulatory intervention is to enforce the obligation of landlords to provide sanitation for their tenants.

An alternative strategy is to provide support to the marketing of sanitation. This strategy can be undertaken in a number of ways that are not feasible for the existing producers, mainly artisan builders and small component manufacturing workshops. Those interventions would aim principally at overcoming the constraints to the expression of effective demand for sanitation and could include the following:

• advertising and other forms of promotion

• facilitation of building regulation approval

• brokerage to put potential purchasers in touch with providers

• quality assurance and guarantee schemes

• training in low-cost construction techniques and in marketing

• centralized production of essential components

• provision of pit emptying and desludging services.

Promotion of improved hygiene practices, including appropriate use and maintenance of the sanitation facilities, is another possible intervention by the public sector. All of those measures will help increase sanitation coverage and health benefits and are appropriate interventions for the health sector. The costs of several of them are recoverable (after an initial launch period) as fees, so that public intervention need not require public expenditure.

Costs of Promotion

The costs of promotion and administration found in two government-run rural sanitation programs documented by the World Bank were US$16.80 (Zimbabwe) and $20.00 (the Philippines) per latrine, respectively (Cairncross 1992). Because these costs are largely fixed, the cost per unit falls as the number of units built increases. Unit costs will therefore be high in relatively unsuccessful programs. Successful programs, on the other hand, often engender the construction of more latrines than they can account for, which also gives an upward bias to the promotional costs per unit built. For example, for every latrine built by Lesotho's rural sanitation program in the late 1980s, four others were built independently but as a result of its promotional activities.

More recently, successful sanitation programs managed by nongovernmental organizations (NGOs) have documented slightly lower unit costs for promotion. For example, the Zimbabwean NGO AHEAD (Applied Health Education and Development), working through district-level health staff and a network of community health clubs, achieved the construction of 3,400 latrines in Makoni district within two years at a total promotional cost of US$45,660, or US$13.43 per unit, equivalent to US$2.24 per household member served (Waterkeyn 2003). In Bangladesh, WaterAid and its partner, a local NGO named VERC (Village Education Resource Centre), have developed an approach that has successfully achieved 100 percent sanitation coverage and the elimination of open defecation in more than 100 villages in six districts at a cost of US$8 per household, or US$1.50 per capita (Allan 2003). Both programs also promoted domestic hygiene practices in addition to the construction and use of latrines. In Bangladesh, all (and in Zimbabwe, most) of the costs of latrine construction were paid by the population themselves.

The programs in Bangladesh and Zimbabwe were particularly successful and well managed. The promotion cost is taken as US$2.50 per capita for cost-effectiveness calculations, which is slightly above the higher of the two, to allow for the imperfections of sanitation programs in the real world.

Diarrheal Disease

The effect of sanitation on diarrhea morbidity has already been mentioned. Table 41.3 shows the results of Esrey and others' (1991) review, attributing a median reduction in incidence of 36 percent to sanitation. Although this figure is the median of the five "better" studies, it must be interpreted with great care because almost all the known studies on the health effects of sanitation are observational studies that use self-selected exposure groups. Confounding by a sense of hygiene is likely to be a significant problem in any such study. From Brazil to Bangladesh, the owners of latrines have been observed to behave more hygienically than their neighbors in practices such as hand washing that are not affected by the presence of a latrine (Hoque and others 1995—see table 41.6; Strina and others 2003). It is thus impossible to prove, except by an intervention study, that any health benefit associated with latrine ownership is due to the latrine and not to the hygiene habits of latrine owners.

Table 41.6

Factors Associated with Hand-Washing Behavior by 90 Women in Bangladesh.

The overall reduction in diarrhea from sanitation quoted by Esrey and others (1991) likely disguises considerable heterogeneity in terms of the context rather than the type of sanitation technology. For example, sanitation is likely to have a greater effect on diarrheal disease in high-density urban areas, where open defecation leads to gross fecal pollution of the neighborhood, and less effect in rural communities, where all but the youngest children use communal defecation sites some distance away from their homes.

For example, Moraes and others (2003), working in urban favelas in northeast Brazil, found that diarrhea incidence among children in households with a toilet was half that in households that did not have one. This comparison is likely to be affected by confounding because the households with toilets were a self-selected group. Comparison between communities is less likely to be affected by confounding, but Moraes and others found a greater reduction. The mean incidence of diarrhea in young children in communities with sewers was only one-third of that in the communities that, for administrative and technical reasons, did not have sanitary drainage.

Thus, although the quality of the studies reviewed by Esrey and others (1991) was in general poor and the range of reductions wide, little doubt exists that excreta disposal can be associated with significant reductions in diarrhea morbidity. Studies showing that proximity to open or overflowing sewers (Moraes and others 2003), failure to dispose hygienically of children's stools (Traoré and others 1994), or the presence of excreta on the ground in the household compound (Bukenya and Nwokolo 1991) is a risk factor for fecal-oral infections provide supporting evidence for the likely effect of sanitation infrastructure, particularly in urban settings, on diarrheal disease transmission.

In conclusion, there are some reasons, such as the likelihood of confounding, to believe that Esrey and others' (1991) median reduction is an overestimate, but reasons exist also to believe that the reductions measured were not as great as they might have been had the provision of sanitation been accompanied by hygiene promotion to ensure that the facilities were fully and appropriately used (especially by young children) and maintained. A systematic review of the effect of sanitation on diarrheal disease is urgently required. Meanwhile, and on balance, Esrey and others' median reduction of 36 percent in diarrhea incidence is the most authoritative estimate available.

Interaction with Water Supply

The results of Esrey and others' (1991) review suggest that the effect of water supply and sanitation combined is no greater than that of either on its own. However, that conclusion is based on only two studies, and the percentage reductions found in the individual studies of each type of intervention exhibit a wide range. Reflection on how in practice each of the two interventions interrupts the transmission of fecal-oral pathogens would suggest that their effects would be largely independent: whereas water supply helps prevent contamination of drinking water, hands, and food, excreta disposal helps prevent contamination of the household yard and surroundings, including children's play areas. Esrey and others (1990) reported three other studies in which sanitation and water supply had a greater effect together than individually, but the reductions in diarrhea incidence in those studies could not be calculated.

For the purpose of burden of disease calculations, therefore, the effects of water supply and sanitation improvements on diarrhea are considered here to be independent and additive, which has the advantage of simplicity.

Effect on Other Disease Categories

The first evidence for the health benefits of excreta disposal related not to its effect on diarrheal disease but on intestinal helminths.

A prolonged series of in-depth studies from 1920 to 1930 by researchers of the Rockefeller Foundation established beyond doubt that promiscuous defecation, especially in the household surroundings and particularly by children, played a major role in the transmission of Ascaris spp., Trichuris spp., and hookworms in a range of settings from Panama to China and the southeastern United States. By implication, the use of sanitary toilets should interrupt transmission by that route.

However, more recent attempts to measure the reductions in parasite prevalence or intensity attributable to improved sanitation have often suffered from the same shortcomings as the studies of their impact on diarrheal disease; many have been cross-sectional studies and, therefore, subject to confounding.

Esrey and others (1991), in reviewing this literature, found that water supply and sanitation reduced the prevalence of ascariasis by a median of 28 percent (range 0 to 83 percent) and of hookworm infection by 4 percent (0 to 100 percent). Those reductions are likely caused by the sanitation rather than by the water-supply improvements. Indeed, three of the nine positive studies of ascariasis and three of the five positive studies of hookworm involved sanitation alone. It is also likely that the effect of excreta disposal on Trichuris infection is similar to that on ascariasis (Henry 1981).

Much emphasis has been placed in recent years on chemotherapy as a control intervention for intestinal helminths, particularly the chemotherapy of schoolchildren. However, that option is not always sustainable because the children are quickly reinfected by the eggs and larvae that remain in the environment. Sanitation, particularly school sanitation, has been adopted by the major international donor agencies as an integral component of the FRESH (Focusing Resources on Effective School Health) framework to ensure its sustainability.

A study in Bangladesh (Mascie-Taylor and others 1999) suggested that chemotherapy was more cost-effective (though less effective) as a helminth control intervention than a health education program that included the promotion of sanitation. However, the health education program was excessively labor intensive and, therefore, expensive; it involved the constant deployment of six health educators and a supervisor in each study area of only 550 households, resulting in a cost of Tk 1600 (US$30) per household, compared with Tk 330 (US$6) per year for chemotherapy. That cost compares with the total cost of US$8 per family for WaterAid's successful "100 percent sanitation" approach in rural Bangladesh (Allan 2003). Whereas the promotion of sanitation is a one-time cost, the cost of chemotherapy is a recurrent annual expenditure. Allowing for such a sanitation promotion initiative once every five years—and using the chemotherapy costing of Mascie-Taylor and others (1999)—sanitation promotion is more cost-effective against helminths in Bangladesh than is chemotherapy. If the cost were apportioned between the effect on diarrheal disease and the effect on helminths, sanitation would be far more cost-effective than chemotherapy.

Sanitation can also help prevent trachoma. More than 70 percent of the incidence of this infection has been shown to be caused by flies, mainly of the species Musca sorbens, which breeds preferentially in scattered human feces. Pit latrines have been shown to reduce the population of these flies by depriving them of their breeding sites (Emerson and others 2004).