Improvement of public health is the strongest and most frequent argument put forward in support of water and sanitation projects (1). Inadequate water and sanitation services adversely affect the health and socioeconomic development of communities, and the two outcomes are related (2, 3). An integrated package of activities aimed at improving water supply and sanitation facilities, providing appropriate hygiene education, and building local capacity for the management of water and sanitation resources is more likely to show significant health benefits than a programme that concentrates on one area alone (2, 4-7). Expected health benefits rely on water and sanitation programmes containing a combination of components most appropriate to the local context. During 1997-2001, the German Technical Cooperation Agency (KFW) contracted and funded the Water and Sanitation Extension Programme 0VASEP) of the Aga Khan Development Network (AKDN) to undertake a project in selected villages in Northern Areas and Chitral in northern Pakistan. The aim was to improve potable water supply at village and household levels, sanitation facilities and their use, and awareness and practices about hygiene behaviour. The health and hygiene education topics addressed included traditional concepts about diseases; promotion of latrine use and the safe disposal of faeces; domestic, environmental, and personal hygiene; food preparation, handling, and storage; transmission routes and prevention of waterborne diseases; and operation and maintenance of water sources.
The Northern Areas and Chitral is a mainly rural, mountainous region approximately 72 496 [km.sup.2]; the population of 900 000 people live in villages that typically comprise 50-200 households. Villages differ by location in topography and geography, as well as by language, ethnicity, and socioeconomic development. The main health facilities are run by the government and the Aga Khan Health Services, Pakistan (AKHS,P) of AKDN. Initially, WASEP selected 109 villages (population 100 000) based on pre-set criteria. At the end of 2001, 99 villages (population 88 000) were enrolled in the project: 68 from Northern Areas and 31 from Chitral. Intervention components delivered by WASEP using appropriate technologies in an integrated manner were water supply, water quality, drainage, sanitation, and school-and community-based hygiene education.
The planning and implementation methodologies have been described elsewhere (8). Implementation schedules were flexible, so the programme could respond to the characteristics and dynamics within each village. The Department of Community Health Sciences, Aga Khan University (AKU-CHS), also of AKDN, helped WASEP in 1998 to develop a planning and monitoring approach for the project. In 2000, AKU-CHS was contracted by WASEP to determine whether, using diarrhoea as the health outcome, the project influenced the health status of villages served by the project. Other institutions conducted evaluations on different aspects of the project, such as hardware performance related to water supply and water quality (9).
In the planning stage, WASEP specifically identified a reduction in incidence of diarrhoea as a project health outcome. The AKUCHS team focused the evaluation on children, as this age group suffers the greatest health burden from diarrhoea (10-12). In children, malnutrition, lack of safe water, poor sanitation, poor hygiene, early motherhood, breastfeeding practices, and inadequate health care are among known risk factors for diarrhoea (13, 14). A case-control design was adopted (15-17) to determine whether, after controlling for selected confounders, children aged <6 years with diarrhoea were more or less likely to reside in villages that participated in the project than in villages that did not participate. This approach to the evaluation treated the combined "package" of interventions as the main exposure variable--because implementation schedules and sequencing for component interventions were flexible it was difficult to assess the contribution of each component independently. …