In all sites, standard methods were adapted to measure the MCE indicators and to collect the data needed to support cost-effectiveness analysis. Depending on the evaluation at each site, data were collected at household, community, district and national levels.

The tools developed for the MCE may be useful to others interested in evaluating child health programmes. These generic tools are available from the Department of Child and Adolescent Health and Development.

An important characteristic of the MCE is the use of a stepwise design. The number and nature of steps were tailored to each site, but in general the following steps were included:

• The strength of IMCI implementation (training coverage, geographical spread, etc.) and health-system characteristics (accessibility, utilization rates, etc.) were assessed at national level. 

• If necessary, a baseline survey of mortality, nutrition and coverage of key interventions was carried out.

• Health-facility surveys were performed, to compare the quality of care provided by IMCI-trained and other health workers. Information on health facility costs was also collected. 

• Information were obtained on health systems support for IMCI (drugs, supervision, etc.) and on the coverage of community and household interventions – either from specially designed household surveys or from secondary sources. Costs at national, district and household level were also ascertained. 

• Using results from the previous steps, a decision was made about whether IMCI implementation is sufficiently strong that an impact on mortality and nutrition is likely. 

• Implementation, including costs, continued to be monitored for two or more years, to allow sufficient time for an impact on mortality to be measurable. 

• A household survey investigated whether coverage with key child-survival interventions was adequate, and whether child nutritional status improved. Costs at the household level were measured again.

• Changes in mortality were assessed by a demographic survey or from results of mortality surveillance.

As of late 2005, the evaluation ended in Tanzania, where all eight of the above steps were completed. In Brazil, Peru and Uganda, the evaluation was stopped at step 5 because there was evidence that IMCI implementation in the whole study area was not sufficient to reach a measurable impact on child survival and nutrition. Nevertheless, in Brazil and Peru there were smaller areas with stronger implementation, and exploratory impact analyses were carried out using secondary data (see publications). In Bangladesh, the study is under way, currently in phase 6.

More information on the design of the MCE, the sampling methods, and the impact model used to define the key indicators is available in the "MCE Design" page.