Tanzania has high underlying mortality levels (under-five mortality rates 130-150 per thousand) and the five IMCI target diseases account for around 70% of under-five deaths. Deaths from malaria are very common, accounting for around 50% of under-five deaths. IMCI implementation began in 1997 in two pioneer districts (Morogoro Rural and Rufiji) with support from the Canadian-funded TEHIP (Tanzania Essential Health Interventions Project). The existence of two similar, contiguous comparison districts where IMCI implementation began in 2002 (Kilombero and Ulanga) provided a special opportunity for a plausibility design approach. Indicators of process and impact in the two intervention districts were collected from 1999 to 2002 and allowed a comparison between these districts with the control districts.

Kilombero, Ulanga and Morogoro Rural Districts lie in Morogoro Region in south-western Tanzania, 200–400km from Dar-es-Salaam. Rufiji District is south of Dar-es-Salaam in Coast Region, bordering Kilombero district.

Tanzania MCE is unique in that implementation is very strong, and all four districts have areas that have been under continuous mortality surveillance for several years; this permitted a study of the impact of IMCI on child mortality. In addition, TEHIP included a strong research component in its two target districts, covering a detailed description of changes in provision and utilization of health services, and behavioural changes at population level.

The specific objectives of the Tanzania MCE were:

• To measure the impact of the IMCI strategy on under-five mortality;

• To assess the effect of the IMCI strategy on child health indicators at household level;

• To assess the effect of the IMCI strategy on child health care at health facility level;

• To document the implementation of IMCI in the two intervention districts;

• To describe other relevant activities in all four districts under study, particularly programmes and activities not involving IMCI;

• To estimate the economic cost of implementing and maintaining IMCI from a societal perspective (i.e., cost to the government provider as well as to the client will be estimated).

The evaluation included six related but distinct operational sub-studies.

Sub-study 1. Continuous demographic surveillance of under-five mortality. 
Mortality data were continuously generated in the four districts, as part of three collaborating projects: Ifakara Centre Demographic Surveillance system, AMMP (Adult Morbidity and Mortality Project), and TEHIP (Tanzania Essential Health Interventions Project). The surveillance systems are further described in the Mortality Measurement section of the Resource Book.

Sub-study 2. "IMCI Phase-in" and "IMCI Follow-up" household surveys. 
A cross-sectional survey in July/August 1999 in the four districts concerned, at the end of the phase-in of IMCI implementation, resulted in the selection of 30 clusters of 20 households in each district except Kilombero, where 35 were chosen so as to over-sample the smaller rural population. About 500 under-five children were examined in each district. Information was obtained on levels of morbidity and malnutrition (including anthropometry and haemoglobin measurement), coverage of preventive interventions, and care-seeking patterns. The survey covered client costs associated with care-seeking; distance from each cluster to the nearest health facility was assessed with a Global Positioning System. For nearly all variables studied, intervention and control districts were remarkably similar; insecticide-treated nets were commoner in non-IMCI districts. A follow-up survey in July/August 2002 gave an estimate of the effect of IMCI implementation on process and impact indicators.

Sub-study 3. Health facility survey.  
As IMCI implementation in the health facilities of the two IMCI districts was already advanced, a baseline survey was not possible. A cross-sectional comparison was carried out, therefore, in August 2000 to compare the characteristics of health facilities, health workers’ characteristics and performance, and related provider costs in 15–20 facilities in each of the four districts. Case management was assessed by observation, re-examination, and exit interviews of about six children in each selected facility. Children in IMCI districts were found to be receiving better care than those in comparison districts. Data were also collected on availability of drugs, supplies and services for child health, availability and accessibility of child health care (opening days/hours, outreach clinics, etc.), and human and physical resources.

Sub-study 4. Documentation of IMCI implementation.  
In the two intervention districts, TEHIP research data have been used for a detailed description of IMCI implementation activities. TEHIP documented the evolution of IMCI and its impact in four research programmes:

1. Health Systems Research: context, content and stakeholders of district health plans and budgets; resource allocation and control; implementation; provider compliance and satisfaction; and coverage.

2. Health Behaviour Research: longitudinal quantitative and qualitative data on health-seeking behaviour, with special reference to IMCI as a tracer intervention; these show trends in IMCI utilization from the household perspective, and monitor perceptions of child illnesses and IMCI interventions, household decision-making and expenditure; user compliance and satisfaction; perceived quality; social mapping; access; and equity.

3. Health Impact Research: monitoring of trends in IMCI-preventable mortality burden, through demographic surveillance (Sub-study 1 above); information is also generated on health-seeking behaviour during terminal illness.

4. District Planning Tools: including a cost tracking system for determining facility level costs per case managed for all essential health interventions, including IMCI.

Sub-study 5. Description of other relevant activities in all four districts.
As in all studies using a plausibility design in which the unit of analysis is the district, IMCI and non-IMCI areas require special attention to explain differences arising from contextual factors. Information has been collected on activities and indicators related to child health, with emphasis on intervention programmes carried out by international and nongovernmental institutions.

Sub-study 6. Estimation of the economic costs of IMCI. 
This sub-study addressed costs from a societal perspective, by estimating client and government provider costs associated with child health care at district, facility and household levels in the four districts concerned. Data were collected through specific data-collection instruments in sub-studies 2, 3 and 5. A full report is available: in summary, there is no evidence that IMCI has been associated with higher costs at the time of the study.

A considerable effort was made to establish supportive partnerships for this evaluation. Several meetings of stakeholders have included representatives of local research institutions, the Ministry of Health, WHO and UNICEF, District representatives, and national IMCI stakeholders, to seek initial input, and to review and develop the draft design and results as they become available.

The MCE findings showed that IMCI was associated with improved quality of care in health facilities at no additional cost. Underfive mortality levels were 13% lower in the IMCI districts at the end of the two years of the evaluation, and there was also a significant reduction in stunting. The Tanzania MCE results were fed back to local policy makers and received wide attention. IMCI is now being expanded to all districts in the country. 

The reports, journal publications and policy briefs coming out of the Tanzania MCE are available under publications.