Impact of IMCI health worker training on routinely collected child health indicators in Northeast Brazil

The Integrated Management of Childhood Illness (IMCI) is a global strategy including improvements in case management at health facilities, strengthening health systems support and improving key family and community practices relevant to child health. In Brazil, IMCI was introduced in 1997, being largely restricted to training health workers in case management. IMCI training of doctors and nurses took place in many municipalities, but implementation of the other two components of IMCI was very limited. We analyze the impact of IMCI health worker training on infant mortality in three states in north-eastern Brazil, by comparing three groups of municipalities over the period 1999 to 2002: 23 with training coverage of 50% or greater, 216 with lower training coverage, and 204 without any IMCI training. Two sources of mortality data are used: vital registration of deaths and births, and the community health workers' (CHW) demographic surveillance system. The latter resulted in a larger number of deaths being reported and in more stable mortality rates over time than the former. Infant mortality rates (IMR) declined rapidly according to both sources of information, during the study period. After adjustment for confounding factors, there was no association between IMCI training coverage and infant mortality measured through either information system. According to the CHW data, the adjusted annual changes were of -7.2 deaths per 1000 births in the high IMCI training coverage group, -4.6 in the low IMCI training coverage and -5.0 in the no IMCI group (p=0.46). According to vital statistics, the corresponding average annual changes were -5.0, -4.2 and -2.8 deaths per 1000 births (p=0.16). The negative findings from the Brazil evaluation suggest that IMCI clinical training, in the absence of the other two components of IMCI, and in an area with infant mortality under 50 per 1000, is unlikely to lead to a measurable impact on mortality.