Assessing the mortality impact of an integrated health programme: lessons from Matlab Bangladesh.

The Matlab maternal-child health/family planning program in rural Bangladesh is an important and rare example of an integrated health program where the mortality impact has been carefully and conclusively evaluated. This chapter presents a chronology and description of the health interventions undertaken by this program. The key findings from the evaluation of the impact of the major health interventions are summarized and some of the key lessons learned about such an evaluation are presented. The Matlab study area was created to accommodate a cholera vaccine trial in 1963. Oral rehydration therapy was introduced in 1979 but failed to lower significantly diarrheal-related infant mortality rates in the intervention area. A maternal tetanus immunization program for pregnant women was started in 1978 and was modified in 1989 to expand coverage to all women. Evaluation of this intervention found that two maternal doses of tetanus vaccine reduced infant mortality by 42%. Measles vaccination began in 1982 and vaccination status lowered mortality risks by 46% providing the most benefit to the most vulnerable children. An acute respiratory infection program launched in 1988 reduced mortality by 32%. Family planning services were offered in late 1977 and led to a significant reduction in maternal mortality rates but not in maternal mortality ratios (the risks associated with pregnancy and childbearing). Formation of a more comprehensive maternal care program in 1987 led to significantly lower direct obstetric mortality ratios. The other health interventions of the Matlab program have yet to undergo systematic evaluation. The lessons learned about evaluation are that 1) there are conditions that serve as prerequisites for systematic assessments of an integrated program 2) a concurrent control area or population is vital 3) increasingly complex programs undermine the usefulness of a comparison area 4) phased introductions of services are the best way to isolate the unique effects of various interventions 5) the goals of rapidly reducing mortality and of rigorously evaluating the impact of interventions sometimes conflict and cause tension 6) evaluation opportunities is lost if a phased study design is abandoned 7) integrated programs eventually lose the ability to assess individual interventions and 8) cause-specific mortality measures can sometimes be useful for evaluation.