Abstract
Technical interventions for maternal healthcare are implemented through a dynamic social process. Peoples' behaviours--whether they be planners, managers, providers, or potential users--influence the outcomes. Given the complexity and unpredictability inherent in such dynamic processes, the proposed cause-and-effect relationships in any one context cannot be directly transferred to another. While this is true of all health services, its importance is magnified in maternal healthcare because of the need to involve multiple levels of the health system, multiple types of care providers from the highly skilled specialist to community-level volunteers, and multiple technical interventions, without the ability to measure significant change in the outcome, the maternal mortality ratio. Patterns can be followed however, in terms of outcomes in response to interventions. From these case studies of implementation of maternal health programmes across five states of India, Pakistan, and Bangladesh, some patterns stand out and seem to apply virtually everywhere (e.g., failure of systems to post staff in difficult areas) while others require more data to understand the observed patterns (e.g., response to financial incentives for improving maternal health systems; instituting available accessible safe blood). The patterns formed can provide guidance to programme managers as to what aspects of the process to track and micro-manage, to policy-makers as to what features of a context may particularly influence impacts of alternative maternal health strategies, and to governments more broadly as to the factors shaping dynamic responses that might themselves warrant intervention.