A novel approach to estimate Maternal Mortality Rate and its determinants by statistical modelling

摘要

Maternal Mortality Rate (MMR) is a key indicator of international development, and its reduction has been a challenge in many developing countries, despite existence of effective interventions. The United Nations Millennium Development Goals (MDG 5) called for a three-quarters reduction in MMR between 1990 and 2015. In current literature and practice, population-based estimates of distinctive reasons contributing to MMR are principally lacking, although some of the studies have attributed to socioeconomic determinants. The present study explores to address these gaps through an innovative model that uses clinical, hospital utilization, and lab and referrals data to predict MMR. The application was proposed in the state of Tamil Nadu in India covering 32 districts covering a population of approximately 72.1 million. The study presents the data collection and consolidation methods, data quality management, sampling methods and data transformations required for building the model. The choice, validation and accuracy of the model across sample sets is established vividly. Significant parameters contributing to MMR across different determinants are given, which will assist in quantified target intervention e.g. Number of antenatal mothers with hemoglobin less than 7 grams explains maternal mortality well. The paper concludes by enumerating the best practices in prediction of MMR and future directions.