BACKGROUND: While information about 4 million neonatal deaths worldwide is limited, even less information is available for stillbirths (babies born dead in the last 12 weeks of pregnancy) and there are no published, systematic global estimates. We sought to identify available data and use these to estimate the rates and numbers of stillbirths for 190 countries for the year 2000, and provide uncertainty estimates. METHODS: We assessed three sources of stillbirth data according to specified inclusion criteria: vital registration; demographic and health surveys (DHS), based on a new analysis of contraceptive calendar data; and study reports that include published studies identified through systematic literature searches of more than 30,000 abstracts and unpublished studies. A random effects regression model was developed to predict national stillbirth rates and associated uncertainty intervals. FINDINGS: Data from 44 countries with vital registration (71,442 stillbirths), 30 DHS surveys from 16 countries (2989 stillbirths), and 249 study populations from 103 countries (93,023 stillbirths) met the inclusion criteria. Model-based estimates were used for 128 countries. For 62 countries, the observed values were adjusted by a correction factor derived from the model. The resultant stillbirth rates ranged from five per 1000 in rich countries to 32 per 1000 in south Asia and sub-Saharan Africa. The estimated number of global stillbirths is 3.2 million (uncertainty range 2.5-4.1 million). In light of the data limitations and the conservative approach taken, the real number might be higher than this. INTERPRETATION: The numbers of stillbirths are high and there is a dearth of usable data in countries and regions in which most stillbirths occur, with under-reporting being a major challenge. Although our estimates are probably underestimates, they represent a rigorous attempt to measure the numbers of babies dying during the last trimester of pregnancy. Improving stillbirth data is the first step towards making stillbirths count in public-health action.
