Neonatal and perinatal mortality in the urban continuum: a geospatial analysis of the household survey, satellite imagery and travel time data in Tanzania

**Background:** Recent evidence suggests the traditional 'urban advantage' in child survival may be reversing in sub-Saharan Africa, with Tanzania showing urban neonatal mortality (40/1000 live births) twice that of rural areas (20/1000 live births). However, methodological challenges include misclassification of neonatal deaths and stillbirths, oversimplification of urban environments, and reliance on administrative definitions of urban/rural. This study aimed to address these limitations using satellite-derived urbanicity measures and examining both neonatal and perinatal mortality.

**Methods:** The study analyzed 8915 pregnancies (8739 live births, 176 stillbirths) among 6156 women from the 2015-2016 Tanzania Demographic and Health Survey (DHS). The coordinates of 527 DHS clusters were spatially overlaid with the 2015 Global Human Settlement Layer (GHS-SMOD) to create a three-category urbanicity measure: core urban (61 clusters, 11.6%), semi-urban (224 clusters, 42.5%), and rural (242 clusters, 45.9%). Travel time to the nearest public hospital was modeled using a least-cost path algorithm. Bivariate and multilevel multivariable logistic regression models were constructed, adjusting for confounders including geographical zone, household wealth, maternal education, age, anaemia, and pregnancy/birth characteristics. Four separate multivariable models were run due to variable availability.

**Key Results:** Both neonatal and perinatal mortality rates were highest in core urban areas. Neonatal mortality was 39.8/1000 live births in core urban, 24.8/1000 in semi-urban, and 21.9/1000 in rural areas (p=0.037). Perinatal mortality was 56.4/1000 pregnancies in core urban, 37.9/1000 in semi-urban, and 35.9/1000 in rural areas (p=0.028). Bivariate analysis showed significantly higher odds of neonatal death (OR=1.85; 95% CI 1.12-3.08) and perinatal death (OR=1.60; 95% CI 1.12-2.30) in core urban compared to rural clusters. In multivariable models, the direction and magnitude of association persisted but was no longer statistically significant: adjusted OR for neonatal death in core urban ranged from 1.26 (95% CI 0.57-2.79) to 2.36 (95% CI 0.69-8.04) across models, and for perinatal death aOR=1.71 (95% CI 0.89-3.27). Travel time to the nearest hospital was not associated with mortality. Notably, >95% of neonatal deaths in core urban occurred in the first week of life, compared to 19% in semi-urban and 14% in rural areas. The stillbirth-to-early-neonatal-death ratio was 0.52 in core urban, indicating substantial under-reporting or misclassification of stillbirths.

**Clinical Implications:** The study challenges the assumption of universal urban advantage in child survival and demonstrates that densely populated urban areas in Tanzania face disproportionately high neonatal and perinatal mortality. The findings suggest that urban populations are heterogeneous, and certain neighborhoods or subgroups may be disproportionately affected. Addressing urban mortality is critical for Tanzania to meet the Sustainable Development Goal target of neonatal mortality <16/1000 live births by 2030. The authors call for more granular primary data collection to disentangle the contributions of pregnancy factors, living conditions, and quality of care in urban birthing facilities, and recommend using satellite technologies for more accurate urbanicity classification in future research.