The association between cold extremes and neonatal mortality in Swedish Sápmi from 1800 to 1895

Background: Studies in which the association between temperature and neonatal mortality (deaths during the first 28 days of life) is tracked over extended periods that cover demographic, economic and epidemiological transitions are quite limited. From previous research about the demographic transition in Swedish Sápmi, we know that infant and child mortality was generally higher among the indigenous (Sami) population compared to non-indigenous populations. Objective: The aim of this study was to analyse the association between extreme temperatures and neonatal mortality among the Sami and non-Sami population in Swedish Sápmi (Lapland) during the nineteenth century. Methods: Data from the Demographic Data Base, Umeå University, were used to identify neonatal deaths. We used monthly mean temperature in Tornedalen and identified cold and warm month (5th and 95th) percentiles. Monthly death counts from extreme temperatures were modelled using negative binomial regression. We computed relative risks (RR) with 95% confidence intervals (CI), adjusting for time trends and seasonality. Results: Overall, the neonatal mortality rate was higher among Sami compared to non-Sami infants (62/1,000 vs 35/1,000 live births), although the differences between the two populations decreased after 1860. For the Sami population prior 1860, the results revealed a higher neonatal incidence rate during cold winter months (<-15.4°C, RR = 1.60, CI 1.14-2.23) compared to infants born during months of medium temperature. No association was found between extreme cold months and neonatal mortality for non-Sami populations. Warm months (+15.1°C) had no impact on Sami or non-Sami populations. Conclusions: This study revealed the role of environmental factors (temperature extremes) on infant health during the demographic transition where cold extremes mainly affected the Sami population. Ethnicity and living conditions contributed to differential weather vulnerability.