Background: Understanding of the impacts of climatic variability on human health remains poor despite a possibly increasing burden of vector-borne diseases under global warming. Numerous socioeconomic variables make such studies challenging during the modern period while studies of climate-disease relationships in historical times are constrained by a lack of long datasets. Previous studies have identified the occurrence of malaria vectors, and their dependence on climate variables, during historical times in northern Europe. Yet, malaria in Sweden in relation to climate variables is understudied and relationships have never been rigorously statistically established. This study seeks to examine the relationship between malaria and climate fluctuations, and to characterise the spatio-temporal variations at parish level during severe malaria years in Sweden 1749-1859.

Methods: Symptom-based annual malaria case/death data were obtained from nationwide parish records and military hospital records in Stockholm. Pearson (r(p)) and Spearman's rank (r(s)) correlation analyses were conducted to evaluate inter-annual relationship between malaria data and long meteorological series. The climate response to larger malaria events was further explored by Superposed Epoch Analysis, and through Geographic Information Systems analysis to map spatial variations of malaria deaths.

Results: The number of malaria deaths showed the most significant positive relationship with warm-season temperature of the preceding year. The strongest correlation was found between malaria deaths and the mean temperature of the preceding June-August (r(s) = 0.57, p < 0.01) during the 1756-1820 period. Only non-linear patterns can be found in response to precipitation variations. Most malaria hot-spots, during severe malaria years, concentrated in areas around big inland lakes and southern-most Sweden.

Conclusions: Unusually warm and/or dry summers appear to have contributed to malaria epidemics due to both indoor winter transmission and the evidenced long incubation and relapse time of P. vivax, but the results also highlight the difficulties in modelling climate-malaria associations. The inter-annual spatial variation of malaria hot-spots further shows that malaria outbreaks were more pronounced in the southern-most region of Sweden in the first half of the nineteenth century compared to the second half of the eighteenth century.