Disentangling the influence of temperature and season on weekly birth rates in Vienna, Austria

Abstract

Background: Before examining other environmental predictors of weekly birth- and conception rates, we set out to investigate seasonal and meteorological predictors in a time series of births given by Viennese women between 1999 and 2019. Methods: We extracted week of birth and duration of gestation in weeks to calculate number of conceptions and of births per week. For both outcomes (birth and conception numbers) we first applied negative binomial regression models. But, when we saw that over-dispersion was negligible, we reverted to Poisson regression models. As predictors, we included a linear term, sine-cosine functions with a wave-length of one year, and another one with a wavelength the average duration of a lunar cycle. In addition, we included temperature as a linear and a quadratic term. Results: On average 341.4 births and 344.1 conceptions occurred per week. The number of births and conceptions increased over the years and within each year seasonal variation was visible. Birth numbers according to the annual sine-cosine model were lowest in January and February and highest in July and August. The number of conceptions were lowest in May and highest in November. Adding another sine-cosine function for the lunar cycle did not improve the model fit much. Nevertheless, that parameter was still significant, with the highest birth numbers occurring roughly between the waxing crescent and the first quarter moon, and the highest conception numbers conceptions occurring at the full moon. Extremer temperatures (both hot and cold) decreased conception numbers and increased birth numbers. Conclusion: While the effects of temperature and the annual variation resembled our expectations, the lunar influence was unexpected. This latter finding should be treated with care, because it was not hypothesis driven. KEYWORDS: annual & lunar cycle, births, temperature, births, conceptions