Not only cold but also heat: the effect of maternal exposure to high temperatures during gestation on neonatal mortality in pre-transitional Casalguidi, 1819–1859

The prevalence of neonatal mortality in pre-transitional demographic systems has been a subject of significant interest, particularly within the demographic history of Ancien Régime Italy (Bellettini, 1981, 1987; Del Panta, 1984), where poor living conditions, poor hygiene, and lack of medical assistance have contributed to creating conditions for extremely high levels of mortality within the first month of life. Among the risk factors, several studies have also highlighted the significant role played by weather conditions. Being born in months characterized by cold and harsh weather has been shown to significantly increase the risks of neonatal death, with respiratory diseases during the winter season proving to be fatal in pre-transitional populations (Breschi & Livi Bacci, 1986; Breschi et al., 2000, 2004; Dalla Zuanna & Rosina, 2011; Huck, 1997; Mooney, 1994; Van Poppel et al., 2018; Williams, 1992; Wrigley et al., 2010). Some authors have also supported the so-called “hypothermia hypothesis,” which suggests that hypothermia may be a direct or indirect cause of increased mortality risk in the first days of life (WHO, 1997), especially for pre-term or low-weight newborns with important thermoregulation impairments (Costello, 2000; Hazan et al., 1991; Knobel & Holditch-Davis, 2007; Lunze et al., 2013; Pellegrino et al., 2022). In this theoretical framework, the malnutrition mothers may have experienced during gestation represented a further risk factor (Dalla Zuanna & Rosina, 2011; Derosas, 2009). Furthermore, hot weather conditions and inadequate food preservation during the transition from the first to the second year of life have been associated with increases in mortality, particularly during the weaning period in the summer and when resulting in the exacerbation of children’s gastrointestinal problems (Breschi & Livi Bacci, 1986; Breschi et al., 2000).

Recent research has also shed light on the role of maternal exposure to heat during pregnancy as a potential factor influencing neonatal mortality, with studies indicating that high temperatures may lead to adverse birth outcomes. However, this association has primarily been studied in contemporary populations, and there is limited research on its specific impact on neonatal mortality within historical contexts (Cil & Kim, 2022; Conte Keivabu & Cozzani, 2022; Kuehn & McCormick, 2017; Rekha et al., 2022). Analyses carried out both in developed and developing contexts have indeed demonstrated that exposure to high temperatures during pregnancy may affect the risk of various adverse birth outcomes, such as pre-term births, low-weight newborns, and stillbirth (Chersich et al., 2020; Dalugoda et al., 2022; McElroy et al., 2022). The aforementioned birth effects are generally considered leading causes of infant mortality, particularly neonatal mortality, but only a few studies have so far investigated the association between heat exposure during pregnancy and death in the first week of life (Dalugoda et al., 2022).

The current study aims to test this hypothesis in the pre-transitional population of Casalguidi, Tuscany, between 1819 and 1859. This is not trivial, as the fragility of pre-transitional populations could suggest that exposure to heat during pregnancy may have strongly impacted and shaped the neonatal mortality pattern. On the other hand, the increase in mortality associated with cold in the first stages of life may be so large and pervasive as to make other weather factors and mechanisms marginal. By considering the potential impact of exposures to both heat and cold, this paper provides a new and comprehensive description of the impact of weather conditions on children’s survival in historical populations.

Using data from parish registers and other historical documentation, we investigated whether high temperatures may have affected both early neonatal mortality (0–6 days) and late neonatal mortality (7–29 days). This subdivision was necessary because many of the abovementioned causes of death, not to mention stillbirths and spontaneous abortions after the 28th week of gestation, are assumed to occur more frequently precisely in that period.

Furthermore, the research delves into the potential variations in this relationship based on socioeconomic status (SES), considering that differential access to resources and the impact of parents’ working activities may have modified the association between maternal heat exposure and neonatal mortality. Specifically, the involvement of pregnant women in outdoor farming activities before and after recognition of pregnancy is explored, as it may have exacerbated the impact of heat stress on maternal health and birth outcomes. The study is anticipated to contribute valuable insights into the interplay between weather conditions, maternal heat exposure, and socioeconomic factors in shaping neonatal mortality in historical populations, offering a more nuanced understanding of this complex relationship.

In recent years, numerous studies have explored and validated potential adverse birth outcomes associated with exposure to elevated temperatures during pregnancy. Recent reviews and meta-analyses focusing on contemporary populations have identified pre-term birth as the most prevalent adverse effect, with stillbirth and low birth weight being other common events linked to gestational exposure to high temperatures (Chersich et al., 2020; Dalugoda et al., 2022). Although the precise physiological mechanisms underpinning these effects remain incompletely understood (Roos et al., 2021; Samuels et al., 2022), they are believed to involve the mother’s general inability to regulate body temperature during extreme heat, potentially leading to dehydration and hyperthermia (Dalugoda et al., 2022; Edwards, 2006; Konkel, 2019). Furthermore, the higher body temperature resulting from fetal metabolism in pregnant women, coupled with elevated ambient temperatures, may impede the body’s cooling mechanisms, reducing uterine blood flow and causing fetal oxygen deficiency. Consequently, these altered physiological conditions and poor health status during gestation may trigger inadequate fetal growth and placental insufficiency (Cowell et al., 2023; Samuels et al., 2022), culminating in low birth weight and/or premature birth (Bonell et al., 2022; Krishna & Bhalerao, 2011; Walker et al., 1969). Additionally, exposure to heat during pregnancy has been associated with the development of pre-eclampsia or eclampsia in mothers, conditions that elevate the risk of pre-term delivery, small-for-gestational-age birth, and stillbirth (Mao et al., 2023; Part et al., 2022). Based on observations that heat strain can induce oxidative stress and the release of various inflammatory markers (Selkirk et al., 2008; Wang et al., 2015), some researchers suggest that pre-term births could also be triggered by a robust inflammatory response at the maternal–fetal interface (Gomez-Lopez et al., 2020; Miller et al., 2023; Schifano et al., 2013).

However, these reviews have not conclusively determined the specific temporal window of pregnancy most vulnerable to extreme heat. Some studies suggest a U-shaped effect (Zheng et al., 2018), whereas others indicate different effects based on the trimester of gestation.

The extent to which high temperatures can influence the health of mothers and newborns can be influenced by individual, family, or contextual vulnerabilities. Among these, the family’s socioeconomic status is the most frequently discussed in the literature, although some authors use maternal education level as a socioeconomic status proxy. The prevailing consensus is that higher family wealth affords better living conditions, nutrition, and access to resources, protecting both the mother during pregnancy and the child in their early life from external hot weather and high temperatures (Bakhtsiyarava et al., 2022; Basu et al., 2018; Chersich et al., 2020; Ngo & Horton, 2016; Sexton et al., 2021). However, this finding has not been consistently replicated in studies of pre-transitional populations. For instance, Scalone and Samoggia (2018) identified higher risks of neonatal mortality during winter among nineteenth-century rural Emilia day laborers, and Karlsson et al. (2021) highlighted higher risks in winter months among farmers in northeastern Sweden at the turn of the twentieth century. In contrast, Fornasin and Rizzi (2022) found higher winter mortality among the upper class in Udine, Italy, attributed to the custom of well-off families sending their children away to be nursed. Contradictory results on the role of socioeconomic status were also found within the context of the hypothermia hypothesis. While Derosas (2009) did not find evidence of infant mortality differentials across socioeconomic groups during winter in Venice, Dalla Zuanna and Rosina (2011) attributed the highest neonatal mortality rates in winter to malnourishment in the poorest strata of the Veneto population. This inconsistency of results in historical populations is sometimes explained as a consequence of the intrinsic limitations (geographic, socioeconomic, etc.) of the populations analyzed.

Due to a later onset of demographic transition compared to other European countries, Italy experienced, during the nineteenth century, higher rates of infant mortality than the most important European countries (Bellettini, 1987; Vallin, 2002). Variations in infant mortality rates were notable across different Italian regions (Breschi & Livi Bacci, 1986; Del Panta, 1984; Manfredini & Pozzi, 2004). Central-Northern regions such as Lombardy (2,930 deaths per 10,000 live births, between 1823 and 1846), Emilia (2,550 in 1863–66), and especially Veneto (3,030 in 1854, Dalla Zuanna, 2017) experienced significantly higher mortality rates within the first year of life compared to other Italian regions, particularly Tuscany (Breschi, 1990; Breschi & Livi Bacci, 1986). Tuscany saw a decline in infant mortality rates from approximately 2750 in the Napoleonic period to about 2160 per 10,000 by the mid-nineteenth century and further to 1730 between 1883 and 1886 (Breschi, 1990; Del Panta, 1984, 1994), during a period when the national mean rate was around 2000 per 10,000. These geographic differences characterized the Italian demographic history for a long time, so much so that in 1910–1912, the average infant mortality rate for Tuscany was 1140 per 10,000, approximately 20% lower than the national average of about 1420 per 10,000 (Del Panta, 1984).

Regarding the structure of the infant mortality model, Tuscany is usually included in a group of regions characterized by high values of neonatal mortality and a marked difference in the probability of death between the winter-born generation and the others, especially those born in the summer (Breschi & Livi Bacci, 1986; Manfredini & Pozzi, 2004). In reality, both Breschi and Livi Bacci for 1872–1879 (1986) and Del Panta for 1863–1870 (1994) show that Tuscany had a neonatal mortality rate that was slightly lower not only than the national average but also than the rate of various regions of Central-Northern Italy. Their findings also confirm and support the presence of a clear survival disadvantage for the winter cohort, often associated with the widespread occurrence of respiratory diseases.

Beyond obvious differences in environmental and climatic factors among regions, the aforementioned differentials in both infant and neonatal mortality are also the result of socioeconomic factors and life conditions. Indeed, numerous studies on infant mortality, mainly based on individual-level data, have consolidated and confirmed this association, not only in the Italian context (Breschi & Pozzi, 1997; Breschi et al., 2000, 2004; Dalla Zuanna & Rosina, 2011; Derosas, 2004, 2009; Scalone & Samoggia, 2018) but also internationally (Bengtsson, 1999; Edvinsson, 2004; Gardarsdottir, 2002; Reid, 1997). For Tuscany, laborers, seasonal workers, and the poorest rural classes showed higher levels of infant mortality due to a combination of poverty, misery, and vulnerability to the effects of cold weather. These geographic and socioeconomic differences greatly influenced infant and child survival, highlighting the complex interplay of historical, environmental, and social factors in shaping the infant mortality patterns of that time.

In the mid-nineteenth century, Casalguidi was a community under the Grand Duchy of Tuscany, situated a few kilometers from the cities of Pistoia, Prato, and Florence. The territory comprised three distinct zones: hilly, plain, and the village (Fig. 1). The hilly and plain areas housed small-sized farms — 5.9 hectares on average in 1835 (Detti & Pazzagli, 2000) — scattered farmhouses, and, especially in the plain, some proto-industrial activities. Most of this territory was characterized by the typical Tuscan sharecropping system. The village, on the other hand, consisted of smaller houses, shops, and artisan workshops, accommodating a small bourgeoisie and many artisans. Numerous families of day laborers also resided in the village in poor and very tiny houses.

The majority of households in Casalguidi were engaged in agriculture, with sharecroppers (48.2%), small landowners (18.2%), and day laborers (12.4%) making up about 80% of the population. Sharecroppers lived on the farm of an absent landowner, renting the land for half of the crop. The contract tied the whole household, whose structure and size should fit the farm size. Work was organized rigidly, with specific roles assigned to each member (Kertzer & Hogan, 1989). Males, in particular, held a privileged position, as they not only ensured the household the crop and the food but also because the renewal of the contract depended on the presence of a large household male labor force (Doveri, 2000).³ Sharecroppers were therefore very attentive to babies and children as they represented their chances to remain on the same farm (Manfredini & Breschi, 2013). Day laborers, on the other hand, did not have access to land. They were hired on a temporary basis and lived in small, mobile nuclear family units (Manfredini, 2003; Nani, 2012).

The rural economy of Casalguidi was primarily based on the cultivation of wheat, oil, and wine, alongside a growing artisanal sector. This led to a community with a certain level of socioeconomic differentiation. Despite this, families faced precarious living conditions and economic situation, as evidenced by the high proportion (35.2%) of households exempt from paying the family tax due to manifest indigence, and by over 51% of households who paid the lowest rates (Manfredini & Breschi, 2008a, 2008b).

From a demographic perspective, the population of Casalguidi grew from 1906 inhabitants in 1819 to 2690 in 1859, with an increase of 41.1% over 40 years. The underlying demographic system presents the typical characteristics of pre-transitional populations, with high fertility (TFT = 5.2 children per woman) and low survival (e₀ = 35.3 years) due to high infant mortality, around 2000 per 10,000 live births (Breschi et al., 2004). Migration also played a significant role in this demographic pattern. If the net migration rate was “only” 195 per 10,000 inhabitants over the entire period, the mean turnover rate was by far higher, over 1300 per 10,000. It was a consequence of the large number of rural landless families (sharecroppers, agricultural wage-earners, seasonal workers, and day laborers), which were subject to strong and rapid turnover, especially among day laborers (Manfredini, 2003; Nani, 2012).

The rural nature of the community, along with its poor living conditions, made it particularly sensitive to economic fluctuations, especially regarding wheat prices. This sensitivity had notable effects on the demographic behaviors of the population. Recent studies (Manfredini et al., 2022) reveal that periods of high prices often triggered short-term crises, leading to increased migrations and mortality rates, particularly among the vulnerable and impoverished families of day laborers. Additionally, these crises resulted in changes to marriage behaviors, particularly among farmers, who frequently delayed marriages during times of high prices. Throughout the period here considered, significant price increases were often associated with major epidemic events, such as the typhoid epidemic of 1816–1817 and the cholera epidemic of 1855 (Fig. 2).

Figures 2 and 3 display the trend of average annual temperatures and monthly temperatures in the area of Casalguidi. The climate in this area is typically temperate, with significant variations in temperatures and other climate variables (e.g., humidity, rainfall) throughout the year and over the years. The annual trend shows a sharp and rapid drop in mean annual temperatures from over 18 °C in 1819–1820 to around 14.5 °C in the period 1823–1838. After that, temperatures leveled off between 15 and 16 °C.

As for seasonal variations (Fig. 3), Casalguidi experiences very hot summers (especially July and August) and cold and harsh winters (January, February, and December), with spring and fall seasons fluctuating between the other two most extreme seasons and showing the largest variability of mean monthly temperatures. This pattern resulted in significant weather variations, such as between 1819 and 1859, when mean monthly temperatures floated between the cold January (median value of 5.7 °C) to the hot July (median = 25.6 °C). Furthermore, the same plot also highlights the outliers, months with mean monthly temperatures that were either extremely cold or extremely hot compared to the others included in the temperature time series used here. In June 1835, September 1828, and December 1851, the mean temperature was much lower than expected for those months, while the weather recorded in August 1820 was particularly hot according to the temperature series for that summer month.

Past populations were extremely vulnerable to adverse weather conditions, as evidenced by a notable increase in infant mortality rates during harsh winter conditions that numerous studies have extensively documented. The disproportionally heightened vulnerability of winter-born generations compared to their summer counterparts was so pronounced that scholarly attention has primarily focused on this correlation, thus leaving other potential weather-related mechanisms and factors influencing infant mortality significantly underexplored in scientific literature.

According to findings from contemporary populations, this study examines the impact of maternal exposure to heat during pregnancy on neonatal mortality in a pre-transitional Tuscan population, with potential combined effects alongside those from cold weather. The historical-demographic nature of the study limited the precise characterization of potential birth outcomes, as typically done in contemporary populations for pre-term births, low-weight births, and stillbirths, which are the most common health consequences for newborns associated with maternal heat exposure during gestation. Nonetheless, this research provides valuable insights into how environmental factors influenced neonatal mortality in a demographic system significantly shaped by the intensity and characteristics of early-life mortality.

The analysis of the population of Casalguidi between 1819 and 1859 essentially confirms that maternal exposure to high temperatures during pregnancy had a negative impact on newborns. Specifically, a significant increase in the risk of death was observed in the first week of life, while no significant changes were noted in the rest of the first month. This could be partly attributed to the inclusion of events in the dataset where it was unclear whether they were stillbirths or deaths occurring shortly after birth. Nevertheless, this finding aligns with existing evidence linking maternal heat exposure during pregnancy to adverse birth outcomes such as stillbirths, pre-term births, and low-weight births, which present high incidence in the perinatal and early neonatal period (Basta et al., 2022; Beck et al., 2010; Kannaujiya et al., 2022; Oza et al., 2015; Wondie et al., 2023). For example, Lawn and colleagues (2006) estimated that approximately 28% of perinatal deaths that occurred in 2000 were a result of prematurity.

Delving into the mechanisms behind that association, it was shown that high temperatures led to harmful and fatal consequences in the first week of life if the mother had been exposed to heat in the second and, to a lesser extent, the third trimester of pregnancy. While this is not a novelty, as previous research has already emphasized this time frame for pre-term births (Li et al., 2021; Shankar et al., 2023), stillbirths, and spontaneous abortions (Li et al., 2018), as well as low-weight births (Basu et al., 2018; Ngo & Horton, 2016), it does help to reinforce this evidence. This is crucial due to the limited number of studies on this topic, the different methodologies used to measure and assess environmental and weather conditions, and the diverse social and economic factors at play, which currently hinder a comprehensive understanding of which gestational period is most closely linked to adverse events resulting from prenatal heat exposure. To cite a few, in a recent study on 15 US hospitals, found a heightened risk of pre-term birth when the mother had experienced heat exposure in the first trimester of pregnancy, whereas Wang and colleagues (2019) did not find any clear indication of periods of heightened vulnerability during gestation for adverse birth outcomes.

The model used in this study has also provided insights into the influence of socioeconomic status on the risk of early neonatal mortality. The findings reveal that infants born to sharecropper couples experienced a significantly lower risk of death in the first week of life compared to children of day laborers when the mother was exposed to particularly high temperatures during the second trimester of pregnancy. Although the poor living conditions of day laborers likely contribute to this outcome, as evidenced by approximately 43% of them being exempt from taxes due to evident indigence, a similar exemption also applied to 32% of sharecropping families, which demonstrated a more pronounced protective effect on newborns than the relatively better-off smallholders (16% exempted). Hence, the lower mortality risks observed in children of sharecropping couples cannot be solely ascribed to their living conditions, nor can they be attributed to the direct effects of household characteristics, as indicated by our models and in alignment with other studies conducted on the same community (Breschi et al., 2000). Following Kertzer and Hogan (1989), the value of children within sharecropping dynamics potentially influenced their lower infant mortality risks. Sharecroppers were able to allocate additional care and attention to children not solely for their participation in farming activities but also because they bolstered the family’s bargaining power by ensuring the future family labor force. In essence, these findings emphasize that family welfare and protection was not simply a matter of parents’ economic conditions, but also involved socio-cultural considerations, even in communities with relatively minor differentials in SES. Ultimately, in pre-transitional societies, the survival of children in their earliest stages of life was jeopardized not only by cold weather conditions and associated diseases but also by potential maternal exposure to high temperatures during pregnancy.