We conducted a prospective monocentric analysis in 1018 neonates at a tertiary university hospital. The study consists of 3 interrelated studies (16–18), in which all neonates where uniformly treated regarding blood glucose (BG) screening and thermal management throughout the recruitment period.

Neonates were enrolled from 05/2020 to 09/2022 at the University Hospital Düsseldorf. Neonates with one or more of the following risk factors were assigned to the risk group: small or large for gestational age (SGA; LGA) (birth weight <10^th/>90^th centile); fetal growth restriction (FGR) [Gordijn et al. (19)]; maternal/gestational diabetes; prematurity (35 + 0-36 + 6 weeks of gestation); maternal risk factors (including preeclampsia, hypertension, medication with beta-blockers, antidepressants or metformin); perinatal stress (asphyxia, respiratory distress, arterial umbilical cord blood pH <7.1, vacuum extraction/unplanned cesarean due to pathological cardiotocography); hypothermia. Neonates without a risk factor were assigned to the control group. Neonates at-risk for hypoglycemia received a standardized BG screening and management according to local standard operating procedure (SOP) (18). The control group received at least two BG measurements at the age of 2–3 hours and at the age of 36–72 hours. In case of low BG levels further measures were taken according to the hypoglycemia SOP (18). As part of the clinical routine rectal body temperature was measured in the delivery room (age: 1–3 hours), after transferal to the maternity or pediatric ward, and at clinical discretion. Hypothermia was defined as a rectal temperature <36.5° (3); mild hypothermia: 36.0-36.4°C; moderate/severe hypothermia: <36.0°C. Neonates without a documented temperature measurement (n=7) were assigned to the non-hypothermic group as it was assumed that there were no clinical signs of hypothermia that would have warranted a temperature measurement. Ambient outside temperature values for Düsseldorf were derived from www.wetterkontor.de ; Deutscher Wetterdienst (Access date: March 26, 2025). If recruitment took place in the same month in two or three years, the mean temperature was used for analysis. The maternity ward had no air conditioning.

Clinical data were obtained from the medical files. Data were recorded with Claris FileMaker Pro version 19 (Claris International Inc).

The studies were approved by the ethics committee of the Medical Faculty of the Heinrich Heine University Düsseldorf according to the Declaration of Helsinki. Both parents of each neonate gave written informed consent.

Statistical analyses were performed using IBM SPSS Statistics version 29.0.2.0 (20) (IBM Corp). χ² and Fisher exact tests were used to analyze categorical variables. Medians were compared using Mann-Whitney U test with Bonferroni correction. Medians with IQRs are reported.

A mixed linear model was conducted to estimate whether body temperatures correlate with the BG levels, using subjects and age (minutes) as random effects. Binary logistic regression analyses were used to predict hypothermia onset, based on different hypoglycemia risk factors. Spearman correlation was computed to test for an association between the ambient temperature in the birth month and the percentage of neonates with hypothermia in the respected month.

In the hypothermia group there was a significant higher percentage of SGA and/or FGR neonates compared to the non-hypothermic neonates (40.3% vs. 18.4%, P<.001) ( Table 2 ). The comparison of the percentage of premature neonates and neonates with perinatal stress in the hypothermic vs. non-hypothermic neonates did not reach statistical significance after Bonferroni correction (22.5% vs. 14.8%, P=.007; 25.4% vs. 17.1%, P=.004) ( Table 2 ). The percentage of LGA neonates was lower in the hypothermic compared to the non-hypothermic neonates (5.1% vs. 18.2%, P<.001) ( Table 2 ). In line with these results, the multiple binary logistic regression analysis showed that the risk factors SGA and/or FGR, prematurity and perinatal stress were positively associated with an increased risk to develop hypothermia, whereas LGA was associated with a decreased risk to develop hypothermia ( Table 3 ).

In the hypothermic group more neonates were born vaginally compared to the non-hypothermic group (49.6% vs. 37.6%, P=.001) ( Table 2 ).

A total of 1350 BG/temperature pairs were analyzed. BG levels correlated positively with the body temperature (F(1, 1332.999)=99.191, P<.001). 202/566 (35.7%) BG values ≤2.5 mmol/l (≤45 mg/dl) and 29/54 (54%) BG values <1.7 mmol/l (<30 mg/dl) had a concomitant temperature documented.

The number of infants who developed at least one hypoglycemic episode ≤2.5 mmol/l (≤45 mg/dl) and <1.7 mmol/l (<30 mg/dl) was significantly higher in the hypothermic group compared to the non-hypothermic group (53.4% vs. 26.2%, P<.001 and 12.7% vs. 1.4%, P<.001) ( Table 2 ). Moreover, the total number of BG levels between 2.6-3.0 mmol/l (46–54 mg/dl) and ≤2.5 mmol/l (≤45 mg/dl) was higher in hypothermic compared to non-hypothermic neonates ( Table 2 ). Correspondingly, more neonates in the hypothermic group had to be transferred to the children’s hospital for hypoglycemia treatment compared to non-hypothermic neonates (17.4% vs. 4.1%, P<.001). Additionally, more neonates in the hypothermic group compared to the non-hypothermic group were treated with glucose gel (51.7% vs. 27.4%, P<.001) ( Table 2 ). Fewer infants in the hypothermic group received exclusively breastmilk compared to the non-hypothermic group (3.0% vs. 12.9%, P<.001), and more neonates in the hypothermic group required transient feeding via a nasogastric tube (11.4% vs. 4.7%, P=.001) ( Table 2 ).

The percentage of neonates that had at least one hypoglycemic episode ≤2.5 mmol/l (≤45 mg/dl) or <1.7 mmol/l (<30 mg/dl) and the lowest measured BG level did not differ significantly between mild and moderate/severe hypothermia ( Supplementary eTable1 ).

In a subgroup analysis of control neonates without hypothermia and neonates with hypothermia as the only hypoglycemia risk factor, the minimal measured BG level was significantly lower in the hypothermic neonates compared to the control neonates without any hypoglycemia risk factor (2.6 (2.2; 2.9) vs. 3.1 (2.6; 3.6) mmol/l (46 (40; 52) vs. 56 (47; 64) mg/dl) P<.001) ( Supplementary eTable 2 ). Furthermore, the percentage of neonates who had to be transferred to the children’s hospital was higher in the hypothermic vs. the non-hypothermic control neonates (18% vs. 2%, P=.002) ( Supplementary eTable 2 ). None of the hypothermic neonates (0%) compared to the non-hypothermic control group (23%) received breastmilk only (P=.002). Correspondingly, the rate of neonates who received breastmilk plus formula was significantly higher in the hypothermic compared to the non-hypothermic control group (96% vs. 67%, P<.001) ( Supplementary eTable 2 ). Treatment with glucose gel and transferal rate to the children’s hospital due to hypoglycemia were higher in the hypothermic group compared to the non-hypothermic control group but failed to reach statistical significance after Bonferroni correction ( Supplementary eTable 2 ). Other clinical characteristics did not differ between both groups ( Supplementary eTable 2 ).

SGA and/or FGR neonates with and without hypothermia were compared. Clinical characteristics and the number of other risk factors did not differ between both the two groups ( Table 4 ). However, the number of infants who developed at least one BG level ≤2.5 mmol/l (≤45 mg/dl) and <1.7 mmol/l (<30 mg/dl) was significantly higher in the hypothermic SGA and/or FGR neonates compared to the non-hypothermic SGA and/or FGR neonates (58% vs. 35%, P<.001 and 15% vs. 4%, P=.003) ( Table 4 ). Moreover, the minimal ever measured BG level was lower in the hypothermic compared to the non-hypothermic SGA and/or FGR neonates (2.4 (2.0; 3.0) vs. 2.7 (2.3; 3.1) mmol/l (44 (36; 54) vs. 48 (41; 56) mg/dl), P=.002) ( Table 4 ).

As the number of neonates with different risk factors varied throughout the study period, the analyses regarding the outside temperature in the birth month was only performed for the subgroup of SGA and/or FGR neonates. Figure 1 shows the percentage of SGA and/or FGR neonates with hypothermia for each birth month and the mean monthly outside temperature. There was no direct correlation between the mean outside temperature in the birth month and the percentage of neonates with hypothermia in the corresponding month (Spearman correlation coefficient [r _s] = -0.294, P = .35). However, the number of neonates with hypothermia was higher in the two coldest months (December/January, mean outside temperature 5.0°C) compared to the two warmest months (July/August, mean outside temperature 19.7°C) (48.4% (15/31) vs. 19.6% (16/61), P=.04).

Neonates born SGA and/or FGR, premature or with perinatal stress were at higher risk of developing hypothermia, all of which are established risk factors for this complication (5, 14, 24). On the other hand, LGA neonates had a lower risk of developing hypothermia.

No correlation was found for the average monthly outside temperature and the percentage of children who developed hypothermia. However, more children had hypothermia in the two coldest compared to the two warmest months. This is consistent with several studies especially from developing countries that have shown an association between environmental temperature and onset of neonatal hypothermia (25–27). This is supported by data showing that increasing the temperature in the operating room to 23°C compared to 20°C during a cesarean delivery can reduce the rates of hypothermia (28). Thus, it might be worth further increasing the attention to newborn temperature management during the coldest months, at least if the care units are not strictly temperature controlled/conditioned.

Interestingly, in contrast to existing literature which reports a higher incidence of hypothermia among neonates delivered via cesarean section (24), hypothermia was more frequent in neonates born vaginally in our cohort. Possible explanations could be: 1) All children who are delivered via cesarean at our facility are promptly transferred to a specific room that is constantly maintained at 30-32°C for initial care of high-risk newborns (dried off, wrapped in warm towels, and provided with a cap). In contrast, after vaginal deliveries the children are usually initially left with their mothers in the delivery room for bonding, and the provision of a hat and warm towels may be delayed. 2) Children of mothers who had a cesarean often receive more frequent care by the nursing staff in the first few days due to the immobility of the mothers: Nursing staff might be more vigilant than parents in preventing heat loss, for example by using a radiant heater for every diaper change.

To treat hypothermia or prevent further heat loss 25.6% neonates were temporarily placed in a warming bed/incubator in the maternity ward nursery. Of note, some neonates were placed in the warming bed as a preventive measure but did not experience hypothermia. So more than one in four neonates was temporarily separated from their mother. This, in conjunction with the high rate of hypoglycemic episodes, might explain the relatively low proportion of infants in the hypothermia group who exclusively received breastmilk (3%). Conversely, reduced breastfeeding may lead to less skin-to-skin contact, which could contribute to an increased risk of hypothermia. While the direction of causality remains speculative, these findings highlight several aspects that offer room for improvement.

In our cohort, body temperature correlated positively with BG levels. Furthermore, neonates with hypothermia were at higher risk to develop hypoglycemia ≤2.5 mmol/l (≤45 mg/dl) and <1.7 mmol/l (<30 mg/dl), had more hypoglycemic episodes and lower BG levels, with a higher need for treatment with glucose gel and transfer to the children’s hospital.

To exemplarily analyze a more homogenous group in terms of hypoglycemia risk factors and other clinical characteristics, we performed a subgroup analysis on SGA and/or FGR neonates. Within this subgroup, nearly 60% of hypothermic neonates developed at least one hypoglycemic episode ≤2.5 mmol/l (≤45 mg/dl) and nearly 15% at least one hypoglycemic episode <1.7 mmol/l (<30 mg/dl) while other clinical characteristics did not differ between hypothermic and non-hypothermic SGA and/or FGR neonates.

The high rates of neonatal hypoglycemia especially in the low range of <1.7 mmol/l (<30 mg/dl) are concerning, as these have been reported to be associated with a suboptimal neurodevelopment with lower IQ scores and worse performance regarding fine motor function and visual-motor integration (7). The potential impact of concurrent hypoglycemia and hypothermia on neurodevelopmental outcomes in neonates has not yet been systematically investigated. It is even conceivable that, in the context of hypoglycemia, hypothermia may exert a neuroprotective effect, similar to its established therapeutic use in hypoxic-ischemic encephalopathy (29). However, this hypothesis remains speculative and further prospective studies are warranted to assess the interactions and synergies between hypoglycemia and hypothermia in the neonatal population. Furthermore, the direction of causality between hypothermia and hypoglycemia in neonates cannot be answered based on the available data. It is known that neonates activate non-shivering thermogenesis via brown adipose tissue and that glucose uptake in brown adipocytes is stimulated by cold exposure (30). The high glucose demand in case of hypothermia may be especially a problem in neonates with limited glycogen reserves (e.g. SGA or preterm neonates). On the other hand, in adults it was shown in a glucose clamp experiment, that hypoglycemia leads to a decrease in body temperature (31) and there are reports that an overdose of antidiabetic drugs can lead to hypoglycemia and hypothermia (32, 33). Still, in neonates it remains unclear whether hypoglycemia leads to hypothermia due to lack of energy reserves or whether hypothermia leads to hypoglycemia due to increased glucose consumption in attempt to stabilize body temperature. Both pathomechanisms are plausible and should be further evaluated in future studies, e.g. evaluating whether the implementation of combined measures to avoid heat loss during the first hours and days of life (such as increasing the cesarean operating room and delivery room temperature, safe skin-to skin care, cap, radiant heater etc.) can reduce the rate of hypothermia and associated hypoglycemia.

The fact that the possible connection between hypoglycemia and hypothermia is underrecognized in everyday clinical practice is also shown by the low rate of temperature documentation in case of hypoglycemia (35.7% at BG ≤2.5 mmol/l (≤45 mg/dl) and 53.7% at BG <1.7 mmol/l (<30 mg/dl).