This was a prospective, quasi-randomized, unblinded clinical trial.

Inborn infants between 240/7 and 336/7 weeks’ gestation, with BW between 500 and 3000 g, were admitted in a level III university neonatal unit.

Infants with major malformation were excluded as well as infants with an estimated birth weight >3000 g since they did not fit in the control bag.

Participants were randomized weekly, from Monday to Sunday, with an initial schedule of 60 weeks, generated by the Statistical Analysis Software (SASࣨ, Cary, NC, USA, version 9.4). The software created a list designating each type of bag according to the week of the study. The authors chose this form of randomization to facilitate the use of the correct bag by the clinical team, who assigned each participant according to the week, avoiding allocation errors.

To estimate the sample size, the authors conducted a 2-year retrospective audit in the neonatal unit. The mean admission hypothermia rate, considering axillary temperature below 36.0 °C, was 18% in premature infants <34 weeks gestation, who were managed at birth using the customary (control) bag. Considering a reduced admission hypothermia rate of 5% in the intervention group as clinically important, the study required a sample of 94 neonates per group at 5% type I error to achieve a power of 80%.

The control group received the plastic bag commonly used in the hospital (Figure 1), made of a non-sterile single layer of polyethylene, measuring 30 × 45 cm, with one open side – that can be closed by a resealable closure zipper – and a closed opposite side, where a collar is made using scissors to allow the newborn's head and neck pass through. Soon after birth, the fontanelle area was dried; then, a 15 × 15 cm plastic cap was placed directly on the newborn's head and a wool cap was on top.

Figure 1.

Intervention and control plastic bag.

(0.26MB).

The intervention group received the NeoHelp™ device (Vygon®, France) (Figure 1), consisting of a sterile double-layer polyethylene bag, available in three sizes (small, medium, and large), with an internal pre-shaped foam cushion, and a self-adjusting double plastic hood attached. According to the manufacturer's recommendations, the infant was placed into the bag, and it was closed by a hermetic Velcro seal. The infant's head was covered by a plastic cap included in the kit, and a wool cap was placed on top.

Both the radiant warmers (target temperature of 36–37 °C) and the transport incubator (target temperature of 35–37 °C) remained ready for use 24 h a day. The servo control probe of the heated warmer was continuously attached to the mattress sheet by microporous tape (Micropore™, 3M™, Brazil), and the temperature was adjusted during stabilization as needed. For promoting heating of the plastic bag, they were extended on the mattress of the radiant warmers, previously the birth.

Infants were attended to in a stabilization room 10 m away from the delivery room (DR), reachable in about two to three minutes. The environmental temperatures of the DR and stabilization room were maintained between 24 °C and 26 °C. According to the local routine, briefly after birth, the preterm infants were placed in a conventional bassinet without drying, wrapped in a heated surgical drape, and transported to the stabilization room. Following the Brazilian Pediatrics Society guidelines applicable at the time of the study,6 preterm infants were placed inside the plastic bag, under a radiant warmer, and resuscitated or stabilized as needed. The bag was continuously closed, being removed only in the neonatal unit when the body temperature reached >36.5 °C. The right upper limb was left out, where the pulse oximeter sensor was placed. Heated gases for ventilation were unavailable.

The infant's temperature was measured in the axillary region using the BD® thermometer (Sparks, Maryland, USA) with a minimum recorded temperature of 32 °C and a 0.1 °C resolution, in two-time points: immediately before the newborn left the stabilization room and at admission to the neonatal unit, still inside the transport incubator. The maternal temperature was measured just before delivery (15 min, on average), using the same thermometer. Temperatures of the delivery and stabilization rooms were registered by Herweg® thermometers (model 1610; Timbo, Brazil).

Hyperthermia rate of >15% in the intervention group.4,15

The primary outcome was admission hypothermia, defined as an axillary temperature at admission to the neonatal unit of <36.0 °C. The secondary outcome was the intra-hospital mortality rate.

Hyperthermia was considered an adverse effect, defined as an axillary temperature of >37.5 °C at admission.

Continuous variables were presented as the median and interquartile range (IQR) due to non-parametric distribution and compared by Mann–Whitney test. Categorical variables were expressed by percentage (%) and compared by χ2 tests, or Fisher's exact test.

To assess the factors associated with admission hypothermia, the authors performed a univariable logistic regression analysis. Then, the authors conducted a multivariable logistic regression analysis using the forward stepwise regression. This method can be used in settings where the number of variables under consideration is larger than the sample size, as in the present study, and it minimizes correlated variables (such as GA and BW). The results were expressed as odds ratios (OR) and 95% confidence interval (95% CI).

To verify the different effects of the interaction between the type of bag and the BW and GA, the authors adjusted the logistic regression model considering the independent interaction between these two variables vs. the bag.

To evaluate the effect size, the authors calculated 95% CI for frequency, risk ratio, the relative and absolute reduction of risk rates, and the number needed to treat.

Data analysis was performed using SAS version 9.4 (Cary, NC, USA), Level of significance was set at 5%.

The Research Ethics Committee approved this study with authorization number CAAE 88760918.8.0000.5404. Before they signed the written consent form, all parents were informed about the study data and intervention to be performed throughout the study.

From June 2018 to September 2019, this level III neonatal unit registered 187 births at <34 weeks gestation, and 185 were eligible (two died in the DR). Of these, the authors excluded 14 infants: seven with major malformations, two with BW >3000 g, and five born outside the DR. The final sample, thus, comprised 171 infants randomized into two groups: 76 in the intervention group and 95 in the control group (Figure 2). There was no allocation bias between the groups.

Figure 2.

Flow diagram.

(0.29MB).

Both groups showed similar baseline demographic characteristics, including BW and GA, but the control group had a significantly higher frequency of cesarean delivery (Table 1). Infants <1000 g corresponded to 19.3% of the total sample and those <28 weeks gestation to 17.5%.

The overall percentage of admission hypothermia was 9.3%, the median temperature at admission was 36.6 °C (IQR, 36.2–37.0 °C), and the hyperthermia rate was 4.7%.

Admission hypothermia (AH) was significantly less frequent in the intervention group (2.6% vs. 14.7%, P = 0.007), which also presented a higher median admission temperature (36.8 °C IQR 36.5–37.1 °C vs. 36.5 °C IQR 36.1–36.9 °C, P < 0.001) (Supplementary appendix – Table 1). A subgroup analysis considering the endpoint temperature of 36.5 °C showed that the prevalence of admission hypothermia was also significantly different between the intervention and control group (23.6% vs. 47.3%, P = 0.001).

When the study reached the expected number of participants in the control group (95), only 76 participants were enrolled in the intervention group. As a result of the proposed randomization schedule, the following weeks would be dedicated to the control bag. Due to the unpredictable weekly birth rate and having reached the pre-specified study enrollment, the authors analyzed the present data, and given the statistical difference between the AH rates between the groups (2.6% vs. 14.7%, p = 0.007), we calculated the size effect indices: 95% CI for intervention bag was 0.0–6.2% vs. 7.6–21.8% for control bag; the risk ratio for the intervention group was 0.18 (95% CI 0.04–0.79); the relative risk reduction was 82%, and the absolute risk reduction was 12.1% in favor of the intervention bag; and an NNT of 8. Thus, the authors decided to interrupt the allocation of patients.

Regarding environmental variables associated with admission hypothermia, both the maternal temperature (36.1 °C vs. 35.7 °C, p = 0.011) and newborn temperature at the stabilization room (36.7 °C vs. 36.5 °C, p = 0.044) were higher in the intervention group. Other environmental temperatures were adequate and similar in both groups. Transport duration and age of admission at the neonatal unit were also similar in both groups (see the supplementary appendix – Table 1).

The intervention group showed a higher rate of hyperthermia (9.2% vs. 1%, p = 0.023) (supplementary appendix – Table 1). When compared with the non-hyperthermic group, the hyperthermic infants showed no statistical difference in BW – 1677 g (IQR 1308–1950) vs. 1420 g (IQR 1040–1795), p = 0.195 – nor in GA – 31.5 weeks (IQR 28.5–32.5) vs. 31.0 weeks (IQR 28.0–32.0), p = 0.862. Other maternal and neonatal characteristics were also similar. Variables related to temperature, however, were different: the hypothermic group had higher maternal temperatures 36.7 °C (IQR 36.2–37.2 °C) vs. 35.8 °C (IQR 35.3–36.2), p < 0.001; higher DR temperature 26.1 °C (24.7–26.4 °C) vs. 23.4 °C (22.3–26.0 °C), p = 0.039; and a higher temperature at stabilization room 37.3 °C (IQR 36.9–37.8 °C) vs. 36.5 °C (IQR 36.1–36.9 °C), p < 0.001. The temperature returned to a normal range around two hours of life. The authors observed no significant clinical event, except self-limited tachycardia.

Supplementary appendix – Table 2 presents a stratified assessment by BW and GA vs. the bag used. Regarding admission hypothermia, the intervention bag showed to be more effective in newborns with BW ≥1000 g (0 vs. 7, p = 0.015) and gestational age ≥28 weeks’ gestation (0 vs. 11, p = 0.003). Hyperthermia was more prevalent in the intervention group for BW ≥1000 g (7 vs. 1, p = 0.025). The authors found no differences related to GA subgroups (see the supplementary appendix – Table 2).

The univariate logistic regression analysis showed that the intervention group (OR, 0.16; 95% CI, 0.03–0.71), GA (OR, 0.80; 95% CI, 0.67–0.97), BW (OR, 0.99; 95% CI, 0.99–0.99), SGA (OR, 3.38; 95% CI, 1.16–9.83), and need for tracheal intubation in the DR (OR, 3.70; 95% CI, 1.29–10.6) were positively associated with AH (Table 2).

In the multivariable logistic regression, the intervention bag remained independently associated with admission hypothermia, showing an 86% risk reduction of the condition (OR, 0.14; 95% CI, 0.03–0.64), and with BW, presenting a 30% chance reduction for every 100-g increase in BW (OR, 0.997; 95% CI, 0.996–0.999).

After adjusting the multiple models, considering the independent interaction between BW and GA and polyethylene bag, the authors found that the interaction with BW showed an approximately 81% reduced risk of admission hypothermia. Conversely, the interaction between GA and the device resulted in a 71% reduced chance of admission hypothermia (see the supplementary appendix – Table 3).

The global mortality rate was 4.7% (8/171), and the death rate was similar in both groups (p = 0.464). None of the hyperthermic infants died.