Quality Improvement Approach to Reducing Admission Hypothermia Among Preterm and Term Infants

Our efforts took place at a freestanding, 298-bed, Midwestern children’s hospital. The hospital houses a 70-bed level IV NICU. Inborn infants come from a birth center that is connected to the children’s hospital. The NICU has nearly 800 admissions annually, with approximately half of those admissions coming from the birth center, where >3000 infants are delivered each year. Although physically connected, the birth center represents a separate entity from the children’s hospital, with independent aims, financial interests, and leadership. Stakeholders for this project included attending neonatologists, neonatology fellows, neonatal nurse practitioners (NNPs), pediatric residents, as well as >250 staff nurses and >20 respiratory therapists. Additional stakeholders included obstetricians, anesthesiologists, and birth center nurses and technicians.

Our quality improvement team included the NICU medical director, a NICU fellow, the director of our NNP program, another NNP, a charge nurse, our clinical nurse educators, our clinical coordinator, and the birth center nursing coordinator. The team was formed to include hypothermia prevention champions across multiple areas.

We developed process maps and fishbone diagrams to better understand our current state and used this information, combined with previously published quality improvement efforts in this area, to devise our drivers of change. In our key driver diagram (Fig 1), we identified staff knowledge, access to the necessary supplies to support normothermia, and alignment between the children’s hospital and birth center as the primary drivers of change.

Interventions included audits of practice both to understand current tendencies and to serve as an educational reminder of hypothermia prevention guidelines. We completed several expansions to our use of chemical heat mattresses and polyethylene bags. Chemical heat mattresses, initially reserved for infants aged <34 weeks, were expanded first to all infants during transportation between the birth center and the NICU, then to infants aged <35 weeks, and finally to any infant prenatally diagnosed with a congenital anomaly requiring NICU admission. Polyethylene bags were initially used for infants aged <28 weeks, and this cutoff was increased to 32 weeks. Finally, we intensified staff education and reminders to support compliance through in-person education, guideline badge cards, and placard reminders on radiant warmers. These interventions were devised and used through 4 separate plan-do-study-act (PDSA) cycles. Interventions for cycles were developed on the basis of information gathered in the previous cycles, as detailed in the results section.

Of note, our interventions were applied to those deliveries for which the NICU team was present. In our institution, the NICU team is present for the delivery of infants with a gestational age of <35 weeks, assisted deliveries, cesarean deliveries, the delivery of infants with known congenital anomalies, and multiple gestation deliveries or at the discretion of the obstetrics provider. NICU admission is anticipated for infants aged <34 weeks’ gestation, for infants requiring more than a nasal cannula for respiratory support, and for infants with congenital anomalies requiring immediate assessment and/or intervention. Outcomes were not tracked for those infants remaining at the birth center.

Our outcome measure was admission hypothermia, defined as an admission axillary temperature <36.5°C. Infants were excluded if they were transferred from an outside institution or were admitted to the nursery before transfer. Data were collected initially via audit forms and subsequently through chart review and analyzed on a monthly basis. We identified 2 process measures: delivery room ambient temperature and staff knowledge of the hypothermia prevention guidelines. Our goal delivery room ambient temperature was 74°F and was assessed via audits in 2015 and 2016. We evaluated staff knowledge via surveys before and after educational efforts in PDSA cycle 3. We used staff knowledge as a surrogate for guideline compliance because auditing deliveries directly was not feasible. We selected admission hyperthermia, defined as an admission temperature >38°C, as a balancing measure.

Data trends were analyzed by using Shewhart control p-charts and g-charts. These charts were generated by using the QI Macros for Microsoft Excel 2016 (KnowWare International Inc, Denver, CO). Shifts were defined as ≥8 consecutive points above or below the centerline, with control limits set at 3 σ.

The hospital Human Research Protection Program determined this quality improvement effort to be exempt from formal institutional review board approval.

In total, we assessed admission temperatures for 1097 infants between October 2013 and October 2019. Results of our improvement efforts are displayed in our control chart (Fig 2).

The first PDSA cycle during the spring of 2016 targeted compliance with hypothermia prevention guidelines, which included the use of polyethylene bags and chemical heat mattresses below gestational age cutoffs of 28 and 34 weeks, respectively, and recommended delivery room temperature be maintained at ≥74°F. This cycle involved audit forms completed by nursing staff after every admission to the NICU from the birthing center. Using these audit forms, we evaluated several guideline metrics and asked staff to measure ambient temperature in the delivery room, as well as the infant’s temperature, before leaving the delivery room. We hoped that completing these forms after a delivery would help improve knowledge of the guidelines for future deliveries. Unfortunately, the average delivery room temperature was 71.5°F, below the World Health Organization–defined target of 74°F. PDSA cycle 1 did not result in a detectable change in our hypothermia rate, but it did reveal a number of infants who were normothermic in the delivery room but hypothermic on admission.

The second PDSA cycle in the summer of 2016 focused on hypothermia occurring during transportation between the birthing center and the NICU and was based on the findings of PDSA cycle 1. We instituted the use of chemical heat mattresses during transportation, regardless of gestational age. Additionally, the responsibility for audit form completion was transitioned to the neonatology fellows to try to improve compliance because this was a much smaller group, and fellows attend nearly all extremely premature deliveries. We detected special-cause variation starting in April of 2016 temporally related to PDSA cycle 2, shifting our incidence of admission hypothermia to 15.4%. As audit form completion dropped to 42%, these forms were abandoned.

A potential further improvement in limiting hypothermia on transportation occurred in the spring of 2017 because the NICU was relocated within the hospital. Travel time between the birth center and the NICU was reduced from ≤15 minutes to as short as 2 minutes. There was not a shift in admission hypothermia associated with the move; however, the improvements detected in cycle 2 were sustained.

Although we achieved some success with PDSA cycle 2, our hypothermia rate remained above our target of 10%. To better understand residual predictors of hypothermia within our unit, we completed a bivariate analysis of patient characteristics related to admission hypothermia. The characteristics we investigated included gestational age at birth, being a multiple at birth, birth via cesarean delivery, and the presence of a congenital anomaly requiring NICU admission. We placed those factors that were significant on bivariate analysis into a multivariate logistic regression model. In Table 1, we display the results of this analysis, which we repeated each time we reached special-cause variation to determine targets for subsequent intervention. These phases also mirror changes in our thermoregulatory interventions. Phase 1 was defined as the time period before the first shift in our admission hypothermia rate, whereas phase 2 represents the period after PDSA cycle 2 when our hypothermia rate improved to 15.4%. Phase 3 encompasses the time period after PDSA cycle 3, which again resulted in a shift in our hypothermia rate, as will be described subsequently. Our regression model revealed increased odds of hypothermia for our infants with a gestational age 33 to 36 6/7 weeks, as compared with term infants for every phase (odds ratio [OR] 4.56 [1.55–13.42]; P = .006, 2.73 [1.12–6.63]; P = .027, and 1.90 [1.01–3.56]; P = .046, respectively). Interestingly, there was not increased odds of hypothermia for more premature infants, as compared with term infants. Birth via cesarean delivery decreased odds of hypothermia in phase 1 (OR 0.40 [0.18–0.86]; P = .019) but increased odds of hypothermia in phase 2 (OR 2.29 [1.09–4.82]; P = .029).

Additionally, we wanted to better understand our staff’s knowledge of our guidelines. Staff knowledge was a surrogate for compliance because auditing deliveries was prohibitively onerous. We sent staff a quiz on guidelines in November 2017 and found that only 9% could identify gestational age cutoffs for using hypothermia prevention tools.

The third PDSA cycle occurred in the spring of 2018. We expanded the use of polyethylene bags and chemical heat mattresses to <32 and <35 weeks, respectively. To address knowledge gaps, we also undertook extensive educational efforts, including flyers, e-mails, discussions, badge cards, and placards on the radiant warmers. PDSA cycle 3 resulted in special-cause variation, decreasing the admission hypothermia incidence to 9.9%. Our process metric of staff knowledge improved from 9% to 45% in May of 2018 after these interventions (P < .0001; Fig 2).

Finally, the fourth PDSA cycle started in the winter of 2018. We had noticed several infants admitted because of prenatally identified congenital anomalies with admission hypothermia. These anomalies included diagnoses such as gastroschisis or omphalocele, congenital diaphragmatic hernia, myelomeningocele, and complex congenital heart disease. In phase 3 of our regression analysis, admissions related to a congenital anomaly conferred increased odds of hypothermia (OR 1.95 [1.09–3.47]; P = .024). We expanded the use of chemical heat mattresses during resuscitation to infants requiring admission because of a congenital anomaly, regardless of gestational age. As simple maneuvers were potentially interrupted because of a more involved resuscitation, we hoped the use of an additional heat source would mitigate this interruption. We did not see another shift with PDSA cycle 4 or a change in the incidence of hypothermia among infants admitted because of congenital anomalies, which has remained at 13.8%.

Our g-chart to assess changes in our balancing measure, hyperthermia, did not reveal evidence of special-cause variation (Fig 3). Over the course of the study, 3.6% of infants arrived to the NICU with temperatures >38°C.

We successfully decreased the prevalence of admission hypothermia from 39.8% to 9.9% among all infants born at the birth center and directly admitted to our NICU. We attribute our success in reducing admission hypothermia to the expanded use of chemical heat mattresses and polyethylene bags, as well as staff education and attention to hypothermia during transportation to the NICU. As education is a challenging intervention to sustain, we have continued to distribute badge cards with guidelines to staff and maintain guideline placards on radiant warmers.

In contrast to other reports, we did not see a change in the rate of admission hyperthermia, our balancing measure, despite using a combination of chemical heat mattress and polyethylene bag.^12,19–21  This may relate to our practice of avoiding direct skin contact with the chemical heat mattress by placing a single blanket between the infant and the chemical mattress. Certainly, in several other studies, researchers have successfully reduced admission hypothermia without a concurrent increase in hyperthermia.^22–24 

Despite attention to delivery and operating room temperature, we were not able to achieve the World Health Organization–defined goal of 74°F, attaining an average temperature of 71.5°F during our data collection phase. We attribute the inability to adjust room temperature at delivery, at least partially, to challenges balancing the comfort of the mother during labor and the unpredictability of birth timing. Fortunately, we were able to achieve improvement in our outcome measure without achieving our targets on this process measure.

Interestingly, throughout our improvement efforts, our infants at highest risk for hypothermia were born between 33 and 36 6/7 weeks’ gestation. This group of infants is often excluded in quality improvement efforts to limit admission hypothermia, which are generally focused on a population aged <32 weeks’ gestation and/or <1500 g. We hypothesize several reasons for our findings. First, guidelines that focus on more premature infants may give the impression that more mature infants are not at risk and lead to more laxity in applying measures like hats, blankets, and skin-to-skin care. Second, these infants may spend more time in the delivery room because providers observe longer to determine placement in the NICU versus a level II nursery. Additionally, infants may be held by their parents before transfer to the NICU, which results in the infant being out from under the radiant warmer before transfer. Our data are in line with Russo et al²³  who reported an initial hypothermia rate of 35% in infants aged <28 weeks, 44% in infants aged between 29 and 32 weeks and 79% in infants aged between 33 and 34 weeks. This baseline data drove them to expand the use of chemical heat mattresses and occlusive wraps to all infants aged <35 weeks. Similarly, Andrews et al⁷  investigated hypothermia occurring among infants aged >35 weeks’ gestation and 1750 g in a mother-infant unit and found a hypothermia rate of 29.1% among those infants born between 35 and 37 weeks. On the basis of these studies and our data, we suggest the need to expand hypothermia prevention efforts to more mature infants.

We additionally identified those infants with prenatally recognized congenital anomalies requiring NICU admission as an at-risk group for admission hypothermia. These infants are frequently excluded from published hypothermia prevention efforts. The presence of a congenital anomaly emerged as a predictor of admission hypothermia. There are several features that could make these infants more vulnerable to hypothermia. Infants with abdominal wall defects have increased evaporative heat losses, which may not be completely mitigated by placement in a bowel bag. Other infants may require an extensive resuscitation, which could limit thermoprotective efforts. A focus on the presence of a congenital anomaly may distract from attention to maintaining normothermia, and, in cases in which immediate postnatal resuscitation is less intensive, these infants may stay in the delivery room for a more prolonged period of family bonding. We have yet to see a shift in admission hypothermia for this group, and they will continue to be a target of future efforts.

The present project has several important limitations. First, data collected from 2013 to early 2016 were based on audit forms, and infants for whom no audit form was completed would have been missed. We anticipate that this selection bias would have likely underreported hypothermia because those infants for whom audit forms were not completed were less likely to receive thermoregulation care in accordance with our guidelines. With the onset of PDSA cycle 2, we transitioned to extracting data from the electronic chart. Additionally, there were several lapses in data collection due to transitions to other quality improvement efforts; therefore, we cannot comment on trends that may have occurred during these time periods. Data collection has been continuous from March of 2018 onward. Finally, as with most quality improvement studies, this is a single-center study, and results are dependent on the character and context of our institution.

We achieved and sustained our aim of a hypothermia rate of <10% for all inborn admissions. We feel that expansion of chemical heat mattresses and polyethylene bags beyond extremely preterm infants is important to improving hypothermia rates. Additionally, the education and awareness of guidelines is vital. Although extremely premature infants continue to be at risk for developing hypothermia, older infants and those with prenatally identified congenital anomalies represent a previously understudied group of infants who would benefit from targeted thermoregulation efforts.

We thank the Children’s Wisconsin–based Excellence in Quality Improvement Program and, in particular, our quality improvement mentor Mary Beth Miranda. We are grateful for our collaborative relationship with the Froedtert Birth Center and, in particular, Allison Bursiek, the birth center nurse manager.