Dexmedetomidine During Therapeutic Hypothermia: A Multicenter Quality Initiative Free

This QI initiative was conceptualized and implemented at 2 level IV NICUs that routinely perform TH for HIE. One is a 54-bed NICU in the eastern United States and the other is a 150-bed NICU in the Midwestern United States; both are major regional TH referral centers. Historically, both centers have used opioids (fentanyl and morphine, respectively) for sedation and comfort during TH.

We organized a multidisciplinary team to identify key drivers in this QI initiative aiming to decrease cumulative opioid exposure during TH (Fig 1). Our primary intervention was the creation and implementation of a standardized sedation algorithm using dexmedetomidine as the first-line agent, with dosing and titration/discontinuation guidelines based on sedation level, HR, and clinical assessment of perfusion (Fig 2). Education regarding the practice change including medication indications and side effects was provided to faculty, fellows, residents, advanced practice practitioners, and bedside nurses. The algorithm (Fig 2) was posted at the bedside of any patient who was undergoing TH. Sedation level, medications, and dose titrations were reviewed daily on rounds. If dexmedetomidine was discontinued because of low HR early in the TH course, providers could reinitiate at a later time if clinically indicated once the HR had stabilized at a rate >70 beats per minute and with no clinical signs of decreased perfusion. Other interventions included emphasizing nonpharmacologic interventions such as encouraging parents to hold their infant while on the cooling blanket and adjusting bedside monitor bradycardia alarms to a threshold of <70 beats per minute instead of the standard 100 beats per minute. Order sets in the electronic medical record (EMR) were created or modified to facilitate consistency between providers and included dexmedetomidine as the first-line sedative agent with dose titration guidance, and opioid (fentanyl or morphine) as needed on the basis of standardized clinical assessment.

Baseline data were collected by retrospective review of the EMR for the 2 years before QI implementation (pre-QI period): October 1, 2018, to September 30, 2020. A preparation and education period was conducted for 1 month before implementation (October 2020). The QI initiative was started November 1, 2020, and continued through November 1, 2022. During the QI period, patient charts were reviewed for the collection of clinical data and medication dosing. Plan-Do-Study-Act (PDSA) cycles were conducted at ∼6-month intervals to identify barriers and optimize the intervention at both centers.

We measured the process by percentage adherence to the sedation guideline, including initiation of dexmedetomidine as the first-line agent and adherence to the recommended starting dose of 0.3 mcg/kg per hour. The primary outcome measure was mean cumulative opioid dose throughout TH. Each institution had a different standard opioid (fentanyl and morphine); therefore, cumulative exposures were measured separately. We excluded opioid boluses that were used in the setting of a procedure such as intubation or peripherally inserted central catheter placement. The mean cumulative exposure during TH for each 5 sequential patients was calculated. The primary balancing measure was need for discontinuation of dexmedetomidine because of bradycardia (sustained HR <70 beats per minute for at least 30 minutes), because this was previously a limitation to dexmedetomidine use in this population.¹⁴  We collected 6 additional balancing measures: (1) requirement for invasive mechanical ventilation, (2) unplanned extubation as an indicator of inadequate sedation, (3) requirement for vasopressor support, (4) time to full enteral feeds, (5) duration of central line, and (6) length of hospital stay. Control charts were generated for length of stay, time to full enteral feedings, and central line duration, with each point representing the mean for every 5 sequential patients. Notably, with respect to feeding and central line duration, trophic volume feeds (20 mL/kg per day) were started during TH when clinically indicated (hemodynamically stable and resolved acidosis) and incrementally increased after rewarming by 30 mL/kg per day as tolerated.

All neonates with HIE undergoing TH were screened for inclusion in the analysis. We excluded infants who died during TH and those with major congenital anomalies or documented neonatal opioid withdrawal syndrome (NOWS). We also excluded infants with severe respiratory failure who required high-level sedation, defined as high frequency ventilation (HFV), inhaled nitric oxide (iNO), or extracorporeal membrane oxygenation (ECMO).

Both participating NICUs prospectively archive physiologic data from bedside monitors (CareScape B650, GE HealthCare, Chicago, IL; MP70 or MX800, Philips Medical, Andover, MA) using a common platform (BedmasterEX, Hill-Rom, Chicago, IL). We used these data to assess the impact of sedation on HR patterns. Hourly mean HR was calculated for hours 0 to 120 after birth, to capture the entire cooling, rewarming, and stabilization period. Infants were classified in groups by exposure (fentanyl, morphine, dexmedetomidine) and group averaged. All data were processed using a custom script written in MATLAB (The MathWorks Inc., Natick, MA).

Median (interquartile range) or number (%) is reported for clinical characteristics unless otherwise stated. Clinical characteristics were compared using Mann–Whitney U and Fisher’s exact tests for continuous and categorical variables, respectively. We used QI Macros software (KnowWare International, Inc., Denver, CO) to create statistical process control charts to examine the impact of the intervention over time and to evaluate for special cause variation during this initiative.

This QI was developed and reviewed by a multidisciplinary team with close monitoring for adverse effects. The use of these sedation medications is standard of care.¹⁵  Data collection was approved by the institutional review boards at both institutions.

In the pre-QI period from 2018 to 2020, there were 177 neonates who underwent TH at the 2 institutions, with 154 neonates meeting inclusion criteria and serving as the baseline cohort (51 at 1 center and 103 at the other). Twenty-three neonates were excluded from the pre-QI cohort for ECMO (3), iNO or HFV (13), redirection of care during TH (5), or NOWS (2).

During the QI period from November 2020 to November 2022, 173 neonates received TH and resulted in a total of 135 neonates included in the QI cohort (52 at 1 institution and 83 at the other, Supplemental Fig 5). Thirty-eight were excluded from analysis because of need for ECMO (3), iNO or HFV (24), major congenital anomaly (3), redirection of care during TH (6), or NOWS (2).

Demographics and clinical characteristics were similar between the pre- and post-QI periods (Table 1). Infants’ birth weights were greater in the QI cohort with a median of 3350 g (2920, 3645), and 235 g greater than in the pre-QI cohort (P = .048). Similar Apgar scores and degree of acidosis were seen in both cohorts.

Our primary intervention was implementation of the standardized sedation guideline, which was updated on the basis of PDSA cycle learnings. On the basis of the results of the first PDSA cycle, the emphasis of the second PDSA cycle was education on the standard starting dose of dexmedetomidine and titration in increments of 0.1 mcg/kg per hour. Education also included emphasis on maximizing nonpharmacologic comfort measures. Changes were made to the EMR dexmedetomidine order set to reflect these recommendations and improve compliance. The third PDSA cycle revealed the need for ongoing education of new groups of providers rotating through the NICU, specifically regarding management of bradycardia in an otherwise clinically stable patient. In addition, the importance of discussing sedation level, dexmedetomidine dose, and possible side effects on daily rounds was emphasized. The fourth PDSA cycle emphasized the importance of weaning or discontinuing dexmedetomidine at the start of rewarming.

Compliance with dexmedetomidine initiation during the QI period was high (99%). The only patient who did not receive dexmedetomidine had severe HIE and was not started on any continuous sedation to evaluate neurologic status more accurately. The process measure of adherence to the predetermined dexmedetomidine start dose of 0.3 mcg/kg per hour increased from 70% in the first year to 91% by the end of year 2. The patients who deviated from the protocol had their dexmedetomidine infusion started at a lower dose of 0.2 mcg/kg per hour. The median maximum dose of dexmedetomidine during TH was 0.3 mcg/kg per hour (0.3, 0.4 mcg/kg per hour). At the completion of TH, the median dexmedetomidine dose was 0.2 mcg/kg per hour (0.2, 0.3 mcg/kg per hour).

Our primary outcome of cumulative opioid exposure during TH significantly decreased at both institutions after algorithm introduction (Fig 3). Fentanyl use decreased from a mean cumulative exposure of 30 mcg/kg (SD 27.4) to 1.7 mcg/kg (SD 5.9). Morphine decreased from a mean cumulative exposure of 648 mcg/kg (SD 211) to 85 mcg/kg (SD 113). Over the 2-year QI period, cumulative fentanyl exposure decreased by 94% and morphine decreased by 87%. Control charts revealed a special cause variation after guideline implementation for both fentanyl and morphine groups, for which the center line was shifted. There was sustained control around the new mean center line for the remainder of the initiative. Two patients at the baseline fentanyl unit were treated with morphine infusion outside of usual practice and were included in the morphine control chart, creating the special cause variation at the beginning of cycle 3 in Fig 3. These 2 patients were initially started on dexmedetomidine, but had it discontinued because of bradycardia and subsequently had a morphine infusion started (as opposed to fentanyl) because of provider preference. Both patients continued on low-dose morphine infusions at 0.02 mg/kg per hour throughout TH. Excluding this special cause variation, the mean cumulative morphine exposure in the QI period was 54 mcg/kg (SD 44.5) and was a decrease of 92% compared with pre-QI. The use of continuous opioid infusions decreased from 90% of the pre-QI cohort to 4% after QI implementation. The median number of nonprocedural opioid boluses decreased from 2 (range 0–11) to 0 (range 0–7). Overall, the Specific, Measurable, Achievable, Relevant, Time-Bound aim was met by the end of the 2-year QI period with a decreased cumulative opioid exposure during TH by >90% at both institutions.

Dexmedetomidine was discontinued because of low HR in 13 of 135 patients (9.6%) during the QI period, with a median maximum dose of 0.4 mcg/kg per hour (0.2, 0.4 mcg/kg per hour) and despite weaning the dose per the algorithm before discontinuation. Of note, none of these patients displayed clinical signs of decreased perfusion with bradycardia. Six of these patients received intermittent opioid boluses after discontinuation of dexmedetomidine and 5 were started on a continuous opioid infusion. Discontinuation of dexmedetomidine was temporary in 4 of the 13 patients at median age of 34 hours (range 24–43) and was restarted a median of 27 hours later (range 15–47) without recurrence of bradycardia.

A total of 202 infants (70%) from the pre-QI and QI periods had physiologic recording data including 23 in the fentanyl group (45%), 63 in the morphine group (61%), and 116 in the dexmedetomidine group (86%). The mean HR nadir for all 3 groups occurred between 24 and 30 hours after birth. The dexmedetomidine group had a lower nadir of HR (93 beats per minute) when compared with the fentanyl (97 beats per minute) and morphine (100 beats per minute) groups and reached the nadir at a later time point (30 hours after birth). Recovery of the HR to a pre-TH, presedation baseline occurred by the end of the monitoring period for all groups (Supplemental Fig 6).

Length of stay was the only balancing measure for which a change was identified. Special cause variation leading to change in mean was noted with PDSA cycle 2, shifting the mean length of stay from 18 days in the pre-QI period to 14 days in the QI period (Fig 4). Other balancing measures remained stable between the pre-QI and QI periods. Both groups had an average time to full enteral feeds of 12 days and central line duration of 6 days (Fig 4). There was no significant difference in the percentage of patients requiring invasive mechanical ventilation, and there were no unplanned extubation events for included infants at either institution before or during the QI periods. Vasopressor use was rare in both cohorts and did not show significant change over time.

In this 2-center QI initiative, we were able to successfully decrease opioid exposure during TH without alteration of important balancing measures by implementing a standardized sedation algorithm utilizing dexmedetomidine as the first-line sedative agent. The use of continuous opioid infusions was rare in the QI period compared with nearly universal use in the pre-QI period. Dexmedetomidine was well tolerated by most patients. Significant bradycardia (sustained HR <70 beats per minute) occurred in about 10% of patients; however, this was not associated with clinical symptoms of compromised perfusion. There was a decreased average length of hospitalization noted in the QI period. We did not observe differences in other short-term clinical outcomes with this practice change.

QI initiatives have been used in neonatal intensive care to reduce opioid exposure in vulnerable neonates. Multicenter projects have successfully optimized nonpharmacologic comfort measures and standardized pain assessments in the overall NICU population.^16–18  The authors of the current report coauthored and strictly adhered to the Newborn Brain Society Guideline on the management of comfort and sedation in neonates treated with TH, including optimization of nonpharmacologic comfort care.¹⁵  Consequently, our joint QI initiative focused on reducing opioid exposure through optimization of alternative pharmacotherapeutic agents. By focusing on a specific subpopulation of neonates with relatively homogenous physiology, the success of our initiative was higher than comparable projects.¹⁹  The high level of compliance in the QI cohort demonstrates that neonatal intensive care is amenable to optimization through targeted guidelines and highlights the challenge of unitwide QI research given the intrinsic diversity of pathophysiology navigated by NICU providers.

Preclinical data support that adequate sedation during TH optimizes its beneficial effects, and dexmedetomidine may represent an alternative to opioids.^5,20  Several studies have shown an association between dexmedetomidine use and decreased need for other sedative medications, decreased time on mechanical ventilation, and decreased time to full enteral feeds when compared with sedation with opioids.^8,9  Using QI methodology, we were able to create and implement a standardized sedation algorithm to successfully introduce dexmedetomidine as the first-line sedation agent in patients undergoing TH, significantly decreasing cumulative opioid exposure for these high-risk patients. In our cohort, dexmedetomidine alone typically provided adequate sedation at low doses. Opioid boluses, if used, were given early during dexmedetomidine titration. Pharmacokinetic studies of dexmedetomidine during TH indicate slower rise in plasma concentrations and longer time to plasma clearance, consistent with our observation of higher sedation needs at the beginning of cooling and ability to wean toward the completion of cooling.^21,22  No patients required both an opioid infusion and dexmedetomidine simultaneously. We did not observe a benefit of dexmedetomidine with respect to the requirement for invasive mechanical ventilation, requirement for vasopressor support, or the time required to achieve full enteral feedings, but did observe a decreased average length of stay.

Recent reports support the safety of dexmedetomidine use during TH, with bradycardia being the most commonly documented side effect.^8,14,20  Although severe bradycardia occurred in a small number of patients, it resolved with discontinuation of dexmedetomidine, and no patients had adverse clinical effects related to bradycardia. An HR of 70 beats per minute or above was tolerated in our population during TH if the patient was otherwise clinically stable, and an isolated low HR may not be an indication to decrease or discontinue dexmedetomidine infusion. In our cohort, we noted that, for most patients, the HR nadir occurred within the first 30 hours of cooling (Supplemental Fig 6). Some patients who did not initially tolerate dexmedetomidine because of low HR were able to tolerate it when it was reintroduced at a later time during TH.

Although dexmedetomidine appears to be a safe alternative to opioids with regard to acute adverse effects and short-term clinical outcomes, neurodevelopmental follow-up was outside the scope of this study. Additionally, we did not include the full range of patients requiring TH, particularly the sickest infants with respiratory failure or those with NOWS, because they often require additional sedation, specifically with opioids, to treat comorbid conditions. Further studies are needed to assess long-term safety and efficacy of dexmedetomidine use in the diverse population of neonates undergoing TH.

This collaborative QI initiative demonstrates the safe implementation of dexmedetomidine as an opioid-sparing agent in neonates treated with TH. Multidisciplinary design and follow-up ensured a high rate of guideline compliance, reinforcing this vital framework for all practice changes in neonatal intensive care. Presentation of a safe sedation guideline with high prescriber compliance provides 1 potential arm for a randomized trial. The relative safety of both dexmedetomidine and opioids in this population and the uncertain impact of these agents on neurodevelopmental outcomes highlight the vital nature of randomized controlled trials comparing sedation regimens and evaluating long-term outcomes of neonates with HIE undergoing TH.