Midazolam is a benzodiazepine sedative used in NICUs. Because benzodiazepine’s effects include respiratory depression and potential detrimental developmental effects, minimizing exposure could benefit neonates. Dexmedetomidine is routinely used for sedation in older pediatric populations. We implemented a quality improvement initiative with the aim of decreasing midazolam infusions by 20% through use of dexmedetomidine.
A multidisciplinary committee created a sedation guideline that included standardized dexmedetomidine dosing escalation and weaning. Baseline data collection occurred from January 2015 to February 2018, with intervention from March 2018 to December 2019. Percentage of sedation episodes with dexmedetomidine initiated was followed as a process measure. Outcomes measures were percentage of eligible infants receiving midazolam infusions and midazolam-free days per sedation episode. Bradycardia with dexmedetomidine, unplanned extubation rates, and morphine dosage were monitored as balancing measures.
Our study included 434 episodes of sedation in 386 patients. Dexmedetomidine initiation increased from 18% to 49%. The intervention was associated with a significant reduction in midazolam initiation by 30%, from 95% to 65%, with special cause variation on statistical process control chart analysis. Midazolam-free days per sedation episode increased from 0.3 to 2.2 days, and patients receiving dexmedetomidine had lower midazolam doses (1.3 mg/kg per day versus 2.2 mg/kg per day, P = 5.97 × 10−04). Bradycardia requiring discontinuation of dexmedetomidine, unplanned extubation rates, and morphine doses were unchanged.
Implementation of a quality improvement initiative was successful in reducing the percentage of patients receiving midazolam infusions and increased midazolam-free days per sedation episode, revealing an overall reduction in benzodiazepine exposure while maintaining adequate sedation.
Adequate sedation of neonates requiring mechanical ventilation may reduce stress, enhance ventilator synchrony, and decrease adverse events such as unplanned extubation. Benzodiazepines, along with opioids, are commonly used in NICUs as a sedative. Animal studies on the use of anesthetics and sedatives, including benzodiazepines, have raised concerns that lengthy or repeated exposures to these medications may negatively affect neurodevelopment in this vulnerable population.1–4 Clinical trials in humans are limited and have varied results, although some have revealed lower cognitive scores that correlate with cumulative benzodiazepine dose exposure.5–7 Balancing the need for sedation with potential neurodevelopmental concerns warrants minimizing exposure to these agents when able and seeking alternatives.
Dexmedetomidine, a highly selective, centrally acting α2 adrenergic agonist, has been revealed to provide adequate sedation with minimal respiratory depressant effects compared with other sedating agents. Dexmedetomidine was approved by the Food and Drug Administration for adult use in 1999, although pediatric use remains off-label.8,9 The activation of α2 adrenergic receptors in the medullary vasomotor center leads to reduction in norepinephrine turnover and sympathetic nervous system signaling from the locus coeruleus, leading to increased endogenous GABAergic activity, which causes sedation.10–13 Separate release of substance P from the dorsal horn of the spinal cord leads to analgesia and can potentiate the effect of opioids.14
Several trials have reported the successful use of dexmedetomidine in pediatric settings.15–18 Additional reports have described the use of dexmedetomidine in preterm neonates and infants with hypoxic ischemic encephalopathy without significant complication.19–26 In particular, dexmedetomidine has promise as a neuroprotective sedative agent that prevents neuronal apoptosis,27–33 in contrast with the potential negative effects of benzodiazepines on neuronal migration.34 Previous studies have revealed the ability of dexmedetomidine to reduce postoperative patient exposure to sedative agents, such as midazolam and morphine, among patients with congenital heart disease.35
With the goal of reducing benzodiazepine exposure among infants requiring sedation in our NICU, we developed a sedation guideline (Supplemental Information) and implemented a quality improvement project in March 2018 with the goal of reducing benzodiazepine exposure during sedation episodes from 95% to 75% by December 2019.
Methods
Context
The Boston Children’s Hospital NICU is an urban 24-bed tertiary and quaternary care NICU in Boston, MA that cares for ∼650 patients annually, with 80% admitted to the medical service and 20% admitted to the surgical service. All patients are outborn.
Intervention
We developed a multidisciplinary team that included a clinical nurse specialist, nurses, pharmacists, and physicians and undertook a quality improvement initiative to decrease benzodiazepine usage in our NICU through multiple interventions: (1) creation of a guideline describing use of dexmedetomidine in our NICU patient population (November 2017–February 2018); (2) multidisciplinary educational launch (March 2018); (3) Educational outreach (March 2018–November 2018); (4) presentation at the NICU Steering Committee (May 2018); (5) daily rounds discussions on mechanically ventilated patients to discuss best sedation plan with consideration of dexmedetomidine as an initial agent if eligible (July 2019); and (6) presentation at the NICU quality improvement meeting (October 2019). The guideline also included guidance about opioids and acetaminophen use for analgesia. For sedative agents, the guideline focused on benzodiazepines and dexmedetomidine and did not include alternative sedatives. Patients were considered eligible for dexmedetomidine if they were at least 35 weeks’ postmenstrual age, anticipated <14 days of sedation, and needed sedation for mechanical ventilation. Infants <35 weeks’ postmenstrual age were excluded because benzodiazepines were not routinely used at that age. Short-term sedation was covered by the guideline as a first population for targeted recommendation of dexmedetomidine. Exclusion criteria included hypoxic ischemic encephalopathy, potential need for extracorporeal membrane oxygenation, uncontrolled hypotension, sustained bradycardia, and hepatic failure.
Study of the Intervention
A multidisciplinary committee developed a dexmedetomidine guideline for NICU patients. Baseline data were collected from January 2015 to February 2018. Our quality improvement interventions began in March 2018. Dosage and usage data were abstracted from the electronic medical record through December 2019.
Measures
Sedation episodes were defined as patients receiving continuous infusions of midazolam or dexmedetomidine, as identified in the electronic medical record, with maximum duration of 20 days and individual sedation episodes separated by a minimum of 3 days between exposures. The process measure was the percentage of eligible patients who had dexmedetomidine infusions initiated. The outcome measures were the percentage of patients eligible to receive dexmedetomidine who received midazolam infusions and total days without midazolam per sedation episode. The multidisciplinary team determined several key potential unintended effects of increased dexmedetomidine infusions and chose the following balancing measures to follow: (1) the number of dexmedetomidine infusions that were discontinued within 24 hours of initiation, (2) monthly rates of unplanned extubations, and (3) morphine usage among infants who received sedation. All cases of early dexmedetomidine discontinuation were investigated for evidence of intolerance, ineffectiveness, or inability to maintain separate intravenous access.
Analysis
Comparison of cohort characteristics was completed by using t tests. To determine if special cause variation was found after changes, we developed statistical process charts (SPCs) using Chartrunner software to make c and p charts. Statistics are presented as percentage or mean ± SD, and patient-level values before and after intervention implementation periods were compared by Fischer’s exact test.
Ethical Considerations
Preparation of guidelines regarding patient care within the NICU is handled by multidisciplinary teams and incorporates evidence-based medicine, feedback from local providers, and involvement of relevant consultant services. Implemented guidelines do not contain practices out of the scope of ordinary practice used in the NICU. Institutional review board approval was obtained to collect patient data (IRB-P00020194).
Results
Sedation Episodes
During the study period, there were 434 episodes of sedation among 386 patients (Table 1). The mean duration of sedation episodes was 7.2 days, and the duration of sedation episodes was not significantly different between the preintervention and postintervention periods (7.6 days and 6.6 days, respectively; t test P value = .08). Although no difference was seen in the pre- and postintervention analysis, duration of sedation episodes did reveal a significant increase in August 2017 to 8.9 days from a baseline of 6.2 days, followed by a decrease in January 2019 to 6.3 days with special cause variation on the SPC i-chart (Supplemental Fig 4). Mean gestational age, birth weight, weight at sedation episode, and length of stay were comparable between the 2 time periods. Postmenstrual age at the time of the sedation episode was lower in the postintervention period by ∼10 days.
Patient Demographics
| Measure | Preintervention January 2015–February 2018 Mean (Range) | Postintervention March 2018–December 2019 Mean (Range) | P |
|---|---|---|---|
| No. episodes | 261 | 173 | .28 |
| Mean episode duration (range), d | 7.56 (1–42) | 6.57 (1–40) | .08 |
| Mean birth wt (range), g | 2357 (370–4370) | 2454 (415–4820) | .35 |
| Mean wt at episode (range), g | 3897 (1500–8595) | 3861 (1500–9360) | .78 |
| Mean gestational age at birth (range), wk | 35 1/7 (23 0/7–42 0/7) | 35 1/7 (23 0/7–42 0/7) | .92 |
| Mean postmenstrual age at episode (range), wk | 46 0/7 (34 1/7–77 6/7) | 44 1/7 (34 0/7–70 2/7) | .03 |
| Mean length of stay (range), d | 32 (2–170) | 28 (1–191) | .30 |
| Indication for sedation, n (%) | |||
| Airway procedure | 60 (23) | 25 (14) | >.99 |
| Cardiac procedure | 5 (2) | 7 (4) | >.99 |
| Gastrointestinal procedure | 66 (25) | 49 (28) | >.99 |
| Other procedure | 18 (7) | 70 (40) | >.99 |
| Mechanical ventilation | 112 (43) | 22 (13) | >.99 |
| Measure | Preintervention January 2015–February 2018 Mean (Range) | Postintervention March 2018–December 2019 Mean (Range) | P |
|---|---|---|---|
| No. episodes | 261 | 173 | .28 |
| Mean episode duration (range), d | 7.56 (1–42) | 6.57 (1–40) | .08 |
| Mean birth wt (range), g | 2357 (370–4370) | 2454 (415–4820) | .35 |
| Mean wt at episode (range), g | 3897 (1500–8595) | 3861 (1500–9360) | .78 |
| Mean gestational age at birth (range), wk | 35 1/7 (23 0/7–42 0/7) | 35 1/7 (23 0/7–42 0/7) | .92 |
| Mean postmenstrual age at episode (range), wk | 46 0/7 (34 1/7–77 6/7) | 44 1/7 (34 0/7–70 2/7) | .03 |
| Mean length of stay (range), d | 32 (2–170) | 28 (1–191) | .30 |
| Indication for sedation, n (%) | |||
| Airway procedure | 60 (23) | 25 (14) | >.99 |
| Cardiac procedure | 5 (2) | 7 (4) | >.99 |
| Gastrointestinal procedure | 66 (25) | 49 (28) | >.99 |
| Other procedure | 18 (7) | 70 (40) | >.99 |
| Mechanical ventilation | 112 (43) | 22 (13) | >.99 |
Process Measure
The monthly proportion of infants receiving any dexmedetomidine increased after intervention from 18% to 49% with demonstration of special cause variation (Fig 1). In addition, the monthly percentage of infants receiving only dexmedetomidine without midazolam increased from 4% to 33% after quality improvement interventions (Supplemental Fig 5).
The process measure reveals an increase in the monthly proportion of sedation episodes in which infants received any dexmedetomidine. Set 1: upper confidence limit (UCL), 60.81; mean, 17.49; lower confidence limit (LCL), 0. Set 2: UCL, 100; mean, 49.43; LCL, 0.
The process measure reveals an increase in the monthly proportion of sedation episodes in which infants received any dexmedetomidine. Set 1: upper confidence limit (UCL), 60.81; mean, 17.49; lower confidence limit (LCL), 0. Set 2: UCL, 100; mean, 49.43; LCL, 0.
Outcome Measures
The monthly proportion of infants receiving any midazolam significantly decreased and revealed special cause variation on SPC chart analysis and decreased from 95% to 65% after the project start (Fig 2). In addition, the percentage of patients receiving only midazolam without dexmedetomidine decreased from 85% to 51% after implementation of the intervention (Supplemental Fig 6). There was a significant increase in midazolam-free days per sedation episode, from 0.3 to 2.2 days from the preintervention to postintervention period (Fig 3), despite there being no differences in the mean duration of sedation episodes.
The outcome measure reveals a decrease in the monthly proportion of sedation episodes in which infants received any midazolam. Set 1: upper confidence limit (UCL), 100; mean, 95.06; lower confidence limit (LCL), 70.34. Set 2: UCL, 100; mean, 64.77; LCL, 14.11.
The outcome measure reveals a decrease in the monthly proportion of sedation episodes in which infants received any midazolam. Set 1: upper confidence limit (UCL), 100; mean, 95.06; lower confidence limit (LCL), 70.34. Set 2: UCL, 100; mean, 64.77; LCL, 14.11.
The outcome measure reveals an increase in the mean monthly midazolam-free days per sedation episode from 0.3 to 2.2 days. Set 1: upper confidence limit (UCL), 0.85; mean, 0.28; lower confidence limit (LCL), 0. Set 2:UCL, 4.8; mean, 2.25; LCL, 0.
The outcome measure reveals an increase in the mean monthly midazolam-free days per sedation episode from 0.3 to 2.2 days. Set 1: upper confidence limit (UCL), 0.85; mean, 0.28; lower confidence limit (LCL), 0. Set 2:UCL, 4.8; mean, 2.25; LCL, 0.
Reduced Benzodiazepine Dosage With Dexmedetomidine
Finally, throughout the study period, infants receiving dexmedetomidine infusions during a sedation episode received less daily midazolam compared with those not receiving dexmedetomidine (1.3 mg/kg per day versus 2.2 mg/kg per day, P = 5.97 × 10−04).
Balancing Measures
Reported side effects of dexmedetomidine include bradycardia and hypotension.36–38 Intolerance of dexmedetomidine due to hypotension, bradycardia or inability to maintain separate venous access was monitored as a balancing measure because dexmedetomidine needs to be given via a dedicated intravenous line per our unit protocol. In the preintervention period, 4 episodes were associated with <24 hours of dexmedetomidine use, whereas only 1 episode was associated with <24 hours of dexmedetomidine in the postintervention period. Three infants had their dexmedetomidine discontinued for inadequate sedation and 1 for episodes of bradycardia without hemodynamic compromise in each period. Of note, both bradycardia episodes did not resolve after discontinuation of dexmedetomidine. Unplanned extubation rates did not significantly change between the baseline and project period (0.99–0.58 per 100 ventilation days, P = .06). Days of morphine exposure (4.2 vs 3.1, P = .06) and morphine dose (1.0 mg/kg per day versus 1.3 mg/kg per day, P = .14) did not vary.
Discussion
We have demonstrated the effective use of dexmedetomidine for sedation among neonates associated with decreased exposure to midazolam. With quality improvement interventions, including introduction of a standardized protocol for dexmedetomidine use and educational outreach, we decreased the proportion of infants who received any midazolam infusions and increased midazolam-free days among infants in the setting of no difference in length of sedation episode between periods. The dexmedetomidine was generally well tolerated, with only 1 infant having an infusion stopped within 24 hours of initiation after standardized dosing guidelines were implemented.
Summary
We have demonstrated that routine use of dexmedetomidine as a sedative among NICU patients >35 weeks’ postmenstrual age resulted in decreased midazolam exposure. There was no increase in adverse events after intervention; with standardized dosing recommendations, there were no episodes of dexmedetomidine discontinuation for lack of efficacy.
Interpretation
For sedation of mechanically ventilated neonates, standardized dexmedetomidine use among near term-equivalent NICU patients resulted in decreased midazolam exposure, both in terms of proportions of patients exposed and days of exposure.
Limitations
Study limitations include the modest patient numbers, single-center design, and limited covariate data to compare across the pre- and postintervention groups. Although our process measures revealed improvement in dexmedetomidine use, we only achieved a 30% reduction in midazolam use, indicating the need for future Plan-Do-Study-Act (PDSA) cycles. In particular, we hypothesize that our current practice of requiring dedicated intravascular access for dexmedetomidine is a significant barrier, which could be addressed with additional studies of compatibility with common NICU medications. In addition, some providers were hesitant to increase dexmedetomidine infusion rate per the protocol, leading to increased use of midazolam; this is being addressed with further bedside educational outreach. There was also a decrease in dexmedetomidine use noted in August 2019, a month during which the majority (5 of 7) of patients received sedation during mechanical ventilation rather than peri-procedure, indicating that a future PDSA cycle could focus on this indication. We will also continue to follow duration of sedation episode for further decreases because dexmedetomidine has a favorable, shorter weaning period, and, if future PDSA cycles are successful, important decreases in duration of sedation episode may occur. Future work is needed to determine the generalizability of dexmedetomidine use at lower gestational ages, which fell outside the scope of this project because benzodiazepines are not routinely used <35 weeks’ postmenstrual age in our unit. Finally, although decreasing midazolam exposure was the goal of the current study, further work is needed to determine if efforts to use dexmedetomidine instead of midazolam will improve neurodevelopment among neonates.
Conclusions
Standardized use of dexmedetomidine for sedation among NICU patients can reduce midazolam exposure.
FUNDING: No external funding.
Drs Morton and Leeman and Ms Labrecque conceptualized and designed the study, coordinated and supervised data collection, conducted the initial analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Mr Moline designed the data collection instruments, collected data, conducted the initial analyses, and reviewed and revised the manuscript; Dr Hansen conceptualized and designed the study, supervised data collection, and critically reviewed and revised the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Comments