To improve outcomes in infants with neonatal opioid withdrawal syndrome (NOWS) admitted to NICU by implementing a quality improvement (QI) initiative incorporating “eat, sleep, console” (ESC) as a withdrawal evaluation tool and promotion of nonpharmacological interventions. Secondarily, we evaluated the impact of the coronavirus disease 2019 pandemic on QI initiative and outcomes.
We included infants born ≥ 36 weeks gestation and admitted to NICU with a primary diagnosis of NOWS between December 2017 and February 2021. (preintervention; December 2017–January 2019, postintervention; February 2019–February 2021). We compared cumulative dose, duration of opioid treatment, and length of stay (LOS) as our primary outcomes.
The average duration of opioid treatment decreased from 18.6 days in the preimplementation cohort (n = 36) to 1.5 days in the first-year postimplementation (n = 44) with a reduction in cumulative opioid dose from 5.8 mg/kg to 0.6 mg/kg and decrease in the proportion of infants treated with opioids from 94.2% to 41.1%. Similarly, the average LOS decreased from 26.6 to 7.6 days. In the second-year postimplementation during the coronavirus disease 2019 pandemic (n = 24), there was an increase in average opioid treatment duration and LOS to 5.1 and 12.3 days respectively, but cumulative opioid dose (0.8 mg/kg) remained significantly lower than the preimplementation cohort.
ESC-based quality improvement initiative led to a significant decrease in LOS and opioid pharmacotherapy in infants with NOWS in NICU setting. Despite the impact of the pandemic, some of the gains were sustained with adaptation to ESC QI initiative.
Neonatal opioid withdrawal syndrome (NOWS) is a constellation of withdrawal symptoms that occur in the first few days after birth in infants with in-utero opioid exposure. The incidence of NOWS has increased by 82% in the last decade in the United States with nearly 80 infants diagnosed every day.1 NOWS is associated with prolonged hospital stay and increased healthcare associated cost.1 At our institution, around 220 infants are delivered annually with intrauterine polysubstance exposure with around 50 admissions to the NICU with NOWS.
Finnegan Neonatal Abstinence Scoring Systems (FNASS) and its modified versions are symptom-based scoring system used for evaluation of withdrawal symptoms in infants with NOWS.2,3 Limitations of FNASS include lack of objectivity, consistency, and poor psychometric rating. Treating infants with NOWS solely based on symptoms may result in increased and prolonged pharmacological treatment.4 Recently, a functional based “Eat Sleep and Console” (ESC) approach has been evaluated as an assessment tool in management of infants with NOWS based on their ability to feed, sleep, and be consoled.4,5 It has been shown to correlate with modified FNASS to detect withdrawal symptoms.6 ESC-based quality improvement (QI) initiatives in multiple settings, including community hospitals, children’s hospitals, and state-wide hospital consortiums, have been shown to improve outcomes in infants with NOWS with decrease in the use of opioids, hospital length of stay (LOS), and hospital costs.7–15
We implemented a comprehensive QI initiative in February 2019 in the NICU with emphasis on promotion of functional-based assessment and use of nonpharmacological intervention in infants with NOWS. Our specific aim was to decrease pharmacological treatment and LOS by 50% in 12 months. Following implementation, we compared outcome measures between preintervention and ESC cohorts. During the pandemic, some resources were affected because of changes in parental visitation policies and resource reallocation, after which we made changes to our QI bundle and evaluated its impact.16
Methods
Settings and scope of problem: our institution is a large referral hospital with a 65-bed level 3 NICU with over 4000 deliveries annually. Around 220 infants are born to mothers with polysubstance use during pregnancy, of which around 50 infants are admitted with symptoms of NOWS. Our NICU is located over 2 floors with an open bay layout on 1 floor and 10 single and double bed private rooms on the other floor. In the baseline preintervention period, neonates with in-utero polysubstance exposure were admitted first to a mother and baby couplet care nursery and monitored for signs of drug withdrawal using modified FNASS assessment tool for 3 to 5 days based on type of maternal substance use. Any infant with severe withdrawal symptoms (>2 scores of 8 or 1 score of 12) would be transferred to the NICU for further monitoring and consideration of pharmacological treatment. On transfer to NICU, infants were continued to be scored based on modified FNASS assessment tool with treatment considered for 3 consecutive scores of 8 or 2 scores above 12. Scheduled morphine or methadone was considered as first choice for NOWS and clonidine or phenobarbitone used as adjunct therapy based on clinician’s preference. There was no protocol for pharmacological treatment during the preintervention period and dose was managed and weaned based on FNASS scores and clinician’s discretion. The infants were monitored for 48 to 72 hours post cessation of pharmacological treatment before being discharged from NICU.
Development and Implementation of ESC-based QI Initiative
We implemented a multidisciplinary QI initiative in February 2019. Our specific aim was to decrease pharmacological treatment and LOS by 50% in 12 months. The key drivers for our initiative included: (1) transition to ESC as a functional based withdrawal assessment tool from FNASS, (2) promotion of nonpharmacological interventions; and (3) decreasing the use of scheduled pharmacotherapy. Interventions identified for the key drivers are listed in Fig 1 and the timeline of key interventions are summarized in Supplemental Table 2.
Patients and Measures
We identified all infants born during the study period between December 2017 and February 2021 with history of in-utero substance and drugs of addiction exposure using International Classification of Diseases, 10th Revision codes; P04.49, P96.1 and P04. 9.17 Electronic medical records of identified infants were reviewed. We included only infants born at ≥36 weeks gestation with in-utero opioid exposure and requiring admission to the NICU with a primary diagnosis of NOWS. We excluded any infants with NOWS admitted with other diagnoses, including respiratory distress, concern for infection, and feeding difficulty not associated with NOWS. The primary outcomes were opioid treatment including duration of treatment (day of the first dose to the last day of receiving opioids), total cumulative dose, and LOS during the admission. We included treatment with adjunctive pharmacologic agents (phenobarbital or clonidine), hospital costs, breast feeding and discharge home with parents as secondary outcome measures. We identified seizures, desaturations, and apnea episodes attributed to opioid treatment as safety measures. We collected data on primary outcomes prospectively during the implementation of ESC program. We retrospectively collected baseline maternal and infant characteristics and data on secondary outcome measures from electronic medical records.
Statistical Analysis
For the primary outcomes, we used statistical process control charts to depict average LOS, duration of opioid treatment, cumulative opioid doses, and proportion of infants treated with opioids over the time course of quality improvement process and the study period. Special cause variation in the outcome measures was determined to have occurred when 8 consecutive measurements were noted on either side of the center line on statistical process control charts. For baseline characteristics and secondary outcomes across 3 time periods: preintervention (December 2017–January 2019), ESC (February 2019–February 2020), and the pandemic cohort (March 2020–February 2021), we used independent t tests and Mann-Whitney test for analysis of continuous variables and χ-square and Fischer’s exact tests for analysis of categorical variables. Statistical significance was set at 2-sided P value below .05. We used IBM SPSS Statistics, Version 28.0. Armonk, NY: IBM Corp and QI Macros for the analysis. The study received exemption status as a quality improvement initiative from Dignity health institutional review board (PHXNR-21-500-250-73-21).
Results
A total of 712 infants with intrauterine drugs of addiction or substance exposure were born during the study period, of which 144 infants with history or confirmed opioid exposure were admitted to the neonatal unit. Of these, 104 infants met the inclusion criteria of ≥36 weeks birth gestational age with intrauterine opioid exposure and admission for NOWS with no respiratory distress or other reasons for admission. (Preimplementation cohort: n = 36, ESC cohort: n = 44, and pandemic period cohort: n = 24). There were no differences in maternal age, race and ethnicity, birth weight and gestational age of infants between the 3 cohorts. However, there was a difference in type of opioid exposure with decrease in methadone use in ESC and pandemic cohort and increase in fentanyl use seen in the pandemic cohort compared with preimplementation cohort P = .01(Table 1). There was no difference in maternal use of tobacco or tetrahydrocannabinol use between the cohorts.
Comparison of Maternal and Infant Characteristics and Infant Outcomes
. | Preimplementation Cohort (n = 36) . | ESC Cohort (n = 44) . | Pandemic Cohort (n = 24) . |
---|---|---|---|
Maternal agea | 28.9 (±4.6) | 29.0 (±5.8) | 28.7 (±4.4) |
Race and ethnicity | |||
Whiteb | 23 (62.2) | 34 (77.2) | 15 (62.5) |
Hispanicc | 7 (18.9) | 4 (9) | 2 (8.3) |
Blackc | 3(8.1) | 5 (11.3) | 3 (12.5) |
Native Americanc | 1(2.8) | 1(2.3) | 1 (4.2) |
Asianc | 1(2.8) | 0 | 1(4.2) |
Otherc | 1(2.8) | 0 | 2 (8.4) |
Maternal opioid use | |||
Methadone, n (%)c | 28 (77.8) | 21(47.7)* | 8 (33.3)*** |
Heroin, n (%)c | 16 (44.4) | 12(27.3) | 3 (12.5)*** |
Oxycodone, n (%)c | 4 (11.1) | 7 (15.9) | 3 (12.5) |
Morphine, n (%)c | 2 (5.6) | 0 (0) | 0 (0) |
Buprenorphine, n (%)c | 1 (2.8) | 6 (13.6) | 3 (12.5) |
Fentanyl, n (%)c | 1(2.8) | 4 (9) | 14 (58.3)**,*** |
Others, n (%)c | 1(2.8) | 4 (9) | 2 (8.3) |
Maternal tobaccob | 14 (38.9) | 9 (20.5) | 8 (33.3) |
Maternal THCb | 7(19.4) | 14 (31.8) | 8 (33.3) |
Infant gestational agea | 38.4 (±1.6) | 37.8(±1.3) | 38.5(±1.7) |
Infant wta (grams) | 3095 (±528) | 2996 (±509) | 3118 (±360) |
Infant male genderb | 18 (50) | 27 (61.3) | 13 (54.1) |
Nonopioids, n (%)c | 17 (47.2) | 0 (0)* | 0 (0)*** |
Any maternal milk n (%)c | 7 (19.4) | 17(38.6) | 2 (8.3)** |
Discharge with parents n (%)b | 24 (66.7) | 20 (45.5) | 10 (41.7) |
Hospital stay cost, $, median (IQR)d | $87 277 ($56 005–$143 446) | $24 797 ($16 285–$46 595) * | $71 592 ($49 390–$102 184)** |
. | Preimplementation Cohort (n = 36) . | ESC Cohort (n = 44) . | Pandemic Cohort (n = 24) . |
---|---|---|---|
Maternal agea | 28.9 (±4.6) | 29.0 (±5.8) | 28.7 (±4.4) |
Race and ethnicity | |||
Whiteb | 23 (62.2) | 34 (77.2) | 15 (62.5) |
Hispanicc | 7 (18.9) | 4 (9) | 2 (8.3) |
Blackc | 3(8.1) | 5 (11.3) | 3 (12.5) |
Native Americanc | 1(2.8) | 1(2.3) | 1 (4.2) |
Asianc | 1(2.8) | 0 | 1(4.2) |
Otherc | 1(2.8) | 0 | 2 (8.4) |
Maternal opioid use | |||
Methadone, n (%)c | 28 (77.8) | 21(47.7)* | 8 (33.3)*** |
Heroin, n (%)c | 16 (44.4) | 12(27.3) | 3 (12.5)*** |
Oxycodone, n (%)c | 4 (11.1) | 7 (15.9) | 3 (12.5) |
Morphine, n (%)c | 2 (5.6) | 0 (0) | 0 (0) |
Buprenorphine, n (%)c | 1 (2.8) | 6 (13.6) | 3 (12.5) |
Fentanyl, n (%)c | 1(2.8) | 4 (9) | 14 (58.3)**,*** |
Others, n (%)c | 1(2.8) | 4 (9) | 2 (8.3) |
Maternal tobaccob | 14 (38.9) | 9 (20.5) | 8 (33.3) |
Maternal THCb | 7(19.4) | 14 (31.8) | 8 (33.3) |
Infant gestational agea | 38.4 (±1.6) | 37.8(±1.3) | 38.5(±1.7) |
Infant wta (grams) | 3095 (±528) | 2996 (±509) | 3118 (±360) |
Infant male genderb | 18 (50) | 27 (61.3) | 13 (54.1) |
Nonopioids, n (%)c | 17 (47.2) | 0 (0)* | 0 (0)*** |
Any maternal milk n (%)c | 7 (19.4) | 17(38.6) | 2 (8.3)** |
Discharge with parents n (%)b | 24 (66.7) | 20 (45.5) | 10 (41.7) |
Hospital stay cost, $, median (IQR)d | $87 277 ($56 005–$143 446) | $24 797 ($16 285–$46 595) * | $71 592 ($49 390–$102 184)** |
IQR, interquartile range; THC, tetrahydrocannabinol.
Mean (± SD); sample independent t test.
χ-square test.
Fischer exact test.
Mann-Whitney test.
P value < .05 (preimplementation versus ESC),
P value < .05 ESC and Pandemic period),
P value < .05 (preimplentation and pandemic period).
Primary Outcomes
The average LOS decreased from 26.6 days in the preintervention period to 7.6 days after implementation of QI initiative with special cause variation noted in February 2019 associated with the transition to ESC based QI bundle. Special cause variation was again noted in March 2020 after introduction of visitation restriction with an increase in average LOS to 12.3 days (Fig 2). The average duration of opioid treatment was 18.6 days in the preimplementation period, which decreased to 1.5 days in February 2019 with implementation of ESC and nonpharmacological intervention bundle. The average opioid treatment days increased during the pandemic period to 5.1 days (Fig 3). The average cumulative dose of morphine decreased from a baseline of 5.8 mg/kg to 0.6 mg/kg in February 2019 with an increase to 0.8 mg/kg noted again in the pandemic period (Fig 4). The average proportion of infants treated with any opioids decreased from 94.2% to 41.4% post implementation of ESC and nonpharmacological bundle with special cause variation noted in June 2019. During the pandemic period, the proportion of infants treated with any opioids increased to 79.1%. (Supplemental Figure 5). Multiple data points were noted outside the upper control lines for all outcomes measures with more outliers noted during the preimplementation period for LOS and cumulative opioid dose.
Secondary Outcomes
The percentage of infants treated with adjunct medications, including phenobarbitone and clonidine, decreased from 47.2% to 0 in ESC-cohort. (P < .01) (Table 1). The median hospital cost was $87 277 in the preimplementation cohort, which decreased to $24 797 in the ESC cohort, and an increase was noted in pandemic cohort to $71 592. (P < .05) (Table 1). The rates of maternal milk use decreased in the pandemic cohort when compared with ESC cohort (38.6% vs 8.3%, P < .01). There was no difference in the rate of discharge with parents between the 3 cohorts (Table 1) There were no reported incidence of seizures or episodes of apnea or desaturations related to opioid treatment.
Discussion
We report the outcomes following implementation of a comprehensive multidisciplinary QI initiative to improve outcomes in infants with NOWS admitted to our NICU. In the first year after the implementation of the ESC program, we noted improved outcomes with more than 50% decrease in length of stay, proportion and days of treatment with opioids, and cessation of use of adjunct medications with lower hospital associated costs.
Our QI initiative included successful improvements documented in the literature, including a multidisciplinary approach, education and engagement of families across the peripartum spectrum, implementation of ESC with associated care-team huddles, and infant-centered care including bundling care times and vital signs and minimizing environmental stimuli.7–15 Although the majority of the published studies involved care of infants with NOWS in pediatric inpatient settings in private rooms with the family, our QI initiative included infants admitted to the NICU. When compared with care in NICU, a recent meta-analysis reported that managing infants in pediatric in-patient settings and rooming-in with families decreased the need of pharmacotherapy by 63%.18 A recent clinical report from the American Academy of Pediatrics on NOWS states that “admission to a NICU only for opioid exposure or NOWS is not required.”19 However, our institution does not have a pediatric inpatient unit and all newborns requiring cardiorespiratory monitoring or opioid medications are transferred to the NICU. Our results with decreased length of stay and pharmacotherapy show that individualized care bundles can be provided to infants with NOWS in NICU settings with improved outcomes similar to other published studies.
However, the COVID-19 pandemic had a significant impact on our patient population and caused disruption to our QI initiative. We noted that there was a decrease in medication assistant programs for mothers with drug use disorder during pregnancy and an increase in maternal fentanyl use during the pandemic. Withdrawal symptoms from in-utero fentanyl, an immediate release opioid, is likely to have an earlier onset and shorter duration of withdrawal symptoms and may have an impact on some of our results.19 During the early phase of pandemic, visitation policies were restricted to 1 parent at bedside, no volunteers were allowed as cuddlers, and our single and family rooms, where majority of our infants with NOWS were cared for, were temporarily reassigned to accommodate the rapidly escalating adult population with COVID-19.16 Reassigning rooms may have some impact as more infants were treated in open bay area next to infants with different clinical conditions and acuity. Subsequently, we noted an increase in length of stay and pharmacological treatment in infants with NOWS. We attempted to adapt to these changes by having ancillary staff, including health care assistants, occupational, and physical therapists volunteering as cuddlers to help parents and nurses in holding and consoling infants. Pharmacist and nursing clinical supervisors provided oversight and reviewed all infants started on opioids for NOWS and ensured rapid weaning as appropriate while emphasizing nonpharmacological interventions. Despite an increase in proportion of infants treated with opioids during the pandemic, the cumulative opioid dose, total days of opioid treatment, and length of stay remained significantly lower than the preimplementation cohort. This suggested that, despite the impact on the resources required for the QI initiative, a cultural change in managing the infants with functional assessment and prioritization of nonpharmacological interventions and our adaptation of the program during the pandemic may have contributed to sustainment of some of the gains made in the first year after the implementation of the QI initiative.
We noted multiple outliers suggesting special cause variations above the upper control limits. This may be related to variation among the providers in use of medications to treat NOWS before our standardized approach with ESC tool or difference in severity of NOWS in individual infants. We noted some of the outliers decreased in the first year of implementation. This may be secondary to decreased variability in practice with standardized approach. Similar findings of higher outliers are reported by other QI initiatives in their baseline period, which improved after implementation of the standardized program.7,9 There was increase in the outliers in pandemic period in opioid treatment days likely indicating increased variability during the pandemic.
Few quality improvement programs based on ESC assessment tool have emphasized the role of families in the care of the infants and has shown increased parental presence and role in care of the infants.7,10 Rooming-in with families have been shown to improve outcomes in infants with NAS.18,20 However, our rates of discharge with families did not improve with the QI initiatives and rates of maternal milk use decreased during the pandemic period, which we believe may be related to pandemic related change in visitation policies.16 We were limited by the inability to extrapolate data on parental presence accurately from electronic medical records. Despite these findings, our results show that by having a comprehensive QI initiative incorporating functional assessment and nonpharmacological bundle, outcomes can be improved when parental presence is limited.
Our study has few limitations. Interpretability of our comparative analysis of baseline characteristics and secondary outcomes are limited by small sample size, however our report of change and improvement in outcomes with the control charts is consistent with other QI initiatives.7,9,10 Our unique patient population, availability of resources, our physical layout of the NICU, and specific components of our QI initiative may limit generalizability of our results. As significant proportion of our infants were discharged under the care of foster families or child protective services, we were unable to follow these infants to evaluate readmission rates, which would be an important balancing measure, however previous studies have not reported increase in readmission rates with ESC based QI initiatives.
Conclusions
Our study adds to the strong evidence that a functional, rather than symptom-based, assessment tool and prioritization of nonpharmacological interventions via ESC based QI initiative in a NICU setting is associated with significant decrease in LOS and opioid pharmacotherapy in infants with NOWS with reduced hospital costs. Our report on the impact of the pandemic on the resources and outcome of the ESC based QI initiative and our steps to mitigate the impact may have implications for centers with similar experience.
Acknowledgments
We are extremely grateful to the editor and the reviewers for their thorough review of our manuscript and for their insightful suggestions and comments that have improved the quality of the manuscript.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no conflicts of interest relevant to this article to disclose.
Dr Muniraman conceptualized and designed the study, designed the data collection instrument, obtained ethical and institutional research board approval, codrafted the initial manuscript, and revised the manuscript; Drs Pham and Small collected data, codrafted the first manuscript draft, reviewed and revised the manuscript; Ms Turbenson collected data and reviewed and revised the manuscript; Ms Coleman and Dr Manuel developed ESC program as core focus group members and implemented ESC program and critically reviewed 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