Abstract
Objective
To determine if newborns receiving morphine or methadone as the primary pharmacologic treatment for neonatal opioid withdrawal syndrome (NOWS) tolerate and receive fewer days of opioid using an accelerated wean protocol (15% decrements) compared with using a slower wean protocol (10% decrements).
Study design
Newborns ≥ 36 weeks of gestation receiving morphine or methadone for NOWS were enrolled in a pragmatic blinded, randomized multicenter trial. Newborns underwent protocol-driven weaning with decreasing opioid doses of either 15% or 10% decrements. Weaning was encouraged every 24 hours and if signs of NOWS worsened, the preceding dose was resumed. To maintain blinding, the last 3 dose levels of the 15% decrements were placebo. The primary outcome was the number of days of opioid treatment from the first weaning dose to cessation of opioids.
Results
Slow enrollment prompted early trial closure; 189 newborns were randomized, 98 (51.9%) to 15% decrements (mean ± standard deviation, 38.8±1.2 weeks gestation, 59.8% male) and 91 (48.1%) to 10% decrements (38.8±1.3 weeks gestation, 61.5% male). Morphine was used most commonly. Intention to treat analysis included all but 4 infants withdrawn in the 15% decrement group. The durations of opioid treatment during weaning were 8.2 (7.2, 9.5) (adjusted mean [95% confidence interval]) and 11.2 (9.7, 12.9) days for 15% and 10% decrement groups, respectively (p<0.001). Adverse events were few in both groups.
Conclusion
Pharmacologic treatment of NOWS using an accelerated wean protocol (15% decrements) was well tolerated with fewer days of opioid treatment compared with 10% decrements.
Trial registration
ClinicalTrials.gov number: NCT04214834
Keywords: morphine, methadone, newborn, abstinence, medication reduction
The management of newborns with neonatal opioid withdrawal syndrome (NOWS) continues to evolve with widespread acceptance that initial treatment should be non-pharmacologic1. Enhanced non-pharmacologic interventions as implemented in the Eat, Sleep and Console (ESC NOWS) trial were associated with decreased pharmacologic treatment and a shortened duration of hospitalization2. As emphasized by the American Academy of Pediatrics, when non-pharmacologic treatment does not control signs of NOWS, opioid replacement treatment is indicated3. There are two principal approaches to opioid replacement, a structured regimen and symptom-based (spot) dosing. Each approach may have an important role in the treatment given the wide range of severity of NOWS.
The trial reported herein focuses on a structured opioid replacement regimen which is triggered when a preset threshold on a scoring tool to assess signs of NOWS (eg, Finnegan Neonatal Abstinence Scoring System [FNASS]4, or the ESC care approach) is exceeded. Pharmacologic treatment is initiated and if necessary, the dose is escalated to minimize signs of NOWS (initiation). Once signs are managed, the dose is maintained (stabilization), followed by serial drug reductions (weaning) until off medication5. The weaning period is the longest duration of pharmacologic treatment and provides the greatest opportunity to reduce the duration of treatment, hospitalization, and separation of infants and caregivers. To our knowledge there have been no prospective randomized trials of the weaning interval of pharmacologic treatment for newborns with NOWS. This trial was designed for newborns receiving an opioid (morphine or methadone) as the primary pharmacological treatment for NOWS to test the hypothesis that an accelerated wean protocol (serial 15% reductions) would be tolerated and reduce the days of opioid treatment compared with a slower wean protocol (serial 10% reductions).
Methods
Design and Oversight
This was a prospective, randomized, double-blind parallel design trial conducted at 23 hospitals in the Advancing Clinical Trials in NOWS (ACT NOWS) Consortium (ClinicalTrials.gov: NCT04214834). The University of Arkansas for Medical Sciences approved the trial and served as the central institutional review board (IRB) with reliance agreements for all hospitals. Written informed consent was required by a parent or legal guardian for all participants.
Participants
Eligibility criteria included a parent or legal guardian and their newborn born ≥36 weeks of gestation, receiving therapy with morphine or methadone as the primary treatment for NOWS, and tolerating enteral feeds and medications by mouth. Exclusion criteria included major birth defects, major surgery, hypoxic-ischemic encephalopathy, seizures from etiologies other than NOWS, respiratory support for > 72 hours, non-NOWS treatment with opioids, planned hospital discharge on opioids, and prior initiation of opioid weaning.
Intervention
The study team reviewed weaning strategies of potential participating hospitals and recent clinical trials6,7 to plan the intervention. The most frequent practice was weaning morphine by serial 10% decrements. A common approach to weaning methadone was not evident. The study team chose serial reductions of 10% for the slower wean arm to reflect clinical practice and recent trials6,7. Serial reductions of 15% were chosen for the accelerated weaning arm as this comparison could reduce hospital days, cost, and caregiver-infant separation.
This was a pragmatic trial that performed the intervention within each hospital’s model of care for NOWS. The trial allowed use of hospital protocols for pharmacologic treatment of NOWS, including scoring tool, choice of opioid (morphine or methadone), and guidelines for treatment. Other aspects of care, including a non-pharmacologic bundle were administered according to each hospital’s practice. Opioid exposed newborns were identified through hospital-specific practices. During initiation or stabilization, newborns were randomly assigned to either treatment group by the Neonatal Research Network Data Coordinating Center with stratification by hospital. Multiple births were independently randomized. Only hospital pharmacy personnel had access to group assignments.
Participants underwent protocol-driven weaning from the stabilization dose by serially decreasing dose levels by 15% (accelerated wean protocol) or 10% (slower wean protocol) per clinician’s decision to wean. Clinical teams were encouraged to wean opioids daily. Escalation of opioid dose was per hospital guidelines, and the preceding dose was resumed. Medication was terminated at 25% or 20% of the stabilization dose in the accelerated versus the slower wean arms, respectively. To maintain blinding and ensure an identical treatment duration in the absence of dose escalations, the last 3 dose levels of the accelerated wean protocol were placebo (normal saline or dextrose water). Adjunctive non-opioid medications to manage NOWS were initiated per hospital practice before or during protocol weaning. Symptom-based spot dosing was not permitted since it would prevent comparison of two fixed weaning regimens.
Details of the intervention (ie, weaning, escalations, medication preparation) are described elsewhere8. Once weaned off study drug, participants underwent 48 hours of monitoring for recurrence of NOWS signs. If the infant’s NOWS signs met hospital criteria for pharmacologic therapy, the prior dose was restarted (25% or 20% of the stabilization dose for the 15% or 10% decrement group, respectively). During the interval off opioid treatment, a masked, trained examiner administered the NeoNatal Neurobehavioral Scale, 2nd edition (NNNS-II)9. Hospital discharge was at the discretion of the clinical team. Participants are engaged in ongoing follow-up for neurodevelopmental outcomes through 2 years of age, with initial contact at 4 weeks post-discharge. The Food and Drug Administration determined that this protocol was exempt from a sponsor-investigator Investigational New Drug.
Predefined adverse events were used to evaluate tolerance of weaning and included seizures, increased stool output requiring intravenous fluid therapy, respiratory disturbances requiring pulmonary support, a change in feeding resulting in intravenous fluids, or any clinical change deemed important by the site investigator. Participants exited the intervention without unblinding if they experienced an adverse event deemed serious, failed to wean off study drug by 35 days, or per request of the physician, parent, or legal guardian. Data were collected by medical chart review for maternal (including race and ethnicity as mandated by the National Institutes of Health), and newborn variables and NOWS treatment. FNASS scores and ESC assessments were not standardized nor collected.
Outcomes
The primary outcome was the duration of opioid treatment from the first weaning dose to opioid cessation in each group. The primary outcome accounted for time without opioid exposure (eg, resumption of opioid treatment during the 48-hour observation), and the cumulative time was expressed in days. Secondary outcomes included the duration of treatment with morphine or methadone, the proportion of participants with an escalation or resumption of opioids, the total amount of opioid, the proportion of infants with adverse events, the NNNS-II scores, and the length of hospital stay.
Sample Size and Statistical Analysis
A sample size of 502 newborns (251/intervention arm) was estimated using a 2-day treatment difference, a standard deviation of 6.9 days7, an alpha error of 0.05, and 90% power. An independent Data Safety Monitoring Board (DSMB) was to review the trial after 25, 50, and 75% of enrolled newborns were ready for discharge, but slow recruitment prompted sample size re-estimation (without unblinding) using aggregate results of the first 116 newborns enrolled. The DSMB accepted a recommendation to reduce the sample size to 302 participants based on 80% power and an alpha of 0.05 to detect a 2-day treatment difference and stipulated that enrollment would continue but end on December 31, 2023.
Categorical variables were summarized with frequency distributions. Continuous and count variables were summarized using mean and standard deviation (SD) or median and interquartile range (IQR) for skewed data. To compare the primary outcome between groups, we employed linear mixed models with hospital as a random effect and treatment arm as a fixed effect. The primary outcome variable was log-transformed to account for the highly skewed data. We report adjusted means, relative means, and their respective 95% confidence intervals (CIs). We calculated the absolute mean difference along with its 95% CI using bootstrapping with 1,000 resamples. A similar approach was used for continuous secondary outcomes. For count and dichotomous secondary outcomes, generalized linear mixed models with Poisson distribution were applied10.
Pre-specified sensitivity analyses included: 1) treatment heterogeneity by sex and race, and 2) adjustment for potential confounding variables before the intervention, including maternal medication for opioid use disorder (MOUD with methadone or buprenorphine), newborn opioid stabilization dose, and feeding method. A post-hoc sensitivity analysis adjusted for initiation of phenobarbital and/or clonidine prior to the weaning intervention.
Analyses for the primary outcome were conducted with the intent to treat sample. Infants withdrawn before or during the intervention without permission for continued data collection were excluded from the primary analysis. For infants withdrawn during the intervention with permission for continued data collection, we used the sum of days of opioid treatment on (hour of last dose known) and off protocol (day of last dose known) to estimate the primary outcome. As per protocol, the number of days from first weaning dose to opioid cessation was truncated to 35 days. To evaluate the impact of this truncation and include data from withdrawn infants, we conducted a sensitivity analysis using survival analysis methods. Analyses were conducted using SAS software, and Graphs were created using R software.
Results
From 9/29/2020 through 12/31/2023, 1,946 newborns were screened, 685 were eligible, and 189 were randomized to 15% (n=98) or 10% decrements (n=91) (Figure 1). Most mothers were single, on public health insurance, and had polysubstance exposure and multiple psychiatric diagnoses (Table I). Maternal characteristics were balanced across groups except for more frequent hypertensive disorders, more buprenorphine, and less methadone MOUD treatment in the 15% decrement group. Newborn characteristics were similar between groups (Table II).
Figure 1:
CONSORT flow diagram
a Newborns withdrawn from the trial without permission for further data collection occurred during stabilization (n=2) and during weaning (n=2). Withdrawals from the trial with permission for further data collection occurred during weaning (n=7) and after 35 days of study intervention (per protocol, n=1). An additional 5 infants were withdrawn after weaning was completed.
b Newborns withdrawn from the trial with permission for further data collection occurred during weaning (n=1).
c Newborns withdrawn from the trial with permission for further data collection occurred during weaning (n=6). An additional 4 infants were withdrawn after weaning was completed.
d Newborns withdrawn from the trial with permission for further data collection occurred during weaning (n=1) and after 35 days of study intervention (per protocol, n=1).
Table I.
Maternal characteristics of enrolled newborns
| 15% Decrement N=98c (51.9%) | 10% Decrement N=91c (48.1%) | |
|---|---|---|
| Age, years | 30.9 (5.0) | 31.6 (5.2) |
| Marital status: Married | 12/97 (12.4) | 10/89 (11.2) |
| Single | 81/97 (83.5) | 75/89 (84.3) |
| Unknown | 4/97 (4.1) | 4/89 (4.5) |
| Ethnicity: Hispanic | 17/97 (17.5) | 18/90 (20.0) |
| Non-Hispanic or Latino | 77/97 (79.4) | 66/90 (73.3) |
| Unknown | 3/97 (3.1) | 6/90 (6.7) |
| Race: American Indian or Alaskan Native | 3/97 (3.1) | 1/90 (1.1) |
| Black | 16/97 (16.5) | 10/90 (11.1) |
| Black and White | 0/97 (0.0) | 1/90 (1.1) |
| White | 74/97 (76.3) | 74/90 (82.2) |
| Unknown | 4/97 (4.1) | 4/90 (4.4) |
| Highest level of education | ||
| High school diploma or less | 31/97 (32.0) | 31/90 (34.4) |
| Partial college or associate/technical school | 18/97 (18.6) | 15/90 (16.7) |
| College degree | 2/97 (2.1) | 2/90 (2.2) |
| Graduate school | 0/97 (0.0) | 2/90 (2.2) |
| Unknown | 46/97 (47.4) | 40/90 (44.4) |
| Mother’s medical insurance: Public | 84/97 (86.6) | 77/90 (85.6) |
| Private | 8/97 (8.2) | 7/90 (7.8) |
| Other | 1/97 (1.0) | 1/90 (1.1) |
| Unknown | 4/97 (4.1) | 4/90 (4.4) |
| Uninsured | 0/97 (0.0) | 1/90 (1.1) |
| Prenatal care: Adequate | 53/97 (54.6) | 42/90 (46.7) |
| Limited | 32/97 (33.0) | 33/90 (36.7) |
| None | 9/97 (9.3) | 13/90 (14.4) |
| Unknown | 3/97 (3.1) | 2/90 (2.2) |
| Multiple gestation: Yes | 1/97 (1.0) | 3/90 (3.3) |
| No | 96/97 (99.0) | 87/90 (96.7) |
| Diabetes prior to pregnancy: Yes | 2/97 (2.1) | 0/90 (0.0) |
| No | 93/97 (95.9) | 89/90 (98.9) |
| Unknown | 2/97 (2.1) | 1/90 (1.1) |
| Gestational diabetes: Yes | 4/97 (4.1) | 1/90 (1.1) |
| No | 87/97 (89.7) | 86/90 (95.6) |
| Unknown | 6/97 (6.2) | 3/90 (3.3) |
| Hypertension prior to pregnancy: Yes | 30/97 (30.9) | 17/90 (18.9) |
| No | 64/97 (66.0) | 73/90 (81.1) |
| Unknown | 3/97 (3.1) | 0/90 (0.0) |
| Pre-eclampsia: Yes | 17/97 (17.5) | 9/90 (10.0) |
| No | 76/97 (78.4) | 80/90 (88.9) |
| Unknown | 4/97 (4.1) | 1/90 (1.1) |
| Chorioamnionitis: Yes | 0/97 (0.0) | 1/90 (1.1) |
| No | 93/97 (95.9) | 85/90 (94.4) |
| Unknown | 4/97 (4.1) | 4/90 (4.4) |
| Mode of delivery: Vaginal | 71/97 (73.2) | 61/90 (67.8) |
| Cesarean | 26/97 (26.8) | 29/90 (32.2) |
| Psychiatric diagnosisa: Any | 72/97 (74.2) | 69/90 (76.7) |
| Bipolar | 13/97 (13.4) | 11/90 (12.2) |
| Anxiety | 58/97 (59.8) | 47/90 (52.2) |
| Depression | 53/97 (54.6) | 54/90 (60.0) |
| Schizophrenia | 1/97 (1.0) | 0/90 (0.0) |
| None | 22/97 (22.7) | 18/90 (20.0) |
| Unknown | 3/97 (3.1) | 3/90 (3.3) |
| Medication for opioid use disorder | ||
| Methadone | 32/97 (33.0) | 38/90 (42.2) |
| Buprenorphine with/without naloxone | 40/97 (41.2) | 30/90 (33.3) |
| None/Unknown | 25/97 (25.8) | 22/90 (24.4) |
| Maternal exposures: Any nicotine use | 53/97 (54.6) | 48/90 (53.3) |
| Any alcohol use | 4/97 (4.1) | 4/90 (4.4) |
| Polysubstance exposureb: None | 0/97 (0.0) | 0/90 (0.0) |
| 1 exposure | 5/97 (5.2) | 10/90 (11.1) |
| 2 exposures | 20/97 (20.6) | 11/90 (12.2) |
| 3 exposures | 23/97 (23.7) | 19/90 (21.1) |
| > 3 exposures | 49/97 (50.5) | 50/90 (55.6) |
Note: Categorical variables are represented as n/N (%). Continuous variables are represented by mean (SD)
These categories are not mutually exclusive
Exposures were based on maternal report and/or drug screens. Most frequent exposures were nicotine, fentanyl, methadone, buprenorphine, marijuana, amphetamine, opioid, benzodiazepine, heroin, methamphetamine, selective serotonin receptor inhibitor, cocaine, gabapentin, and oxycodone
Data from 2 participants were missing, resulting in a maximum of 187 contributing responses
Table II.
Infant characteristics
| 15% Decrement N=98 | 10% Decrement N=91 | |
|---|---|---|
| Infant characteristics after birth | ||
| Gestational age (weeks) | 38.8 (1.2) | 38.8 (1.3) |
| Weight at birth (kg) | 3.0 (0.5) | 3.1 (0.4) |
| Proportion of infants with birth weight < 10th percentilea | 24/97 (24.7) | 16/91 (17.6) |
| Head circumference at birth (cm) | 33.3 (1.7) | 33.7 (1.6) |
| Proportion of infants with head circumference < 10th percentilea | 23/97 (23.7) | 15/91 (16.5) |
| Length at birth (cm) | 48.4 (2.9) | 49.0 (2.2) |
| Proportion of infants with length < 10th percentilea | 25/97 (25.8) | 12/91 (13.2) |
| Outbornb | 16 (16.5) | 18 (19.8) |
| Apgar scores, median (IQR) n | ||
| 1 minute | 8.0 (7.0, 8.0) 94 | 8.0 (8.0, 8.0) 89 |
| 5 minute | 9.0 (8.0, 9.0) 95 | 9.0 (9.0, 9.0) 89 |
| Sex | ||
| Male | 58/97 (59.8) | 56/91 (61.5) |
| Female | 39/97 (40.2) | 35/91 (38.5) |
| Breast fedc | 15/91 (16.5) | 8/88 (9.1) |
| Toxicology screen: yes | 89 (91.8) | 79 (86.8) |
| Fetal exposuresd: | ||
| None | 7/89 (7.9) | 4/79 (5.1) |
| 1 exposure | 21/89 (23.6) | 18/79 (22.8) |
| 2 exposures | 28/89 (31.5) | 24/79 (30.4) |
| 3 exposures | 19/89 (21.3) | 21/79 (26.6) |
| > 3 exposures | 14/89 (15.7) | 12/79 (15.2) |
| Infant characteristics at the start of the intervention | ||
| Age at start of intervention (days) | 5.8 (2.5) | 6.3 (3.0) |
| Weight (kg) | 2.9 (0.5) | 3.0 (0.4) |
| Treatment medication | ||
| Morphine | 79/98 (80.6) | 66/91 (72.5) |
| Methadone | 19/98 (19.4) | 25/91 (27.5) |
| Stabilization dose (mg/kg/dose) | ||
| Morphine | 0.07 (0.03) | 0.08 (0.05) |
| Methadone | 0.10 (0.04) | 0.11 (0.06) |
| Second- or third-line drugs started prior to interventione | 18 (18.8) | 19 (20.9) |
| Phenobarbital | 8/18 (44.4) | 4/19 (21.1) |
| Clonidine | 4/18 (22.2) | 9/19 (47.4) |
| Morphine | 8/18 (44.4) | 9/19 (47.4) |
| Methadone | 0/18 (0.0) | 1/19 (5.3) |
| Location of care at start of intervention | ||
| Newborn nursery | 17/96 (17.7) | 15/91 (16.5) |
| Special care nursery | 14/96 (14.6) | 13/91 (14.3) |
| NICU | 58/96 (60.4) | 54/91 (59.3) |
| Regional NICU | 7/96 (7.3) | 8/91 (8.8) |
| Pediatric unit | 0/96 (0.0) | 1/91 (1.1) |
| Co-enrollment in Eat-Sleep-Console Trialf | 3/98 (3.1) | 4/91 (4.4) |
Note: Categorical variables are represented as n/N (%). Continuous variables are represented by mean (SD) unless otherwise indicated
Percent of infants with measurement < 10% using Fenton curves
Born at a different facility then the one providing care for opioid treatment
Infant placed to the breast for some or all of the feeds
Exposures were based on newborn toxicology of urine, meconium, hair, or umbilical cord blood or tissue. Most frequent exposures were fentanyl, methadone, buprenorphine, and marijuana
These categories are not mutually exclusive; 3 newborns received both phenobarbital and clonidine
Young LW et al, N Engl J Med. 2023;388:2326–37 (reference 2)
The FNASS or modified FNASS were used in 93% of the hospitals. The opioid stabilization dose was nearly identical between groups (Table II). Prior to weaning, phenobarbital was used more frequently in the 15% decrement group, and clonidine was used more frequently in the 10% decrement group as second-line medications. The primary outcome was shorter in the 15% compared with the 10% decrement group (unadjusted mean ± SD, 9.3±6.3 vs 11.7±6.1 days for 15% vs 10% decrements, respectively, p=0.009). Group differences persisted when hospital was added as a random effect (adjusted mean, [95% CI], 8.2 [7.2, 9.5] vs 11.2 [9.7, 12.9] days for the 15% vs 10% decrement group, respectively, p<0.001, Table III) with an absolute mean difference (95% CI) of 3.0 days (1.7, 4.3). The differences between groups persisted in sensitivity analyses after adjustment for potential covariates, and when adjusted for use of phenobarbital or clonidine prior to weaning (Table III). There were no effects of sex or race on the primary outcome. The survival model indicated a 24% decrease in the days of opioid treatment in the 15% decrement compared with the 10% decrement group (Figure 2). Seventeen newborns were withdrawn during the intervention with ongoing data collection (9 and 8 newborns from the 15% and 10% decrement groups, respectively). The most common reasons for withdrawals were parental requests (eg, concerns about blinding, dose escalation), change in custody where wards of the state were prohibited from research participation, and adverse events.
Table III.
Days of opioid treatment from the first wean to cessation of opioid
| Outcomesa | 15% Decrement Adjusted mean (95% CI) |
10% Decrement Adjusted mean (95% CI) |
Adjusted relative mean (95% CI) |
Absolute difference (95% CI)e |
P valuef |
|---|---|---|---|---|---|
| Primary outcome | |||||
| Days of opioid during weaningb | 8.2 (7.2,9.5) | 11.2 (9.7, 12.9) | 0.74 (0.64, 0.85) | 3.0 (1.7, 4.3) | <0.001 |
| Sensitivity analysis | |||||
| Days of opioid during weaning with covariatesc | 8.2 (7.0, 9.7) | 10.7 (9.0, 12.7) | 0.77 (0.67, 0.89) | 2.5 (1.3, 3.9) | <0.001 |
| Days of opioid during weaning adjusted for prior use of non-opioid medicationsd | 8.3 (7.1, 9.8) | 11.2 (9.5,13.2) | 0.74 (0.64, 0.85) | 3.0 (1.7, 4.4) | <0.001 |
A logarithmic transformation was applied to outcomes of all models and hospitals were added as random effects to account for variation between them
Sample size was 185 out of 189 randomized, consented newborns. Four newborns were not included in the analysis due to withdrawal before the end of the intervention without permission for further data collection. Of 185 newborns, 17 were withdrawn with permission for data collection. Two of the 17 had days of opioid treatment above 35 (37 and 38 days) and their study medication was truncated at 35 days per protocol. The days of opioid treatment for these 17 withdrawn newborns were calculated by combining days of treatment per protocol before withdrawal with the days of opioid use after withdrawal. Additionally, 9 newborns were withdrawn after completion of the study intervention and had days of opioid per protocol.
The model was adjusted for MOUD (methadone vs buprenorphine), opioid stabilization dose, and mode of feeding (breast vs bottle) as fixed effects and hospitals as random effects. Sample size was 177 out of 185 due to missing mode of feeding (n=7) and MOUD (n=1). All covariates were non-significant.
Model adjusted for a dichotomous variable indicating whether phenobarbital and/or clonidine were initiated prior to weaning (n=185). The use of phenobarbital and clonidine were non-significant.
Absolute difference and 95% CI were derived from 1,000 bootstrap resamples.
p-values are based on the original model and not the bootstrap resampling.
Figure 2:
Time to discontinuation of opioid medication during weaning
A survival model for 15% decrements (black) and 10% decrements (red) of opioids was run with a parametric method using a log normal distribution since the proportional hazards assumption for a semi-parametric model was not met. A beta coefficient (95% confidence interval) of 0.76 (0.66, 0.87, p<0.001) reflected the expected decrease of approximately 24% in days of opioid treatment between the 15% decrement and 10% decrement groups. Censored newborns are indicated by the symbol +.
The primary outcome was shorter in the 15% decrement group for morphine; results for methadone paralleled morphine but were not statistically significant (Table IV). Escalations during the 48-hour observation period were more frequent and unexpected in the 10% decrement group. Newborns in the 15% vs the 10% decrement group were exposed to a lower total amount of opioid during the intervention (9.3±8.1 vs 15.3±13.4 morphine mg equivalents/kg, respectively, p<0.001). In each group there were few adverse events (Table IV), a similar frequency of protocol violations and deviations (Table V), and similar newborn characteristics and NNNS-II scores at hospital discharge (Table VI). Medical visits for NOWS symptoms at 4 weeks post-discharge were infrequent in both groups (Table VI).
Table IV.
Secondary outcomes
| 15% Decrement N=98d | 10% Decrement N=91d | Adjusted relative effect (95% CI)e | |
|---|---|---|---|
| Days of opioid during weaning when treated with morphine mean (N, SD)a | 8.9 (75, 5.8) | 11.6 (66, 5.6) | 0.70 (0.60, 0.82) |
| Days of opioid during weaning when treated with methadone mean (N, SD)a | 10.8 (19, 8.0) | 12.1 (25, 7.2) | 0.89 (0.59, 1.32) |
| Proportion of infants with an escalation during weaning n/N, (%)b | 56/96 (58.3) | 47/91 (51.6) | 1.13 (0.91, 1.40) |
| Proportion of infants with an escalation above stabilization dose during weaning n/N (%)b | 3/96 (3.1) | 5/91 (5.5) | 0.57 (0.15,2.15) |
| Proportion of infants with an escalation during the 48h observation period n/N (%)b | 3/93 (3.2) | 17/91 (18.7) | 0.17 (0.05, 0.59) |
| Number of escalations per infant during weaning, median (IQR) Nb | 1.0 (0.0, 3.0) 96 | 1.0 (0.0, 2.0) 91 | 1.28 (1.02, 1.61) |
| Total amount of opioid during weaning (morphine mg equivalents/kg), mean (N, SD)a,c | 9.3 (85, 8.1) | 15.3 (83, 3.4) | 0.59 (0.49, 0.72) |
| Initiation of 2nd or 3rd line medications during weaning n/N, (%)b | 9/96 (9.4) | 11/91 (12.1) | 0.78 (0.32, 1.91) |
| Type of 2nd and 3rd line medication initiated during weaning | |||
| Phenobarbital n/N, (%) | 6/8 (75.0) | 3/4 (75.0) | 1.85 (0.46,7.52) |
| Clonidine n/N, (%) | 3/8 (37.5) | 1/4 (25.0) | 2.45 (0.25, 24.1) |
| Adverse events | |||
| Seizure | 1 | 0 | |
| Bloody stool | 2 | 1 | |
| Respiratory disturbance | 2 | 3 | |
| Feeding difficulties | 0 | 0 | |
| Otherf | 1 | 3 | |
| Total adverse eventsg | 6 | 7 |
A logarithmic transformation was applied to the outcome, and a linear model with hospital as random effect was used to account for variation between hospitals
Poisson model with hospital as a random effect was used to account for variation between hospitals
Morphine milligram equivalents were derived as described in reference 32
Unadjusted reported values
For continuous outcome adjusted relative means are reported and for binary outcomes adjusted relative risks are reported
Other: over-sedation (clinical team inadvertently did not stop opioids when study medication was initiated), epidural hematoma/parietal skull fracture (scalp swelling noted at the start of weaning and neuroimaging revealed the findings), and an inguinal hernia repair
All adverse events in the 15% decrement group occurred during weaning and 4 of 7 adverse events occurred during weaning in the 10% decrement group. The skull fracture/epidural hematoma (same infant) was felt to be birth related and the inguinal hernia was reported after the infant was withdrawn from the study but prior to discharge.
Table V.
Protocol violations and protocol deviations
| 15% Decrement | 10% Decrement | |
|---|---|---|
| Number of randomized newborns with protocol violations or deviationsa | 46/98 (46.9)b | 36/91 (39.6) |
| Newborns with non-pharmacy protocol violations or deviationsc | 43/98 (43.9) | 31/91 (34.1) |
| Newborns with pharmacy protocol violations or deviations | 12/98 (12.2) | 8/91 (8.8) |
Categories are not mutually exclusive, newborns could have both protocol violations or deviation
Results are n/N (%)
The most common protocol violation or deviation was an incorrect time of medication administration (too early, too late, missed, 48/145 [33%] violations/deviations). For context, a hospital that enrolled 10 participants receiving morphine, could administer greater than 800 doses of medication in total. Variables such as nurse to patient ratio (pandemic, off-shifts, etc) may have contributed. The next most common protocol violation or deviation was an incorrect gap dose (given but not indicated, not given when indicated, 38/145 [26%] violations/deviations).
Table VI.
Characteristics at discharge and at 4 weeks post-discharge
| 15% Decrement N=98 | 10% Decrement N=91 | |
|---|---|---|
| Age at discharge (days), mean (N, SD) | 22.7 (94, 11.0) | 23.4 (90, 11.0) |
| Weight at discharge (kg), mean (N, SD) | 3.3 (93, 0.7) | 3.4 (90, 0.6) |
| Head circumference at discharge (cm), mean (N, SD) | 34.6 (84, 1.8) | 34.9 (81,1.6) |
| Length at discharge (cm), mean (N, SD) | 49.9 (83, 2.8) | 50.9 (81, 2.4) |
| Neonatal Neurobehavioral Scale (NNNS-II) mean (N, SD) | ||
| Attention | 4.9 (54, 1.6) | 4.8 (47, 1.6) |
| Handling | 5.7 (62, 2.6) | 5.7 (58, 2.4) |
| Self-regulation | 4.3 (62,1.5) | 4.0 (57, 1.2) |
| Arousal-reactivity | 5.4 (63, 1.7) | 5.8 (59, 1.7) |
| Tone | 5.3 (61,0.7) | 5.3 (57, 0.6) |
| Non-optimal reflexes | 3.1 (60, 0.9) | 3.2 (57, 1.0) |
| Quality of movement | 6.4 (61, 1.4) | 6.2 (57, 1.3) |
| Stress abstinence | 2.3 (63, 0.8) | 2.4 (59, 0.7) |
| Length of hospital stay after start of weaning (days), mean (N, SD) | 16.9 (94, 10.1) | 17.1 (90, 10.1) |
| Infant discharged to: n/N (%) | ||
| Both parents | 24/94 (25.5) | 27/89 (30.3) |
| Mother | 31/94 (33.0) | 27/89 (30.3) |
| Father | 5/94 (5.3) | 3/89 (3.4) |
| Relative | 16/94 (17.0) | 14/89 (15.7) |
| Foster placement | 13/94 (13.8) | 11/89 (12.4) |
| Other | 5/94 (5.3) | 7/89 (7.9) |
| Discharged in state custody, n/N (%) | ||
| Yes | 23/94 (24.5) | 19/90 (21.1) |
| No | 71/94 (75.5) | 71/90 (78.9) |
| 4 weeks post-discharge | ||
| Any emergency department, acute care visit, or readmission where symptoms were related to NOWS, n/N (%) | 5/72a (6.9) | 1/63a (1.6) |
denominators are reduced due to withdrawn infants (n=30) and no shows (n=24)
Discussion
There is no standard approach to weaning opioids used to treat NOWS. Providers commonly wean morphine by 10% reductions from the stabilization dose but weaning methadone is more variable5. The findings of this trial indicate that accelerated (15% decrements) compared with slower weaning (10% decrements) resulted in fewer days of opioid treatment without an increase in adverse events, and similar neurobehavioral assessments. The results were consistent among unadjusted, adjusted, and sensitivity analyses of days of opioid treatment during weaning. There was no difference in the length of hospital stay, reflecting 3 dose levels of placebo in the 15% decrement arm to maintain blinding. The results of this trial support that compared with 10% decrements in morphine or methadone (as used in clinical practice or randomized trials 5–7), 15% decrements from the stabilization dose reduce the duration and total exposure to opioids and has the potential to decrease the length of medically indicated hospitalization, and separation of infants and caregivers.
Participating hospitals used morphine most frequently, consistent with a survey of care practices for NOWS from 2016–201711. A systematic review concluded that morphine is the least effective pharmacologic treatment for NOWS12, however widespread use continues and the results of this trial reflected primarily the use of morphine. Fewer infants were treated with methadone and the study results for methadone were not significant, but paralleled those of morphine. Structured opioid replacement with methadone has used population pharmacokinetic modeling to optimize a dosing regimen13 and was associated with a decrease in the length of treatment for NOWS in a pre-post cohort14. We noted variability in methadone weaning regimens among potential participating hospitals, which may reflect differences in hospital adjustments to weaning regimens as well as the high pharmacokinetic variability of methadone15. In contrast, randomized trials of methadone used a structured regimen with a fixed percent decrement during weaning7,16,17 which prompted our decision to wean morphine and methadone in a similar fashion.
To our knowledge there are no randomized trials of weaning opioids using structured opioid replacement to provide evidence-based guidelines for this phase of NOWS management. There are multiple randomized trials of pharmacologic treatments as the primary therapy for newborns with NOWS6,7,16–21, and each of these trials compared two medications across initiation, stabilization, and weaning. All except two trials17,18 weaned medications by decrements of 10% of the stabilization dose. Our results provide evidence that weaning can be done faster without an increase in adverse events. The intervention was complemented by encouragement to wean dosing daily to overcome a common clinical tendency to wean at longer intervals.
Quality initiatives have also provided important observations about weaning22. For example, a multicenter, collaborative cohort study performed in Ohio reported that adherence to a treatment protocol had the greatest impact on the use and duration of pharmacologic therapy, and length of hospital stay23,24. Use of a stringent 8-step weaning protocol was the most important predictor of short-term outcomes and reflects that weaning is the longest interval of pharmacologic treatment.
Prior to weaning there were group imbalances in maternal hypertensive disorders, MOUD programs, and adjunctive medications for NOWS treatment. We are not aware of any association between maternal hypertensive disease and NOWS. In a randomized blinded trial maternal MOUD with buprenorphine compared with methadone resulted in less morphine to treat NOWS and a shorter hospital stay25. However, our pre-specified sensitivity analysis including adjustment for MOUD type was consistent with the primary outcome. For infants treated with adjunctive medications, a retrospective cohort study and an older small, partially randomized trial suggest that phenobarbital may reduce treatment duration with morphine26,27. This prompted a post-hoc analysis which indicated that adjustment for non-opioid medication initiated prior to weaning did not alter the results.
A strength of this trial was that newborns were randomized and stratified by hospital given the marked center variation in the care of newborns with NOWS11,28,29. We implemented this as a blinded intervention to minimize bias associated with assessment tools (eg, FNASS) for NOWS30. The pragmatic features allowed conduct of the trial within the framework of individual hospitals’ models of care for NOWS. Limitations included a reduced sample size reflecting hospital-level changes in the number of mothers with opioid use disorder and decreasing numbers of pharmacologically treated newborns with NOWS, limited pharmacy support for an interventional trial with active management 24 hours a day, state-specific legal restrictions on research involving opioid exposed newborns, the impact of COVID-19 on staff availability, and a short interval to obtain consent from guardians with complex psychosocial circumstances who may have been under high stress after birth. Maintaining equipoise was a challenge, given that some clinicians wished to wean slower than per protocol. Finally, a common weaning regimen for methadone treatment is not apparent, and some hospitals wean medications by larger decrements than the study protocol.
The care of newborns with NOWS continues to evolve and has focused on implementing non-pharmacologic interventions, adopting ESC assessments for NOWS, and reducing pharmacologic treatment2,31. Irrespective of whether ESC or FNASS is used as a scoring tool, there will be newborns in need of opioid replacement treatment for NOWS. The results of this trial provide evidence that an accelerated weaning protocol of 15% compared with 10% decrements in opioid therapy has the potential to reduce hospital days and minimize the separation of caregivers and infants among pharmacologically treated newborns with NOWS.
Supplementary Material
Role of the sponsor:
Although NICHD and NIH ECHO staff (Environmental Influences on Child Health Outcomes) had input into the study design, conduct, analysis, and manuscript drafting, the comments and views of the authors do not necessarily represent the views of NICHD, the ECHO program, the National Institutes of Health, the Department of Health and Human Services, or the U.S. Government.
Abbreviations:
- ESC
Eat, Sleep and Console
- FNASS
Finnegan Neonatal Abstinence Scoring System
- MOUD
medication for opioid use disorder
- NOWS
neonatal opioid withdrawal syndrome
- NRN
Neonatal Research Network
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Publisher's Disclaimer: longer the Accepted Manuscript, but it is not yet the definitive Version of Record; we are providing this early version to give early visibility of the article. Please note that Elsevier’s sharing policy for the Published Journal Article applies to this version, see: https://www.elsevier.com/about/policies-and-standards/sharing#4-published-journal-article. Please also note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
This work was presented at the Pediatric Academic Society, April 2025 in Honolulu, Hawaii.
Conflicts of interest: None of the authors have conflicts of interest related to this trial.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data sharing:
The data supporting this study are available on the NICHD Data and Specimen Hub (DASH). Researchers interested in accessing the data can submit a request through the DASH platform.
References
- 1.Wachman E, Schiff D, Silverstein M. Neonatal Abstinence Syndrome: Advances in Diagnosis and Treatment. JAMA. 2018;319(13):1362–1374. [DOI] [PubMed] [Google Scholar]
- 2.Young LW, Ounpraseuth ST, Merhar SL, Hu Z, Simon AE, Brenner JY et al. , N Engl J Med. 2023;388:2326–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Patrick SW, Barfield WD, Poindexter BB, AAP COMMITTEE ON FETUS AND NEWBORN, COMMITTEE ON SUBSTANCE USEAND PREVENTION. Neonatal Opioid Withdrawal Syndrome. Pediatrics. 2020; 146(5):e2020029074. [DOI] [PubMed] [Google Scholar]
- 4.Finnegan L, Connaughton J, Kron R, Emich J. Neonatal abstinence syndrome: Assessment and management. Addict Dis 1975;2:141–158. [PubMed] [Google Scholar]
- 5.Kraft W, Stover M, Davis J. Neonatal abstinence syndrome: Pharmacologic strategies for the mother and infant. Seminars in Perinatology. 2016; 40:203–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kraft W, Adeniyi-Jones S, Chervoneva I, Greenspan JS, Abatemarco D, Kaltenbach K, et al. Buprenorphone for the Treatment of the Neonatal Abstinence Syndrome. N Engl J Med. 2017; 376:2341–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Davis J, Shenberger J, Terrin N, Breeze JL, Hudak M, Wachman EW, et al. Comparison of Safety and Efficacy of Methadone vs Morphine for Treatment of Neonatal Abstinence Syndrome: A Randomized Clinical Trial. JAMA Pediatr.2018; 172(8):741–748. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Czynski A, Laptook A, Das A, Smith B, Simon A, Greenberg R et al. Pragmatic, randomized, blinded trial to shorten pharmacologic treatment of newborns with neonatal opioid withdrawal syndrome (NOWS). Trials 10.1186/s13063-023-07378-x. [DOI] [Google Scholar]
- 9.Lester BM and Tronick EZ (eds). The Neonatal Neurobehavioral Scale (NNNS-II). Unpublished manual. 2024; Brown Center for Children. Women and Infants Hospital. [Google Scholar]
- 10.Zou Guangyong. A modified Poisson regression approach to prospective studies with binary data. American Journal of Epidemiology. 2004; 159(7): 702–706. [DOI] [PubMed] [Google Scholar]
- 11.Young L, Hu Z, Annett R, et al. Site-Level Variation in the Characteristics and Care of Infants with Neonatal Opioid Withdrawal. Pediatrics. 2021; 147(1): e2020008839. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Disher T, Gullickson C, Singh B, Cameron C, Boulos L, Beaubien L et al. , Pharmacological treatments for neonatal abstinence syndrome. A systematic review and network meta-analysis. JAMA Pediatr. 2019; 173(3):234–243. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wiles J, Isemann B, Mizuno T, Tabangin ME, Ward LP, Akinbi H et al. Pharmacokinetics of oral methadone in the treatment of neonatal abstinence syndrome: A pilot study. J Pediatr 2015; 167(6): 1214–1220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hall ES, Meinzen-Derr J, Wexeblatt SL. Cohort analysis of a pharmacokinetic-modeled methadone weaning optimization for neonatal abstinence syndrome. J Pediatr 2015;167(6):1221–1225. [DOI] [PubMed] [Google Scholar]
- 15.McPhail BT, Emoto C, Fukuda T, Butler D, Wiles JR, Akinbi H et al. Utilizing pediatric physiologically based pharmacokinetic models to examine factors that contribute to methadone pharmacokinetic variability in neonatal abstinence syndrome patients. J Clin Pharmacol. 2020;60(4):453–465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Brown MS, Hayes MJ, Thornton LM. Methadone versus morphine for treatment of neonatal abstinence syndrome: A prospective randomized clinical trial. Journal of Perinatology. 2015.35;278–283. [DOI] [PubMed] [Google Scholar]
- 17.Sutter MB, Watson H, Yonke N, Weitzen S, Leeman L. Morphine versus methadone for neonatal withdrawal syndrome: a randomized controlled pilot study. BMC Pediatrics. 2022.22;345, DOI: 10.1186/S12887-022-03401-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Jackson L, Ting A, Mckay S, Galea P, Skeoch C. A randomized controlled trial of morphine versus phenobarbitone for neonatal abstinence syndrome. Arch Dis Child Fetal Neonatal Ed. 2004; 89: F300–F304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bada H, Sithisarn T, Gibson J, Garlitz K, Caldwell R, Capilouto G et al. Morphine Versus Clonidine for Neonatal Abstinence Syndrome. Pediatrics. 2015.135;e383, DOI: 10.1542/peds.2014-2377. [DOI] [PubMed] [Google Scholar]
- 20.Nayeri F, Sheikh M, Kalani M, Niknafs P, Shariat M, Dalili H et al. Phenobarbital versus morphine in the management of neonatal abstinence syndrome, a randomized control trial. BMC Pediatrics. 2015;15:57, DOI: 10.1186/s12887-015-0377-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Bada HS, Westgate PM, Sithisarn T Yolton K, Charmigo R, Pourcyrous M et al. Clonidine as monotherapy for neonatal opioid withdrawal syndrome: A randomized trial. Pediatrics. 2024;154(5): e20230656610 [Google Scholar]
- 22.Patrick SW, Schumacher RE, Horbar JD, Buss-Frank ME, Edwards EM, Morrow KA et al. Improving Care for Neonatal Abstinence Syndrome. Pediatrics. 2016;137(5):e20153835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hall E, Wexelblatt S, Crowley M, Grow JL, Jasin LR, Klebanoff MA et al. A Multicenter Cohort Study of Treatments and Hospital Outcomes in Neonatal Abstinence Syndrome. Pediatrics. 2014.134;e527. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hall E, Wexelblatt S, Crowley M, Grow JL, Jasin LR, Klebanoff MA et al. Implementation of a Neonatal Abstinence Syndrome Weaning Protocol: A multicenter Cohort Study. Pediatrics. 2015.136;e803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Jones HE, Kaltenbach K, Heil SH, Stine SM, Coyle MG, Aria AM et al. Neonatal abstinence syndrome after methadone or buprenorphine exposure. N Engl J Med. 2010.363;2320–2331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Merhar SL, Ounpraseuth S, Devlin LA, Poindexter BB, Young LW, Berkey SD et al. Phenobarbital and clonidine as secondary medications for neonatal opioid withdrawal syndrome. Pediatrics. 2021;147(3):e2020017830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Coyle MG, Ferguson A, Lagasse L, Oh W, Lester Bl. Diluted tincture of opium (DTO) and phenobarbital versus DTO alone for neonatal opiate withdrawal in term infants. J Pediatr. 2002;140:561–564. [DOI] [PubMed] [Google Scholar]
- 28.Patrick SW, Kaplan HC, Passarella M, Davis MM, Lorch SA. Variation in treatment of neonatal abstinence syndrome in US Children’s Hospitals, 2004–2011. Journal of Perinatology. 2014; 34: 867–872. [DOI] [PubMed] [Google Scholar]
- 29.Milliren C, Gupta M, Graham d, Melvin P, Jorina M, Ozonoff A. Hospital Variation in Neonatal Abstinence Syndrome Incidence, Treatment Modalities, Resource Use, and Cost Across Pediatric Hospital in the United States, 2013 to 2016. Hospital Pediatrics. 2018; 8(1):15–20. [DOI] [PubMed] [Google Scholar]
- 30.D’Apolito K. Assessing Neonates for Neonatal Abstinence: Are you Reliable? J Perinat Neonat Nurs. 2014;28:220–231. [Google Scholar]
- 31.Hahn PD, Graham DA, Ozonoff A, Milliren CE. Neonatal opioid withdrawal syndrome treatment guidelines and birth hospital utilization. Pediatrics. 2024. 154(1):e2023063635. [DOI] [PubMed] [Google Scholar]
- 32.Devlin LA, Zhuopei H, Merhar SL, Ounpraseuth ST, Simon AE, Lee JY et al. Influence of eat, sleep and console on infants pharmacologically treated for opioid withdrawal: A post hoc subgroup analysis of the ESC-NOW randomized clinical trial. JAMA Pediatrics. Doi: 10.1001/jamapediatrics2024.0544. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data supporting this study are available on the NICHD Data and Specimen Hub (DASH). Researchers interested in accessing the data can submit a request through the DASH platform.