Standard Fixed-Schedule Methadone Taper Versus Symptom-Triggered Methadone Approach for Treatment of Neonatal Opioid Withdrawal Syndrome

This was a single-center QI study with monthly prospective data collection. Infants who were cared for between July 2016 and May 2018 from an urban academic medical center in Boston, Massachusetts, were considered for inclusion.

Our center is a tertiary-care urban medical center with a specialized prenatal program for women with substance use disorders. More than 90% of women in our clinic are on methadone or buprenorphine for opioid-use disorders (OUDs). Our institution practices a rooming-in model of care in which infants stay with their mothers on the postpartum unit until maternal discharge, and then mothers stay with their infants on the inpatient pediatric unit for continued monitoring and NOWS treatment. At maternal discharge, mothers are provided with a place to sleep to continue to room-in. Mothers are encouraged to breastfeed if they are enrolled in an addiction treatment program, have attended prenatal care visits, and have no urine toxicology evidence of illicit and/or unprescribed drugs close to the time of delivery. Our institution practices nonpharmacologic care as first-line treatment with a standardized nonpharmacologic care bundle, including encouragement of skin-to-skin contact and breastfeeding, a quiet environment, a cuddler program, and clustering of care.⁶  Parental and other caregiver presence is tracked every 4 hours in the medical record. Infants are monitored for 5 to 7 days in the inpatient setting for need for medication treatment. After discharge, the majority of infants are managed in a specialized pediatric primary care clinic for substance-exposed infants.

Inclusion criteria for data analysis included opioid-exposed infants who were (1) ≥35 weeks’ gestational age, (2) required pharmacologic treatment with methadone for NOWS, and (3) received care at our institution for their entire hospitalization. Exclusion criteria also included infants unable to take enteral methadone.

Between May 2016 and June 2016, our institution implemented a nonpharmacologic care bundle that included parental engagement, focus on nonpharmacologic care as primary treatment, and a switch to the eat, sleep, console (ESC) NAS assessment method.⁶  This nonpharmacologic care bundle was maintained during the study’s baseline period between July 2016 and November 2017.

All infants are assessed with the ESC NOWS care tool every 4 hours with scores documented in the electronic medical record (Supplemental Fig 3).^6,7  Infants with “yes” responses for difficulty eating, sleeping, or consoling per the ESC definitions have a team huddle. Nonpharmacologic care is optimized first. Infants who continued to have “yes” responses for poor eating, sleeping, or consoling are initiated on pharmacologic treatment. All nurses at our institution underwent a standardized ESC training with use of a training video with documentation of certification (≥80% interrater reliability) through an electronic health learning system in 2017. Annual recertification is required with documentation of ≥80% reliability on standardized training cases.

Between June 2016 and November 2017, first-line pharmacologic treatment was FSMT in which methadone was dosed every 8 hours (0.2–0.8 mg/kg per day; Fig 1). Methadone was titrated to effect, with increases permitted once every 16 hours (every other dose) then weaned by 10% of the maximum dose daily until the infant was down to 20% of the maximum dose before discontinuation. Infants who did not tolerate a wean could be reescalated by 10% and restabilized before resuming the wean. The infants were monitored for 48 hours off methadone before discharge from the hospital. Second-line adjunctive medication treatment of infants who reached maximum doses of methadone with continued poor ESC scores consisted of phenobarbital (for benzodiazepines, illicit drug, and/or polydrug coexposures) or clonidine (for selective serotonin reuptake inhibitor [SSRI] coexposure). Patients were weaned off clonidine as inpatients after weaning off the methadone, while phenobarbital patients were weaned as outpatients, starting by 20% per week 48 hours after methadone was stopped.

Our NOWS QI committee met on a monthly basis during the intervention period. The committee consisted of physician and nursing leaders from each of the pediatric inpatient units, staff nurses, obstetrical leaders from our substance use in pregnancy prenatal program, social workers, lactation specialists, and pediatric pharmacy leaders. The NOWS QI team developed plan-do-study-act (PDSA) cycles, and outcomes were tracked on a monthly basis. The ESC assessment method and nonpharmacologic care bundle were identical to the baseline period without further interventions to attempt to decrease pharmacologic treatment rates.

Just before the start of our intervention, all attending physicians on our pediatric inpatient units were educated during faculty meetings about the new STMA protocol (Fig 1). The NOWS STMA protocol was approved by the pediatric pharmacy committee and placed on the hospital intranet, and the electronic health record NOWS order set was modified to include STMA dosing as the default setting before STMA launch. Nursing staff on all units were notified about the new protocol via in-person huddles, staff meetings, and e-mail. All resident physicians were in-serviced on the NOWS treatment protocol at the start of each month by the NOWS QI team through a standardized lecture and handouts.

All infants cared for between December 2017 and May 2018 who met criteria to initiate medication per protocol, as defined by “yes” responses to ESC despite optimal nonpharmacologic management, were initiated on the novel STMA protocol. One dose of enteral methadone 0.07 mg/kg was given, which could be repeated up to every 8 hours for continued “yes” responses to ESC after a team huddle. Infants were monitored as inpatients for 48 hours after their last symptom-triggered dose before discharge from the hospital. Infants who received >4 symptom-triggered doses, particularly those who received medication intermittently for >4 days, with continued “yes” responses to ESC were considered for conversion to the FSMT beginning at 0.2 mg/kg per day. Infants who were converted from STMA to FSMT were continuing to have escalating NOWS symptoms despite STMA dosing. There was a concern for the risk of development of iatrogenic withdrawal in infants who received methadone for more than 4 days; thus, these infants were placed on a FSMT. The criteria for escalation of daily methadone dose and initiation of second-line agents were identical to the FSMT group.

During the baseline period, once infants were initiated on pharmacologic treatment, they were placed on both a continuous cardiac and oximetry monitoring. Infants who were initiated on medication while in the postpartum unit were moved to the nursery because of a lack of central monitoring capacity in the postpartum rooms and then were transferred to the pediatric unit once their mothers were discharged. On the pediatric unit, continuous cardiac and oximetry monitoring were provided for all infants, regardless of pharmacologic treatment, as is the unit’s standard practice.

In January 2018, the institutional cardiac-monitoring protocol was modified to intermittent cardiorespiratory monitoring for 4 hours after each STMA dose while in the nursery. Infants were permitted to be discontinued from cardiac monitoring and returned to the mother’s postpartum room after 4 hours if there was an absence of apnea, respiratory rates were between 20 and 70 breaths per minute, oxygen saturation was >93%, and heart rate was >80 beats per minute. Infants would continue to receive hourly vital signs in the mother’s postpartum room for 8 hours after each STMA dose. Cardiac-monitoring practices on the pediatric inpatient unit remained unchanged given that rooming-in was already possible with the availability of central monitoring.

Study staff reviewed the charts of all mother-infant dyads. Baseline characteristics were collected, including demographics, OUD treatment in pregnancy, psychiatric medications, illicit drug exposures in the third trimester (via urine toxicology screens), and pregnancy outcomes. For the infants, birth parameters and complications, breastfeeding (defined as any amount of breast milk consumed by the infant during the hospitalization), and details of NOWS treatment were collected. Data were hand abstracted and entered into an electronic database.

Outcome measures for NOWS included (1) median LOS (primary outcome), (2) median total methadone treatment days, (3) median total milligrams of methadone received, and (4) any adjunctive medication treatment of NOWS (yes or no). Adverse outcomes that were tracked as balancing measures included (1) 30-day readmissions, (2) 30-day emergency department visits, (3) seizures, (4) adverse cardiorespiratory events, and (5) escalations in care (ie, transfers to intensive care) for NOWS management. Lastly, the balancing measure of pharmacologic treatment rates for all opioid-exposed term infants was monitored on a monthly basis.

The NOWS QI team monitored all cases for protocol adherence on a monthly basis. Specifically, all cases in which STMA was used were reviewed by the lead physician on the team and presented monthly during NOWS QI meetings, with feedback given to the clinical team involved in the cases. Feedback was also sought out on a monthly basis from the resident, attending physicians, and nursing leaders on the inpatient units. The percentage of infants who were converted from STMA to FSMT was monitored on a monthly basis.

Demographic and maternal exposure frequencies were calculated among the 2 cohorts. NOWS outcome measures were then compared between the 2 cohorts by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). The primary outcome measure for NOWS was median LOS. Medians were chosen instead of means because of the nonnormality of the data and because of the significant variability in the outcome data, which may skew the means. We compared NOWS outcome measures in 2 ways: (1) as an intention-to-treat analysis, comparing all infants initiated on STMA (including those who ultimately failed STMA) versus those whose treatment initiated on FSMT during the baseline period and (2) an as-treated sensitivity analysis, comparing the subset of infants successfully treated with STMA versus those treated with exclusive FSMT during the baseline period.

We also created statistical process control charts to identify changes in continuous NOWS outcomes over time and to identify special cause variation, with use of shift rule (defined as 8 consecutive data points above or below the center line) and trend rule (6 successive points increasing or decreasing) to identify significant variability.¹⁶  We included interventions by start date in the statistical process control charts to infer the impact of the intervention on the outcomes.

We compared potential confounders between the 2 groups known to impact NOWS outcomes, including maternal opioid agonist therapy, illicit drug exposure, smoking, psychiatric diagnoses and medications, parental presence at the bedside (percentage of time over the hospitalization), infant sex, nonparental custody, NICU admission, and breastfeeding initiation. Each potential covariate was assessed with bivariate analyses for differences between groups and for association with NOWS outcomes.

Last, we compared adverse events, including hospital readmissions, emergency department visits within 30 days of discharge, seizures, and transfers to higher levels of care between cohorts using bivariate analyses.

All infants born at our institution during the study time period were eligible for the intervention. This study was approved by the Boston University Medical Campus Institutional Review Board as an exempt QI protocol. No informed consent was obtained because all infants received routine care for their NOWS management.

Of the 197 infants with prenatal opioid exposure cared for at our institution between July 2016 and May 2018, 76 met our eligibility criteria for data analysis. Ten (5.1%) were excluded because of prematurity with <35 weeks’ gestational age, 1 was transferred to another hospital before pharmacologic treatment was complete, and 110 (55.8%) did not require any pharmacologic treatment, consistent with other rooming-in studies.⁵  There were 48 infants in the FSMT group and 28 infants in the STMA group.

The demographics and maternal exposure of the 2 cohorts are shown in Table 1. Two-thirds of the infants were exposed to methadone in utero and the remaining one-third to buprenorphine. Coexposure to nicotine, selective SSRIs, and illicit drugs was similar between the 2 groups. There were fewer infants exposed to benzodiazepines in the STMA group, likely secondary to variation in psychiatric medication–prescribing practices over the course of the study (P = .004). Forty-three percent of the infants were transiently admitted to the NICU after birth for reasons unrelated to NOWS, such as late prematurity, transient tachypnea, or hypoglycemia, but were subsequently managed on the inpatient pediatric unit or the mother-infant unit after stabilization. One-third of the infants were in nonparental custody at the time of discharge.

In covariate assessment, only benzodiazepine exposure was associated with more methadone received by the infant (6.5 [SD 5.8] vs 3.9 mg [SD 4.0]; P = .04) but not significantly associated with any other NOWS outcome measure. Maternal opioid use, breastfeeding, nicotine exposure, SSRI exposure, illicit drug exposure, and parental presence were not significantly associated with any NOWS outcomes in bivariate analyses.

NOWS outcome measures for the FSMT and STMA cohorts are shown in Table 2. Infants initiated on the STMA protocol had a median LOS of 10.5 days (interquartile range [IQR] 10.5) versus 17.0 days (IQR 3.9; P = .003) in the FSMT group, with a 9.2-day difference in median methadone treatment days (2.5 [IQR 9.0] vs 11.7 [IQR 4.0]; P = .0001). There were fewer milligrams of methadone received in the STMA group and no significant difference in adjunctive agent use between the 2 cohorts.

Figure 2 shows the process control charts for NOWS outcomes across the study period. Special cause variation for LOS (Fig 2A), methadone treatment days (Fig 2B), and total milligrams of methadone received (Fig 2C) all occurred in January 2018, 1 month after the start of the STMA intervention. The outliers in Fig 2 between December 2017 and May 2018 represent the infants who were converted from STMA to FSMT.

Pharmacologic treatment rates increased from a baseline rate of 40% to 70% during the first 2 months of the intervention then went back down to 40% during the final months of the intervention. For our other balancing measures, there were no documented seizures during the entire study period and no transfers to higher levels of care for NOWS management or complications. No infants who were placed on intermittent cardiac monitoring had any documented adverse cardiorespiratory events. There was 1 hospital readmission related to NOWS in the FSMT group and none in the STMA group. There were 3 infants in the FSMT group and 2 infants in the STMA group who were seen in the emergency department within 30 days of discharge for reports of fussiness, reflux, and/or colicky behavior, which could possibly have been related to NOWS.

Within the STMA cohort, the average number of symptom-triggered doses received was 2.1 (SD 1.0) given on average day of life 4.3 (SD 1.5) for the first dose and day of life 6.6 (SD 3.5) for the last symptom-triggered dose. Six (21%) of the 28 infants in the STMA cohort were subsequently converted to a FSMT because of escalating signs of NOWS on STMA. For our process measure, the percentage of infants who were successfully treated with STMA increased over the course of the 6-month intervention period from 0% in December 2017 to 75% to 85% between January 2018 and April 2018 and to 100% in May 2018. Protocol adherence was high throughout the intervention period without notable deviations. Demographics of those who were successfully managed with STMA versus those for whom it failed did not differ significantly, with the exception of lower mean parental presence in the failure group (32.6% vs 65.0%; mean difference 32%; 95% confidence interval 9%–56%; P = .008). In the sensitivity analysis, outcomes of the 22 successful STMA infants were compared with the FSMT cohort (Table 3). Median LOS, methadone treatment days, and total milligrams of methadone received were significantly shorter in the successful STMA group versus the FSMT group.

In this study we demonstrated the success of a novel pharmacologic treatment approach for NOWS associated with a shorter LOS and less postnatal opioid exposure. The STMA was successful in most cases and was associated with a 9-day reduction in median methadone treatment days and 6.5-day reduction in median LOS compared with FSMT. In adjusted analyses, the mean difference in methadone treatment days remained significant, with 5.6 fewer days of methadone treatment in the STMA group. No additional hospital initiatives were ongoing to decrease LOS or methadone treatment days during the intervention period; thus, the benefits can be attributed to STMA.

NOWS treatment approaches are rapidly changing with revision of decade-old treatment paradigms. Starting with a fixed-schedule opioid taper may not be necessary to adequately treat an infant’s acute withdrawal symptoms and could unintentionally lead to the development of iatrogenic withdrawal with unnecessary prolonged and costly hospitalizations. The symptom-triggered medication approach is well aligned with the accumulating consensus that nonpharmacologic care should be first-line treatment of NOWS.^1,5,9  The symptom-triggered approach may allow the infant to room-in with the parent for a longer period of time. Our intermittent cardiac-monitoring pilot for 4 hours after each STMA dose allowed the infant to quickly return to the mother’s postpartum room without any documented adverse events. In addition to the short-term benefits of shorter hospitalizations, there may be long-term benefits of less exposure to opioid medications in the first month of life.¹⁷  Previous studies have revealed improved neurobehavioral profiles at 6 weeks of age in infants who did not require pharmacologic treatment versus those who received FSMT.¹⁷ 

A subset of our infants failed STMA and were subsequently converted to FSMT. This may be due to lack of comfort by the providers in giving multiple symptom-triggered doses or there being a subset of infants who will require a more prolonged course of opioids because of a more severe phenotype of NOWS. Because of staff concerns for escalating symptoms despite frequent STMA dosing and the possible creation of iatrogenic withdrawal with additional STMA dosing, this subset of infants were converted to FSMT. Previous studies in infants in an intensive care setting have revealed that 50% of infants exposed to 5 days or more of opioids and 100% exposed to 9 or more days of opioids will develop iatrogenic withdrawal.¹⁸  Methadone has a long half-life of 15 to 60 hours; thus, the 48-hour observation period after abrupt cessation of repeated doses of methadone was thought to be insufficient to ensure the infant would not subsequently develop withdrawal after discharge. Although our sample size was too small to make any definitive conclusions, these infants who were converted from STMA to FSMT tended to have less parental presence, with a trend toward more illicit drug exposure. This group of infants should be targeted in future studies to improve on the symptom-triggered treatment approach for all infants with NOWS. This could include additional optimization of nonpharmacologic interventions, the development of guidelines for when to convert infants to fixed-schedule dosing, the use of higher methadone dosing for STMA, or the use of buprenorphine instead of methadone for symptom-triggered treatment. In the current study, despite the STMA failures, overall LOS for the STMA-initiated cohort was shorter than the FSMT cohort, and we argue that the symptom-triggered approach should be first-line treatment on all infants and conversion to fixed-schedule dosing be second-line treatment.

A strength of this study is that it was the first to compare NOWS hospitalization outcomes by using a symptom-triggered treatment approach with a fixed-schedule approach as well as the first study to examine the use of symptom-triggered methadone. The novel idea to dose opioids for NOWS on a symptom-triggered basis was first proposed by Grossman et al⁸  as part of a QI initiative. However, no statistical comparison of NOWS outcomes was performed, and no standardized treatment algorithm was presented. We elected to use methadone instead of morphine, given that it was the standard of care of our institution for NOWS and the growing evidence that methadone is superior to morphine based on recent randomly assigned controlled trials.^14,15  Methadone is fast acting and long lasting in comparison with morphine. Using approaches from alcohol detoxification, longer-acting benzodiazepines are typically preferred to shorter-acting agents because they are thought to help patients autotaper, requiring an overall lower number of doses.¹³ 

This study has limitations that warrant mentioning. The retrospective nature of the study is limiting because there may be other confounding variables and bias that would be eliminated with a randomized control trial design. Also, generalizability may be limited because not all hospitals have well established nonpharmacologic care bundles with rooming-in capacity in which symptom-triggered dosing will likely be more effective. All our mothers were also on prescribed medication for OUD; thus, the response to symptom-triggered treatment of infants exposed to shorter-acting opioids or illicit opioids is unclear. Also, we did not examine longer-term outcomes of the symptom-triggered treatment approach. Lastly, this study is limited by the small sample size.

STMA was associated with decreased hospital LOS, opioid treatment days, and total amount of opioid medication required to treat NOWS in comparison with FSMT. Larger, more rigorous studies using this symptom-triggered medication approach that include long-term infant outcomes as well as comparison trials using different symptom-triggered medication regimens are warranted. This has the potential to become the new treatment paradigm for NOWS.

We acknowledge Camilla Farrell, PharmD, for her role in developing the STMA treatment protocol and the research assistants who helped with data collection for this project, notably Jean Devera, Angela Nolin, and Leonie Hoyo. We also acknowledge our entire Boston Medical Center NOWS QI team and all the nurses and physicians on the inpatient pediatric units for participation in this initiative. We thank our obstetrical partners in Project RESPECT for their continued support.