Implementation of an Opioid Weaning Protocol at a Tertiary Care Children’s Hospital

This single-center retrospective cohort study was conducted at the Children’s Hospital of Illinois from September 2012 to December 2020. Children’s Hospital of Illinois has roughly 7000 total pediatric and neonatal inpatient admissions per year, with 700 admissions to the PICU. Approval for this study was obtained from the institutional review board through the University of Illinois College of Medicine at Peoria. Informed consent was waived. Subjects were identified via medication utilization reports for methadone orders during the study period. Children were included if they had received opioid for 5 or more days in the PICU, age 18 years or less at the time of admission, and the attending service was the pediatric intensivist or hospitalist. Patients were excluded if they were admitted to the neonatal unit, expected to have a wean duration of > 48 hours remaining on discharge, required transfer to another facility before wean completion, had a prolonged hospitalization of >72 hours or more after wean completion, or if they died before discharge.

The study was conducted in 3 phases: preintervention (phase 1), standardized pharmacist-managed methadone wean protocol (phase 2), and addition of the WAT-1 tool (phase 3) in concordance with implementation of quality improvement projects. The phase 1 preintervention group included patients admitted between September 2012 and March 2014 during which time there was not a standardized weaning protocol, assessment tools, or nursing interventions. Patients in phase 1 were weaned off methadone at the discretion of the attending provider. Phase 2 occurred between April 2014 and February 2016 during which time there was an implementation of a pharmacist-managed methadone wean protocol based on the patient’s risk of withdrawal (Supplemental Document 2). Three standardized weans were developed with wean selection dependent on the duration of opioid exposure: low risk with <15 days of opioid infusion, moderate risk with 16 to 30, and high risk with ≥ 31 days.^5–7  Phase 3 occurred between March 2016 and December 2020 with the addition of the twice daily nursing assessments using the WAT-1 tool, which was integrated into the electronic medical record system. The WAT-1 is a validated tool to identify iatrogenic withdrawal syndrome in children age 2 weeks to 18 years in the PICU or general pediatric unit.¹  Institutional WAT-1 scoring was started on the first day of opioid weaning and continued twice daily until 72 hours after the last opioid dose. WAT-1 scores were used to expedite, hold, or extend methadone tapers (Fig 1). The guidelines included recommendations for an as needed enteral opioid dose if the patient experienced symptomatic withdrawal. Withdrawal was defined as a WAT-1 score > 3. A WAT-1 score has a high sensitivity (87%) and specificity (88%) to predict withdrawal in validation studies.¹  Although a WAT-1 score of 3 or more is associated with withdrawal, institutional weaning guidelines recommend to hold for a score of 4 or more. The use of a higher score was intended to allow for low levels of withdrawal and expedite weans. Diagnosis of withdrawal, holding methadone weans, and escalation of methadone doses were performed at the discretion of the primary care team before Phase 3. Each chart was reviewed for documentation of withdrawal, a delay in weaning, or an increased dose by medical personnel. If the patient was described in a note as suffering from withdrawal, a wean did not occur in 2 days because of withdrawal (wean held), or the dose was increased for any reason, they were classified accordingly.

Data collected included patient demographics (age, sex, weight), admitting diagnosis, concurrent lorazepam and/or clonidine wean, history of previous opioid exposure, overall length of stay (LOS) in days, PICU LOS in days, opioid infusion duration in days, methadone duration in days, number of enteral opioid rescue doses, methadone wean duration in days, initial risk group (low, moderate, or high), and documented withdrawal events. Previous opioid exposure was defined as any previous orders for opioids from birth to the current admission that were documented in the health-systems electronic medical record. The methadone duration included the entirety of methadone prescribing including transition from a continuous infusion. The methadone wean duration started with the first dose reduction including transitions from intravenous to enteral therapy that were not adjusted to account for bioavailability. Withdrawal was subdivided into 3 categories: withdrawal without taper modification, withdrawal necessitating a temporary hold on wean, and withdrawal necessitating dose escalation.

Analysis was performed by using the open source statistical program R (version 4.0.0) against a 2-sided alternative hypothesis with a significance level of 5% (P < .05), unless otherwise specified. Nonparametric Kruskal-Wallis tests were used to compare numeric variables, such as LOS. χ² tests, or Fisher’s exact tests were used to compare categorical or binary variables, such as documented withdrawal events. Post hoc, pairwise comparisons were also performed between groups, and P values were adjusted for multiple comparisons by using a Holm correction. The adjusted P values are reported as the “adjusted p.”

Data were analyzed for 115 patients from 3 phases with 36 patients in phase 1, 43 patients in phase 2, and 36 patients in phase 3. No significant differences were found between phases in patient age, weight, sex, PICU LOS, previous opioid exposure, and total opioid exposure before wean initiation (Table 1). Concurrent lorazepam weans declined (86% in phase 1, 74% in phase 2, and 56% in phase 3) throughout the course of the study, whereas the percentage of patients with a clonidine wean increased (36% in phase 1, 60.5% in phase 2, and 64% in phase 3) (Table 2). Adjusted pairwise comparisons of opioid exposure before wean initiation, which was used to define risk, found no significant difference between phase 1 and phase 2 (adjusted P = .149), phase 1 and phase 3 (adjusted P = .767), or phase 2 and phase 3 (adjusted P = .149) (Table 3).

A significant difference was found between phases for overall LOS (P = .009). Pairwise Wilcoxon rank tests with a Holm correction for multiple comparisons found the LOS from phase 1 (median: 25.2 days, interquartile range [IQR]: 17.6–28.8) was significantly higher than phase 3 (median: 17.2 days, IQR: 14.8–22.2), adjusted P = .005. A statistically significant difference was not identified between phase 1 to phase 2 and phase 2 to phase 3 (Tables 2 and 3).

The methadone wean duration was also significantly different between the 3 phases (P = .004). The standardization of weaning regimens and implementation of WAT-1 monitoring corresponded with a 3-day decrease in methadone weans between phase 1 and phase 3 (median: 8.62 days, IQR: 6.30–13.90 versus median 5.16 days, IQR: 3.61–7.55, adjusted P < .001). Neither the pharmacist-managed methadone wean protocol nor the addition of WAT-1 scores to the protocol as single interventions were associated with a statistically significant reduction in wean durations (Tables 2 and 3).

The proportion of patients experiencing withdrawal differed significantly between the 3 phases (P = .003) with 20 patients (56%) in phase 1, 15 patients (35%) in phase 2, and 6 patients (17%) in phase 3 described as having withdrawal. Pairwise comparisons found a significant difference between phase 1 (56%) and phase 3 (17%), adjusted P = .004. Once again, no statistical differences were noted between phase 1 to phase 2 or phase 2 to phase 3 (Tables 2 and 3).

The proportion of patients requiring an as needed “rescue” dose also differed significantly between 3 phases (P = .005) with 11 patients (31%) in phase 1, 12 patients (28%) in phase 2, and 1 patient (3%) in phase 3. Corrected, pairwise Fisher’s exact tests found a significant difference between phase 1 and phase 3, adjusted P = .009, and also phase 2 and phase 3, adjusted P = .009. There was no significant difference identified between phase 1 and phase 2 (Tables 2 and 3).

The proportion of patients experiencing withdrawal that necessitate holding a wean or increasing a dose differed significantly between phases (P = .006 and 0.014, respectively). Approximately 44% (n = 16) of children in phase 1, 26% (n = 11) in phase 2, and 11% (n = 4) in phase 3 suffered withdrawal that necessitated holding the wean. Methadone doses were increased in 11 patients (31%) in phase 1, 6 patients (14%) in phase 2, and 2 patients (6%) in phase 3 (Table 2). Pairwise comparisons found phase 1 was significantly higher than phase 3 for both assessments (adjusted P = .011 and 0.036, respectively). A statistically significant difference was not elucidated between phase 1 to phase 2 or phase 2 to phase 3 for either assessment (Table 3).

This study describes associated outcomes of a standardized pharmacist-managed methadone wean protocol combined with an objective bedside assessment tool of signs and symptoms of opioid withdrawal. The combination of the standardized weaning regimen and nursing assessment via the WAT-1 tool was associated with a statistically significant decline in overall length of stay and wean duration, whereas the individual interventions themselves were not associated with identifiable changes in these quality measures. Despite a more rapid opioid wean duration, documented withdrawal declined dramatically from 55.6% of all children having withdrawal in phase 1 to only 16.7% of all children having withdrawal in phase 3. Furthermore, it was also demonstrated that there was no statistical difference in rates of withdrawal, wean holds, or withdrawal necessitating increased opioids doses with individual interventions. The standardized opioid wean and nursing evaluation using WAT-1 in tandem was associated with significant changes in wean duration and withdrawal events when evaluated together. The only individual measure that was statistically different between phase 1 to phase 2, and phase 2 to phase 3 was a decline in administered rescue doses (P = .009).

Our results are similar to previous reports in which standardized opioid weans were implemented for pediatric inpatients. Steineck and colleagues described their experience with a 4-tiered stratified pharmacist-managed methadone taper. The transition to a pharmacist managed taper was associated with a decrease in total LOS, methadone taper duration, and as needed opioid doses. Despite these improvements, they were unable to identify a change in withdrawal rates.⁵  The pharmacist managed wean protocol used in the current study was derived from the protocol described by Steineck et al with a few revisions. One such revision included avoiding opioid tapering when children had received <5 days of opioids. In this case, opioid sparing therapies such as a clonidine taper or as needed opioid administrations may have been used. Amirnovin and colleagues described a reduction in opioid weans by 11 days with a 4-tiered weaning regimen that used 10% to 20% weans in a pediatric cardiac ICU setting.⁸  Other studies with patient populations ranging from neonates to adults have been unable to identify a change in length of stay with interventions.^9,10  Although these results are in slight contrast to our current study, we suspect that the combination of a pharmacist driven weaning schedule paired with objective bedside assessment of withdrawal is what led to significant changes in all patient- and hospital-level metrics we assessed.

Several different tools to assess iatrogenic withdrawal in pediatrics have been used in previous studies. In addition to the WAT-1, the Opioid and Benzodiazepine Withdrawal Score and Sophia Observation Withdrawal Symptoms scale have been used and previously validated.^10–12  The Modified Finnegan Neonatal Abstinence Syndrome monitoring tool has been used as well for evaluation of neonatal abstinence syndrome after in-utero opioid exposure.⁹  WAT-1 scores were used to identify rates of withdrawal in this study given the tool’s utility across a wide age range and ease of use. This study identified a greater association with declines in length of stay, wean duration, as needed opioid administration, and rates of withdrawal with combined WAT-1 bedside assessment and a standardized weaning regimen. The use of a WAT-1 score has been validated to identify withdrawal with scores of 3 or higher. In this study, methadone weaning continued as long as scores were < 4. This score was used in an effort weigh risk of withdrawal the risks of prolonged opioid exposure. Therefore, a WAT-1 score of 4 or greater was used to define opioid withdrawal. The institutional guidelines reflect this and could account for the decline in rates of withdrawal, increased doses, and wean holds. However, the greatest decline in the rates of withdrawal, halting weans, and increased methadone doses was between phase 1 and phase 3, potentially reflecting that using both standardized weans and objective withdrawal monitoring will have the greatest success in improving outcomes.

To our knowledge, this is the first study to evaluate both a pharmacist-managed methadone wean protocol with objective bedside assessment utilizing the WAT-1 tool. There are limitations to the interpretation of the outcomes seen in our current study, including the retrospective design at a single children’s hospital. Inability to identify a difference between groups may have been attributable to the small sample sizes within each phase rather than a lack of actual difference. In an effort to limit confounding by external factors on the length of stay associated with opioid weaning, all children who remained in the institution for > 72 hours after the completion of their wean were excluded from the analysis. Comorbidities such as feeding intolerance, surgical procedures, imaging, or social factors may delay discharge. The exclusion of these children, could limit the applicability of the results to all patients. In addition, other institutional changes or changes in hospital-based therapies may have contributed to the results of this study. For example, over the 8 years of data collection, improvements in care such as enhanced use of noninvasive mechanical ventilation of increased utilization of dexmedetomidine for sedation may have contributed to the outcomes and length of patient accrual seen in this study between phase 1 and phase 3. The majority of children were on concomitant medication weans which could have confounded withdrawal rates. In these instances, withdrawal from 1 agent would have been indistinguishable from the other. Withdrawal from any agent would have been documented as withdrawal with the potential for holding the opioid wean or increasing methadone doses. Finally, it is unclear from the current study what affect the WAT-1 tool alone may have had in our patient population, given that in phase 3 of our study period, a standardized pharmacist-managed weaning protocol was already in place at the time of WAT-1 implementation. However, as patient and hospital level quality measures differ from previous studies, the combination of the paired interventions could have been associated with the outcomes seen in this study.

In conclusion, this study suggests that a standardized, pharmacist-managed methadone wean in conjunction with a structured objective bedside assessment of opioid withdrawal using the WAT-1 tool may be associated with decreased methadone wean duration, patient withdrawal symptoms, and overall hospital LOS. Our findings could have broad implications for pediatric patients requiring an inpatient opioid wean as institutions seek standardized approaches to improve patient outcomes and hospital measures of quality.

Thank you to Dr Lauren Cummings for her involvement in creating the study protocol.