Neonatal Neurobehavior Predicts Neonatal Opioid Withdrawal Syndrome Severity Prior to Treatment Interventions

Abstract

Objective

To utilize the NeoNatal Neurobehavioral Scale (NNNS-II) as a standardized and reliable tool to help guide treatment of neonatal opioid withdrawal syndrome (NOWS) with the goal of improving the consistency and quality of care provided to opioid-exposed newborns.

Study design

We examined NOWS severity through several indicators, including the number of pharmacological treatment medications, maximum dose of medications, length of treatment, and maximum score on NOWS observational tools to create an overall severity score for each newborn. Then we examined whether NOWS risk could be detected as early as 24 hours after birth using the NNNS-II to predict NOWS severity.

Results

Newborn neurobehavior prior to signs of NOWS predicted NOWS severity. Newborns who were more hypertonic and had poorer quality of movement had higher severity scores.

Conclusion

These findings indicate that NOWS risk detection may be enhanced with an early newborn neurobehavioral assessment.

Newborns exposed to opioids in utero are at risk for neonatal opioid withdrawal syndrome (NOWS). Some newborns become physiologically dependent while in utero and experience drug withdrawal following the cessation of exposure after delivery. These newborns can experience autonomic instability, high pitched cry, irritability, tremors, and difficulty feeding and sleeping.¹ The onset of signs of NOWS can vary as a function of maternal substance use with symptoms typically beginning within 12–72 hours after birth, although withdrawal signs can be delayed 5–7 days.² The American Academy of Pediatrics guidelines for NOWS hospital observations are: a minimum 3 days for immediate release opioids, 4–7 days for newborns exposed to buprenorphine and sustained-release opioids, and 5–7 days for newborns exposed to methadone.³ Although this stay is often essential for identifying signs of NOWS, this prolonged care places a significant burden on clinicians and contributes to higher health care costs.⁴

To guide treatment of NOWS, clinicians use observational tools to evaluate newborn central nervous system, metabolic, vasomotor, respiratory, and gastrointestinal disturbances.^1,5 Scoring scales for NOWS monitor progression and guide pharmacological treatment. The Finnegan Neonatal Abstinence Scoring System (FNASS) is 1 of the most widely utilized NOWS observational tools used every 4 hours to monitor signs of NOWS¹; however, it is also 1 of the longest scoring systems which may add a clinical burden. The Neonatal Withdrawal Inventory (NWI) is a less popular but shorter 8-point check list.⁵ If newborns reach a predefined threshold on either of these observational tools, pharmacological treatment is recommended.⁶ These observational tools are often subjective and not evidenced-based,⁷ and the choice of observational tool often depends on independent hospital protocols, with scoring varying between hospitals and hospital staff.⁴

Without a predictive model for NOWS onset or severity, most diagnoses occur at least 36 hours after birth, with signs sometimes delayed up to a week. Factors such as the opioid’s half-life, exposure dose, and/or polysubstance exposure can contribute to this delay, potentially postponing timely care. However, this extended observation may be unnecessary for approximately half of newborns who may not develop NOWS. Although prolonged monitoring is appropriate in some cases, early discharge could be a safe and practical option if supported by thorough risk assessment, enhanced outpatient follow-up, and comprehensive nonpharmacological care. In addition, the binary nature of a NOWS diagnosis oversimplifies its severity, limiting its predictive value and clinical utility.

The absence of early, reliable, and valid assessments for the treatment of NOWS may limit medical professionals’ ability to provide the most effective, timely, and appropriate medical interventions based on the latest evidence and best practices to opioid exposed newborns. Consequently, we examined whether NOWS risk could be detected around 24 hours after birth (prior to most signs of withdrawal) using a standardized neurobehavioral assessment—the NeoNatal Neurobehavioral Scale, Second Edition (NNNS-II)—to predict NOWS severity throughout the newborns’ hospital stay.

Methods

This study was approved by the Institutional Review Board at the University of Utah (00 124 221) on September 14, 2019 and Women and Infants Hospital (1 479 081–3) on February 13, 2020. Written consent was obtained for all participants. Enrollment and data collection began in 2019 and will be completed in 2025.

Participants

Participants included 211 newborns with prenatal opioid exposure who were part of a prospective longitudinal study examining the predictive validity of NOWS clinical markers. Data were collected from 2 hospital sites, Women and Infants Hospital of Rhode Island and the University of Utah Hospital. Mothers were recruited during their third trimester or soon after the birth of their newborn. A subsample of infants (n = 38, 18.0%) was excluded when hospital protocol switched to the Eat, Sleep, Console (ESC) assessment and treatment at the University of Utah hospital. Thus, the final study sample included 173 newborns. Demographics are described in Table I. Mothers and newborns were enrolled if opioid use was identified prenatally or at birth via medical records. Opioid use was defined as the use of prescription medications, like morphine, oxycodone, and fentanyl, prescribed for pain management, and/or illicit drugs, like heroin. Expecting mothers used an average of 2.8 (SD = 2.3) substances reported in Table II, consistent with other studies.^8,9 Newborns with congenital abnormalities, genetic syndromes, metabolic disturbances, or serious medical illnesses, such as sepsis, asphyxia, seizures, or respiratory failure, were excluded. Newborns were also excluded if they were unable to take oral medications or if their caregiver was unable to provide informed consent.¹⁰

Table I.

Demographic Information by Study Site

Demographic characteristics	University of Utah (n = 53)	Women and infants hospital (n = 120)
Newborn sex
Female	27 (50.9%)	62 (51.7%)
Male	26 (49.1%)	58 (48.3%)
Newborn gestational age, Mean (SD)	38.2 (1.8)	38.2 (1.7)
Income, N (%)
Less than $15,000	10 (23.3%)	17 (15.0%)
$15,000-$29,999	4 (9.3%)	15 (13.3%)
$30,000-$39,999	3 (7.0%)	9 (8.0%)
$40,000-$49,999	1 (2.3%)	5 (4.4%)
$50,000-$59,999	4 (9.3%)	1 (0.9%)
$60,000-$69,999	3 (7.0%)	5 (4.4%)
$70,000-$99,999	1 (2.3%)	3 (2.7%)
$100,000-$199,999	1 (2.3%)	3 (2.7%)
$200,000 or more	1 (2.3%)	1 (0.9%)
Unknown/prefer not to say	15 (13.9%)	54 (47.8%)
Maternal education, N (%)
8th grade or less	1 (2.0%)	2 (1.8%)
9th to 12th grade (did not graduate)	16 (31.4%)	21 (19.1%)
High school graduate/received diploma	19 (37.3%)	37 (33.6%)
Trade or technical school	4 (7.8%)	10 (9.1%)
Partial college or associate’s degree	11 (21.6%)	21 (19.1%)
College degree	0 (0.0%)	15 (13.6%)
Graduate degree	0 (0.0%)	4 (3.6%)
Maternal ethnicity
Hispanic/Latino	12 (25.5%)	22 (18.8%)
Not Hispanic/Latino	35 (74.5%)	95 (93.8%)
Maternal race
American Indian or Alaskan Native	1 (2.4%)	0 (0.0%)
Black	0 (0.0%)	6 (5.2%)
Multiracial	5 (11.9%)	23 (20.0%)
White	36 (85.7%)	86 (74.8%)
Maternal age, Mean (SD)	31.4 (4.2)	32.8 (4.9)

Table II.

Reported Maternal Drug Use During Gestation via Toxicology and Self-Report

Reported drug use	N (%)
Opioid treatment
Methadone	61 (35.3%)
Buprenorphine	77 (44.5%)
Both	11 (6.4%)
None	24 (13.9%)
Selective serotonin reuptake inhibitor use
Yes	55 (31.8%)
No	118 (68.2%)
Other opioid use
Fentanyl
Yes	51 (29.5%)
No	122 (70.5%)
Oxycodone
Yes	29 (16.8%)
No	144 (83.2%)
Morphine
Yes	26 (15.0%)
No	147 (85.0%)
Heroin
Yes	41 (23.7%)
No	132 (76.3%)
Other drug use
Alcohol
Yes	17 (9.8%)
No	156 (90.2%)
Nicotine
Yes	117 (67.6%)
No	56 (32.4%)
Marijuana
Yes	71 (41.0%)
No	102 (59.0%)
Illicit drug use*
Yes	63 (36.4%)
No	110 (63.6%)

We included the most prevalent drug categories; however, there may be additional drug use not reported. Maternal drug use is not mutually exclusive.

^*Illicit drugs include cocaine, phencyclidine, methamphetamine, and an illicit “other” category.

NOWS Measures

Treatment for NOWS was assessed using the modified FNASS¹ (n = 117, 67.6%) at Women and Infants Hospital and the NWI⁵ (n = 36, 20.8%) at the University of Utah Hospital. We were missing assessments for 20 newborns (11.6%). The ESC was used with 38 newborns, however, these infants were excluded due to their considerably differing NOWS observational tool, their reduced reliance on pharmacological treatments, and their typically shorter hospital stays, which limited the observable indicators of clinical severity. Treatment criteria for NOWS varied by site and included: three or more consecutive scores greater than 7 or 2 or more consecutive scores greater than 11 on the FNASS; one or more scores of 8 or higher on the NWI.

Other NOWS variables were collected from the medical records, including the need for pharmacological intervention, maximum dose of treatment medication (mg), number of additional medications to treat NOWS (eg, phenobarbital) and length of NOWS treatment (nonpharmacological and pharmacological).

Newborn Neurobehavior

Neonatal neurobehavior was assessed shortly after birth (M = 1.25 days old, SD = 0.70; Range = 18 hours to 120 hours [4 days]; Table S1 available at www.jpeds.com) by trained examiners using the NNNS-II. The NICU Network Neurobehavioral Scale (NNNS) examiners were blinded to treatment status and NOWS scores as the NNNS occurred prior to a NOWS diagnosis/treatment. The NNNS-II is a standardized comprehensive evaluation of newborn neurobehavioral performance that includes neurologic integrity, motor and behavioral function, and signs of stress/abstinence. The NNNS-II includes the most reliable and valid items from the original neurobehavioral assessment—the NNNS¹¹—which was developed for the at-risk infant and has been empirically established in studies of prenatal drug exposure.^12,13

Covariates

Additional demographic variables were collected at the birth visit or were extracted from medical records. We included study site, maternal education level, maternal race and ethnicity, maternal depression and/or anxiety diagnosis, maternal selective serotonin reuptake inhibitor use, and newborn sex as covariates in the analysis.

Statistical Analysis

All data analyses took place in R (version 4.3.2) and RStudio (version 2023.09.1).

We standardized NOWS observational total scores within each assessment (ie, FNASS or NWI scores) for all assessments that the infants were administered across their hospital stay. We then identified the maximum score for each infant and combined the 2 assessments.

Next, confirmatory factor analysis was used to evaluate the extent to which the observed indicators (eg, length of NOWS treatment) correlate with the hypothesized, underlying latent factor (NOWS Severity Score). Factor loadings were inspected to quantify the strength and direction of the relationships between the indicators and the latent factor. We used empirically-informed, standard components that are tracked regularly throughout NOWS care as indicators: Maximum score on the NOWS assessment, number of medications to treat NOWS (eg, addition of phenobarbital), maximum dose of primary treatment medication, and length of NOWS treatment (nonpharmacological and pharmacological), assessed throughout newborns’ hospital stay (M = 9.03 days; SD = 6.52). Newborns not treated with pharmacology were assigned a zero on maximum dose and number of medications, indicating no dose of any treatment medication. We derived the latent factor score from these indicators, weighted by the factor loadings.

Finally, structural equation modeling with maximum likelihood estimation was used to estimate full structural models. The first model examined the relations between the NNNS-II subscales and the NOWS Severity Score latent factor, while a second model adjusted for study covariates.¹⁴ We excluded newborns who showed withdrawal signs on their most recent NOWS assessment prior to the NNNS to focus the analyses on newborn before their onset of symptoms (n = 30, 17.3% assessed before 48 hours; n = 18, 10.4% assessed after 48 hours; n = 2, 1.2% missing). Thirty-eight (30.0%) of the newborns in the analysis were administered treatment after they were assessed and 85 (49.1%) were not treated with NOWS medications (Table S1 available at www.jpeds.com). Model fit for all analyses was assessed via the chi-square goodness-of-fit test, Root Mean Square Error of Approximation (RMSEA < .08), and Confirmatory Fit Index (CFI > .90).¹⁵

Results

We tested a one-factor model of NOWS Severity using confirmatory factor analysis with our 4 severity indicators. Acceptable model fit was achieved, χ² (2) = 3.43, P = .180, RMSEA = .064, CFI = .997 (Figure S1 available at www.jpeds.com). The factor loadings were statistically significant (standardized ls = 0.558–0.954, all ps < .001).

Next, we estimated a full structural model to examine the relations between the NNNS-II subscales and the NOWS Severity Score latent factor; Model fit was acceptable, χ² (26) = 34.36, P = .126, RMSEA = .067, CFI = .956. NNNS-II Hypertonicity, Quality of Movement, and Arousal significantly predicted the NOWS Severity Score (Figure 1). For full results see Table S2 available at www.jpeds.com. We then adjusted for covariates—study site, maternal education level, maternal race and ethnicity, maternal depression and/or anxiety diagnosis, maternal selective serotonin reuptake inhibitor use, and newborn sex—χ²(56) = 73.67, P = .057, RMSEA = .073, CFI = .901. Adjusting for covariates, Hypertonicity and Quality of Movement remained predictive of NOWS Severity Score (Figure 2). For full results see Table S3 available at www.jpeds.com.

Figure 1.

Predictive relations between the NNNS-II summary scores and the NOWS Severity Score latent factor. Note. *P < .05; Standardized estimates are presented. Higher hypertonicity and lower quality of movement scores predicted higher NOWS severity scores. Maximum NWI or FNASS R² = .47, maximum dose of treatment medications R² = .90, number of treatment medications R² = .91, length of treatment R² = .20. Solid black lines represent statistically significant paths and dashed gray lines represent nonsignificant paths.

Figure 2.

Predictive relations between the NNNS-II summary scores and the NOWS Severity Score latent factor adjusting for covariates. Note. *P < .05, **P < .01; Standardized estimates are presented. Lower quality of movement and attention scores predicted higher NOWS severity scores. Maximum NWI or FNASS R² = .55, maximum dose of treatment medications R² = .95, number of treatment medications R² = .87, length of treatment R² = .22. Solid black lines represent statistically significant paths and dashed gray lines represent nonsignificant paths. Model is adjusted for study site, maternal education level, maternal race and ethnicity, maternal depression and/or anxiety diagnosis, maternal selective serotonin reuptake inhibitor use, and newborn sex.

Discussion

Dysregulated newborn neurobehavior, measured around 24 hours after birth, predicted NOWS severity throughout the newborns’ hospital stay. Specifically, newborns with more hypertonicity and lower quality of movement had higher NOWS severity scores. These findings indicate that NOWS risk may be evident before typical withdrawal signs are present, indicating that the NNNS-II may be an early clinical indicator of NOWS risk. Although the NNNS-II does not have the same large body of evidence as the NNNS, given its recent implementation, we suggest that the NNNS-II can also be reliably administered with opioid-exposed newborns to evaluate the neurobehavioral organization and functioning of newborns requiring pharmacological treatment for NOWS.^7,16,17

The current study extends the small but growing literature linking NOWS severity with newborn quality of movement, as well as with later infant cognition and motor development.^13,18 Exposure to opioids in utero can disrupt the development of neural circuits that are implicated in attention regulation and motor control.¹ Our results indicate that we may be able to improve early detection of NOWS risk by using newborn assessments specifically designed to identify these early neurobehavioral disruptions.

Standardized and rigorous assessments, such as the NNNS-II, may aid in the identification of newborns who need pharmacological treatment or extended observation while receiving nonpharmacological interventions, from infants who could be discharged sooner to recover in their home environment. At home, where infants experience greater comfort and fewer disruptions, nonpharmacological interventions may be especially effective. These supportive measures not only promote maternal-newborn bonding but also empower families to actively participate in their infant’s care. Furthermore, by facilitating earlier discharges, hospitals can reduce the costs associated with prolonged NICU stays, freeing up critical resources for other patients. Improved protocols for managing NOWS also lessen the strain on hospital staff, as infants with NOWS often require intensive monitoring and care. A standardized early identification protocol could lead to greater consistency in care across institutions, reducing variability in treatment approaches and outcomes. However, further research is needed before adopting the NNNS-II or similar rigorous tools to guide NOWS treatment decisions. Finally, factors such as the need for supplementary support, including the development of a Plan of Safe Care, should be carefully evaluated when considering earlier discharges.

A notable strength of our study is its comprehensive approach, which evaluates multiple indicators of NOWS severity beyond the need for pharmacological treatment. This approach addresses variability in symptoms, capturing severity scores in infants who don’t require medication but are still receiving necessary monitoring and caregiver support on nonpharmacological care strategies. Although some infants may have more severe signs of NOWS requiring additional medications and larger doses with extended hospital observations, others may recover with less intensive care. This broader perspective allows for more variability in addressing severity, going beyond simply indicating whether an infant is on pharmacology. Nonetheless, we acknowledge limitations to our findings. Although developing a severity measure was a strength of this study, the metric may present challenges for clinical interpretability. Further, the observational tools used to assess NOWS varied within and across hospitals in our study, which may impact the consistency of severity scoring. Currently, there is no standardized method for combining these observational measures. In addition, we found that assessing severity in newborns with ESC management posed unique challenges due to the differences in observation and treatment protocols, decreased reliance on pharmacological treatments, and shorter hospital stays. As the ESC approach becomes more widely implemented, developing standardized methods for integrating observational tools and identifying appropriate indicators of NOWS severity will be crucial for future NOWS research.

Identifying NOWS severity 24 hours after birth could help streamline early care for newborns exposed to opioids, reducing the necessity for extended hospital stays. Additional studies with larger, more geographically diverse samples are necessary to validate these findings. With these expanded data, the NNNS could be refined into a more practical, streamlined version for clinical use. The NNNS-II has the potential to further support this patient population and their families.

Glossary

CFI

Confirmatory Fit Index

ESC

Eat, Sleep, Console

FNASS

Finnegan Neonatal Abstinence Scoring System

NNNS

NICU Network Neurobehavioral Scale

NNNS-II

NeoNatal Neurobehavioral Scale

NOWS

Neonatal opioid withdrawal syndrome

NWI

Neonatal withdrawal inventory

RMSEA

Root mean square error of approximation