Prevalence and Predictors of Early Intervention Referral Among Substance-Exposed Newborns Free

This study is a secondary analysis of data collected through the CHoSEN QIC. Previously published specifics about CHoSEN QIC include the initiation of eat, sleep, console techniques; reduced use of pharmacologic therapy; recommendations for breastfeeding eligibility; and optimizing both prenatal counseling and parental engagement.³  Participating birthing hospitals report deidentified patient-level data, including discharge data and EI referral, to the collaborative on a quarterly basis to preserve anonymity for infants born in hospitals with few SEN admissions. SEN data collected by participating birthing hospitals are maintained in a Health Insurance Portability and Accountability Act-compliant centralized Research Electronic Capture database hosted at the University of Colorado.¹⁴ 

Mother–infant dyads from the CHoSEN database were included in analysis if infants had prenatal opioid ± substance exposure, as defined below, and if EI referral status during birth hospitalization was documented in the CHoSEN database. Opioid ± substance exposure was defined at the hospital level on the basis of maternal prenatal history, verbal screening, biologic testing during birth hospitalization, and/or infant biologic testing based on maternal prenatal history. Dyads were excluded from analysis if data were missing for infant sex, gestational age, and birth hospitalization length of stay (Fig 1).

The outcome of interest was whether an infant received a referral to EI during their birth hospitalization. Hospitals participating in CHoSEN received recommended practices for the care of SENs, including the recommendation for newborn providers to refer SEN to EI during birth hospitalization.

EI referrals in Colorado can be made by health care providers; health care professionals including therapists, social workers, and case managers; and by parents or caregivers. A formal assessment and evaluation are completed within 45 days of referral. Infants and children with an “established condition,” a diagnosis associated with a high probability of developmental delay, as determined by a panel of physicians, automatically qualify to receive EI services. This eligibility for services is independent of a documented developmental delay. For those without an automatically qualifying diagnosis, including infants with prenatal substance exposure, receipt of EI requires a delay of 33%+ in 1 of the 5 developmental domains (social–emotional, cognitive, communication, motor, adaptive). Although SEN is not currently an automatically qualifying diagnosis for EI services in Colorado, specific education and guidance were provided to participating CHoSEN hospitals regarding the risks of developmental delay and recommendation to submit an EI referral for all SENs during birth hospitalization. Referrals to EI were documented in the hospital’s electronic medical record and subsequently abstracted by local hospital staff into the Research Electronic Capture database as yes, no, or unknown.

Variables of interest included maternal sociodemographic characteristics including race, ethnicity, primary language, insurance at the time of delivery, and prenatal substance use, as well as infant clinical variables including gestational age, sex, birth hospitalization length of stay, receipt of pharmacologic therapy for withdrawal symptoms, level of hospital care received, receipt of in-hospital social work resources or child protective services (CPS) referrals, and discharge disposition. All variables, including demographic variables of race, ethnicity, and primary language, were defined by the local participating hospital and abstracted directly from the local electronic medical record. Birth hospitalization length of stay, captured as a continuous variable, was categorized according to average length of stay by delivery mode: 1 to 3 days for vaginal deliveries, 4 to 6 days for cesarean deliveries, and >6 days for prolonged hospitalizations. Level of care was captured according to the American Academy of Pediatrics levels of neonatal care and ranged from Level I well baby nursery to Level III NICU.¹⁵  Substance exposure(s) were categorized as previously published by the CHoSEN collaborative and included the following categories:

1. medications for opioid use disorder (MOUD) including prescribed methadone or buprenorphine;

2. illicit opioids including nonprescribed buprenorphine or methadone, fentanyl, and/or heroin;

3. nonopioid illicit substances including cocaine; methamphetamines; methylenedioxymethamphetamine, commonly referred to as ecstasy; and phencyclidine;

4. prescription medications including opioids, selective serotonin reuptake inhibitors, benzodiazepines, nonopioid medications for pain, etc;

5. cannabis and/or alcohol; and

6. nicotine.¹⁶ 

The CHoSEN database collects specific substances individual; substances were subsequently categorized into the 6 categories described here for descriptive purposes and into the following 3 categories for analysis: single MOUD exposure, single non-MOUD exposure, and polysubstance exposure. Given that the CHoSEN data system captures birthing individuals with opioid ± other illicit substances, no infants were solely exposed to cannabis, alcohol, nicotine, or selective serotonin reuptake inhibitors alone without the presence of opioids.

We used χ², Fisher’s exact, and analysis of variance tests, as appropriate, to assess for statistical differences in maternal and infant characteristics by EI referral status. We developed Poisson regression models with robust error variance to assess the independent association of maternal and infant characteristics with risk of no EI referral during birth hospitalization. On the basis of significant bivariate results, we included the following maternal and infant characteristics in the regression model: insurance status, infant gestational age, length of stay, type of substance exposure(s), pharmacological agents for neonatal abstinence syndrome, level of care, and receipt of social work and CPS services during birth hospitalization. Birth year was included to account for increase in EI referral rates over time; birth hospital was included as a random effect to account for different referring patterns by site. Maternal primary language and race were excluded from the regression model because of high rates of missingness. All statistical analysis was conducted using SAS 9.4 (Cary, NC). This study was reviewed and deemed exempt by the institutional review board.

Fifteen Colorado hospitals enrolled in CHoSEN QIC reported data on 1611 SENs born between quarter 2 of 2017 and quarter 4 of 2022. After exclusions, the final analyzed cohort included 1222 mother–infant dyads (Fig 1) with dyads from each participating CHoSEN QIC hospital. Maternal and infant characteristics by EI referral status are presented in Table 1. Significant differences in EI referrals were identified by maternal race, insurance, infant gestational age, location of neonatal care, type of prenatal substance exposure(s), receipt of pharmacologic treatment of withdrawal symptoms, length of stay during birth hospitalization, and receipt of social work and CPS services. Frequency of maternal substance use by category is outlined in Fig 2 with MOUD (49%), illicit opioids (52%), and nonopioid illicit substances (48%) being the most common substance exposure. Categories are not mutually exclusive and 75% of mothers had documented polysubstance use. Figure 3 demonstrates overall percentage of infants receiving EI referrals over time, ranging from 23% for infants born in quarter 2 2017 to a high of 62% for infants born in quarter 1 2020, whereas Fig 4 outlines overall EI referral rate by site.

In the multivariable analysis (Table 2), those with private insurance were 26% (adjusted risk ratio [aRR] 1.26, 95% confidence interval [CI] 1.03–1.55) less likely to receive a referral compared with their publicly insured counterparts. Compared with infants exposed to MOUD only, those exposed to a single non-MOUD substance (aRR 2.15, 95% CI 1.67–2.76) and those with polysubstance exposure (aRR 1.58, 95% CI 1.26–1.97) were significantly less likely to be referred to EI. Families that did not receive social work services during birth hospitalization were 34% (aRR 0.66, 95% CI 0.53–0.82) more likely to receive an EI referral than those who received social work support. Conversely, those without involvement of CPS during birth hospitalization were 21% (aRR 1.21, 95% CI 1.01–1.44) less likely to receive EI referral. There was no difference in EI referral status by infant length of birth hospitalization when adjusting for all other including covariates.

Among this cohort of mother–infant dyads impacted by substance use during pregnancy, we demonstrated that, overall, EI referral rates during birth hospitalization are low for this statewide SEN population, with less than half of infants receiving referrals. EI referrals differed substantially among this cohort on the basis of several demographic and clinical characteristics, including insurance, type of substance exposure(s), and social work or CPS involvement.

Overall, this cohort demonstrates a significantly lower referral rate than other published EI referral data evaluating the SEN population. In Massachusetts, 77% of 256 eligible SENs were referred and 48% of referred infants enrolled for services.¹⁷  Significant state variability exists in which at-risk infants qualify for EI services under the Part C EI program. States are given considerable autonomy to determine eligibility criteria for infants on the basis of diagnosed or risk of developmental delay, including those with perinatal substance exposure. Few states (Massachusetts, New Hampshire, New Mexico, and West Virginia) have expanded those at risk of developmental delay to be eligible for Part C services, without documentation of existing delay(s).¹⁸  Referral and receipt of EI services are associated with restrictiveness of state-determined eligibility criteria, as documented by McManus et al.¹⁹  In communication from EI Colorado, 2020–2021 data demonstrate that 8972 of the 14811 (60.6%) children who had referrals to EI were evaluated and 6021 (67.1% of those evaluated) infants were deemed eligible for services. Among those eligible, 418 withdrew from the program before receipt of any EI services. In sum, 37.8% of those referred ultimately received services.²⁰  State data regarding the number of overall eligible children for EI referral, or specifics regarding perinatal substance exposure among those referred and evaluated, are not known.

In a recently published qualitative study of mothers in recovery for opioid use disorders, several barriers to initial and ongoing engagement with EI were noted.²¹  Mothers report intentionally limiting engagement with health care providers and governmental-supported programs because of experiences of judgment and stigma related to their substance use. Mothers also noted the guilt associated with their substance use during pregnancy, and despite their concern for the potential impact of their substance use on their infant’s postnatal development, they experienced significant barriers to EI engagement. The coexistence of maternal mental health conditions, especially anxiety and depression, may additionally limit engagement in EI, a system not currently structured to address maternal mental health and well-being.²² 

Inequities in EI referral based on infant or family demographics has been well established in the broader pediatric and NICU populations. A retrospective evaluation of nearly 15 000 children in the Denver metro area with developmental disability or delay demonstrated disparities in referrals, access, and receipt of EI services across several demographics. Black, non-Hispanic children and those with low annual income were significantly less likely to receive EI referrals and ultimately receive EI services, compared with their non-Hispanic white and more financially resourced counterparts.²³  In combination with our findings among the SEN population, it is clear that inequities for EI referral, access, and participation remain a challenge for all populations who may benefit from EI evaluation and services.²⁴ 

Regardless of substance exposure type, other programs may meet the needs of families and address infant development in the postnatal period, either in concert with EI or as a standalone program. In Colorado, non-EI programs including Child First and Nurse Family Partnership exist to cultivate strong and nurturing relationships. Plans of Safe Care, a federally mandated and state-administrated program, is designed to identify and establish a community safety net for families affected by substance use. A Plan of Safe Care, ideally developed during pregnancy and implemented during the birth hospitalization, addresses the following components: physical health, behavioral health, infant health and development, and parenting/family support.²⁵  EI or another modality of developmental follow-up referral is an integral part of this plan. For infants not referred to EI during birth hospitalization, the ideally warm handoff to the outpatient provider is an opportunity to share recommendations for outpatient referral. Additional work to determine the optimal provision of postnatal support for families affected by substance use is needed and may include a combination of EI and other home-based community support programs, on the basis of individual circumstances and needs.

Strengths include data from a large cohort across multiple hospital systems representing 60% of all Colorado births, making this data more generalizable at the state level. Limitations to this work should also be noted. These data capture EI referral during birth hospitalization; granular data on who completed the referral and follow-up data for infants referred posthospital discharge are not known. As previously published, referral to EI for evaluation and services does not equate to receipt of services with known high attrition rates.¹⁷  Our cohort included variable referral rates by site with many infants with missing EI referrals. This may represent underreporting of infants not referred to EI and thus acknowledge that our results are likely an overestimation of EI referrals. Cohort differences based on known versus unknown EI referral status are outlined in Supplemental Table 3. Missing EI referral status was disproportionately among Spanish-speaking families; combined with the large number of unknown and other maternal race categories, analysis to evaluate these disparities is limited.

EI referral during birth hospitalization was the primary outcome for this study and was measured by the presence of the referral in electronic health record at local sites. We recognize that some referrals may have occurred without electronic health record documentation and that some participating hospitals may not have a systematized workflow that includes consistent documentation of EI referrals. Indeed, sites who reviewed their local CHoSEN data with low EI referral rates have now embarked on process mapping efforts to clearly understand the EI referral workflow with the goal to standardize referral and documentation.

SENs are at risk for developmental delay and require a reliable and standardized process to ensure referral to EI services during birth hospitalization, a nearly universal touchpoint for mother–infant dyads with the health care system. Overall, there remains a paucity of literature evaluating the direct impact of EI services on the SEN population, 1 that remains at high risk for developmental and cognitive delays. Further development of site-specific quality improvement work to refer infants and state-level systems to follow-up referred infants to determine receipt of services and to assess the impact of EI on this population is crucial, as is the need for equitable EI referral across subpopulations.

We thank the families affected by substance use who were cared for in hospitals participating in the CHoSEN QIC, including those who contributed data for this analysis: Children’s Hospital Colorado - Colorado Springs, Denver Health Medical Center, Greeley Medical Center, Lutheran Medical Center, Medical Center of the Rockies, Memorial Hospital, North Suburban Medical Center, Parker Adventist Hospital, Parkview Medical Center, Platte Valley Medical Center, Saint Joseph Hospital, St Mary’s Medical Center, San Luis Valley Health, University of Colorado Hospital, and Valley View Hospital. We also thank the following steering committee members, without which the ongoing work of CHoSEN would not be possible: Dr Lauren Bruns, Dr Laurie Halmo, Dr Danielle Smith, Colleen Wheeler, PA-C, and Dr Erica Wymore. CHoSEN is supported by the University of Colorado School of Medicine Medicaid Upper Payment Limit Program, Colorado Perinatal Care Quality Collaborative, and Illuminate Colorado.