Neonatal Abstinence Syndrome and Childhood Mental Health Conditions, 2009–2015: Commercial Versus Medicaid Populations

Data from the IBM MarketScan Commercial Database from January 1, 2009, to September 30, 2015, were used to identify all paid claims for children ages 1–5 who had at least 5 years of insurance eligibility, as assessed by Healthcare Effectiveness Data and Information Set standards (i.e., an average of 87.6% covered days over the 5-year eligibility period)(4). Children were identified as having NAS if a diagnosis of NAS (ICD-9, code 779.5) appeared in the child’s administrative claims history at any point during the study period; all other children meeting the eligibility criteria were used as a comparison group. This method of identification deviates from Sherman et al., who identified NAS by using the NAS code and a code for a live birth event (ICD-9, codes V30–V39) (2). We did so to address sample size issues created by concerns surrounding discontinuity of health insurance from the live birth event in this population, considering the differences in Medicaid eligibility criteria for low-income pregnant women and children. Additionally, because our population identified children with claims in the database between ages 1 and 5, we assumed all birth events were live births. The ICD-9 codes identified by Sherman et al. (2) were used to ascertain mental health conditions. Pharmacy claims were also analyzed for the commercially insured population; no information on pharmacy claims was analyzed for the Medicaid population. Pharmacy prescriptions were obtained with the Medi-Span Generic Product Identifier (5) codes to identify drug group classifications for psychotropics and anxiolytics, antidepressants, antipsychotics and mood stabilizers, and stimulants. Frequency analyses were conducted to ascertain intrapopulation differences in diagnosis and differences between the commercially insured and Medicaid-insured populations.

A total of 190 (<0.1%) children with NAS and 1,405,522 children without NAS met the eligibility criteria in the commercially insured population. The Medicaid-insured population identified by Sherman et al. (2) expressed rates of NAS that were 28.7 times higher than those expressed in the commercially insured population. In a post hoc sensitivity analysis, when the requirement for a participant’s continuous insurance eligibility was relaxed to 3 years and 1 year, the rate of NAS increased <0.1% under both scenarios (see online supplement).

The results of the frequency analysis are shown in Table 1. In the commercially insured population, the prevalence of any mental health condition was higher among children with NAS compared with children without NAS (35.8% compared with 22.3%). This disparity appears to be driven by the high prevalence of children diagnosed as having developmental delays (25.3% versus 8.2%, respectively). A small sample of children with NAS limits a more robust comparison and thus requires cautious interpretation, but prevalence of the other mental health conditions examined in the commercially insured population was largely comparable to rates in the Medicaid population. In the commercially insured population, the likelihood of a hyperkinetic syndrome of childhood (which includes attention-deficit hyperactivity disorder [ADHD]) diagnosis tended to be higher among children without NAS than among children with NAS (7.3% compared with 6.8%, respectively). This finding deserves further attention, considering other published literature reported higher rates of ADHD (6) or hyperkinetic syndrome of childhood (2) among children diagnosed as having NAS.

Among the 68 children with NAS in the commercial population who were diagnosed as having one of the studied mental health conditions, the top five drug classifications of filled prescriptions were anti-infective agents and penicillin (N=47, 69.1%), antiasthmatic and bronchodilator agents (N=37, 54.4%), cephalosporins (N=35, 51.5%), macrolides (N=35, 51.5%), and dermatological agents (N=33, 48.5%). Among the 68 children with NAS and a co-occurring mental health condition, 12 children (17.6%) had a paid claim for any psychotropic medication. Of the 190 children with NAS, 13 children (6.8%) had paid claims for psychotropics, 11 (5.8%) for stimulants, 4 (2.1%) for anxiolytics, 2 (1.1%) for antipsychotics, and 1 (0.5%) for antidepressants.

The prevalence of NAS in the population with Medicaid identified by Sherman et al. (2) was 28.7 times higher than in the commercially insured population. Although previous literature has estimated that the population of infants with NAS served by public payer sources exceeds the population covered by commercial insurance (3), the small sample in this study could indicate a gap in continuous coverage by commercial insurance for children with a history of NAS. From a research perspective, analyzing one commercially insured population may yield an insufficient sample size to adequately study questions related to NAS.

In both the commercially insured and Medicaid-insured populations (2), young children with a history of NAS had higher rates of early neurodevelopmental diagnoses compared with children without NAS. Our results are consistent with prior studies that have found neurodevelopmental risks among young children exposed to opiates (7). Detection bias may influence the higher rates demonstrated in both populations, given that many infants with NAS have access to follow-up in a specialized developmental clinic or early intervention program as well as ongoing screening and referral for developmental or mental health concerns (8).

Understanding the variation in the prevalence of co-occurring mental health conditions and in prescription histories between commercially insured and Medicaid-insured populations of children with NAS has important implications. Even when the prevalence of childhood mental health conditions is similar by payer type (which one would expect in the absence of other selection bias), important payer-related variation may exist (9). For instance, differences in ADHD diagnostic patterns between children with and children without NAS in both the Medicaid-insured population and its commercial counterpart may indicate important disparities in health care practice that require further research, similar to the work assessing disparities in access to dental care for children as a result of Medicaid payer status (10). Payer source may indicate overdiagnosis of these syndromes in commercially insured populations, which has been hypothesized for thyroid cancer diagnoses (11).

Although our data do not elucidate the detailed exposures that may underly the differences in rates of mental health conditions between the two insurance populations, we note that differences in these outcomes may be associated with exposure to opioids during in utero developmental stages. In recent years, NAS has become synonymous with neonatal opioid withdrawal syndrome, which is distinct from NAS related to other substances. Although recent literature has examined rates of prescription opioid use among pregnant women (12), a gap in the literature exists in quantifying the rates of opioid use among pregnant women by insurance status. Payer source may be an important variable to consider in future analyses when assessing mental health conditions developing in childhood and may act as a proxy for the differences in early childhood exposures (e.g., foster care) (13) and social determinants of health that influence a mother’s likelihood of drug use as well as treatment-seeking behaviors (14).

Although absolute counts were low, the data also indicate that psychotropic agents are being prescribed in the population of children with NAS and a co-occurring mental health condition. Although no information was analyzed in the Medicaid population (2), the estimated 6.8% prescription rate in the commercially insured population is higher than rates found in a previous analysis of children age 4 and younger who were insured by Medicaid in 36 states (15) and lower than the rates found in a previous study of foster care children age 6 and younger (16). The use of psychotropics for young children remains a controversial issue given the lack of evidence surrounding the efficacy of these medications, and the use of psychotropics among children who have a history of NAS should be examined in future research.

As NAS rates rise, the need for research on childhood and developmental outcomes is urgent. Careful consideration of statistical power is warranted when using commercially insured populations to address this need. Nonetheless, elevated rates of mental health conditions were observed in the population of commercially insured children with NAS, consistent with the findings in a claims-based analysis of a Medicaid-insured population. Understanding the use of psychotropics among children with NAS is an area of future research.

The authors acknowledge Dr. Daniel Harris for data support.